tropical legumes for sustainable farming systems in ...aciar.gov.au/files/node/555/proc115.pdf ·...

Tropical Legumes for Sustainable Farming Systems in Southern Africa

and Australia

Editors:

Anthony M. Whitbread and Bruce C. Pengelly

Australian Centre for International Agricultural ResearchCanberra 2004

Department ofPrimary Industries and Fisheries

The Australian Centre for International Agricultural Research (ACIAR) was establishedin June 1982 by an Act of the Australian Parliament. Its mandate is to help identify agri-cultural problems in developing countries and to commission collaborative researchbetween Australia and developing country researchers in fields where Australia has aspecial research competence.

Where trade names are used this constitutes neither endorsement of nor discriminationagainst any product by the Centre.

ACIAR PROCEEDINGS

This series of publications includes the full proceedings of researchworkshops or symposia organised or supported by ACIAR. Numbers in thisseries are distributed internationally to selected individuals and scientificinstitutions.

© Australian Centre for International Agricultural Research, GPO Box 1571,Canberra, ACT 2601.

Whitbread, A.M. and Pengelly, B.C., ed., 2004. Tropical legumes for sustainablefarming systems in southern Africa and Australia. ACIAR Proceedings No. 115,180p.

ISBN 1 86320 419 9 (print)1 86320 420 2 (electronic)

Cover design: Design ONE Solutions; photographs by A. Whitbread and B. PengellyTop row: farmers inspecting tropical legume,

Lablab purpureus

;preparing animal feed made from maize meal and lablab; farmers in cropof

Mucuna pruriens

. Middle row: farmers carrying straw used for animalfeed, animal bedding and straw brooms;

Mucuna pruriens

(centre).Bottom row: preparing animal feed from maize meal; Obert Jiri withfarming family collaborators; farmer and

Lablab purpureus

.

Technical editing and typesetting: Clarus Design Pty Ltd

Printing: Pirion Printing

3

Foreword

In much of southern Africa, large rural-based populations rely on subsistence andsmallholder cropping and livestock enterprises. The productivity and sustainability ofthese livelihoods are threatened by various forms of land degradation. As well, boththe quality and quantity of forage crops restrict animal production.

ACIAR funded a project in southern Africa to develop sustainable animal andcropping systems based on the integration of well-adapted forage plants and appro-priate agronomic and animal-management practices. The project was carried out byCSIRO Sustainable Ecosystems and collaborating institutions in South Africa andZimbabwe.

The project team identified legumes that can be used in rotation or intersown withcrops to improve the nitrogen status of soils, and used the farming systems modelAPSIM to investigate these dynamics. In particular, they selected forage plants thatcould be introduced into the degraded grasslands of South Africa and Zimbabwe toimprove animal production.

Through a participative action research approach, the project was very successfulin obtaining farmer adoption of forage and dual-purpose legumes in several farmingsystems in southern Africa.

Most of the papers in these proceedings come from a final meeting and a successfulreview of the ACIAR project. They focus on aspects of forage and ley legumes andinclude the interactions with animal and crop production, and farmer evaluation andadoption of new technologies in the various farming systems found in South Africaand Zimbabwe.

There is currently little information available to scientists and extension workers,so this publication, which describes some of the recent developments in farmingsystems research for South Africa and Zimbabwe, will be a valuable resource.

Peter CoreDirectorACIAR

Tropical legumes for sustainable farming systems in southern Africa and Australia, edited by A.M. Whitbread and B.C. Pengelly. ACIAR Proceedings No. 115.

(Printed version published in 2004)

5

Contents

Foreword 3

Preface 7

The Environment 11

The Farming Systems of the Zimbabwean District of Wedza and the Role and Adoption of Forage Legumes in Small-scale Crop–Livestock Enterprises 12

B. Chigariro

Description of the Biophysical Environment of Three Maize-producing Areas in the Limpopo Province of the Republic of South Africa and the Validation of APSIM to Simulate Maize Production 17

A.M. Whitbread and K.K. Ayisi

Evaluation of Forage Technologies 27

Tropical Forage Research for the Future — Better Use of Research Resources to Deliver Adoption and Benefits to Farmers 28

B.C. Pengelly, A. Whitbread, P.R. Mazaiwana and N. Mukombe

Selecting Potential Fodder Bank Legumes in Semi-arid Northern South Africa 38

G. Rootman, M. Mabistela and B. Pengelly

Assessment of the Variation in Growth and Yield of Diverse Lablab (

Lablab purpureus

) Germplasm in Limpopo Province, South Africa 44

K.K. Ayisi, M.P. Bopape and B.C. Pengelly

Evaluation of Herbaceous Forage Legumes Introduced into Communally Managed Rangelands in Zimbabwe 51

P.H. Mugabe, D. Majee, X. Poshiwa, B. Chigariro, B. Ndlovu, F. Washayanyika and N. Mukombe

Evaluation and Screening of Forage Legumes for Sustainable Integration into Crop–Livestock Farming Systems of Wedza District 58

R. Nyoka, N. Chikumba, I. Chakoma, P. Mazaiwana, N. Mukombe, and N. Magwenzi

Identification and Development of Forage Species for Long-Term Pasture Leys for the Southern Speargrass Region of Queensland 64

R.L. Clem and B.G. Cook

Farming Systems Research 81

The Role of Dual Purpose and Forage Legumes to Improve Soil Fertility and Provide High Quality Forage: Options in a Maize-based System in Zimbabwe 82

O. Jiri, B.V. Maasdorp, E. Temba and A.M. Whitbread



6

Growth and Symbiotic Activities of Cowpea Cultivars in Sole and Binary Cultures with Maize 92

K.K. Ayisi and P.N.Z. Mpangane

Lablab Density and Planting-Date Effects on Growth and Grain Yield in Maize–Lablab Intercrops 99

H.M. Maluleke, K.K. Ayisi and A.M. Whitbread

Grain Yield of Maize Grown in Sole and Binary Cultures with Cowpea and Lablab in the Limpopo Province of South Africa 106

P.N.Z. Mpangane, K.K. Ayisi, M.G. Mishiyi and A. Whitbread

Grain Sorghum Production and Soil Nitrogen Dynamics Following Ley Pastures on a Vertosol in Queensland, Australia 115

A.M. Whitbread and R.L. Clem

Mucuna, Lablab and Paprika Calyx as Substitutes for Commercial Protein Sources Used in Dairy and Pen-fattening Diets by Smallholder Farmers of Zimbabwe 126

C. Murungweni, O. Mabuku and G.J. Manyawu

Animal Production from Legume-based Ley Pastures in Southeastern Queensland 136

R.L. Clem

Participative Methodology and Adoption of Technologies 145

Identifying the Factors that Contribute to the Successful Adoption of Improved Farming Practices in the Smallholder Sector of Limpopo Province, South Africa 146

T.S. Ndove, A.M. Whitbread, R.A. Clark and B.C. Pengelly

Contrasting Adoption, Management, Productivity and Utilisation of Mucuna in Two Different Smallholder Farming Systems in Zimbabwe 154

B.V. Maasdorp, O. Jiri and E. Temba

A Socioeconomic Evaluation of an ACIAR-funded Project on the Use of Forage Legumes in Integrated Crop–Livestock Systems on Smallholder Farms in Zimbabwe 164

G.J. Manyawu, O. Jiri, B. Chigariro, P.R. Mazaiwana, R. Nyoka and D. Chikazunga

Using the Agricultural Simulation Model APSIM with Smallholder Farmers in Zimbabwe to Improve Farming Practices 171

A.M. Whitbread, A. Braun, J. Alumira and J. Rusike



7

Preface

‘Tropical forage and ley legume technology for sustainable grazing and croppingsystems in southern Africa’, ACIAR project number AS2/96/149, originated fromdiscussions held in South Africa and Zimbabwe from 1994 to 1996 and at a work-shop held in Polokwane (formerly Pietersburg), Republic of South Africa (RSA) inJune 1997.

Tropical regions of both countries support rural communities that depend oncereal and cash-cropping, and livestock production systems. These agriculturalsystems are under pressure because of the increasing population and, as a conse-quence, are often not sustainable. Forage-legume technology has the potential toimprove the sustainability and productivity of some of these farming systems by pro-viding a source of nitrogen (N). This can potentially be achieved by integratingforage legumes into the cropping components of mixed crop–livestock systems,intensive livestock systems such as dairying, and through the enhancement of thecommunal grazing lands.

These initial ideas were eventually formed into a four-year project that began on 1January 1999 with Zimbabwean, South African and Australian partners. It was initiallyhoped that Mozambique could be considered as a potential partner during the earlydeliberations, but the difficulties in identifying appropriate in-country collaboratorsmade this impossible.

The aims of the project were:

• to identify a range of ley legumes that can be used in rotation with crops, or as inter-row legumes with crops, to improve the N status of soils and to reduce soil erosion,resulting in sustainable and acceptable cropping practices

• to select forage germplasm that can be introduced into the degraded grasslands ofnorthern South Africa, Zimbabwe and Mozambique to improve animal productionusing procedures appropriate to smallholder systems

• to develop grazing and cut-and-carry livestock-management systems, includingdairying, that are sustainable and acceptable within the socioeconomic frameworksof the selected communities

• to demonstrably improve the decisions and practices of farmers managing forage-legume systems by collating research results from this project and from elsewhereand providing details of cost, benefits and management options to farmers andextension workers in an appropriate format

• to gain adequate understanding of seed production of the most-promising lines toenable the development of a local seed industry.

Several agencies have collaborated in this project. The Department of Research andSpecialist Services (DR&SS) in Zimbabwe, the Department of Agriculture and Envi-ronment in the Limpopo Province of RSA and the Commonwealth Scientific and



8

Industrial Research Organisation’s (CSIRO) Division of Tropical Agriculture (nowCSIRO Sustainable Ecosystems) in Australia have been the official partners in theproject. In addition to these contracted agencies, AGRITEX and the University of Zim-babwe have been important and strongly committed collaborators in Zimbabwe, ashave the University of the North and the Agricultural Research Council’s (ARC)Range and Forage Institute in South Africa, and the Queensland Department ofPrimary Industries in Australia.

The project had two sites in Zimbabwe: at Zana II and Dendenyore, both of whichare near to Wedza in Region IIB about 150 km southeast of Harare. There were severalreasons for the selection of these sites. These included:

• the climate, which has sufficient rainfall for relatively reliable cropping

• enthusiastic farmers

• the reliance of communities on rangeland (veld) as a base for animal production

• animal production practices in which dairy and beef production were already estab-lished commercial enterprises for smallholders

• proximity to established extension and research teams

• the contrasting land tenure (Zana II is a resettlement area and Dendenyore a com-munal land community) provided a diverse farming system.

The project also operated at two sites in South Africa: Tarantaaldrai and Dan.Tarantaaldrai was chosen because:

• it was an existing dairy cooperative, already producing milk for local consumption

• it had what appeared to be an enthusiastic group of farmers/cooperative members

• dairy production in this semi-arid region was based on grazing unimproved veld

• there was a well-established infrastructure including arable land for forage crop-ping

• there were on-site extension staff

The Dan site was chosen as it:

• represented a typical maize-based, smallholder farming system in Limpopo Prov-ince, in which legume pulses, especially groundnut and cowpea, were importantcrops used in intercropping or in maize rotations

• was close to established extension and research teams

• had an enthusiastic farmer community.

The project operated in three sites in southeastern Queensland. The QueenslandDepartment of Primary Industries’ Brian Pastures Research Station near Gayndahoffered an established grazing trial in which ley and phase legumes had been utilisedfor a number of years. This enabled the dynamics of soil nitrogen and organic matterin forage legume–cereal cropping production systems to be studied in some detail. Cin-nabar and Broad Creek are both commercial beef-cattle properties in the Burnett regionand were used in an experimental program to select new forage-legume genotypes thatare adapted to the subtropical environment.



9

A participative model of research, in which farmers and extension and research pro-fessionals developed research priorities associated with one of the many legume tech-nologies at their disposal, or at least could be associated with soil fertility issues, wasused at all sites. The diversity of farming systems addressed in the project, resulted ina wide range of legume technologies being researched. These ranged from compari-sons of new legume germplasm, to studies on the dynamics of soil N and carbon, theimpact of ley, phase and green manuring on subsequent cereal crops to the impact oflegume feeds on animal production. With the exception of the work at the Brian Pas-tures Research Station, all research was conducted on-farm.

The project included a number of formal training initiatives including the fol-lowing:

• Mr Richard Clark of the Queensland Department of Primary Industries designedand conducted a workshop on participative research for the entire project team atMurewa, Zimbabwe at the start of the project in April 1999.

• University of Queensland accredited studies on action learning were successfullyundertaken by several African team members over the course of the project.

• Team members Mrs Beatrice Chigariro, Mr Temba Elliot and Mr Brian Ndlovuundertook research within the project to meet requirements for undergraduatestudies at the University of Zimbabwe.

• Several postgraduate students were supported by the project and Mr JairusNkgapele and Ms Gifty Mishiyi (through the University of the North, South Africa)and Mr Obert Jiri (through the University of Zimbabwe) successfully completedMaster of Agricultural Science degrees or equivalent within the project.

• Team members Mr Owen Mabuku from Zimbabwe and Mr Terries Ndove fromSouth Africa were awarded John Allwright Fellowships by ACIAR to undertakepostgraduate studies at the University of Queensland.

The project has undertaken a wide range of legume-based research in the threecountries. The outcomes of that research have led to new insights into legume adapta-tion and productivity, soil N/soil organic-matter dynamics and the impact of legumeuse on cereal and animal production. In some instances, particularly in Zimbabwe, theproject has been able to achieve at least the first stages of the adoption of legume tech-nologies by farmers. The project has also enabled formal and informal training ofresearch and extension professionals to be undertaken, and the training here has sub-sequently resulted in further training and career opportunities for several of the Africanproject team members. Importantly, the project activities have exposed all teammembers to a much wider range of biophysical and social issues and environments,developed better understanding of the farming systems, and fostered what we hope willbe long-standing professional relationships.

Many of the papers in this publication were sourced from the final projectmeeting and review held at the Magoebaskloof Hotel in Limpopo Province, SouthAfrica on 7–9 October 2002. The editors refereed the papers and Mr Dick Jones isacknowledged for his substantial contribution to the reviewing process. Drs MervProbert, Michael Robertson and Bob McCown of CSIRO Sustainable Ecosystems



10

are also acknowledged for reviewing various manuscripts. The paper by Pengellyet al. was reprinted with permission from the journal Tropical Grasslands, Volume37, No. 4, pages 207–216.

The papers within this publication are presented under the following themes: TheEnvironment; Evaluation of Legume Technologies; Farming Systems Research; andParticipative Methodology and Adoption of Technologies. We hope that this publica-tion will prove to be an excellent resource for many other research and developmentinitiatives in progress in the Limpopo Province and for the wider R&D community ofthe tropics and subtropics.

Dr Bruce C. Pengelly (Project Leader)Dr Anthony M. Whitbread (Principal Investigator)Brisbane, AustraliaApril 2004



The Environment

11



12

The Farming Systems of the Zimbabwean District of Wedza and the Role and Adoption of Forage

Legumes in Small-scale Crop–Livestock Enterprises

B. Chigariro*

Abstract

An ACIAR project entitled “Tropical forage and ley legume technology for sustainable grazing and croppingsystems in southern Africa” (Project no. AS2/96/149) conducted a farmer participatory research project withcommunal (Dendenyore ward) and resettlement (Zana ward) farmers in the District of Wedza in Zimbabwe from1999 until 2002. The main aim of this project was to test and introduce a suite of legume technologies into themixed crop–livestock systems that predominate in these regions. A range of constraints that include poor soilfertility, lack of dry season feed for livestock and variable rainfall, seriously limit production and food securityin Wedza. Diversifying the range of crops, selling cash crops such as paprika and replacing weedy fallows withimproved lablab or mucuna fallows, were all methods employed by farmers to improve production. Livestockfarmers formed cooperative organisations to sell their products and build capacity in livestock production. Manyfarmers were able to substitute bought-in feed concentrates with home grown legume-based feed sources. Thispaper sets the scene for several subsequent papers that describe in detail the participative action research programbased around the theme of using well-adapted legumes to improve the farming system.

Zimbabwe has a total land area of about 39,000 km

2

.Five zones or ‘natural regions’ are recognised (Vin-cent and Thomas 1957). They are based largely onthe mean annual rainfall, together with other physicalparameters such as soil type (Table 1).

The country has a tropical continental climate in which rainfall is usually restricted to the period between November and March. The winter months between May and August are cool, because of the countries elevation (much of it above 1200 m alti-tude) and four distinctive seasons can be recognised:• summer – rainy season (November to mid-March)

• autumn – post rainy season (mid-March to mid-May)

• winter – cool dry season (mid-May to mid-August)• spring – hot dry season (mid-August to

November).

Wedza District

Zimbabwe is comprised of nine provinces with thedistrict of Wedza located in the Mashonaland EastProvince, about 120 km southeast of Harare. The dis-trict had a population of about 82,000 in 2001 andhad a range of smallholder and commercial-scalefarming operations, of which the communal farmingsystems and the large-scale commercial operationscovered the largest land areas in 2001 (Table 2).

* Agricultural Technical and Extension Services, PO Box30, Wedza, Zimbabwe.



13

The ACIAR project operated in two wards inWedza District — Dendenyore and Zana. Descrip-tions of the communal and resettlement systems canbe found in Maasdorp et al. (2004). Dendenyorecovers an area of 14,028 ha and is part of the com-munal farming sector. In 2001, Dendenyore hadabout 3000 households and a population of about15,000. Zana covers an area of 5187 ha and is withinthe resettlement sector. In 2001, it had 315 house-holds and a population of about 1900. The two wardsare located in natural region IIb between 18°15' and19°15'S and 31°15' and 31°20'E, at an altitude ofabout 1400 masl. The temperature range in summer isusually 16°C to 30°C; in winter there is a moderatefrost incidence.

The soils in the Wedza district are largely derivedfrom granite, with categories on a typical catena rec-ognised on the basis of soil moisture-holdingcapacity, waterlogging, weed burden and soil fer-tility. All these parameters increase down slope fromthe dry topland granitic sands of the upland ridgesand valley slopes to the hydromorphic vlei-marginand valley bottom vlei soils. Size and texture of soilparticles range from fine/medium grained sands overmedium to coarse-grained sandy loams at the toplandsites to medium to coarse-grained sandy loams in thevleis and vlei margins (J. Dimes, pers. comm.).Pockets of dolerite intrusions present in Dendenyore

have led to the formation of red clay soils. The soilsfound in the topland, upland ridges and valley slopesare commonly used for crop production and are gen-erally highly acidic, with the pH (CaCl

2

) below 4.4and very low in the status of base cations K

+

, Ca

++

and Mg

++

(Jiri et al. 2004), reflecting very low soilorganic matter (C

T

< 0.4%) and an associated lowcation-exchange capacity. The combination of lowwaterholding capacity of the soils and the relativelyimpermeable nature of the underlying granite,renders such soils, even on the uplands, susceptible tohigh water tables for short periods during many rainyseasons (Thompson and Purves 1978).

The Smallholder Farming Systems in Wedza

In both Zana and Dendenyore, intensive mixed crop–livestock farming is practised. Maize is a major crop,while there are minor crops of sunflower, soybeans,groundnuts, cowpea millets, sorghum and sweetpotatoes. The Dendenyore farmers have a subsist-ence agricultural focus and sell any surplus locally.The resettlement farmers of Zana, on the other hand,have a larger arable area available and employ thatextra land in cash cropping. The two most importantcash crops are tobacco and paprika.

Table 1. Mean annual rainfall for each of the five Natural Regions recognised in Zimbabwe. Source:Vincent and Thomas (1957).

Natural region Mean annual rainfall range (mm)

Region IRegion IIRegion IIIRegion IVRegion V

Specialised and diversified farming regionIntensive farming regionSemi-intensive farming regionSemi-extensive farming regionExtensive farming region

> 1000650–1000650–800450–650

<650

Note. Region II is divided into two sub-regions. Sub-region IIa receives an average of 18 rainy pentads per year, and normally experiences reliable conditions. It rarely experiences dry spells in summer. Sub-region IIb receives an average of 16–18 rainy pentads per season and is subject to more severe dry spells during the rainy season, or to occurrences of relatively short rainy seasons (Vincent and Thomas 1957).

Table 2. Farming sectors, number of farmers and properties and the total area of eachsector in Wedza District Zimbabwe in 2001.

Sector No. of farmers No. of properties Total area (ha)

Large-scale commercialSmall-scale commercialResettlement areaCommunal area

76475758

22,036

76486NANA

8,056145,900

NA108,400

NA = data not available.



14

Arable farming by resettlement farmers in Zanahas been undertaken since 1986. Each farmer has atotal arable area of 5 ha, of which, after 2001, 0.4 hato 0.8 ha was typically being used to produce fodderfor their dairy production, while crop production wasbeing carried out on about 3 ha. Dendenyore has beenunder cultivation since the 1930s, and the typicalward holding of arable land ranges from 0.8 ha to 5ha. Grazing in both farming sectors is communal.

The farming system of Zana does have thecapacity to employ fallows and there is some croprotation carried out, with a common crop sequencebeing maize–tobacco–maize or maize–paprika–maize. Fallows are seldom employed in Dendenyorewhere the smaller arable land parcels per farmernecessitate cropping on all land every summer. Inmore recent years, and since the inception of legumeresearch with farmers in the district, the rotationsbeing employed by dairy and beef farmers in Zanaand Dendenyore are being modified to include aforage legume phase of lablab or mucuna, with themost prevalent rotations being lablab or mucuna fol-lowed by maize, maize–sunflower, tobacco orpaprika.

Livestock Production in Wedza

Livestock and crop production in Zana and Dende-nyore are integrated enterprises, with livestock beingimportant sources of traction and manure, and cropresidues and fodder crops providing important feedfor both beef and dairy production. Cattle fattening inthe Wedza district dates back to the 1950s, with theformation of the Wedza Feeders Association. Theassociation has over >200 members, from almostevery one of the 14 wards of the district. Dendenyoreward has about 2400 farmers who own cattle, ofwhich 65 are pen fatteners and members of theWedza Feeders Association and another 54 areaspiring dairy farmers who are members of theWedza Dairy Association. Dairy farmers in Zanahave been producing milk since 1992.

All dairy cows are milked manually, and usuallyonly once per day, first thing in the morning fol-lowing separation of the cow and calf overnight. Asmall number of farmers who have purebred or cross-bred dairy cattle milk twice per day. The majority ofestablished dairy farmers milk crossbred cows.

Marketing of livestock products is carried outwithin the ward, with beef farmers selling theiranimals through local butcheries and abattoirs. Milk

from dairy enterprises is either marketed throughthe Wedza Dairy Association or sold locally. Directselling in 2001 was realising prices of about $75/L

1

compared with about $50/L when sold through theWedza milk centre which is managed by the Agri-cultural Rural Development Authority–DairyDevelopment Programme (ARDA–DDP) in collab-oration with Agricultural Research and Extension(AREX).

Resources for Extension

An important part of the Wedza district farmingsystems is the presence and role of the agriculturalextension team that operates at the ward level. Thedistrict extension arm is within the Zimbabwe Min-istry of Lands, Agriculture and Rural Resettlementand has a staff of 45, with a ratio of extension officerto farmers of about 1 to 1200 in the communal areasand 1 to 400 in the resettlement areas.

The extension personnel play a key role in the dis-semination of technologies to farmers. Their role is tofacilitate agricultural activities and train farmers onall aspects of agriculture. Several extension strategiesare employed to aid technology diffusion, with the‘master farmer’ training program being amongst themost widely used. By 2002, 2059 farmers in the dis-trict had qualified as master farmer at ordinary andadvanced level, with 125 farmers gaining these qual-ifications during the project period.

Advances in Feed Supply

The natural rangeland grazing is the major source ofruminant livestock feed in the smallholder and com-mercial sectors of farming in Zimbabwe. However,this source is particularly degraded in the smallholdersector because of a history of overgrazing, and it isunable to supply sufficient feed quantity and qualitythroughout the year. Crude protein levels in the dryseason can fall to below 5%. Commercial feed supple-ments are used by both dairy and beef-fattening enter-prises to partially overcome this protein shortage, butthey are expensive (Murungweni et al. 2004).

1 These are Zimbabwe dollars (ZWD). The exchange ratein late 2002 was USD1 = ZWD55. Since then, hyper-inflation has reduced the value of the ZWD andsubstantially increased the cost of inputs.



15

Both the beef-fattening and dairy-production farmergroups have adopted fodder bank technologies duringthe period 1999 to 2002, using lablab and mucuna.Hay is produced from both legumes at the end of thegrowing season. Since the start of the ACIAR project,dairy farmers in Zana resettlement area and beef pro-ducers in Dendenyore are adopting the practice ofgrowing legumes and bana/napier grass as foddercrops in their crop rotations in an attempt to reducefeed constraints. Over 70 farmers are now using lablaband mucuna as a feed source, and there has been arapid increase in the number of farmers using plantedcut-and-carry bana/napier grass which is fertilisedwith manure (Table 3). The use of this legume tech-nology has spread to four resettlement villages neigh-bouring the participating community in Zana, to theSengezi resettlement area, about 40 km from Zana,and to other nearby communal areas. Animal produc-tion has increased considerably, with milk yieldsalmost doubling (Table 4). The farmers in the dairysector have indicated that there has been a markedincrease in the butterfat content of the milk. Dende-nyore farmers have reduced their dependence on com-mercial feeds for pen-fattening by about 25%.

Conclusion

The Wedza communal and resettlement farmingsystems are complex. They produce maize and othercrops for household consumption and, in the case ofthe Zana resettlement area, paprika and tobacco ascash crops. Many farmers also have commerciallyfocused dairy and beef enterprises. There has beenadoption by a large number of farmers within thefarming communities originally participating in theproject. The higher production and greater incomebeing obtained by these farmers, and the enthusiasmand skills of the local extension staff, have resulted inthe use of legumes being adopted in other Wedzawards. In all of these communities, the use oflegumes in the farming system has enabled farmers toincrease their incomes.

There remains a large number of issues that requirefurther investigation. In the dairy enterprises forinstance, the focus has been on feeding lactating cows,but measures to enable farmers to monitor the growthrate of the calves have not been implemented. Simi-larly, there have been no targeted measures of thechanges in conception rate resulting from legume tech-

Table 3. Changes in the number of farmers employing various feed sources in dairy and beef productionenterprises between 1999 and 2002 in Zana and Dendenyore wards, Wedza district, Zimbabwe.

Feed type Zana (dairy producers) Dendenyore (beef producers)

1998–99 1999–2000

2000–01 2001–02 1998–99 1999–2000

2000–2001

2001–2002

LablabMucunaBana/napierUrea/stoverNative grass haySoybean/groundnut hayMaize stoverPaprika calyx

Nil20022

136

468656

2317

11131248

122823

23211249

123823

nil321322–

45234

114–

1331380

132613–

545438–

3626

153–

Table 4.

Average on-farm milk production (l/cow/day) trends between 1999 and 2002 at Zana resettlement areain Wedza District, Zimbabwe, as a result of incorporation of legume hay into the diet of cows.

Season and lactation stage 1999–2000 2000–01 2001–02

b

Summer productionBirth to 7 months7–8 months

4–61–3

6–17

a

2–85–104–6

Winter productionBirth to 7 months7–8 months

3 –41 – 2

5 –72–6

5 – 72–6

a

Purebred cow;

b

drought year.



16

nology introductions or the role of legumes inincreasing production in small stock such as goats. Inthe cropping systems, screening of new legumes hasshown lablab and mucuna to be well-adapted, but theirimpact and options in rotations with crops such aspaprika have not been documented, and even the impacton maize has not been widely demonstrated on a widerange of farmers’ fields. These and many other issuesstill require further research on both the communal andresettlement areas of Wedza, and in adjacent districts.

References

Jiri, O., Maasdorp, B.V., Temba, E. and Whitbrerad, A.M.2004. The role of dual purpose and forage legumes toimprove soil fertility and provide high quality forage:options in a maize-based system in Zimbabwe. Theseproceedings.

Maasdorp, B.V., Jiri, O. and Temba, E. 2004. Contrastingadoption, management, productivity and utilisation ofmucuna in two different smallholder-farming systems inZimbabwe. These proceedings.

Murungweni, C., Mabuku, O. and Manyawu, G.J. 2004.Mucuna, lablab and paprika calyx as substitutes forcommercial protein sources used in dairy and pen-fattening diets by smallholder farmers of Zimbabwe.These proceedings.

Thompson, J.G and Purves, W.D. 1978. A guide to the soilsof Rhodesia. Causeway, Salisbury, Rhodesia,Information Services, Department of Research andSpecialist Services, Ministry of Agriculture, RhodesiaAgricultural Journal, Technical Handbook No. 3, 31p.

Vincent, V and Thomas, R.G. 1957. An agricultural surveyof Southern Rhodesia. Part I: Agro-ecological Survey.Ministry of Agriculture, Government of the Federation ofRhodesia and Nyasaland, Salisbury, Southern Rhodesia.



17

Description of the Biophysical Environment of Three Maize-producing Areas in the Limpopo

Province of the Republic of South Africa and the Validation of APSIM to Simulate Maize Production

A.M. Whitbread* and K.K. Ayisi

†

Abstract

The focus of this paper is to provide background biophysical information and test the performance of theAPSIM (Agricultural Production Systems sIMulator) maize model on several data sets collected at the fieldsites used in the Limpopo Province of South Africa during the ACIAR project AS2/96/149 described in theseproceedings. The project targeted one community of smallholder farmers at Dan in the Mopani district ofLimpopo, with the aim of improving their dryland cropping systems. A site called Syferkuil, located on theUniversity of the North’s experimental farm, and a nearby site called Dalmada, on a farmer’s field, were alsothe location of several maize-intercropping/relay planting field trials. Using the data of the sole-maize controltreatments from these trials, APSIM-Maize V3.2 predicted the biomass accumulation with a high degree ofprecision (r

2

= 0.82). Factors such as grazing damage, severe frost and striga infection that reduced theobserved grain yield resulted in general over-prediction of simulated grain yield. Simulating maize productionwithout N inputs under dryland conditions at Dan, using long-term weather records (1975–2002), indicatespoor maize grain yields (<1000 kg/ha) and several crop failures. The application of 30 or 60 kg/ha N fertiliserincreased maize yields to 1700 and 2600 kg/ha in 50% of seasons.

The Limpopo Province is in the northern part ofSouth Africa, with an area of 12.3 million hectaresand a population of 4.9 million in 1996 (Anon 2003).Although maize production accounts for only 4% ofthe province’s gross agricultural income, maize isconsidered to be Limpopo’s most important drylandcrop in terms of the extent of production, the area uti-lised and the number of farmers involved (~519 000farmers). Maize is the staple food of most of the ruralpopulation and the crop residues are a feed resource

for animals. Despite the low and unreliably distrib-uted rainfall in many parts of the province, maize willbe an important crop for food security in the foresee-able future.

Most of the smallholder farming sector inLimpopo is located on infertile degraded soils, wherenutrient deficiencies, predominantly of N (nitrogen)and P (phosphorus), limit crop production. As thecash reserves of these mainly subsistence farmers arelimited, little or no inorganic fertiliser is applied tothe maize crops and grain yields are commonly<500 kg/ha. The traditional farming practice of inter-cropping maize with grain legumes such as ground-nuts (

Arachis

spp.) may help reduce the risk of cropfailure of one of the species and add some N to thesystem through biological N fixation. Papers in these

* CSIRO Sustainable Ecosystems, 306 Carmody Road, StLucia, Queensland 4067, Australia.

† School of Agricultural and Environmental Sciences, Plant Production, University of the North, Private Bag X1106, Sovenga, 0727 Republic of South Africa.



18

proceedings by Ayisi and Mpangane (2004),Maluleke et al. (2004) and Mpangane et al. (2004)describe various intercropping and relay-plantingexperiments using lablab (

Lablab purpureus)

andcowpea

(Vigna unguiculata).

The impact of the inter-crops on maize production and their potential for Nfixation were investigated with the aim of providingfarmers with options for cash cropping, forage pro-duction and improvements in soil fertility.

The APSIM (Agricultural Production SystemssIMulator) software system provides a flexible struc-ture for the simulation of climatic and soil manage-ment effects on the growth of crops and changes inthe soil resource (Keating et al. 2003). Use of APSIMto investigate maize production systems in climati-cally risky environments has been extensively testedin Kenya and Zimbabwe (Keating et al. 1999; Sha-mudzarira and Robertson 2002).

The focus of this paper is to describe the biophysicalenvironment and test the performance of the APSIMmaize model on several field data sets from the sitesused in the Limpopo Province during the ACIARproject AS2/96/149 described in these proceedings.

Materials and Methods

A range of experiments was conducted at the threesites described in this paper to investigate variousaspects of maize–legume technologies. Most of theseexperiments included a sole maize control treatmentthat has been used to validate the APSIM model forthe prediction of maize growth and grain yield. Thecontrol treatments are sourced from experimentsdescribed in Malueke et al. (2004) and Ayisi andMpangane (2004), and unpublished data fromMishiyi (2003).

The experimental site at Syferkuil (23

∞

85'S29

∞

67'E, 1250 masl) was on the University of theNorth’s field station. The experimental site atDalmada (23

∞

87'S 29

∞

53'E, 1334 masl) was 15 kmfrom Polokwane (Pietersburg) and was rented from alocal farmer. Irrigation was available at both thesesites, and the experiments were entirely researchermanaged. The on-farm site at Dan (23

∞

90'S 30

∞

27'E,650 masl) was 13 km east of Tzaneen town in theMopani district, on a 200 ha cropping area farmed byabout 300 farmers. Each farmer had access to 0.5–1.0 ha of arable land under a ‘permission to occupy’arrangement. Full details of this site and the commu-nity can be found in Ndove et al. (2004). The exper-iments conducted at Dan were also essentially

researcher managed, although the farmers activelyassisted in many field operations.

Field experiments

Syferkuil and Dalmada.

All experimental areaswere ploughed, disced and harrowed 1–2 weeks priorto planting. At planting, a basal P application of30 kg/ha as single superphosphate and a K applica-tion of 30kg/ha as potassium chloride was applied.Maize (SNK2147) was planted by hand into 90 cmrows at 15 kg/ha in the first week of December in2001 and 2002, and thinned to 3 plants/m

2

afteremergence. N was applied as urea at 15 kg/ha atplanting and at 30 days after planting (DAP). Supple-mentary irrigation was applied at both sites in 15 mmapplications at planting, and at 30 and 45 DAP. The2002–03 experiments at Syferkuil received 120 mmof irrigation in total, applied in eight 12–15 mmapplications at 10 day intervals from planting. The2002–03 experiments at Dalmada received 10–12 mm of irrigation at 40, 55, and 60 DAP. All treat-ments were replicated three times. There were six5 m rows of maize. The plots were located in a dif-ferent part of the field in each season.

Dry matter samples were taken at 54, 67 and 88DAP at Syferkuil, and at 50, 71 and 95 DAP for allcrops at Dalmada. At each of these samplings, threemaize plants from each plot were cut at ground leveland dried. At final harvest, maize cobs were har-vested from the central four rows, with 1 m borders ateach end of the plots. Stover yield was determined bytaking 10 plants randomly from the harvested area.

Dan.

At Dan during the 2000–01 season, an exper-iment was conducted on a farmer’s field to determinethe effects on maize production of applications of 0,30 and 60 kg/ha of N. The experiment was arrangedas a randomised complete block design, with threereplicates and plot sizes of 5

¥

8 m. No data were asrecorded, because a herd of animals demolished thetrial before harvest. After the construction of a sturdyfence, this experiment was continued in the 2001–02season.

In both seasons, land preparation and basal fertili-sation was as described for the Syferkuil andDalmada sites. Due to the low pH (Table 1) an addi-tion of 500 kg/ha of dolomitic lime was also appliedduring the land preparation activities. Maize(SNK2147) was planted as described above on1 November 2001. In the 30 and 60 kg/ha N treat-ments, half the fertiliser was delivered as urea atplanting and the other half at 30 DAP. No supple-



19

mentary irrigation was available. The plots were har-vested on 6 April 2002 using the same procedures asat Syferkuil and Dalmada.

Biophysical characterisation of the sites

Soil characterisation was undertaken for the soilprofiles at Dan, Syferkuil and Dalmada. Measure-ments of drained upper limit (DUL), crop lowerlimit (CLL) and bulk density, and the calculation ofplant available water capacity (PAWC), at the threefield sites were undertaken using the methodsdescribed in Dalgleish and Foale (1998) (Table 2).This soils information, along with the chemicalanalyses for the specific experiments (Table 5),were used to parameterise the SOILWAT2 andSOILN2 modules that determine the dynamics ofwater, carbon and nitrogen within APSIM. The ini-tialisation information (mineral N, starting soilwater, residues and roots) is specific to each exper-iment and is described below.

Weather information was obtained through theAgricultural Research Council Institute for Soil,Climate and Water, the South African Bureau ofMeteorology and via International Rainman V4.1(Clewett et al. 2002).

Simulation of maize growth

Syferkuil and Dalmada.

All simulations werestarted and initialised at day 182 in the same year thatexperiments were sown, using the data in Tables A2and A3. Soil water and soil mineral N at sowing weretherefore determined by APSIM. It was assumed thatmost surface plant residues from the previous seasonwere removed, and the simulations were initialised at400 kg/ha. The timing of tillage events, sowing, Nfertilisation, and irrigation were as for the field exper-iments described above. The SC401 short-seasonvariety available in APSIM was found to best repre-sent the growth of SNK2147. Harvesting of the mod-elled data took place when the simulated crop

Table 1. Soil chemical analysis for the soil profiles at Dan, Syferkuil and Dalmada.

Depth(cm)

pH Pa

mg/kgCab Mgb Kb Nab Mn

mg/kgZnc

mg/kgcmol(++++)/kg

Dan 0–1515–3030–6060–90

4.85.55.65.8

6222

2.423.563.703.12

1.271.732.022.02

0.090.050.050.04

0.030.040.080.10

174.141.830.417.1

2.52.2

13.115.9

Syferkuil 0–1515–30

6.67.0

3022

2.862.51

2.822.43

0.280.19

0.260.20

n.d.n.d.

n.d.n.d.

Dalmada 0–1515–3030–60

7.67.97.5

43278

4.084.494.86

4.184.515.02

1.020.950.73

0.020.040.14

n.d.n.d.n.d.

n.d.n.d.n.d.

n.d. = not determineda 1:7.5 extractant Bray 2b 1:10 extractant ammonium acetate 1 mol pH7c 1:4 extractant 0.1 mol HCl

Table 2. Soil water characteristics and mineral N of the soil profiles at Dan, Syferkuiland Dalmada.

Depth

(cm)

DUL

(mm)

CLL (maize)(mm)

PAWC

(mm)

Mineral N

(kg/ha)

DanSyferkuil Syferkuil – deepDalmada

12090

12090

330137184158

20685

11883

124526675

28444810

CLL = crop lower limit; DUL = drained upper limit; PAWC = plant available water capacity.



20

reached physiological maturity. All simulationsassumed that P was not limiting. This is a reasonableassumption because basal applications of P fertiliserwere applied to all field experiments.

Dan.

Maize in the 2001–02 season was simulatedfor the treatments that received N fertiliser at 0, 30and 60 kg/ha. The soil mineral N measured prior tosowing (Table A1) was initialised in the model atday 300. Soil water was also initialised to the DULat this time. It was assumed that most surface plantresidues from the sorghum crop season had beengrazed, and surface residue was initialised at400 kg/ha. Fertiliser N was added as described forthe field experiment. Harvesting of the modelleddata took place when the simulated crop reachedphysiological maturity.

Long-term simulation of dryland maize at Dan

.Using the soil characterisation for Dan (Table A1)and weather data (Letaba Letsitele station 19935)that contained daily records from 1975 until April

2002, the production of sole maize was simulatedeach season. While the soil nitrate and soil waterwere initialised in the first year of the simulation as inTable A1, the model subsequently determined theseparameters. A soil tillage event took place on Julianday 304 each year. Maize (SC401) was sown on ayearly basis using a rainfall-based sowing rule. Inorder to trigger the sowing event within the sowingwindow from 15 November to 15 January, rainfall ofat least 20 mm over five days was required beforeplanting would take place. Maize was planted at 3plants/m into 90 cm rows. A grass weed (ashort-season annual grass) was also sown at25 plants/m at the same time as the maize, to mimicthe effect of weeds on the maize crop. The weedswere removed by tillage at 33 DAP. Separate simu-lations received 0, 30 or 60 kg/ha N fertiliser as ureain a split application at sowing and 33 DAP. Maizewas harvested when it reached physiological matu-rity. Grain weight is expressed at 12% moisture.

Table 3. Rainfall (mm) at Dan, 1998–2002, and the long-term average (1975–2001).

Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Year

1998–991999–002000–012001–02Average

2525

119

0200

10

210

166

21

4367528652

18092

133252

99

283272142202142

150248

1846

141

159554247

7117

202300

57n.a.106

16170

10n.a.40

3099

n.a.16

798

n.a.6

11141749

692601759

n.a. = data not available at this time.

Table 4. Rainfall (mm) at Pietersburg since 1998 and the long-term average (1904–1996).


1998–991999–002000–012001–02Averagea

05003

00003

8410

12

6640435541

1199067

21579

12475824689

100105

12102

87

3188

692173

404074

461

574172329

1120102011

124

345

477667377488493

a Long-term average calculated from 87 years of records sourced from Rainman International (Clewett et al. 2002).

Table 5.

Rainfall (mm) at Syferkuil, 1998–2002, and the long-term average (1984–2002).


1998–991999–002000–012001–022002–03Average

520007

000004

646067

563142651441

64106

94239

091

72856277

11376

77120

0136

7384

6238

72155872

544211334452

2127

3021

029

12151514

012

052

4007

353822336600308480

Source: University of the North experimental station records.



21

Results and Discussion

Biophysical environment

Dan

. Based on the 1975–2001 weather files fromthe nearby Letaba Letsitele 19935 weather station,the annual average rainfall at Dan is 759 mm. Therainfall pattern is strongly summer dominant, with86% of the rainfall received from October to the endof March (Table 3). The amount and distribution ofrainfall is critical, as agricultural production at Dan israin fed. Minimum temperatures of 6.5

∞

C are reachedin June and July, and maximums of 32∞C are reachedin December and January (data not shown).

Dalmada. Weather at Dalmada, which is approxi-mately 30 km from Pietersburg, is based on theweather records obtained from the Pietersburg sta-tion. At Pietersburg, there is an annual average rain-fall of 493 mm, received mainly from October toMarch (Table 4). Maximum temperatures of 28∞Coccur during January, with minimums of 4.4∞Cduring July (data not shown). Supplementary irriga-tion was available to the field experiments at this site.

Syferkuil. The minimum and maximum tempera-ture data obtained from the University of the North’sexperimental site at Syferkuil were similar to thosedescribed from the Pietersburg weather station. Frostoccurs rarely, but caused the death of all legumeexperiments in May 2002. While the long-term rain-fall averages were similar to those of Pietersburg, theamount of rainfall received was lower in theOctober–January period and higher in the February–April period (Table 5). The Syferkuil site is closer toa mountain range and its weather is obviously influ-enced by this.

Soils

Dan. The main soils used for crop production atDan range from coarse-grained sandy soils to sandyloams derived from granitic parent materials. Thedepth of these soils, which overlie clay, varies from60 cm to >150 cm. Much of the cropping land is sus-ceptible to waterlogging during heavy rains, andsubsoil drainage is poor because of the underlyingdense clay layer. The sandy nature of the soil resultsin high bulk density and PAWC is 124 mm to therooting depth of 120 cm (Table 2).

Available phosphorus concentrations are below6 mg/kg, which is severely limiting to plant growth(Table 1). These soils are acidic, with pH values

below 5.8. The basic cations Ca, Mg and K are in theadequate range for most grain crops. High concentra-tions of Mn in the surface soils at Dan indicate pos-sible toxicity under waterlogged conditions.

Syferkuil. The soil at Syferkuil is a sandy loam(77–81% sand in the 0–60 cm depth). Soil depthvaries from 90 to 120 cm, resulting in a PAWC thatvaries from 52 to 66 mm for maize (Table 2). Thesoils at the Syferkuil site are described as part of theHutton series (Soil Classification Working group,1991) with an orthic A horizon and red apedal Bhorizon overlying unspecified material (BoyeMashotole, pers. comm.). Apedal is defined as mate-rials that are well aggregated, but well-formed pedscannot be detected macroscopically. Regular super-phosphate fertiliser applications have resulted in ade-quate to high available P (Table 1). The basic cationsCa, Mg and K are in the adequate range for mostgrain crops.

Dalmada. The soils at the Dalmada site aredescribed as part of the Bainsvlei Form (Soil Classi-fication Working group, 1991) with an orthic Ahorizon and red apedal B horizon overlying a hardplinthic B horizon at a depth of 80–90 cm. This con-sists of an indurated zone of accumulation of iron andmanganese, which could not be cut with a spade evenwhen wet and which limits rooting to this depth. ThePAWC for maize is 75 mm (Table 1). Impededdrainage through this layer could result in waterlog-ging, although this did not occur during our experi-mental program. Adequate to high available P wasalso measured at this site and the basic cations Ca,Mg and K are in the adequate range for most graincrops (Table 2). This site had been commerciallyfarmed and probably received fertiliser applicationsin the past.

Specification of the APSIM SOILWAT2, SOILN2 and RESIDUE2 modules

The inputs required to specify the SOILWAT2 andSOILN2 modules are given in Appendix 1, TablesA1, A2 and A3. Other parameters that relate to evap-oration and run-off are found in Table A4, andparameters that relate to the initialisation of residueare in Table A5. These parameters were largelybased on the measured values described above. Otherparameters that are not measurable, but required bythe model, were estimated by consulting a range ofliterature.



22

Sole maize field experiments

Dalmada and Syferkuil. Biomass accumulation ispresented for the Dalmada experiment in 2002–03 andshows a slow accumulation until 50 DAP, followed bya rapid growth phase until maturity (Figure 1).

The simulated and observed data for all biomassmeasurements made at intervals during the growth ofmaize in the 2001–02 and 2002–03 seasons at bothsites were combined (Figure 2). The simulatedbiomass accumulation was generally underestimatedlater in the season (i.e. when biomass values werelargest), but the overall r2 = 0.83 is high.

Grain yield was similar at the two sites in 2001–02and increased considerably at Syferkuil in 2002–03in response to more irrigation (Table 6). A severefrost late in the season at Syferkuil in 2001–02reduced the harvested yield and may account for theoverestimation of the simulated grain yield. AtDalmada in the 2002–03 season, uncontrolled wildanimals damaged the plots by grazing and reducedharvested yield. This explains much of the variationbetween the observed and simulated data.

Dan. After maize was planted across all treatmentsin 2001–02, 252 mm of rain fell during Novemberwith a further 202 mm during December, resulting ingood vegetative maize growth. Very hot and dry con-ditions corresponded with floral initiation around8 January 2002, with temperatures above 35∞C aroundflowering time in late January. With the exception of

42 mm of rain in the last two days of January, droughtconditions prevailed throughout January and Feb-ruary. These hot and dry conditions resulted in a nearlycomplete failure in grain filling and very low grainyield in all treatments (Table 7). According to the sim-ulations, N stress affected all treatments from 10 Jan-uary 2002 until maturity and water stress was limitingmaize yield from February 5 until maturity.

There was no significant difference in grain yield orstover + husk yield between treatments. The most grainwas produced in the treatment that received 60 kg N/ha, while no grain was produced in the treatment thatreceived no N fertiliser (Table 7). The simulations pre-dicted the biomass production well but overestimatedthe grain produced on the fertilised N treatments. Theeffects of Striga hermonthica (witchweed), theft ofgreen cobs (mealies) and losses due to late harvestingwere all factors that may have contributed to this dis-crepancy in the treatments that received N fertiliser.

Predicted

Observedbiomass

00

2000

4000

6000

8000

10000

20 40 60 80

Days after planting

Bio

mas

s yi

eld

(kg

/ha)

100 120 140

1: 1

00

2000

4000

6000

8000

10000

12000

2000 4000 6000 8000

Observed

Pred

icte

d

10000 12000

y = 0.7x + 624

r2 = 0.83

Figure 1. The measured values of biomass at 54, 68,82 and 128 days after planting and thesimulated biomass accumulation of maize atDalmada in 2002–03

Figure 2. The relationship between the observed andpredicted data for all biomass measurements(kg/ha) made during experiments at theDalmada and Syferkuil sites.

Table 6. The observed and simulated maize grainyield (kg/ha) of the sole maize treatments.

Year Site Observed Simulated

2001–02 Syferkuil 1238 1627

Dalmada 1185 1200

2002–03 Syferkuil 5181 3843

Dalmada 1674 3733



23

Long-term simulated performance of maize production at Dan

The simulation of maize production without Ninputs under dryland conditions at Dan, using theweather data for the period 1975–2002, indicatesextremely poor maize grain yields (<1000 kg/ha) inmost seasons, and two crop failures (Figure 3).

With the application of 30 kg/ha of N, the poorestseason yielded 587 kg/ha and 50% of the seasonsyielded >1700 kg/ha. The application of 60 kg/ha ofN resulted in maize yielding >2600 kg/ha in 50% ofseasons (Figure 3).

Nitrogen use efficiency (NUE) was in the range16–59 kg grain/kg N for applications of 30 kg/ha,and 18–44 kg grain/kg N for applications of 60 kg/ha

(data not presented). These generally high NUEfigures result from the model’s assumption that therewas no limitation caused by deficiencies of othernutrients, or by the other factors described above thataffected the field trials.

Conclusion

The APSIM maize model was able to simulatebiomass production with a high degree of precision.The observed grain yields were often affected byfactors that are unmodellable (e.g. theft, damage bygame) and factors that are now becoming modellable(e.g. parasitic weeds). A new version of the maizemodel that is responsive to P is being tested in the

0N

30N

60N

00

0.20

0.40

0.60

0.80

1.00

1000 2000 3000 4000

Maize grain yield (kg/ha)

Cu

mu

lati

ve p

rob

abili

ty

Table 7. The observed and simulated grain and biomass (kg/ha) at finalharvest at Dan in 2001–02.

Grain Biomass

Observed Predicted Observed Predicted

M0 153 0 2735 2345

M30 320 888 3763 4547

M60 591 1300 4621 5403

Significance n.s. n.s.

Figure 3. Cumulative distribution functions for maize productionwith 0, 30 or 60 kg/ha N applied (1975–2002) at Dan.



24

ACIAR Risk Management Project (LRW2/2001/028) in Zimbabwe, and will be applicable to thelow-P soils of South Africa.

The soil characterisations presented in this paperare common to large areas of arable land in theLimpopo Province. Utilising the characterisationinformation presented in this paper and modifying itto suit other sites will help expedite other modelapplications in the region.

The production of maize at Dan in the absence of Ninput resulted in a high risk of crop failure and a verylow potential yield target. This matches the experi-ences of farmers from the community, presented byNdove et al. (2004). The application of moderateamounts of fertiliser N (30 and 60 kg/ha) resulted inlower risk and higher yield potential. Although thereason for not applying fertiliser is said to a lack ofavailable cash, the costs of ploughing large and oftenunused tracts of land are readily paid by the farmers.Efforts by the agricultural department extensionofficers to help farmers reprioritise their crop inputscould be rewarded by better food security.

References

Anon 2003. Abstract of agricultural statistics. Directorate:Agricultural Statistics of the National Department ofAgriculture. Agricultural Information Services, PrivateBag X144, Pretoria 0001, RSA.

Ayisi, K. and Mpangane P.N.Z. 2004. Growth andsymbiotic activities of cowpea cultivars in sole and binarycultures with maize. These proceedings.

Clewett, J.F., George, D.A., Ooi, S.H., Owens, D.T. andPartridge, I.J. 2002. Rainman International (version 4.1):An integrated software package of rainfall informationfor better management. QI02023, Queensland,Department of Primary Industries.

Dalgliesh, N. and Foale, M. 1998. Soil matters: monitoringsoil water and nutrients in dryland farming. CSIROAustralia, 71.

Keating, B.A., Carberry, P.S., Hammer, G.L., Probert, M.E.,Robertson, M.J., Holzworth, D., Huth, N.I., Hargreaves,

J.N.G., Meinke, H., Hochman, Z., McLean, G.,Verburg,K., Snow, V., Dimes, J.P., Silburn, M., Wang, E., Brown,S., Bristow, K.L., Asseng, S., Chapman, S., McCown,R.L., Freebairn, D.M., Smith, C.J. 2003. An overview ofAPSIM, a model designed for farming systemssimulation. European Journal of Agronomy 18, 267–288.

Keating, B.A., Waddington, S., Grace, P., Rohrbach, D.,Dimes, J., Shamudzarira, Z., Carberry, P. and Robertson,M. 2000. Exploring farmer options for maize productionstrategies via scenario analyses using the APSIM model— an example of the approach. SOILFERTNET/CIMMYT, Risk Management Working Paper Series, No.2000/02.

Maluleke, M., Ayisi, K. and Pengelly, B.C. 2004. Lablabdensity and planting date effects on growth and grainyield in maize lablab intercrops. These proceedings.

Mishiyi, S.G. 2003. Nodulation dry matter accumulationand grain yielding of cowpea and lablab varieties undersole and intercropping systems with maize. Degree ofMasters in Agricultural Management, University of theNorth, Republic of South Africa.

Mpangane, P.N.Z., Ayisi, K.K., Mishiyi, M.G. andWhitbread, A. 2004. Grain yield of maize, grown in soleand binary cultures with cowpea and lablab in theLimpopo Province of South Africa. These proceedings.

Ndove, T.S., Whitbread, A.M., Clark, R.A., Pengelly, B.C.2004. Identifying the factors that contribute to thesuccessful adoption of improved farming practices in thesmallholder sector of Limpopo Province, South Africa.These proceedings.

Probert, M.E., Dimes, J.P., Keating, B.A., Dalal, R.C. andStrong, W.M. 1998. APSIM’s water and nitrogenmodules and simulation of the dynamics of water andnitrogen in fallow systems. Agricultural Systems, 56, 1–28.

Shamudzarira, Z. and Robertson, M.J. 2002. Simulating theresponse of maize to nitrogen fertiliser in semi-aridZimbabwe. Experimental Agriculture, 38, 79–96.

Soil Classification Working Group 1991. Soil classification:a taxonomic system for South Africa. Memoirs on theagricultural natural resources of South Africa, No. 15.ISBN 0-621-10784-0. Obtainable: Director, DirectorateAgricultural Information, Private Bag X144, Pretoria0001, Republic of South Africa.



25

Appendix 1

Specification of APSIM SOILWAT2 and SOILN2 modules

The parameter swcon determines the proportion ofwater above the DUL that will be drained each day.In a well-drained sandy soil that becomes saturated, itcould be expected that the soil water content willreturn to the DUL over 2–3 days. In Probert et al.(1998), swcon = 0.2 for the ‘Warra’ vertisol soil indi-cating much slower drainage than swcon = 0.7 for thesandy soils described here.

Finert describes the proportion of initial organiccarbon assumed to be inert. Assuming that all organicC measured at depth is essentially inert, this quantityis assumed to remain the same at all depths.

Fbiom describes the initial biom as a proportion ofnon-inert C. These values are based on Probert et al.1998 and other published data sets.

Table A1. Soil properties and initial values at the Dan site, by layer

Layer number 1 2 3 4 5

Layer depth (mm)Air_dry weight (mm/mm)ll15 (mm/mm)Dul (mm/mm)Sat (mm/mm)swcon Bulk density (g/cm3)Organic carbon (%)pHnh4 (mg/g)no3 (mg/g)FinertFbiom

1500.03

0.1200.2300.400

0.71.470.70

5.00.50

0.90.570.03

1500.03

0.1200.2300.400

0.71.460.60

5.60.50

0.70.660.02

3000.03

0.1800.2900.400

0.71.460.50

5.60.50

1.50.80

0.015

3000.03

0.2000.2900.400

0.71.460.40

5.80.40

1.00.99

0.010

3000.03

0.2000.2900.400

0.71.460.40

6.00.40

1.00.99

0.010

Table A2. Soil properties and initial values at the Syferkuil site, by layer.

Layer number 1 2 3 4 5

Layer depth (mm)Air–dry weight (mm/mm)ll15 (mm/mm)Dul (mm/mm)Sat (mm/mm)swcon Bulk density (g/cm3)Organic carbon (%)pHnh4 (mg/g)no3 (mg/g)FinertFbiom

1500.03

0.0540.1300.403

0.71.450.87

7.00.5011.00.460.03

1500.03

0.0720.1560.403

0.71.450.87

7.00.50

9.00.460.02

3000.03

0.1100.1570.403

0.71.450.70

6.90.50

5.00.57

0.015

3000.03

0.1100.1570.403

0.71.450.60

6.90.40

5.00.67

0.010

3000.03

0.1100.1570.403

0.71.450.50

6.90.40

3.00.80

0.010



26

Table A3. Soil properties and initial values at the Dalmada site, by layer.

Layer number 1 2 3 4

Layer depth (mm)Air_dry weight (mm/mm)ll15 (mm/mm)Dul (mm/mm)Sat (mm/mm)swcona

Bulk density (g/cm3)Organic carbon (%)pHnh4 (mg/g)no3 (mg/g)Finertb

Fbiomc

1500.03

0.0680.1820.432

0.71.370.87

6.90.503.000.460.03

1500.03

0.1000.1710.436

0.71.360.87

7.00.502.000.460.02

3000.03

0.1000.1750.428

0.71.380.70

6.90.501.000.57

0.015

3000.03

0.1000.1750.428

0.71.380.60

6.90.400.500.67

0.010

Table A4. Soil water parameters for the experimental sites at Dan, Syferkuil and Dalmada.

APSIM code Definition Values

u Stage 1 soil evaporation coefficient (mm) 3.0

cona Coefficient for stage 2 soil evaporation (mm day –0.5) 3.5

cn2 Run-off curve number 80a

cn_red Maximum reduction in cn2 due to presence of surface residues 20

cn_cv Percentage residue cover at which maximum reduction in cn2 occurs 80

salb Bare soil albedo 0.1

diffus_const Coefficient defining diffusivity 88

diff_slope Coefficient defining diffusivity 35.4a The curve number at Syferkuil and Dalmada was set as 72 to represent lower potential run-off from these very flat

sites than from the sloping and hard-setting Dan site.

Table A5. Soil nitrogen parameters for experimental sites at Dan, Syferkuil and Dalmada.

APSIM code Definition Values

soil_cn C:N ratio of soil 14.5

root_wt Initial root residues (kg/ha) 400

root_cnr C:N ratio of root residues 45

residue_wt Initial surface residues (kg/ha) 100

residue_cnr C:N ratio of surface residues 80

residue_type Type of surface residues, which determines the specific area and contact factor used by the residue module

maize

pot_decomp_rate Potential decomposition rate of surface residues under optimal conditions (day–1)

0.05



Evaluation of Forage Technologies

27



28

Tropical Forage Research for the Future — Better Use of Research Resources to Deliver

Adoption and Benefits to Farmers1

B.C. Pengelly,* A. Whitbread,* P.R. Mazaiwana† and N. Mukombe†

Abstract

Successful adoption of forage technology is frequently associated with a need to increase production andincome generation. Farmers might be expected to ‘demand’ new forages only when they can see a financialbenefit in the short to medium term. Opportunities for farmers to generate income from livestock productionare increasing dramatically as demand for animal products increases across Asia and Africa. Most of thisincreased demand will be met from mixed cropping–livestock enterprises, in which most tropical livestock arecurrently raised and production usually depends on low-quality crop residues. Forage research in the futurewill need to provide farmers with the means to meet the increased demand for livestock products. Thechallenge will be to develop research strategies that identify well-adapted forages that can improve livestockproduction and can be grown within the spatial and temporal constraints of complex and resource-limitedmixed cropping–livestock farming systems; in addition, it will be necessary to provide appropriate informationon the management and economic benefits of these forages. This paper presents two possible approaches.

In a participatory action research program on the use of forage legumes in cropping systems in Zimbabwe,the keys to successful forage adoption in rural communities are seen as: the emerging market for livestockproducts; a motivated and educated extension service working with a range of research specialists; andopportunities for beneficial synergies to be exploited from a mixed livestock–maize production system.Focused benchmarking of the communities identified farmers with sufficient resources and appropriatelivestock systems to benefit from improved forage.

In a rice-based Indonesian farming system, simulation of forage growth and livestock production before on-farm research begins is being used to examine the possible whole-of-farm impacts of using planted forages.Identifying the most likely spatial and temporal opportunities for growing forages and incorporating them intothe feed calendar can potentially avoid costly on-farm research on practices that have doubtful economicimpact; moreover, this approach enables a wide range of options to be compared rapidly.

To avoid the mistakes of the past, researchers need to provide supporting evidence that investment in newforages makes a difference not only to livestock production but also to household income. They will need tofocus on farming systems, such as mixed crop– livestock systems, where farmers have total control over theforage they produce and where adoption can be shown to be economically sustainable. Researchers also needto identify and target those farmers within rural communities with the resource capacity to invest in newfarming practices.

1 This paper was first printed in Tropical Grasslands,volume 37, pages 207–216. It is reprinted here with thepermission of the journal.

* CSIRO Sustainable Ecosystems, Queensland Biosciences Precinct, St Lucia, Queensland 4067, Australia.

† AGRITEX, Wedza, Zimbabwe.



29

Livestock production is an important component ofmany smallholder farming systems throughout thetropics. As well as providing food, such as fish, meat,milk and eggs, livestock are also critical in providingdraft power and manure for fertiliser and/or fuel;moreover, livestock often have wider socioeconomicroles within the community, such as providing finan-cial security. The feed sources for animals in small-holder farming systems are extremely variable butusually comprise a mixture of grazing on commu-nally owned grasslands, cut-and-carry forages fromoff-farm, and crop residues.

Without substantial additions to the diet in theform of mineral, protein and energy supplements,both reproductive rates and animal production arealmost universally poor within these systems.Despite this low level of animal production, dietarysupplements including sown forages have seldombeen used until recently because of farmers’ percep-tions that they were not needed, the lack of capital orplanting materials, or the absence of a financialincentive to invest labour and capital in forage pro-duction. For many years, the lack of incentive forforage investment resulted in poor adoption of almostall new forage technologies in smallholder systems intropical Africa and Asia. This is despite manydecades of pasture and forage research, and develop-ment in these regions. As noted by Squires et al.(1992) and Thomas and Sumberg (1995), the adop-tion of planted forages by farmers in sub-SaharanAfrica has, in the main, been limited. Before consid-ering any further commitment to new tropical foragescience in smallholder farming systems, researchersneed to address the following three key questions:• Given the history of forage adoption, why should

an increase occur now in the use of improvedforages in smallholder systems?

• If such an increase is to occur, what farmers aremost likely to adopt forage technology and in whatfarming systems?

• What have been the constraints to forage adoptionand utilisation by these farmers?This paper attempts to answer these questions in

the light of international and regional trends in live-stock production and consumption, socioeconomicconstraints on farmers, and the large body of infor-mation already available on forage species adapta-tion and forage management.

International Demand for Livestock Products

The past 10 years have seen remarkable changes indemand for livestock products across the tropicalworld, and the increased demand is predicted tobecome even more obvious in the future. It has beenestimated that milk consumption across the tropicswill increase by about 3.2% per year until 2020 (Del-gado et al. 1999). Similarly, beef and pork consump-tion is expected to double in developing countriesbetween 1993 and 2020. This increase in demand isalready having, and will continue to have, majorimpacts on household, farm and even regional econ-omies throughout the tropics. The mixed crop–live-stock farming systems of the tropics will be mostaffected by this increase in demand for livestockproducts. These systems already produce more thanhalf the meat and most of the world’s milk supply(Blackburn 1998; CAST 1999); more than 85% ofthe world’s cattle, sheep and goats are held in thesemixed systems in the tropics (Gardiner and Devendra1995).

In several regions of Africa, Asia and the Amer-icas, dairying has become an important and econom-ically attractive enterprise for smallholder farmers.For example, by 1996, more than 400 000 small-holder dairy farmers in Kenya produced about 70%of the country’s market milk (Reynolds et al. 1996).In Thailand, dairy cattle numbers increased fourfoldduring the 1990s to meet the increase in demand formilk; by 1999, there were >19 000 smallholder dairyfarms in existence (Hare et al. 1999). Similarincreases in demand for milk have been recorded inColombia, South America (Rivas and Holmann2000). With these rapid increases in livestock pro-duction comes demand for new sources of livestockfeed.

The increase in demand for livestock products isalso having major effects on regional economies andeven government policy. The increase in demand forbeef in Java, the most populous Indonesian island,has resulted in a rapid decline in local herd sizes onthe Indonesian islands of Lombok and Sulawesi.High beef prices have encouraged farmers to marketa significant proportion of the breeding herd forslaughter. Consequently, beef cattle numbers inSouth Sulawesi have declined from 1.23 million in1991 to only 0.84 million in 1997 (FAO 1999). Thedecline in herd size has had various effects. Anumber of provincial governments in Indonesia have



30

embarked on research and development programs toimprove reproduction and livestock production inBali cattle, with emphasis on developing a betterforage basis for the industry. Some provincial gov-ernments have also imposed restrictions on livestockexports from their islands.

The rapid increase in demand for livestock prod-ucts is providing opportunities for farmers to deriveincome from livestock production and improve theeconomic sustainability of their farming enterprises.However, it is also providing threats as farmers over-stretch their dependence on limited resources such ascommunally grazed grasslands, manure and theirbreeding herds. Improved planted forages have thecapacity to provide better quality feed for livestock.How then should the research community prioritisethe limited forage research budget to assist farmers totake full advantage of the changing marketplace?

Avenues available to improve livestock productionSmallholder livestock systems throughout the

tropics have traditionally been based on low-qualityroughage sources from communally owned grass-lands and/or crop residues (Ranjhan 1986; Moog1986; Dzowela 1993). With both of these sources,total forage available and forage quality are usuallylimiting for considerable periods of the year. Forexample, crude protein concentrations in grass andmaize stover can be as little as 3% and 6%, respec-tively (Ndlovu and Sibanda 1996; Makembe andNdlovu 1996). Consequently, peak milk yields ofindigenous animals in Zimbabwe can be as little as 3–5 kg/d (Mutukumira et al. 1996; Pedersen andMadsen 1998). Improving animal production willnecessitate additional forage and/or some method ofimproving diet quality at strategic times throughoutthe year.

As potential economic benefits from improvedanimal nutrition become apparent, farmers arebecoming far more interested in their feeding optionsand are making larger investments in animal feed. InThailand, the purchase of animal feed constitutesalmost 60% of a dairy farmer’s direct costs and stillproductivity is largely limited by feed supply (Hare etal. 1999). Farmers have a number of options forimproving diet quality. Feeding concentrates such asurea, phosphorus and molasses can greatly improveoverall diet quality and improve livestock produc-tion. Such technology has been widely adoptedthroughout much of the tropics where low-qualitycrop residues are in abundance and form a significantpart of the diet. In Zimbabwe, many livestock pro-

ducers already feed concentrates, while in many partsof Asia, urea–molasses blocks are commonly used.

Over 40 years of research has clearly demonstratedthat animal performance can also be improved byproviding better nutrition via higher quality grasses,such as fertilised cut-and-carry grasses, or by theinclusion of legumes in the diet (e.g. Humphreys1991). Despite the well-known benefits of improvedforages, adoption in smallholder systems, and indeedin extensive production systems in developed coun-tries, has frequently been poor. However, a numberof success stories do exist.

In formulating policy on where forage researchmight be directed, it is important to consider factorsthat might affect adoption and at least attempt tocharacterise the farmers most likely to adopt newforage technologies, and their farming systems.

Farming Systems and Control of Utilisation

Given their restricted resources, smallholder farmersare necessarily cautious about new investments; theywould probably invest in new forages only when theyhave control over the resultant utilisation. Conse-quently, investment in new forages for communallygrazed grasslands is unlikely until pastoral commu-nities develop strategies for grazing management(Squires et al. 1992). Rather, improved forages aremore likely to be planted in mixed cropping–live-stock systems where grazing control is usually prac-tised, at least for part of the year. In these farmingsystems, the challenge for farmers, extension special-ists and forage scientists is to identify suitable foragespecies/cultivars, and to identify the spatial and tem-poral opportunities to grow those forages.

Which Farmers Might Invest in Forage Technology?

The availability of well-adapted species and soundscientific evidence that improved forages improvelivestock production has been insufficient toencourage farmers to make significant investment inforages. Thomas and Sumberg (1995) attributedsome part of this reluctance to grow forages to pro-ducers being unfamiliar with the concept of investinglabour and capital in forages rather than staple crops.Farmers have not been accustomed to consideringforage production as a part of subsistence agriculture.



31

Horne and Stür (1997) argued that, in Asia, the lackof adoption of what are clearly well-adapted pastureplants was, in the main, the result of researchers notaddressing farmers’ requirements. They stressed theneed for farmers to be partners in forage research anddevelopment, so that researchers can understandbetter the complexities and forage priorities withinfarming systems.

Over the past decade, there have been two signifi-cant changes that affect these views. The first was thealready described development of smallholder live-stock enterprises that have income generation, ratherthan wealth, social status or food security as a corner-stone. The second has been the use of a participatoryprocess in agricultural research, particularly in agri-cultural research targeting smallholders in devel-oping countries. In fact, participatory action research(PAR) has become commonplace and almost a pre-requisite of research programs. Appropriatelyapplied PAR is, by definition, inclusive of thefarming community and enables all research partnersto be part of the process of developing project aims.Farmers can then see first hand the impact of variousexperimental treatments on the production of theirown crops and livestock.

Despite the increase in income-generating live-stock enterprises and the use of PAR, investment inforage technology would not be appropriate for allsmallholder livestock farmers. To adopt forage tech-nology, it might be expected that farmers would meetat least the majority of the following criteria:• They would already have a relatively stable food

source. Farmers who struggle to support theirfamilies’ needs for rice or maize or other staplefood would almost certainly see investment inforage as a high-risk strategy. If these subsistencefarmers had available capital for on-farminvestment, it would most likely be targeted atimproving food production through the purchaseof fertiliser, labour for weeding etc.

• They would already have expertise in livestockproduction and would be in the position torecognise the potential benefits of increasedanimal production.

• They would have capital available for theinvestment.

• They would have land available for growing theforage and could envisage a management systemthat could fit the forages into a cropping cycle.Farmers who perceive that they already haveinsufficient land for crop production would not be

in a position to consider using valuable land fornew forages, unless totally new scenarios of landuse became available.

• They would already have available to them anestablished or emerging market system forlivestock products.

Forage Research for Smallholders — Where to from Here?

Identifying new well-adapted forages?

Farmers can produce higher quality forage throughthe use of improved grass species, such as napiergrass, in conjunction with fertiliser or manure use, orprovide high-quality forage through the incorpora-tion of legumes into animal diets. Both strategieshave a role and are being used in a range of environ-ments throughout the tropics.

For instance, in Kenya, rapid advancement in thedairy industry has been largely based on the use ofnapier grass (Pennisetum purpureum) or other Pen-nisetum spp. Throughout Asia and the Americas,Pennisetum hybrids also form a major component ofimproved forages, but several other species such asPaspalum atratum, Panicum maximum and Brachi-aria spp. are also used in various farming systems inthe tropics (Horne and Stür 1999).

Legumes to supplement crop residues and grassesare also being adopted much more widely; again, arange of species is already being used. In WestAfrica, Stylosanthes scabra has been adopted as asupplementary feed and S. guianensis is being usedwidely in Southeast Asia. In sub-Saharan Africa,velvet bean (Mucuna pruriens) and lablab (Lablabpurpureus) are being used, while in the Americas,Arachis pintoi and Desmodium ovalifolium havebeen adopted by smallholder farmers. Horne and Stür(1999) list a range of other legume species that arebeing used or have potential as forage in smallholdersystems in Southeast Asia.

In general, the large number of well-adaptedspecies identified over many years of research andthe range of planted forages being used in some wayin diverse environments suggest that finding newforage species and cultivars is not a priority researchactivity, at least for now.

If the knowledge base for forage adoption can beconsidered at least adequate, what approaches mightthe research community take to enhance adoption?



32

Forage research — approaches to encourage adoption

Low adoption rates over the past 40 years indicatethat many farmers do not see investment in forages asa priority. If farmers are to take advantage of foragetechnology to meet livestock market demands, newapproaches need to be applied to targeting, designingand conducting research, and providing outcomes tofarmers. These approaches include: better identifica-tion of those farmers for whom forage planting is asustainable practice to enhance livestock productionand farm income; and better delivery of informationon which judgments on planted forages and changesin practice can be made. Two contrasting researchprojects, which address these approaches in southernAfrica and Indonesia, are outlined below.

Forage Research Using Teams, Participatory Research and a

Systems Approach: an Example from Southern Africa

This model of research has been widely used over thepast 10 years (e.g. Horne and Stür 1997; Thorpe1999) and so cannot be considered novel. Its key ele-ments are its truly participatory approach and the useof a team of scientists and extension specialists whocan evaluate the role, management, and biologicaland economic benefits and costs of new plantedforages within the whole farm enterprise.

In 1999, a project was initiated to develop foragetechnology for mixed cropping– livestock systems insouthern Africa. Aims of the project, funded by theAustralian Centre for International AgriculturalResearch (ACIAR), included:• the identification of well-adapted legumes for use

in rotations with cropping• the development of strategies for feeding legumes

to supplement maize stover and other low-qualityforage

• the development of strategies to make the best useof soil nitrogen accumulated during the foragelegume phase of the rotation.Although the project operates in the Limpopo

Province of South Africa and near Wedza in Zim-babwe, only the results from Zimbabwe will be dis-cussed here. Farmers have been partners in mostaspects of the research program, all of which is takingplace on-farm. Most importantly, the project team

restricted its research program to groups of farmerswho were perceived to have the capacity to change.Communities that might participate in the projectwere selected according to a number of criteria, themost important of which were:• Farmers were already producing livestock or milk

for marketing.• It was believed that some members of the

community had sufficient capital to considerincluding inputs in their farming systems.

• At least some farmers had sufficient land to enablethem to fallow an area of cropping land each year.

• The community had access to experiencedextension officers who understood their farmingpractices.Two communities in close proximity to each other

are participating in the project. The Zana II commu-nity is a ‘resettlement’ community and has beenfarming its land, over which the community hastenure, for more than 20 years. The Dendenyore com-munity is on communal land, with none of thefarmers having tenure over the land.

Experienced local extension officers who had adetailed knowledge of the local community under-took initial benchmarking of these communities atthe beginning of the project. The aim of the bench-marking was to determine farming practices, produc-tion levels, and the consumption and/or sale ofagricultural outputs. Several groups of farmers(about five farmers per group) were interviewed ineach of the communities. Farmers on the communalarea of Dendenyore produced maize and some othercrops but also produced beef cattle for market. Incontrast, farmers from Zana II produced a greaterarray of crops, were investing in cash crops (such aspaprika) and were involved in livestock productionthat focused on milk production.

The farmers perceived their communities toconsist of three distinct groups based on wealth. Theywere able to describe in detail how they saw meas-ures of wealth, based on capital and farming equip-ment. For instance, farmers in ‘wealth group 1’ hadcultivators, harrows and carts, while the pooresthouseholds (wealth group 3) had no farming imple-ments. Crop yields varied across wealth groups. Mostof the wealthy farmers had up to 1.2 ha of fallow landper year (see Tables 1 and 2). In Zana II, the numberof cows per farmer ranged from two to four, whilemilk yields ranged from 3 L to 6 L/day. Off-farmlabour was used in the resettlement community ofZana II but not in Dendenyore. In Dendenyore, there



33

was a vast range in the age of finished bullocks (3–10 years) and sale price (Zim$2000–8000/animal).

The benchmarking provided a detailed overviewof wealth and farming practice and the relationshipsbetween these two factors. It enabled the identifica-tion of smallholder farmers in each community whomight be best situated to adopt new practices becauseof their greater wealth and access to inputs. It alsoprovided baseline criteria, such as level of animalproduction, cropped area and grain yields. Thesebaseline data are essential to assessment of projectimpact in terms of productivity changes; impactassessment will be made at the project’s completionin 2003.

After three years of project implementation,research indicates that velvet bean and lablab areamongst the best-adapted legumes for use in rotationwith maize on the acid, light-textured cropping soils;that there are large benefits to subsequent maize

crops when legumes have been included in a rotation;and that both beef and milk production are increasedby including legumes in feeds. While none of theseresults is new, importantly, they have been obtainedon-farm, with farmers participating in most aspectsof the work. Farmer response to these research resultsis promising. Farmers who have been part of theproject since the beginning, and their neighbours, aresowing lablab or velvet bean in their fallow land attheir own expense and initiative, i.e. in addition toplanting as part of the research program (seeTable 3). They are investing considerable labour intomaking and storing hay from these legumes in earlyautumn, and are feeding it to lactating cows andpenned beef animals during the dry season. Supple-menting maize stover with lablab hay has signifi-cantly increased milk yields (measured by thefarmers) from 4–6 to 6–17 L/day. Similar improve-ments in liveweight gains have been recorded when

Table 1. Resources, crop yields and number of animals for each of three wealth groups within the Dendenyorecommunal farming community (n = 49 households).

Resource or crop Wealth group 1 (wealthiest)

Wealth group 2 Wealth group 3 (poorest)

Bank account >Zim$5000a Zim$1–4000 <Zim$3000

Arable land 3–5 ha 2–4 ha <2 ha

Maize yields 2.0 t/ha 1.5 t/ha 0.8 t/ha

Sweet potato yields 15 t/ha 2 t/ha 1 t/ha

Groundnuts 75 bags/ha 50 bags/ha 12 bags/ha

Oxen 4 2 Nil

Cattle >10 5–9 <5a US$1.00 = Zim$60, May 2001

Table 2. Resources, crop yields and number of animals for each of three wealth groups within the Zana IIresettlement farming community (n = 75 households).

Resource or crop Wealth group 1 (wealthiest)

Wealth group 2 Wealth group 3 (poorest)

Bank account Zim$10,000–100,000a Zim$1000–9000 <Zim$1000

Maize yields 3 t/ha 2 t/ha 2 t/ha

Tobacco 2 t/ha 1.5 t/ha Nil

Paprika 2 t/ha 1.25 t/ha 1.0 t/ha

Farm labour Up to 3 employed Occasional Nil

Oxen >4 2 Nil

Dairy cows >3 2 2a US$1.00 = Zim$60, May 2001



34

low-quality forage has been supplemented withlablab hay. Changes in farmer practices, not directlyassociated with the forage research results per se,have also occurred because of farmers’ perceptionsof what can be achieved through better practice (seeTable 4). These changes in practice are possiblymore important than the research results themselves,because farmers are demonstrating that they arewilling to make significant investment to achievechange and improve production.

We consider that several factors have been impor-tant in achieving the enthusiastic partnership with thefarmers in this project and achieving change in prac-tice. These include the use of a participatory

approach, the success of well-adapted forages, and afocus on the more wealthy farmers who wereexposed to existing dairy and beef marketing prac-tices. Primarily, it is these farmers who haveincreased the areas sown to legumes outlined inTable 3. However, the teamwork of researchers andextension specialists has been equally important. Theteam of researchers with expertise in plant adapta-tion, soil nutrition, crop agronomy and livestocknutrition has enabled a range of issues to beaddressed. The role of the extension specialists in theproject has enabled researchers to better identifyfarmers’ needs and priorities. Most importantly, theresearchers and extension specialists have worked

Table 3. Changes in the number of farmers using lablab, velvet bean or bana grass and making legume hay fordairy or beef production in Zana II and Dendenyore communities in Zimbabwe since the beginning ofthe project in 1999.

Forage/practice Zana II community Dendenyore community

1998–99 1999–00 2000–01 1998–99 1999–00 2000–01

Lablab Nil 4 11 Nil 4 13

Velvet bean 2 6 13 3 5 31

Bana/napier grass 0 8 12 2 2 38

Hay making 2 5 8 3 4 13

Table 4. Descriptive changes in practice as indicators of adoption of legume technology in the Zana II andDendenyore farming communities in Zimbabwe.

Change in practice from 1999 to 2001 Comments

Thirty-eight farmers in the community now hold ‘dairy production’ certificates

The demand for dairy certificates has been farmer-driven and was not envisaged in the initial project design.

Three farmers from within the project have attended ‘train the trainers’ courses

This development has been encouraged both by the farmers and by the extension specialists

Fourteen fact sheets have been produced in relation to legumes in the cropping system; four of these have been produced by farmers in partnership with AGRITEX

Some key farmers are recognising their own expertise in certain areas of legume production and use, and have willingly provided key information for these fact sheets.

Some farmers are purchasing additional dairy animals as production increases

This is a strong indication of adoption: buying these animals requires substantial investment.

Farmers are replacing high-cost dairy concentrates with legume hay

Initially, the project aimed to improve production of dairy and beef animals by supplementing animals that produced milk primarily from veld grazing. However, more affluent dairy producers have taken the opportunity to reduce costs by replacing concentrates with legume hay while maintaining or increasing milk production.

Farmers in adjacent communities are adopting forage options without the project having any on-farm study in those communities

This is another strong indicator of adoption: the on-farm results are being communicated by farmers themselves.



35

with farmers to transform research results into on-farm practice.

At this early stage, there is still some doubt as towhether this initial adoption will be sustained, butevidence of new practice and farmer investment sug-gests that some of the changes will be.

Evaluation of Forage Value in Farming Systems Before On-farm Research: a Project in Sulawesi,

Indonesia

The previously outlined decline in cattle numbers inSulawesi will inevitably mean that this region will beunable to maintain its role as an exporter of cattle.This outcome will impact on the agricultural andoverall economy in a region that is already one of thepoorest of the 26 provinces of Indonesia. Based oncattle prices of about 3–4 million rupiah (Rp) peranimal (US$1.00 = Rp11,200 at June 2001), theexport industry has a gross worth of aboutUS$75 million. For individual farmers, cattle salesare extremely important because they provide ameans of obtaining a significant cash income in addi-tion to income from crop production, which is oftenlimited to subsistence production levels.

Previous research has provided considerableunderstanding of the complexities of forage and live-stock production in rice-based farming systems inSoutheast Asia (Perkins et al. 1986; Ranjhan 1986),particularly in Sulawesi (Rachmat et al. 1992). How-ever, there has been no quantitative evaluation of theeconomic costs, benefits and risks of using improvedforages to increase livestock production.

In this project, a farming systems analysisapproach is being adopted to evaluate the potentialimpact of improved forages on production of cattle.A simulation modelling approach is being used toprovide estimates of forage production, its impact onlivestock production, and the possible biophysicalinteractions between livestock and crop production.Finally, a whole-of-farm economic analysis will beused to evaluate the economic benefits and risks offorage adoption. This combination of simulation andeconomic analysis provides information to thefarmer on the possible on-farm consequences offorage adoption and also provides researchers andfunding agencies with evidence on which to basedecisions about research investment.

The project focuses on smallholder farmers oper-ating mixed farming systems of: (a) rice, maize,groundnut and cattle; and (b) estate crops, rice andcattle. To date, farmers in these systems rely almostexclusively on cut-and-carry forage, and communalgrazing of naturally occurring vegetation and cropresidues, as feed for their animals. This is largelybecause of intensive cropping in the farming systemsand limited opportunity for spatial and temporalplacement of improved forages within the systems.

Nevertheless, there are various possible opportuni-ties for placement of forages into these farming sys-tems, including placement of forages in communallygrazed areas, placement in estate cropping systems,use of herbaceous forages with field cropping (suchas relay cropping), and even the placement of forageson bunds. Improving forages in communally grazedareas is notoriously difficult because of the absenceof grazing or harvesting management; moreover,sowing forages on bunds in rice paddies offers littlepotential to provide substantial amounts of forage.The best opportunities for substantially increasingforage production are likely to be the use of herba-ceous legumes and grasses in relay cropping (espe-cially in those areas where only one or two crops ofrice, peanuts, maize or soybean are grown annually)or in estate crops (especially newly establishedorchards). The project is concentrating on these twooptions.

The key questions being addressed are:• What benefits, if any, to animal production and

whole-farm income might be realised from theadoption of high-quality forages?

• What are the related costs of and constraints on thatadoption?

• Is the use of planted forages ever a viableeconomic option, and, if so, when and where in thecropping systems might they be produced?

• If planted forages are a viable option, do farmersfind the provision of information relating toadoption, in terms of potential economic costs andbenefits, useful in decision making?Providing answers to these questions should move

the research and development of tropical forages insmallholder farming systems forward. The topicneeds to progress from species evaluations and thedescription of farming systems, to a consideration ofthe broader issue of technology adoption by small-holders. In particular, it needs to concentrate on thepotential use and impact of high-quality forages on



36

livestock production and whole-farm income in rain-fed cropping systems.

Conclusions

Despite almost 50 years of investment in forageresearch in the tropics, forage adoption has been rel-atively poor across all tropical farming systems, yetthere is overwhelming evidence that planted foragescan make a substantial impact on livestock produc-tion. When considered together, these two factssuggest serious flaws in the research community’sunderstanding of smallholder farming systems and ofhow farmers perceive forages. The lack of adoptionof well-adapted plants also demands that researchersand extension specialists themselves adopt a newparadigm — one that ‘recognises’ the socioeconomicfactors affecting adoption and is also dominated bysocioeconomic considerations. Forages are not acommodity in themselves but a means to providinglivestock products. As such, they are not usually highon a farmer’s list of priorities. It is now time forforage researchers to place more emphasis on pro-viding evidence to farmers of the economic benefitsand costs of forages.

As most livestock in the tropics are in mixed crop–livestock systems, research and development pro-viders must prioritise these systems. Of course, this isnot a new approach, and undoubtedly most forageresearch being conducted today is aimed at thesefarming systems. It is in these systems that forageshave the potential to play a large role in supple-menting the crop residues that in many cases form thebulk of feed resources. The major question is howmuch, if any, resources should be aimed at opengrazing/communal land situations where, in thedeveloping world at least, there is rarely any controlover grazing. In such circumstances, any judiciousfarmer would be unwilling to invest resources, as thereturn on that investment would be shared withothers. In these situations, until grazing can be con-trolled, further research and development on newforage plants or other management practices cannotbe justified. Instead, investment to demonstrate todecision makers the benefits of implementingchange, especially in grazing control, within the soci-oeconomic constraints of the community would be apotentially better use of research and developmentresources.

Given past difficulties in achieving adoption,research and development providers may need to

relinquish their often-desired aim of introducingforage development amongst the most resource-poorfarmers. These farmers, by definition, are the mostsensitive to risk and are most unlikely to adopt newforage technologies. In most smallholder systems,these poorest farmers have little or no land in additionto their cropping land; they do not have land on whichto grow forages and seldom have the cash reservesnecessary for new initiatives. Instead, more receptivetargets for adoption might be farmers who haveaccess to resources for investment and who are moreable to take risks.

The forage research environment has changed con-siderably over the past decade. Research funding forforage genetic resources, species evaluation andforage management has decreased greatly. At thesame time, new forage research and developmentprograms that consider socioeconomic factors arebecoming more prevalent, as evidenced by severalpapers in these proceedings. An even more recentresearch theme has been the use of simulation mod-elling of forage and animal production. The judiciouscombination of research on socioeconomic factors,simulation modelling and economic analysis appearsto offer the best hope for helping farmers to developfarming practices that enable them to take advantageof the rapid increase in consumption of livestockproducts, and of improved market conditions.

References

Blackburn, H. 1998. Livestock production, the environmentand mixed farming systems. In: Nell, A.J., ed., Livestockand the environment. Proceedings of the InternationalConference on Livestock and the Environment, Ede/Wageningen, Netherlands, 16–20 June 1997, organisedby the World Bank, the Food and AgricultureOrganization and the International Agricultural Centre.Wageningen, The Netherlands, InternationalAgricultural Centre, 114–123.

CAST (Council for Agricultural Science and Technology)1999. Animal agriculture and global food supply, Chapter4. Ames, Iowa, CAST, 92 p. <http://www.cast-science.org> (accessed 18 July 2000).

Delgado, C., Rosegrant, M., Steinfeld, H., Ehui, S. andCourbois, C. 1999. Livestock to 2020: the next foodrevolution. Food, Agriculture and the EnvironmentDiscussion Paper No. 28, Washington DC, InternationalFood Policy Research Institute, 72 p.

Dzowela, B.H. 1993. Advances in forage legumes: a sub-Saharan African perspective. Proceedings of the XVIIInternational Grassland Congress, Palmerston North,



37

New Zealand and Rockhampton, Australia, 8–21February 1993, 2115–2119.

FAO (Food and Agriculture Organization of the UnitedNations) 1999. Livestock industries of Indonesia prior tothe Asian financial crisis. FAO RAP publication 1999/37.Rome, Italy, FAO, 198 p. <http://www.fao.org/DOCREP/004/AB986E/AB986E00>.

Gardiner, P. and Devendra, C., ed., 1995. Global agenda forlivestock research. Proceedings of a consultation held atthe International Livestock Research Institute (ILRI),Nairobi, Kenya, 18–20 January 1995. Nairobi, Kenya,ILRI, 118 p.

Hare, M.D., Thummasaeng, K., Suriyajantratong, W.,Wongpichet, K., Saengkham, M., Tatsapong, P.,Kaewkunya, C. and Booncharern, P. 1999. Pasture grassand legume evaluation on seasonally waterlogged andseasonally dry soils in north-east Thailand. TropicalGrasslands, 33, 65–74.

Horne, P. and Stür, W.W. 1997. Current and futureopportunities for forages in smallholder farming systemsof south-east Asia. Tropical Grasslands, 31, 359–363.

Horne, P. and Stür, W.W. 1999. Developing foragetechnologies with smallholder farmers — how to selectthe best varieties to offer farmers in Southeast Asia.ACIAR Monograph No. 62. Cali, Colombia, CIAT. 80 p.

Humphreys, L.R. 1991. Tropical pasture utilisation.Cambridge, UK, Cambridge University Press, 206 p.

Makembe, N.E.T. and Ndlovu, L.R. 1996. Dolichos lablab(Lablab purpureus cv. Rongai) as supplementary feed tomaize stover for indigenous female goats in Zimbabwe.Small Ruminant Research, 21(1), 31–36.

Moog, F.A. 1986. Forages in integrated food croppingsystems. In: Blair, G.J., Ivory, D.A. and Evans, T.R., ed.,Forages in Southeast Asian and South Pacific agriculture:proceedings of an international workshop held at Cisarua,Indonesia, 19–23 August 1985. ACIAR Proceedings No.12. Canberra, Australia, Australian Centre forInternational Agricultural Research, 152–156.

Mutukumira, A.N., Dube, D.M.J., Mupunga, E.G. andFeresu, S.B. 1996. Smallholder milk production, milkhandling and utilization: a case study from the Nharira/Lancashire farming area, Zimbabwe. Livestock Researchfor Rural Development, 8(1). <http://www.cipav.org.co/lrrd/lrrd8/1/mutuku.htm> (accessed 30 June 2003).

Ndlovu, L.R. and Sibanda, L.M. 1996. Potential of dolichoslablab (Lablab purpureus) and Acacia tortilis pods insmallholder goat kid feeding systems in semi-arid areasof Southern Africa. Small Ruminant Research, 21, 273–276.

Pedersen, C. and Madsen, J. 1998. Constraints andopportunities for improved milk production and calfrearing in Sanyati communal farming area, Zimbabwe.Livestock Research for Rural Development, 10(1).<www.cipav.org.co/lrrd/lrrd10/1/pede101.htm>(accessed 30 June 2003).

Perkins, J., Petheram, R.J., Rachman, R. and Semali, A.1986. Introduction and management prospects forforages in Southeast Asia and the South Pacific. In: Blair,G.J., Ivory, D.A. and Evans, T.R., ed., Forages inSoutheast Asian and South Pacific agriculture:proceedings of an international workshop held at Cisarua,Indonesia, 19–23 August 1985. ACIAR Proceedings No.12. Canberra, Australia, Australian Centre forInternational Agricultural Research, 15–23.

Rachmat, R., Stür, W.W. and Blair, G.J. 1992. Cattle feedingsystems and limitations to feed supply in South Sulawesi,Indonesia. Agricultural Systems, 39, 409–419.

Ranjhan, S.K. 1986. Sources of feed for ruminantproduction in Southeast Asia. In: Blair, G.J., Ivory, D.A.and Evans, T.R., ed., Forages in Southeast Asian andSouth Pacific agriculture: proceedings of an internationalworkshop held at Cisarua, Indonesia, 19–23 August 1985.ACIAR Proceedings No. 12. Canberra, Australia,Australian Centre for International AgriculturalResearch, 24–28.

Reynolds, L., Metz, T. and Kiptarus, J. 1996. Smallholderdairy production in Kenya. World Animal Review, 87(2),66–73.

Rivas, L. and Holmann, F. 2000. Early adoption of Arachispintoi in the humid tropics: the case of dual-purposelivestock systems in Caquetá, Colombia. LivestockResearch for Rural Development, 12(3). <http://www.cipav.org.co/lrrd/lrrd12/3/riva123.htm> (accessed30 June 2003).

Squires, V.R., Mann, T.L. and Andrew, M.H. 1992. Problemsin implementing improved range management on commonlands in Africa: an Australian perspective. Journal of theGrassland Society of Southern Africa, 9, 1–7.

Thomas, D. and Sumberg, J.E. 1995. A review of theevaluation and use of tropical forage legumes insub-Saharan Africa. Agriculture, Ecosystems andEnvironment, 54, 151–163.

Thorpe, W. 1999. ILRI’s smallholder dairy systemsresearch: experiences and lessons from collaborativeR&D in eastern Africa. In: Thornton, P.K. and Odera,A.N., ed., Workshop on ecoregional research atInternational Livestock Research Institute (ILRI):Proceedings of a workshop held at Addis Ababa,Ethiopia, 5–8 October 1998. Addis Ababa, Ethiopia,ILRI, 69–82.



38

Selecting Potential Fodder Bank Legumes in Semi-arid Northern South Africa

G. Rootman,* M. Mabistela† and B. Pengelly§

Abstract

Two experiments were conducted in Limpopo Province, northern South Africa, to determine the suitability of54 tropical forage legume accessions for use in fodder banks. In a small-plot evaluation trial, accessions ofStylosanthes scabra, Macroptilium bracteatum, M. atropurpureum and Desmanthus pubescens persisted forthree years and there was seedling recruitment in the annual legume Chamaecrista rotundifolia. In a secondexperiment, S. scabra cv. Seca was managed as a fodder bank to determine its yield and persistence in thissemi-arid environment. Overall establishment was poor, but total dry-matter yield ranged from 1 to 1.5 t/ha,even in a very dry year. In contrast to the performance of cv. Seca in Australia, there was almost no seedlingrecruitment during the three-year experiment.

The species listed above may provide options for fodder banks in this region, although the combination ofa semi-arid, unreliable climate, sandy soils, and continuous grazing by indigenous game is likely ensure thatfodder bank technology is a relatively unreliable source of animal feed in this environment.

The Limpopo Province of northern South Africahas a semi-arid tropical climate, with 70% of theprovince receiving rainfall of less than 600 mm peryear. Agriculture is the biggest provider of liveli-hoods, with large-scale commercial farming ofcattle, maize and horticultural crops as well assmallholder farming of beef cattle and maize beingimportant. In addition to large-scale commercialand smallholder farmers, recent provincial govern-ment programs to redistribute land have resulted ina growing number of medium-scale animal pro-ducers (emerging farmers) who grow beef cattle formarket, primarily feedlot markets, rather than forhome consumption. There are also a small numberof dairy producers.

At all scales of enterprise, animal production isdependent on native forages and is severely constrainedby low forage quality for much of the dry season (Marchto November). The integration of well-adapted legumesinto animal production systems has improved foragequality in other tropical countries. In Australia, augmen-tation of native pastures with legumes such as stylos(Stylosanthes hamata and S. scabra) has been widelyadopted by beef producers and has resulted in annualliveweight gains increasing by about 50% (Coates et al.1997). Forage legumes such as S. scabra have also beensuccessfully used as standing feed that can be strategi-cally fed during the dry season. In West Africa, about27 000 farmers have adopted stylo fodder banks for usein the dry season (Elbasha et al. 1999). The integrationof forage legumes into livestock production systems inLimpopo has the potential to play a similarly pivotalrole in reducing the impact of low forage quality onanimal production.

The aim of the study described here was to identifyforage legumes that could be used to provide high-

* Towoomba ADC Private Bag X1615, Bela-Bela,Republic of South Africa.

† Department of Agriculture & Environment, Private Bag X9487, Polokwane 0700, Republic of South Africa.

§ CSIRO Sustainable Ecosystems, 306 Carmody Road, St Lucia, Queensland 4067, Australia.



39

quality forage in fodder banks grown on cultivatedland adjacent to the natural pastures or veld. Thisfodder bank option for providing legume-based feedwas seen as preferable to augmentation of native pas-tures in Limpopo Province because of the low rainfallof the region, and because it is much simpler to operateand is better suited to smallholder systems, partlybecause farmers have greater control of the use of thelegume feed as a supplement throughout the year.

A second trial was sown to determine the potentialyield and persistence of S. scabra cv. Seca whensown for a fodder bank in the Limpopo Provinceenvironment. This cultivar was selected on the basisof its performance in Australia in semi-arid environ-ments; it was deemed the most likely to persist andproduce significant yield.


Project site

An existing cooperative dairy scheme was selectedas the experimental site. The Batlokwa dairy schemeat Tarentaaldraai was founded during the early 1980sand continues to produce milk for sale in nearby com-munities. Tarentaaldraai is in the northwest ofLimpopo Province in the Matoks area (23∞27'S29∞37'E), halfway between Polokwane (Pietersburg)and Machedo (Louis Trichardt), and is on the Tropicof Capricorn at an elevation of 1048 m. The native

vegetation is locally described as arid sweet bush-veld. Rainfall is about 500 mm per year; althoughaccurate long-term data for the site are not available,long-term records from the nearby Mara ResearchStation are in Table 1. Soils are light-textured loamysands (>1 m depth), neutral in reaction in the Ahorizon (0–10 cm) and have low phosphorus levelsbelow 10 cm depth. The basic cations Ca, Mg and Kare in the adequate range (Table 2).

Legume adaptation

Fifty-four legumes accessions (Table 3) werechosen for inclusion in this evaluation study based ontheir history of adaptation elsewhere in dry or semi-arid tropical and subtropical environments. Inocu-lated seed of all legumes was sown into a preparedseedbed to which 22 kg/ha P was applied as singlesuperphosphate on 3 March 1999. Seeds werecovered with soil using a rake. Each plot consisted ofa single 4 m row with 1 m between rows. Sowingdensity depended on seed size with the aim being toobtain at least 20 plants per metre of row. Yieldratings were taken one month after establishment,and legume presence and absence was recorded eachyear for three years.

Stylosanthes fodder bank

Seca stylo seed was sown into a cultivated seedbedon 28 February 2000 at a rate of 4 kg/ha. Two sowing

Table 1. Mean (1960–1999) maximum and minimum temperature (∞∞∞∞C) and mean (1936–1999)monthly rainfall (mm) for Mara Research Station (23∞∞∞∞9'S 29∞∞∞∞34'E, elevation 894 m),situated 40 km north of Tarentaaldraai.

Month Maximum temperature

(∞∞∞∞C)

Minimum temperature

(∞∞∞∞C)

Rainfall

(mm)

JanuaryFebruaryMarchAprilMayJuneJulyAugustSeptemberOctoberNovemberDecemberMean temp/total rainfall

30.429.628.726.824.922.422.624.627.328.228.229.626.9

18.017.716.212.9

8.14.64.67.1

10.914.116.117.312.3

79.757.954.635.012.5

5.03.12.1

13.632.865.880.6

442.7



40

treatments were used: seed was broadcast and eitherpressed or not pressed into the soil with a tractorwheel. Each treatment was replicated twice. The sitewas on the same soil type described in Table 2 andhad a slight slope.

Plant density was recorded after establishment,after the first rainfall of the second summer, andagain at the end of the third summer, using fifty0.25 m ¥ 0.25 m quadrats per replicate. Because ofpoor and uneven plant establishment and survival,biomass was only recorded in the higher density sec-tions of the experiment in the second and third year,from cuts taken from about 20 quadrats.


Rainfall in March 2000 (approximately 150 mm) andfollowing overcast conditions resulted in good emer-gence in both trials. Total rainfall in year 1 (1999–2000) and year 2 (2000–2001) was at or near the meanannual rainfall. However, in the third year of the trialthere was a severe drought, with almost no rainfallrecorded from December 2001 to late April 2002.

Legume adaptation

Most of the 54 legumes established well. The latesowing time and grazing by indigenous game meantthat no species were high yielding, but one accessionof A. americana (CPI 93624) and a number of Vignaand Macroptilium accessions had the highest yields(Table 3). Almost all legumes were heavily frosted inJune 2000, with only Lotononis bainesii cv. Miles

being unaffected, while Desmanthus leptophyllus cv.Bayamo and TQ 90 and D. pubescens cv. Umanmade some growth during the dry season. Mostaccessions failed to persist into the second season;those that persisted and regrew in November 2000are noted in Table 3. By March 2001, at the end of thesecond summer, only seven accessions (Stylosanthesscabra, Macroptilium bracteatum, M. atropur-pureum and Desmanthus pubescens) had persistedunder heavy grazing by indigenous game (Table 3),although seedlings of Chamaecrista rotundifolia hadgerminated and established. Most of these sevenlegumes survived the extremely dry summer of2001–02, but production was minimal.

While the persistence of Seca stylo was expected,the persistence of Desmanthus pubescens cv. Umanon the light-textured soils suggests that this speciesmight also have a role in this environment. Similarly,the production and persistence of the Macroptiliumaccessions suggest that M. atropurpureum andM. bracteatum might be considered as legumes forshort-term pastures or fodder banks. Macroptiliumaccessions might have the advantage over some ofthe productive annual Vigna accessions, as they havethe ability to persist for at least two years.

Seca fodder bank

Heavy rainfall soon after sowing resulted in con-siderable surface wash, loss of seed and unevenestablishment. Establishment at the bottom of theslope was poorer than at the top. Nevertheless, therewas a significant difference in plant density betweenpressed (32 plants/m2) and unpressed (9 plants/m2)treatments (Table 4). This may have been because ofbetter contact between soil and seed, or becausewater concentrated at the bottom of the tractor wheeltrack, providing better soil moisture conditions forgermination and establishment. While the differ-ences in plant density between treatments remainedthrough to years 2 and 3, those differences were notsignificant.

Only 22% and 16% of established seedlings in therolled and unrolled treatments, respectively, survivedinto the second summer of the trial, but most of thosethat did were still present towards the end of the thirdsummer (February 2002; Table 4). Although therewas some seedling establishment at the beginning ofthe second summer (November 2000), this was prob-ably from seed that had failed to germinate at sowingrather than from new seed set in the first year. No

Table 2. Soil chemical properties at Tarentaaldraai.

Property Soil depth (cm)

0–10 10–50

pH (H2O) a 7.6 5.5

Phosphorus (mg/kg) b 22 10

Calcium (cmol(+)/kg) c 6.65 2.38

Magnesium (cmol(+)/kg) c 1.98 1.10

Potassium (cmol(+)/kg) c 0.97 0.56

Exchangeable acidity (me%) 0.09 0.06

Clay (%)d 14 18a 1:5 soil:waterb 1:7.5 extractant Bray 2c 1:10 extractant ammonium acetate 1N, pH7d Near infrared determination



41

seedlings were observed after November 2000, butplants flowered and there was some seed set during

the trial. In a number of observations, that seed wasnot well filled.

Table 3. Legume accessions sown in an adaptation trial at Tarentaaldraai in March 2000; ticks indicate the highestyielding accessions in year 1, and those persisting into the second summer and at the end of the thirdsummer of the trial.

Species Accession No. plants establishedMarch 2000

High yield accessionsApr–May

2000

Regrowth or seedlingsNov 2000

Persistence Apr 2002

Aeschynomene americana CPI 56282CPI 93624cv Leecv Glenn

IntermediateIntermediateIntermediateIntermediate

✔

Aeschynomene falcata ATF 2194ATF 2196cv Bargoo

PoorPoorPoor

Aeschynomene histrix CPI 93599CPI 93636CPI 93638ATF 2191ATF 977

PoorPoorPoorPoorVery good

Aeschynomene villosa cv Reidcv KretschmerTP 188

Very goodVery goodVery good

Alysicarpus rugosus CPI 30034CPI 30187CPI 51655CPI 76978CPI 94489

Very goodVery goodVery goodVery goodVery good

Centrosema pascuorum CPI 64950cv Cavalcade Q10050CPI 55697

PoorPoorPoorVery good

Chamaecrista pilosa CPI 57503 Very good

Chamaecrista rotundifolia CPI 93094cv Wynn

Very goodVery good

✔

✔

✔

✔

Clitoria ternatea cv Milgarra Intermediate

Desmanthus pubescens cv Uman Poor ✔ ✔

Desmanthus leptophyllus cv Bayamo TQ 90

Very goodVery good

✔

✔

Lotononis bainesii cv Miles Very good ✔

Macroptilium atropurpureum cv Aztec Very good ✔ ✔ ✔

Macroptilium bracteatum CPI 27404CPI 55769CPI 68892


✔

✔

✔

✔

✔

✔

✔

✔

Macroptilium gracile CPI 84999CPI 93084cv Maldonado


✔

✔



42

Because of the uneven density throughout the trial,dry matter yield was measured from only the highestdensity areas to give an estimate of ‘potential’ yield.The potential yield was 1.5 t/ha in March 2001, 0.8 t/ha in November 2001, 0.9 t/ha in December 2001 and1.0 t/ha in January 2002.

Despite the uneven establishment and very dryenvironment, especially in the summer of 2001–2,this trial demonstrated that Seca stylo can establish,persist and produce at least 1–2 t/ha/annum oflegume biomass in this environment, and that this canbe expected to continue for at least three years.Although this suggests that Seca stylo may have arole in improving animal production as a fodder bankspecies, a major restriction could be the failure ofSeca to produce significant amounts of seed and soallow for seedling recruitment to replace the originalplants, which inevitably die. Perhaps, if fodder banksneed to be resown every three or four years, higheryields may be necessary to make their use attractive.

Failure to set seed might have several causes, such asfrost (which can occur as early as May), drought andcontinuous grazing by game throughout the year(which seem unlikely, given the history of stylo inAustralia), or disease (there was no sign of anthrac-nose in the experiment).

Conclusions

Macroptilium atropurpureum and M. bracteatummay have potential as fodder bank species in thisregion, although both species would probably need tobe resown every two to three years. AlthoughS. scabra persisted for three years, the failure to setseed, at least in this experiment, suggests that it haslittle advantage as a fodder bank species over theMacroptilium species, especially as its feed qualitycan be less than that of most other legumes (Jones etal. 2000). Chamaecrista rotundifolia seeded, andthere was recruitment of new plants in the legume

Macrotyloma daltonii CPI 60303 Very good

Stylosanthes guianensis cv Oxley Poor

Stylosanthes hamata cv Amigacv Verano

IntermediateIntermediate

Stylosanthes mexicana CPI 87484CPI 87847

IntermediateIntermediate

Stylosanthes scabra cv Seca Intermediate ✔ ✔

Vigna lasiocarpa CPI 34436 Very good ✔ ✔

Vigna luteola cv Dalrymple Very good ✔ ✔

Vigna oblongifolia Q 25362 Very good ✔ ✔

Vigna unguiculata var. dekindtiana CPI 121688 Intermediate

Table 3. (cont’d) Legume accessions sown in an adaptation trial at Tarentaaldraai in March 2000; ticks indicatethe highest yielding accessions in year 1, and those persisting into the second summer and at the end of thethird summer of the trial.

Species Accession No. plants establishedMarch 2000

High yield accessionsApr–May

2000

Regrowth or seedlingsNov 2000

Persistence Apr 2002

Table 4. Seedling and plant density (/m2) of Stylosanthes scabra cv. Seca in the rolled and unrolled treatments atTarentaaldraai.

Planting method Seedlings25/3/00

Plants28/11/00

Seedlings28/11/00

Plants15/2/02

Rolled 32.5 (15.5) a 7.0 (4) 3.5 (2.5) 5.5 (3.5)

Unrolled 9.0 (5) 1.5 (0.5) 1.0 (0) 2.5 (1.5)a Standard error of the mean in parentheses



43

adaptation trial. While reliable regeneration fromseed might negate the need to re-establish fodderbanks every two to three years, C. rotundifolia haspoor leaf retention (Jones et al. 2000), which mightmake it one of the least desirable legumes for a fodderbank.

Whichever species might be considered, however,continuous grazing by small indigenous game in theregion is a major constraint on using fodder banks,and it will be necessary to use effective fencing tomake the most of fodder bank technology. Theannual occurrence of frost, often as early as May,suggests that farmers would need to cut and conserveforage at the end of April to gain the most benefitfrom the technology.

References

Coates, D.B., Miller, C.P., Hendricksen, R.E. and Jones,R.J. 1997. Stability and productivity of Stylosanthespastures in Australia. II. Animal production fromStylosanthes pastures. Tropical Grasslands, 31, 494–502.

Elbasha, E.H., Thornton, P.K. and Tarawali, G. 1999. An ex-post economic assessment of fodder banks in WestAfrica. ILRI Impact Assessment Series No. 2. Nairobi,Kenya, International Livestock Research Institute, 61.

Jones, R.M., Bishop, H.G., Clem, R.L., Conway, M.J.,Cook, B.G., Moore, K. and Pengelly, B.C. 2000.Measurements of nutritive value of a range of tropicallegumes and their use in legume evaluation. TropicalGrasslands, 34, 78–90.



44

Assessment of the Variation in Growth and Yield of Diverse Lablab (Lablab purpureus) Germplasm

in Limpopo Province, South Africa

K.K. Ayisi,* M.P. Bopape* and B.C. Pengelly†

Abstract

Lablab (Lablab purpureus) is a dual-purpose legume with the potential to be incorporated into the smallholdermaize monoculture system which predominates in the Limpopo Province of South Africa. Thirty-threeintroduced accessions and three local plant types were evaluated in a farmer’s field near Polokwane during the2002/03 growing season for agronomic characteristics, biomass production and grain yield. Days to 50%flowering ranged from 51 to over 150 days and days to maturity ranged from 90 to 197 days among the entries.Approximately 50% of the accessions produced between 1000 and 5000 kg/ha of total biomass at 87 days afterplanting (DAP), with two of the entries producing over 7000 kg/ha. The maximum grain yield obtained wasbetween 500 and 600 kg/ha, produced by about 14% of the accessions. Generally, the early flowering andmaturing types produced the highest grain yields. This diversity in growth characteristics suggests potential forlablab to be a multi-purpose crop for use in the smallholder systems of South Africa. Further on-farmevaluation in partnership with farmers is suggested as the next step in testing promising accessions of lablab.

Lablab (Lablab purpureus) is a legume that is pres-ently grown in certain parts of Africa, namely, Cam-eroon, Sudan, Ethiopia, Uganda, Kenya, Tanzania,Swaziland and Zimbabwe, primarily as a food crop,with both the grain and the immature pods being con-sumed (Smartt 1985). Lablab is poorly known inSouth Africa but initial trials indicate that it can besuccessfully grown in the Limpopo Province, whichis in the northern part of the country. Lablab is usefulas a forage crop, a cover crop that can producenitrogen through biological nitrogen fixation,improve soil nitrogen status, and suppress weeds(Schaafhausen 1963; Humphreys 1995). Recognised

for its drought hardiness (Hendricksen and Minson1985; Cameron 1988), lablab has the potential to begrown in semi-arid parts of the Limpopo Provincewhere the mean annual rainfall is commonly < 650mm. Following an initial study with smallholderfarmers in rural communities, it was discovered thatthe fresh or dried lablab leaves could be cooked andconsumed as leafy vegetables and as well as beingused as a grain legume. Stimulating the interest inlablab as an alternative crop in these predominantlysmallholder farming systems requires evaluation andan active promotion of the crop within the province.Currently, the only commercial lablab cultivar in thecountry is Rongai, which is a long-duration foragetype (>150 days to maturity). There is an importantneed to identify other lablab types for integration intothe rural farming systems of the Limpopo Province.

The objective of this research was to assessbiomass production, yield performance and agro-

* School of Agricultural and Environmental Sciences,Plant Production, University of the North, Private BagX1106, Sovenga 0727, Republic of South Africa.

† CSIRO Sustainable Ecosystems, 306 Carmody Road, St Lucia, Queensland 4067, Australia.



45

nomic characteristics of diverse lablab germplasm inthe Limpopo Province relative to locally sourcedmaterial.


A field experiment was established on 10 December2002 on a clay loam soil (Bainsvlei Form accordingto the Soil Classification Working Group, 1991) on afarmer’s field at Dalmada, near Polokwane in theLimpopo Province. The soil at that location wasneutral pH and low in both phosphorus and potas-sium (Table 1). Further details of the soil can befound in Whitbread and Ayisi (2004).

The experiment was established as a randomisedcomplete block design with three replications. Eachexperimental unit was 3 m ¥ 3.6 m, consisting of fourrows of lablab with an inter-row spacing of 90 cm and22.5 cm between plants. Thirty-three lablab accessionswere selected from a collection of the species held atthe Australian Tropical Forage Genetic ResourcesCentre, Commonwealth Scientific and IndustrialResearch Organisation (CSIRO), Brisbane, Australia,on the basis of their relatively early flowering or appro-priate grain types as noted in a recent characterisationof the germplasm collection (Pengelly and Maass2001). These accessions were CPI 29399, CPI 29400,CPI 29803, CPI 34777, CPI 34780, CPI 35771, CPI35893, CPI 35894, CPI 36019, CPI 36093, CPI 30701,CPI 52504A, CPI 52504B, CPI 52506B, CPI 52508,CPI 52513, CPI 52514, CPI 52518B, CPI 52530, CPI52533, CPI 52535, CPI 52551, CPI 52552, CPI 52554,CPI 57314, CPI 57315, CPI 60795, CPI 81364, CQ3620, CQ 3621, Q 5427, Q 6880B, and Q 6988. Threelocal forage types that are locally known as ‘Rongaiblack’, ‘Rongai brown’, and ‘Rongai white’ wereassigned as controls in each block. All seeds were inoc-ulated with an appropriate commercial Bradyrhizo-bium inoculant before planting. The seeds were sown

by hand at a depth of 4–6 cm, then thinned 2 weeksafter emergence to the desired spacing. Weeds werecontrolled by hand-hoeing. A total of 297 mm of rain-fall was received at the site and 178 mm of irrigationwas applied (Table 2).

Data collected were emergence, mid-season dry-matter production, days to flowering, physiologicalmaturity, grain yield and its components. Emergencepercentage was recorded 3 days after seedling emer-gence. Dry matter was collected from two plants in themiddle rows of each plot, at 0.5 m, 87 days afterplanting (DAP) and dried in an oven at 65°C to constantmoisture content. Dried samples were weighed todetermine biomass production. Flowering was scoredwhen 50% of plants on a plot had flowered and physi-ological maturity was recorded when 90% of the podswithin an experimental unit were brown. At harvest,two 2 m lengths of the middle rows of each experi-mental plot were harvested for the determination ofgrain yield, number of pods/plant and seed weight/pod.

The data were analysed using the Statistical Anal-ysis System (SAS). Analysis of variance was used totest significance of treatment effect, and differencesdue to varieties were compared using Duncan’s mul-tiple range test (DMRT) procedure.


Weather

The total rainfall received at Dalmada fromDecember 2002 to July 2003 when plants weregrowing was 297 mm, of which about 91% fell inDecember, January and February (Table 2). The totalamount of irrigation applied was 178 mm. Maximumtemperatures remained above 28°C during the periodfrom December to April. The minimum temperatureremained above 12°C during this period but fell to4.4°C in June (Figure 1).

Agronomic characteristics

EmergenceAll accessions had in excess of 80% establishment

which was sufficient for excellent plot cover andreflected the favourable soil and weather conditionsat sowing.

FloweringThere were significant differences in days to flow-

ering among the lines. The earliest flowering accessionswere CPI 35894, CPI 52508 and CPI 52513, which

Table 1. The pH and phosphorus and potassium levelsat three depths prior to sowing of the lablabevaluation trial at Dalmada in December2002.

Soil depth (cm) 0–15 15–30 30–60

pHNitrogen (mg/kg)Phosphorus (mg/kg)Potassium (cmol(+)/kg)

6.83

261.00

6.91

140.84

6.817

0.64



46

flowered before 60 DAP. Accessions CPI 30701, Q5427, CPI 34780, CPI 29400, CPI 52518B, and Rongaiwhite, Rongai black and Rongai brown, were late flow-ering (Table 3). The majority of the lines floweredbetween 60 and 100 DAP (Figure 2). The three types ofthe local Rongai cultivar flowered later than 150 DAPas did CPI 34780, CPI 30701 and Q 5427.

Physiological maturity

Days to physiological maturity ranged from 90 to197 DAP. The accessions CPI 35771, CPI 35894,CPI 52504B, CPI 52508, CPI 52513, CPI 52533, CPI52535, CPI 52552, CPI 60795 and CQ 3620 maturedbefore 100 days and this constituted about 28% of thelines studied. The late-maturing genotypes wereRongai white, Rongai black and Rongai brown, CPI34780, CPI 30701 and Q 5427 which all matured

after 180 DAP. CPI 34780, CPI 30701 and Q 5427are clearly long-duration types, comparable to thelocal cultivar, Rongai. Evaluation of maturity acrossthe lablab entries indicated that the majority of thelines (72%) matured between 90 and 120 days, whilst11.0% and 17% matured between 120 and 150 DAPand after 180 DAP, respectively (Figure 3).

Table 2. Total monthly rainfall and irrigation waterapplied (mm) between December 2002 andJuly 2003 at Dalmada.

Months Rainfall Irrigation Total

DecemberJanuaryFebruaryMarchAprilMayJuneJulyTotal

10811051–––

28–

297

8693225–

102212

178

1161798325–

105012

475

0

5

10

15

20

25

30

35

Decem

ber

January

February

Marc

hApril

May

June

Tem

per

atu

re (°

C)

Max Min

Figure 1. Minimum and maximum temperaturesbetween December 2002 and July 2003 atDalmada.

Tem

pera

ture

(°C

)

Table 3. Flowering and physiological maturity (daysafter planting) of lablab accessions duringthe 2002/03 growing season at Dalmada.

Accession Flowering Maturity

CPI 29399CPI 29400CPI 29803CPI 34777CPI 34780CPI 35771CPI 35893CPI 35894CPI 36019CPI 36903CPI 30701CPI 52504ACPI 52504BCPI 52506BCPI 52508CPI 52513CPI 52514CPI 52518BCPI 52530CPI 52533CPI 52535CPI 52511CPI 52552CPI 52554CPI 57314CPI 57315CPI 60795CPI 81364CQ 3620CQ 3621Q 5427Q 6880BQ 6988Rongai blackRongai brownRongai whiteSignificanceCV (%)

94.3 c101.7 b90.7 c81.7 d158.0 a

63.3 g–k76.7 e59.7 k80.3 de83.0 d158.6 a65.0 g–j65.7 g–i76.0 e59.7 k51.7 l

65.0 g–j101.3 b66.7 fg

61.7 h–k64.7 g–j66.7 fg60.3 jk67.0 fg61.3 i–k61.7 h–k60.7 jk60.7 jk

63.0 g–k71.0 f

158.3 a64.7 g–j66.3 gh156.7 a157.7 a158.7 a

**2.9

125.0 e138.3 b115.7 g121.3 e197.0 a90.3 r

111.7 h99.7 o–q119.3 f111.0 h197.0 a

102.0 lm99.7 o–q107.0 j

99.7 o–q99.0 q

101.0 l–o127.3 c104.7 k99.3 pq

99.7 o–q101.7 l–n99.3 pq104.7 k

101.3 l–n100.7 m–p

99.0 q102.3 l99.0 q109.0 i197.0 a

101.7 l–n100.3 n–q

197.0 a197.0 a197.0 a

**0.6

Note: Numbers followed by the same letter or letters within a

column are not significantly different from each other (p < 0.01);

** = highly significant, p < 0.01; CV = coefficient of variation.



47

Biomass accumulation

There were significant differences (p < 0.01) intotal biomass accumulation of the plants as well as inthe leaf, stem and root fractions at 87 DAP (Table 4).CPI 29399 produced more than 7000 kg/ha biomass,which was comparable to that of Rongai brown. Thebiomass yield of Rongai black, CPI 30701, CPI52506B and CPI 81364 ranged form approximately5000 to 7000 kg/ha. Forty-seven percent of acces-sions produced biomass between 1000 and 5000 kg/ha by 87 DAP. CPI 52513 accumulated the least totaldry matter, which was less than 1000 kg/ha. Biomassproduction was generally related to flowering time.For instance CPI 29399 and Rongai brown were

amongst the late-flowering, high-yielding types andCPI 52513 was the earliest to flower and had a lowbiomass yield.

Grain yield and yield components

There were significant differences (p < 0.01) ingrain yield among the treatments studied. CPI 52514,CPI 52552, CPI 60795, CQ 3620 and Q 6880B pro-duced yields of >500 kg/ha, whereas CPI 29399, CPI29803, CPI 34780, CPI 35893, CPI 30701, CPI52518B, Q 5427, Rongai black, Rongai brown andRongai white produced grain yield of less than 50 kg/ha. The other lines were between the two yields(Table 5). The high-yielding lablabs were all early

51.5

9.1 9.16.1 6.1

0 0 0 0

18.2

0

10

20

30

40

50

60

Time of flowering (DAP)

%

(60–

70)

(70–

80)

(80–

90)

(90–

100)

(100

–110

)

(110

–120

)

(120

–130

)

(130

–140

)

(140

–150

)

(150

–160

)

Time of maturity (DAP)

0

1711

72

01020304050607080

%

(90–

120)

(120

–150

)

(150

–180

)

(180

–210

)

Figure 2. Frequency distribution of flowering in lablab accessionsduring the 2002/03 growing season at Dalmada.

Figure 3. Frequency distribution of maturity in lablab accessionsduring the 2002/03 growing season at Dalmada.



48

flowering. With the exception of CPI 60795, theyalso produced relatively lower total biomass indi-cating less investment in vegetative growth, andhigher numbers of pods per plant and heavier seedsper pod (Table 5). The poor-yielding types generallyflowered between 90 and about 100 DAP andmatured late. The flowering date of the late typesoccurred in March, which exposed all reproductivestages of these lines to much lower minimum and

maximum temperatures (Figure 1). This could havecontributed to lower seed production in the late-maturing types.

Conclusions

There was considerable physiological variationamongst the lablab accessions studied here. The ear-liest-flowering types gave promising total grain

Table 4. Dry-matter production (kg/ha) at 87 days after planting of lablab accessions during the 2002/03 growingseason at Dalmada.

Accession Leaf Stem Root Total

CPI 29399CPI 29400CPI 29803CPI 34777CPI 34780CPI 35771CPI 35893CPI 35894CPI 36019CPI 36903CPI 30701CPI 52504ACPI 52504BCPI 52506BCPI 52508CPI 52513CPI 52514CPI 52518BCPI 52530CPI 52533CPI 52535CPI 52511CPI 52552CPI 52554CPI 57314CPI 57315CPI 60795CPI 81364CQ 3620CQ 3621Q 5427Q 6880BQ 6988Rongai blackRongai brownRongai whiteSignificanceCV (%)

3231 a1666 de499 mn491 mn835 g–j1349 f491 mn640 i–m1013 g

798 g–k2414 b1345 f517 i–n2501 b

565 k–n339 n

656 i–m858 g–i

699 h–m655 i–m765 g–l1597 e

709 h–m1604 e866 g–l777 j–k1647 de2144 c1855 d942 gh

730 h–m588 j–m730 h–m2075 c3376 a1548 ef

**11.03

4440 a1229 i–o481 op

584 n–p3822 ab1631 h–l864 k–p1859 f–j1876 f–j

1315 h–m2563 d–f2108 e–h747 m–p3937 ab746 m–p

271 p1201 i–o1394 h–n1699 g–k1003 k–p1046 j–p2471 d–g1116 i–o2657 c–f1408 h–n1524 h–m3061 cd2858 c–e2879 c–e1488 h–m727 m–p771 m–p802 i–p2806 ce3393 bc1916 f–i

**23.7

325.9 a155.8 h–k166.6 g–i233.7 c226.3 c

177.6 f–h229.8 c

149.8 i–k87.8 op

118.5 k–m277.9 b

148.5 i–k71.5 pq

198.7 d–f75.1 pq

174.5 f–i92.2 op

93.9 n–p139.9 j–l161.2 h–j

132.6 k–m188.0 e–g95.5 n–p171.5 g–i188.3 e–g74.1 pq

210.2 c–e102.6 no215.1 cd

111.4 m–o74.7 pq

70.30 pq49.3 q227.4 c257.6 b

130.8 k–m**9.1

7796 a3050 f–i1146 op

1274 n–p4882 cd3158 f–h1585 k–p2583 g–k2977 f–i1623 k–p

5255 c3601 e–g1335 m–p

6637 b1296 n–p

784 p1941 i–o

2345 h–m2538 g–l1819 k–p1939 i–o4257 c–e1918 j–o4432 c–e2462 h–m2608 g–k4917 cd5045 c

3133 d–f2542 g–l1531 k–p1429 l–p1581 k–p

5129 c7117 ab

3595 e–g**

18.7

Note: Numbers followed by the same letter or letters within a column are not significantly different from each other (p < 0.01); ** =

highly significant, p < 0.01; CV = coefficient of variation.



49

yields, sometimes in excess of 500 kg/ha in this verydry environment. The diversity in growth character-istics suggests potential for the crop in smallholdersystems in South Africa, as a multi-purpose cropwhich might supply green pods as vegetables as wellas grain, and where leaf and stem postharvest might

offer improved forage to the small number of grazinganimals in the farming systems. Further studies arerequired to confirm their adaptability and find suit-able agronomic practices for cultivation. Equallyimportant are further studies in partnership withfarmers to assess lablab eating qualities.

Table 5. Grain yield and yield components of lablab accessions during the 2002/03 growingseason at Dalmada.

Accession no. Grain yield (kg/ha)

Number of pods/plant Seed weight/pod

(g)

CPI 29399CPI 29400CPI 29803CPI 34777CPI 34780CPI 35771CPI 35893CPI 35894CPI 36019CPI 36903CPI 30701CPI 52504ACPI 52504BCPI 52506BCPI 52508CPI 52513CPI 52514CPI 52518BCPI 52530CPI 52533CPI 52535CPI 52511CPI 52552CPI 52554CPI 57314CPI 57315CPI 60795CPI 81364CQ 3620CQ 3621Q 5427Q 6880BQ 6988Rongai blackRongai brownRongai whiteCV (%)

1.1 d79.1 a–d14.3 d

132.5 a–d5.0 d

106.6 a–d49.5 cd

187.2 a–d353.2 a–d92.3 a–d

7.2 d111.8 a–d156.1 a–d190.9 a–d412.2 a–d440.4 a–d518.7 a–c

1.1 d185.3 a–d347.5 a–d359.5 a–d439.6 a–d

576.2 a382.2 a–d90.8 a–d

112.7 a–d570.5 a–b99.7 a–d574.2 ab53.0 cd12.2 d

531.9 a–c71.2 b–d

2.6 d3.3 d7.4 d125.1

2.9 ef8.8 c–f3.2 ef

8.2 d–f1.4 f

12.5 b–f2.3 ef

8.1 d–f12.0 b–f5.1 d–f2.7 ef

6.9 d–f6.9 d–f10.1 c–f13.4 b–f29.2 a

24.9 ab0.8 f

6.8 d–f12.5 b–f8.9 c–f

23.2 a–c18.6 a–d15.1 b–f4.6 d–f9.1 c–f24.6 ab6.6 d–f

16.8 a–d5.3 d–f5.8 d–f

12.9 b–f8.1 d–f

14.6 b–f6.1 d–f7.0 d–f

73.1

0.3 fg1.6 b–g0.6 d–g2.3 b–g0.3 fg

2.9 b–d1.7 b–g2.5 b–f2.3 b–g2.3 b–g0.3 fg

2.8 b–e2.9 b–d

3.1 b2.7 b–e

6.4 a2.5 b–f0.5 e–g2.1 b–g2.5 b–f1.8 b–g2.8 b–e3.0 bc2.6 b–f2.7 b–e2.6 b–f3.2 b

2.6 b–f2.7 b–e1.7 b–g0.7 c–g3.5 b

2.3 b–g0.1 g0.2 fg0.2 g56.5

Note: Numbers followed by the same letter or letters within a column are not significantly different from each

other (p < 0.01); ** = highly significant, p < 0.01; CV = coefficient of variation.



50

References

Cameron, D.G. 1988. Tropical and subtropical pasturelegumes. Queensland Agriculture Journal, March–April,110–113.

Hendricksen, R.E. and Minson, D.J. 1985. Lablabpurpureus — a review. Herbage Abstracts, 55, 215–227.

Humphreys, L.R. 1995. Diversity of productivity of tropicallegumes. In: D’Mello, J.P.F. and Devendra, C., ed.,Tropical legumes in animal nutrition. Wallingford, UK,CAB International, 1–21.

Pengelly, B.C. and Maass, B.L. 2001 Lablab purpureus (L.)Sweet — diversity, potential use and determination of acore collection of this multi-purpose tropical legume.Genetic Resources and Crop Evolution, 48, 261–272.

Schaafhausen, R.V. 1963. Economic methods of using thelegume Dolichos lablab for soil improvement and feed.Turriabla, 13, 172–178.

Smartt, J. 1985. Evolution of grain legumes. II. Old and NewWorld pulses of lesser economic importance.Experimental Agriculture, 21, 1–18.

Soil Classification Working Group 1991. Soil classification:a taxonomic system for South Africa. Memoirs on theAgricultural Natural Resources of South Africa No. 15.ISBN 0-621-10784-0 Obtainable: Director, DirectorateAgricultural Information, Private Bag x144, 128 p.

Whitbread, A.M. and Ayisi, K.K. 2004. The description ofbiophysical environment of three maize-producing areasin the Limpopo Province of South Africa and thevalidation of APSIM to simulate maize production. Theseproceedings.



51

Evaluation of Herbaceous Forage Legumes Introduced into Communally Managed Rangelands

in Zimbabwe

P.H. Mugabe,1 D. Majee,2 X. Poshiwa,2 B. Chigariro,3 B. Ndlovu,1

F. Washayanyika1 and N. Mukombe3

Abstract

Eighteen legume accessions were tested over three growing seasons on a previously controlled grazing site anda communal grazing management site. Within sites, there were two soil wetness blocks — namely, vlei andtopland. Generally, the vleis produced better establishment in season 1. Most legume counts drastically declinedin seasons 2 and 3 on all sites except for uncontrolled topland where some legume accessions increasedsignificantly, including Aeschynomene villosa cv. Reid, A. villosa cv. Kretschmer, Chamaecrista rotundifolia cv.93094 and C. rotundifolia cv. Wynn. Biomass production was very low for most legumes across all sites and thepromising accessions were C. rotundifolia cv. 93094, C. rotundifolia cv. Wynn, C. pilosa, A. villosa cv. Reid andA. villosa cv. Kretschmer. While the results were not very encouraging, the C. rotundifolia accessions are worthpursuing on communal toplands, while the A. villosa accessions could be tried on the other sites.

The use of forage legumes for rangeland improvementin Zimbabwe has been researched for decades and hasbeen shown to be an option for improving degradedrangelands. For instance, Clatworthy and Madakadze(1988) recommended Desmodium uncinatum cv. Sil-verleaf for reinforcing topland rangelands on heavysoils and Macroptilium atropurpureum cv. Siratro,Stylosanthes guianensis cv. Oxley and Macrotylomaaxillare cv. Archer for sandy soils. Muchadeyi andChakoma (1995) and Majee and Chikumba (1995)identified Trifolium spp., Arachis pintoi cv. Amarillo,Chamaecrista rotundifolia and Aeshynomene ameri-

cana, among others, as potential legumes for use onsandy vlei sites.

Such research has been confined mainly toresearch stations and commercial farms with limitedwork on communally managed rangelands. Thecurrent research was done to evaluate the potential ofa range of herbaceous tropical forage legumes forimproving forage production on communallymanaged rangelands. Specific research objectiveswere to evaluate the establishment, persistence andbiomass production of a range of herbaceouslegumes on communal rangelands.


Site description

The study was carried out in Zana 2 Ward inWedza district in the Mashonaland East Province ofZimbabwe, located 70 km west of Marondera. The

1 Animal Science, University of Zimbabwe, PO Box MP167, Mount Pleasant, Harare, Zimbabwe.

2 Grasslands Research Station, Agriculture Research and Extension Services (AREX), P. Bag 3701, Marondera, Zimbabwe.

3 Agricultural Technical and Extension Services, PO Box 30 Wedza, Zimbabwe.



52

area receives 750–1000 mm of rainfall annually, andthe soils are predominantly sandy. Monthly rainfallreceived in the years 1996/97 to 2001/02 is displayedin Figure 1.

Experimental design

Test legumes were sown in two sites: a fenced, 7 hapaddock, under controlled grazing by one bull (con-trolled); and an unfenced communally managedgrazing area outside and adjacent to the 7 ha paddock(communal). Within each site, there were two soilwetness blocks — namely, topland (dry) and vlei(wetland). There was no replication of the soilwetness blocks within a site but the legume treat-ments were replicated three times per block. Eighteenlegume accessions of Aeschynomene, Chamaecrista,Lotononis, Desmodium, Macroptilium and Stylosan-thes species (Table 1) were sown on 26 November1999 in plots measuring 2.7 ¥ 2.8 m with 0.5 mbetween plots and 1m borders between replicates.Existing natural forage was burned and sowing rowsprepared with an ox-drawn plough at 0.5 m spacing.Sowing rates varied between 1 and 5 kg/hadepending on seed size and seed availability. Dolo-mite lime and single superphosphate were applied atplanting at 1000 kg/ha and 250 kg/ha, respectively,based on recommendations from soil analyses. Thecommunal site was temporarily fenced and notgrazed during the first two growing seasons to allowlegumes establishment.

Establishment and persistence assessment

Plant counts for establishment and persistenceassessments commenced on 17 December 1999 andwere repeated fortnightly until the end of March inthe first season and three times in each of the secondand third seasons. Legume plants were counted along1 m in each of the two middle rows per plot. Changesin plant counts both within and between seasons rep-resented persistence.

Biomass assessment

Biomass production was assessed in two 0.5 ¥ 0.5m quadrats per plot. Herbage was clipped to 5 cmabove the ground, separated into botanical compo-nents, oven-dried at 60°C for 72 hours and weighedto determine the dry matter yields. Biomass samplingwas done over two rainfall seasons on 17/11/00, 15/12/00, 12/01/01, 26/01/01, 08/02/01, 11/12/01, 03/02/02 and 27/04/02.

Analysis of variance (ANOVA) on establishment,persistence and biomass was done using the GeneralLinear Model procedures of SAS (1998).

Results

Establishment and persistence

Controlled vleiAll 18 legumes on the controlled vlei established

and persisted to the end of season 1 (1999–2000),

Rain

fall

(mm

)

500

400

300

200

100

0

1996/1997

1997/1998

1998/1999

1999/2000

2000/2001

Sept Oct Nov Dec Jan Feb

Month

Mar Apr May June July Aug

Figure 1. Rainfall data for Zana 2 Ward, Wedza district, Mashonaland East Province, Zimbabwe,for 1996–2001.



53

with Stylosanthes scabra cv. Seca, S. hamata cv.Verano, Aeschynomene villosa cv. Kretschmer, A.falcata TP 188, S. mexicana 87484 and A. falcata cv.Bargoo having more than 10 plants/m (Table 1). A.villosa cv. Reid had very poor establishment but hadbetter plant counts at the end of season 2 than otherlegumes. All legume plant counts decreased drasti-cally with zero counts for all in season 3 except for A.villosa cv. Reid and Lotononis bainesii cv. Miles.

Communal vleiEstablishment of legumes as shown by mean plant

counts was similar to that on controlled vlei in season1, with greater than 10 plants/m for S. scabra cv.Seca, S. mexicana 87484, S. hamata cv. Verano, A.falcata cv. Bargoo, A. falcata TPI 88, A. villosa cv.Kretschmer, A. histrix ATF 977 and A. histrix 93638.All legume counts had dropped by the end of seasons2 and 3 except for Lotononis bainesii cv. Miles whichincreased from 1.7 plants/m in season 1 to 6.1 plants/m in season 3.

Controlled toplandAll legumes except L. bainesii cv. Miles estab-

lished and persisted to the end of season 1. Noaverage counts greater than 10 plants/m wererecorded. Most counts declined to zero counts inseasons 2 and 3, but L. bainesii cv. Miles increasedfrom 0 plants/m in season 1 to 4 plants/m in season 3.

Communal toplandAgain, no average plant counts of greater than 10

plants/m were recorded, although the majority estab-lished and persisted to the end of season 1. Whileplant counts in season 2 were generally low, mostlegume plant counts had significantly (P < 0.05)increased by the end of season 3. Greater than 10plants/m were recorded for A. villosa cv. Reid,Chamaecrista rotundifolia cv. Wynn, C. rotundifolia93094, L. bainesii cv. Miles and Desmodium unci-natum cv. Silverleaf at the end of season 3 (Table 1).

Biomass production

Controlled vleiIn season 2, the highest legume biomass of 492 kg/

ha was recorded for A. villosa cv. Kretschmer. Theother significantly higher (P < 0.05) biomass per-formances were for A. villosa cv. Reid, L. bainesii cv.Miles and S. scabra cv. Seca. Biomass production forall legumes was negligible in season 3 on the con-trolled vlei (Table 2).

Communal vleiIn season 2, the highest biomass production of

1191.6 kg/ha was recorded for A. villosa cv. Reidwith A. villosa cv. Kretschmer, S. scabra cv. Seca,Macroptilium atropurpureum cv. Siratro and L.bainesii cv. Miles being the other significant (P <0.05) top performers (Table 2). The rest of thelegumes had negligible biomass production. Alllegumes had negligible biomass production in season3.

Controlled toplandA. villosa cv. Reid had the highest biomass produc-

tion of 111.2kg/ha followed by L. bainesii cv. Miles(72 kg/ha) and S. scabra cv. Seca (68.8 kg/ha) (Table2). The rest of the legumes had negligible biomass inseason 2. All legumes had negligible biomass inseason 3 (Table 2).

Communal toplandIn season 2, much higher biomass production was

recorded for some legumes compared to the con-trolled topland. C. rotundifolia cv. 93094, C. rotun-difolia cv. Wynn, M. atropurpureum cv. Siratro andC. pilosa had significantly (P < 0.05) higher total drymatter yields of 2094.2 kg/ha, 1700 kg/ha, 658.8 kg/ha and 517 kg/ha, respectively (Table 2). All legumeson communal topland had negligible biomass pro-duction in season 3.

Discussion

There was differential establishment of legumesdepending on previous grazing management and soilwetness. Generally, higher establishment wasachieved on the communal site. This could have beendue to the difference in the native forage soil seed-banks of the two sites. The communal site wouldhave had a smaller seed bank of the natural herba-ceous vegetation because of the previous over-grazing, resulting in reduced seed production by thenative species. The amount of litter and senescentherbage from previous seasons was obviously not aproblem for legume germination since all sites wereburnt before sowing.

Within grazing management sites, establishmentwas highest on vlei for most accessions — the highestestablishments being for the communal vlei. Thissuggests an interaction of previous grazing manage-ment and soil wetness. The soil moisture in the sandytoplands was apparently not adequate to support theinitial year establishment of most accessions.



54

Tab

le 1

. A

vera

ge le

gum

e co

unts

for

the

cont

roll

ed a

nd c

omm

unal

gra

zing

sit

es f

rom

sea

son

1 (1

999–

2000

) to

sea

son

3 (2

001–

2002

).

Con

trol

led

Com

mun

al

vlei

topl

and

vlei

topl

and

Spe

cies

Sea

son

1999

–20

0020

00–

0120

01–

0219

99–

2000

2000

–01

2001

–02

1999

–20

0020

00–

0120

01–

0219

99–

2000

2000

–01

2001

–02

Aes

chyn

omen

e fa

lcat

a cv

. Bar

goo

10.7

2.5

0.06

2.3

0.2

0.5

21.7

1.

5 0.

64.

0 4.

2 6.

2

A. f

alca

ta A

TF

219

61.

0 0

05.

0 1.

2 1.

17.

7 0.

8 0.

93.

3 2.

3 8.

2

A. f

alca

ta T

PI

8814

.3

0 0

2.7

0.2

0.5

14.7

0

0.06

0.7

0 2.

6

A. h

istr

ix 9

3638

3.0

0.5

00.

30.

3 0.

310

.3

0 0.

31.

0 0.

2 1.

3

A. h

istr

ix A

TF

977

9.0

0.5

0.5

2.3

0 0.

910

.7

2.7

0.2

2.0

0.3

2.6

A. v

illo

sa c

v. R

eid

7.0

11.2

2.

41.

3 1.

32.

87.

0 10

.7

0.8

0.3

0 18

.9

A. v

illo

sa c

v. K

rets

chm

er16

.3

3.6

0.7

4.0

3.2

3.2

12

2.2

2.9

2.3

05.

7

Cha

mae

cris

ta r

otun

difo

lia

cv. W

ynn

6.7

0.2

0.6

3.3

0.8

0.3

8.7

0.8

0.1

5.3

4.0

14.2

C. r

otun

difo

lia

9309

42.

7 0.

2 0.

061.

3 1.

01.

73.

3 0.

7 0.

43.

0 4.

5 23

.3

C. p

ilos

a C

PI

5750

33.

3 0.

2 0

3.3

0.2

0.2

4.0

0 0.

14.

7 4.

5 9.

8

Lot

onon

is b

aine

sii c

v. M

iles

5.7

5.3

2.0

0 0

4.1

1.7

0.8

6.1

0 0

12.0

Des

mod

ium

unc

inat

um c

v. S

ilve

rlea

f1.

00.

3 0.

40.

3 0.

32.

20.

30.

51.

10

0.3

19.2

Mac

ropt

iliu

m a

trop

urpu

reum

cv.

Sir

atro

1.3

0.5

00.

3 0

1.4

0.3

0.3

0.2

0.3

0.3

0.5

Styl

osan

thes

ham

ata

cv. V

eran

o22

.70.

2 0

4.7

0.2

022

.7

0 0.

062.

3 0

3.4

S. h

ippo

cam

poid

es c

v. O

xley

3.7

0 0.

38.

0 0.

50.

18

1.0

0.3

1.7

2.8

4.6

S. m

exic

ana

8748

413

.30

0.06

4.3

0.8

0.4

25

1.8

1.6

8.7

0.3

2.5

S. s

cabr

a cv

. Sec

a31

.311

.3

0.6

3.7

1.3

2.4

33.3

19

.30.

22.

7 3.

5 9.

5

S. s

cabr

a cv

. Fit

zroy

7.0

2.3

0.3

1.7

1.0

0.4

9.7

5.0

0.1

3.3

1.8

3.2

LS

D5.

081.

417.

755.

081.

417.

755.

081.

417.

755.

081.

417.

75



55

Tab

le 2

. Leg

ume

biom

ass

prod

ucti

on (

kg/h

a) in

con

trol

led

and

com

mun

al g

razi

ng s

ites

for

sea

son

2 (2

000–

2001

) an

d se

ason

3 (

2001

–200

2).

Con

trol

led

Unc

ontr

olle

d

vlei

topl

and

vlei

topl

and

Spe

cies

Sea

son

2000

–01

2001

–02

2000

–01

2001

–02

2000

–01

2001

–02

2000

–01

2001

–02

Aes

chyn

omen

e fa

lcat

a cv

. Bar

goo

4.8

0 0

0 2

0 18

0 0.

03

A. f

alca

ta A

TF

2196

0 0.

012.

8 0

0 0

47.2

0

A. f

alca

ta T

PI

880

00

0 0

0 0

0

A. h

istr

ix 9

3638

0 0

25.2

0

0 0

3.6

0

A. h

istr

ix A

TF

977

2 0

0 0

0 0

35.2

0

A. v

illo

sa c

v. R

eid

397.

2 0

111.

2 0

1191

.6

0.5

27.6

0

A. v

illo

sa c

v. K

rets

chm

er49

2 0

4 0.

07

322.

4 0.

10

0

Cha

mae

cris

ta r

otun

difo

lia

cv. W

ynn

0 0

5.2

0 6.

9 0

1700

0

C. r

otun

difo

lia

9309

429

.2

0 0

0 0

0.06

2094

.2

0.02

C. p

ilos

a0

02.

8 0.

06

0 0

517.

6 0

Lot

onon

is b

aine

sii c

v. M

iles

388.

4 0

72

0.4

98.8

0.

7 19

.2

0

Des

mod

ium

unc

inat

um c

v. S

ilve

rlea

f0

0 0

0 0

0 0

0

Mac

ropt

iliu

m a

trop

urpu

reum

cv.

Sir

atro

0 0

9.2

0.5

172

0 65

8.8

0

S. h

amat

a cv

. Ver

ano

1.2

00

0 0

0.03

0 0

S. h

ippo

cam

poid

es c

v. O

xley

0 0

0 0

0 0.

0719

2.4

0.4

S. m

exic

ana

8748

43.

7 0

0 0

0 0

47.2

0

S. s

cabr

a cv

. Sec

a13

4.4

0.01

68.8

0.

2 38

4.4

0 13

0.4

0

S. s

cabr

a cv

. Fit

zroy

2.8

0 0

0 14

.4

0 96

0

LS

D18

.13

0.23

18.1

30.

2318

.13

0.23

18.1

30.

23



56

The poor persistence of most accessions across allsites could have been due to competition from nativevegetation. This has been documented for Chamaec-rista rotundifolia and Stylosanthes accessions insemi-arid sites in Nigeria (Tarawali 1994), and forStylosanthes spp. in Mozambique (Muir and Abrao1999). According to Clatworthy and Muyotcha(1980), some legumes — especially those that havean erect growth habit — can die as a result of shading,and legumes such as the Stylosanthes species can bebadly affected. A. villosa and C. rotundifolia acces-sions increased dramatically in plant counts in thethird year on communal topland. While the reasonsare not clear, it can be postulated that while theseaccessions are slow establishers on well-drainedsemi-arid sites, they can persist under intensivegrazing and competition from native vegetation.

The biomass production was very poor for mostaccessions. This has also been the experience in otherstudies. In a 2-year study of nine accessions in threesemi-arid sites in Nigeria, C. rotundifolia accessionsyielded better in the second season, with yields of0.3–1.7 t/ha, while S. hamata generally yielded betterin the establishment year, with a yield range of 0.7–1.8 t/ha (Tarawali 1994). In Mozambique, Muir andAbrao (1999) observed poor yields of only 298 kg/hawhich declined progressively over 5 years. They con-cluded that the quantity of Stylosanthes speciesforage produced may not compensate for the cost ofestablishing forage banks, although they argued thatthis forage contribution may be significant on a rangesite with very little dry-matter production in the dryseason. The only ‘reasonable’ biomass production inthis study was for A. villosa and C. rotundifoliaaccessions and even then, there was no consistencyamong the sites. Edye et al. (1998) argue that ade-quate plant densities for high dry-matter and seedyields must occur regularly to ensure the long-termpersistence of legumes to ensure their colonisingability during favourable seasonal conditions can berealised. In this study, this argument may holdbecause of the poor establishment and persistenceresults obtained.

The legumes in this study were tested in singlelegume plots and not legume mixtures. Tarawali(1994) suggested that legumes could be considered inmixtures so as to exploit the good aspects of each spe-cies. This might give a spread of production in dif-ferent seasonal conditions. Some legume seed, e.g. L.bainesii cv. Miles, seemed to survive in the soilacross seasons only to germinate under more favour-

able conditions. This is a good attribute for anylegume to be used on these climatically variablesavannah rangelands.

The two grazing management systems tested couldhave been either too lenient or too excessive to permitsufficient legume growth. It is necessary to test inter-mediate grazing intensities for the communallymanaged rangelands with introduced forage leg-umes.

Conclusion

The establishment persistence and biomass produc-tion success of the 18 legumes being tested wereaffected by previous grazing management of the siteand long-term soil wetness. These are realistic varia-bles for communally managed rangelands, whichvary from site to site.

At this point, it would seem that on the basis ofestablishment, persistence and biomass production,the following should be further observed and evalu-ated for improving communally managed rangelandswith characteristics similar to the test sites: A. villosacv. Reid, A. villosa cv. Kretschmer, C. rotundifoliacv. 93094, C. rotundifolia cv. Wynn and C. pilosaCPI 57503.

The minimal performances by the legumes shouldnot be used to totally abandon legume evaluations oncommunally grazed rangelands because as per theargument of Muir and Abrao (1999), their contribu-tion might still be of significance on these range-lands.

The research must be continued to evaluate variousgrazing management strategies’ effects on the per-sistence and biomass production. These strategiesmust not only include grazing intensity but must alsoencompass local farmer community dynamics andcommunity institutional options for managing therangelands common property.

Acknowledgments

The authors wish to thank the Australian Centre forInternational Agricultural Research (ACIAR) forfunding this project. Additional funding was pro-vided by the University of Zimbabwe, Department ofAnimal Science; Department of Research and Spe-cialist Services, Grasslands Research Station; andZimbabwe Open University. Research partners fromthe Commonwealth Scientific and Industrial



57

Research Organisation (CSIRO) Australia, in partic-ular Drs Bruce Pengelly, Anthony Whitbread, DickJones, and Bruce Cook, and Richard Clark from theDPI, Godfrey Manyawu from Grasslands ResearchStation and Mr Mazaiwana from AgriculturalResearch and Extension (AREX), Wedza providedinvaluable technical advice throughout the projectduration and are sincerely thanked.

References

Clatworthy, J.N. and Muyotcha, M.J. 1980. Effects oflegumes sown in reverted veld or in pastures of animalproduction. Annual report of Division of Livestock andPastures, Department of Research and SpecialistServices, Zimbabwe Ministry of Agriculture, 1978–79,174–181.

Clatworthy, J.N. and Madakadze, C. 1988. Some thoughtson the collection, introduction and evaluation of pasturelegumes in Zimbabwe. In: Dzowela, B.H., ed., Africanforage plant genetic resources, evaluation of foragegermplasm and extensive livestock production systems.

Addis Ababa, Ethiopia, International Livestock Centrefor Africa, 21–36.

Edye, L.A., Hall, T.J., Clem, R.L., Graham, T.W.G.,Messer, W.B. and Rebgetz, R.H. 1998. Sward evaluationof eleven “Stylosanthes seabrana” accessions and S.scabra cv. Seca at five subtropical sites. TropicalGrasslands, 32(4), 243–251.

Majee, D. and Chikumba, L.P. 1995. Introduction andscreening of forage legumes in Zimbabwe. Harare,Zimbabwe, Department of Research and SpecialistServices (DR&SS) annual reports 1992–1993.

Muchadeyi, R. and Chakoma, I. 1995. Evaluation of foragelegumes on vleis. Harare, Zimbabwe, Department ofResearch and Specialist Services (DR&SS) annualreports 1995.

Muir, J.P. and Abrao, L. 1999. Productivity of 12Stylosanthes in semi arid Mozambique. TropicalGrasslands, 33, 40–50.

SAS 1998. SAS Users Guide: statistics. Cary, NorthCarolina, SAS institute.

Tarawali S.A. 1994. The yield and persistence of selectedforage legumes in subhumid and semi-arid west Africa.Tropical Grasslands, 28(2), 80–89.



58

Evaluation and Screening of Forage Legumes for Sustainable Integration into Crop–Livestock

Farming Systems of Wedza District

R. Nyoka,* N. Chikumba,* I Chakoma,* P Mazaiwana,† N. Mukombe† and N. Magwenzi†

Abstract

A forage-legume screening trial was carried out on 34 annual and perennial legumes during the period 1999 to2001 in the Zana and Dendenyore wards of Wedza district in Zimbabwe. Performance of legumes was based onthe assessment of establishment, resistance to pests and disease, and dry matter production. Based on researchresults and appraisal by the farmers, the legumes that appear to have potential for a variety of uses in the localfarming systems and require further evaluation include the following: Alysicarpus rugosus CPI 76978,Centrosema pascuorum Q 10050, Centrosema pascuorum cv. Cavalcade, Desmodium intortum cv. Greenleaf,Desmodium uncinatum cv. Silverleaf, Lablab purpureus cv. Endurance, Macroptilium atropurpureum cv.Siratro, Macroptilium bracteatum CPI 27404, Macroptilium gracile cv. Maldonado, Macroptilium gracileCPI 93084, Vigna lasiocarpa CPI 34436, Vigna unguiculata CPI 121688 and Macrotyloma axillare cv. Archer.

Smallholder farming systems in Zimbabwe are char-acterised by a high degree of interdependence betweencrop and livestock activities at farm level. Crops suchas maize, sorghum, finger millet and pearl millet aregrown in association with livestock rearing (mostlycattle, goats, sheep and poultry). Crop production ismost commonly on a subsistence basis and is severelylimited by, among other factors, a shortage of inputs,lack of draught power and inherent poor soil fertility.Ruminant livestock depend largely on native pasturethat is usually in short supply and of poor nutritivequality during the prolonged dry season. Crop stoversmake up the bulk of livestock feed during the dryseason. The dung from livestock is frequently used asmanure in the arable lands.

Increasing rural resettlement has led to a rapid andgeneral decline in the areas of grazing and arablelands available to each farm family. Therefore, theneed to intensify and optimise both livestock andcrop production is the major challenge being facedby smallholder farmers.

Legumes tend to have higher protein levels thangrasses and have a low natural occurrence in most ofthe natural pastures. However, forage-legume tech-nology has the potential to improve the sustainabilityand productivity of the smallholder farming systemsby providing a protein source that can boost livestockproduction (through increased availability of high-quality forages) and crop production (through bio-logical nitrogen-fixation). For the successful use offorage legumes, climatically and edaphically adaptedaccessions that produce reasonable quantities ofquality forage need to be selected. This paper reportson the performance of 34 legume accessions evalu-ated in the Wedza district of Zimbabwe.

* Grasslands Research Station, P. Bag 3701, Marondera,Zimbabwe.

† Agricultural Technical and Extension Services, PO Box 30, Wedza, Zimbabwe.



59


Farmer group description

The group of farmers involved with the trialsincluded households headed by both males andfemales. Their ages were in the range 40–65 years.Farming activities were maize-based (Table 1).

Legume accessions

Thirty-four perennial legume accessions wereplanted in November 1999, for agronomic and phe-nological evaluation in the Zana and Dendenyorewards of Wedza district during the period 1999 to2001. Table 2 lists accessions and species.

Experimental sites

The trials were conducted at two and four sites,respectively, in the Zana and Dendenyore wards ofWedza district. Wedza lies in agro-ecological regionIIb at an altitude of 1200 m. Mean annual rainfall is inthe range 600–900mm with about 80% fallingbetween November and March. Mean maximum tem-peratures of 31.1°C occur in October and mean mini-mums of 8.4°C in July. Soils in the district arepredominantly acidic (pH 4.5) deep brown sands tosandy loams of low fertility derived from granite. Thevegetation is wooded shrubland with Terminaliasericea and Burkea africana, in association withvarious Combretum and Acacia species, being promi-nent. Brachystegia spiciformis, Julbernadia globifloraand Brachystegia boehmii may occur in places. Grass-lands are dominated by species of Hyparrhenia.Further details of the farming systems common tothese areas can be found in Chigariro (2004).

Experimental design and layout

The accessions were planted during the 1999–2000 season in a randomised complete block with

seven replications, with each replicate on a differentfarm. Each plot was a 6 m long row, with 1 mbetween rows.

Land preparation, sowing and seedling establishment

Seed of the 34 accessions was scarified using sandpaper and inoculated with appropriate Rhizobiumstrains before planting. All sites were prepared to afine tilth by using ox-drawn ploughs and discing.Before discing, the row plots received a basal appli-cation of 500 kg/ha of dolomitic lime and 250 kg/haof single superphosphate. In November 1999, inocu-lated seed were drilled in rows and lightly coveredwith soil. Farmers were assisted by research andextension personnel during sowing and were respon-sible for the general upkeep of the experimental sites.The sites at Zana were kept almost weed free, whilethere was varying levels of weed competition at theother sites.

Data collection

A team comprising the farmer, researcher andextension personnel visited the plots fortnightly tomonitor the performance of the accessions. Farmerswere also encouraged to visit the plots on their ownand write comments in the notebooks issued at theinception of the project and to provide feedback toother partners during the fortnightly visits. At the endof the evaluation period, farmers were asked to rankthe legumes in terms of performance.

The establishment density was assessed at 12weeks after planting, and monitoring of vigour andpest and disease incidence during the growingseason. Establishment was rated visually on a one tofive scale (1 – poor; 2 – fair; 3 – good; 4 – very good;5 – excellent). Similarly, pest and disease incidencewere rated on a one to five scale (1 – least severe; 2 –fairly severe; 3 – severe; 4 – very severe and 5 –

Table 1. Gender of household head, main agricultural activities and ages of farmers in the forage evaluationgroup in Zana and Dendenyore wards in Wedza District.

Farmer Ward Agricultural activities Approximate age (years) Household head

ChingwaruChingwaChiwanzaMakwarimbaMapfumoMakondo

DendenyoreDendenyoreDendenyoreDendenyoreZana IIZana II

MaizeMaizeMaizeMaize and beefMaize, soya bean and paprikaMaize and paprika

50–5545

50–6045 5540

Male/femaleMaleMale/femaleMaleMale/femaleMale



60

extremely severe). In May 2001, dry matter yield wasmeasured by cutting two randomised 1 m2 quadrants5 cm above ground level from each plot. Cut materialwas oven dried at 60°C and weighed.

Statistical analysis

The general linear models (GLM) procedurewithin SAS (1994) was used to analyse data.


Plant density

Plant density at 12 weeks after planting differedsignificantly across sites (P < 0.05), with the countson a red soil site at Dendenyore and the two sandysoils at Zana being higher than on the four sandy soilat Dendenyore (data not presented). The good estab-lishment observed in Zana could be attributed to thefact that the farmers were generally more committedto the trial and kept their plots free of weeds at thiscritical stage in the growth of the legumes. Regularweeding significantly reduced competition betweenthe test species and the weeds. At other sites whereestablishment was relatively poor, little attention waspaid to the plots and the plants suffered from weedcompetition.

Considerable differences in plant density were alsoobserved among the legume species (P < 0.01).Legumes in the genera Aeschynomene, Alysicarpusand Macroptilium generally exhibited good emer-gence compared with other genera, with scoresranging from 2.70 to 3.50. Accessions with thehighest establishment rating were Aeschynomeneamericana cv. Lee, Alysicarpus rugosus CPI 76978and Macroptilium gracile cv. Maldonado (Table 2).

Forage dry-matter yields

Forage dry-matter yields were taken only in year 2of the trial, so the significant differences betweenaccessions in yield (Table 2) would be expected toreflect a strong discrimination against annual species.Dry-matter yields greater than 1500 kg/ha wereobtained from Alysicarpus rugosus CPI 51655, Alys-icarpus rugosus CPI 94489, Centrosema pascuorumCPI 65950, Desmodium intortum cv. Greenleaf,Macroptilium atropurpureum cv. Siratro, Vignaunguiculata CPI 121688, Macroptilium gracile cv.Maldonado and Macrotyloma axillare cv. Archer(Table 2).

Intermediate forage dry-matter yields in the range1000–1500 kg/ha were obtained from Alysicarpusrugosus CPI 30034, Macroptilium bracteatumCPI 55769 and Neonotonia wightii cv. Cooper.

The least amounts of forage material were har-vested from Desmodium uncinatum cv. Silverleaf,which gave an average of 237 kg/ha over the 2000–2001 season. The yields were atypical of this usuallyhigh-yielding accession. Yields were low becauseharvesting was done during the dry season (May)when leaf loss was at its peak. Generally, competitionfrom weeds prevented most species from realisingtheir optimum yielding potentials.

Most of the accessions listed above are perennialspecies, although the results for V. unguiculataCPI 121688 are of special interest. This ‘wild’ formof the species has performed very differently to theannual domesticated cowpea, indicating that thisform (probably V. unguiculata spp. dekindtiana) haspotential in these systems, as material may be able topersist for more than one year.

Disease and pest tolerance

Legume species differed significantly (P < 0.01)in their tolerance to diseases. Stem and root rot, inaddition to leaf sclerosis, were recorded in somecowpeas, particularly in cv. Gauche and the localstrains (Table 2). The incidence of stem and root rotwas particularly high between October and Marchwhen hot, wet and humid conditions prevailed. Fun-gicides were sprayed to curtail the disease but withlittle success. In view of the high cost of fungicidesand other crop chemicals, it is unlikely that thesecowpeas would be adopted for use as livestockforage, but might still be persevered with for humanfood production. Desmodium intortum cv. Greenleafwas severely affected by a disease causing numerousbrown necrotic lesions on the leaves. The conditionwas particularly serious during flowering, andresulted in heavy loss of leaf material. Legumesfrom the other genera were generally less susceptibleto fungal diseases. Species exhibiting impressivetolerance to diseases were Macroptilium bracteatumCPI 55769 and CPI 68892, Macroptilium gracileCPI 84999 and CPI 93084, and Stylosanthes hamatacv. Verano. Cheaper alternatives of controlling dis-eases by cultural methods, particularly within thegenus Vigna, need to be sought.

Three accessions of V. unguiculata (cv. Gauche,and the local strains) were the most susceptible to



61

aphids, followed by Neonotonia wightii cv. Cooperand Macroptilium gracile cv. Maldonado. Symp-toms of nematode attack were also evident in cowpeacultivars. Problems of pests were generally high atsites where weed management of the plots was poor.

An appraisal by the farmers

At the end of the evaluation period, a group discus-sion involving the researchers, extension personneland the participating farmers was held to consolidatefeedback from the farmers and map out future strate-

gies. At the end of the 2000–2001 growing season,each farmer was also invited to name their five best-performing legume accessions. The farmers’appraisals were based mainly on the visual assessmentof the herbage yield, tolerance to diseases, pests andgrazing by livestock (Table 3). The opinions offarmers tallied with observations made by researchers,but again their selections have probably discounted theperformance of annual legumes in the first year of theevaluation trial. Consequently, both the research andfarmer appraisals have to be seen as being useful in the

Table 2. Means (n = 6) of performance attributes of legume accessions planted in the Zana and Dendenyorewards of Wedza district.

Species Establishment Pest Disease Dry matter yield (kg/ha)

Aeschynomene americana CPIa 56282Aeschynomene americana CPI 93624Aeschynomene americana cv.b Glenn Aeschynomene americana cv. Lee Alysicarpus rugosus CPI 30034Alysicarpus rugosus CPI 30187Alysicarpus rugosus CPI 51655Alysicarpus rugosus CPI 76978Alysicarpus rugosus CPI 94489Centrosema pascuorum CPI 65950Centrosema pascuorum Qc 10050Centrosema pascuorum cv.CavalcadeClitoria ternatea cv. MilgarraDesmodium intortum cv. GreenleafDesmodium uncinatum cv. SilverleafLablab purpureus cv. EnduranceMacroptilium atropurpureum cv. SiratroVigna oblongifolia Q 25362Macrotyloma daltonii cv. Ex-OdziVigna unguiculata CPI 121688Macroptilium bracteatum CPI 55769Macroptilium bracteatum CPI 27404Macroptilium gracile cv.Maldonado Macroptilium gracile CPI 84999Macroptilium gracile CPI 93084Macrotyloma daltonii CPI 60303Neonotonia wightii cv. Cooper Stylosanthes hamata cv. VeranoVigna lasiocarpa CPI 34436Vigna luteola cv. DalrympleVigna unguiculata Local 2Vigna unguiculata Local 1Vigna unguiculata cv. Gauche

2.22.02.52.72.01.01.73.12.81.91.41.70.51.71.91.91.82.71.30.41.71.02.71.01.81.10.72.81.41.81.82.01.6

3.8

3.4

3.84.2

6.1

2.9

4.1

3.7

3.8

4.4

3.7

3.8

3.1

1.3

2.92.54.05.72.93.84.64.23.84.55.03.32.54.93.74.02.92.12.5

3.53.43.74.23.92.94.13.73.43.93.73.73.01.32.72.63.63.62.73.44.54.23.84.54.93.22.84.83.63.82.81.92.5

––––

116814931704

83422331807

315956647

1717237

–1911

939734

23831116

8351614

525840

–1034

–613893

–––

a Australian Commonwealth Plant Introduction number.b Commercial cultivar.c Queensland Department of Primary Industries introduction number.



62

selection of legumes for use in a two-year legumefallow rather than a single-year legume rotation.

In addition to information on legume adaptationand attractiveness to farmers, the research projectprovided other benefits. Farmers considered that theforage-legume screening program had been instru-mental in forming closer links between farmers,extension staff and researchers, and all participatingfarmers present at the evaluation meetings at the endof the experimental program reported that theircontact with extension agents had improved substan-tially since the start of the project. They indicated thatthe project opened the way for more frequent visitsby extension staff and other government officialsfrom the district, as well as other stakeholders inter-ested in rural development. The farmers also indi-cated that the project had a big impact on the waythey now perceive and manage their pastures.

Despite the feedback from farmers, however, thereremains a number of constraints to the use of foragelegumes in these farming systems. These includeshortage of seed, lack of adequate financial resources,

and ownership patterns of grazing lands. Farmersagreed to consider establishing seed-production plots.They indicated that for this to be successful, sometraining on the basics of seed production and mar-keting was necessary. The aspirations of the farmerswere incorporated into the activities of the 2001–2002season. Seed plots of the best-bet legume species wereestablished during this season by farmers, withminimum assistance from researchers and extensionpersonnel. However, the dry conditions that persistedin the area in that summer adversely affected establish-ment and subsequent seed production. Despite thedrought, farmers hoped to produce seed in the forth-coming season if conditions are favourable.

Conclusion

Evidence from the experiment has resulted in thefarmers, extension staff and researchers agreeing onthe selection of some legume species for further testingin Wedza district. The materials recommended forfurther testing under different farming systems are:

Table 3. The five top-performing legumes as appraised by the farmers.

Farmer Ward Legume species ranking

Mapfumo Zana II 1. Macroptilium atropurpureum cv. Siratro2. Macroptilium bracteatum CPI 688923. Desmodium uncinatum cv. Greenleaf4. Vigna species5. Lablab purpureus cv. Endurance

Makwarimba Dendenyore 1. Macroptilium bracteatum CPI 68892 2. Macrotyloma axillare cv. Archer3. Macroptilium bracteatum CPI 274044. Macroptilium atropurpureum cv. Siratro5. Desmodium uncinatum cv. Greenleaf

Chingwa (loams) Dendenyore 1. Vigna species2. Lablab purpureus cv. Endurance3. Alysicarpus rugosus CPI 769784. Centrosema pascuorum Q 100505. Stylosanthes hamata cv. Verano

Chingwa (red soils) Dendenyore 1. Vigna species2. Macrotyloma axillare cv. Archer3. Lablab purpureus cv. Endurance4. Alysicarpus rugosus CPI 76978 5. Vigna lasiocarpa CPI 34436

Makwarimba Dendenyore 1. Macroptilium bracteatum CPI 688922. Macrotyloma axillare cv. Archer.3. Macroptillium atropurpureum cv. Siratro4. Desmodium uncinatum cv. Silverleaf5. Desmodium intortum cv. Greenleaf



63

Aeschynomene americana cv. Glenn, Aeschynomeneamericana cv. Lee, Alysicarpus rugosus CPI 76978,Centrosema pascuorum CPI Q10050, Centrosemapascuorum cv. Cavalcade, Desmodium intortum cv.Greenleaf, Desmodium uncinatum cv. Silverleaf,Lablab purpureus cv. Endurance, Macrotyloma axil-lare cv. Archer, Macroptilium atropurpureumcv.Siratro, Macroptilium bracteatum CPI 68892,Macroptilium bracteatum CPI 27404, Macroptiliumgracile cv. Maldonado, Vigna lasiocarpa CPI 34436,Vigna unguiculata CPI 121688, and the local cow-peas. While research is still needed to assess and quan-

tify the impact of forage legumes on soil fertility andcrop and livestock production, the concept of legumeuse in these systems is of considerable interest tosmallholder farmers in Wedza district.

References

Chigariro, B. 2004. The farming systems of theZimbabwean district of Hwedza and the role and adoptionof forage legumes in small-scale crop–livestockenterprises. These proceedings.

SAS 1994. SAS Institute Inc., Cary NC, USA.



64

Identification and Development of Forage Species for Long-Term Pasture Leys for the Southern

Speargrass Region of Queensland

R.L. Clem and B.G. Cook*

Abstract

Well-managed pasture leys using species that are adapted to the climate and soils are an option for redressingpasture and soil degradation in areas of the speargrass region of southern Queensland. To identify the mostpromising pasture plants and compare them with current cultivars, some 180 legumes and 50 grasses weretested in small swards at two sites on farmers’ properties at Cinnabar near Kilkivan and Broad Creek nearDurong.

Since planting in February 2000, rainfall has been below average but most of the legumes and grassesestablished. Density of legumes has generally declined, but the grasses have improved to the extent that someare now covering 100% of the plots. Of the grasses, creeping bluegrasses (Bothriochloa insculpta cvv. Hatch,Bisset and CPI 69517) are well adapted to both sites as are digit grasses (Digitaria eriantha cv. Premier andD. milanjiana cv. Strickland and accessions such as D. eriantha ATF 605). There are also Urochloa spp. thatare equally as well adapted and leafier than U. mosambicensis cv. Nixon.

The annual legumes, lablab and Centrosema pascuorum Q 10050 were the highest yielding in the first yearat both sites and Macroptilium bracteatum cvv. Juanita and Cadarga have persisted at both sites, producingsome forage in the third season. These could be effective short-term ley legumes in cropping programs.Legumes that are strongly perennial and could be used in longer-term leys and for grazing are Stylosanthesscabra cvv. Seca and Siran and S. guianensis var. intermedia cv. Oxley. Desmanthus glandulosus CPI 90319Ais also promising at both sites but a range of Desmanthus leptophyllus, including CPI 38820, CPI 63453, CPI92806, CPI 92809, CPI 92818 and TQ 90, while promising at Cinnabar, were less successful at Broad Creek.Aeschynomene falcata ATF 2194 is a low-growing but promising long-term legume for grazing that haspersisted and produced more forage than A. falcata cv. Bargoo at Broad Creek.

The southern speargrass region of Queensland com-prises an area of some 6.2 million hectares south of24°S latitude in coastal and near-coastal areas withannual rainfall of 700 to 1000 mm (Weston et al.1981). The main land use is grazing of native andsown pastures with some 1.5 to 2 million beef cattle.Beef cattle breeding, cattle finishing and mixedfarming (cattle/cropping) are the main rural enter-

prises, with a smaller area of the more fertile soilsbeing used for annual crops. Sorghum, maize, barley,soybeans, navy beans and peanuts are the main crops,while forage crops of oats, forage sorghum andlablab are common in cattle finishing and mixed-farming enterprises.

Pasture and soil degradation is widespread, with20% of the southern black speargrass area assessed asdegraded and some 60% deteriorating and requiringsome change in management to prevent further deg-radation (Tothill and Gillies 1992).

* Queensland Department of Primary Industries, PO Box395, Gympie, Queensland 4570, Australia.



65

Cropping practice varies greatly depending oneconomic circumstances, land and soil suitability,and farmer preference, and is constantly underreview. Deeper, more-fertile soils with minimalslope might be cropped each year, whereas some ofthe shallower, less-fertile soils are cropped irregu-larly and many are being resown to pasture. A suc-cessful cropping phase can increase a farmer’s per-hectare returns, but at a cost. Each cultivation reducessoil structure and fertility, and largely destroysexisting pasture and ground cover. Well-managedpasture leys using species that are adapted to the cli-mate, soils and farmer’s need are an option forredressing these adverse consequences.

Legumes for short-term pasture leys in areassubject to frequent cultivation have been evaluated insouthern areas of Queensland (Lloyd et al. 1991;Weston et al. 2000) and northern Australia (Jones etal. 1991). These were often species that had beenrejected from evaluations that sought persistentpasture plants. Since the sown ley was only requiredto persist for 2 to 3 years, these less persistent species,which were often very productive, were more accept-able. In the southern speargrass lands, cropping isopportunistic and, on many properties, old croppinglands are being returned to pasture. For manyfarmers, unless there is a change in commodity pricesin favour of grain, future cropping is unlikely. Forthis reason, plants that can persist for the medium tolong term are more desirable than higher-producingshort-term plants. Further, it may also be important toincorporate a grass into the ley, in the interest of sup-pressing weeds, improving ground cover, enhancingorganic-matter accretion, and generally providinggreater stability to the system.

Species investigated in this project were drawnfrom the most promising of the short-term group,

pasture cultivars recommended for the area, andother productive accessions that are likely to persistin the longer term.

Our specific objectives were to assess the potentialof a range of sown legumes and grasses to fulfil a rolein longer-term pasture leys in farming systems in theregion and to encourage farmers to develop and eval-uate pasture systems based on the more promisingspecies.


Sites and sowing proceduresPaddocks with soils and land-types representative

of the district were selected on farmers’ properties atBroad Creek via Durong and Cinnabar via Kilkivan.These areas had previously been used for croppingbut had been sown to, or recolonised with, a range ofpasture grasses and legumes, and in more recentyears had been grazed with beef cattle. The mainpasture species included Heteropogon contortus,Bothriochloa bladhii, Bothriochloa decipiens,Chloris gayana, other Chloris spp. and Eragrostisspp. Soils were clay loams that varied in colour andfertility (Table 1).

Areas of about 2 ha at each site were fenced toexclude grazing stock and cultivated before sowing. InFebruary 2000, 139 legume and 50 grass accessionsand in February 2002, 45 legumes and 5 grassesincluding controls (Appendixes I and II) were sown byhand into the surface of dry soil. The exceptions werethe larger-seeded Arachis and Lablab spp., whichwere sown into furrows and covered. All legumeswere inoculated with appropriate strains of eitherRhizobium or Bradyrhizobium spp. and seed wasmixed with a small quantity of dampened sawdust to

Table 1. Location, rainfall and some soil characteristics of the experimental sites.

Datum Broad Creek Cinnabar

Latitude and longitude 26°27'S, 151°27'E 26°07'S, 152°09'E

Average annual rainfall (mm) 780 880

Soils (0–10 cm) Brown sandy clay loam Dark clay loam

pH (1:5 H2O) 6.3 6.6

Available P (bicarbonate) (ppm) 18 36

Exchangeable K (ppm) 0.73 1.1

Available S (ppm) 6.8 7.2

Organic C (%) 1.5 2.6



66

aid spreading. At the Cinnabar site, plots were rolledusing a rubber tyred roller to firm the coarse seedbedand improve soil–seed contact. At Broad Creek, thelighter textured soil surface was loose and dusty and nopost-sowing treatment was undertaken.

Plots were 5 ¥ 2m with a 1 m border between therows to allow access and demarcation. To make rec-ognition of plots easier, the grasses and legumes wereallocated rather than randomised so that accessionsof the same genus were not sown in adjoining plots.Two replicates were sown.

In the 2002 planting, as well as the small plots,some 20 accessions, including commercial cultivars,were sown in larger (100 m2) plots. At both sites, thelegume plots in this series were oversown with digitgrass (Digitaria eriantha cv. Premier) at 2 kg/ha.

Management and sampling procedure

2000 plantingEstablishment was determined by counting

emerged seedlings in two quadrats of 0.25 m2/plot atCinnabar and Broad Creek in March 2000. Plot yield,composition, frequency and vigour of the sownaccessions was measured in May 2000. Plot yieldwas assessed by ranking all plots on a scale of 1 to 5and then calculating yields (kg/ha) from regressionequations derived by cutting, drying and weighing 10selected plots across the range of yields and plottingthese against the rated yields. Botanical compositionwas determined by estimating the percentage of sownspecies and other species in each plot and calculatingthe yield of the sown species.

Composition and vigour were measured inDecember 2000 and March 2002. In legume plots,density was measured by counting old plants and

seedlings (two categories) and, in the grass plots, byestimating the ground cover (0 to 100%). Yield, com-position and vigour were measured in March 2002 asfor the sampling in May 2000. No yield measure-ments were made in autumn 2001 because of the verydry conditions.

Grazing stock had access to the experimental areafollowing sampling in May 2000 to October 2000,from December 2000 to February 2001, and againfrom May 2002 to September 2002.

Sodium molybdate was sprayed on foliage inNovember 2000 at a rate supplying 100 g/ha ofmolybdenum.

2002 plantingDensity of established plants was measured in May

by counting emerged seedlings and vigour was ratedon a scale of 1 to 3.

Results

Seasonal conditions

Rainfall at the two sites for 2000, 2001 and 2002(Table 2) was below average, with some extended dryperiods. For the 2000 planting, subsoil moisture accu-mulated over the preparatory cultivation phase and aneffective rainfall event in February resulted in goodestablishment and early pasture growth. However,establishment from the 2002 planting was poor becauseof very low rainfall following planting in late February.

Establishment and density

2000 plantingAll accessions established at both sites, although

initial plant densities were higher at Cinnabar, where

Table 2. Rainfall (mm) at Cinnabar and Broad Creek for 2000 and 2001 and long-term averages for Kilkivan andDurong.

Location Year Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Total

Cinnabar 200020012002

514835

7299

114

376929

41340

492321

443

37

134011

76

16

015

14164

62217

11254

629672

Kilkivan long-term average

134 122 97 58 49 47 44 34 38 68 73 117 881

Broad Creek 200020012002

128370

4414799

03280

9404

201420

210

40

5290

083

015

8784

52172

9770

463648

Durong long-term average

86 83 64 40 36 33 35 27 30 58 70 102 664



67

rainfall was marginally better. Mean plant densitieswere, for legumes, 52 plants/m2 at Cinnabar and 34plants/m2 at Broad Creek and, for grasses, 270 plants/m2 at Cinnabar and 116 plants/m2 at Broad Creek(Table 3). In March 2002, of the 139 legumesplanted, 111 were still present at Cinnabar but only79 at Broad Creek.

Legume density declined markedly betweenplanting and March 2002 at both sites, with dry con-ditions over late winter and spring, but weeds such asblack pigweed (Trianthema portulacastrum), Sidaspp., a range of chenopods and summer-growinggrasses also depleted soil moisture and affected allbut the fastest-growing legumes. Plant density ofmost accessions was adequate to allow comparisonsand some accessions had moderate to high densities.There was also large variation between accessions

within each genus, which more probably reflect dif-ferences in adaptation. At Cinnabar, there were, inall, 36 accessions with plant densities greater than 7plants/m2. They were Aeschynomene falcata (3accessions), Aeschynomene histrix (2), Arachis par-aguariensis (2), Desmanthus spp. (18), Macroptiliumbracteatum cv. Juanita and Stylosanthes spp. (10).Plant densities were lower at Broad Creek, with onlyArachis paraguariensis, Desmanthus virgatus CPI37538, Stylosanthes scabra CPI 110116 and cvv.Seca and Siran, S. guianensis var. intermedia cv.Oxley (fine-stem stylo) and S. seabrana cvv. Primarand Unica having more than 7 plants/m2.

In contrast to the legumes, all the grasses grew wellat both sites, competing better with weeds andresponding to nitrogen mineralised by cultivation. ByMarch 2002, the estimated cover of all grasses except

Table 3. The number of accessions from each genus that established (Mar 2000) and the number surviving (Mar2002) at Cinnabar and Broad Creek and the plant density (mean for each genus) in March 2000 andMarch 2002.

Cinnabar Broad Creek

Accessionssown

AccessionsMar 2002

Density (plants/m2) Accessions Density (plants/m2)

Genus Mar 2000 Mar 2002 Mar 2002 Mar 2000 Mar 2002

LegumesAeschynomeneAlysicarpusArachisCentrosemaClitoriaDesmanthusLablabMacroptiliumMedicagoNeonotoniaStylosanthesVigna

166431

813418

111

50401

800415

110

96641577

9424356

101187542

5060150410

110

60200

540412

100

40173459

130274846117516

105001031070

Mean of 138 accessions 52 2.8 34 1.5

GrassesBothriochloaCenchrusDichanthiumDigitariaPanicumPaspalumUrochloa

863

1351

14

863

1351

14

220189

92257319404163

86a

605873344068

863

1351

14

173123

79155

59139

66

87a

778480707076

Mean of 48 accessions 270 60 116 78a Figures for grasses in March 2002 are estimates of cover.



68

Panicum and Paspalum spp. had increased at Cin-nabar and Broad Creek. At Cinnabar, cover for somegrasses was 100%.

2002 plantingOf the 48 accessions planted in small plots at Cin-

nabar, all but 3 established, although a further 23 hadpopulations of less than 2 plants/m2. Best establish-ment was for Aeschymonene americana ATF 1086,Desmanthus sp. ATF 3930, Macroptilium atropur-pureum cv. Aztec, Stylosanthes guianensis var. inter-media cv. Oxley and Stylosanthes seabrana ATF2523 and ATF 2539. All grasses established at den-sities that should allow some evaluation. At BroadCreek, 16 accessions did not establish and only 4 leg-umes, Aeschynomene americana ATF 1086, Clitoriaternatea cv. Milgarra and CPI 47187 and Stylosan-thes seabrana ATF 2539 had densities greater than 2plants/m2. Except for Panicum coloratum cv. Bam-batsi, all grasses established poorly.

In the larger plots at Cinnabar, all accessions estab-lished but some had low populations. Legumes with

the highest establishment were Aeschymonenefalcata ATF 2914 and Bargoo, Desmanthus glandu-losus CPI 90319A, Desmanthus leptophyllus CPI38820, Macroptilium bracteatum cvv. Cardaga andJuanita, Stylosanthes guianensis var. intermedia cv.Oxley and S. seabreana cv. Unica. At Broad Creek,most accessions failed to establish and the best hadpopulations of less than 5 plants/m2.

Yield and other attributes

Legumes — 2000 plantingAnnual legumes such as Lablab and some acces-

sions of Aeschynomene, Alysicarpus, Centrosemaand Macroptilium produced higher yields in the firstseason than the perennial legumes which were gener-ally slower growing at both sites (Tables 4 and 5).

Production was higher at Cinnabar than at BroadCreek. At Cinnabar, Lablab purpureus cv. Rongaiyielded over 4000 kg/ha and some accessions of Cen-trosema pascuorum over 3000 kg/ha. Macroptilium

Table 4. Forage yields measured in May 2000 (2000), December 2000 (2001) and March 2002 (2002) of thebetter adapted legume cultivars and accessions at Cinnabar (ranked by yields for 2002).

Genus

Species

Acc/Cvv Yield (kg/ha)

2000 2001 2002

DesmanthusStylosanthesDesmanthusDesmanthusDesmanthusDesmanthusDesmanthusDesmanthusStylosanthesDesmanthusDesmanthusDesmanthusDesmanthusDesmanthusDesmanthusDesmanthusDesmanthusDesmanthusDesmanthusMacroptiliumLablabLablabCentrosemaLablab

leptophyllusscabrasp.glandulosusleptophyllusvirgatusvirgatusleptophyllusguianensis var. intermedialeptophyllusleptophyllus leptophyllusleptophyllusleptophyllusleptophyllusleptophyllusleptophyllusbicornutusvirgatusbracteatumpurpureuspurpureuspascuorumpurpureus

92809Seca

AC 1190319ATQ 90914919132663453Oxley

Bayamo3835192806928183882087876878609272290906

ATF 3009JuanitaRongai

HighworthQ 10050

Endurance

136320452045136320451818159015901818181818181818204518182045159024992272204524994545386336363409

21318138083

23339

287130125273100139170157304143172167838869590

59

6560552950613686324931243061303029672674266225612374234322371843129910621062687



69

bracteatum cv. Juanita persisted and produced a yieldof over 600 kg/ha in year 3 but the other annual andshort-term legumes which yielded well in the firstyear did not survive or regenerate (Table 4).

In the second season, legume yields were lowwhen measured in December 2000 because of com-petition with weeds. Forage yields in the thirdsummer were high, with Stylosanthes scabra cv.Seca and S. guianensis var. intermedia cv. Oxley pro-ducing 3000–5000 kg/ha. A range of Desmanthusspp. grew well, with several accessions performingbetter than available cultivars (Table 4).

Desmanthus leptophyllus CPI 92809, TQ 90 andCPI 63453 are strong leafy types that grew 0.8–1.0 mtall and outyielded D. leptophyllus cv. Bayamo. D.leptophyllus CPI 38351 is a tall, woody type. Otheraccessions such as D. leptophyllus CPI 92722,although not as high yielding, were leafy and morecompact and should be well suited to grazing. Des-manthus glandulosus CPI 90319A and Desmanthusbicornutus CPI 90906 are tall, ‘stemmy’ types whileDesmanthus virgatus was variable from D. virgatusCPI 91491 and 91326, which grew to 0.8 m, to themore compact ATF 3009, which also looks a goodgrazing type. Many of the lower-yielding Desman-thus spp. were small and unthrifty plants, suggesting

that rhizobial associations were not fully effective.Aeschynomene falcata ATF 2194 was the best of thejoint vetch accessions producing over 1000 kg/ha butforming a dense sward. Arachis paraguariensis CPI91419 with a yield of 500 kg/ha was the best of theforage peanuts and is now starting to spread from thesown rows. These are accessions with some potentialbut are low-yielding, especially in establishmentyears. Only accessions with higher yields are shownin Table 4.

At Broad Creek, maximum yields in the first yearwere lower than at Cinnabar, ranging from 1000 to2000 kg/ha, but the order of ranking was similar. Withthe exception of Macroptilium bracteatum cv. Juanita,legumes that produced high yields in the first seasonfailed to persist or regenerate. In the second season,Stylosanthes scabra (cvv. Seca and Siran and CPI110116) and Stylosanthes seabrana (cvv. Primar andUnica) were the highest yielding. These and S. guian-ensis var. intermedia cv. Oxley also produced highyields in the third season (Table 5).

Desmanthus appeared to be less well adapted tothis site than to the Cinnabar site, but some acces-sions, such as Desmanthus leptophyllus CPI 38820and 63454, gave high yields at both sites. Similarly,Arachis paraguariensis CPI 91419 was spreading at

Table 5. Forage yields measured in May 2000 (2000), December 2000 (2001) and March 2002 (2002) of thebetter adapted legume cultivars and accessions at Broad Creek (ranked by 2002 yields).

Genus Species Acc/Vvv Yield (kg/ha)

2000 2001 2002

StylosanthesStylosanthesStylosanthesStylosanthesMacroptiliumStylosanthesDesmanthusStylosanthesMacroptiliumDesmanthusDesmanthusDesmanthusArachisDesmanthusAeschynomeneLablabCentrosemaLablabLablab

scabrascabrascabraguianensis var. intermedia atropurpureumseabranaleptophyllusseabranabracteatumleptophyllusglandulosusleptophyllusparaguariensispernambucanusfalcatapurpureuspascuorumpurpureuspurpureus

SecaSiran

110116OxleyAztecPrimar38820UnicaJuanita63453

90319ABayamo9141983566

ATF 2196Rongai

Q 10050EnduranceHighworth

7141071892

109110701249178510712142714892

1607892

124919642142196419641964

126555230221

1922449589

225117723654522122140

986618

691

7498508239573041245823321541137412501250833750666533108



70

this site and Aeschynomene falcata ATF 2194 grewinto a dense sward early in the summer but wasadversely affected later by dry weather.

Grasses — 2000 plantingMany of the available cultivars of grasses are pro-

ductive and well adapted to grazing in permanent andley pasture systems. Bothriochloa insculpta cvv.Hatch and Bisset were high yielding at both sites andin all years. After the third summer, they were almostpure swards (Table 6). Similarly, some of the Uro-chloa spp. were high yielding and formed denseswards. An objective of this study, however, was toidentify grasses that would be more compatible withlegumes in medium to longer-term leys in the sub-tropical areas. To this end, some of the other, less-productive grasses in Table 6 are of interest.

Bothriochloa insculpta CPI 69517 is very similarto the cultivar Bisset but is earlier flowering by about2 weeks. This could be an advantage in central andsouthern areas where Bisset is often frosted beforeseed matures. Earlier flowering should mean morereliable seed production, which could improve thesupply of seed and reduce its cost to farmers. Alsobeing from a drier environment at Bulawayo, thisaccession could extend the range of B. insculpta.

Digitaria and Urochloa varieties are also good col-onising and productive grasses. The digit grasses areknown to be very palatable, which could be anadvantage where higher animal growth rates aresought. More palatable grasses could also be helpfulin managing to manipulate the composition of thepasture. Several accessions (including D. erianthaATF 2109 and ATF 605 and D. milanjiana CPI59828 and CPI 59777) could be used as alternativesto the cultivars Premier and Strickland. Urochloamosambicensis CPI 46876 and 60127 are leafiertypes than Nixon but are likely to have similarweedy characteristics (heavy seeding, good colo-nising ability and lower palatability when mature)that are not favoured by some farmers. Dichanthiumannulatum CPI 50819 and 84148.2 and Panicumcoloratum ATF 714 have been less aggressive andlower yielding but they are very leafy and maycombine well with legumes in leys.

Discussion

While the observations and measurements to date canonly be regarded as preliminary, they indicate thatother legumes are unlikely to be more productivethan Lablab purpureus in the first year. Centrosema

Table 6. The percentage of plots covered by and yields of well-adapted cultivars and accessions of grasses atCinnabar and Broad Creek in March 2002

Genus Species Cinnabar Broad Creek

Acc/Cvv % (kg/ha) % kg/ha

BothriochloaBothriochloaBothriochloaBothriochloaDichanthiumDichanthiumDigitariaDigitariaDigitariaDigitariaDigitariaDigitariaDigitariaPanicum Panicum UrochloaUrochloaUrochloaUrochloa

bladhiiinsculptainsculptainsculptaannulatumannulatumerianthaerianthaerianthamilanjianamilanjianamilanjiananatalensiscoloratumcoloratummosambicensismosambicensismosambicensisstolonifera

SwannHatchBisset69517

84148.250819

ATF 2109ATF 605Premier59828

Strickland5976159752

BambatsiATF 714

46876Nixon6012747173

8595

97.592.5655080709085

82.582.5955555957080

87.5

4248949754974591456136243435293633734185412335604185337428117185493648734966

9092.510097.5

not sown45

82.592.592.597.597.540904040

97.582.59598

59151003999988956

not sown37497498845669147289808126664498266634998956766571648123



71

pascuorum CPI 65950 and Q 10050 appear betteradapted to this environment than the cultivar Caval-cade and some Alysicarpus rugosus could be consid-ered. However, for these to be serious contenders,they would need to regenerate with some reliabilityfrom seed in most years. This has not occurred overthe last 3 years, but rainfall has been lower thanaverage. Burgundy bean (M. bracteatum), presentlybeing developed for use in leys in cropping systems(Pengelly and Conway 2000) has again demonstratedit can produce high yields in the first season andpersist at least into the second and third season.

Some of the more promising perennial legumes,although they have not been widely tested in thisregion, merit further consideration. The larger plotsplanted in 2002 at these sites will provide someopportunity for this, although the establishment wasless than optimum. It is important now to confirm tofarmers that these legumes can contribute better-quality leguminous forage to the grass pasturesystems and show how they can benefit subsequentcrops. Seed production of a range of legumes to allowtesting in larger plots should be a priority — theseinclude Aeschynomene falcata ATF 2194, Desman-thus glandulosus CPI 90319A, Desmanthus bicor-nutus CPI 90906, Desmanthus leptophyllus CPI92809, 38820 and 63453, Desmanthus virgatus CPI91326, 91491 and ATF 3009, and Desmanthus sp.AC 11.

A range of well-adapted and productive grass cul-tivars is available and farmers recognise their value.The main issue now is to encourage the use of themost appropriate grasses and better define their man-agement requirements to meet specific needs. Theseneeds are likely to vary widely from minimum inputsin extensive grazing to very complex requirementsfor more intensive livestock/cropping systems.

There has been ongoing interaction with farmerssince the project commenced. Researchers and exten-sion officers have met with farmer groups on severaloccasions to canvass their views and to determinepriority areas. These farmers have always been sup-portive and are now using pasture information fromthis project in decision-making on issues that affecttheir properties and enterprises.

Farmers have been involved with a range of activ-ities including meetings, field days, site selection andin compilation of the list of species entries, throughrecommending local standards and describing thetypes of plants required for their various production

systems, as well as preparation for planting and stockmanagement. Farmers at the Broad Creek site ini-tially indicated they do not want accessions ofChamaecrista rotundifolia and Urochloa panicoidesincluded in the evaluation. In the case of Wynn cassia(C. rotundifolia cv. Wynn) however, some farmers,through a better understanding of the role of legumes,now believe it can benefit their enterprises and aremore comfortable to use it, knowing they can manageit to their advantage.

The cooperator at Broad Creek has also planted asmall area (20 ha) of Leucaena leucocephala cv. Tar-ramba and is now planting Premier digit grass, agrass he was not familiar with before the project com-menced, between the rows. Adjoining farmers areusing larger areas of lablab in cattle finishing pro-grams and others are considering lablab and leucaenaas alternatives to planting oats. Fine-stem stylo is nat-uralised over areas on some properties and its value isbeing increasingly recognised. Other farmers in thedistrict have planted areas of Stylosanthes seabranaon the basis of visits to the project sites. These out-comes confirm that these farmers now have a betterunderstanding of the current and future role oflegumes and grasses in their farming and grazingenterprises.

Acknowledgments

We are thankful to farmers at Broad Creek — DanMoloney (for providing land and resources), BobBlack, Tom Hoare and Hugh Campbell, and at Cin-nabar — Gordon McGill (for providing land andresources), Gerry Hewson, Paul Stumm, RussellSmyth, Ian Fitzgerald and Sidney Innes for theirsupport and interest. R. Tyler, Brian PasturesResearch Station, Gayndah and A.J. Greenup, BjelkePetersen Research Station, Kingaroy helped withextension activities and A.G. Salmon provided tech-nical support. The Australian Centre for InternationalAgricultural Research provided financial support.

References

Jones, R.K., Dalgleish, N.P., Dimes, J.P. and McCown, R.L.1991. Ley pastures in crop-livestock systems in the semi-arid tropics. Tropical Grasslands, 25, 189–196.

Lloyd, D.L., Smith, K.P., Clarkson, N.M., Weston E.J. andJohnson, B. 1991. Ley pastures in the subtropics. TropicalGrasslands, 25, 181–188.



72

Pengelly, B.C. and Conway, M.J. 2000. Pastures oncropping soils: which tropical pasture legume to use.Tropical Grasslands, 34, 162–168.

Tothill, J.C. and Gillies, C. 1992. The pasture lands ofnorthern Australia — their condition, productivity andsustainability. Occasional Publication No. 5. TropicalGrassland Society of Australia.

Weston, E.J., Doughton, J.A., Dalal, R.C., Strong, W.M.,Thomas, G.A., Lehane, K.J., Cooper, J.C., King, A.J. and

Holmes, C.J. 2000. Pastures on cropping soils: managinglong-term fertility of cropping lands with ley pastures insouthern Queensland. Tropical Grasslands, 34, 169–176.

Weston, E.J., Harbison, J., Leslie, J.K., Rosenthal, K.M. andMayer, R.J. 1981. Assessment of the agricultural andpastoral potential of Queensland. Agriculture BranchTechnical Report No. 27. Brisbane, QueenslandDepartment of Primary Industries, 1–195.



73

Appendix I

List of legume species planted at Cinnabar (C) and Broad Creek (BC) in 2000 and 2002. A rating of the performance of each accession/cultivar is given as

0 = did not establish, 1 = established but poor (no promise), 2 = fair, 3 = good, 4 = very good (promising).

Species Sites

C2000 C2002 BC2000 BC2002

Aeschynomene americana CPI 56282 1 1

Aeschynomene americana CPI 93624 1 1

Aeschynomene americana ATF 1086 3 1

Aeschynomene americana cv. Glenn 1 1

Aeschynomene americana cv. Lee 1 1

Aeschynomene falcata ATF 2194 2 3 2 2

Aeschynomene falcata ATF 2196 1 2

Aeschynomene falcata cv. Bargoo 1 3 2 2

Aeschynomene hystrix CPI 93599 1 1



Aeschynomene hystrix ATF 977 1 2

Aeschynomene hystrix ATF 2191 2 2

Aeschynomene villosa CPI 37235 1 1

Aeschynomene villosa TPI 88 1 1

Aeschynomene villosa cv. Kretchmer 1 1

Aeschynomene villosa cv. Reid 1 1

Alysicarpus monilifer CPI 52359 1 1

Alysicarpus rugosus CPI 30034 2 1





Arachis paraguariensis CPI 91419 2 2 2 0

Arachis paraguariensis Q 25237 2 2

Arachis stenosperma CPI 91420 1 1

Arachis stenosperma ATF 377 1 1

Centrosema pascuorum CPI 65950 1 2

Centrosema pascuorum Q 10050 2 2

Centrosema pascuorum cv. Cavalcade 1 1

Clitoria ternatea CPI 28810 1 1






74




Clitoria ternatea CPI 51380B 1 1


Clitoria ternatea cv. Milgarra 1 1 1 2

Desmanthus acuminatus CPI 78379 1 1

Desmanthus acuminatus CPI 78380 1 1

Desmanthus acuminatus ATF 2986 2 1

Desmanthus bicornutus CPI 81336 1 2





Desmanthus bicornutus CPI 91162 3 4 1 0

Desmanthus glandulosus CPI 90319A 4 4 3 3

Desmanthus leptophyllus CPI 38351 3 2 2 0


Desmanthus leptophyllus CPI 55719 4 1









Desmanthus leptophyllus TQ 90 4 2

Desmanthus leptophyllus cv. Bayamo 4 2 3 2

Desmanthus paspalaceus CPI 78385 2 1

Desmanthus paspalaceus CPI 83570 2 1

Desmanthus paspalaceus ATF 3015 1 1

Desmanthus paspalaceus ATF 3023 2 1

Desmanthus pernambucanus CPI 21825 1 1







Species Sites

C2000 C2002 BC2000 BC2002



75




Desmanthus pubescens CPI 92804 2 1




Desmanthus pubescens cv. Uman 2 1

Desmanthus tatuhyensis CPI 25840 1 1



Desmanthus tatuhyensis ATF 2240 2 1



Desmanthus virgatus CPI 67643 3 1





Desmanthus virgatus CPI 78372 2 2 2 0












Desmanthus virgatus CPI 91326 3 1 2 2





Desmanthus virgatus ATF 2250 1 1




Species Sites

C2000 C2002 BC2000 BC2002



76

Desmanthus virgatus Q 9153 1 2 2 0

Desmanthus virgatus cv. Marc 2 2

Desmanthus sp. CPI 121840 1 1

Desmanthus sp. CQ 3650 1 2

Desmanthus sp. AC 10 4 1

Desmanthus sp. AC 11 4 1

Desmanthus sp. ATF 3785 2 1

Desmanthus sp. ATF 3786 2 1 1 0

















Desmanthus comp cv. Jaribu 1

Lablab purpureus cv. Endurance 4 4

Lablab purpureus cv. Highworth 4 4

Lablab purpureus cv. Rongai 4 4

Macroptilium atropurpureum cv. Aztec 3 3 3 2

Macroptilium atropurpureum CPI 84989 3 3

Macroptilium bracteatum CPI 55750 3 3

Macroptilium bracteatum CPI 55758 3 3

Macroptilium bracteatum cv. Cadarga 3 3 3 3

Macroptilium bracteatum cv. Juanita 3 3

Medicago sativa cv. Trifecta 1 3

Neonotonia wightii CPI 111766 1 1





Species Sites

C2000 C2002 BC2000 BC2002



77




Neonotonia wightii cv. Cooper 1 2

Stylosanthes guianensis var. intermedia ATF 3067 3 2

Stylosanthes guianensis var. intermedia ATF 3070 3 3 2 0

Stylosanthes guianensis var. intermedia ATF 3071 1 1

Stylosanthes guianensis var. intermedia cv. Oxley 3 2 3 2

Stylosanthes mexicana CPI 87469 2 1



Stylosanthes scabra CPI 110116 3 3

Stylosanthes scabra ATF 3076 1 2

Stylosanthes scabra ATF 3077 2 0

Stylosanthes scabra cv. Seca 4 3 4 2

Stylosanthes scabra cv. Siran 4 4

Stylosanthes seabrana ATF 2523 1 2

Stylosanthes seabrana ATF 2539 2 2

Stylosanthes seabrana cv. Primar 3 3

Stylosanthes seabrana cv. Unica 3 2 3 2

Vigna oblongifolia Q 25362 1 1

Vigna oblongifolia CPI 60433 1 1

Vigna trilobata CPI 13671 1 1

Vigna unguiculata CPI 121688 1 0

Species Sites

C2000 C2002 BC2000 BC2002



78

Species Sites

C2000 C2002 BC2000 BC2002

Bothriochloa bladhii CPI 104802A 2 1

Bothriochloa bladhii cv. Swann 2 2

Bothriochloa insculpta CPI 69517 4 4

Bothriochloa insculpta CPI 106671 2 2

Bothriochloa insculpta cv. Bisset 4 4

Bothriochloa insculpta cv. Hatch 4 4

Bothriochloa pertusa cv. Keppel 2 1

Bothriochloa pertusa cv. Medway 2 1

Cenchrus ciliaris CPI 71914 2 1

Cenchrus ciliaris CPI 73393 2 1

Cenchrus ciliaris cv. American 3 2

Cenchrus ciliaris cv. Bella 3 1

Cenchrus ciliaris cv. Gayndah 3 2

Cenchrus ciliaris cv. Viva 2 3

Chloris gayana ATF 3964 2

Dichanthium annulatum CPI 50819 2 1

Dichanthium annulatum CPI 84148.2 3 ns

Dichanthium aristatum cv. Floren 2 1

Digitaria eriantha CPI 125659 2

Digitaria eriantha CPI 125659A 2

Digitaria eriantha CPI 125659B 2

Digitaria eriantha ATF 605 3 2 4

Digitaria eriantha ATF 616 2 3



Digitaria eriantha cv. Premier 3 3 4 3

Digitaria milanjiana CPI 59755 3 2





Digitaria milanjiana cv. Jarra 2 1

Digitaria milanjiana cv. Strickland 4 3

Digitaria natalensis CPI 59752 3 1

Panicum coloratum CPI 16796 1

Appendix II

List of grass species planted at Cinnabar (C) and Broad Creek (BC) in 2000 and 2002. A rating of the performance of each accession/cultivar is given as

0 = did not establish, 1 = established but poor (no promise), 2 = fair, 3 = good, 4 = very good (promising)



79

Panicum coloratum ATF 714 2 1

Panicum coloratum ATF 3957 1

Panicum coloratum cv. Bambatsi 2 1 1

Panicum maximum cv. Gatton 2 3

Panicum maximum cv. Petrie 3 3

Panicum sp. aff. infestum cv. C1 1 1

Paspalum simplex ATF 3128 1 1

Urochloa mosambicensis CPI46876 4 4

Urochloa mosambicensis CPI 47167 4 4




Urochloa mosambicensis cv. Nixon 4 3

Urochloa mosambicensis cv. Saraji 2 3

Urochloa oligotricha CPI 47122 3 1




Urochloa oligotricha CPI 73436 2 ns

Urochloa stolonifera CPI 47173 4 4

Urochloa stolonifera CPI 47178 3 2

Species Sites

C2000 C2002 BC2000 BC2002



Farming Systems Research

81



82

The Role of Dual Purpose and Forage Legumes to Improve Soil Fertility and Provide High Quality

Forage: Options in a Maize-based System in Zimbabwe

O. Jiri,* B.V. Maasdorp,* E. Temba* and A.M. Whitbread†

Abstract

Mucuna pruriens and Lablab purpureus have emerged as successful forage or green-manure legumes for usein the smallholder crop–livestock systems of Zimbabwe. Two experiments were established in the 1999–2000wet season at on-farm sites within a mixed livestock–cropping farming system in Wedza district (averagerainfall 650–900 mm) located on acidic (pH < 5) and infertile sandy and sandy-loam soils. Two additional siteswere located on more-fertile clay soils.

Experiment 1 showed that improved fallows of mucuna grown for 19 weeks produced between 4.7 and 11.2t/ha DM and generally doubled that produced by cowpea or a first-year Macrotyloma axillare cv. Archer–Chloris gayana perennial pasture. Weedy fallow treatments, which represent typical farmer practice, produced3.3–6.3 t/ha DM. After harvest of the mucuna system, the biomass was either removed as hay or ploughed inas green manure. A maize crop was then grown on these treatments and on the weed-fallow treatments in thefollowing 2000–2001 wet season. On sandy sites where no P fertiliser was applied to the previous mucunaphase, a maize grain yield of 2.3 t/ha was achieved following the mucuna green-manure system; this was 64%higher than the maize yield following the weedy fallow and 100% higher than the maize yield following themucuna ‘removed’ hay system. Averaged across the seven sites, grain yield declined from 2810 to 1680 kg/ha when mucuna residues were removed as hay. The fertilised (93 kg/ha N, 18 kg/ha P) continuous maizesystem produced 2920 and 4460 kg/ha on the sandy and clay sites, respectively.

Experiment 2 investigated the maize-grain response to the application of 0, 30, 60, 90 and 120 kg/ha offertiliser N, and compared it with maize production following a one-year weed fallow or forage lablab. Resultsshowed that, by integrating forage lablab into arable fallow land, subsequent maize grain yield could be from8 to 57% higher than maize following a weedy fallow. Farmers could rotate forage lablab and maize andbenefit from large amounts of high-quality fodder from the lablab above-ground biomass and also gain soilfertility amelioration from the residual effect of the lablab below-ground biomass.

* Crop Science Department, University of Zimbabwe, POBox MP 167, Harare, Zimbabwe.

† CSIRO Sustainable Ecosystems, 306 Carmody Road, St Lucia, Queensland 4067, Australia.



83

For improved fallow management to be adopted bycrop–livestock farmers, the practice must enhanceboth dry-season feed supply and soil fertility for sub-sequent cropping. An improvement of weed fallowmanagement systems with sown legumes has thepotential to enhance the restoration of soil fertilitythrough the fixation of atmospheric N2, and/or throughan improvement in soil-physical properties (Wilson1988). Suitable legumes also have the potential to alle-viate feed constraints, especially for cattle and duringthe dry season, through their higher nutritive valuecompared with natural fallow (Minson 1984).

Mucuna (Mucuna pruriens var utilis) is a vigorous,twining annual legume whose primary roles are soil-fertility maintenance, soil protection and weed sup-pression (Carsky et al. 1998). The biomass can alsobe used as fodder, and the seed as feed or food (Duke(1981) and Olaboro (1993), quoted by Buckles et al.(1998)). Dry-matter production in mucuna, as withall other crops, is affected by soil fertility. Dry matterand ground cover will also depend on the occurrenceof drought or waterlogging (Buckles et al. 1998).There is thus a need to evaluate mucuna dry-matterproduction under the poor soil-fertility, sub-humidconditions that prevail in the smallholder farmingsector of Zimbabwe, where it could have potential inalleviating the problems of poor soil fertility andinadequate animal feed.

Lablab (Lablab purpureus) is also a twiningannual legume capable of producing large quantitiesof biomass. It makes excellent hay if harvested at theright time when most of the leaf is still intact. Itscrude protein content is approximately 12% (Mucha-deyi 1998). Lablab tolerates drought, a common phe-nomenon in most cropping environments inZimbabwe. Like mucuna, it is large-seeded, henceeasy to handle and establish, and its deep and exten-sive root system contributes to soil organic mattercontent when decomposed and improves aerationand soil structure. As with mucuna, it is an excellentnitrogen fixer, hence has the potential to maintain andimprove available soil nitrogen (Ayoub (1986),quoted in Haque et al. (1986)). It would appear,therefore, that lablab grown for forage on arablelands also has the potential to enhance soil fertilitywhile at the same time supplying large quantities ofgood-quality fodder.

Two, on-farm, multi-site experiments that wereconducted in the Wedza district of Zimbabwe aredescribed in this paper. Experiment 1, drawn fromthe MPhil thesis of Jiri (2003), was established to

compare the relative benefits of a weedy fallow and asystem where mucuna was grown and either removedfor hay (for animal feeding) or ploughed in as a greenmanure, on the yield of a subsequent maize-croppingphase. Comparisons were also made with continuousfertilised maize. The response to P and lime applica-tion by mucuna, and to P application by cowpea anda perennial grass–legume system, were also investi-gated. Experiment 2, based on the studies of Temba(2001), investigated the effect of a lablab or weedfallow, with the biomass used for forage, on thegrowth of maize, as compared with the application ofinorganic fertiliser N.

Methods and Materials

Both experiments were conducted in the smallholderfarming community of Wedza (18°41’S latitude,31°42’E longitude; 1400 masl), which lies in NaturalRegions IIb and III (Surveyor-General 1984) and islocated 150 km southeast of Harare in Zimbabwe.Mean annual rainfall ranges from 600 to 900 mm,with most falling during November–March. Detailsof the environment and farming system at Wedza canbe found in Chigariro (2004).

Rotation experiment 1

An on-farm experiment was established at nineseparate farmers’ fields across a range of soil types.Difficulties in the field activities at two of these sitesresulted in data from only seven of the sites beingreported here. The aim of this experiment (Jiri 2003)was to determine the residual effects of year 1 treat-ments (Table 1) established in November–December1999 on the subsequent maize yield in the 2000–2001season. At each site, treatment plots of 8 ¥ 9 m werelaid out randomly in a block with no on-farm replica-tion. While farmers were responsible for tillage oper-ations, weeding, and assisted in the planting andharvesting operations, the experiments were essen-tially researcher-designed and managed on-farmtrials.

1999–2000 season. Weeds on the weed fallowtreatments grew unchecked from November 1999until they were measured for biomass on 11–13 April2000 and removed (simulating dry-season grazing).

Mucuna (Mucuna pruriens var. utilis) and cowpea(Vigna unguiculata, a trailing variety ex Matopos)fallow and continuous maize (Zea mays var. SC501)were planted (two seeds per station) from 25–30



84

November 1999 in 90 cm rows with 30 cm spacingbetween plants. All legume seed was inoculated withrhizobium before planting. The maize, mucuna andcowpea plants were thinned to one plant per stationwhen the maize had reached the three-leaf stage. Phos-phorus was applied as single superphosphate (8% P,12% S) at a rate of 16 kg/ha to +P mucuna and +Pcowpea treatments before planting. An additionalmucuna treatment that received 16 kg/ha P (as singlesuperphosphate) and 500 kg/ha of dolomitic lime wasincluded at each site to test the response of mucuna tolime.

The continuous maize plots, also started in the1999–2000 season, received 18 kg/ha P and 24 kg/haN banded near the plants in the form of ‘CompoundD’ fertiliser (8:6:6:6.5% N:P:K:S) at planting and 69kg/ha N applied 5 weeks after planting (WAP) in theform of ammonium nitrate (34.5% N). Hand-weeding of all plots occurred at the same time as thetopdressings were applied. Carbaryl and dimethoatewere used as required to control leaf eaters andaphids.

Macrotyloma axillare cv. Archer mixed withChloris gayana (Katambora rhodes grass) wasplanted in year 1 into shallow furrows and lightlycovered with soil. An additional treatment thatreceived 16 kg/ha P at planting was also established.Being a perennial system, the archer–rhodes grassmix was grown for 2 years, and its forage productioncompared with mucuna and cowpea in year 1.

All treatments were harvested 11–13 April 2000(~19 WAP) using a net plot area (42 m2) formed afterremoving a 1 m border area for mucuna and 43.2 m2

for maize. The archer–rhodes grass plots were cut at10 cm above the ground to facilitate regrowth. Plantmaterial from the net plots was weighed, subsampledand dried at 60°C for 48 hours to determine drymatter. In the continuous maize treatments, cobswere removed from the stalk, dehusked and shelledfor grain. Plant residues from the mucuna, cowpeaand archer–grass hay treatments were removed fromthe plots to simulate a hay system. Plant residuesfrom the mucuna green-manure treatment werereturned to the plots and immediately incorporatedby ploughing with an animal-drawn implement.Mucuna pods were considered as biomass.

Grazing livestock were allowed access to theseareas during the dry season (May–October), as thesefarms are situated in the communal farming areaswhere livestock roam freely during the non-croppingseason.

Maize crop 2000–2001. Maize was planted acrossall treatments (except archer–grass) in the 2000–2001 season. All fields were ox-ploughed by thefarmers by 11 October 2000. Maize var. SC501 wasplanted into all plots on 28–29 November 2000 and,other than fertiliser application, managed as in year 1.The continuous maize treatments received 18 kg/ha Pand 24 kg/ha N banded near the plants in the form ofCompound D at planting and 69 kg/ha N split-appliedat 3 and 6 WAP in the form of ammonium nitrate. Allother plots received 30 kg/ha of N as ammoniumnitrate, split-applied at 3 and 6 WAP. Hand-weedingof all plots was undertaken at the same time as thetopdressings were applied. Harvesting took place on

Table 1. List of treatments used in experiment 1 conducted by Jiri (2003).

Year 1 treatments N (kg/ha) P (kg/ha) Year 2 treatments N (kg/ha) P (kg/ha)

Weed fallow Maize 30

Maize (full fertiliser rate) 93 18 Maize 93 18

Mucuna (ley) 16 Maize 30

Mucuna (ley) Maize 30

Mucuna (green manure) 16 Maize 30

Mucuna (green manure) Maize 30

Mucuna (ley) + lime 16 Maize 30

Cowpea 16 Maize 30

Cowpea Maize 30

Perennial ley 16

Perennial ley



85

10 April 2001 using the same procedures asdescribed for year 1.


Full details of this experiment can be found inTemba (2001). In the 1999–2000 season on two farmsin the Wedza district with granitic sandy soils, a blockwas planted to lablab (with 27 kg/ha P as SSP and 500kg/ha lime) and an adjacent area remained under weedyfallow (no inputs). At the end of the season, lablab washarvested by cutting at the base of the stem (wholeplants cut with sickles) and made into hay for feedingdairy animals. The roots of lablab and their crowns,together with fallen leaves, remained in situ. The lablabgrew well, but no estimate of either the above or below-ground biomass was measured. In May 2000, the siteswere tilled with an ox-drawn plough.

In mid-November 2000, a short season maizevariety (SC407) was sown with 6 fertiliser treatments¥ 3 replicates marked out in each of the lablab andweed areas at the 2 farms (plots size 5 ¥ 4 m). Thelablab treatments received 0, 30 and 60 kg/ha N (asammonium nitrate) ± basal fertiliser (16.7 kg/ha Pand 22 kg/ha S as SSP, and 15 kg/ha K as potassiumchloride). The weed treatments received 0, 30, 60, 90and 120 kg/ha N (as ammonium nitrate) + basal fer-tiliser as above, and an additional 0 N plot receivedno basal fertiliser (Table 2). The basal fertilisers wereapplied before sowing and the N fertiliser was splitand applied at 4 and 7 WAP.

The maize was planted in 90 cm rows with 30 cmbetween plants for a target population of 37,000plants/ha. At maturity, net plot areas of 2.7 ¥ 2.8 m,which excluded border rows and plants, were har-vested.

Soil types. Experiment 1 was conducted at 7 sites,5 of which were located on sandy or sandy loam soil,all of which were derived from granite and classifiedas Ferralic Arenosol (FAO–UNESCO 1974) (Table3). The ‘Gunzvenzve’ site was located on a red clayclassified as a Chromic Luvisol derived from doler-itic parent material, and the black clay ‘Dzunza’ sitewas classified as a Vertisol. Experiment 2 was con-ducted at two sites, one of which was located adja-cent to the Mbavha sandy site and the other at thefarm of Muparadzi located on a sandy loam (Table3). Both soils are classified as Ferralic Arenosols.

Soil and plant analyses

In experiment 1, soil samples (0–15 and 15–30 cmdepth) collected from each of the seven sites at thestart of the experiment were dried, ground (< 2 mm)and analysed for total organic carbon using theWalkley and Black method with a correction factorof 1.3 (Walkley 1947), and for plant available P usingthe Bray I (0.03M ammonium fluoride in 0.025MHCl) method (Bray and Kurt 1945). The basic cationsK+, Ca2+ and Mg2+ were determined using atomicabsorption spectrometry on extractions in 1M ammo-nium chloride at pH 7 (Rayment and Higginson1992). The soil pH was determined using a pH meterin a 1:5 soil/0.1M CaCl2 solution. In experiment 2,soil samples were collected from the weed fallow andlablab treatments following the 1999–2000 season.


In year 1, the mucuna hay and green-manure treat-ments and an additional mucuna treatment that was tobe resown as a two-season hay crop were used as

Table 2. List of treatments used in experiment 2 conducted by Temba (2001).

Treatment number

Maize following lablab Treatment number

Maize following one year weed fallow

N fertiliser1

(kg/ha)±Basal

fertiliser2N level(kg/ha)

±Basalfertiliser

123456

00

30306060

+–+–+–

789

101112

00

306090

120

–+++++

1 Ammonium nitrate applied at 4 and 7 weeks after planting.2 16.7 P:22 S:15 K kg/ha applied as SSP and potassium chloride.



86

three replicates in an analysis of variance (ANOVA).Treatment differences in the growth of mucunabetween the –P, +P and +P+lime treatments weretested in an ANOVA using the 7 sites as replicates.Significant difference was used to calculate meanseparation based on Duncan’s multiple range test(DMRT). In year 2, treatment differences in maizegrain and biomass were tested using the sites as rep-licates in an ANOVA. and least-significant differ-ence was used to calculate mean separation based onDMRT.

Results

Soil properties

The sandy and sandy-loam soils were, in general,highly acidic, with the pH (CaCl2) below 4.4 andvery low in the status of base cations K+, Ca++ andMg++ (Table 3), reflecting very low soil-organicmatter (total carbon, CT < 0.4%) and an associatedlow cation-exchange capacity. According to Gourley(1999), a soil K concentration below 0.2 cmol(+)/kgis considered to be low for tropical pastures, while avalue of <0.3 cmol(+)/kg is the critical concentrationfor maize. Aitken and Scott (1999) considered 0.21–0.27 cmol(+)/kg to be the critical soil concentrationof Mg for maize growth. The two clay soils wereslightly less acidic, contained more soil organic Cand cations. According to the Zimbabwean Classifi-cation of Bray extracted P, available P was consid-

ered to be low (less than 7 µg/g Bray-1 P) at all sites,with the exception of the Mbavha site in experiment1, where P status was considered medium (Zvomuya(1996), quoted in Chenje et al. (1998)). This confirmsthe infertility of these over-cropped soils.


Year 1 treatments (1999–2000). Average rainfallreceived across the seven sites during the growingseason was 660 mm. There was a significant differ-ence in biomass between sites (Figure 1). The highestbiomass measurements were obtained on the sandy-loam and clay soil types. These heavier soil types con-tained the highest concentrations of soil C and cations(Table 3), and most likely had a higher water-holdingcapacity. In the continuous maize treatments, theaverage grain and stover yield of the clay sites wasalmost double that of the sandy and sandy-loam sites(Table 4). Weed biomass produced on the clay siteswas also double the biomass grown on the sandy sites.

Averaged across all sites, mucuna biomass, rela-tive to the mucuna treatments that received no inputs,increased by 31% with the application of 16 kg/ha ofP, and by 52% with the addition of 16 kg/ha of P and500 kg/ha of lime (Figure 2). There were no signifi-cant differences, however, between the +P and+P+Lime treatments.

Year 2 maize response (2000–2001). Averagerainfall received across the seven sites during thegrowing season was 918 mm; a mid-season drought

Table 3. Soil properties of the seven on-farm sites (0–30 cm depth).

Location Soil type pH(CaCl2)

Pa

(µg/g)Cb

(%)K+ Mg2+ Ca2+

___________cmol+/kg__________

Experiment 1MbavhaMumvanaMapira AChikumbirikeMapira CGunzvenzveDzuna

SandSandSandSandSandy loamRed clayBlack clay

4.24.24.34.24.44.85.2

11.57.05.53.57.01.01.0

0.290.270.360.360.340.961.22

0.090.110.110.110.170.190.23

0.440.531.000.130.327.678.10

0.090.100.260.120.144.406.21

Experiment 2Mbavha

Muparadzi

LablabWeedLablabWeed

4.64.14.34.6

3.41.01.51.0

––––

0.160.090.040.04

0.320.170.190.17

0.460.310.260.28

a Fluoride extractable phosphorus (Bray 1-P).b Walkley–Black (Walkley1947)



87

occurred. An ANOVA using the sites as replicates(n = 7) revealed a significant difference between rep-licates, though when the 5 sandy and 2 clay sites wereanalysed separately, no significant difference wasfound between them. The weed fallow was used asthe control for the treatments that did not receive P inthe previous season. On the sandy sites, the mucunagreen-manure treatments produced 64 and 100%(p < 0.05) more than the maize produced following aweed fallow or mucuna that was removed for hay,respectively (Figure 3). No significant difference wasfound between treatments at the clay sites.

The fertilised continuous maize control was usedas the control for the mucuna treatments that receivedP in year 1 of the experiment. At the sandy sites, nosignificant difference in maize yield was found

between the treatments (Figure 4). At the clay sites,however, maize grain yield was greatest with full fer-tiliser application, and was significantly furtherreduced where the mucuna was removed for hay inthe previous season (Figure 4).

There was a significant interaction between Papplication and the management of the mucuna resi-dues on the 2000–01 maize grain yield. There was nodifference in maize growth where P had been appliedto the mucuna hay or green-manure treatments in the1999–2000 season (Figure 5). However, where ferti-liser P was not applied, compared with maize fol-lowing mucuna green manure, maize yield wassignificantly reduced following the mucuna haytreatment in the 1999–2000 season (Figure 5). Papplied to a preceding mucuna hay crop significantlyincreased maize grain yield, but there was no maizeresponse to P applied to mucuna green manure.

The maize that followed the cowpea treatments didnot respond to the P applications that were applied tothe cowpea crop (data not shown). There were also nosignificant differences between cowpea sites: meangrain yield following cowpea was 3710 kg/ha. Therewere no results for the maize crop following the two-year perennial ley system due to drought in 2001–02.


Following a weed fallow, there was a response upto 30 kg/ha N at Muparadzi site and up to 60 kg/ha N

Table 4. The yield (kg/ha) of maize grain and stover,and the biomass (kg/ha) of weeds in theweedy fallow treatments, produced duringthe 1999–2000 season.

Sand/sandy loam Clay

Maizea grainstoverWeed

2160b (690)2888 (870)3308 (660)

5750 (350)6550 (250)6280 (70)

a (Maize received 93, 18, 18, 19.5 kg/ha N, P, K, S respectively).b Standard error in brackets calculated for sand (n = 5) and clay

(n = 2)

0

2000

4000

6000

8000

10000

12000

Mu

cun

a b

iom

ass

(kg

/ha) LSD p < 0.05

Dzuna (C

lay)

Gunzvenzv

e (Clay)

Mapira

C (S

andy Loam)

Chikumbiri

ke (Sand)

Mapira

A (S

and)

Mum

vana (Sand)

Mbavha (S

and)

0

2000

4000

6000

8000

10000

12000

Bio

mas

s yi

eld

(kg

/ha)

–P

+P

+P + lime

LSD p < 0.05

Mucuna Perennial Ley Cowpea

Figure 1. The biomass of mucuna produced at 7 on-farm sites in Wedza District, Zimbabwe inthe 1999–2000 wet season.

Figure 2. The response of mucuna, the perennial ley andcowpea to P fertilisation and mucuna to 500kg/ha of dolomitic lime in 1999–2000. (LSDrefers to the treatment mean differenceswithin the mucuna treatment only.)



88

at Mhuka site (Figure 6). This lack of furtherresponse to higher levels of N indicates that limita-tions other than nitrogen may have affected plantgrowth. The basal fertilisers containing P, K and Sdid not overcome these limitations in the weed treat-ments.

Maize following a lablab hay crop to which P, Sand lime had been applied, showed no significantresponse to the basal P, K and S (hence ± means areshown). Maize following the lablab at the Mhuka siteyielded significantly more grain than that followingthe weed fallow. The effect of lablab on maize yieldwas much smaller at the Muparadzi site.

Maize yield was significantly lower with the treat-ments that received no N or basal fertiliser grownafter the weed fallow (Figure 6).

Discussion

Experiment 1

The biomass production of mucuna was excellentat all sites, and exceptional on the soils that had ahigher clay content (9500–11200 kg/ha) and conse-quently greater water-holding capacity and more cat-ions. Mucuna was able to grow across a range of soilswhere the available P status was medium (Mbavha)to low (all other sites), it responded significantly tothe addition of P and lime. A separate study reportedin Jiri (2003) showed some evidence that, with dolo-mitic lime application, mucuna was responding to theincrease in the availability of P, rather than to anincrease in the supply of Mg or Ca. This warrantsfurther investigation.

The fact that mucuna produced more biomass thancowpea and the archer–rhodes grass system is impor-tant for farmers who need to replenish soil fertilityand produce fodder for livestock. However, forfarmers who value relish (young leaf) more thanfodder, cowpea would be more important.

The growth of maize in 2000–2001 was influencedby an interaction between soil type, fertiliser applica-tion and management of the previous season’s resi-dues. Where fertiliser P was not added to either thefallow or maize phase, significantly more maize wasgrown where mucuna was green-manured (Figure 5),particularly on the sandy sites (Figure 4), presumablythe result of P uptake by mucuna and recycling duringthe maize phase. The removal of the mucuna residuesfor hay generally resulted in a negative effect at allsites where fertiliser P had not been added. Under the

0

1000

2000

3000

4000

5000

Sand Clay

Gra

in y

ield

(kg

/ha)

LSD p < 0.05

n.s.

Weed fallow

Mucuna hay (-P)

Mucuna green manure (-P)

Figure 3. The effect of a weed fallow, mucuna (–P)grown as hay or green-manure in the 1999–2000 season on maize grain yield (kg/ha)in the 2000–2001 season. (LSD refers tothe treatment mean differences within thesandy site only.)

Figure 4. The effect of maize (fully fertilised),mucuna (+P) grown as hay or green-manurein the 1999–2000 season (+P) on maizegrain yield (kg/ha) in the 2000–2001season. (LSD refers to the treatment meandifferences within the clay site only.)

0

1000

2000

3000

4000

5000

Sand Clay

Gra

in y

ield

(kg

/ha)

LSD p < 0.05

n.s.

Maize +P+N

Mucuna hay (+P)

Mucuna green manure (+P)



89

marginal to very low available soil P concentrationsmeasured at all sites (Table 3), maize was highlyresponsive to P additions. Carsky et al. (2001) foundthat maize response to a lablab fallow that received 9kg/ha P was similar to maize that received 30–60 kg/haN. Studies showing the effect of recycled P on subse-quent crops are rare. Breman (1998), cited by Carskyet al. (2001), suggests that the introduction ofimproved legume fallows is not an adequate strategywhere P is limiting.

Experiment 2

The growth of forage lablab can result in superiorperformance by a subsequent maize crop in terms ofgrain yields when compared to a one-year weed fal-lowed land, by an average of 57% at the Mhuka siteand 8% at the Muparadzi site (Figure 6). Followinglablab at the Mhuka site, the maize yield without Nwas similar to that achieved with 60 kg/ha N afterweed fallow. Nitrogen fixed by lablab must havepartly contributed to the higher maize grain yieldsachieved when maize followed forage lablab whencompared to yields obtained on land fallowed for oneyear. Another factor contributing to this could beresidual effects of the SSP and lime applied to thelablab. Removal of above-ground biomass (for haymaking) implied that N that could have contributed tosuperior performance of maize following lablab

would have been sourced from that which remainedin roots, crowns and fallen leaves of the lablab.

The decomposition of lablab roots, crowns and leaflitter would also have led to addition of soil organicmatter, an important component of a good soil whichimproves water-holding and cation exchange capac-ities of the soil and, in consequence, an improved soilphysical-environment for subsequent crops (Muhr etal. 1999). In the mucuna trials, it was found that insome cases there were no significant differences infollowing maize yields between above-groundmucuna biomass removal and incorporation, indi-cating that it would be more judiciously used asfodder, with the remaining below-ground biomassstill contributing significantly to soil improvement.

Thus, farmers in the smallholder sector, who areusually on poor soils in terms of fertility, could applyup to 60 kg/ha N less to maize if forage lablab is usedin an improved fallow system instead of leaving landunder weed fallow.

Conclusion

In year 1 of the experiment, mucuna was able toproduce significant biomass across a range of soiltypes containing marginal to very low available soilP. Mucuna produced the highest biomass on soilswith the highest water-holding capacity andresponded significantly to applications of P fertiliser(16 kg/ha P) at planting. In year 2 of the experiment,where maize was grown across the treatments, on thesandy sites in the absence of fertiliser P additions,maize grain production was much higher followinggreen-manuring practices than after a weedy fallowor mucuna ‘removed’ as hay.

The lablab forage system, with P, S and limeapplied, has the potential to restore soil fertility to agreater extent than a weed fallow and contribute toincreased subsequent crop yield. By growing foragelablab, farmers could reduce the amount of fertiliserN from organic sources and cut down on productioncosts. Farmers could apply 30 kg N/ha and be able toproduce satisfactory maize grain yields followinglablab and simultaneously obtain quality hay foranimal feed.

Leaving land under weedy fallow for a season doeslittle to restore soil fertility, having little effect on theyield of a subsequent crop. The results clearly showthat smallholder farmers could benefit more by inte-grating SSP-fertilised forage legumes into fallowlands as leys (for hay), or as green manure if unferti-

0

1000

2000

3000

4000

5000

Gra

in y

ield

(kg

/ha)

LSD p < 0.05

Hay (–P)

Green m

anure (–

P)

Hay (+P)

Green m

anure (+

P)

Figure 5. The effect of P-fertilised or unfertilisedmucuna grown as hay or green-manure(1999–2000) on the grain yield of maize in2000–2001 averaged across all sites.



90

lised. Instead of leaving arable lands under weedyfallow, farmers could crop them to a forage legume,such as mucuna or lablab. They can harvest the largequantities of high-quality above-ground biomass forlivestock feed and still benefit the following seasonin terms of soil-fertility improvement and increasedyield, with reduced inorganic fertiliser inputs.

Recommendations for future research

1. Further investigations of the options of usingleguminous hay and green-manure crops andapplying various rates of fertilisers using a farmingsystems model such as the Agricultural ProductionSystem sIMulator (APSIM) (Keating et al. 2003).

2. Further investigations of the lablab and mucunaforage systems on different soils with replicationsat sites. This would lead to more conclusiveevidence than the single-site data derived here formucuna.

3. It would also be important to design a system inwhich the potential conflicts between mucuna orlablab integration systems are fully compared andcontrasted in terms of costs and benefits in aneconomic analysis.

4. There is need to assess and quantify the foragelegume root contribution to N for the subsequentcrop.

Acknowledgments

The Australian Centre for International AgriculturalResearch (ACIAR) provided funding through projectAS2/96/149 for this study.

References

Aitken, R.L. and Scott, B.J. 1999. Magnesium. In: Peverill,K.I., Sparrow, L.A. and Reuter, D.J., ed., Soil analysis aninterpretation manual. Melbourne, Australia, CSIROPublishing, 255–262.

Bray, R.H. and Kurtz, L.T. 1945. Determination of total,organic and available forms of phosphorus in soils. SoilScience, 59, 39–45.

Buckles, D., Triomphe, B and Sain, G. 1998. Cover crops inhillside agriculture. In: Farmer innovation with mucuna.Mexico, IDRC and CIMMYT, 13, 23–27.

Carsky, R.J., Oyewole, B. and Tian, G. 2001. Effect ofphosphorus application in legume cover crop rotation onsubsequent maize in the savanna zone of West Africa.Nutrient Cycling in Agroecosystems, 59, 151–159.

0

1000

2000

3000

4000

5000

6000

7000

8000

9000

N rate (kg/ha)

0 30 60 90 120

Mai

ze g

rain

(kg

/ha)

Mhuka lablab ± basalMhuka weed + basalMuparadzi lablab ± basalMuparadzi weed + basal

Muparadzi weed (–basal)Mhuka weed (–basal)

Figure 6. Effect of forage lablab on subsequent maize grain yield withdifferent fertiliser levels relative to a one-year weed fallow atMhuka and Muparadzi sites. (The LSD bar refers to thetreatment mean differences between the lablab and weed fallowtreatments at each of the sites separately.)



91

Carsky, R.J., Tarawali, S.A., Becker, M., Chiloye, D., Tian,G and Sanginga, N. 1998. Mucuna – herbaceous coverlegume with potential for multiple uses. InternationalInstitute of Tropical Agriculture, Resource and CropManagement Research Monograph No. 25.

Chenje, M., Sola, L. and Paleczny, D., ed. 1998. The state ofZimbabwe’s environment 1998. Harare, Zimbabwe,Government of the Republic of Zimbabwe, Ministry ofMines, Environment and Tourism.

Chigariro, B. 2004. The farming systems of theZimbabwean district of Hwedza and the role and adoptionof forage legumes in small-scale crop–livestockenterprises. These proceedings.

FAO–UNESCO 1974. Soil map of the world 1:5000000.Volume I, legend. Paris, UNESCO.

Gourley, C.J.P. 1999. Potassium. In: Peverill, K.I., Sparrow,L.A. and Reuter, D.J., ed., Soil analysis an interpretationmanual. Melbourne, Australia, CSIRO Publishing, 229–245.

Haque, I., Jutzi, S. and Neate, P.J. H. 1986. Potentials offorage legumes in farming systems of sub Saharan Africa.Proceedings of a workshop held at the InternationalLivestock Center for Africa, Addis Ababa, Ethiopia,1985. Addis Ababa, ILCA.

Jiri, O. 2003. Integration of forage legumes into smallholderfarming systems, with emphasis on mucuna [Mucunapruriens (L) DC var. utilis (Wall. Ex. Wight) Baker exBurck]. MPhil thesis, University of Zimbabwe, Harare,Zimbabwe.

Keating, B.A., Carberry, P.S., Hammer, G.L., Probert, M.E.,Robertson, M.J., Holzworth, D., Huth, N.I., Hargreaves,J.N.G., Meinke, H., Hochman, Z., McLean, G., Verburg,K., Snow, V., Dimes, J.P., Silburn, M., Wang, E., Brown,S., Bristow, K.L., Asseng, S., Chapman, S., McCown,R.L., Freebairn, D.M. and Smith, C.J. 2003. An overviewof APSIM, a model designed for farming systems

simulation. European Journal of Agronomy, 18, 267–288.

Minson, D. J. 1984. Nutritional values of tropical legumes ingrazing and feeding systems. In: Forage legumes forenergy-efficient animal production. Proceedings of atrilateral workshop held in Palmerston North, NewZealand, 30 April 30–4 May 1984.

Muchadeyi, R. 1998. Herbage yields, chemical compositionand in vitro digestibility of dual purpose legumesintercropped with maize for dry season fodder supple-mentation. MSc thesis, University of Zimbabwe, 5–6.

Muhr, L., Peters, M., Tarawali, S.A. and Schultze-Kraft, R.1999. Forage legumes for improved fallows inagropastoral systems of subhumid Africa. TropicalGrasslands, 33, 220–256.

Rayment, G.E. and Higginson, F.R. 1992. Australianlaboratory handbook of soil and water chemical methods.Melbourne, Australia, Inkata Press, 138p.

Surveyer-General 1984. Zimbabwe 1:1,000,000 map ofnatural regions and farming areas, second ed. Harare,Zimbabwe, Surveyor-General.

Temba, E. 2001. Residual effect of forage lablab (Lablabpurpureus) on performance of a succeeding maize crop ina resettlement area in natural region IIA. A dissertationsubmitted for the BSc Agricultural Management Degree,University College of Distance Education, University ofZimbabwe.

Walkley, A. 1947. A critical examination of a rapid methodfor determining organic carbon in soils — effect ofvarious digestion conditions and of inorganic soilconstituents. Soil Science, 63, 251–64.

Wilson, J.R. 1988. Advances in nitrogen cycling inagricultural ecosystems. In: Proceedings of thesymposium on advances in nitrogen cycling inagricultural ecosystems, Brisbane, Australia, 11–15 May1987. Wallingford, UK, CAB International, 333.



92

Growth and Symbiotic Activities of Cowpea Cultivars in Sole and Binary Cultures with Maize

K.K. Ayisi and P.N.Z. Mpangane*

Abstract

Nitrogen is a major limiting plant nutrient in cereal production in smallholder farming systems in the LimpopoProvince of South Africa and the practice of intercropping the cereal with legumes is usually proposed toenhance nitrogen nutrition in the system. The benefit of symbiotic nitrogen fixation in intercropping systemshas, however, been variable over diverse environments and needs further clarification. Alternate intercroppingstudies of maize and four distinct cowpea cultivars were conducted in 1998/99 and 1999/2000 at Syferkuil andThabina/Dan in the Limpopo Province to determine the effect of the system on growth, nodulation, N2 fixationand nitrogen uptake of the component crops. Maize dry-matter accumulation was generally not influenced byintercropping and nitrogen uptake of intercropped maize was only improved relative to the sole culture maizeat Syferkuil in 1999/2000. On average, the amount of nitrogen fixed by the cowpea cultivars was higher in theintercropped cowpea than the sole crops, except in 1999/2000 at Syferkuil, where the amount fixed was 8.6%higher in the sole than the intercropped cowpea. The amount of nitrogen fixed by the legumes ranged from 20to 69 kg N/ha and 87 to 217 kg N/ha at Syferkuil in 1998/99 and 1999/2000, respectively, whereas at Thabina/Dan, the range was 103 to 175 kg N/ha and 83 to 147 kg N/ha, in 1998/99 and 1999/2000, respectively. Noduleformation differed among the cowpea cultivars tested, with the long-season cowpea cultivars generallyproducing heavier nodules at both locations in the two growing seasons. The different performance of thecowpea cultivars in nitrogen fixation at the two locations emphasises the importance of the environment onsymbiotic nitrogen fixation activities in cowpea.

Intercropped systems consisting of cereals andlegumes are common throughout Africa, including insmallholder farming systems in the Limpopo Prov-ince of South Africa. However, in this province, theproportion of the legume is relatively small com-pared to the cereal, since maize is the priority crop.The widespread practice of intercropping cereals andlegumes may be due to some of the established andspeculated advantages of intercropping, such ashigher grain yields (Harris et al. 1987; Putnam andAllan 1992), greater land-use efficiency per unit landarea (Baker and Blamey 1985; Harris et al. 1987) and

improvement of soil fertility by addition of nitrogenthrough symbiotic fixation and excretion from com-ponent legume species (Patra et al. 1986).

The dominance of maize in the smallholder crop-ping system of the Limpopo Province has resulted inpoor soil fertility, particularly in nitrogen and phos-phorus, in many rural communities of the province.Nitrogen is a nutrient that is required in largeramounts among plant nutrients and therefore farmersmust apply large amounts of inorganic fertilisers tomaintain crop yields. The use of artificial nitrogenfertilisers plays an important role in supplying cropnutrient needs but continuous application is costlyand many smallholder farmers cannot afford ferti-lisers or can only apply marginal quantities (Jonesand Wendt 1995). In addition, excessive application

* School of Agricultural and Environmental Sciences,Plant Production, University of the North, Private BagX1106, Sovenga, 0727 Republic of South Africa.



93

can sometimes cause environmental problems, suchas contamination of surface and underground waterthrough run-off and leaching. Intercropping a cerealwith a legume can benefit an associated cereal crop(Patra et al. 1986) or a following cereal in a rotationalsystem. The nitrogen benefit from a legume in anintercropping system will depend on its active sym-biotic activity under such system. In an intercroppingsituation, the quantity of nitrogen derived from bio-logical fixation from a legume is influenced by anumber of factors, such as soil nitrogen (Patra et al.1986; Stern 1993), species or cultivar of legume(Stern 1993; Watiki et al. 1993), soil moisture, andlegume density in the intercrop (Stern 1993; Watikiet al. 1993). The importance of legume variety innitrogen nutrition in the maize–cowpea intercroppingsystem in the Limpopo Province and many parts ofSouth Africa has not yet been documented. Some ofthese cowpea cultivars exhibit differences in theirsymbiotic activities as sole crops in the province(Ayisi et al. 2000) but their fixation in an intercrop-ping situation is yet to be established. The objectivesof this study were to: (i) assess dry-matter productionand nitrogen uptake of the associated maize crop inan intercropped system with different cowpea varie-ties; and (ii) determine the effect of the intercroppingon biomass production, nodulation, and nitrogen fix-ation of the cowpea varieties in the system.

Materials and methods

Field experiments were carried out at two locations inthe Limpopo Province of South Africa — namely, theUniversity of the North experimental farm at Syfer-

kuil and a communal field at Thabina/Dan during the1998/99 and 1999/2000 growing seasons. The exper-iments were established at different sites within eachlocation. Syferkuil has relatively higher soil fertilitycompared to that of Thabina/Dan (Table 1).

The experiments were established as randomisedcomplete block designs with four replications at eachlocation. Treatments examined were four cowpeacultivars — namely, Pan 311, Pan 326, BechuanaWhite and Agrinawa — which were either inter-cropped in alternate 90 cm rows with maize varietySNK2147 or planted as sole cultures. Thus, thecowpea replaced a maize row in an alternate mannerin the intercropped plots. Maize in sole culture wasincluded as separate treatment. The final densitieswere 30,000 and 15,000 plants per hectare for thesole and intercropped maize, respectively, and60,000 and 30,000 plants per hectare for the sole andintercropped cowpea, respectively. Pan 311 and Pan326 are short-duration types whereas BechuanaWhite and Agrinawa are medium- and long-durationcultivars, respectively. The cowpea seeds were inoc-ulated with commercial Bradyrhizobium strainCB756 just before planting. The data collected wereseasonal dry-matter accumulation, nodulation andnitrogen fixation. Dry-matter samples of the compo-nent crops were taken twice from two 1 m lengthrows of each experimental unit in both growingseasons and at each location. During sampling, maizeplants were cut at ground level to determine theabove-ground dry matter, whereas in cowpea, whole-plant samples, including roots, were taken. Thecowpea plants were dug carefully to maintain theirroot systems and then immersed in water to remove

Table 1. Initial soil analysis results at the experimental sites at all seasons.

Location Season Depth(cm)

pH(H2O)a Nb Pc Kd

mg/kg mg/kg (cmol(+)/kg)

Syferkuil 1998/99

1999/2000

0–1515–300–15

15–30

7.17.06.87.0

13.514.014.313.2

50.446.544.340.2

0.320.310.300.31

Thabina/Dan 1998/99

1999/2000

0–1515–300–15

15–30

6.86.35.65.4

8.77.76.95.7

6.44.21.61.1

0.170.170.120.11

a 1:5 soil:water.b NH4 + NO3 1:5 Extractant 0.1 N K2SO4.c 1:7.5 extractant Bray 2.d 1:10 extractant ammonium acetate 1 N, pH 7.



94

bound soil. The samples were carefully washed undera tap on a sieve to recover all loose roots, after whichroot nodules were detached from each plant. Bothmaize and cowpea plant materials were oven-dried at60°C to constant moisture content and weighed.

Nitrogen fixation in the legume was assessed usingthe 15N natural abundance technique with sole maizeas the reference crop (Unkovich et al. 1993). Delta(d) 15N values were determined using mass spec-trometer model NA15000NC (CHNN) analyser. Theproportion of plant nitrogen derived from the atmos-phere (%Ndfa) was calculated as follows:

where d15N ref plant is the value of a non-leguminousreference plant grown simultaneously in close prox-imity to the legume and sampled simultaneously; andB is the d15N value of cowpea grown in the glasshouseon sand culture, where the crops depended solely on

symbiotic fixation for their N nutrition. The sole maizecrop within each experimental block was used as a ref-erence crop for the legume in that block, as soil condi-tions within the block were largely similar. Data weresubjected to analysis of variance (ANOVA) using theStatistical Analysis System (SAS). Differencesbetween treatment means were separated using theleast significant difference (LSD) procedure (Gomezand Gomez 1984).


Maize dry-matter accumulation and nitrogen yield

No significant differences in seasonal dry-matteraccumulation of maize were observed among thetreatments at either Syferkuil or Thabina/Dan or ineither year (Tables 2 and 3); except at Syferkuil in1999/2000, 58 days after planting (DAP), wheremaize grown in intercrop with cowpea cultivar

%Ndfa = ( N ref plant – N legume)

( N ref plant – B)∂ ∂

∂¥

15 15

15

1001

Table 3. Dry-matter accumulation and nitrogen yield (kg/ha) of maize, intercropped with different cowpeacultivars Pan 311, Pan 326, Bechuana (BC) White and Agrinawa) at Thabina/Dan in the 1998/99 and1999/2000 growing seasons (M = maize; DAP =days after planting).

Cropping 1998/99 growing season 1999/2000 growing season

Dry mattera N Yielda Dry mattera N yielda

40 DAP 88 DAP 40 DAP 88 DAP 60 DAP 60 DAP

M + Pan 311M + Pan 326M + BC whiteM + AgrinawaSole M

11191226111916101261

63807307685359806726

2024202924

11912412192

135

26082965218025081734

4563374837

a No significant differences at the p £ 0.05 level of testing.

Table 2. Dry-matter accumulation and nitrogen yield (kg/ha) of maize intercropped with different cowpea cultivars(Pan 311, Pan 326, Bechuana (BC) White and Agrinawa) at Syferkuil in the 1998/99 and 1999/2000growing seasons (M = maize; DAP = days after planting).

Crop 1988/89 growing season 1999/2000 growing season

Dry mattera N yielda Dry matter N yield

55 DAP 80 DAP 55 DAP 80 DAP 58 DAP 95 DAP 58 DAP 95 DAP

M + Pan 311M + Pan 326M + BC WhiteM + AgrinawaSole M

15391668136717681476

36093219401532513016

3439334034

5754665848

2320 b2596 b2334 b3755 a2439 b

42914594519957134176

4551438339

73 b77 b85 ab110 a59 b

a Not significant.Note: numbers in each column followed by different letters are significantly different at p £ 0.05.



95

Agrinawa produced the highest dry-matter yield of376 g/m2. This was 55.2% higher than the averageyield of the other treatments and 54% higher than thesole maize yield (Table 2).

The general lack of significant differencesbetween the sole maize and all the intercroppedsystems is an indication that the cowpea speciesstudied can be incorporated into the maize culture inthis alternate intercropping system withoutdepressing the maize growth.

The nitrogen yield response of maize to croppingsystem was significant (p £ 0.05) only at Syferkuil in1999/2000 (Tables 2 and 3) and not at the other loca-tions and years. At both 58 and 95 DAP, at this loca-

tion, the nitrogen yield of maize intercropped withcowpea cultivar Agrinawa was higher than in the solecrop, whereas all other maize intercrops were similarto the sole cropped maize. The intercrops, however,showed a tendency towards higher nitrogen yields,about 44% and 47% higher on average than the soleculture maize at 58 and 95 DAP, respectively (Table2). The increased nitrogen yield of maize inter-cropped with cowpea cultivar Agrinawa could beattributed to the high nodulating ability of thiscowpea cultivar (Tables 4 and 5), indicating the pos-sibility of some nitrogen transfer from the cowpea tothe maize, as has been reported previously (Patra etal. 1986). The nitrogen yield of maize intercropped

Table 4. Whole-plant dry-matter accumulation and nodule mass of cowpea cultivars (Pan 311, Pan326, Bechuana (CB) White and Agrinawa) at Syferkuil during the 1998/99 and 1999/00growing seasons (DAP = days after planting).

Cropping system 1998/99 1999/2000

Dry matter (55 DAP)

(kg/ha)

Nodule mass (55 DAP)(mg/plant)

Dry matter (58 DAP)

(kg/ha)


Sole Pan 311Sole Pan 326Sole BC WhiteSole Agrinawa Intercropped Pan 311Intercropped Pan 326Intercropped BC WhiteIntercropped Agrinawa

1444 a857 c

1592 a1281 ab922 bc995 bc

1100 abc1249 ab

2.1 c2.7 bc6.8 a4.3 a

2.8 bc5.8 abc8.8 a

5.1 abc

1055 a1265 bc2531 a2167 a907 d

1207 cd1750 b1547 b

2.3 c4.4 b10.2 a8.3 ab2.5 c3.0 c

5.9 abc4.9 bc

Note: numbers in each column followed by different letters are significantly different at p £ 0.05).

Table 5. Whole-plant dry-matter accumulation and nodule mass of cowpea cultivars (Pan 311, Pan 326,Bechuana (BC) White and Agrinawa) at Thabina/Dan during the 1998/99 and 1999/2000 growingseasons (DAP = days after planting).


Dry matter (40 DAP)

(kg/ha)


Dry mattera (60 DAP)

(kg/ha)


Sole Pan 311Sole Pan 326Sole BC whiteSole Agrinawa Intercropped Pan 311Intercropped Pan 326Intercropped BC whiteIntercropped Agrinawa

1174 a648 c

1209 a1039 ab805 bc812 bc844 bc

1076 ab

16.9 cd24.3 bc27.7 bc32.2 b7.6 d

29.8 b46.7 a

35.3 ab

668987606

1269750

1022536440

6.8 c9.7 bc9.1 bc

14.7 ab6.9 c

12.0 bc20.5 a

14.0 abca Not significant at the p £ 0.05 level of testing.Note: numbers in each column followed by different letters are significantly different at p £ 0.05).



96

with cowpea cultivar Bechuana White, which is alsoa high-nodulating cultivar, was statistically similar tothat of maize intercropped with Agrinawa, confir-iming possible transfer. The lack of nitrogen benefitbetween intercropped and sole culture maize, asobserved in other seasons and locations has also beenreported previously (Van Kessel and Roskoski1988).

Cowpea dry-matter accumulation

Cowpea dry-matter accumulation differed signifi-cantly among the cultivars at all locations and sea-sons, except at Thabina/Dan in 1999/2000 where itwas not significant (Tables 4 and 5).

With the exception of Pan 311 and BechuanaWhite at Thabina/Dan in 1998/99, where dry-matteraccumulation in the intercrop was reduced by about43 to 46% relative to the sole crops, dry-matter accu-mulation in the other legumes did not differ whetherplanted in sole or in intercropped systems. Legumegrowth suppression by maize in intercroppedsystems has been reported (Clement et al. 1992). Thelonger-duration cultivars (Bechuana White andAgrinawa) also accumulated much more dry matterthan the early-maturing types (Pan 311 and Pan 326)earlier in the growing season.

Nodule mass

Nodule mass differed among the cowpea cultivars,locations, seasons and also whether the cowpeaswere grown as intercrops or sole cultures (Tables 4

and 5). The highest nodule weights were generallyrecorded in Bechuana White and Agrinawa acrosslocations and seasons. The nodules produced bythese two cultivars when grown as intercrops werealso, on average, 25–45% heavier than when grownas sole crops, except in the 1999–2000 season atSyferkuil, where the average sole crop nodule weightof the cultivars was about 41% lower in the inter-crops. Generally, the cultivar Bechuana Whiteappeared to be the most consistent in terms of noduleweight and might be the cultivar with the highestpotential for improved soil fertility.

Percentage nitrogen derived from fixation

The delta 15N value enables one to determine theextent of a plant’s dependence on soil and atmos-pheric nitrogen. The percentage of nitrogen derivedfrom fixation ranged from 4.3% to 11.7% and 16.8%to 50.0% in the 1998/99 and 1999/2000 growing sea-sons, respectively, at Syferkuil (Table 6), whereas atThabina/Dan, the range was 19.3% to 50.6% and71.3% to 92.7%, respectively, during the two seasons(Table 7). The legumes at Syferkuil (experimentalfarm) were largely dependent on soil nitrogen duringthe 1998/99 growing season but this dependence wasrelatively lower in 1999/2000. This is an indication ofthe importance of seasonal variations in the cultivars'symbiotic activities. In general, the percentagedependence on symbiotic fixation was much higherat Thabina/Dan (farmers’ field) than at Syferkuil,which could partly be attributed to the relativelyhigher initial soil nitrogen at the experimental station.

Table 6. Percentage nitrogen derived from symbiotic fixation (%Ndfa = proportion of plant N derived from theatmosphere) and the amount of nitrogen fixed by sole and intercropped cowpea cultivars (Pan 311, Pan326, Bechuana (BC) White and Agrinawa) at Syferkuil, 55 days after planting.

Cropping system 1998–99 1999–2000

%Ndfa N fixed(kg/ha)

%Ndfa N fixed (kg/ha)


6.3 ± 0.38.4 ± 0.14.3 ± 0.3

–11.7± 0.75.3 ± 0.05.7 ± 0.6

–

63.3 ± 328.2 ± 319.5 ± 5

–68.5 ± 824.7 ± 324.3 ± 3

–

31.0 ± 1716.8 ± 8

29.5 ± 1223.0 ± 1612.7 ± 2

33.8 ± 1828.2 ± 150.0 ± 4

124 ± 2787 ± 31

217 ± 58167 ± 2

136 ± 14121 ± 17144 ± 2

148 ± 29

Note: values are means ± SE.



97

Amount of nitrogen fixed

A wide variation in the amount of nitrogen fixed bythe legumes was observed, ranging from as low as19.5 kg N/ha to as high as 217 kg N/ha across loca-tions and seasons (Tables 6 and 7). There were alsoinconsistencies in the amount of nitrogen fixed by theindividual cultivars, but Pan 326 generally appears tobe a lower nitrogen fixer compared to the others.Comparing sole and intercropped cowpea, theamount of nitrogen fixed was on average 6 to 15%greater in the intercropped systems than in the solecultures. This observation is partly similar to the find-ings of Patra et al. (1986), who reported higheramounts of nitrogen fixed by legumes under inter-cropping conditions.

Conclusions

The cowpea cultivars Pan 311, Pan 326, BechuanaWhite and Agrinawa can be incorporated into maizeculture in an alternate intercropping system withoutdepressing maize growth. The accumulation ofnitrogen by maize did not change whether maize wasplanted as a sole crop or as an intercrop and therewere also no differences among maize crops grownwith the different cowpea species, except in 1999/2000 at Syferkuil.

Significant differences in dry-matter accumulationby the cowpea varieties were, however, observedunder both sole culture and intercropping systems,with the longer-season cultivars accumulating muchhigher dry matter at the mid-vegetative growth stage.

Biomass accumulation in the intercrop was reducedcompared to the sole crops. Nodule formation alsodiffered among the cultivars, with Bechuana Whiteand Agrinawa producing many more nodules thanthe short-duration type. In terms of the cultivars’dependence on symbiotic nitrogen fixation, plantsgrown under farmers’ field conditions dependedmore on symbiotic nitrogen for their nitrogenrequirement than those grown at the experimentalfarm. The actual amount of nitrogen fixed by thelegumes ranged from about 20 to 217 kg/hadepending on season, location and cropping system,sole crop or intercrop.

References

Ayisi, K.K., Nkgapele, R.J. and Dakora, F.D. 2000. Noduleformation and function in six varieties of cowpea (Vignaunguiculata L. Walp.) grown in a nitrogen rich soil inSouth Africa. Symbiosis, 28, 17–31.

Baker, C.M. and Blamey, P.C. 1985. Nitrogen fertilizereffects on yield and nitrogen uptake of sorghum andsoybean, grown in sole cropping and intercroppingsystems. Field Crops Research, 12, 223–240.

Clement, A., Chalifour, F.P., Bharati, M.P. and Gendron, G.1992. Effects of nitrogen supply and spatial arrangementon the grain yield of a maize/soybean intercrop in a humidsubtropical climate. Canadian Journal of Plant Science,72, 57–67.

Gomez, K.A., and Gomez, A.A. 1984. Statistical proceduresfor agricultural research, 2nd edition. New York, JohnWiley and Sons, Inc.

Harris, D., Natarajan, M. and Willey, R.W. 1987.Physiological basis for yield advantage in a sorghum/

Table 7. Percentage nitrogen derived from symbiotic fixation (%Ndfa = proportion of plant N derived from theatmosphere) and the amount of nitrogen fixed by sole and intercropped cowpea cultivars (Pan 311, Pan326, Bechuana (BC) White and Agrinawa) at Thabina/Dan, 55 days after planting.





20.2 ± 11.132.6 ± 0.141.8 ± 0.636.1 ± 8.019.3 ± 1.650.6 ± 7.544.8 ± 9.932.9 ± 2.3

168 ± 17103 ± 22175 ± 21107 ± 7151 ± 6

170 ± 10169 ± 53148 ± 12

92.7 ± 689.3 ± 380.7 ± 684.4 ± 1071.3 ± 378.1 ± 1784.3 ± 1279.7 ± 11

124 ± 483 ± 2699 ± 21147 ± 17123 ± 29117 ± 30146 ± 21136 ± 6

Note: values are means ± SE.



98

groundnut intercrop, exposed to drought: dry matterproduction, yield and light interception. Field CropsResearch, 17, 259–272.

Jones, R.B. and Wendt, J.W. 1995. Contribution of soilfertility research to improve maize production bysmallholder farmers in eastern and southern Africa. In:Jewell, D.C., Waddington, S.R., Ransom J.K. and Pixley,K.V., ed., Maize research for stress environments.Mexico, DF, CIMMYT (International Maize and WheatImprovement Centre), 2–14.

Patra, D.D., Sachdev, M.S. and Subbiah, B.V. 1986. 15Nstudies on the transfer of legume fixed nitrogen toassociated cereals in intercropping systems. Biology andFertility of Soils, 2, 165–171.

Putnam, D.H. and Allan, D.L. 1992. Mechanism of over-yielding in sunflower/mustard intercrop. AgronomyJournal, 84, 188–195.

Stern, W.R. 1993. Nitrogen fixation and transfer in intercropsystems. Field Crops Research, 34, 335–356.

Unkovich, M.J., Pate, J.S. and Sanford, P. 1993. Preparationof plant samples for high precision nitrogen isotope ratioanalysis. Communications in Soil Science Plant Analysis,24, 2093–2106.

Van Kessel, C. and Roskoski, J.P. 1988. Row spacing effecton N2-fixation, N yield and soil N uptake of intercroppedcowpea and maize. Plant and Soils, 111, 17–23.

Watiki, J.M., Fukai, S., Banda, J.A. and Keating, B.A. 1993.Radiation interception and growth of maize/cowpeaintercrop as affected by maize plant density and cowpeacultivar. Field Crops Research, 35, 123–133.



99

Lablab Density and Planting-Date Effects on Growth and Grain Yield in Maize–Lablab

Intercrops

H.M. Maluleke,* K.K. Ayisi*,† and A.M. Whitbread§

Abstract

Incorporation of legumes into the predominantly smallholder maize monoculture systems in the LimpopoProvince of South Africa adds to soil fertility and animal and human nutrition. Lablab (Lablab purpureus) hasbeen found to be well adapted to the dry environment of the Limpopo Province and has the potential to beintercropped with maize. Its potential for prolific growth requires that lablab is well managed if the staplemaize yield is to be maintained or enhanced. Field experiments were established in the 2001/02 and 2002/03seasons at two locations to test the effect of two relative planting dates of lablab — namely, simultaneouslywith maize, and 28 days later, as well as lablab planting densities of 2, 4, 6, 8 and 10 plants/m, on the grain yieldof maize and biomass accumulation of the component crops in a row intercropping system. Grain yieldreduction generally occurred when maize was simultaneously planted with lablab, except at densities of twoand four lablab plants/m in one season. When lablab was planted later, the associated maize produced yieldsthat were similar to or enhanced compared to the maize by itself. Maize dry-matter yields were reduced atlablab densities beyond four plants/m whereas lablab dry-matter accumulation increased with increasingdensity.

Introduction

Maize is the major staple food in the Limpopo Prov-ince of South Africa and it therefore dominates thesmallholder farming system of the province. Thispreference for maize has led to continuous culture ofthe crop, which together with low levels of externalinputs, has resulted in severe soil degradation inmany smallholder farming systems throughout theLimpopo Province. Identification of alternative crop-ping systems and practices to break the dominant

maize monoculture system is required to enhancecrop productivity on farmers’ fields. Intercroppingmaize and leguminous species, mainly cowpea(Vigna unguiculata), groundnut (Arachis hypogeae)and bambara groundnut (Vigna subterranean), iscommon amongst farmers but usually the legumecomponent is minimal. Lablab (Lablab purpureus) isa legume species that can be successfully cultivatedin the Limpopo Province and has the potential to beincorporated into the maize monoculture system.However, preliminary studies conducted on lablabindicated that the crop has prolific growth character-istics and, if not well managed, could severely sup-press maize growth and yields in an intercroppingsystem. Planting date and density of lablab are twoimportant management tools that could be exploredto minimise competitive pressure created by a com-


† Corresponding author.§ CSIRO Sustainable Ecosystems, 306 Carmody Road,

St Lucia, Queensland 4067, Australia.



100

ponent crop in an intercropping system (Ofori andStern 1987).

The objectives of this research, therefore, were todetermine the influence of lablab planting date anddensity on grain yield and to determine the agro-nomic characteristics of maize and dry-matter pro-duction of the component crops in an intercroppingsystem.


Field experiments were carried out at two locations inthe Limpopo Province of South Africa — namely, theUniversity of the North experimental farm at Syfer-kuil and a smallholder farmer’s field at Dalmada nearPolokwane during the 2001/02 and 2002/03 growingseasons. The pre-sowing soil fertility status at the twolocations is presented in Table 1.

The experimental fields were ploughed 2–3 daysbefore planting and 30 kg/ha of phosphorus wasapplied in the form of superphosphate at Dalmada,2001, followed by disking to incorporate the ferti-liser. Nitrogen at 30 kg/ha was applied as urea atplanting to maize at this location. No inorganic ferti-liser was applied at Syferkuil due to the relativelyhigh soil fertility at the site. The lablab seeds wereinoculated with a commercial Bradyrhizobium strainjust before planting.

In the 2001/02 season, the experiments wereplanted on 12 and 13 December 2001 at Dalmada andSyferkuil, respectively. In 2002/03, plantingoccurred on 6 and 12 December 2002 at Dalmada andSyferkuil, respectively. The experiments were in fac-torial arrangement as a randomised complete blockdesign with three replications at each location. The

treatments examined were five different densities oflablab — namely, 0 (sole maize), 2, 4, 6 and 8 plants/m in 2001/02 and an additional treatment of 10plants/m in 2002/03. These densities were eitherplanted simultaneously with maize or 28 days afterplanting (DAP). The lablab was planted between the90 cm inter-row spacing of maize, thus creating a dis-tance of 45 cm between the maize and lablab rows.Each experimental unit consisted of six rows ofmaize 6 m long. The maize cultivar used was SNK2147 and that of lablab was Rongai, which is a long-duration type.

Dry-matter samples of the crops were taken peri-odically from a 1.8 m2 area of each experimental unitthroughout the growing season. Maize plants werecut at ground level to determine the above-grounddry matter. With lablab, whole-plant samples weretaken during dry-matter determination. The plantswere dug carefully to maintain their root system andthen immersed in water to remove bound soil. Thesamples were carefully washed under a tap on a sieveto recover all loose roots. The biomass accumulationfor lablab consisted of both the tops and root mate-rial. Both maize and lablab plant materials wereoven-dried at 60°C to constant moisture content, thenweighed.

Weeds were controlled by hand twice during thegrowing season. Days to flowering in both cropswere recorded when 50% of the plants in a plot hadflowered. Physiological maturity of maize wasscored when 90% of the plants in a plot revealed cobswith no milk line (Stoskopf 1981). Grain yieldsamples of maize were taken from the middle fourrows, 2.5 m long, of each plot, leaving one row oneach side as a border row.

Table 1. Initial topsoil (0–15 cm) and subsoil (15–30 cm depth) nutrient status at Syferkuil and Dalmada duringthe 2001/02 and 2002/03 growing seasons.

Location Season pHa Mineral Nb

(mg/g)Pc

(mg/g)Kd

(cmol(+)/kg)

0–15 15–30 0–15 15–30 0–15 15–30 0–15 15–30

Syferkuil

Dalmada

2001/022002/032001/022002/03

6.67.87.66.9

7.07.57.97.1

20.011.03.23.0

12.09.02.82.0

23339

43

14277

27

0.380.291.411.02

0.490.300.820.95

a 1:5 soil:water.b NH4 + NO3 1:5 Extractant 0.1N K2SO4c 1:7.5 extractant Bray 2.d 1:10 extractant ammonium acetate 1 N, pH 7.



101

Data were subjected to analysis of variance(ANOVA) using the Statistical Analysis System(SAS). Differences between treatment means wereseparated using the least significant difference (LSD)procedure (Gomez and Gomez 1984).


Grain yield

There was no lablab grain produced during any ofthe field trials so only the grain yield of maize is pre-sented here. Maize grain yield was influenced byboth lablab planting date and density at both loca-tions and seasons (Tables 2 and 3). The interactioneffect of planting date and density was also signifi-cant at all locations and seasons except at Dalmada inthe 2002/03 season. Grain yield of maize simultane-ously planted with lablab was, on average, reducedby 26% and 57% compared to those intercroppedwith later-planted lablab in 2001/02 at Dalmada andSyferkuil, respectively (Table 2). In 2002/03, theyield reduction was 20% and 41%, respectively(Table 3.). A general trend of decreasing maize grainyield with increase in lablab density under the simul-taneous planting system was observed in bothseasons at the two locations. In 2001/02, greater

maize yield reduction occurred at densities of 6plants/m and above (Table 2). However, in 2002/03,the planting density of 2 plants/m at Syferkuilresulted in similar yields as the sole crop under thesimultaneous planting. In maize–legume intercrops,most researchers have reported yield depression ofthe legume by maize (Ezumah et al. 1987; Ofori andStern 1987; Clement et al. 1992). The yield reductionof maize observed in this study could be attributed toincreased competition created by lablab at higherdensities. Maize grain yield reduction has beenreported in maize–cowpea intercrops (Shumba et al.1990; Siame et al. 1998) and in maize–bean systems(Siame et al. 1998).

When lablab was planted 28 days after maize,grain yield at planting densities of 2 and 4 lablabplants/m was similar to the sole maize in 2001/02 atDalmada, whereas at Syferkuil, a significantly highergrain yield — an average of 59% — was obtained atlablab densities of 2 and 4 plants/m compared to thesole maize yield during the same season (Table 2).During the 2002/03 season at Dalmada, the grainyield of maize intercropped with later-planted lablabat a density of 2 plants/m was 14% higher than for thesole crop, whereas at a planting density of 4 plants/m,grain yield was similar to the sole crop yield. AtSyferkuil, grain yields at 2 and 4 plants/m were

Table 2. Response of maize grain yield (kg/ha) tolablab planting date and density (plants/m) atDalmada and Syferkuil during the 2001/02growing season (Sim = maize and lablabplanted simultaneously, 28 DAP = lablabplanted 28 days after the maize).

Density (plants/m)

Dalmada Syferkuil

Sim 28 DAP

Sim 28 DAP

0246810DateDensityInteraction

1076 a802 b721 b321 c90 d

–******

1076 a999 a971 a539 b464 b

–******

863 a528 b436 c189 d122 d

–******

863 c1615 a 1126 b803 c583 d

–******

Note: means followed by the same letter within columns are

similar statistically; ** = P < 0.01; * = P < 0.05; NS = not

statistically significant.

Table 3. Response of maize grain yield (kg/ha) tolablab planting date and density at Dalmadaand Syferkuil during the 2002/03 growingseason (Sim = maize and lablab plantedsimultaneously, 28 DAP = lablab planted 28days after the maize).

Density (plants/m)

Dalmada Syferkuil

Sim 28 DAP

Sim 28 DAP


1674 a1438 b1217 c733 d819 d572 e

****NS

1674 b1908 a1654 b971 c961 c884 c

****NS

5181 a4729 a3055 b2567 bc2041 bc1583 c

*****

5181 b7550 a7141 a5797 b3985 c2875 c

******

Note: means followed by the same letter within columns are

similar statistically; ** = P < 0.01; * = P < 0.05; NS = not

statistically significant.



102

higher than the sole maize by about 46% and 38%,respectively (Table 3). The greater yield boost underthe later-planted lablab system could primarily beattributed to the better suppression of lablab vigourby the earlier-planted maize. Yield advantage inintercropping can arise when component crops havedifferent growth patterns and make major demandson resources at different times (Harris et al. 1987;Putnam and Allan 1992). In an intercropping system,a component crop can positively modify the growingenvironment for the benefit of the other crop, whichcan lead to an overall yield advantage relative to thesole crop (Vandermeer 1992). In this study, the later-planted lablab, though less competitive with maize,was observed to completely cover the soil late in theseason which could act to suppress weeds, createcooler soil conditions and possibly minimise mois-ture loss compared to the sole crops. These combinedimpacts would contribute to the enhanced yield of theintercropped maize compared to maize sole crops.

Maize biomass accumulation

Planting density of lablab significantly (P £ 0.05)reduced maize dry-matter accumulation at all samplingdates at both Dalmada and Syferkuil during the twoseasons of experimentation, except at 45 and 59 DAP atSyferkuil in 2002/03 (Tables 4 and 5). The interactioneffect between planting density and date was not signif-icant at the two locations and seasons. Pooled acrossplanting date, maize biomass accumulation declined asthe number of lablab plants/m increased at both loca-tions. The decreased biomass accumulation withincreasing density resulted in the lower grain yields athigher densities. However, in 2001/02 at Dalmada,

maize dry-matter accumulation was not reduced byintercropping compared to sole maize when plantedwith lablab at 2 and 4 plants/m (Table 4). At Syferkuil,this effect was observed only at 45 DAP, beyond whichthe dry matter was reduced significantly compared tothe sole crops. In 2002/03, yields of maize intercroppedat 2 and 4 lablab plants/m were similar to the sole cropat all sampling dates at the two locations, except at 54DAP at Dalmada, where dry-matter accumulationintercropped with 4 lablab plants/m was reduced by31% compared to the sole maize (Table 5). On average,maize dry-matter accumulation was reduced at all sam-pling dates when planted simultaneously with lablab,especially at the later stage of growth.

Lablab biomass accumulation

During the 2001/02 growing season, significantdifferences in seasonal lablab dry-matter accumula-tion related to planting date were observed. Biomassaccumulation of later-planted lablab was consistentlyreduced in the intercropping system with maize com-pared to those simultaneously planted with maize(Tables 6 and 7). When lablab was simultaneouslyplanted with maize, a general increase in biomassaccumulation with increasing density of the legumewas observed at all locations and seasons. This iscontrary to the accumulation pattern of maize, wherean increase in lablab density resulted in a decrease inmaize biomass accumulation. When lablab wasplanted 28 days after the maize, biomass accumula-tion did not differ with planting density in the 2001/02 growing season, but in the 2002/03 seasons, therewas an increase in biomass accumulation withincreasing density (Tables 8 and 9).

Table 4. Total maize dry-matter accumulation (kg/ha) at different growth stages as influenced by lablab plantingdensity at Dalmada and Syferkuil during the 2001/02 growing season (DAP = days after planting).

Density(plants/m)

Dalmada Syferkuil

41 DAP 55 DAP 88 DAP 102 DAP 45 DAP 59 DAP 83 DAP 102 DAP


858 a788 a

710 ab579 bc534 c

****NS

2712 a2958 a2655 a1672 b1619 b

****NS

5752 a5508 a4037 b3347 bc2737 c

****NS

8488 c9780 a9314 b4590 d3292 e

****NS

623 a 562 ab633 a

456 bc425 cNS**NS

892 a782 b774 bc737 c683 cNS**NS

4784 a3626 b3560 bc2890 cd2447 d

****NS

7878 a6091 b 5688 b4026 c3563 c

***NS

Note: means followed by the same letter within columns are similar statistically; ** = P < 0.01; * = P < 0.05; NS = not statistically

significant.



103

Table 5. Total maize dry-matter accumulation (kg/ha) at different growth stages as influenced by lablab plantingdensity at Dalmada and Syferkuil during the 2002/03 growing season (DAP = days after planting).

Density(plants/m)

Dalmada Syferkuil

54 DAP 68 DAP 82 DAP 54 DAP 68 DAP 82 DAP


1001 a954 a690 b615 bc596 bc581 c

**NSNS

2616 a2512 a2537 a1734 b1584 bc1336 c

**NSNS

3911 a3896 a3853 a2703 b2686 b2296 b

**NSNS

2630 a 2587 a2475 a1187 b1173 b1028 b

****NS

3956 a3999 a3839 a2546 b2431 b 2307 b

****NS

9673 a9077 a9188 a7205 b7058 bc6482 c

****NS


significant.

Table 6. Total dry-matter accumulation (kg/ha) of lablab at Dalmada as affected by planting density and dateduring the 2001/02 growing season (Sim = maize and lablab planted simultaneously, 28 DAP = lablabplanted 28 days after the maize).

Density(plants/m)

41 DAP 55 DAP 88 DAP 102 DAP

Sim 28 DAP Sim 28 DAP Sim 28 DAP Sim 28 DAP

246810DensityDateInteraction

150.0 b170.8 b225.0 a212.5 a

–******

29.2 a29.7 a29.7 a37.5 a

–******

358 d 441 c608 b754 a

–******

42.5 a59.0 a44.4 a44.0 a

–******

667 c511 c

1020 b1375 a

–******

45 a63 a48 a48 a

–******

671 c562 c

1196 b1546 a

–******

34 a53 a34 a24 a

–******


significant.

Table 7. Total dry-matter accumulation (kg/ha) of lablab at Syferkuil as affected by planting density and dateduring the 2001/02 growing season (Sim = maize and lablab planted simultaneously, 28 DAP = lablabplanted 28 days after the maize).

Density(plants/m)

41 DAP 55 DAP 88 DAP 102 DAP


246810LSD (£0.05)DensityDateInteraction

182 b117 c200 b256 a

–29******

48 a50 a69 a78 a

–29******

621 c575 c746 b921 a

–102******

60 a68 a85 a 92 a

–102******

3480 d4018 c4411 b4784 a

–238****NS

170 a204 a256 a296 a

–238****NS

5589 c6093 b6514 b7421 a

–475*****

239 a308 a331 a357 a

–475*****


significant.



104

Flowering and maturity of maize

Days to flowering of maize ranged from 61–75DAP across locations and seasons (Tables 10 and11). The timing of flowering and maturity wassimilar across all treatments in the 2001/02 season(Table 10). In the 2002/03 season, flowering wasdelayed by 2–10 days at the intercrop densities of 6–10 lablab plants/m. Where the maize and the lablabwere planted simultaneously, maturity was generallydelayed at lablab densities of 6–10 plants/m.

Conclusion

The grain yield of maize simultaneously plantedwith lablab was generally reduced compared to

the sole crop. However, at the intercropped lablabdensities of 2 and 4 plants/m at Dalmada anddensity of 2 plants/m at Syferkuil in 2002/03,maize produced similar yields as the sole cropmaize. There was a general trend of decreasingmaize grain yields as lablab density increasedafter simultaneous planting. When lablab wasplanted 28 days after maize, grain yields of theassociated maize crop were similar to or higherthan the sole crop yield. Maize dry-matter accu-mulation at intercrop densities of 2 and 4 plants/mwas generally similar to the sole crop. Beyond thisdensity, a general decrease in dry matter wasobserved. Dry-matter accumulation of lablab, onthe other hand, increased with increasing densityof the legume.

Table 8. Total dry-matter accumulation (kg/ha) of lablab at Dalmada as affected byplanting density and date during the 2002/03 growing season (Sim = maize andlablab planted simultaneously, 28 DAP = lablab planted 28 days after the maize).

Density(plants/m)

54 DAP 68 DAP 82 DAP

Sim 28 DAP

Sim 28 DAP

Sim 28 DAP


177 c208 c259 b263 b298 a

******

25 c35 bc52 bc60 b101 a

******

709 b744 b770 b852 a925 a

******

62 b59 b90 a

108 a125 a

******

798 d922 c916 c998 b1392 a

******

94 b90 b

115 b140 b190 a

******

Note: means followed by the same letter within columns are similar statistically; ** = P < 0.01; * = P < 0.05;

NS = not statistically significant.

Table 9. Total dry-matter accumulation (kg/ha) of lablab at Syferkuil as affected byplanting density and date during the 2002/03 growing season (Sim = maize andlablab planted simultaneously, 28 DAP = lablab planted 28 days after the maize).

Density(plants/m)

54 DAP 68 DAP 82 DAP

Sim 28 DAP Sim 28 Sim 28 DAP


213 c244 c294 b299 a333 a

******

60 c70 c87 b96 b137 a

******

539 c635 c845 b873 a

1001 a******

84 a96 a

142 a152 a172 a

******

893 d1016 c1011 c1093 b1485 a

******

189 b185 b210 b236 b286 a

******

Note: means followed by the same letter within columns are similar statistically; ** = P < 0.01; * = P < 0.05;

NS = not statistically significant.



105

The flowering date of maize intercropped withhigh-density lablab tended to be longer compared tolower lablab densities as well as the sole maize crop.The planting date of lablab did not influence flow-ering in maize. Similar to flowering, the maturity ofmaize was also delayed with increasing lablab densi-ties. Lablab could therefore be incorporated in thepredominantly maize monoculture without reducingyield of the cereal if planted about a month later. Ifplanted simultaneously with maize, a density of 4lablab plants/m should not be exceeded.

References


Ezumah, H.C., Nam, N.K. and Walker, P. 1987. Maize–cowpea intercropping as affected by nitrogenfertilization. Agronomy Journal, 79, 275–280.

Gomez, K.A. and Gomez, A.A. 1984. Statistical proceduresfor agricultural research, 2nd edition. New York, JohnWiley and Sons, Inc.

Harris, D., Natarajan, M. and Willey, R.W. 1987.Physiological basis for yield advantage in a sorghum/groundnut intercrop exposed to drought: 1. Dry matterproduction, yield and light interception. Field CropsResearch, 17, 259–272.

Ofori, F. and Stern, W.R. 1987. Cereal–legume intercroppingsystems. Advances in Agronomy, 41, 41–86.

Putnam, D.H. and Allan, D.L. 1992. Mechanism for over-yielding in sunflower/mustard intercrop. AgronomyJournal, 84, 188–195.

Shumba, E.M., Dhliwayo, H.H. and Mukoko, O.Z. 1990.The potential of maize–cowpea intercropping in lowrainfall areas of Zimbabwe. Zimbabwe Journal ofAgricultural Research, 28, 32–38.

Siame, J., Willey, R.W. and Morse, S. 1998. The response ofmaize/Phaseolus intercropping to applied nitrogen onOxisols in northern Zambia. Field Crops Research, 55, 73–81.

Stoskopf, N.C. 1981. Understanding crop production, 1st

edition. Reston, Virginia, Reston Publishers Co. Inc.

Vandermeer, J. 1992. The ecology of intercropping.Cambridge, UK, Cambridge, University Press.

Table 10. Days to flowering and maturity of maize at Dalmada and Syferkuil during the 2001/02growing season (Sim = maize and lablab planted simultaneously, 28 DAP = lablabplanted 28 days after the maize).

Density (plants/m)

Dalmada Syferkuil

Flowering Maturity Flowering Maturity


02468

64 65 68 70 68

6464646464

110 111 111 108 110

109110109109110

73 70 71 66 62

73 72 71 69 70

115118120120118

115112115112114

Table 11. Flowering and maturity of maize at Dalmada and Syferkuil during the 2002/03 growingseason (Sim = maize and lablab planted simultaneously, 28 DAP = lablab planted 28days after the maize).

Density (plants/m)

Dalmada Syferkuil



02468

10

66 66 62 75 74 75

66 69 70 71 71 70

118 113 109 122 124 126

118 125 122 124 117 124

65 65 67 65 67 75

65 61 65 71 75 75

110 111 121 120 120 128

110 116 108 112 108 109



106

Grain Yield of Maize Grown in Sole and Binary Cultures with Cowpea and Lablab in the Limpopo

Province of South Africa

P.N.Z. Mpangane,* K.K. Ayisi,*,† M.G. Mishiyi* and A. Whitbread§

Abstract

The smallholder cropping system in the Limpopo Province of South Africa is characterised by predominantlymaize monoculture, low external inputs and poor soil fertility, particularly low nitrogen and phosphorus. Theneed to include significant amounts of leguminous species in the system without sacrificing maize yields isessential. Field studies were conducted to assess the yield performance and agronomic characteristics of maizein sole and intercropped systems with diverse cowpea and lablab cultivars. Two different experiments werecarried out at two locations in the Limpopo Province — namely, the University of the North experimental farmat Syferkuil and at a communal field at Thabina/Dan community during the 1998/99 and 1999/2000 growingseasons, and in the 2001/02 season, at Syferkuil and a farmer’s field at Dalmada. The 1998/99 and 1999/2000trials consisted of four cowpea cultivars (Pan 311, Pan 326, Bechuana White and Agrinawa) intercropped atalternate 90 cm inter-row spacing with the maize variety SNK2147 and their respective sole crops. The 2001/02 trials consisted of two cowpea cultivars (Glenda and Bechuana White) and two lablab cultivars (Rongai andCommon) planted as sole crops or interplanted between two maize rows, spaced 90 cm apart. Differences inmaize grain yield during the 1998/99 and 1999/2000 trials were only observed at Syferkuil in 1999/2000. Theyield of maize intercropped with Pan 311 was superior to the sole crop yield, whereas the others were similar.In the 2001/02 trials, intercropped maize grain yields were generally lower than the sole crops. Seed yield ofthe cowpea cultivars Pan 326, Bechuana White and Agrinawa were similar in sole and intercropped systems,whereas the seed yield of cowpea cultivar Pan 311 was reduced by intercropping.

Intercropping involving cereals and legumes is acommon practice in many developing countries ofAfrica, Asia and South America but the advantagesof intercropped systems over sole cultures are influ-enced by several factors including habitat, soil fer-tility and moisture, and crop species and varieties(Vandermeer 1992). Most researchers have discov-

ered that when intercropping maize and a grainlegume, the maize often depresses the yield of thelegume crop, especially at high levels of soilnitrogen. Combinations in which the legume cropwas depressed by maize include: maize–cowpea(Ezumah et al. 1987; Ofori and Stern 1987), maize–peanut (Searle et al. 1981) and maize–soybean(Searle et al. 1981; Clement et al. 1992). However,other researchers have reported yield depression ofmaize when intercropped with cowpea (Shumba etal. 1990; Siame et al. 1998). This inconsistent per-formance of component crops requires critical inves-tigation in each particular locality if farmers are tobenefit from the practice of intercropping in that


† Corresponding author.§ CSIRO Sustainable Ecosystems, 306 Carmody Road, St

Lucia, Queensland 4067, Australia.



107

locality. The smallholder farming system of theLimpopo Province of South Africa is characterisedby variable seasonal rainfall, predominantly maizemonocultures, low soil organic-matter levels, poorsoil fertility, particularly low nitrogen and phos-phorus, and minimal external inputs into the farmingsystems. The consequent effect is the levelling ofcrop yields at extremely low levels which, if notchecked, will render the farming land in most ruralcommunities unsuitable for sustainable production.Recent efforts to improve soil fertility have beenthrough the introduction of leguminous species intothe farming systems of many rural communities,mainly as intercrops with maize. Cowpea (Vignaunguiculata) is a grain legume that is receiving muchattention from researchers for its improved growthand yield under sole culture. However, cowpea isusually intercropped with maize or grain sorghum onfarmers’ fields, but the proportion of the legume inthe mixture is often too low to constitute a true inter-cropping system. Since maize is also a priority crop,the challenge is to maintain or enhance growth andyield of the cereal under increased proportion oflegumes in a mixture. Lablab (Lablab purpureus) isanother leguminous species that has shown a poten-tial for prolific growth in the province. Both cowpeaand lablab are sources of food for humans as a graincrop and as a leafy vegetable, and also as a foddercrop for animals. The performance of the twolegumes as intercropped species with maize is, how-

ever, not yet established. The objectives of this studywere to determine the effects of intercropping maizewith diverse cowpea cultivars and lablab on grainyields and agronomic characteristics of the compo-nent crops in the Limpopo Province.


Two separate field experiments were carried out atthree locations in the Limpopo Province of SouthAfrica — namely, the University of the North exper-imental farm at Syferkuil and a communal farmers’field at Thabina/Dan during the 1998/99 and 1999/2000 growing seasons, and then at Syferkuil and asemi-commercial farmer’s field at Dalmada in the2001/02 growing season. Syferkuil has relativelyhigher soil fertility than on the farmers’ fields atThabina/Dan and Dalmada due to a long history offertilisation at Syferkuil (Table 1). The low soil fer-tility at the farmers’ fields is typical of the small-holder systems in the province. Thabina/Danreceives higher annual rainfall (750 mm) than Syfer-kuil and Dalmada, where the annual rainfall is 500mm and 600 mm, respectively. The experimentalfields were ploughed 2–7 days before planting and 50kg P/ha was applied in the form of single superphos-phate. This was followed by disking to incorporatethe fertiliser at all sites and years. Nitrogen fertiliserswere applied at planting as urea, at 40 kg N/ha, to themaize at Thabina and Dalmada only.

Table 1. Pre-sowing soil analysis results at the experimental sites at all seasons.

Location Season Depth (cm)

pH(H2O)a Nb

(mg/kg)Pc

(mg/kg)Kd

(cmol(+)/kg)

Syferkuil 1998/99

1999/2000

0–15 15–300–15

15–30

7.17.06.87.0

13.514.014.313.2

50.446.544.340.2

0.320.310.300.31

Thabina/Dan 1998/99

1999/2000

0–15 15–300–15

15–30

6.86.35.65.4

8.77.76.95.7

6.44.21.61.1

0.170.170.120.11

Syferkuil 2001/02 0–15 15–30

7.16.9

10.59.5

3327

0.550.53

Dalmada 2001/02 0–1515–30

7.87.3

4.55.0

1311

1.481.43

a 1:5 soil:waterb NH4 + NO3 1:5 Extractant 0.1 N K2SO4.c 1:7.5 extractant Bray 2.d 1:10 extractant ammonium acetate 1 N, pH 7.



108

The 1998/99 and 1999/2000 trials were establishedon 16 and 26 November 1998 at Syferkuil andThabina/Dan, respectively, and then on 6 and 13December 1999 at the two locations, respectively,during the second season. The experiments wereestablished as randomised complete block designswith four replications at each location under drylandconditions. Treatments examined during the 1998/99and 1999/2000 were four cowpea cultivars, namely,Pan 311, Pan 326, Bechuana White and Agrinawa,which were either intercropped in alternate 90 cmrows with maize variety SNK2147 or planted as solecultures. Thus, the cowpea replaced a maize row inan alternate manner in the intercropped plots. Soleculture maize was included as a separate treatment.The final densities were 30,000 and 15,000 plants/hafor the sole and intercropped maize, respectively, and30,000 and 60,000 plants for the intercropped andsole culture cowpea, respectively. Pan 311 and Pan326 are short-duration types whereas BechuanaWhite and Agrinawa are medium- and long-durationcultivars, respectively.

In the 2001/02 trials at Syferkuil and Dalmada, thetreatments included two cowpea cultivars, namely,Glenda and Bechuana White, and two lablab culti-vars, Rongai and Common, grown as sole cultures orintercropped with maize. The trials were establishedon 6 and 10 December 2001 at the two locations,respectively. The lablab cultivars were both long-duration types obtained from a commercial seedcompany in South Africa. Unlike the 1998/99 and1999/2000 trials, where the legume was planted at 90cm inter-row spacing from the maize, the legumes inthe 2001/02 trials were planted between two maizerows spaced 90 cm apart. Thus, the legumes were 45cm from the maize. The maize density in this trialwas 60,000 plants/ha and that of the legume was25,000 plants/ha in both the intercrop and sole cropsystems. The legume seeds in all trials were inocu-lated with the commercial Bradyrhizobium strainCB756 just before planting. The data collected wereagronomic characteristics, seasonal dry-matter pro-duction of the legumes in the 2001/02 season and thegrain yield of the component crops.

Weeding was done twice during the growingseason in all the trials by hoeing and hand-pulling.All the legumes were sprayed once with Metasystoxduring the growing season to control aphids. Days toflowering for all crops were recorded when 50% ofthe plants in a plot had flowered. Physiological matu-rity of the legumes was scored when 90% of the

plants in a plot revealed pods that rattled whenshaken and of maize when 90% of the plants had nomilk line (Stoskopf 1981). Seed yield samples ofmaize and cowpea were taken from the middle rowsof each plot, leaving one row on each side as a borderrow. Harvested maize and the legumes samples wereoven-dried at 65°C to constant moisture content(14%) and then weighed to determine grain yield.


Maize grain yield

In the 1998/99 and 1999/2000 trials, maize grainyield was significantly (p £ 0.05) affected by crop-ping system only in 1999/2000 at Syferkuil. Theyield at this location ranged from 4956–7757 kg/hawith the highest yields occurring in maize inter-cropped with Pan 311, Pan 326 and Agrinawa (Figure1). Maize intercropped with Pan 311 was 41.7%superior to the sole maize whereas intercrops withPan 326, Agrinawa and Bechuana White were similarto the sole maize. Maize yield in 1998/99 averagedacross the five treatments was 2813 kg/ha at Syfer-kuil and 3378 kg/ha at Thabina/Dan. Maize yield in1999/2000 at Thabina/Dan was unavailable due toexcessively high rainfall late in the season, whichprevented access to the experimental site andseverely damaged the crops.

0

2000

4000

6000

8000

M+Pan31

1

M+Pan32

6

M+BC w

hite

M+Agrina

wa

Sole m

aize

See

d yi

eld

(kg/

ha)

ab

bc

a

abc

c

Figure 1. Grain (seed) yield of maize (M) in soleculture and intercropped with cowpeacultivars Pan 311, Pan 326, Bechuana (BC)White and Agrinawa at Syferkuil in the1999/2000 growing season.



109

The highest performance of maize intercroppedwith cowpea cultivar Pan 311 could be due to this cul-tivar being small and generally early maturing — thusoffering minimal competition to the maize. The aboveresults agree with other researchers' findings that yieldadvantages in intercrops may arise when the growthduration of the component crops differs. Each cropwill be exposed to greater resources because they willmake major demands for those resources at differenttimes (Ofori and Stern 1987; Putnam and Allan 1992;Ayisi and Poswall 1997). The lack of significant dif-ference in the grain yield of the sole and intercroppedmaize is an indication that maize could be successfullyintercropped with these cowpea cultivars without sac-rificing the maize grain yield.

Grain yield in the 2001/02 trials, where the legumewas planted between maize at 90 cm inter-rowspacing, was influenced by the cropping system atboth locations. At Syferkuil, the yield of all the inter-cropped maize was severely reduced relative to thesole maize (Table 2). However, at Dalmada, the grainyield of maize intercropped with cowpea cultivarGlenda was similar to the sole crop, whereas the yieldof maize in the other intercrops was lower than that ofthe sole maize. The generally poor grain yield per-formance of the intercropped maize at both locationsindicates that the presence of all legumes in the maizeculture created a severe competitive condition, whichprevented effective growth and yield of the associ-ated maize. Increased competition in an intercrop-ping system could either result from the increased

number of plants per unit area compared to the soleculture, or an effective reduction in resources as bothplant species must ‘share’ them — if the resourcesare lower than the combined demands of the partici-pating species, yield reduction is bound to occur(Vandermeer 1992). Other researchers have reportedthat yield advantages through intercropping mayarise if the growth duration of the component cropsdiffer. Each crop will be exposed to higher resourcesbecause they will make major demands for resourcesat different times (Putman and Allan 1992; Ayisi andPoswall 1997). The generally lower yields of theintercropped maize were due to intense competitionfrom the legume resulting from to narrower inter-rowspacing of the 2001/02 trials.

Cowpea grain yield

Cowpea grain yield was influenced by croppingsystem in both the 1998/99 and 1999/2000 trials atSyferkuil, but only in 1998/99 at Thabina/Dan. AtSyferkuil in 1998/99, the highest grain yield wasrecorded in sole cropped Pan 311, with a yield of 532kg/ha, followed by maize intercrops with Pan 311and Pan 326 (Figure 2). The lowest yields wererecorded in sole and intercropped Bechuana Whiteand Agrinawa. Comparing the effect of croppingsystem on individual cowpea yields, grain yields ofPan 326, Bechuana White and Agrinawa remainedunchanged whether they were intercropped orplanted as sole cultures. However, the yield of Pan311 was reduced by 27% when intercropped. Acrossthe various cropping systems, intercropping reducedcowpea seed yield by an average of 13% relative tothe sole culture at Syferkuil in 1998/99. During the1999/2000 growing season at this location, thehighest yields were recorded in Bechuana Whiteunder both sole and intercropping systems, followedby sole cropped and intercropped Pan 311 andAgrinawa (Figure 3). These results are contrary to theprevious season’s observations where BechuanaWhite and Agrinawa produced the lowest yields —thus indicating the importance of season on cowpeaperformance in the intercropped system. Similar tothe 1998/99 data, the yields of Pan 326, BechuanaWhite and Agrinawa were not dependent on whetherthey were planted as sole crops or as intercrops.However, the yield of Pan 311 was reduced by 53%when intercropped with maize. Yield reduction ofintercropped Pan 311 could be attributed to the

Table 2. Grain yield (kg/ha) of sole and intercroppedmaize at Syferkuil and Dalmada during the2001/02 season (M = maize; BC White =cowpea cv. Bechuana White; Glenda =cowpea cultivar; Rongai and Common =lablab cultivars).

Cropping system Syferkuil Dalmada

Sole MM + BC WhiteM + GlendaM + RongaiM + CommonLSD (p £ 0.05)CV (%)

1238 a172 b330 b172 b44 b43456

1420 a455 b1213 a595 b168 b46232

Note: LSD = least significant difference; CV = coefficient of

variation; means followed by the same letter or letters within a

column are not significantly different from each other (p £ 0.05).



110

increased competitiveness of the associated maizecrop as indicated by its superior yield performance.

At Thabina/Dan, a significant difference wasonly recorded in 1998/99 and not in 1999/2000. In1998/99, the highest seed yield of 1547 kg/ha wasrecorded in sole cropped Pan 311 (Figure 4), whichwas similar to the results at Syferkuil during thesame year. The seed yield of cowpea cultivar Pan311 was reduced by 24% when intercropped.Similar to Syferkuil, yields of cultivars Pan 326,Bechuana White and Agrinawa remainedunchanged whether they were intercropped orgrown as sole crops. According to Ayisi et al.

(2000), Pan 311 is a high-yielding cultivar, but fromthe present study it did not perform very well underthe intercropped system with maize.

Intercropping usually results in increased yield ofthe more competitive component crop with a subse-quent reduction in the weaker crop. However, this wasnot observed under maize intercropped with cowpeacultivars Pan 326, Bechuana White or Agrinawa. Rel-atively high rainfall throughout the growing seasons atboth locations might have favoured the medium- tolong-duration cowpea cultivars to increase yield underthe intercropped system. The consistent performanceof cultivars Pan 326, Bechuana White and Agrinawa

0

500

1000

1500

2000

Seed

yie

ld (k

g/h

a)

c c

abb

b

c

a

b

M+Pan311

M+Pan326

M+BC w

hite

M+Agrin

awa

Pan311

Pan326

BC white

Agrinawa

0

500

1000

1500

2000

Seed

yie

ld (k

g/h

a)

b bc

d d dd

c

a

M+Pan311

M+Pan326

M+BC w

hite

M+Agrin

awa

Pan311

Pan326

BC white

Agrinawa

Figure 3. Seed yield of cowpea cultivars Pan 311, Pan 326, Bechuana (BC) Whiteand Agrinawa as sole crops and intercropped with maize (M) at Syferkuilin the 1999/2000 growing season.

Figure 2. Seed yield of cowpea cultivars Pan 311, Pan 326, Bechuana (BC) Whiteand Agrinawa as sole crops and intercropped with maize (M) at Syferkuilin the 1998/99 growing season.



111

indicates their ability to maintain high productivityunder intercropping situations and hence their poten-tial as intercrop varieties in the maize–legume inter-cropped system. However, the inconsistency of thecultivars from season to season needs to be addressedthrough additional experimentation before final rec-ommendations are made. Our results also indicate thatmaize can be intercropped with some of the cowpeacultivars in this particular system with no yield sacri-fice of either crop, since seed yield was not affected byintercropping compared to sole cropping.

During the 2001/02 trials, the seed yield of cowpearesponded significantly (p £ 0.05) to cropping systemat Dalmada but not at Syferkuil (Table 3). At Dal-mada, the highest seed yields were recorded in solecropped cultivar Glenda and sole Bechuana White.

On average, the seed yield of the cowpea varietieswas 56% lower in the intercrops than in sole crops. Thegreatest reduction occurred in intercropped Glenda inwhich the yield was only 27% that of the sole cropyield. The observation that intercropping maize andgrain legumes leads to severe depression of the legumeyield is in agreement with reports by Searle et al.(1981) and Siame et al. (1998) who observed similarlower legume yields in maize–cowpea intercrop trialsand also by Ezumah et al. (1987) and Ofori and Stern(1987) in maize–soybean intercrops. Yield depressionof both component crops, such occurred in this study,suggested severe moisture and nutrient deficiencyconditions. No grain was produced by lablab duringthe course of experimentation.

Flowering and physiological maturity of maize

No significant differences in days to flowering andphysiological maturity of maize due to croppingsystem was observed at either location during the1998/99 and the 1999/2000 trials. Maize floweringdate ranged from 64 to 71 days after planting (DAP)and that of maturity from 122 to 128 DAP across theseasons and locations.

Flowering and physiological maturity of cowpea

Days to flowering in the cowpea cultivars rangedfrom 54 to 112 DAP and 63 to 84 DAP at Syferkuil

0

500

1000

1500

2000

M+Pan

311

M+Pan

326

M+BC w

hite

M+Agr

inawa

Pan31

1

Pan32

6

BC whit

e

Agrina

wa

See

d yi

eld

(kg/

ha)

bbb

cc c

c

a

Figure 4. Seed yield of cowpea cultivars Pan 311, Pan 326, Bechuana(BC) White and Agrinawa as sole crops and intercroppedwith maize (M) at Thabina/Dan in the 1998/99 growingseason.

Table 3. Seed yield (kg/ha) of sole and intercroppedcowpea cultivars Glenda and Bechuana (BC)White at Syferkuil and Dalmada during the2001/02 season.


Sole BC WhiteSole GlendaIntercropped BC WhiteIntercropped GlendaLSD (p £ 0.05)CV (%)

8851009938740NS27

878 a1010 a543 b280 c18514

Note: LSD = least significant difference; CV = coefficient of variation; NS = not significant (p £ 0.05); means followed by the same letter or letters within a column are not significantly different from each other (p £ 0.05).



112

for the 1998/99 and 1999/2000 growing seasons,respectively (Table 4). At Thabina/Dan (1999/2000),flowering of the cowpea cultivars ranged from 51 toabout 70 DAP (Table 4). Based on days to physiolog-ical maturity, two groups of varieties could be iden-tified, the early (Pan 311 and Pan 326) and medium tolate (Bechuana White and Agrinawa) maturing types.Intercropping generally did not influence floweringor maturity in the cowpea cultivars.

During the 2001/02 trials, differences in days toflowering in maize due to cropping system werealso not observed at the two locations. Floweringdate ranged from 70 to 74 DAP at Syferkuil andfrom 74 to 79 DAP at Dalmada. The lack of a sig-nificant difference in days to flowering in maize isan indication that the presence of legume in the

maize microenvironment did not alter the phe-nology of the crop.

Flowering and physiological maturity of cowpea

Similar to the previous season’s trials, differencesin days to flowering were observed among thecowpea and lablab cultivars (p £ 0.05) at both Syfer-kuil and Dalmada (Table 5).

Flowering of the legume cultivars ranged from 54to 133 DAP at Syferkuil. The cowpea cultivarsGlenda and Bechuana White flowered between 54and 64 DAP, which is in the range reported by Ayisiet al. (2000) for several cowpea cultivars at theselocations. The lablab cultivars Rongai and Commontook longer to flower; from about 129 to 133 DAP.

Table 4. Flowering and physiological maturity (days after planting) of sole and intercropped cowpea cultivars(Pan 311, Pan 326, Bechuana (BC) White and Agrinawa) at Syferkuil and Thabina/Dan in 1998/99 and1999/2000.

Cropping system Syferkuil 1998/99 Thabina/Dan 1998/99 Syferkuil 1999/2000

Flowering Maturity Flowering Maturity Flowering Maturity

Sole Pan 311Sole Pan 326Sole BC WhiteSole AgrinawaIntercropped Pan 311Intercropped Pan 326Intercropped BC WhiteIntercropped Agrinawa

60.6 b70.3 b118.6 a111.9 a63.1 b54.8 b110.0 a111.9 a

83.0 bc84.0 bc162.4 a164.2 a82.0 c85.0 b162.4 a163.9 a

54.3c51.0c69.0a62.8b52.3c53.3c69.7a61.5b

90.5 c93.2 c

112.5 a107.5 ab

89.0 c90.0 c

111.5 bc106.7 b

63.1b65.1b84.0a80.4a63.3b64.0b79.1a81.8a

113.5 bc118.5 b123.9 a126.9 a109.3 c

113.8 bc123.9 a125.4 a

Note: means followed by the same letter or letters within a column are not significantly different from each other (p £ 0.05)

Table 5. Flowering and physiological maturity (days after planting) of cowpea cultivars (Bechuana (BC)White and Glenda) and lablab cultivars (Rongai and Common) as influenced by the croppingsystem at Syferkuil and Dalmada during the 2001/02 growing season.



Sole BC WhiteSole GlendaSole RongaiSole CommonIntercropped BC WhiteIntercropped GlendaIntercropped RongaiIntercropped CommonLSD (p £ 0.05)

64.7 b59.3 bc130.7 a130.3 a55.3 c54.0 c

133.0 a129.7 a

9.2

93.396.7

––

93.392.4

––

NS

65.7 c61.0 c

114.3 b124.3 ab

64.3 c64.0 c

124.3 ab126.0 a

10.1

93.393.3

––

90.096.7

––

NS

Note: statistical significance for physiological maturity was tested only in cowpea, as the lablab was killed by frost before it matured; LSD = least significant difference; NS = not significant (p £ 0.05); means followed by the same letter or letters within a column are not significantly different from each other (p £ 0.05).



113

No differences in maturity of cowpea cultivars wasobserved and lablab matured before being killed byan early season frost at 146 DAP.

Total biomass accumulation of the legumes

A significant effect of cropping system on total dryweight of the legumes was observed at all samplingdates except 67 DAP at Syferkuil (Table 6). Biomassaccumulation by 67 and 71 DAP at the two locations,respectively, generally did not differ much betweenthe sole crops and intercrops for the individual leg-umes. However, at 54 DAP, for cv. Glenda, total dryweight was enhanced under intercrop by about 60%relative to the sole crop.

Conclusions

Intercropping maize with cowpea cultivar Pan 311enhanced maize yield over the sole culture yield, butthe yield of the legume was reduced by 27% whenintercropped compared to the sole crop of thiscowpea. From our study, it is clear that cowpea cul-tivars Pan 326, Bechuana White and Agrinawa couldbe successfully intercropped with maize in alternate,single 90 cm rows without sacrificing grain yields ofthe two crops in the system. Differences in days toflowering and physiological maturity due to croppingsystem were observed between cowpea cultivars butnot in maize. The early-flowering cowpea cultivarswere Pan 311 and Pan 326. In general, days to flow-ering days to maturity of all the cowpea cultivars did

not vary between the sole cultures or the intercropswith maize.

When cowpea and lablab were planted betweentwo rows of maize spaced at 90 cm, seed yields ofboth the maize and the cowpeas were generallyreduced compared to the sole crops. However, whenmaize was intercropped with cowpea cultivarGlenda, the maize yield was comparable to the solecrop at one location but not the other. Biomass accu-mulation of the legumes generally did not differbetween the crops grown in sole culture and those inintercropped systems. Further studies involvingvarying densities of the legumes will help identify thebest combinations of the cereal and the legumes forintercropping.

References

Ayisi, K.K. and Poswall, M.A.T. 1997. Grain yield potentialof maize and dry bean in a strip intercropping system.Applied Plant Science, 11(2), 56–58.

Ayisi, K.K., Nkgapele, R.J. and Dakora, F.D. 2000. Noduleformation and function in six varieties of cowpea (Vignaunguiculata L. Walp.) grown in a nitrogen rich soil inSouth Africa. Symbiosis, 28, 17–31.


Ezumah, H.C., Nam Ky Nguyen and Walker, P. 1987.Maize–cowpea intercropping as affected by nitrogenfertilization. Agronomy Journal, 79, 275–280.

Table 6. Total dry-matter accumulation (kg/ha) of cowpea cultivars Bechuana (BC) White and Glenda and lablabcultivars Rongai and Common at Syferkuil and Dalmada, as influenced by cropping system during the2001/02 growing season (DAP = days after planting).

Cropping system Syferkuil54 DAP

Syferkuil67 DAP

Syferkuil88 DAP

Dalmada50 DAP

Dalmada71 DAP

Dalmada95 DAP

Sole BC WhiteSole GlendaSole RongaiSole CommonIntercropped BC WhiteIntercropped GlendaIntercropped RongaiIntercropped CommonLSD (p £ 0.05)CV (%)

1172 ab834 b

1079 b751 b643 b1744 a1130 b666 b58733

25142438277819403378250520523131NS34

5750 a5160 ab5562 a5488 a2165 b3442 ab3539 ab4936 ab

327141

771 abc971 a

726 abc558 abc862 ab403 c

462 bc484 bc

44639

3312 ab4000 ab3013 ab3210 ab4161 a

2375 ab1865 b2740 ab

226342

2268 bc2982 bc5709 a

3902 ab978 c

1629 c2648 bc2854 bc

226145

Note: LSD = least significant difference; CV = coefficient of variation; NS = not significant (p £ 0.05; means followed by the same letter or letters within a column are not significantly different from one another (p £ 0.05).



114

Ofori, F. and Stern, W.R. 1987. Cereal–legume intercroppingsystems Advances in Agronomy, 41, 41–90.

Putnam, D.H. and Allan, D.L. 1992. Mechanism of over-yielding in sunflower/mustard intercrop. AgronomyJournal, 84, 188–195.

Searle, P.G.E., Comudom, Y., Sheddon, D.C. and Nance,R.A. 1981. Effect of maize + legume intercroppingsystems and fertilizer nitrogen on crop yields and residualnitrogen. Field Crops Research, 4, 133–145.

Shackel, K.A. and Hall, A.E. 1984. Effect of intercroppingon the water relations of sorghum and cowpea. FieldCrops Research, 8, 381–387.

Shumba, E.M, Dhliwayo, H.H. and Mukoko, O.Z. 1990.The potential of maize–cowpea intercropping in lowrainfall areas of Zimbabwe. Zimbabwe Journal ofAgricultural Research, 28, 32–38.

Siame, J., Willey, R.W. and Morse, S. 1998. The response ofmaize/Phaseolus intercropping to applied nitrogen onoxisols in northern Zambia. Field Crops Research, 55,73–81.

Vandermeer, J. 1992. The ecology of intercropping.Cambridge, UK, Cambridge, University Press.



115

Grain Sorghum Production and Soil Nitrogen Dynamics Following Ley Pastures on a Vertosol

in Queensland, Australia

A.M. Whitbread* and R.L. Clem†

Abstract

Highly productive sown pasture systems can result not only in high growth rates of beef cattle, but also leadto increases in soil nitrogen and the production of subsequent crops. The response of grain sorghum to variousperiods of annual legumes and grasses were investigated in a no-till system in the South Burnett district ofQueensland. Two-to-four years of the tropical legumes Macrotyloma daltonii and Vigna trilobata (two self-regenerating annual legumes) and Lablab purpureus (a resown annual legume) resulted in soil nitrate nitrogen(0–90 cm depth) ranging from 36 to 102 kg/ha compared with 5 to 11 kg/ha after grass only pastures. Grainsorghum produced in up to four crops following the legume pastures exceeded 3000 kg/ha of grain in eachseason. Simulation studies utilising the farming systems model APSIM (Agricultural Production SystemssIMulator) indicated a high correlation with observed grain and biomass data (r2 > 0.68) and soil N dynamics.In simulated sorghum crops (1954–2000), grain yield was unlikely to exceed 2000 kg/ha of grain in 60% ofseasons following a grass pasture while following 2-year legume leys, grain exceeded 3000 kg/ha in 80% ofseasons. It was concluded that mixed farming systems that utilise short-term, legume-based pastures for beefproduction in rotation with crop production enterprises can be highly productive.

In the northern grain belt of Australia, which takes innorthern New South Wales and southern and centralQueensland, lower crop yields and grain proteincontent as a result of declining soil fertility haveprompted many farmers to consider the role of leypastures as a source of forage and a way to improvesoil fertility. The traditional winter-growing legumespecies of medic and lucerne have been used as leypastures in the southern areas of Queensland wheresome winter rainfall is received. However, in thecentral Queensland areas, most of the rainfall isreceived during the hot summer months and tropicalspecies of legumes are better adapted.

The increasing interest in legumes for use infarming and grazing systems is generated by theincreasing need to provide nitrogen to the soil for sub-sequent crops in environmentally acceptable ways,and the need to provide high-quality forage to increasegrowth rates of stock to produce beef of better eatingquality. Both longer-term perennial legumes that canpersist with grass and forage legumes used in shorter-term leys can provide high-quality forage to finishcattle for premium beef markets.

The duration, species selection and managementof pastures will depend on the soil type and farmingenterprises. Farmers located on cropping countrywhere animals are of minimal importance may lookto ley pastures to provide soil fertility inputs (phys-ical, biological and chemical) for subsequent crop-ping activities, while farmers involved with mixed-farming operations can utilise the leys as valuable

* CSIRO Sustainable Ecosystems, 306 Carmody Road, StLucia, Queensland 4067, Australia.

† Queensland Department of Primary Industries, Gympie, Queensland, Australia.



116

forage as well as for improving the soil of the crop-ping country. They are more likely to economicallyjustify pasture for longer durations and offset thelosses in grain production with animal production.

A number of summer-growing legumes are nowknown to be well adapted to lower rainfall (700 to800 mm MAR — mean annual rainfall) areas incentral and southern Queensland. They have a rangeof characteristics that could complement or providealternatives to the annual lablabs and cowpeas andinclude such legumes as Milgarra butterfly pea (Clit-oria ternatea cv. Milgarra), Endurance lablab(Lablab purpureus cv. Endurance) and burgundybean (Macroptilium bracteatum cvv. Cardaga andJuanita). Other self-regenerating legumes such asVigna trilobata and Macrotyloma daltonii, tested bythe Queensland Department of Primary Industriesand the Commonwealth Scientific and IndustrialResearch Organisation (CSIRO) for many years,remain unreleased but may provide useful germ-plasm in the future.

The study described in this paper investigates thedynamics of soil nitrogen and the production of grainsorghum following various ley legume pastures. Usingthese data to validate the farming systems modelAPSIM (Agricultural Production Systems sIMulator)potential production of grain sorghum across a rangeof seasons was also analysed. The animal productionphase of this trial is described in Clem (2004).

Methodology

Soil type

The experiment was conducted on the Brian Pas-tures Research Station on a moderately to stronglyself-mulching Vertosol (black earth) on a 3–10%slope, Ug5.34 (Northcote 1979) with weak linearGilgai (Reid et al. 1986). The three legume treat-ments were situated lower on the slope and had a soil

depth of 0.9–1.2 m, a plant available water capacity(PAWC) of 171 mm (Table 1) and total organiccarbon of 1.7% and 1.4% at 0–15 cm and 15–30 cmdepth, respectively. The soil on which the grasspasture treatment was situated was only 90 cm deepwith a PAWC of 115 mm and total organic carboncontents of 2% and 1.5% at 0–15 cm and 15–30 cmdepth, respectively.

Pasture phase

From the pastures established in the summer of1997/98 and described by Clem (2004), four pasturetreatments were selected to test the response ofsorghum production after the termination of these leyphases. The treatments selected were the improvedgrass pasture (Bothriochloa insculpta cv. Bisset,Dichanthium sericeum and Panicum maximum var. tri-choglume cv. Petrie), the annual legume pastureLablab purpureus cv. Highworth, and the self-regener-ating legume pastures Macrotyloma daltonii CPI60303 and Vigna trilobata CPI 13671 systems. Subse-quent to the initial pasture plantings in 1997/98, L. pur-pureus had been resown on 7 December 1998, 4January 2000 and 13 December 2000.

Sorghum phase

In November 1999, following a heavy grazing bycattle, approximately one-third of the 2 ha replicatedtreatments previously sown to M. daltonii, V. trilo-bata and L. purpureus and a smaller area (~500 m2)in the grass pasture was sprayed with herbicide, keptfree of weeds and planted to sorghum at the datesindicated in Table 2. Similarly, in the followingseason, these same areas and a further one-third of theplots were grazed, sprayed with a knockdown herbi-cide and planted to sorghum. In the fourth season(2002/03), only the M. daltonii and L. purpureustreatments were continued. The areas that hadalready been planted with sorghum for the past one or

Table 1. Soil properties of the experimental site near Mt Brambling, Brian Pastures.

Depth (cm) Bulk density(g/cm3)

Drained upper limit (%)

Sorghum lower limit (%)

Saturated (%) Plant available water (mm)

0–1515–3030–6060–9090–120

1.231.291.281.301.44

4543444338

2526242428

4846474641

3125595728



117

two seasons, and the remaining third of the plotswhich were still sown to pasture were heavily grazed,sprayed and planted with sorghum.

Within each of the areas sown to sorghum, a rep-resentative test area (10 ¥ 30 m) was selected towhich all crop and soil measurements were confined.These test areas were divided into two plots (10 ¥ 15m) and one-half received 80 kg of N/ha as urea in twosplits (Table 2) during the sorghum crops.

Three additional plots (10 ¥ 5 m) established in theM. daltonii test areas received 20, 40 and 120 kg/haof N applied as urea in a split application at the datesindicated in Table 2. All fertiliser was broadcast byhand to the soil surface. These plots were used to testthe sorghum N response in each season.

Grain sorghum was planted into the legume andsorghum stubble that remained in the plots aftergrazing using a no-till planter. No estimate of theremaining residue at sowing was made. In the 1999/2000 season, sorghum (Pioneer S75) was planted at3.2 kg/ha into 25 cm rows. In the three subsequentseasons, sorghum was planted at 3.0–3.6 kg/ha into90 cm row spacings using sorghum variety BusterMR in 2000/01 and Pioneer M43 in 2001/02 and2002/03.

Cattle grazed the sorghum residues remaining afterharvest. Sorghum regrowth was sprayed with aknockdown herbicide and after the withholdingperiod, cattle grazed the dead plant residues.

At physiological maturity, sorghum grain andstover were harvested for the determination of yieldand nitrogen concentration. Two separate, centrallylocated 5 m rows within each plot were selected andthe number of plants and grain heads were counted.The heads were cut from the plants, removed and theremaining stover was cut at ground level andremoved. Stover was dried in a forced-air dehydratorat 80°C for 48 hours and weighed to determine drymatter. The sorghum heads were dried, threshed toremove grain and weighed. The panicle straw weightwas added to the stover weights.

Soil samples were taken in each plot beforeplanting and following harvest of each crop and ana-lysed for nitrate nitrogen and water content. Allsamples were taken using a hydraulic soil samplerwith cores to 90 cm being sampled and then parti-tioned into 0–15, 15–30, 30–60 and 60–90 depthincrements for analysis. Soil nitrate nitrogen wasdetermined using the method of Keeny and Nelson(1982) described in Rayment and Higginson (1992).

APSIM set-up

Soil was characterised (Table 1) following the pro-cedures of Dalgliesh and Foale (1998). The drainedupper limit (dul) was determined by slowly wettingthe soil to saturation, allowing drainage to take placeover 2–3 weeks and determining the moisture con-tent. Bulk density was calculated at the dul using therelationship that exists between measured bulkdensity and gravimetric moisture content describedby Gardner (1988) and referred to by Hochman et al.(2001). The crop lower limit (cll) for sorghum wasdetermined by placing a rain-out shelter over asection of the 1999/2001 sorghum crop at floweringand determining the soil water content after the crophad reached maturity.

Simulation of measured sorghum phases

Daily climatic records from the research station(maximum and minimum temperatures, rainfall,solar radiation) were used to simulate sorghumgrowth. All management parameters in the modelmimicked the actual field experimentation. Analysisof the modelled data was confined to the treatmentswhere two seasons of legume pasture had precededthe sorghum phase and only the 1999/2000, 2000/01and 2001/02 sorghum crops were simulated. Thesorghum simulations in the 1999/2000, 2000/01 and2001/02 seasons were preformed using the actualplanting dates, the timing of nitrogen fertiliser appli-cations and management information (Table 2). Noattempt was made to include weeds in the model sim-

Table 2. Timing (dates) of cropping operations for the sorghum test phase.

Season Spraying Planting N split 1 N split 2 Harvest

1999/20002000/20012001/20022002/2003

12 Nov 9913 Nov 0017 Aug 0212 Dec 02

7 Jan 003 Jan 01

13 Dec 0112 Feb 03

28 Jan 009 Feb 0124 Jan 027 Mar 03

3 Mar 0012 Mar 0013 Feb 0216 Apr 03

18 Apr 003 May 0122 Apr 0216 Jun 03



118

ulations and it was assumed that the low weed popu-lations in the field during the sorghum crops did notlimit growth.

Long-term simulations

To consider the long-term effect of seasonal condi-tions on sorghum production in the first season fol-lowing the termination of a pasture, APSIM wasinitialised with the soil water and mineral nitrogenconditions that were measured following each of thepasture treatments on Julian day 319 (November 15)each year. Simulations were then run from 1954–2001using the Brian Pastures long-term weather data. Eachyear, the sorghum crop was sown between November15 and January 15 when at least 30 mm of rain wasreceived over five consecutive days. As all soil condi-tions were reset to the same parameters each season,the effect of climate on sorghum grain production waspredicted following each pasture treatment. The grasspasture treatment was also included with the sameprofile depth and PAWC as the soils on which thelegume treatments were established.


The depth of soil under the grass treatment wasfound to be about 30 cm less than the other treat-ments. Sorghum production on this treatment couldnot be compared with that of the other treatments andwas therefore not included in the statistical analysisof sorghum production.

In the first year of the sorghum phase, data wereanalysed as a fully randomised factorial design of 3previous pasture treatments ¥ 2 fertiliser rates ¥ 2replicates. In subsequent years, the number of pre-vious seasons to sorghum was also included as a fac-torial treatment in the design. Mean separation wastested using Duncan’s multiple range test (DMRT) atP £ 0.05. All data were tested for the assumption ofcommon variance and transformed if necessary.

Total soil nitrate nitrogen data were found to havelarge difference in variance between replicates of dif-ferent treatments and were statistically analysed afterlog transformation.

Results

Seasonal conditions

The previous legume pasture did not significantlyinfluence soil water measured prior to sowing of thesorghum crops. The soil profile (0–90 cm) containedbetween 34 and 56% of the total PAWC (171 mm) atsowing over the four seasons (data not shown). Rain-fall was generally near or below average over the fivegrowing seasons (Table 3). The in-crop rainfall was138 mm in 1999/2000, 222 mm in 2000/01, 276 mmin 2001/02 and 232 mm in 2002/03.

Sorghum production

Sorghum grain yield production in all years washighest following the L. purpureus pasture, howeverthis was statistically significant only in the 2000/01 and2002/03 sorghum crops (Table 4). Total biomass pro-duction (grain + stover) in the 2000/01 sorghum cropwas highest in the L. purpureus treatment, and in 2001/02 was significantly lower in V. trilobata treatmentsthan M. daltonii and L. purpureus treatments (Table 5).

There was no response in grain or biomass yield tonitrogen fertiliser application (0 or 80 kg N/ha) in anyof the treatments or any of the years so the data pre-sented in Tables 4 and 5 are averages of these ferti-liser treatments. There was also no response insorghum production to the additional nitrogen ferti-liser treatments (0, 40, 120 kg/ha N) included withinthe M. daltonii treatments.

The sorghum grown on the grass treatment yieldedvery poorly in the first year (1999/2000) and con-

Table 3. Monthly rainfall received at Brian Pastures Research Station from July 1998 and the long-term average(1963–2002).

Year Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Year

1998/991999/002000/012001/022002/03Average

2284

546

035

2614

31

9129

11933

038

029

21107171

336461

107118

70155

2470

9053665258

105

3578

718

0110

5161

138164113

94

4927

120606371

53332

56137

302746494340

3726

1701029

592661659691527710



119

tinued to yield less than the other treatments in sub-sequent seasons (Table 4 and 5).

There was a decline in sorghum grain yield with anincrease in the number of preceding sorghum phases.This decline became significant in the final sorghumphase where grain yield was 18% lower in the thirdyear of sorghum and 27% lower in the fourth year ofsorghum compared with the treatment where one pre-vious sorghum phase had been grown (Table 6).There was no significant effect of the precedinglegume species so the data shown in Table 6 are themean of these treatments.

Soil nitrate

The effect of pasturesSoil nitrate measured after the termination of the

two, three or four seasons of pasture was significantlylower following the grass treatments than following

the legume treatments (Table 7). There was no signif-icant difference in soil nitrate found between thelegume treatments and there was no significantincrease in soil nitrate with additional seasons oflegume growth (tested using the repeated measuresanalysis of pasture treatment ¥ replicate ¥ year).

Residual value in the sorghum phaseIn the first season following 2 years of legume pas-

tures, soil nitrate nitrogen ranged from 36 to 86 kg Nin the profile (0–90 cm) and declined to virtually zerofollowing four sorghum crops (Figure 1). Althoughthe amount of nitrogen mineralised between theharvest of a sorghum crop and the sowing of the nextsorghum crop was similar in each year (averagedacross the legume treatments it was 26, 43, 39 kg/hanitrate nitrogen in 2000, 2001 and 2002, respec-tively), the starting soil nitrate successively declinedwith the removal of nitrogen in each sorghum crop

Table 4. Grain production (kg/ha) of the sorghum phases over the four seasons.

Pasture phase 1999/2000 2000/01 2001/02 2002/03

Macrotyloma daltoniiVigna trilobataLablab purpureusGrassa

2825 a2401 a4098 a995 (244)

3641 b3860 b5362 a2405 (339)

3053 a3113 a3738 a3053 (1159)

3352 b–

4004 a–

a Brackets indicate standard error of the mean in the grass treatments only.Note: means followed by a different letter within columns are significantly different according to Duncan’s multiple range test at P £ 0.05.

Table 5. Biomass production (kg/ha) of the sorghum over the 3 seasons.

Pasture phase 1999/00 2000/01 2001/02 2002/03

Macrotyloma daltoniiVigna trilobataLablab purpureusGrassa

6357 a6182 a

10407 a4477 (919)

11243 b11612 b16706 a

7379 (1923)

11499 a7178 b10662 a

6635 (1068)

6949 a–

8369 a–

a Brackets indicate standard error of the mean in the grass treatments only.Note: means followed by a different letter within columns are significantly different according to Duncan’s multiple range test at P £ 0.05.

Table 6. Sorghum grain yield in areas preceded by legume pastures and 1, 2 and 3 years of sorghumcrops.

Sorghum yield 1999/2000 2000/01 2001/02 2002/03

First yearSecond yearThird yearFourth year

3108 4562 a4013 a (12%)

3738 a3113 a (17%)3053 a (18%)

–4318 a3556 b (18%)3159 b (27%)

Notes: means followed by a different letter within columns are significantly different according to Duncan s multiple range test at P £ 0.05. Percentages shown in brackets in bold: Decline in grain yield = (grain yield of new sorghum phase — grain yield of older sorghum phase)/grain yield of new sorghum phase.



120

(Figure 1). Soil nitrate nitrogen declined from thepresowing to postharvest measurement in all seasonsdue to the uptake of nitrogen in sorghum.

The application of 80 kg/ha of fertiliser nitrogenhad a significant effect on soil nitrate at the harvestsamplings in 2000/01 and 2001/02. At harvest in the2000/01 season, total soil nitrate nitrogen increasedfrom 15 kg/ha with no fertiliser to 74 kg/ha with fer-tiliser applied in the two previous seasons. Similarlyat the 2001/02 harvest sampling, soil nitrate nitrogenincreased from 7 kg/ha to 48 kg/ha with nitrogen fer-tiliser additions. By the final harvest, soil nitrate wasnegligible in the treatments that did not receive

nitrogen fertiliser and 24 kg N/ha was available in thetreatments that had received the yearly fertilisernitrogen application of 80 kg/ha.

A significant relationship was found between soilnitrate at sowing and grain yield in 1999/2000 and2000/2001 of the treatments that did not receive fer-tiliser nitrogen, indicating that grain yield wasincreased with higher soil nitrate nitrogen followingthe pasture treatments (Figure 2). Similar correla-tions were not found in other years or between othertreatments, including the fertiliser comparisons. Thissuggests that soil nitrate nitrogen was limitingsorghum yield, although the nitrogen rate trials didnot indicate this effect.

APSIM simulation

The observed and predicted data for the sorghumcrops (1999/2000, 2000/01 and 2001/02 seasons) thatwere grown on the area that had been sown to legumepasture for two seasons were compared and an r2 =0.68 was achieved for the simulation of sorghumgrain yield (Figure 3) and an r2 = 0.79 was achievedfor the simulation of sorghum biomass (Figure 4).

Figure 5 displays the APSIM predicted soil nitratecontent for the L. purpureus and V. trilobata zeronitrogen treatments from November 1999 until May2002. The soil nitrate values in the model are reset atthe measured pre-sowing value each year. With the

Table 7. Soil nitrate (kg N/ha) measured prior tosowing sorghum following two, three or fourseasons of pasture.

Years of pasture 2 3 4

Date sampled 24 Nov 1999

30 Nov 2000

6 Dec 2001

Macrotyloma daltonii 64 a 67 a 83 a

Vigna trilobata 36 a 40 a 46 a

Lablab purpureus 86 a 102 a 69 a

Grass 5 b 3 b 11 b

Note: means followed by a different letter within columns are significantly different according to Duncan s multiple range test at P £ 0.05; tested on log transformed data.

0

25

50

75

100

Nit

rate

N to

90

cm (k

g N

/ha)

M. daltoniiV. trilobataL. purpureusP. maximum

LSD p < 5%

99/00Sowing Harvest Harvest Harvest Harvest

00/01Sowing

01/02Sowing

02/03Sowing

Figure 1. Total soil nitrate nitrogen (0–90 cm) measured before sowing andpostharvest in each season from 1999/2000 to 2002/03 in the treatmentsthat received no additional nitrogen fertiliser.



121

exception of the year 1 harvest nitrate value fol-lowing the lablab pasture, each of the measuredvalues of nitrate at harvest is in close agreement withthe simulated data. There is a general decline in theamount of nitrate in the profile after harvest with eachsucceeding sorghum crop. This indicates the deple-

tion of nitrate nitrogen as shown in the field data inFigure 1.

Resetting soil water and soil nitrate to the samevalues that were measured in 1999 after 2 years of thepasture treatments allowed the effect of seasonal con-ditions experienced in each season (1954–2001) to

y = 26x + 2110

r2 = 0.52

0

1000

2000

3000

4000

5000

6000

7000

0 50 100 150

Nitrate-N at sowing

Gra

in y

ield

(kg

/ha)

0

1000

2000

3000

4000

5000

6000

7000

0 1000 2000 3000 4000 5000 6000 7000

Pred

icte

d g

rain

yie

ld (k

g/h

a)

Observed grain yield (kg/ha)

y = 1.301x – 369

r2 = 0.68

Figure 2. The relationship between nitrate nitrogen measured atsowing on the treatments that did not receive fertilisernitrogen and grain yield in the 1999/2000 and 2000/01sorghum crops.

Figure 3. The observed and predicted data for sorghum grainyield in the 1999/2000, 2000/01 and 2001/02 seasons.



122

determine the sorghum growth (Figure 6). The prob-ability of achieving no grain production following agrass pasture occurred during 15% of seasons. Fol-lowing a grass pasture, grain production did notexceed 3000 kg/ha and approximately 60% of thetime, grain production was between 1000 and 2000kg/ha. Significantly higher grain yields wereachieved following the legume pastures with muchlower risk levels. Grain production was slightly morerisky following the L. purpureus pasture, with 20%of seasons producing 1500–2600 kg/ha of grain com-pared with less than 10% for V. trilobata and M. dal-tonii. Grain production was between 3000 and 5300kg/ha in at least 80% of the seasons between 1954and 2001 following all of the legume pastures.

Discussion

Sorghum production

Legume treatmentsIn the 10 seasons prior to 1992–1993, average

sorghum production in the South Burnett was 2.3 t/haand ranged from 1.6 t/ha in 1983 to 2.6 t/ha in 1987and 1989 (Crosthwaite and Harvey 1992). In the trialdescribed here, sorghum production exceeded thisaverage in all treatments following the legume pas-

tures and grain production was exceptional followingthe L. purpureus pastures in most seasons (Table 4).

Sorghum production declined as the number ofcrops since a legume pasture increased. This declinewas 12% in the second sorghum phase and increasedto 27% after four sorghum phases (Table 6). Theminimum yield remained above 3053 kg/ha (Tables 4and 6), which is considered high by local standards.While a strong relationship between soil nitratenitrogen at sowing and grain yield was found in theunfertilised treatments in the first 2 years of thesorghum production (Figure 2), this relationship didnot exist in later crops. As soil nitrate nitrogendeclined at each successive sowing, so too did grainyield. While the application of 80 kg/ha of nitrogenas urea significantly increased the after-harvest soilnitrate concentrations, there was no effect on grainyield as soil nitrogen declined with successivesorghum crops. Fertiliser was applied by hand to thesoil surface and may not have been as effective as ifapplied to the root zone at or before planting, as is theusual farmer practice.

Grass treatmentsSorghum growth in the first and second season fol-

lowing grass pastures was very poor and mainly theresult of very low soil nitrate nitrogen concentrations

0

2000

4000

6000

8000

10000

12000

0 2000 4000 6000 8000 10000 12000

Pred

icte

d b

iom

ass

(kg

/ha)

Observed biomass (kg/ha)

y = 0.7866x + 1322.9

r2 = 0.7884

Figure 4. The observed and predicted data for sorghum biomass in the1999/2000, 2000/01 and 2001/02 seasons.



123

present at sorghum sowing. Nitrogen deficiency ingrassland pastures has been well documented in theBrigalow landscapes of central Queensland by Rob-ertson et al. (1993a,b, 1994). Robertson et al. (1993a)found that up to 20 t/ha of surface litter and root res-idues of high C:N ratio accumulated under these pas-tures. Using incubation techniques, these authorsfound that pasture soil respired more CO2 and miner-

alised less nitrogen than cultivated soils. Since theC:N ratio of pasture residues is high (>50), immobi-lisation of mineral nitrogen occurs.

In the system described in this study, sorghum wasestablished using a no-till planter. This results inminimal soil disturbance and slow decomposition ofsurface and buried plant residues. The highest soilnitrate nitrogen concentrations were found at the sowing

0

20

40

60

80

100

120

140

Nov-99 Feb-00 Jun-00 Oct-00 Feb-01 Jun-01 Oct-01 Feb-02

Nov-99 Feb-00 Jun-00 Oct-00 Feb-01 Jun-01 Oct-01 Feb-02

Nit

rate

nit

rog

en (0

–90

cm) (

kg N

/ha)

0

20

40

60

80

100

120

140

Nit

rate

nit

rog

en (0

–90

cm) (

kg N

/ha)

(a)

(b)

Figure 5. Soil nitrate (0–90 cm) during a cropping phase preceded by a) Lablabpurpureus and b) Vigna trilobata. Solid line is the simulated value,The open points are the measured soil nitrate value before sorghumsowing and APSIM (Agricultural Production Systems sIMulator) hasbeen reset to this value. The closed points are the measured soil nitratevalue after harvest.)



124

of the 2001/02 sorghum phase (Figure 1) indicating thatthe mineral nitrogen was becoming available after along period of immobilisation. Also contributing to thelower sorghum yield in this treatment was the shallowersoil profile and a PAWC of 115 mm compared with 171mm found in the other treatments.

Nitrogen dynamics

Since soil samples were not available from theseexperiments before the legume pastures were sown,the soil nitrogen status is not known. Before thelegume pastures, successive sorghum crops and grassfodder crops had been planted for more than 20 years,so it could be reasonably assumed that soil nitrogenwas depleted as studies on similar soil types haveshown (Dalal and Mayer 1986; Whitbread et al.1998). Soil nitrate nitrogen content (0–90 cm) rangedfrom 36 to 86 kg/ha after 2 years of legume pasturesand significant increases in these concentrations wasnot found with further years of pasture. The studies ofArmstrong et al. (1999) on vertosol soils in centralQueensland found the percentage of nitrogen derivedfrom atmosphere (%Ndfa) to be strongly negativelycorrelated to the amount of soil nitrate in the profilefor two annual and two perennial legumes, includingL. purpureus. They also found %Ndfa in L. pur-pureus to peak at 72% in the first year of a rotation,

declining to <13% and 50% in the third and fourthyear of the rotation, respectively. Soil sampling of theother treatments within this study described by Clemet al. (2004) indicates that where grass is also a com-ponent of the pasture, mineral nitrogen concentra-tions are kept low (A. Whitbread, unpublished data)and %Ndfa of the legume component may remainhigh.

APSIM simulations

The APSIM sorghum model was able to simulatebiomass growth and grain production with a highdegree of precision with the assumption of resettingsoil water and soil nitrate to the measured pre-sowingvalues. Through the removal of nitrogen in sorghumgrain, soil nitrate declined with each successivesorghum crop and was similar to the measured data.While the field trial described in this paper allowedmeasurement of the response of sorghum to variousphases of legume pasture in four seasons, the simula-tion was able to extend this to the many different cli-matic conditions experienced between 1954 and2001. Although the legume pasture treatments wereestablished in the field on a deeper soil profile thanthe grass treatment, the PAWC and depth of all soilprofiles could be made the same in the simulationstudy. All treatments could therefore be compared.

0

0.20

0.40

0.60

0.80

1.00

0 1000 2000 3000 4000 5000 6000

Yield (kg/ha)

Cu

mu

lati

ve p

rob

abili

ty

L. purpureus

Grass

M. daltonii

V. trilobata

Figure 6. The probability of sorghum grain production using the soil water andnitrogen starting conditions of the various rotations simulated from1954 to 2001.



125

Other simulation studies are under way to examineissues that may affect the pasture–grain system.These include: determining the precision of APSIMwhere soil water and nitrate are not reset on a yearlybasis; determining the recharge of soil moisture aftera pasture phase; determining crop production and soilnitrogen dynamics and potential responses tonitrogen fertiliser that may occur past the 4-yearperiod examined in these field studies.

Conclusions

The potential for highly productive grain systemssustained by phases of highly productive pasturesand beef production have been demonstrated throughthis paper and the study of Clem et al. (2004). Therewere no benefits in terms of nitrogen accumulationfrom continuing the sole legume pasture phases pasttwo seasons. Although soil nitrate nitrogen declinedwith successive sorghum crops, the mineralisation ofnitrogen in the time between the harvest of one cropand the sowing of the next remained high. The crop-ping component of this study did not continue forlong enough to conclude for how long adequate soilnitrogen would be maintained. Sorghum grain yielddid decline with each successive crop relative to the‘new’ areas sown to sorghum each year, howevercrop responses to the addition of fertiliser nitrogenwere highly variable. APSIM was able to simulatethe sorghum growth and soil nitrogen dynamics ofthis cropping phase with a high degree of precisionand could be applied to a range of other scenarios notable to be tested in the field. APSIM showed thatsorghum production following a grass pasture wasrisky and unlikely to exceed 2000 kg/ha of grain in60% of seasons. Sorghum production following 2-year legume leys was highly productive with >3000kg/ha of grain produced in 80% of seasons.

References

Armstrong, R.D., McCosker, K., Johnson, S.B., Walsh,K.B., Millar, B., Standley, J. and Probert, M.E. 1999.Legume and opportunity cropping systems in centralQueensland. 1. Legume growth, nitrogen fixation, andwater use. Australian Journal of Agricultural Research,50, 909–924.

Clem, R.L. 2004. Animal production from legume-based leypastures in south-eastern Queensland. Theseproceedings.

Crosthwaite, I. and Harvey, J. 1992. South Burnett CropManagement Notes 1992–93. Kingaroy, QueenslandDepartment of Primary Industries, J. Bjelke-PetersenResearch Station, 171 p.

Dalal, R.C. and Mayer, R.J. 1986. Long-term trends infertility of soils under continuous cultivation and cerealcropping in southern Queensland. I Overall changes insoil properties and trends in winter cereal yields.Australian Journal of Agricultural Research, 24, 265–279.

Dalgliesh, N. and Foale, M. 1998. Soil matters: monitoringsoil water and nutrients in dryland farming. CSIROAustralia, 71.

Gardner, E.A. 1988. Lecture 10 In: Soil water inunderstanding soils and soil data. Brisbane, AustralianSociety of Soil Science Inc., Queensland Branch, 153–185.

Hochman, Z., Dalgliesh, N.P. and Bell, K.L. 2001.Contributions of soil and crop factors to plant availablewater capacity of annual crops on black and greyvertosols. Australian Journal of Agricultural Research,52, 955–961.

Northcote, K.H. 1979. A factual key for the recognition ofAustralian soils, 4th edition. Glenside, South Australia,Rellim Technical Publications.

Rayment, G.E. and Higginson, F.R. 1992. Australianlaboratory handbook of soil and water chemical methods.Melbourne, Inkata Press, 41 p.

Reid, R.E. Sorby, P. and Baker, D.E. 1986. Soils of the BrianPastures Research Station Gayndah, Queensland.Brisbane, Queensland Department of Primary Industries.Research Establishments Publication QR86004, ISSN0813-4391.

Robertson, F.A., Myers, R.J.K. and Saffigna, P.G. 1993a.Distribution of carbon and nitrogen in a long-termcropping system and permanent pasture system. AustralianJournal of Agricultural Research, 44, 1323–1336.

Robertson, F.A., Myers, R.J.K. and Saffigna, P.G. 1993b.Carbon and nitrogen mineralisation in cultivated andgrassland soils in subtropical Queensland. AustralianJournal of Soil Research, 31, 611–619.

Robertson, F.A., Myers, R.J.K. and Saffigna, P.G. 1994.Dynamics of carbon and nitrogen in a long-term croppingsystem and permanent pasture system. Australian Journalof Agricultural Research, 45, 211–221.

Whitbread, A.M., Lefroy, R.D.B. and Blair, G.J. 1998. Asurvey of the impact of cropping on soil physical andchemical properties in north-western New South Wales.Australian Journal of Soil Research, 36, 669–681.



126

Mucuna, Lablab and Paprika Calyx as Substitutes for Commercial Protein Sources used in Dairy

and Pen-fattening Diets by Smallholder Farmers of Zimbabwe

C. Murungweni, O. Mabuku and G.J. Manyawu*

Abstract

Five experiments were carried out to determine the role of Lablab purpureus (lablab) hay, Mucuna pruriens(mucuna) hay, paprika (Capsicum anuum L) calyx and commercial feeds on profitable and sustainableruminant feeding systems. Three of the five experiments were carried out under a participatory researchapproach in two smallholder areas of the Wedza district in Zimbabwe. The other two were carried out undermore controlled conditions at Grasslands Research Station in Marondera, Zimbabwe. The first on-farmexperiment was carried out in Zana 2 ward in Wedza. Results determined the minimum amount of mucuna orlablab hay required to supplement cattle for maintaining liveweight to be equivalent to 1.5% of the animal’sbodyweight on as-is basis. The second on-farm experiment determined the effect of lablab hay and paprikacalyx on milk yield, compositional quality of milk and liveweight gain. In this experiment, four indigenousMashona cows in early lactation were used in a crossover design of 21-day periods. All the four diets increasedmilk yield and improved milk quality. The third on-farm experiment tested the effect on liveweight gain ofdiets based on hay from lablab and mucuna. Six castrated indigenous Mashona cattle were used, three at eachof the two sites. The animals were randomly allocated to the three diets. Replication was by site. Theexperimental animals were allowed an uninterrupted period of 73 days access to the assigned diet, withfortnightly weight recordings. Results showed no significant differences (P > 0.05) between the experimentaldiets. Animals feeding on these diets gained between 1.0 and 1.2 kg of liveweight per day. However, the dietwith hay from mucuna resulted in highest total weight gain and gross margin. The last two experiments werecarried out at Grasslands Research Station. The first on-station experiment had two parts. The first determinedvoluntary feed intake of diets based on hay from lablab and mucuna to be respectively, 3.0% and 3.1% of theanimal’s bodyweight on a dry matter basis. Levels above 2.6% of bodyweight caused metabolic problemsresulting in diarrhoea in sheep. The second part of this experiment determined the effect of diets based on hayfrom lablab and mucuna on nitrogen balance and dry matter digestibility. This experiment used sheep inmetabolism crates, with 21 days for adapting to the diets and 7 days of collection. Results showed higher drymatter digestibility (P < 0.05) in commercial feed (14% CP) and lablab-based diet (14% CP) than in a mucunahay based diet (15% CP). No significant differences were observed between the three diets in terms of nitrogenbalance and liveweight gain. The last on-station experiment determined the effect of diets based on lablab hayand mucuna hay on milk yield and milk quality of Jersey cows. Nine cows in early lactation were used in acompletely randomised design, with three cows assigned to each diet. The experimental period was 28 days.

* Grasslands Research Station, Agriculture Research andExtension Services (AREX), P. Bag 3701, Marondera,Zimbabwe.



127

Results showed higher milk yields, protein, lactose and non-fat solids for cows fed either commercial ormucuna-based diets compared with those fed lablab-based diets. No differences were observed in butterfat.Profits were 120% higher with mucuna-based rations and 76% higher with lablab-based diets than when usingcommercial rations. In addition, mucuna and lablab options reduced maize usage by 24% and 30%respectively, compared with grain in commercial feed. Most smallholder farmers received these legume-basedtechnologies very well because they are cheaper and convenient. The overall results from these trials generateda lot of interest from most farmers within and outside Wedza, resulting in overwhelming response in numbersof farmers now using these technologies in Zimbabwe. The recorded numbers of cattle farmers using thesetechnologies increased from 9 in the 1999–2000 season, through 26 in the 2000–01 season, to 132 in the 2002–03 season.

The Wedza district lies between longitudes 31∞10' and32∞00'E and latitudes 18∞25' and 19∞10'S. Long-termannual rainfall is 750 mm, falling predominantlybetween November and March. Temperatures rise tomean maximum of 28∞C in October and fall to a meanminimum of 5∞C in July. Soils vary from graniticsandy soils in most parts of Dendenyore Ward to somemoderately deep red clay in Zana resettlement area.Zana resettlement area is subdivided into three wards,Zana 1, 2 and 3. The target group for this project wasfarmers in Zana 2 Ward and Dendenyore communalarea. Households in Zana 2 own up to 6 ha of land.Most of these farmers grow tobacco and paprika ascash crops, and maize for human consumption andlivestock feeding. Over 50% of the households areaspiring milk producers. Milk is either sold within thecommunity or to the nearby Wedza Dairy Center. Incontrast, most farmers in Dendenyore fatten beef cattlein pens for sale both to local markets and to commer-cial abattoirs. Some of the households have been doingthis for over 10 years.

Animal production in Zimbabwe falls into two cat-egories: wet season and dry season. During the wetseason, ruminants normally consume enough nutri-tious green grass to ensure a balanced nutrient supplyto post-ruminal sites for maintenance and produc-tivity. A decline in nutrient level in summer (espe-cially a decrease in crude protein) results in ruminantsfailing to eat enough of the basal diet to sustain pro-ductivity and, in some instances, maintenance. Crudeprotein falls from an average 12% in summer tobetween 3% and 4% in winter (Topps and Oliver1993). Neutral detergent fibre was observed to riseabove 30% during winter. However, earlierresearchers (Ward 1968; Ngongoni and Manyuchi1993; Topps and Oliver 1993) determined that proteinsupplementation of poor-quality forage (as is the casewith dry season grass) can meet energy and protein

requirements for both maintenance and production.Nutritional stress is a major cause for poor fertility ofanimals in rural areas (Hamudikuwanda 1999). This isdue to poor ovulation patterns resulting from animalsfailing to reach the target 65% mature weight atbreeding. Product output is naturally low in winterbecause of reduced intake resulting from poor digesti-bility of tough, low nutrient-level, dry season grass.Appropriate protein supplements are necessary forsmallholder farmers to be successful cattle farmers.

Smallholder farmers in Zana 2 Ward and Dende-nyore communal area, like most of their counterpartsaround Zimbabwe, use commercial feed sources indairy and pen-fattening programs. However, the ever-spiralling cost and the irregular availability of thesecommercial feeds are two of the major challengesfacing most farmers. Farmers in Zana 2 and Dende-nyore have been growing and feeding lablab andmucuna to cattle since the inception of the AustralianCentre for International Agriculture Research(ACIAR) project in 1998. The ACIAR project testedvarious feeding strategies and came up with recom-mendations on how best to use the two forage legumeswhen feeding for maintenance and for production.Experiments were carried out in on-farm farmer par-ticipatory research in Zana 2 and Dendenyore, as wellas in controlled environments at Grasslands ResearchStation. Research and extension officers closelyassisted the farmers in such operations as mixing ofdiets, feeding cattle, weighing, and sampling milk. Theproject went on to analyse the economic benefits ofusing these two forage legumes in cattle diets.

On-farm Experiments

Various treatments were formulated using plant mate-rial (chemical compositions shown in Table 1). Paprika



128

calyx and hay from lablab and mucuna is high in crudeprotein (>17% CP) and high in ash (>8%). Digestibilityof these three feed sources is above 54%. The crudeprotein content of paprika calyx, mucuna hay andlablab hay makes these three feedstuffs ideal for formu-lating diets for animals in production phase, especiallywhen mixes use maize grain. The energy content ofmaize is high (>12 MJ ME/kgDM) and the acid deter-gent fibre content of maize is low (<6%). Mixing maizewith any of the three defined protein sources will resultin rations of high digestibility (>65%).

Experiment 1: Determination of the minimum level of mucuna hay required to maintain the liveweight of free-ranging cattle during a dry winter season

The objective of this experiment was to establishthe minimum level of supplementing cattle on rangewith hay of mucuna to avoid liveweight loss duringthe eight-month dry season in Zimbabwe.


AnimalsTwelve indigenous cattle were used in this experi-

ment, which was carried out at three different sites inDendenyore. Four castrated male animals were used ateach site under the management of the cattle owner.

Diets and experimental designMucuna was grown in November 2000 and har-

vested at flowering stage in March–April 2001. Har-vesting was done using a sickle to cut the foragematerial at the bottom of the stem, close to theground. The harvested forage was air-dried byhanging on A-frame rakes for two weeks, turningover the hay only once to avoid loss of leaf. The air-

dried hay was chopped to an average length of 5 cmto make it easier to use both leaf and stem.

The cattle free-ranging on veld were supplementedwith three levels of hay, equivalent to 1%, 1.5% or2% of the animal’s liveweight, on as-is basis. Theanimals were then randomly allocated to these treat-ments. ‘As is’ means that hay of 91–92% dry matter,cut and dried under conditions explained above, wasused. The animals were given their supplement allo-cation in two equal portions: in the morning beforefree ranging, and in the afternoon.

Results

Animals supplemented with hay equivalent to1.5% of their bodyweight maintained weight(Figure 1). Those eating an equivalent of 1% lostalmost 20 kg of liveweight in three months, whileanimals that were not receiving any supplement at alllost twice as much. Animals that were allocated anequivalent of 2% of their bodyweight gained almost20 kg in three months.

Discussion

Hay from mucuna can significantly benefit small-holder farmers, and especially those who rely oncattle as draught power. It is necessary for suchfarmers to supplement their animals during the drywinter season if they are to retain the draughtcapacity of the animals for the following season.Farmers who supplement animals normally reach theproductive season with their animals in good shape.Topps and Oliver (1993) reported that protein sup-plementation for ruminant livestock, at levels highenough to facilitate intake of mature native pasturehay and prevent liveweight loss, is imperative. Butprotein supplementation is usually not practised,

Table 1. Dry matter and chemical composition of lablab hay, mucuna hay, paprika calyx, dairy meal, maizegrain, maize stover, veld hay and wheat bra

Lab-lab hay

Mucuna hay

Paprika calyx

Dairy meal

Maize grain

Maize stover

Veld hay

Wheat bran

Dry matter (g/kg)Ash (g/kgDM)Crude protein (g/kgDM)Crude fibre (g/kgDM)Ether extracts (g/kgDM)Acid detergent fibre (g/kgDM)Dry matter digestibility (g/kgDM)Dry organic matter digestibility (g/kgDM)

915100173

–––

543587

91485.6203

–––

575584

881125195

–29

392546600

89044.915115851.0195815835

890112115

–48.555.7805812

92054.350.91406.10436560580

9304342–

15.0633344354

93064.547.44029.70494390410



129

especially in the smallholder sector, because of thehigh cost and/or non-availability of commercialprotein supplements in times of need. Perhaps greaterreliance on home-grown forage/browse nitrogensources for ruminant animal production is more cost-effective than use of conventional commercialprotein supplements (Matizha et al. 1997).

Mucuna proved to be a cost-effective foragelegume. It is relatively easy to grow, yields well (over3 t/ha of dry matter), and is useful to cattle, as wasshown by the results. Alternative feeding strategiesthat rely on protein sources that are cheap and avail-able on-farm would be more attractive to the small-holder farmers.

Experiment 2: Evaluation of lablab hay and paprika calyx as protein sources in diets for lactating indigenous Mashona cows

This experiment had two objectives. The first wasto determine yield and compositional quality of milkfrom lactating indigenous Mashona cows that wereeating diets based on lablab hay, paprika calyx andcommercial diet (15% CP) over a period of 21 days.The second objective was to determine weightchanges in cows eating diets based on lablab andpaprika calyx, as well as weight changes in calvessuckling from these cows.


AnimalsFour indigenous Mashona cows in early lactation

were used in the Zana 2 experiment. Two of the cows

were in second parity and the other two were in thirdparity. The cows were milked once per day, in themorning between 07.00 and 08.00. A calf was used tosuckle the cow before milking to assist in the initia-tion of milk letdown. The cows were then left withtheir calves for the rest of the day, which is thenormal practice in most smallholder areas in Zim-babwe.

Two other cows were observed under the normalpractice of morning milking followed by freegrazing, with minimum supplementation and usingresidues of any amount and type. Milk yield andcomposition of these cows were also analysed. Theywere then used as negative controls.

Diets and experimental designLablab and calyx from paprika grown and har-

vested in Zana 2 Ward were used in diets formulatedwith crushed maize grain as the major source ofenergy. Commercial dairy meal (15% CP) was usedas the control diet. One farmer managed the four lac-tating cows used in this experiment. Four diets, for-mulated as shown in Table 2, contained crude proteinof at least 14% and energy level of at least 10.7 MJME/kgDM — the general requirement for lactatingcows. Lablab hay and paprika calyxes were milled topass through a 5 mm screen before mixing, to ensureeasy mixing of ration constituents. The cows werethen allowed treatment diets equivalent to 2.5% ofbodyweight plus 10% of milk yield (ARC 1984).

The cows were randomly allocated to the treat-ments and then fed half the total quantity at milking(07.00) and the remainder at 14.00. After the animalshad completely consumed these diets, they were left

250

260

270

280

290

300

310

320

330

0 2 4 6 8 10 12

Week number

Live

wei

gh

t (k

g)

1% velvet hay

1.5% velvet hay

2% velvet hay

No velvet hay

Figure 1. Liveweight changes in cattle supplemented with mucuna hay during the drywinter season in Wedza district, Zimbabwe

1% mucuna

1.5% mucuna

2% mucuna

No mucuna



130

with unlimited access to maize stover. Each cow wasfed all four diets, with a changeover period of 21days. Data recorded during the last seven days ofeach period was used in statistical analysis. Duringthe collection period, daily milk yield was recordedbefore samples were taken from each cow for com-positional analysis at Zimbabwe Dairy ServicesAssociation, Harare. Liveweight for each cow and itscalf was recorded at the beginning and end of eachexperimental period. The data was analysed by thegeneral linear model procedure in SAS (1994).

Results

The diets based on lablab hay, paprika calyx andthe commercial dairy meal (15% CP) had no signifi-cant effect on the milk yield or milk composition(Table 3). Milk yield from indigenous cows wasdouble that of cows on common feeding practice.Results also showed an improvement in milk compo-sition. Milk from cows eating experimental diets hadnon-fat solids levels above the standard minimum of8%. Somatic cell counts (SCCs) were highest in milkfrom cows on diets that incorporated lablab hay.However, cows eating diets that incorporated bothlablab and paprika calyx had SCCs that were a thirdof those eating the diet based on lablab alone. Never-theless, the SCCs recorded were far below themaximum legal level of 1.0 ¥ 106 cells per mL.

Discussion

Similarities in milk yield and milk compositionbetween cows eating home-grown protein-based dietsand commercial protein-based diet suggest that thediets have similar nutrient release patterns. However,the average yield was far below the minimumexpected for commercial dairy farming in Zimbabwe(which is at least 7 L per cow per day), probablybecause the cows were only milked once a day and leftwith their calves for the rest of the day. This differsfrom commercial dairy practice, in which cows aremilked twice a day and calves are separated from theirmothers usually 4 days after birth. The low milkyields, which may have contributed to the lack of sig-nificant differences between the diets, might haveresulted from genetic and/or environmental factorssuch as poor milk genes, poor nutrition and poor healthmanagement. Animals gradually adapt to adverse con-ditions, with a negative impact on production, and thesame diets can produce different results even frommedium- or high-producing breeds in optimal health.

Given the cost of commercial concentrates and theresulting total milk yields, it is not reasonable to feedbought-in concentrates to indigenous Mashona cattlefor commercial milk production. The Wedza farmersin this project deduced that one month’s productioncosts using commercial feed would equal one term ofschool fees for a child attending a local secondary

Table 2 Ration formula and chemical composition of experimental diets testing lablab and paprikacalyx as dairy feeds.

Plant material/nutrient being considered Ration 1 Ration 2 Ration 3 Ration 4

Crushed maize grain (%)Milled lablab hay (%)Milled paprika calyx (%)Dairy meal 15% CP (%)

560

440

604000

5323.523.5

0

000

100

Table 3. Yield and composition of milk from indigenous cows eating diets based on lablab, paprika calyx andcommon practice during the dry winter season in Zimbabwe.

Diet Milk yield (kg)

Fat (%) Protein (%) Lactose (%) Non-fat solids (%)

SCC (cells per mL)

1234

CPa

2.101.781.751.190.8

2.522.172.502.091.58

2.993.022.843.122.75

4.844.884.974.965.07

8.959.028.959.259.94

6.0 ¥ 103

1.95 ¥ 104

6.3 ¥ 103

6.0 ¥ 102

6.1 ¥ 103

a CP = common practice of smallholder farmers in Zimbabwe



131

school. However, if other, indirect benefits (such asthe better physical condition of cows, improvedhealth, improved reproductive performance, andincreased weight gain of suckling calves) are takeninto account, there might be a net benefit.

Although there was an improvement in milk com-position, the percentage of milk fat was far below theminimum of 3% set for commercial milk producersin Zimbabwe. Milk fat is determined by the geneticpotential of the animal as well as by nutrition.Despite the lower fat level in milk from indigenouscattle, farmers who keep these breeds and uselegumes may benefit from improvement in severalother factors that affect cow’s lifetime productivity,including increased milk yield, better milk quality,increased length of lactation, early return to oestrus,shorter calving intervals and earlier breeding ofheifers. In fact, most farmers who have been part ofthe feeding trials since 1999 stated that their cows arenow giving a calf every year. In addition to this, milkfrom cows feeding on rations based on foragelegumes was richer in taste compared to milk fromcows outside the experiment.

Experiment 3: Evaluation of lablab hay and mucuna hay as protein sources in diets for pen fattening cattle

The objective of this experiment was to determineliveweight gain by indigenous beef cattle eating dietsbased on mucuna hay and lablab hay over a period oftwo months. A diet based on the commercial diet(14.5% CP) was used as the control.


AnimalsSix castrated indigenous Mashona cattle were used

at two experimental sites in Dendenyore. One farmermanaged the three animals at each site.

Diets and experimental design

Three test diets were formulated as follows:Diet 1 — 5:3 maize grain : mucunaDiet 2 — 3:2 maize grain : lablab hayDiet 3 — 9:1 maize grain : commercial concentrate.

The mucuna hay and lablab hay were grownlocally from December 2000 and harvested in May2001. The three animals at each site were randomlyallocated to the three diets. The treatment allowancefor each animal was equivalent to 2.5% of its body-

weight (Topps and Oliver 1993) given in two equalmeals at 08.00 and 14.00. The animals were fattenedin pens measuring an average of 3 ¥ 4 m. Eachanimal was fed for 73 days and had an unlimitedaccess to maize stover and water. Liveweight wasrecorded at the beginning of the experiment andevery fortnight thereafter. The data were analysed bythe general linear model procedure in SAS (1994).

Results

There were no significant differences (P > 0.05) intotal weight gain between animals eating diets basedon the home-grown forage legumes and those eatingthe commercial diet (Table 4).

Discussion

Mucuna and lablab hay can substitute for the com-mercial feed (14.5% CP) commonly used by mostcommercial farmers for pen fattening. This resultshould, however, be used cautiously because it wasderived from only two experimental sites with alimited number of experimental units and still requiresproper characterisation. Some positive characteristicsof the forage legume diet are a good level of protein,digestibility above 54% and a high (>8%) ash content.It is also known that the buffering capacity of foragelegumes creates a more stable rumen environment(Van Soest 1994), and this may be one factor contrib-uting to the good performance of diets based on lablaband mucuna hays. Maize grain is normally used as asource of good-quality energy in pen-fattening meals,but maize is also the most important staple food inZimbabwe. Farmers are often reluctant to includeenough maize in cattle diets containing commercialprotein concentrates: some end up reducing the levelfar enough to cause asynchronous release of energyand protein in the rumen, resulting in poor utilisationof the diets. It was observed that most farmers inDendenyore reduce the 9:1 maize:concentrate mix to7:1 or less during poor harvest seasons.

The energy level of the two forage legumes isabove 8 MJ ME/kgDM. Although the quality ofenergy in the forage legumes may differ with that ofmaize, these results show that their incorporation inpen-fattening meals reduces the use of maize grainwithout compromising weight gain. By using thesehome-grown protein sources, farmers will alsoreduce production costs. Farmers will avoid short-ages of commercial feeds, due to either ill-timeddeliveries or unavailability, which sometimes occur,



132

especially during drought years such as 2001–02.They will welcome any savings on maize grain whenthe savings are accompanied by increased profitsfrom using low-cost, home-grown forage legumeslike mucuna and lablab.

Results of a gross margin analysis made for thisexperiment by O. Jiri et al. (unpublished data) for thewhole fattening period of 73 days suggest that small-holder farmers will be discouraged from pen fat-tening. The savings per beast over that period usingthe most economic diet (mucuna based) will beenough to buy one only 50 kg bag of maize grain atthe current market rate. Using commercial rations inthe current Zimbabwean economic environmentwould drain farmers of their financial resources andlabour. However, most small-scale farmers do notcost their own labour, and therefore see lablab andmucuna forage as a great benefit in pen fattening.Because of these findings, the number of farmersgrowing mucuna and lablab for feeding animals isincreasing in Dendenyore. The 2002–03 droughtfurther increased farmers’ interest, as these twoforage legumes are relatively drought-tolerant.

On-station Experiments

Experiment 1: Determination of the maximum level of intake of diets made by mixing hay from mucuna or lablab with crushed maize grain

This experiment had two parts; determination ofmaximum and optimum intake of fattening diets, andthe dry matter digestibility of diets that were formu-lated using hay from lablab and mucuna.


AnimalsTwelve ear-tagged Doper rams aged between four

and ten months were used in these trials. The rams

were dosed with valbazen against internal parasitesand then vaccinated against pulpy kidney two weeksbefore the experiment began. They were stratifiedaccording to age, randomly allocated to the threetreatments, and weighed on the morning before thebeginning of the experiment.

Diets and experimental designMucuna and lablab were grown at Grasslands

Research Station and harvested at flowering stage.The forage material was air-dried on A-frame racksfor two weeks and then milled to allow proper mixingwith crushed maize grain at ratios of 3:2 maize grain:lablab hay for ration 1 and 5:3 maize grain: mucunahay for ration 2. The control diet was a commercialfattening diet (15% CP). These diets had alreadybeen tested in Wedza during participatory pen-fat-tening experiments. During the experiment, theanimals were initially housed in individual pens, andhad ad libitum access to the assigned treatment diets.After an adaptation period of 14 days, data were col-lected for seven days to determine maximum intake.In the second part of the experiment, the young ramswere individually housed in metabolism crates andallowed treatment diets equivalent to 2.5% of eachanimal’s bodyweight, twice daily in equal propor-tions (Topps and Oliver 1993), and wheat branwithout restriction. The diets were introduced at07.30 and at 14.00. Clean water was always availablefrom buckets.

Results

No significant differences (P > 0.05) in maximumintake were observed between the commercial dietand the two forage legume based diets. The rams ate3.1% of their bodyweight of the ration containing hayfrom mucuna, 3.0% of the ration containing hay fromlablab, and 3.0% of the ration containing commercialfeed (all values on dry-matter basis). Nevertheless,forage legume based diets were seen to lead to meta-bolic disorders manifested as diarrhoea if given in

Table 4. Total weight gain by indigenous Mashona cattle eating diets based on hay frommucuna and lablab and a commercial diet (14.5% CP) during two dry wintermonths in Dendenyore, Zimbabwe.

Diet 1Mucuna based

Diet 2Lablab based

Diet 3Commercial

CVa

Total gain (kg) 88.5 a 74.5 a 76.0 a 4.94a CV = coefficient of variation.Note: differences between means followed by the same letter are not statistically significant.



133

excess of 2.6% of the animals’ bodyweight. Afterseveral adjustments to the feeding structure, wefound it ideal to feed the rations in equal proportionsin the mornings and afternoons and to introducepoor-quality roughage such as wheat bran or maizestover. Dry-matter digestibility of the three rationswas measured at ≥70%.

Discussion

Establishment of intake is necessary because penfattening usually reaches a stage when food is givenwithout limit (i.e. ad libitum). That levels of theforage-based diets over 2.6% of bodyweight causeddiarrhoea may suggest that their roughage contentcould not adequately sustain proper rumen motility.

Results on intake are encouraging. Animals in pen-fattening programs normally consume at least 2.5%of their bodyweight (CPA 1998), and this level didnot result in any noticeable metabolic problems. Anyadditional feed would be given in the form of poor-quality roughage like maize stover or wheat bran,improving the rhythmic movements of the rumen andaiding the proper formation of wastes. The highintakes of feeds containing lablab and mucuna haysuggest a high degradability for such diets. Legumesalso have a good buffering capacity, necessary forappropriate microbial activity and crucial to themaintenance of a stable rumen environment.

Dry matter digestibility of the two forage legumeswas high (≥70%). Any differences in production maybe the result of different levels of maize grain and/orthe possible associative effects of maize with aprotein source. The mineral composition of the diets,which was not analysed in this experiment, may alsocontribute to production performance. Other factorslike vitamins may have an effect.

Animals in production require bypass protein to bebetween 35% and 40%, so partitioning of nutrientsbetween the rumen and the lower gut should be char-acterised.

Experiment 5: Yield and composition of milk from Jersey cows eating diets based on hay from lablab and mucuna

This experiment was designed to evaluate theeffect of hay of lablab and mucuna on the yield andcomposition of milk from Jersey cows, as well as theeconomic benefits of using such diets.


AnimalsNine Jersey cows in early lactation and in their

second parity were used in this experiment. The cowswere kept in individual 3 ¥ 2 m pens with free accessto clean tap-water, and were milked twice each day.

Diets and experimental designAir-dried hay of mucuna and lablab was used in

rations formulated to meet the minimum energy andprotein requirements for lactating dairy cows, as fol-lows:Ration 1 — commercial dairy meal (16% CP)Ration 2 — crushed maize grain and milled lablabhay mixed in the ratio of 1:1Ration 3 — crushed maize grain and milled mucunahay mixed in the ratio of 5 maize: 4 mucuna hay.

The ration with commercial dairy meal was used asthe control diet. The hay was milled to about 5 mmlong before mixing with crushed maize grain. Thethree treatment diets were randomly allocated to thenine lactating cows, with three cows per diet. Thecows were allowed treatment diets equivalent to0.5% bodyweight plus 0.1 ¥ milk yield, and unlim-ited access to maize stover. Treatment diets weregiven in two equal meals in the morning and after-noon for 28 days. Milk samples were taken from eachcow during the last seven days and sent for composi-tional analysis at the Zimbabwe Dairy Services Asso-ciation in Harare. Milk recorded during the last sevendays of the experiment was used in statistical anal-ysis. The chemical composition of the rations used inthese trials is shown in Table 5.

Results

Milk yield from cows eating a ration with hay frommucuna was significantly higher (P < 0.05) than theyield from cows eating lablab-based diets, as shownin Table 6. There were no significant yield differ-ences between the mucuna-based diet and commer-cial feed. Protein, lactose and non-fat solids weresignificantly higher (P < 0.05) in milk from cowseating commercial feed and mucuna based diets com-pared with milk from cows eating lablab; differencesbetween the commercial diet and the mucuna-baseddiet were not significant. Butterfat content washigher in lablab-based diets, although no significantdifferences were observed between the three diets.SCCs were below maximum legal levels of 1.0 ¥ 106

cells/mL.



134

Results of a gross margin analysis (Table 7) showthat using hay of lablab and mucuna increases profitsin commercial dairy farming. Costs of producinglablab and mucuna hay were calculated as labour,chemical inputs and opportunity costs for 1 kg of hayat 5 t dry matter per hectare for lablab and 4 t/ha formucuna. These are the estimated yields obtained atGrasslands Research Station during 2001–02, aseason of poor rainfall distribution.

Discussion

Hay of mucuna and lablab forage is economic forcommercial milk production. Farmers using thelablab-based diet made 76% more daily profit thanthose using the commercial diet, while those usingthe mucuna-based diet made 121% more. Farmers in

Wedza welcomed the results and acknowledgedthem by increasing the cropping area allocated tothese legumes. The high level of adoption of thesetechnologies (Maasdorp et al. 2004) may be attrib-uted to the comparable milk yields of the mucuna-based ration and the commonly used commercialfeed, the convenience of using these legumes, and theever-increasing cost of commercial feed.

With the abnormal economic situation in Zim-babwe, farmers will rely more on home-grown feedsthan commercial concentrates. From November 2002to November 2003, milk producer price increased10-fold from Zim$77/L to $800/L, while the price of15% CP commercial feed rose 27-fold from $55/kgto $1500/kg. This resulted in hyperdemand formucuna and lablab seed for the 2003 season. Grass-lands Research Station failed to cope with farmers

Table 5. Chemical composition of experimental diets testing the role of lablab and mucuna hay based diets indairy feeding.

Plant material/nutrient being considered Ration 1Commercial

Ration 2Lablab-based

Ration 3Mucuna-based

Dry matter (g/kg)Crude protein (g/kgDM)Crude fibre (g/kgDM)Acid detergent fibre (g/kgDM)Dry matter digestibility (g/kgDM)Dry organic matter digestibility (g/kgDM)

890147103125870900

893140139182800820

891155130167810824

Table 6. Milk yield and composition and somatic cell counts of Jersey cows eating diets based on lablab hay,mucuna hay and commercial feed.

Milk yield(kg)

Butter fat(%)

Protein(%)

Lactose(%)

Non-fat sol-ids(%)

Somatic cell counts

Commercial feed (15% CP)Lablab-based feedMucuna-based feedCoefficient of variation

10.9 a7.0 b9.0 a13.4

4.2 a4.4 a3.9 a19.4

3.7 a3.2 b3.6 a6.5

4.8 a4.5 b4.8 a2.1

9.5 a8.7 b9.5 a3.3

1.0 ¥ 104

1.2 ¥ 105

3.7 ¥ 103

Note: Means in the same column with different letters are significantly different (P < 0.05).

Table 7. Gross margin (Zim$/day) for dairy programs using lablab and mucuna-based diets.

Milk yield (kg)

Milk pro-ducer

price/L

Gross/L Daily production

cost

Daily gross margin

Maize grain (%)

Commercial rationLablab-based rationMucuna-based ration

10.97.79.0

777777

839593693

602177170

237416523

805056



135

seed requirements and is in the process of encour-aging farmers to establish seed banks at farm level.

Using forage legumes in dairy diets reduces thecost of production and results in higher profits.Although these legumes require minimum attentionduring the growing season, the harvesting period islabour intensive. These legumes may not, therefore,be ideal in large-scale production without appro-priate mechanisation. However, the Zimbabweangovernment is promoting the involvement of small-holder farmers in commercial dairy farming, andmucuna and lablab have a role to play.

Another crucial point is that research has concen-trated on protein sources and paid less attention toenergy sources in animal production. Maize is themain energy source used for animal production inZimbabwe, but is also the main staple food of thepeople. As a result, both the farmers and the policymakers want to reduce the maize used in cattle dietsas far as possible. Maize usage in dairy diets reducesby 24% in the mucuna diet and by 30% in the lablabdiet, compared with traditional commercial feedingsystems.

It is also worth noting that the Jersey cows eatinglablab- and mucuna-based diets produced good-quality milk that met the expected minimum marketstandards. Mucuna and lablab also reduce externalinputs and alleviate land degradation.

Conclusion

Forage legume based diets are ecologically and eco-nomically sustainable both in dairy and in pen-fatteningprograms for smallholder farmers in Zimbabwe. In thecurrent economic environment, dairy farming is moreprofitable than pen fattening. Using bought protein isuneconomic in the Zimbabwean smallholder system,and farmers should replace commercial protein withforage legumes where possible. Future research shouldexamine the long-term effects of using mucuna andlablab on the reproductive performance of cattle,perhaps focusing on the length and shape of the lacta-tion curve and the effect on fertility.

Acknowledgments

The authors cordially thank ACIAR for funding thisappropriate research. We greatly appreciate thetimely support of the ACIAR team leaders. We alsothank Simon Chibi, Livingstone Svisva and Benny

Kamanga (agriculture assistants, Grasslands), MissC. Nyama (research technician, Grasslands),Mr Mutenga and Mr Munjoma (Chemistry Section,Grasslands), Mr Mazaiwana, Mr Mukombe,Mr Pswarai and Mrs Magwenzi (Agritex extensionofficers in Wedza), and the farmers in Wedza formaking this research work a success.

References

ARC (Agricultural Research Council) 1984. The nutrientrequirements of farm livestock, no. 2, ruminants (suppl.no. 1). CAB International, Farnham Royal.

CPA (Cattle Producers Association) 1998. Beef productionmanual. Harare, Zimbabwe, CPA.

Hamudikuwanda, H. 1999. Is fertility a problem incommercial beef herds in Zimbabwe? Paper presented atthe Cattle Producers Association management school inBulawayo and Harare, 6–9 July 1999. Harare, Zimbabwe,Department of Animal Science, University of Zimbabwe.

Maasdorp, B.V., Jiri, O. and Temba, E. 2004. Contrastingadoption, management, productivity and utilisation ofmucuna in two different smallholder farming systems inZimbabwe. These proceedings.

Manyawu, G.J., Jiri, O., Chigariro, B., Mazaiwana, P.R.,Nyoka, R. and D. Chikazunga, D. 2004. A socioeconomicevaluation of an ACIAR-funded project on the use offorage legumes in integrated crop–livestock systems onsmallholder farms in Zimbabwe. These proceedings.

Matizha, W., Ngongoni, N.T., Topps, J.H. 1997. Effect ofsupplementing veld hay with tropical legumes Desmodiumuncinatum, Stylosanthes guianensis and Macroptiliumatropurpureum on intake, digestibility, outflow rates,nitrogen retention and live weight gain in lambs rates,nitrogen retention and live weight gain in lambs. AnimalFeed Science and Technology, 69, 187–193.

Ngongoni, N.T. and Manyuchi, B. 1993. A note on the flowof nitrogen to the abomasum in ewes given a basal diet ofstar-grass hay supplemented with graded levels of deeplitter poultry manure. Zimbabwe Journal of AgriculturalResearch, 31(2), 135–140.

SAS 1994. SAS user guide: statistics, version 6.10. NorthCarolina, Cary, SAS Institute Inc.

Topps, J.H. and Oliver, J. 1993. Animal foods of centralAfrica. Technical Handbook No. 2 (rev. ed.). Harare,Zimbabwe, Modern Farming Publications.

Van Soest, P.J. 1994. Function of the ruminant forestomach.In: Van Soest, P., ed., Nutritional ecology of the ruminant(2nd ed). Ithaca, NY, USA, Cornell University Press,230–252.

Ward, H.K. 1968. Supplementation of beef cows grazing onveld. Rhodesia Journal of Agricultural Research, 6, 93–101.



136

Animal Production from Legume-based Ley Pastures in Southeastern Queensland

R.L. Clem*

Abstract

As soils used for cropping in central and southeastern Queensland decline in organic matter and availablenitrogen, the benefits of using legume-based ley pastures as an alternative to using fertiliser nitrogen are beingmore widely recognised. However, as well as providing increases in soil fertility for subsequent cropping cycles,ley pastures also need to provide adequate returns to farmers from livestock production. This study at BrianPastures Research Station in southeast Queensland compared forage and animal production from eight tropicalpastures, which included legumes identified in recent evaluation programs and being developed for use byfarmers.

The pastures used were lablab (Lablab purpureus cv. Highworth), which is an annual legume that wasresown each year; Macrotyloma daltonii CPI 60303 and Vigna trilobata CPI 13671, which are annual legumesthat regenerate from seed each year; butterfly pea (Clitoria ternatea cv. Milgarra) and burgundy bean(Macroptilium bracteatum CPI 27404), which are perennial legumes; a grass pasture of green panic (Panicummaximum var. trichoglume cv. Petrie), creeping bluegrass (Bothriochloa insculpta cv. Bisset) and Queenslandbluegrass (Dichanthium sericeum); and the same mixture of grasses with butterfly pea and with caatinga stylo(Stylosanthes seabrana cvv. Primar and Unica).

End-of-season yields ranged from 1 to 5 t/ha, with lablab and the grass–legume pastures producing thehighest yields. M. daltonii and V. trilobata regenerated each year, but regular spraying for weed control wasnecessary. Burgundy bean persisted for three years and butterfly pea was even more persistent. Both butterflypea and Caatinga stylo persisted and combined well with the grasses.

Lablab produced the most liveweight gain/ha, with growth rates from 0.60 to 0.86 kg/head/day. Growth rateson other legumes varied from 0.39 to 0.79 kg/head/day, and there were some differences in the duration ofgrazing. Liveweight gain/ha was similar for butterfly pea and burgundy bean and was higher than forV. trilobata, followed by M. daltonii. On the grass and grass–legume pastures, growth rates ranged from 0.39to 0.71 kg/head/day, with legume-based pasture producing gains of 30 to 70 kg/ha more than the grass-onlypasture over five years.

Beef cattle production and cropping are the majoragricultural industries on large areas of central andsoutheastern Queensland, although the integrationof livestock and cropping has been limited even onproperties where both enterprises are practised.With long-term cultivation and cropping on clay

soils, lower wheat grain protein levels are associ-ated with a decline in soil nitrogen (Dalal et al.1991). At the same time, higher quality forage isneeded to increase cattle growth rates and reduceage at turn-off to meet more demanding beef mar-kets. As a result there is increasing interest in, andmore recognition of, the value of using legume-based ley pastures to maintain or restore soil fer-tility and to provide higher quality forage toimprove livestock performance.

* Department of Primary Industries, PO Box 395,Gympie, Queensland 4570.



137

There are many similarities between this regionand those to the north and south. In northern Aus-tralia, where research in the 1980s investigated thepotential for using tropical legumes in leys in crop-ping and livestock systems (Jones et al. 1991), theircommercial use was limited until the live-cattleexport trade increased demand for high-qualitypasture (Winter et al. 1996). In southern Queensland,although ley pastures based on lucerne (Medicagosativa) and annual medics (Medicago spp.) have beenshown to arrest soil fertility decline and improve cropyield and protein content (Dalal et al. 1991), theiradoption has been slow. While this slow adoption hasbeen attributed to a number of causes (Lloyd et al.1991; Weston et al. 2000), the areas now being sownare increasing.

In central and southeast Queensland until the1990s, the use of leys was limited because well-adapted pasture legumes and the appropriateagronomy to allow reliable establishment and man-agement (Keating and Mott 1987) were not readilyavailable. There is now a good range of tropicallegumes (Pengelly and Conway 2000) and grasses(Cook and Clem 2000) that can be used in leys, andthe value of lablab (Lablab purpureus cvv. High-worth, Rongai and Endurance) is being further recog-nised (McCosker et al. 2000). Larger areas of lablaband butterfly pea (Clitoria ternatea cv. Milgarra) arenow being sown.

To continue and increase the rate of adoption ofleys in farming/livestock systems, the more complexissues of management now need to be addressed.These include the strategies necessary to maximisethe accumulation of nitrogen in the ley and optimiseits use during the subsequent crop, and to improve thevalue of grazing from the pasture phase. This paperreports some aspects of pasture and animal produc-tion from a range of forage and pasture types grownon a vertosol in the Burnett region of Queensland.


Site description

Experiments were located at Brian PasturesResearch Station, Gayndah, Queensland, Australia(25∞39'S, 151∞45'E). The average annual rainfall atthe station is 702 mm, but rainfall is highly variable,both annually and seasonally. On average, thehighest maximum and minimum temperatures are in

January (32.1∞C and 20.2∞C) and the lowest in July(21.6∞C and 6.6∞C) with approximately 20 frostseach year. Low moisture availability during thewarmer months, low winter temperatures andfrosting are the major climatic limitations to pasturegrowth (Fitzpatrick and Nix 1970).

Soils are brown to dark self-mulching clays ofbasaltic origin. Surface pH is 6.5 to 7.5 with an alka-line pH trend with depth. Soil phosphorus levelsrange from 40 to 80 ppm (Colwell 1963, bicarbonateextraction) and soil depth varies from 0.4 to 1.5 m(Reid et al. 1986).

All treatment paddocks had been cleared of treesand cultivated. The original vegetation was openwoodland with Eucalyptus tereticornis, E. melano-phloia and E. crebra being the main species; the pre-dominant grasses were Bothriochloa bladhii,Dichanthium sericeum and Heteropogon contortus.

Design and treatments

Eight pasture treatments, each replicated twice,were sown into 16 paddocks of 2.5 ha each from 4 to6 February 1998. Lablab was sown again on7 December 1998, 4 January 2000 and 13 December2000. The pasture treatments were:1. Lablab (Lablab purpureus cv. Highworth)2. Macrotyloma daltonii CPI 603033. Vigna trilobata CPI 13671.4. Butterfly pea (Clitoria ternatea cv. Milgarra)5. Burgundy bean (Macroptilium bracteatum CPI

27404)6. Grass7. Butterfly pea with grass8. Caatinga stylo (Stylosanthes seabrana cvv. Primar

and Unica) with grass.Legume sowing rates were, in the grass–legume

paddocks, caatinga stylo 3 kg/ha and butterfly pea4 kg/ha and, in the legume-only paddocks, butterflypea 7 kg/ha, burgundy bean 3 kg/ha, M. daltonii4 kg/ha, V. trilobata 3 kg/ha and lablab 32 kg/ha.Lablab, butterfly pea, burgundy bean, M. daltoniiand V. trilobata were sown into moist soil, whilecaatinga stylo and the grasses were surface-sown androlled into the soil surface.

The grass was a mixture of 1 kg/ha creeping blue-grass (Bothriochloa insculpta cv. Bisset), 1 kg/haQueensland bluegrass (Dichanthium sericeum) and2 kg/ha green panic (Panicum maximum var. tri-choglume cv. Petrie).



138

Grazing management

All pastures were grazed with Brahman crossbredsteers at a range of stocking rates. On the grass andgrass–legume pastures, the first grazing com-menced on 11 August 1998 for 293 days to 12 April1999 at a stocking rate of 1.25 ha/steer. Subsequentgrazings, all at the same stocking rate, have beenfrom 7 July 1999 to 26 April 2000 (324 days), from28 November 2000 to 9 July 2001 (288 days) andfrom 13 September 2001 to 29 May 2002 (270days).

The five forage legume-only treatments weregrazed for the same period in year 1, but in later yearsgrazing time has varied depending on forage availa-bility (Table 1). Grazing in the three annual legumetreatments (lablab, V. trilobata and M. daltonii) hasbeen in late summer/early autumn. There was also areduction in the area of these paddocks, as sorghumwas sown into one-third of each paddock in 2000 andtwo-thirds of each paddock in 2001 as part of anotherexperiment. In 2001, a group of 10 steers grazed the0.8 ha for between 6 and 12 days.

The perennial legumes (Milgarra butterfly pea andburgundy bean) provided grazing earlier and over a

longer period (Table 1) and again for 160 to 176 daysfrom 13 December 2001 to 29 May 2002.

All steers were weighed at six-week intervalsexcept when grazing the small areas of legume in2001. Stock were removed from the pastures whenweight loss was imminent (drafts 1 and 4) or as soonas weight loss was measured (drafts 2 and 3).

Pasture measurements

Seedling density was measured after emergence bycounting plants in 100 quadrats in each paddock.Pasture yield and botanical composition and legumedensity were determined each year at the end of thegrowing season. In the annual forage paddocks, thiswas done by cutting 10 quadrats, counting the legumeplants cut, and sorting legumes from grass andweeds. In the perennial pastures, these attributeswere measured using the BOTANAL technique(Tothill et al. 1992).

In the paddocks of butterfly pea, caatinga stylo,burgundy bean, M. daltonii and V. trilobata, theamount of legume seed in the soil has been measuredeach year by taking soil cores following first rains inspring, and washing seed from the soil using the tech-nique of Jones and Bunch (1988).

Table 1. Stocking rates (ha/steer) and grazing days for the annual and perennial forage types over four seasons.

Forage type 1998(18/5–3/8/98)

1999(3/11/98–12/4/99)

2000(6/1–15/6/00)

2001(28/11/00–9/7/01)

ha/steer days ha/steer days ha/steer days ha/steer days

Lablab purpureusMacrotyloma daltonii Vigna trilobata Clitoria ternatea Macroptilium bracteatum

0.50.620.620.620.62

7777777777

0.621.251.251.251.25

81130130160160

0.420.830.830.830.83

101161161161124

0.080.080.080.620.62

1269

215207

Table 2. Rainfall (mm) at Brian Pastures Research Station from 1997–88 to 2001–02, and the long-termaverage.


1997–981998–991999–002000–012001–02Average (48 yr)

382284

54634

72614

31

28

43119

330

3830

3921

107171

3362

126107118

70155

75

5890536652

103

1203578

718

103

1205161

138164

96

34927

1206066

1035

3332

536

753027464939

203726

17029

752592661659624702

Note: Bold ≥ average. Long-term decile 5 is 670 mm/year.



139


Seasonal conditions

Rainfall has generally been near or below averageover the five growing seasons (Table 2). Summerrainfall in 1997–98 was adequate for soil moisturebuild-up and successful establishment, but pastureyields in the first season, except for lablab, were lowand grazing was restricted. Pastures have been main-tained, but long dry periods have resulted in variablepasture yields.

Legume density and soil seed

Legume and grass establishment in all pasturetypes was satisfactory. Density of legumes at estab-lishment ranged from 9 to 30 plants/m2. Establish-ment of lablab was consistent across all years ofsowing, but populations of other legumes have varieddepending on plant survival and recruitment fromseed (Table 3).

The annual legumes, M. daltonii and V. trilobata,have successfully recruited from seed, with large

populations of seedlings emerging on several occa-sions each summer. The success of this regenerationis very dependent on control of weeds and recolo-nising grasses, and can be improved with strategicuse of herbicide. Soil seed levels for V. trilobatadeclined after year 1 but have been maintained atmoderate levels since. In contrast, M daltonii hasmaintained levels of 200 to 300 kg/ha (Table 4).

Of the perennial legumes, butterfly pea was morepersistent than burgundy bean. Most of the originalplants of burgundy bean had died by the end of thefourth year, but there was some regeneration from amoderate amount of seed in the soil in the 2001summer.

The density of butterfly pea declined initially, butthen increased following recruitment in 2001; theamount of soil seed, although low in years 1 and 2,had increased to 30–40 kg/ha by year 4.

Similarly, in the grass–legume pastures, the densityof butterfly pea first declined but increased in 2002(Table 3). Seedling recruitment of caatinga stylo washigh in 2000–02 following heavy seeding (as indicatedby soil seed reserves and seedling density).

Table 3. Legume density of lablab sown each year and changes in density of the other legumes sown in 1998measured at the end of the summer.

Legume density (plants/m2)

1998 1999 2000 2001 2002

Lablab purpureusMacrotyloma daltonii Vigna trilobata Clitoria ternateaMacroptilium bracteatum C. ternatea with grassStylosanthes seabrana with grass

10113013189

16

15––

9 (1)a

12 (5)6

11

11––

8 (1)8 (5)6 (2)

12 (35)

125020

7 (6)1 (9)7 (6)

7 (71)

–––

12 (3)6 (2)

12 (4)37 (16)

a Numbers in brackets are seedlings.

Table 4. Changes in soil seed (kg/ha) of legumes for the range of pastures.

Legume soil seed (kg/ha)

1998 1999 2000 2001 2002

Lablab purpureusMacrotyloma daltonii Vigna trilobata Clitoria ternatea Macroptilium bracteatum C. ternatea with grassStylosanthes seabrana with grass

0216200

043

00

0216

277

2300

0222

160

150

38

0335

4246

62690

0192

223313

12485



140

Pasture yields and composition

Pasture yields varied in response to rainfall andperiods of moisture stress. Lablab produced highyields in the first season, when good subsoil moisturewas available, and again in 2000–01 (Table 5).

Although the annual legumes (M. daltonii andV. trilobata) produced high yields in some years, thetotal legume yield and presumably the amount ofnitrogen accumulated (Jones 1972; Vallis 1972) overfour summers was lower than for lablab. Theselegumes successfully recruited from seed each year,however, and comprised a moderately high propor-tion of the total yield (Table 5).

All pure legume pastures were recolonised withgrass or weeds to varying extents, but butterfly peapersisted well with over 50% legume in the pasture.Burgundy bean had a moderate percentage of legumein the pasture into the third summer, but then the pro-portion of the legume declined to 7%, after which itsvalue for future grazing and any further nitrogeninput would be negligible.

Pasture yields in the grass and grass–legume pad-docks were high, and provided for longer periods ofgrazing than on the legume paddocks. In the grasspaddocks, the percentage of green panic, a grass thatrequires high nitrogen to persist under grazing(Jones et al. 1995), declined from 80% to 15% ofpasture yield, while the percentage of Bisset blue-grass increased to nearly 60%. In the grass–legumepaddocks, green panic also declined; however, itremained at over 40% of yield and the increase inbluegrass was less. The legume componentremained at 20% in the butterfly pea paddocks andincreased to some 40% in the stylo paddocks(Table 5).

Steer liveweight gain

The forage legume paddocks were grazed for var-iable periods over the summer each year at stockingrates generally higher than for the grass–legume pas-tures (Table 1). Grazing days ranged from 77 on allforages in the first year to 215 on butterfly pea in2001. Butterfly pea and burgundy bean paddockswere again grazed for 160–180 days in 2002. On thegrass–legume pastures, grazing began in August1998 at 1.25 ha/steer, and these pastures were grazedat this stocking rate for between 270 and 324 dayseach year.

In the paddocks sown to lablab, M. daltonii andV. trilobata in 2001, only one-third (or 0.8 ha) wasreplanted or allowed to regenerate to legume, as theother two-thirds was sown to sorghum. Six groups of10 head each were used to graze these areas for 6–12days in May 2001 after the crop had been harvested.This avoided the risk of animals breaking into thecrop area, but liveweight gains on the legume areaswere not recorded.

Steer liveweight performance and liveweight gain/ha varied with pasture type and season. On the foragetreatments, liveweight gain/head ranged from 0.35 to0.86 kg/head/day, while gain/ha has varied from 50to over 200 kg/ha/year (Table 6).

Steers grazing lablab consistently gained at highrates of up to 0.86 kg/steer/day, and over four yearslablab produced the most liveweight gain/ha.

Grazing of V. trilobata and M. daltonii com-menced earlier in the summer than for lablab, butoverall grazing days were fewer, with a range of 72–194 animal grazing days/ha for M. daltonii and 104–194 for V. trilobata compared with 130–240 forlablab.

Table 5. Total yield (kg/ha) and percentage legume (in brackets) for each forage and pasture type.

1998 1999 2000 2001 2002a

Lablab purpureusMacrotyloma daltoniiVigna trilobata Clitoria ternateaMacroptilium bracteatum C. ternatea with grassStylosanthes seabrana with grass

4155 (100)1314 (100)1581 (100)1003 (100)923 (100)1009 (33)1960 (6)

2398 (100)4866 (57)2268 (62)2270 (49)2058 (71)3660 (5)5845 (7)

2566 (100)1940 (80)2937 (65)2650b (53)2900b (35)2792 (13)3853 (28)

4990 (100)4437 (87)3253 (57)3494 (65)2285 (7)

3686 (25)3886 (24)

–––

3454 (32)2370 (3)

3903 (20)4328 (43)

a In 2002, these legume areas were sown to sorghum.bThese yields were estimated.



141

Tab

le 6

.L

ive-

wei

ght

gain

of

stee

rs a

nd g

ain/

ha f

or e

ach

of t

he f

orag

e an

d pa

stur

e ty

pes

over

fou

r se

ason

s (a

nnua

l fo

rage

s) a

nd f

ive

seas

ons

(per

enni

al l

egum

esan

d gr

ass–

legu

me

past

ures

).

1998

1999

2000

2001

2002

Tot

al

(kg/

ha/d

)(k

g/ha

)(k

g/ha

/d)

(kg/

ha)

(kg/

ha/d

)(k

g/ha

)(k

g/ha

/d)

(kg/

ha)

(kg/

ha/d

)(k

g/ha

)(k

g/ha

)

Lab

lab

purp

ureu

s0.

6092

0.86

112

0.73

176

0.86

124

505

Mac

roty

lom

a da

lton

ii0.

4050

0.66

690.

3567

0.66

4823

4

Vig

na tr

ilob

ata

0.60

740.

7982

0.55

107

0.79

8634

9

Cli

tori

a te

rnat

ea

0.39

480.

6482

0.49

950.

6422

10.

6518

362

9

Mac

ropt

iliu

m b

ract

eatu

m 2

7404

0.56

690.

6380

0.40

590.

6722

20.

6717

160

1

Gra

ss0.

4410

40.

5213

50.

5713

20.

5712

449

5

C. t

erna

tea

wit

h gr

ass

0.42

980.

5213

50.

6815

70.

6313

652

6

Styl

osan

thes

sea

bran

a w

ith

gras

s0.

3991

0.62

162

0.68

156

0.71

154

563



142

Growth rates of steers on M. daltonii were lessthan those recorded from other pastures despite highlegume yields, and ranged from 0.35 to 0.66 kg/steer/day. These lower growth rates may be due to lowerpalatability of this legume. In comparison, V. trilo-bata was noticeably the most palatable of thelegumes and gave high steer growth rates of 0.55–0.79 kg/steer/day.

Beef production from butterfly pea, which wasbeing recolonised with grass, varied most, from a lowof 48 kg/ha in 1998 to a high of 220 kg/ha in 2001.Butterfly pea is not favoured by short growingseasons such as that in 1998, but over the relativelylong growing season of 2000–01 it maintainedanimals for 215 days at 0.64 kg/head/day. Burgundybean, which is more palatable than butterfly pea, pro-duced similar per animal performance and liveweightgain/ha, although the legume content of the pasturedeclined. In year 1, liveweight gain was higher fromburgundy bean, possibly because of the high legumeyield, its high palatability and its ability to regrowunder severe defoliation (Dalzell et al. 1997).

On the grass–legume pastures, grazing periodswere longer (270–324 days) at the lower stockingrate of 1.25 ha/steer than on the annual and perenniallegume forages. Growth rates ranged from 0.4 to0.71 kg/steer/day and were similar to growth rates onthe butterfly pea and burgundy bean planted withoutgrass. Beef production from the grass–legume pas-tures was slightly less than from the perennial legumepastures, but liveweight gain was probably restrictedby forage availability in the first year of grazing,when steers gained only 120 kg/head/year. The rangefrom 160 to 200 kg/head/year in years 2, 3 and 4 issimilar to that from other tropical grass–legume pas-tures in southeastern and central Queensland (Man-netje and Jones 1990; Partridge et al. 1996). Afterfour years of grazing, the grass–legume pastures pro-vided a useful benefit in animal production over thegrass-only pastures (Table 6). Except in the first year,presumably when the quality of grass was high fol-lowing cultivation and mineralisation of nutrients,steer growth rates were higher on both the grass withbutterfly pea and the grass with caatinga stylo.

Implications and Conclusions

There is potential for the forage and pasture typestested to provide useful grazing for beef cattle whilebeing used as pasture leys. The pastures included anannual legume that was resown each year (lablab),

annual legumes with potential for seedling recruit-ment (V. trilobata and M. daltonii), short-term per-ennial legumes that also recruit from seedlings(butterfly pea and burgundy bean), and two short-term perennial legumes (butterfly pea and caatingastylo) sown with grass.Under the seasonal conditions encountered, in whichrainfall was near or below average, lablab was themost reliable and most easily managed, and also pro-duced the most liveweight gain. Economic analysiswould be needed to determine relative cost-effective-ness. A major cost in such a production system is theresowing of the lablab each year, but this may beoffset against its consistent and reliable establish-ment, which is a critical element of any ley pasturesystem (Bellotti et al. 1991).The annual legumes, V. trilobata and M. daltonii,successfully recruited from seed each spring. In thefirst season there was a high percentage of weeds,with legume contributing only about 60% of yield.With improved control of weeds in subsequent sea-sons, a higher proportion of legume was achieved.However, a chemical spray strategy that gave reliableweed control was difficult to develop under thevarying conditions, since plant growth was dictatedprimarily by the amount and timing of rainfall eventsin spring and early summer. A range of chemicalswas used, but the most successful approach was tospray with glyphosate as needed to control weeds(and kill legume seedlings) until a major rainfallevent. Although this strategy killed some legumeseedlings, there was little risk of re-establishmentfailing because of the high amount of soil seed. Asthe range of weeds with some tolerance to glyphosateincreased — mainly phasey bean (Macroptiliumlathyroides), which itself may have a role in ley pas-tures, and sour thistle (Sonchus oleraceus) —dicamba was added to the chemical mix. Given thedifficulties of managing V. trilobata and M. daltonii,it is doubtful that they offer any advantages over theeasily established lablab or the more perennial bur-gundy bean, butterfly pea and caatinga stylo.

Animal production of 60–120 kg liveweight gain/ha/year from all forage types compares favourablywith production from the higher-producing extensivegrazing areas, such as buffel grass on the fertile claysoils of the brigalow lands of Queensland (Mannetjeand Jones 1990) and productive native pastures over-sown with stylo (Partridge et al. 1996). This returnand the benefits of improved crop yields and proteincontent (McCosker et al. 2000) following lablab are



143

encouraging, given that with refinement in manage-ment there is potential to improve these benefits formany of the forage types.

In considering any possible improvement in leypastures for animal production, it is also necessary toconsider the characteristics that will maximise bene-fits to the subsequent crop phase. It seems desirablethat any ley pasture should have a high proportion oflegume, at least in the early years, to maximise Ncontribution. Choosing the species best suited to theley phase will depend on many factors, including thedesired length of the ley. For short (1–2 year) phases,lablab is well suited because of its easy and reliableestablishment, high production, minimal weed prob-lems and relatively easy management of grazing, andneeds to be more widely promoted to farmers.

Planted in pure swards, the other legumes (but-terfly pea, burgundy bean, M. daltonii and V. trilo-bata) established successfully and produced highlegume yields, especially in the first two years.Summer- and winter-growing weed species requiredregular chemical treatment in both the perennial andannual legumes. This, and the need to adjust grazingtimes to match growth of the single-species forage inresponse to soil moisture, complicated managementand compromised other objectives. Other disadvan-tages of using pure legume stands identified by Joneset al. (1991) are that the stands may not provide ade-quate mulch for control of soil temperatures andcrusting, that nitrogen mineralisation is too rapid, andthat the potential for rapid weight loss of cattle washigh if no alternative feed was available when thelegume deteriorated or became unpalatable. In thelonger term, there is the possible threat of soil acidi-fication (Noble et al. 1998), although this would nothappen quickly in these well-buffered soils.

In the legume–grass leys, which also provided highanimal production, weed invasion and grazing man-agement were less serious problems. However theamount of legume, which is critical to the amount ofN fixation (Hossian et al. 1995), varied considerably.In the case of butterfly pea, legume content declinedfrom 33% to 5% and then increased to 20%, while forcaatinga stylo, legume content increased from 5% to40%. The amount of soil seed also increased to levelsthat should ensure some recruitment. In fact, the soilseed reserve of butterfly pea sown with grass wasconsiderably higher than in the areas where it was notsown with grass, perhaps because of lower stockingrates over the period of seeding. This confirms theimportant role of grazing management in enabling

seed-set adequate for the persistence of annual orshort-term perennial legumes.

The preferred pasture to meet the objectives of leyswould be one in which legume is a high percentage ofyield in the early life of the ley, to maximise theamount of nitrogen fixed in the system. A minorgrass component could increase to benefit from thenitrogen, reduce weed invasion, increase soil organicmatter above that of legume alone, and add grazingvalue as the ley develops. The presence of a grass inthe system also ensures that the resulting organicmatter mineralises more slowly, at a rate commensu-rate with the needs of the ensuing crops.

Technologies identifying legumes for differentlengths of ley, and defining their managementrequirements to achieve the desired early growth andlater persistence, are now well into development.However, to achieve the desired legume–grass mixwill be more difficult because it will require moreprecise control of grass establishment. Sowing atlower seeding rates is unlikely to be effective becauseof the high variability and range of establishmentconditions that can be encountered. Sowing larger-seeded grasses is one option, but the range of desir-able grasses is limited. Purple pigeon grass (Setariaincrassata cv. Inverell) is one such species, althoughits relatively low palatability concerns some farmers.Another alternative is to oversow grass into estab-lished legume, perhaps in the second year, but this islikely to increase the risk of failure because grassseedlings may not be able to compete with estab-lished legume plants for moisture. A further option isto sow less vigorous grasses (Jones and Rees 1997).Queensland bluegrass (Dichanthium sericeum) is onesuch grass that has been used with some success(Clem et al. 2001).

Selection of other grasses better suited to the leyphase, testing of options for establishment and man-agement in crop–livestock systems, and refinementof legume management are needed so that the valueof ley pastures to animal and crop production, andultimately to the sustainability of the whole farmingsystem, can be realised.

Acknowledgments

I am grateful to staff at Brian Pastures ResearchStation for their technical support and to the BrianPastures Board of Governors and the Belmont BrianPastures Committee for their encouragement toproceed with the work. Financial assistance was pro-



144

vided by the Australian Centre for International Agri-cultural Research.

References

Bellotti, W.D., Bowman, A. and Silcock, R.G. 1991.Sustaining multiple production systems. 5. Sown pasturesfor marginal cropping lands in the subtropics. TropicalGrasslands, 25, 197–204.

Clem, R.L., Brandon, N.J., Conway, M.J., Esdale, C.R andJones, R.M. 2001. Early stage evaluation of tropicallegumes on clay soils at three sites in central and southerninland Queensland. Tropical Agriculture TechnicalMemorandum No. 7. Australia, CSIRO.

Cook, B.G. and Clem, R.L. 2000. Pastures on cropping soils:which grass for where? Tropical Grasslands, 34, 156–161.

Colwell, J.D. 1963. The estimation of the phosphorusfertilizer requirements of wheat in southern New SouthWales by soil analysis. Australian Journal ofExperimental Agriculture and Animal Husbandry, 3,190–197.Dalal, R.C., Strong, W.M., Weston, E.J. andGaffney, J. 1991. Soil fertility decline and restoration ofcropping lands in sub-tropical Queensland. TropicalGrasslands, 25, 173–180.

Dalzell, S.A., Brandon, N.J. and Jones, R.M. 1997.Response of lablab purpureus cv. Highworth,Macroptilium bracteatum and Macrotyloma daltonii todifferent intensities and frequencies of cutting. TropicalGrasslands, 31, 107–113.

Fitzpatrick, E.A. and Nix, H.A. 1970. The climatic factor inAustralian grassland ecology. In: Milton Moore, R., ed.,Australian grasslands. Canberra, Australian NationalUniversity Press.

Hossain, S.A., Waring, S.A., Strong, W.M., Dalal, R.C. andWeston, E.J. 1995. Estimates of nitrogen fixations bylegumes in alternate cropping systems at Warra,Queensland, using enriched 15 N dilution and natural15N abundance techniques. Australian Journal ofAgricultural Research, 46, 493–505.

Jones, R.J. 1972. The place of legumes in tropical pastures.ASPAC technical bulletin, No. 9. Taipei, Taiwan.

Jones, R.M. and Bunch, G.A. 1988. A guide to sampling andmeasuring the seed content of pasture soils and animalfaeces. Tropical Agronomy Technical Memorandum No.59. Australia, Division of Tropical Crops and Pastures,CSIRO.

Jones, R.K., Dalgleish, N.P., Dimes, J.P. and McCown, R.L.1991. Ley pastures in crop–livestock systems in the semi-arid tropics. Tropical Grasslands, 25, 189–196.

Jones, R.M., McDonald, C.K. and Silvey, M.W. 1995.Permanent pastures on a brigalow soil: the effect of

nitrogen fertilizer and stocking rate on pastures andliveweight gain. Tropical Grasslands, 29, 193–209.

Jones, R.M. and Rees, M.C. 1997. Evaluation of tropicallegumes on clay soils at four sites in southern inlandQueensland. Tropical Grasslands, 31, 95–106.

Keating, B.B. and Mott, J.J. 1987. Growth and regenerationof summer-growing pasture legumes on heavy clay soilsin southeastern Queensland. Australian Journal ofExperimental Agriculture, 27, 633–641.

Lloyd, D.L., Smith, K.P., Clarkson, N.M., Weston E.J. andJohnson, B. 1991. Ley pastures in the subtropics. TropicalGrasslands, 25, 181–188.

Mannetje, L.’t and Jones, R.M. 1990. Pasture and animalproductivity of buffel grass with siratro, lucerne ornitrogen fertilizer. Tropical Grasslands, 24, 269–281.

McCosker, K., Durkin, P., Schwarz, A. and Kelly, R. 2000.Lablab is a viable option for cropping rotations in centralQueensland. Tropical Grasslands, 34, 308.

Noble, A.D., Thompson, C.H., Jones, R.J. and Jones, R.M.1998. The long-term impact of two pasture productionsystems on soil acidification in southern Queensland.Australian Journal of Experimental Agriculture, 38, 335–43.

Partridge, I., Middleton, C. and Shaw, K. 1996. Stylos forbetter beef. Information series Q1 96010. Brisbane,Department of Primary Industries.

Pengelly, B.C. and Conway, M.J. 2000. Pastures oncropping soils: which tropical pasture legume to use.Tropical Grasslands, 34, 162–168.

Reid, R.E., Sorby, P. and Baker, D.E. 1986. Soils of theBrian Pastures Research Station, Gayndah, Queensland.Brisbane, Department of Primary Industries, ResearchEstablishments Publication QR86004, 62 p.

Tothill, J.C., Hargreaves, J.N.C., Jones, R.M. andMcDonald, C.K. 1992. BOTANAL – A comprehensivesampling and computing procedure for estimating pastureyield and composition. 1. Field sampling. Technicalmemorandum no. 78. Australia, Division of TropicalCrops and Pastures, CSIRO.

Vallis, I. 1972. Soil nitrogen changes under continuouslygrazed legume–grass pastures in subtropical coastalQueensland. Australian Journal of ExperimentalAgriculture and Animal Husbandry, 12, 495–501.

Weston, E.J., Doughton, J.A., Dalal, R.C., Strong, W.M.,Thomas, G.A., Lehane, K.J., Cooper, J.C., King, A.J. andHolmes, C.J. 2000. Pastures on cropping soils: managinglong-term fertility of cropping lands with ley pastures insouthern Queensland. Tropical Grasslands, 34, 169–176.

Winter, W.H., McCown, R.L. and Zuill, D. 1996. Legume-based pasture options for the live cattle trade from theAustralian semi-arid tropics. Australian Journal ofExperimental Agriculture, 36, 947–955.



Participative Methodology and Adoption of Technologies

145



146

Identifying the Factors that Contribute to the Successful Adoption of Improved Farming Practices

in the Smallholder Sector of Limpopo Province, South Africa1

T.S. Ndove,*,† A.M. Whitbread,§‡ R.A. Clark** and B.C. Pengelly‡

Abstract

A formidable challenge that faces the governments of many developing nations is the maintenance of foodsupply to an ever-growing population and the protection of the natural resource base. Despite the existence ofmany relevant and cost-effective technologies, and the continued focus of researchers on finding newtechnologies that enhance crop and livestock productivity, the smallholder farming sector remains relativelyunproductive. Understanding the key factors that contribute to adoption of technologies that improve farmproductivity is paramount if sustainable farming production is to be achieved within the smallholder sector.This paper reports on the key factors that have enabled a participative multidisciplinary R&D project to makesignificant impact in the adoption of fertiliser and legume technologies by a smallholder farming communityin South Africa.

Using semistructured and group interviews, information was collected on the process of practice changeresulting from interventions (farmer team and network development, training, on-farm experimentation,demonstrations) during the period November 2002 to January 2003. The study has revealed that farmers haveadopted improved farming practices, such as fertiliser use, row planting, intercropping of legumes, and pestand disease management. Capacity building of the farmers associated with the project has resulted in thedevelopment of significant leadership and organisational capacity to sustain improvement of farmperformance. Changing the focus from subsistence to producing for profit has been a great achievement withinthis community. Despite the success of the R&D project, the high age of the farmers (who are predominantlywomen older than 60 years) and frequent droughts are likely factors impeding continuity and sustenance ofproject achievements.

1 This paper was first published as: Ndove, T.S.,Whitbread, A.M., Clark, R.A. and Pengelly, B.C. 2003.Identifying the factors that contribute to the successfuladoption of improved farming practices in thesmallholder sector of Limpopo Province, South Africa.In: Robinson, J.B. and King, C.M., ed., Proceedings ofthe First Australian Farming Systems Conference <http://afsa.asn.au>. It is reprinted here with permission.

* MSc candidate, University of Queensland, St Lucia 4067,Queensland, Australia.

† Extension officer, Department of Agriculture, Limpopo, P.O. Box 1273, Letaba 0870, Republic of South Africa.

§ Department of Plant Production Systems, Wageningen University, P.O. Box 430, 6700 AK Wageningen, The Netherlands.

‡ CSIRO Sustainable Ecosystems, 306 Carmody Road, St Lucia 4067, Queensland, Australia.

**Department of Primary Industries, G.P.O. Box 46, Brisbane 4001, Queensland, Australia,



147

The most common worldwide challenge in thetwenty-first century is the supply of the ever-growingpopulation with sufficient food and fibre. Singh et al.(1991) stated that agricultural scientists andresearchers have, over the past decades, developedimproved methods and practices that can increaseproduction. Although many of these new agriculturaltechnologies have helped raise farm production andefficiency, the productivity of farmers in developingcountries still remains poor, and adoption ofimproved technologies is less than desired. Thesehave widened the gap between the performance ofthe new technologies at the experimental station andin the farmer’s field. As noted by Singh et al. (1991),only the large-scale, wealthier farmers have adoptedmodern farming innovations, while the smaller andpoor farmers have been unable to make use of thesame technologies.

The Limpopo Province of South Africa dependseconomically on agriculture, mining and tourism. Themajority of the population in this province lives inrural areas, where agriculture has been the main sourceof their local economies and development for manydecades. Government, non-government, local andinternational research and development (R&D) organ-isations have been engaged in improving the farmingpractices and profit of the smallholder farmers; how-ever, the level of adoption has remained low andunsustainable. Over the years there have been largeR&D inputs to develop appropriate methods that canbe used to achieve improved practices and outcomesfor food security and profit. A recent popular methodis the ‘participatory’ approach.

Although large-scale improvement in smallholderfood security and profit has not been widelyachieved, Singh et al. (1991) recorded that, whereparticipatory methods were implemented, they led toimproved adoption of R&D practices and achieve-ment of increased production by farmers. Socioeco-nomic differences have been found to be importantfactors that influence the adoption of new technology(Lockie and Vanclay 1992). De Sousa Filho (1997)has also postulated that there is a need to study thedeterminants of technology adoption and diffusion ofsustainable agricultural technologies and practices,particularly in the smallholder farming situation.

This study was aimed at a better understanding ofthe key factors contributing to the adoption ofimproved agricultural practices that impact on theachievement of increased smallholder production,profit and capacity.

Overview of Methodologies

Background to the study site

This multidisciplinary R&D project, operatingfrom 1999 to 2002, was a partnership between theAustralian Centre for International AgriculturalResearch (ACIAR) and South Africa’s LimpopoDepartment of Agriculture. The project targeted onecommunity called Dan in Limpopo Province with theaim of improving the dryland cropping and livestockproduction systems of the small-scale farmers. Themain objectives of the project have been to increasethe crop yield and to improve the quality and yield oflivestock through the use of legume species that areadapted to the local environment. The community ofDan was used for a pilot study because it has the rep-resentative features of the Mopani District, which isone of the six districts of Limpopo Province.

Biophysical environment

Dan is situated at 23∞55'S 30∞16'E, 13 km east ofTzaneen town, and falls under the Greater TzaneenMunicipality of the Mopani District. The main soilsused for crop production range from coarse-grainedsandy soils to sandy loams derived from graniticparent materials. The organic carbon content is gen-erally less than 1.0%, and the plant available nitrogenand phosphorus concentrations are severely limitingto plant growth. The soils are acidic with pH valuesbelow 5.8. The sandy nature of the soil results in alow effective cation exchange capacity and theleaching of nutrients, particularly N, S and K. Muchof the cropping lands are susceptible to waterloggingduring heavy rains, and subsoil drainage is poor dueto an underlying dense clay layer.

The annual average rainfall is 759 mm (based on theavailable weather files at the nearby Letaba-Letsiteleweather station). The rainfall pattern is stronglysummer dominant with 86% of the rainfall receivedfrom October to the end of March. The minimum tem-perature of ª6.5∞C is reached in June and July and themaximum of ª32∞C in December and January.

Community structures

Dan has a reported population of 15 000 people.The cropping area is 200 ha with 300 farmers (withindividual allocations of between 0.5 ha and 1.0 ha).Dan also has communal grazing land of 400 ha, withabout 1000 head of cattle (individual ownership



148

ranges between 1 and 100 head). The majority of thefarmers are elderly women with an average age of 65.Their only source of income, apart from agriculture,is from government social grants that amount toR550.00 (A$110.00) per month. The majority of thefarmers belong to the Dan Farmers Association.

Farming systems

Most of the farmers plant maize every season, andit is used as the staple food. Maize is harvested asimmature cobs, ‘green mealies’, or at maturity andmilled for home consumption. Cowpeas and ground-nuts are commonly intercropped with maize. Thecropping area is used as a communal grazing areabetween April and October. Any crop that is stillstanding after April will be grazed by livestock, andthis makes it difficult to plant crops of long duration,such as perennial plants and ley legumes.

While the majority of farmers don’t own livestock,grazing on cropping areas is open to any livestockowner. Land preparation is done by hired tractors andin many cases farmers plant late because tractors arenot available. Broadcasting is the most common wayof sowing seeds. While some farmers use newly pur-chased hybrid seeds, the majority of them still userecycled seed from previous harvests. Cowpea isusually planted by hand between the maize plants afterthe first weeding of the maize. Most of the farmers donot apply fertiliser; however, a growing number applysmall amounts of inorganic and organic fertilisers.Organic fertilisers are usually applied in the form ofkraal manure at a rate of between 0.5 and 1.5 t/ha. Theproduce from the fields is generally used for homeconsumption and very little is traded.

Land tenure

The land is communally owned, and farmers havepermission to occupy (PTO) rights only. The tradi-tional authority allocates and controls the land. Thesystem has not made enough room to accommodatenew entry for new residents or exit strategies for theold participants. There is no strict policy on fallowland, even if it remains unploughed over many years.

Survey methods used

This study was conducted between November2002 and January 2003, was focused on the partici-pants of the ACIAR project, and employed a case-study approach that used both qualitative and quanti-

tative techniques. The case-study approach hasenabled the use of a variety of methods and tech-niques that focus on processes rather than outcome,and more on discovery than confirmation (Burns1997). Multidisciplinary participative research wasused as the model in this study. This method isderived from the principles of action research, whichare described by Zuber-Skerrit (1991) as being prac-tical, participative and collaborative. In this method-ology the researcher is not an outside expertconducting an inquiry with the ‘subjects’, but a co-worker doing research with the people and a practicalproblem. Semistructured interviews with open-endedand closed questions were used with key respond-ents. Group interviews were also used to inform, val-idate and triangulate the result of the semistructuredinterviews, by running four group interviews withsome respondents (Burns 1997). The reason for thisapproach was that the farmers work in groups, and itremains important to validate the individualresponses by going through the same questionnaireusing group interviews. Secondary data obtainedfrom baseline studies, combined with data from theLimpopo Department of Agriculture and the farmers’own records, were used to benchmark the farmers’practices before the start of the pilot project.

Sampling method

Of the 32 pilot project participants, a random selec-tion of 22 of those who were available for both theinterview sessions was chosen to participate in thestudy. The data gathered from the interviews wereclassified, tabulated and analysed in accordance withthe objectives of the study. The main questions askedboth in individual and in group interviews were aimedat gaining understanding of the factors that haveimpacted on the adoption or non-adoption of improvedagricultural technology. The questions focused on soilfertility knowledge (soil fertility management, types offertilisers used), agronomic practice (intercropping,weeding and pest management), capacity and organi-sational capacity, as well as economic questions (eco-nomic objective of farming, knowledge of profit andloss). With regard to practices, farmers were askedabout their knowledge and where and when theyacquired it. Unstructured interviews with key farmerinformants and some other key participants were alsoconducted. Observation of farmers’ practices was alsodone to gain more understanding of their knowledgeand skills.



149


Summation of the project activities

After unstructured and semistructured interviewswith the key farmer informants, some extension per-sonnel and the researchers who participated in theACIAR project at Dan recorded a chronologicalsummary of key inputs between 1999 and 2002(Table 1). These activities were confirmed by all theparticipants who were interviewed as a true reflectionof the activities that the project has embarked upon.Activities such as the educational tours, meetings,and training sessions were repeated at intervals overthree seasons, while others such as the baseline studyand project introductions were done once. Table 1

also reveals that in 1999 most of the activities weregeared at consultation with the stakeholders, whichincluded farmers, extension personnel and commu-nity leaders. The purpose of the project was alsointroduced and negotiated with all relevant stake-holders.

Social classification of the respondents

All but one of the 22 respondents in the semistruc-tured interviews were female. The average age of therespondents was 63 years, which is below the Danfarmers’ average age of 65 (information obtainedfrom the baseline study in 1999). Forty-five per centof the respondents were in the 61–70 age range. Thehigh age of the respondents limits the availability of

Table 1. Summarised timelines of project activities, 1999–2002.

1999 2000 2001 2002

• Introduction of the project(Local extension, farmers and farmer organisation)

• Farmers’ meeting monthly.

• Project team meets bi-monthly

• Assessment and planning• Training need

assessment

• Farmers’ meeting monthly

• Establish another soil fertility team


• Training needs assessment

• Farmers’ meeting monthly

• Continuation of soil fertility team


• Training needs assessment

• Community consultation(Met with farmer organisation, local extension personnel and other stakeholders)

• Soil fertility teams formed

• Training sessions on soil fertility and the role of legumes

• Establishment of farmers’ experiment on soil fertility.

• Training on soil fertility, how to use different fertilisers, deficiency symptoms, application methods

• Continuation of farmers’ experiment on soil fertility. Comparing application of different levels of fertilisers and non-fertiliser application

• Training simple gross margin

• Continuation of farmers’ experiment on soil fertility. Comparing application of different levels of fertilisers and non-fertiliser application

• Meeting of project leadership with the farmers

• Demonstration on row planting, fertiliser application

• Demonstration on strip, inter-row and relay intercropping

• Follow-up on intercropping, soil fertility

• Baseline study done by a consultant

• Educational tours• Participative evaluation

of the year


of the year


of the year

• Needs assessment based on the baseline report

• Annual project workshop• Establishment of

technical on-farm experiment on legumes and intercropping

• Annual project workshop• Technical on-farm

experiment on legumes and intercropping

• Project evaluation workshop

• Technical on-farm experiment on legumes and intercropping



150

labour, and only a few can afford to hire labour whenit is required.

The level of education ranges from very low to noeducation at all (Table 2). Sixty-three per cent of therespondents cannot read or write, while those whocan are limited to the indigenous language. The cre-ativity of the extension personnel therefore becomesthe most important factor in transferring skills andtechnology. The role of extension personnel, withsupport from the researchers, has been very influen-tial in this type of work with the farmers. On-farmexperimentation, demonstrations and tours have beenhighly regarded by farmers.

Responses of farmers

Soil fertility management and fertiliser useThere has been a consistent increase in the use of

inorganic fertilisers by the respondents from 1999 to2001 (Table 3). In 2002, the majority of the respond-ents were using fertilisers. The knowledge of thefarmers about the use of fertilisers has improved con-siderably and can be attributed to the demonstrationsand on-farm experiments that farmers were partici-pating in since 1999. The high price of the fertilisersand the consistent drought have emerged as thegreatest deterrent to the use of fertilisers. However,respondents remain convinced that rainfall will

improve in future and that their yields and profits willimprove. Although lime is less expensive than inor-ganic fertilisers, only four respondents had used it.This low utilisation of lime may be associated with alack of understanding of the benefit of lime whencompared with nitrogenous fertilisers, from whichcrops may show instant responses. Although lime isless expensive per kilogram than inorganic fertiliser,the large amount of lime required for application (inexcess of 500 kg/ha) and transportation costs or dif-ficulties contribute to its reduced use. The consistentincrease in the use of inorganic fertilisers from 1999to 2001 may indicate that farmers have attributed theobserved good performance of crops in demonstra-tion plots to the use of fertilisers. This process offarmers’ practice change in adopting fertiliser use hasalso been observed by Nkonya et al. (1997), whorevealed that demonstration of technologies in thefarmers’ fields seemed to be most appealing tofarmers.

Pest managementThe respondents have observed an increase in the

incidence of aphid infestation of legume crops suchas cowpea and lablab. This was also supported by theextension and research team, who indicated that therewere no aphid or beetle infestations on legumes war-ranting control measures until the use of exotic

Table 2. Social classification: age and educational achievement of respondents.

Age No. ofrespondents

No schooling Primaryschooling

Adult literacy schooling

41–5051–6061–7071–80

16

105

–284

–111

131–

Mean of age: 63.

Table 3. Dan farmers using soil fertiliser amendments prior to and during the project period.

Amendmenta No. of respondentsb

Before 1999 1999 2000 2001 2002

2:3:2LANUreaLimeKraal

21191444

75512

22122

5431–

675––

11–––

a Amendments: 2:3:2 = 11:22:11% N:P:K; LAN = lime ammonium nitrate (34% N); Urea = 46% N; lime = calcium carbonate (±Mg); kraal = manure of various qualities.

b Multiple responses could be obtained from a total of 22 respondents.



151

legume cultivars (Table 4). The indigenous cowpeacultivars never had pest problems until the introduc-tion of the exotic cultivars. Adequate legume growthdepends, to a large extent, on the chemical control ofthese pests. This adds another economic burden to thealready resource-poor farmers. The benefits of pestcontrol measures, generally one spray per growingseason, far outweigh the cost of spraying. Therespondents, realising these advantages, seem to beprepared to use pest control measures, as indicated bythe increasing use of chemicals. Besides poor soil fer-tility and droughts, the largest constraint on maizeproduction is the maize stalk-borer. Increased pesti-cide use for control of maize pests may have alsoresulted from the need to control pests on legumes.

Intercropping and row-sowing practicesThere has been an increase of 50% in the use of row

sowing by respondents since 1999 (Table 5). Themajority of respondents have adopted row sowingover their traditional method of broadcasting seeds.The on-farm experimentation and demonstrations ofrow planting have led to increased adoption of thispractice. The farmers saw row-planting practices ascompatible with other agronomic practices of band-placement of fertilisers and pest and weed control.

While farmers have traditionally practised inter-cropping in various forms, accompanying technolo-gies such as weeding, fertiliser application and pestcontrol measures were not always integral parts ofthe system. The respondents preferred strip inter-cropping over row and relay intercropping, as itseems make it easiest for them to do agronomic man-agement practices such as fertiliser application,disease and pest control and harvesting when thesame crops are grouped together in strips.

The beneficial effects of legumes have alsobecome drivers of the adoption of intercropping.The capacity-building activities (Table 1) have ledto an understanding that the benefit of the legumesto maize is derived from the supply of nitrogen. Theeducational sessions addressed the fact that nitrogensupply is not directly between legumes and maize,but occurs over several seasons through soil fertilitybuild-up. The increase in the farmers’ under-standing of legume technologies has made this prac-tice even more relevant. The fact that farmers havebeen intercropping for decades made it easy forthem to adopt this practice, which was simply mod-ified to further satisfy their original need to producemore crops.

Table 4. Pest control measures applied prior to and during the project.

Pest No. of respondentsa

Before 1999 1999 2000 2001 2002

Stalk-borerAphidsCutwormsBeetlesLocustNo control

17104112

322–12

1––––

33–––

841––

2111–

a Multiple responses could be obtained from a total of 22 respondents.

Table 5. Sowing arrangement practices prior to and during the project.

Type of sowing arrange-ment

No. of respond-

entsa

Before 1999 1999 2000 2001 2002

Strip-intercropb

Row-intercropc

Relay-intercropd

Row sowing — sole crop

2135

17

1––6

21–1

7122

9127

2–11

a Multiple responses could be obtained from a total of 22 respondents.b Strip-intercrop: alternate strips of multiple rows (4–6) of legumes and maize.c Row-intercrop: alternate single rows of legume and maize.d Relay-intercrop: delay in the planting of the legume between rows of maize.



152

The importance of profitability as a driving forceSeventy-six per cent of the respondents have indi-

cated their understanding of profit, and that profit canonly be achieved when input costs are less than theproduct sales (Table 6). Taking into account theirlevel of education and age, this is a remarkablechange in their view of farming. However, they arealso aware that soil fertility, water, and control ofpests and diseases can improve the quantity andquality of their yield and improve their profit margin.Fifty-four per cent indicated that, despite their smallcropping areas, their aim is to farm for food and tosell the surplus production. With an average land-holding of 0.5 ha and frequent droughts, the produc-tion of these farmers is seriously limited. Reducedinput costs and better rainfall patterns can improveboth yield and profit for the farmers if other agro-nomic practices are done. Increased farmer under-standing of economics, particularly profit and loss,has been achieved despite the persistent droughtscausing serious crop losses.

Conclusions

The adoption of improved technology seems to beinfluenced by many factors, ranging from environ-mental factors, farmer type and the methods used byextension agents, to socioeconomics. In this paper wehave tried to determine the key factors that influencethe adoption of improved agricultural technologiesand practices.

The main barrier to adopting such technologies asfertiliser and hybrid seed use has been determined asthe purchase cost. Despite the potential benefits ofthese technologies, farmers are reluctant to invest inthem. As noted by Caswell et al. (2001), cost is abarrier to the adoption of improved technology.These authors found that there is a differencebetween farmers who do not want to adopt practicesand farmers who fail to adopt due to barriers such ascost. Even the utilisation of inputs such as kraal

manure, which is readily available for some farmers,was restricted by the costs of transportation.

Methods used by the extension personnel andresearchers in building capacity for the farmers havecontributed to the increase in adoption of most of thepractices. The use of training sessions, on-farmexperimentation, demonstrations and education tourshave greatly influenced the farmers to adopt technol-ogies such as fertilisers and intercropping, and agro-nomic practices such as pest and weed management.Despite the high age of the farmers, the consistent useof these different methods systematically over threeyears (Table 1) has gradually built confidence forfarmers to adopt improved practices. The use of feed-back loops in the form of regular assessment andplanning sessions (Table 1) has enhanced the effec-tiveness of the methods used to impart knowledgeand skills to the farmers.

The relevance of the new or improved technologyto the farmers and the need of the farmers to improvetheir situation have been observed to be importantfactors of adoption. Most of the practices that wereintroduced were the result of farmers themselvesidentifying problems that reduce their yield, anddemonstrating their willingness to improve theirfarming situation. The need of the farmers to learn isconsistent with the principles of adult learning, whichsay that the adult should feel and recognise the needto learn (Malouf 1994)

Persistent drought has contributed much to the dif-ficulties the farmers have faced in implementingsome of practices learned. Although some of the agri-cultural practices were successful in experimentaltrials, at demonstration sites and on the toured sites,consistent failure due to drought has reduced theadoption of most practices. Environmental factorssuch as rainfall have been found to be very importantin determining adoption and practice change, as allfarmers have highlighted drought or poor rain distri-bution as the major constraint.

In this study, many factors that affect adoption ofimproved practices have been identified. The cost ofthe improved practices, the extension methods, theage of the farmers and poor rainfall have emerged asthe main factors determining adoption. Furtherdetailed studies need to be conducted in otherfarming communities. The farmers at Dan can also beused to motivate other farmers in other R&D pro-grams in the region, because they have achieved somuch under difficult conditions.

Table 6. The end-point use of produce and theconcept of profit.

No. of respondents

Produce for foodProduce for food and sellingKnowledge of profitNo knowledge of profit

101216

6



153

Acknowledgments

First and foremost, we appreciate the level of under-standing and cooperation that we have been given bythe Dan community during the entire period of thestudy. The willingness of Department of Agriculturein Limpopo Province to be supportive and coopera-tive is also highly appreciated. We also acknowledgethe ACIAR, which sponsored this study.

References

Burns, B. 1997. Introduction to research methodology (3rded.). Sydney, Addison Wesley Longman.

Caswell, M., Fuglie, K., Ingram, C., Jans, S. and Kascak C.2001. Adoption of agricultural production practices:lessons learned from the US Department of AgricultureArea Studies Project. Resource Economics Division,Economic Research Service, US Department ofAgriculture. Agricultural Economic Report No. 792.

De Souza Filho, H.M. 1997. The adoption of sustainableagricultural technologies: a case study in the State ofEspírito Santo, Brazil. Ashgate, Aldershot, UK.

Lockie, S. and Vanclay, F. 1992. Barriers to the adoption ofsustainable crop rotation; focus group report. Report toNSW Agriculture, Wagga Wagga. Centre for RuralSocial Research, Charles Sturt University, Australia.

Malouf, D. 1994. ‘Understanding learning’: how to teachadults in a fun and exciting way. Chatswood, Australia,Business and Professional Publishing.

Nkonya, E., Shroeder, T. and Norman, D. 1997. Factorsaffecting adoption of improved maize seeds andfertilisers in northern Tanzania. Journal of Economics,48, 1–2.

Singh, S.W, Vijayaraghavan, K. and Haque, T. 1991.Transfer of technology to small farmers (an analysis ofconstraints and experiences). New Delhi, India, ConceptPublishing Company.

Zuber-Skerrit, O., ed. 1991. Action research for change anddevelopment. Aldershot, United Kingdom, GowerPublishing Co.



154

Contrasting Adoption, Management, Productivity and Utilisation of Mucuna in Two Different Smallholder Farming Systems in Zimbabwe

B.V. Maasdorp, O. Jiri and E. Temba*

Abstract

As part of a project to address smallholder problems of soil infertility and feed shortage in sub-humid areas ofZimbabwe through integration of forage legumes in cropping systems, farmers with typically intensive mixedcrop–livestock production were supplied with 5 kg of mucuna seed (Mucuna pruriens var. utilis) in October2000. The aims were to assess the likelihood of adoption, evaluate mucuna production and use by farmers, andprovide information for future research and extension. The two participating farming communities representedthe two major smallholder farming systems in Zimbabwe, communal and resettlement, and were targetedbecause of their commitment to semicommercial ruminant livestock production. They were in the communalarea of Dendenyore (cattle pen-fattening) and the Zana resettlement area (dairying). Total arable landholdingswere 5 ha in Zana and an average of 2.7 ha in Dendenyore. Whereas the communal farmers of Dendenyorecropped primarily maize at a subsistence level, farmers in Zana achieved higher yields and also cash-croppedtobacco and paprika. Monitoring of the 37 farmers given seed occurred during 2000–01 and of these plus over45 new growers in 2001–02.

Mucuna was produced and used more successfully, and there were more positive signs of adoption, by thecommunal area farmers (higher yields, more timely planting, higher seeding rates, better protection frominvading livestock, increase rather than decrease in area planted, adequate rather than inadequate seedretention). There was no interest in using mucuna as green manure by these livestock producers, and onlyshort-lived interest in intercropping it. They all used mucuna for pen-fattening rations or dairy feed. Attemptsin Zana to make hay were not very successful, as harvesting was late (May) because of competition for labourfrom other crops. Mucuna hay was not easy to protect from marauding livestock and not easy to feed. Hayharvesting was later largely abandoned, and pod harvesting favoured. The communal farmers mostly harvestedpods, let their livestock graze the residues in the field, fed seed and pods, and also sold seed for mucunaplanting (including to some Zana farmers). Despite little soil amendment use on mainly depleted granite sands,mucuna yields were considerable, averaging 3 and 1.5 t/ha for pods and seed, respectively, in Dendenyore in2000–01 with rainfall of 864 mm.

We concluded that mucuna had good potential for integration into smallholder systems in sub-humid areas,particularly to address problems associated with livestock feed. The cash-cropping resettlement farmers hadmore serious labour constraints than the communal farmers, and these reduced their capacity to grow andmanage mucuna well. In contrast to the resettlement farmers, who were better endowed with resources, thecommunal farmers were much more interested in selling mucuna seed.

* Crop Science Department, University of Zimbabwe, POBox MP 167, Harare, Zimbabwe.



155

Poor soil fertility and inadequate forage supplies aremajor constraints in the mixed crop–livestocksystems typical of smallholder farming in Zimbabwe.Soils are largely infertile granitic sands that havecommonly been subjected to prolonged monocrop-ping of maize with minimal fertiliser use. The majorsources of livestock feed are crop residues andnatural grazing, which are insufficient in both quan-tity and quality, particularly in the lengthy dryseason. Commercial feeds are largely unaffordable.Overstocking and degradation of communally grazedrangelands are a disincentive for interventions toimprove their productivity.

Integration of forage legumes into the croppingsystems might alleviate these problems, providinghigh N biomass for soil amendment and/or forage.This applies particularly in the wetter areas withhigher potential for cropping (Natural Regions [NR]2 and 3 with average annual rainfall of 650–1000 mmand 650–800 mm, respectively, falling mostlybetween November and March). Such legumes canbe integrated as a cereal intercrop or in rotation as aley or green manure. Setting aside of arable land as aweedy fallow is a common practice (Nyakanda 2003;Chuma et al. 2000) and would facilitate this use oflegumes. One of the main reasons given for the use ofweedy fallows in Nyakanda’s survey was exhaustedsoil fertility. About 25% of farmers held the view thatfallowing would result in some restoration, thoughchemical analysis and following crop yields did notsupport this. Interventions such as sowing legumes toincrease the effectiveness of fallows therefore seemneeded.

Despite the successful development of foragelegume technologies, adoption in Africa and Asia hasgenerally been poor (Thomas and Sumberg 1995;Horne and Stur 1997). This has been largely attrib-uted to lack of involvement by farmers in develop-ment of research priorities (Horne and Stur 1997).With this in mind, full farmer participation was anintegral part of a project on the introduction and inte-gration of forage legumes into smallholder croppingsystems. The project was conducted in two areas ofcontrasting tenure and land use, communal and reset-tlement, representing the two most common small-holder farming systems in Zimbabwe.

In the communal farming system, households areallocated arable land on a more or less permanentbasis by traditional leadership, while grazing andwoodland resources are used communally. Arablelandholdings vary in size, but are typically 2 ha per

household, and there are no restrictions on livestocknumbers. This farming system is not sustainable andcrop production is largely at a subsistence level.

Apart from the current (2000–03) land redistribu-tion exercise, resettlement areas were created fol-lowing Zimbabwe’s independence in 1980, whencentral government acquired 3.3 million ha of landfrom large-scale commercial farmers to resettlehouseholds from densely populated communal areas(Gore et al. 1992). Model ‘A’ resettlement schemescomprise intensive village settlement with individualarable land allocation and communal access tograzing. In NR 2, each household has access to 5 haof arable land and grazing rights for five livestockunits. The original model ‘A’ blueprint, to providefor a reasonable crop rotation on the lighter, lessfertile soils, was the cropping of only 3 ha and fal-lowing of the other 2 ha (Ivy 1983).

As the 2000–01 rainy season approached, it wasthought that it was about time that farmers were pro-vided with legume seed to grow as they wished. Thiswould be the fourth growing season in which theproject had contact with the target smallholder com-munities. After two seasons during the planningphase, 2000–01 was the second season of on-farmtrials. Farmers were impatient for the project to getunder way in 1999–2000, and had now been exposedto a year of legume fodder production and feedingtrials. The adoption process needed to be facilitatedand encouraged by supplying farmers with legumeseed.

Mucuna was selected as the main legume in theintercropping and ley trials because seed of thislarge-seeded, well-adapted, productive legume waseasy to produce in large yields, and it was readilyavailable. Seed of the smaller-seeded forage legumeshas gradually disappeared from the formal market inrecent years, as large-scale farmers have stopped pro-ducing it because of increasing costs of productionand diminishing demand, amongst other factors.Seed of Lablab purpureus, the large-seeded speciesgrown in the project for the first animal feeding trials,was not readily available either, as it is not easy toproduce a good seed crop unless intensive measuresare taken to control pests during flowering and pod-ding.

Mucuna seed was distributed to promote adoptionof ley legume technologies. Further objectives,achieved through subsequent monitoring, were togain information on productivity under farmer man-agement, to assess the relevance of the legume tech-



156

nologies developed and the likelihood of foragelegume adoption by farmers, and to identify influen-tial factors, particularly constraints arising fromfarming practices and socioeconomic factors. Thisincluded eliciting farmers’ perceptions and prefer-ences about mucuna production and use, andassessing whether their needs are met by the avail-able mucuna technology. Such information would beuseful for decisions about future research and exten-sion to improve the appropriateness and adoption ofthe technology.

Methods

Details of the study areas

This study was conducted in the neighbouringwards of Dendenyore and Zana in Wedza District,about 140 km southeast of Harare.

Dendenyore is in a communal area under cultiva-tion since the 1930s, and Zana is part of a model ‘A’resettlement scheme formed in 1986. They lie inNR 2B, which has less reliable rainfall than NR 2A.The region can be subjected to severe dry spellsduring the rainy season, and occasionally has rela-tively short rainy seasons (Ivy 1978). The area isabout 1400 m above sea level.Soils are mostly acidic sands derived from granite.Chemical analysis revealed no notable differences infertility between Dendenyore and Zana sands; theaverages for six sites are presented in Table 1. Thesoils have low concentrations of all major nutrients, alow cation exchange capacity and very low organiccarbon content.

The population pressure in the communal area isconsiderably greater than in the resettlement area,with total land area and population for Dendenyoreand Zana being 14 028 and 5 187 ha and about15 000 and 1890 persons, respectively (Chigariro

2004). Average household sizes were similar, at fivepersons in Dendenyore and six in Zana.

Intensive mixed crop–livestock farming is con-ducted in both areas. Cropping is rain-fed. In Dende-nyore, the main crop is maize (the staple diet), withyields averaging about 1 t/ha, although the potentialis about 5 t/ha. Other, more minor field crops are mil-lets, sorghum, groundnuts, sweet potatoes, sun-flowers, cowpeas and soybeans. Production is atsubsistence level, with any surplus marketed. In addi-tion to these crops, the Zana resettlement farmers alsoproduce cash crops of tobacco and paprika, andachieve maize yields of at least 2 t/ha. There was nofodder production before the inception of the project(Chigariro 2004).

The average total arable landholding is 5 ha inZana, with about 3 ha cropped each year, and about2.7 ha (0.8–5 ha) in Dendenyore, with about 2 ha(0.8–3.8 ha) cropped. Cattle are mainly kept fordraught, manure and investment purposes. Herd sizesin Dendenyore are 0–4 in the poorest, 5–9 in themiddle and >10 in the wealthier households (Peng-elly et al. 2003). In Zana, farmers own 2–4 dairycows in addition to other cattle. More householdsown goats, kept primarily as a ready source of cashincome. In Zana, land preparation is mostly by tractor(with assistance from the District DevelopmentFund), whereas Dendenyore farmers rely on draughtanimals.

Demographic characteristics differed between thetwo areas. The age of participating farmers rangedfrom 20 to 76 years in Dendenyore and from 20 to 65years in Zana, while the male:female ratios for thesefarmers were about 1:2 and 3:1, respectively(P. Mazaiwana, pers. comm.).

The communities of Dendenyore and Zana weretargeted for the forage legume introduction project,and consequently this mucuna adoption study,because of their pre-existing commitment to beef fat-

Table 1. Soil chemical properties of granite sands averaged over six sandy sites from Dendenyore and Zana.

Soil depth (cm) pH Ka

(cmolc/kg)Mg

(cmolc/kg)Ca

(cmolc/kg)Available

Pb

(mg/g)

Total N(%)

C(%)

0–1515–30

4.314.14

0.110.09

0.060.05

0.220.17

126

0.040.04

0.350.26

a Equivalent to meq/100gb Bray PSource: adapted from Jiri (2003)



157

tening and dairy enterprises, respectively. TheWedza Feeders Association was formed in the 1950sand has a total of over 200 members from nearly all14 wards, including two resettlement areas. Therewere 65 pen-fatteners in Dendenyore. The WedzaDairy Association was formed by Zana farmers in1992, and 54 aspiring dairy farmers from Dende-nyore have now also registered. Before the project,summer milk yields with spring calving were 4–6 L/day. Milk production in the dry season declines to aslittle as 1–2 L/day, limited by lack of adequategrazing and supplementary feeds. With legumefeeding, however, these production levels increasedto 5–17 and 2–6 L/day, for summer and winterrespectively (Chigariro 2004). Similarly, Dende-nyore farmers depended on bought feeds for pen-fat-tening, but have now reduced this input by about25%.

The ratio of agricultural extension officers tofarmers is 1:1200 in the communal area and 1:400 inthe resettlement area (Chigariro 2004).

Seed distribution

Mucuna seed was issued to interested farmers inZana and Dendenyore in October 2000. Fifteenfarmers in Zana and 22 in Dendenyore were givenseed, mostly 5 kg each (for an anticipated 0.1 ha).Farmers were given seed on the understanding thatthey would pass on a similar amount of seed fromtheir own crop to another farmer. All the farmers whohad hosted legume ley trials the previous season (fivein Zana and four in Dendenyore) accepted seed. Theremainder of the farmers given seed were identifiedby government extension staff.

Monitoring

Monitoring of the farmers and their mucuna cropswas done by extension staff during the 2000–01growing season, and more formally by researchersfollowing harvesting in the dry season of 2001.

Further monitoring was done in 2001–02, of the2000–01 growers and ‘new’ adopters (i.e. those whogrew mucuna in 2001–02 from seed they had beengiven or been sold by other farmers). In addition tomonitoring in Dendenyore and Zana, extension staffreported on the diffusion of this legume technologyto other wards within the district.

Agronomic practices were noted, including thesystem of mucuna integration, planting date, seedingrate, soil amendments used, weeding, date of harvest,and any problems encountered. The area planted wasmeasured. Where the mature crop of pods was har-vested (all Dendenyore farmers and a few in Zana)they were weighed (air dried). No systematic attemptwas made to estimate hay and intercrop stover yields.Subsequently, use of the crop was monitored.Farmers were also asked about their future plans forgrowing and using mucuna.

Additionally, in May–June 2001 information wassought about the perceptions of farmers who hadhosted trials for two seasons, and of those who hadstarted to experiment with mucuna on their own. Asimple questionnaire was developed, and 18 farmerswere interviewed.

Results

Table 2 shows the number of farmers given seed byresearchers, the total number of growers, and thenumber of those monitored.

A reasonable amount of rain fell in the 2000–01season (864 mm), although there was a mid-seasondrought. In 2001–02, most of the season was drought,but about 300 mm fell in February, bringing the totalto 733 mm.

Agronomic practice

Cropping systemIn Dendenyore, mucuna was grown as a monocul-

ture in both 2000–01 and 2001–02, except by onefarmer in the first year. In Zana, about half the

Table 2. Numbers of farmers receiving mucuna seed, growing the crop, and monitored.

Zana Dendenyore

2000–01 2001–02 2000–01 2001–02

Number of farmers given seedNumber of growersNumber monitored

151310

328

6

222113

0>61

11



158

farmers intercropped the mucuna in 2000–01, butchanged to monoculture the next year. ‘Too muchcare needed’ was a common observation about inter-cropping. In the second season, there was still nointerest in using mucuna for green manure, even bysecond-year growers who were no longer so con-cerned with seed bulking.

PlantingInformation about planting is presented in Table 3.

Planting in Dendenyore was done at the beginning ofthe rains, but in Zana planting was delayed, in somecases even into January.

The area planted to mucuna varied considerably.The minimum area planted by first-time testers was0.02–0.05 ha, while the maximum was 0.22 ha inDendenyore and 0.48 ha in Zana. In general, Zanafarmers allocated more land to mucuna than Dende-nyore farmers. Among those growers who usedmucuna in both 2000–01 and 2001–02, those inDendenyore tended to increase the area planted in thesecond year, but those in Zana reduced it.

The seeding rate varied considerably, often beingmuch lower than the recommended rate of about50 kg/ha, especially in Zana.

Fertiliser useInorganic fertiliser or manure was applied to about

20% of monocultured mucuna crops in Dendenyoreand 40% of such crops in Zana. The inorganic ferti-lisers used were Compound D (8:6:6:6.5% N:P:K:S),

ammonium nitrate (34.5% N) and single superphos-phate (8:12% P:S). In Dendenyore, application rateswere 43–342 kg/ha fertiliser and 6–18 t/ha manure,and in Zana 13–144 kg/ha fertiliser.

WeedingLittle information was obtained about weeding

practices, other than that in 2000–01 Dendenyorefarmers weeded once, on average at 26 days (range18–31 days) after planting. Experimental plots wereweeded twice.

Harvesting of monocultured mucunaIn 2000–01 all farmers harvested in May, either all

the herbage (all growers in Zana and one in Dende-nyore), or pods only. Some Zana farmers extractedsome pods thereafter. In the following season, withone exception, only pods were harvested and this wasdone even later: in June in Dendenyore and July inZana. Only one farmer in each year, in Zana, har-vested early enough (end of April) to make reason-able quality hay.

Mucuna yields

Pod and seed yieldsTable 4 shows pod yields, excluding data from one

waterlogged site. In 2000–01, the shelling percentagewas 53% for monocultures and the intercrop inDendenyore, where widely spaced maize was smoth-ered by mucuna. In Zana, pods were only harvestedfor two intercrops and the shelling percentage was

Table 3. Planting of mucuna in Zana and Dendenyore in two seasons: date of planting, area, seed planted andseeding rate.

2000–01 2001–02

Previous growers New growers

Zana Dendenyore Zana Dendenyore Zana Dendenyore

Planting dateaveragerange

10/127/12–18/1

30/1112/11–19/12

28/1212/12–10/1

15/1128/10–25/11

5/1(n = 1)

23/1110/11–14/12

Area planted (ha)averagerange

0.240.05–0.48

0.100.05–0.22

0.170.04–0.40

0.150.10–0.18

0.38(n = 1)

0.060.02–0.11

Seed planted (kg)averagerange

4.82.25–10.0

5.0all 5.0

5.03–9

7.55–10

15.0(n = 1)

2.91–5

Seeding rate (kg/ha)averagerange

306–67

6123–96

4313–75

4829–59

39(n = 1)

4839–60



159

36%. Shelling percentage was not determined for the2001–02 monocultures; 53% was used for both sites.

Hay yieldsWith the exception of a single case, it was not pos-

sible to estimate Zana mucuna hay or intercrop stoveryields because of prior grazing in the field and/orbecause hay was already in racks or had been partlyfed. The single estimate was 285 kg hay, equating to3.1 t/ha.

Use of the crop

All farmers used the mucuna, in various forms, asfeed for ruminants. In both seasons in Zana, abouthalf the farmers had their mucuna crop invaded andlargely demolished by uncontrolled livestock. Mostother farmers harvested the crop for ‘hay’, but onlyone farmer cut early enough (end of April) to makereasonable quality hay. The rest harvested after podmaturity and, in 2000–01, extracted some pods forseed retention, and in one case, all pods for feeding.More farmers proposed feeding pods or seed in2001–02, but this was mostly not possible because oflate planting, drought or livestock invasion. Thesingle new grower monitored in 2001–02 had moresuccess, allowed the pods to mature, and grazed theresidues in the field.

In Dendenyore all the farmers harvested the podsin the field. Most then allowed their livestock in,rather than collecting the residues. In 2000–01,feeding pods (crushed or milled) or seed (whole ormilled) were equally popular. The shelled pods,sometimes crushed, were also fed both to cattle andto goats. Where seed was fed, this averaged 101 kgper farmer. In 2001–02, use of the crop in Dende-nyore was similar to the previous year, although mostgrowers now proposed to shell the pods and feed seedand shelled pods separately.

It was observed that there were problems pro-tecting mucuna hay (heaped or in hay racks) frommarauding livestock. In an attempt to protect the hay,some farmers had resorted to storing it indoors (insheds, barns and even in their bedrooms). It was alsoobvious that mucuna hay is not easy to feed unless itis milled.

Farmers were originally given seed on the condi-tion that they pass on a similar amount to anotherfarmer. This was done by most of the Dendenyoregrowers, with some even passing on seed to morethan one farmer and in more than one season. InZana, however, none of the growers monitoredpassed on seed (because not enough seed was beingproduced or harvested), with the exception of onenew grower in 2001–02.

Table 4. Pod yields (kg) of monocultured and intercropped mucuna in Zana and Dendenyore in 2000–01 and 2001–02.

2000–01 2001–02

Zana Dendenyore Zana Dendenyore

MonoculturesPer farmer

averagerange

n.a.nil or very little; 191 (n = 1)

26083–575

n.a.nil;147 (n = 1)

808–146

Per hectareaveragerange

n.a.nil or very little; 2098 (n = 1)

2949814–6188

n.a.nil; 383 (n = 1)

863148–1448

IntercropsPer farmer

averagerange

2719–35 a

241 (n = 1)1215 (n = 1)

––

––

Per hectareaveragerange

450190–710

––

––

––

a Not all collected



160

Dendenyore farmers were more anxious to sellsome of their seed, whereas Zana farmers were lessconcerned about this. Fifteen of the 16 new Zanagrowers in 2001–02 bought seed from Dendenyorefarmers.

Farmers’ perceptions

Reasons for growing mucunaThe main goal of Dendenyore farmers was to

reduce the amount of inorganic fertiliser bought formaize crops, while farmers in Zana had specificallyplanted it for animal feed. All farmers, however,anticipated higher maize yields following mucunathan after weed fallow, from which they expectedlittle soil fertility benefit. Unfortunately, the poorrains in 2001–02 made it impossible for farmers toassess this.

Labour needsThe assessed labour requirement of mucuna technol-ogies differed between the two localities. Onaverage, 76% of the Zana farmers assessed the labourrequirement as high and the remainder as medium. InDendenyore, 20% classified mucuna production aslabour intensive while about 80% assessed it asrequiring low labour inputs. Table 5 shows farmers’assessments of labour inputs. Generally, planting,weeding and harvesting for hay coincided with manyother activities, especially in Zana. The few growersof maize–mucuna intercrops said that the practicerequired too much care. No farmers grew mucuna onits own as green manure.

Extent of future mucuna cultivationFarmers indicated that they would be able to set

aside about 0.4 ha for mucuna production. Followingboth seasons, most growers indicated that theywanted to increase the area planted to mucuna in thenext season. Some expansion was apparently largelyachieved in Dendenyore in 2001–02 (Table 3), withan average increase of 168% for those farmers mon-

itored in both seasons. However, in Zana the areaplanted in 2001–02 by farmers monitored in 2000–01declined by an average of 26%. In the two years ofgrowing their own mucuna, few farmers hadachieved an area of 0.4 ha (27% in Zana and none inDendenyore).

Discussion

Because of the relatively small number of farmersinvolved, any conclusions drawn from this moni-toring should be treated with caution. Nevertheless,valuable insights can emerge from a few carefullychosen case studies (Parton 1992), and some cleartrends emerged.

The average mucuna pod and seed yields achievedby Dendenyore farmers in 2000–01 were 2949 and1564 kg/ha, respectively. These are considerableyields for a dual-purpose crop growing with little soilamendment on infertile soils of low water-holdingcapacity.

In 2001–02, because of the bad drought, pod yieldswere only about 25% of those in the previous season.This is a similar yield reduction to that experienced ina CIMMYT seed-bulking plot near Harare in NR 2A(S.R. Waddington, pers. comm.). Timely planting bythe Dendenyore farmers in 2001–02 (late Octoberand November) ensured that good stands were estab-lished before the drought set in. It should be notedthat, despite the drought, mucuna did produce someyield and was not a complete failure like many othercrops. In contrast to the timely Dendenyore plant-ings, the 2001–02 planting in Zana was late (mid-December to early January), as in the previousseason, and establishment tended to be poor. Zanafarmers also had more problems with invading live-stock, and mostly very little seed, if any, was har-vested (in some cases not even enough forreplanting).

The low seeding rates used by some farmers, par-ticularly in Zana, were likely due to seed shortage,but also indicate that perhaps more advice is neededabout plant spacing, since low densities wouldincrease the weed burden. Farmers seemed preparedto weed their own mucuna crops only once, and thisdid not thoroughly eliminate weed competition.

Most farmers did not apply fertiliser or manure totheir mucuna, and the amounts used by those who didvaried widely. On a per hectare basis, small to rea-sonable amounts were applied. In the on-farm trials,mucuna had shown good responses (1999/2000) to

Table 5. Problems encountered with the cultivation ofmucuna (percentage of farmers asked).

Zana (%)

Dendenyore (%)

Planting is tediousWeeding is difficultWorkload increasedDifficult to work on

57768845

44342423



161

SSP and lime application (Jiri 2003). However, thesefindings had not been taken up by extension staff orfarmers. It was noted that extension staff recorded‘Fertilisers used: N/A’, possibly indicating a lack ofappreciation of this potential of SSP and lime toimprove mucuna growth. The few farmers whoapplied inorganic fertiliser tended to use CompoundD and ammonium nitrate: these contain N, whichshould not be necessary for well-nodulated crops.Non-affordability and poor availability of SSP andlime may also be influential factors.

Farmers seemed to prefer growing mucuna as amonoculture to intercropping. Intercropping is notcommonly seen as a soil fertility management prac-tice (Omiti and Freeman 1999; Bellon et al. 1998,cited by Ayuk and Jera 2001). Chibudu (1998) alsoreported this intercrop to be unpopular, because thetwining mucuna made maize harvesting difficult.The lack of interest in using mucuna as a greenmanure crop, despite the desire in Dendenyore toreplace fertiliser, was more probably caused by thenecessary trade-offs between use for green manure,use of hay or pods as livestock feed, and seed sales,rather than by the high labour requirement for incor-porating green manure (Muller-Samann and Kotschi1994). The main incentive to adopt legumes is theeconomic benefit from sales, rather than benefits insoil fertility (Muza and Mapfumo 1999).

In Dendenyore, farmers appear to have grown andutilised mucuna for livestock feed more successfully,preferring to let the crop grow to maturity and har-vesting pods in May–June. The mucuna was fed tolivestock either as whole or crushed pods, or milledor whole seed; shelled pods were also fed broken,whole or crushed, and the residue grazed in the field.By contrast, in Zana the intact crop was more oftenintentionally or accidentally grazed in the field orcollected for hay. However, only in two cases moni-tored did farmers cut the crop early enough to makehay of reasonable quality. It seems possible that com-petition for labour because of the harvesting of othercrops in April and May militates against earlier har-vesting of the mucuna, reflected in the perception ofZana farmers that mucuna crops have a high labourdemand.

Problems of access by uncontrolled livestock tomucuna as hay seem to be reduced by collecting andstoring pods. Milling and bagging hay would alsoprotect it from livestock and make feed rationingeasier, but would further increase the labour needed.Cases of increased labour demands inhibiting tech-

nology uptake by smallholder farmers in Zimbabwehave already been documented (Huchu and Sithole1994).

For the communal area farmers of Dendenyore,who hoped to sell some seed, mucuna may have arole as a direct source of cash income.

After the 2000–01 season, most growers indicatedthat they intended to increase the area planted tomucuna the next season. This seems to have beenachieved in Dendenyore, but not in Zana, where therewere few expansions and some reductions in areasplanted in the second season. This is probably relatedto seed availability. In Dendenyore, but not in thecash-cropping Zana system, farmers produced andretained adequate seed in both seasons for plantingthe next year. Most Zana growers seem to have pro-duced no seed at all for planting in 2002–03. Unlessfarmers in Zana are willing to buy seed from theirneighbours in Dendenyore, this is not a sustainablesystem. It is usually very easy to produce mucunaseed from a forage crop, so this failure to sustainmucuna seed supplies has negative implications forthe continued production and adoption of other leylegumes.

Baseline data was determined only for Dende-nyore and Zana farmers within the project interestgroups, who were therefore all actual or potentialcattle owners with some fallow land. In very similarcommunal and resettlement farming communities,also in NR 2B, Nyakanda (2003) found respectively36% and 63% of farmers to have weedy fallows,averaging 0.42 and 0.83 ha (11% and 14% of arableland). However, more communal area farmers thanindicated by the proportion with fallow land mighthave enough land to plant legume leys, since onestrategy to counteract risks (of drought, waterlog-ging, livestock damage, lack of labour for weeding,inadequate fertiliser etc.) is to cultivate as much landas possible (Carter and Murwira 1995; Nyakanda2003).

In a season of reasonable rainfall, monoculturedmucuna would produce, on average, yields of 2–3 t/ha pods with 1–1.5 t/ha seed, according to the 2000–01 data for Zana and Dendenyore. An area of just0.1 ha is therefore likely to yield about 200–300 kg ofpods with 100–150 kg seed, which at the rate of 2 kg/day of pods for cattle (Topps and Oliver 1993) couldprovide a high-protein feed input for 100–150 daysfor one animal. Hay yields of just 350 kg from 0.1 ha(3.7–11.8 t/ha in unfertilised trial plots; Jiri 2003)



162

would feed a cow hay at 6 kg/day for about 60 days(C. Murungweni, pers. comm.).

Conclusions

This study revealed the high potential of mucuna forintegration into smallholder farming systems in NR 2of Zimbabwe, particularly as livestock feed. Farmersproduced good mucuna crops and some successfullyproduced seed, which is essential to perpetuate thecrop and to ease adoption by other farmers. Thepotential for mucuna to improve soil fertility, andfarmers’ perceptions and adoption of it for this pur-pose, were less clear: farmers had no opportunity toobserve the impact of mucuna on following maizecrops because of the drought of 2001–02.

Production and use of mucuna differ between theDendenyore communal area and the Zana resettle-ment area, perhaps because of differences in theirfarming systems and their cash economies. The dif-ferent reasons of Dendenyore and Zana farmers forgrowing mucuna clearly reflected the areas’ differentresource bases. It was also quite apparent that reset-tlement farmers, involved in cash-cropping oftobacco and paprika, had serious labour constraints.Although they wanted to also undertake dairying andreplace bought dairy meal with homegrown feeds,Zana farmers’ attempts to incorporate mucuna intotheir farming system were less successful than thoseof Dendenyore communal area farmers. This hasimplications for the adoption of ley legume tech-nology in this sector. The ultimate way of using thecrop for livestock feed, as hay (Zana) or pods/seed(Dendenyore), was also influenced by labour availa-bility.

Dendenyore farmers also saw mucuna seed as asource of cash, in contrast with the cash-croppingZana farmers. Synergy might be created between twosuch communities by encouraging farmers to con-sider marketing mucuna seed between them as a feed.

The most efficient way to feed mucuna to livestockremains to be determined, especially regardinglabour demands, palatability and feed conversion (ofhay, seed, and shelled and unshelled pods; whole,crushed or milled).

Only the first two years of farmer testing ofmucuna were monitored. In Dendenyore, the numberof farmers trying out this new crop tripled in thesecond year, confirming the relevance of mucunaleys to the communal farming system and the likeli-hood of wider adoption by communal farmers.

References

Ayuk, E.T. and Jera, R. 2001. Review of literature onadoption and farmers’ perception studies on soil fertilitymanagement practices. In: Ayuk, E.T., Hove, L.,Kadzere, I., Matarirano, L., Agumya, A. and Mavankeni,B., Annual Report 2001 of the ICRAF–ZimbabweAgroforestry Project. The Zambezi Basin AgroforestryProgramme, 4–11.

Carter, S.E. and Murwira, H.K. 1995. Spatial variability insoil fertility management and crop response in MutokoCommunal Area, Zimbabwe. Ambio, 24(2), 77–84.

Chibudu, C. 1998. Green manuring crops in a maize basedcommunal area, Mangwende: experiences usingparticipatory approaches. In: Waddington, S.R.,Murwira, H.K., Kumwenda, J.D.T., Hikwa, D. andTagwira, F., ed., Soil fertility research for maize-basedfarming systems in Malawi and Zimbabwe. Proceedingsof the SoilFertNet Results and Planning Workshop, 7–11 July 1997, Mutare, Zimbabwe. Harare, Zimbabwe,SoilFertNet and CIMMYT, 87–90.

Chigariro, B. 2004. The farming systems in Hwedza Districtof Zimbabwe. These proceedings.

Chuma, E., Mombeshora, B., Murwira, H.K. and Chikuvire,J. 2000. The dynamics of soil fertility management incommunal areas of Zimbabwe. In: Hilhorst, T. andMuchena, F.M., ed., Nutrients on the move: soil fertilitydynamics in African farming systems. London,International Institute for Environment andDevelopment, 146 p.

Gore, C., Katerere, Y. and Moyo, S. 1992. The case forsustainable development in Zimbabwe: conceptualproblems, conflicts and contradictions. A report preparedfor the United Nations Conference on Environment andDevelopment. ENDA-Zimbabwe, 33.

Horne, P. and Stur, W.W. 1997. Current and futureopportunities for forages in smallholder farming systemsof south-east Asia. Tropical Grasslands, 31, 359–363.

Huchu, P. and Sithole, P.N. 1994. Rates of adoption of newtechnology and climatic risk in the communal areas ofZimbabwe. In: Craswell, E.T. and Simpson, J., ed., Soilfertility and climatic constraints in dryland agriculture.ACIAR Proceedings No. 54, 44–50.

Ivy, P. 1978. The dryland cash-crop production potential ofNatural Region II. Zambezia, VI(ii), 147–160.

Ivy, P. 1983. Resettlement areas. Starvation, subsistence, orsurplus. The Zimbabwe Science News, 17(9/10), 152–153.

Jiri, O. 2003. Integration of forage legumes into smallholderfarming systems, with emphasis on velvet bean [Mucunapruriens (L) DC var. utilis]. MPhil thesis, University ofZimbabwe.

Muller-Samann, K.M. and Kotschi, J. 1994. Sustaininggrowth. Soil fertility management in tropicalsmallholdings. Arnhem, The Netherlands, Technical



163

Centre for Agricultural & Rural Cooperation (CTA)/Eschborn, Germany, Deutsche Gesellschaft fürTechnische Zusammenarbeit (GTZ) GmbH, 254.

Muza, L. and Mapfumo, P. 1999. Constraints andopportunities for legumes in the fertility enhancement ofsandy soils in Zimbabwe. In: Maize productiontechnology for the future: challenges and opportunities.Proceedings of the Sixth Eastern and Southern AfricaRegional Maize Conference, 21–25 September 1998,Addis Ababa, Ethiopia. CIMMYT and EthiopianAgricultural Research Organization, 214–217.

Nyakanda, C. 2003. Effects of Sesbania sesban and Cajanuscajan improved fallows on soil moisture and nutrientdynamics and on maize performance in medium rainfallareas of Zimbabwe. DPhil thesis, University ofZimbabwe, 21–42.

Omiti, J.M. and Freeman, H.A. 1999. Micro-level strategiesto improve soil fertility in maize-based semi-arid farmingsystems. In: Maize production technology for the future:challenges and opportunities. Proceedings of the SixthEastern and Southern Africa Regional Maize Conference,

21–25 September 1998, Addis Ababa, Ethiopia.CIMMYT and Ethiopian Agricultural ResearchOrganization, 338–342.

Parton, K.A. 1992. Socio-economic modelling of decisionmaking with respect to choice of technology by resource-poor farmers. In: Probert, M.E., ed., A search forstrategies for sustainable dryland cropping in semi-arideastern Kenya. ACIAR Proceedings No. 41, 110–118.

Pengelly, B.C., Whitbread, A., Mazaiwanana, P.R. andMukumbe, N. 2003. Tropical forage research for thefuture — better use of research resources to deliveradoption and benefits to farmers. Tropical Grasslands, 37,207–216.

Thomas, D. and Sumberg, J.E. 1995. A review of theevaluation and use of tropical forage legumes in sub-saharan Africa. Agriculture, Ecosystems andEnvironment, 54, 151–163.

Topps, J.H. and Oliver, J. 1993. Animal foods of centralafrica, revised edition. Technical handbook No. 2,Zimbabwe Agricultural Journal, 67–68.



164

A Socioeconomic Evaluation of an ACIAR-funded Project on the Use of Forage Legumes in Integrated

Crop–Livestock Systems on Smallholder Farms in Zimbabwe

G.J. Manyawu,* O. Jiri,† B. Chigariro,§ P.R. Mazaiwana,§ R. Nyoka§ and D. Chikazunga‡

Abstract

A farmer participatory research project was initiated in 1998 in the Wedza District of Zimbabwe to improvesoil fertility and crop and livestock production on smallholder farms in Natural Region IIb (750–900 mmannual rainfall). Based on legume technologies integrated into crop–livestock farming systems, the project ranfor four years. By the end of the project there was widespread adoption of legumes in ley pastures, cereal–legume intercropping and livestock feeding systems. This rapid adoption resulted from the combination ofgood forage technologies, the better practices process and the participatory action research approach. Theproject had a large impact on the community, demonstrated by improved household incomes from the sale ofmilk, beef cattle and seed, and improved crop yields. Forage legumes fitted well into smallholder farmers’integrated crop–livestock systems, and Lablab purpureus and Mucuna pruriens were the most widely usedlegumes. Adoption was hampered by shortage of seed of the recommended varieties. We concluded thatfarmers should produce legume seed at a commercial scale to satisfy local and external markets. Farmers saidthat, for an even bigger impact, the project should help them to obtain crop inputs and better breeds of cattle.Farmers’ comments at the end of the four years showed that the project still had a lot to accomplish. The readyadoption of the project results demonstrated that scientists should develop and run research programs inpartnership or consultation with the end-users of the technology.

The purpose of this research project, initiated in1998, was to improve soil fertility and crop and live-stock production on smallholder farms through well-

adapted and well-chosen legumes. Normally, cropand livestock production in these farming systems isintegrated. Crops provide feed for livestock and live-stock provide draught power and manure for thecropping system (Francis and Sibanda 2000). Thisaspect of the farming system was incorporated intothe project.

The project was conducted in Wedza district andlasted four years. It was developed in associationwith smallholder farmers of Zana II and Dendenyorewards using farmer participatory research methodol-ogies, sponsored by the Australian Centre for Inter-

* Grasslands Research Station, Private Bag 3701,Marondera, Zimbabwe.

† AREX Head Office, Agronomy Institute, PO Box 594, Causeway, Harare, Zimbabwe.

§ AREX, Wedza Office, PO Box 30, Wedza, Zimbabwe.‡ University of Zimbabwe, Agricultural Economics

Department, PO Box MP 167, Mount Pleasant, Harare, Zimbabwe.



165

national Agricultural Research (ACIAR), andimplemented by a multidisciplinary team ofresearchers from Zimbabwe, in conjunction withsome Australian scientists. Farmer-managed experi-ments were conducted to evaluate a range of tropicalforage legume accessions in veld (grassland) rein-forcement, ley pasture production and cereal–legumeintercropping. The project employed the principles ofthe better practices process (McCartney et al. 1998;Clark and Timms 2000) to ensure a positive impactand continuous improvement of the research pro-gram.

In Zana II, the farmers agreed to test 30 species oflegume in veld reinforcement as a communal effort.Mucuna (Mucuna pruriens) was also used in someon-farm trials to demonstrate the benefits, for soil fer-tility and fodder production, of using legumes incereal rotations or intercropping systems. Dairyfarmers from the same ward were also involved inexperiments to test the use of lablab (Lablab pur-pureus) and mucuna seed or foliage as protein sup-plements for lactating cows during the dry season.Similar experiments were conducted in DendenyoreWard. However, farmers in Dendenyore were mostlyinterested in beef cattle farming, particularly pen-fat-tening, and not dairying.

Our report describes the adoption of the technolo-gies and their impact on soil fertility, on fodder, cropand livestock production, and on farmer’s liveli-hoods.


Since 1999, diagnostic surveys to benchmarkprogress were conducted annually between Februaryand March by research and extension staff from theproject. In these benchmarking surveys, research andextension staff spoke with farmers to determinefarmers’ aspirations and achievements. Data fromthese exercises was largely descriptive.

In addition, a socioeconomic survey was con-ducted in February 2002 using a semistructured ques-tionnaire to assess the levels of adoption of the newlegume-based technologies and their impact onhousehold income, capital developments and liveli-hoods. The questionnaire gathered information onhousehold demography, resource endowments andfield crop, fodder and animal production before andafter the introduction of the project. It was given to 46farmers in Dendenyore and 18 in Zana II. Thesefarmers were chosen on the basis of their level of

involvement in the project. In Dendenyore, 24 of thefarmers were founding members of the project and 22were farmers (hereafter referred to as adopters) whowere recruited into the project when they adoptedtechnologies introduced by the project. Similarly, inZana II there were 15 founding members and threeadopters. Five trained enumerators administered thequestionnaire. Data from the questionnaires wascoded and entered into the SPSS analytical program(version 7.5) for analysis.


Brief description of the farming systems in Zana II and Dendenyore wards

The total numbers of households in DendenyoreWard and Zana II were 3000 and 75, respectively. Atthe inception of the project, there were 49 partici-pating households at Dendenyore and 19 at Zana II.The average age of household heads in the two wardswas 58, and the average educational level was tograde 7 in primary school. The average size of thehousehold was bigger in Zana (9.7) compared toDendenyore (6.2). All households in Zana wereheaded by men. In Dendenyore, 38% were headed bywomen. Although most households were headed bymen, women carried out most of the farming opera-tions in both wards.

Farmers in the two wards take farming seriously.Most are members of agricultural associations, suchas the Zimbabwe Farmers Union, the LivestockFeeders Association, the Dairy Association and othercommodity associations for tobacco, mushroom,paprika, maize and soybeans. These associations area vital source of information on agricultural produc-tion and marketing.

When farmers at both sites were asked to rank themost important sources of income, they rated cropproduction, livestock production and off-farmincome as equally important.

Total land holdings of the average farmer inZana II were almost double those in Dendenyore(Table 1). The difference exists because Zana II is arelatively new resettlement area where farmingstarted in 1986. Dendenyore is an old communal set-tlement where most of the land has been continuallysubdivided through family inheritance since the1950s. Up to 75% of the land in both wards wasdevoted to crop production. While farmers in Zana IIhad less land under fallow, there were indications in



166

the past two years that some farmers were beginningto devote more land to fodder production.

Crop production Farmers from Zana II concentrated on crop produc-

tion, with tobacco the most important crop followed indescending order by paprika (Capsicum annuum),soybean (Glycine max), sugar beans (Phaseolus spp.)and maize. Farmers derived this rank order byassessing incomes obtained from these crops and theircontribution to household food security.

Farmers at Dendenyore also carried out intensivecrop production, supplemented with beef productionand horticulture. The most important crops were, indescending order, maize, sugar beans, groundnuts

(Arachis hypogea), Bambara nuts (Voandezia subter-reanea) and sweet potatoes (Table 2). Up to 56% ofthe farmers in Zana II and 79% in Dendenyore inter-cropped sugar beans with maize. Other legumes thatwere mixed with maize were mucuna and lablab.

All farmers indicated that they practised crop rota-tion and that they grew at least one legume, such asBambara nuts, soybean, forage legumes, sugar beansand cowpeas (Vigna unguiculata), to improve soilfertility through biological nitrogen fixation and toreduce costs of N fertilisers.

Livestock productionFarmers in both wards rear beef and dairy cattle,

with a few sheep and goats (Table 3). Cattle were byfar the most widely kept livestock. About 87% of thefarmers in Zana II were involved in commercial dair-ying. Most of the milk was sold at the Wedza DairyCenter and to neighbouring farmers. The milk wasproduced from exotic (Jersey) and indigenous(Mashona) beef crosses. In Dendenyore, farmersonly milked their multipurpose indigenous beefcattle for subsistence; otherwise, the cattle were keptto provide draught power, manure, meat and income.

In each ward, a group of farmers attended initialmeetings to outline problems that needed to be inves-tigated in crop–livestock systems of production. Thissame group of farmers was asked to host and manageon-farm research trials from 1999. As these farmersgained experience, they helped in technology diffu-sion in the wards and in distributing pasture seed toneighbours. Fields belonging to host farmers wereaccessible to all farmers in the program; there were norestrictions. Host farmers also explained their role inthe continuous improvement of the project and howthis led to a sharper focus in resolving productionproblems. Neighbouring farmers who copied technol-ogies from host farmers were termed ‘adopters’.

The farmers hosted the experiments as individualsor in groups. At Zana II, rangeland research washosted by a group of farmers, whereas research on

Table 1. The mean farm size (ha) and the pattern ofland use on properties owned by farmersinvolved in the ACIAR project.

Dendenyore Zana II

Total land size Total land under cropTotal land fallowTotal land under pasture before 1999Total land for pastures after 1999

2.71.50.6

Negligible

0.4

5.03.00.9

0.3

0.4

Table 2. Farmer rankings of the most important cropsgrown in Dendenyore and Zana II Wards.

Crops Dendenyore Zana II

MaizeBeansGroundnutsPaprikaTobaccoSoybeans Bambara nutsSweet potatoes

46 (1)27 (3)44 (2)

–––

26 (4)13 (5)

18 (1)3 (5)

–17 (2)15 (3)6 (4)

––

Table 3. Cattle ownership by average households in Dendenyore and Zana II Wards.

Ward Cattle type Year

1999mean (s.d.)

2000mean (s.d.)

2001mean (s.d.)

Dendenyore BeefDairy

7.7 (6.0)2.5 (1.6)

7.4 (5.3)2.4 (1.6)

7.3 (4.4)2.3 (1.3)

Zana II BeefDairy

13.1 (11.9)1.5 (1.3)

11.8 (11.7)2.6 (1.7)

12.4 (10.9)4.6 (1.6)



167

intercropping and screening of legumes for leypasture production was hosted by individual farmers.At Dendenyore, all the research was hosted at indi-vidual farms, but in some instances (for example, inevaluating legumes suitable for leys) the farmerscame to work at experimental sites in groups.

Adoption of crop production technologies

The main reasons why farmers were adopting theuse of legumes in their cropping systems were toimprove soil fertility and the yield of subsequent maizecrop, and to provide fodder. Results of the surveyshowed that mucuna and lablab were the most widelygrown legumes in Zana II and Dendenyore, and thatthey were increasingly being adopted by farmersinside and outside the project area. The farmers wereimpressed with the drought tolerance of lablab andmucuna. The number of farmers who grew lablabincreased by 20% in Zana II and 68% in Dendenyore,and the numbers who grew mucuna increased bysimilar proportions. The area sown to these legumesalso increased progressively (Table 4) as farmersgained knowledge about how to grow them.

Most farmers adopted the use of these legumesboth in homestead gardens and in the main fields.Many of the adopters were getting seed of the twolegumes from host farmers at prices that ranged fromZim$5 to $45/kg (US$0.09–0.82). Initially, seed wasprovided by the project free of charge, and the projectcontinued to provide free seed to at least five farmerseach year. It was project policy that each year ahosting farmer or adopter who received free seed in

the previous season should pass on at least 5 kg ofseed to another new farmer. At the time of the survey,no arrangements had been made to market the seed tocommercial seed companies. However, there wasoverwhelming demand for seed of the foragelegumes and some farmers showed interest in pro-ducing seed at a commercial scale.

Adoption of livestock production technologies

Although the number of beef cattle reared in thetwo wards did not change (Table 3), the feeding prac-tices of the farmers did. Farmers increasingly usedforage legumes to replace commercial concentrates.In 2001, on average, 5.9 beef and 1.8 dairy cattle perfarm at Dendenyore and 8.8 beef and 1.2 dairy cattleper farm at Zana II were fed lablab or mucuna as hay,fresh material or ground seed to substitute for com-mercial concentrates such as ‘Super 10’.

About 31% of the farmers at Zana II and 16% atDendenyore used commercial supplements. Themajority, 67% and 82% respectively, relied on maizestover and 2% used groundnut tops, pods from indig-enous trees and maize meal as supplementary feedduring the dry season. Some 71% of the Dendenyorefarmers and 45% of Zana II farmers said that theamount of forage legumes grown was sufficient fortheir cattle herds. The other Zana II farmers indicatedthat they did not produce sufficient forage becausethey devoted less land to pastures. They appeared toget better returns from cash crops such as tobacco andpaprika.

Table 4. Production statistics for lablab and mucuna at Wedza.

Parameter measured Dendenyore Ward Zana II Ward

Mean s.d. Mean s.d.

Lablabarea (ha) sown in 2001area (ha) sown in 2000area (ha) sown in 1999Forage yield in 2001(kg DM/ha)Forage yield in 2000(kg DM/ha)Forage yield in 1999(kg DM/ha)

1.130.961.06

976.921026.82929.33

0.440.350.37

580.451052.51633.43

0.850.700.44

766.671058.33400.40

0.580.520.33

560.571330.21270.83

Mucunaarea (ha) sown in 2001area (ha) sown in 2000area (ha) sown in 1999Forage yield in 2001(kg DM/ha)Forage yield in 2000(kg DM/ha)

0.980.801.06

1100.361937.50

0.360.430.26

1746.282431.81

0.690.650.49

644.62345.00

0.500.380.36

736.44353.20



168

Impact of the project

Crop practicesThe project had a huge impact on cropping prac-

tices. Farmers reported that, on average, maize grownafter a legume yielded 4 t grain/ha, whereas maizeunder continuous cultivation yielded 2 t/ha.Although previous research by Manyawu et al.(1995) and Alemseged et al. (1991) indicated thatintercropping maize with legumes depressed maizegrain yields by 18–20%, evidence collected in thecurrent survey demonstrated that in Wedza the yieldreduction was only 3%. Farmers were thereforechanging their cropping practices to include legumesin their cereal rotations. Some farmers even favouredthe idea of intercropping maize with lablab because itwas less aggressive than mucuna.

Livestock productionThe majority of farmers indicated that the project

had a direct impact on their incomes and livelihoods.In Zana II, improved milk yields increased farmers’incomes. The higher yields resulted from betterquality feeds introduced by the project. Table 3shows that the number of dairy animals increased inZana II, and Table 5 shows the increase in milkyields.

Farmers in Zana II were selling a litre of milk atZim$60 (US$1.09) at the Wedza Dairy Center. Anaverage farmer in Zana II who had two dairy cowsthat produced 5 L/day was selling up to 300 L of milkper month, realising Zim$18 000 (US$327). Thefarmers also indicated that improved feeds intro-duced by the project had increased milk fat content.Considering that per capita annual income in Zim-babwe is about Zim$1340, the current income frommilk may be sufficient to sustain an ordinary dairyfarmer in Wedza.

Returns were higher (1:1.9) when farmers usedforage-based diets, as opposed to concentrate-based

diets (1:0.7). Up to 78% of the farmers in Zana IIindicated that income from milk sales was the maincash benefit they were enjoying from the project. Theadditional income helped farmers to meet their dailyhousehold expenses and acquire some farm inputs.

In Dendenyore, beef farmers were increasing theirprofits by using home-grown leguminous fodder tosubstitute for commercial concentrates in cattle fat-tening diets. Because of the low cost of the home-mixed diets, farmers were able to feed their livestockmore effectively and cheaply. Gross margin analysisby Murungweni et al. (2004) showed that the farmersmade a profit of Zim$931.80 (US$16.90) on a steerwhen they used forage-based diets. Farmers whoused commercial concentrates made a loss ofZim$2619.00 (US$47.62) on every steer.

Generally, farmers felt that feeding legumesimproved the health status of their livestock, andthese sentiments were echoed by neighbours whoadmired the cattle. Good nutrition also improvedconception and, consequently, total milk yields.Draught cattle remained in good condition up to theend of the dry season.

Indirect benefitsFarmers indicated that they were getting indirect

savings on fertiliser by growing legumes, becausesubsequent crops required less N fertiliser. Farmersalso noticed that growing mucuna in rotation withmaize was cost-effective in eradicating Striga(witchweed). When legumes were used in cattle-fat-tening diets, the amount of maize grain in the supple-mentary diet was reduced by 30%. A few farmerssaid they roasted and ground mucuna seed to make abeverage similar to coffee, thereby saving money oncoffee.

Constraints

Limitations of legume technologiesDespite the benefits outlined above, some dairy

farmers in Zana II reported that milk from dairy cowsthat were fed mucuna had a tainted taste and turnedsour sooner than usual. Dairy cows sometimes tooktime to get used to the legume, and some animalseven suffered from mild diarrhoea at the onset of thefeeding period. The farmers also added that lablabwas unsuitable in some instances because it suc-cumbed to aphids during droughts. It was prone todamage and death of seedlings at establishmentbecause of cutworms and also to a variety of insectpests at flowering. This made it less suitable than

Table 5. Average daily milk production (L/cow/day)at Dendenyore and Zana II.

Year Dendenyore Zana II

Mean yield

s.d. Mean yield

s.d.

2001200019991998

2.942.942.911.50

2.42.12.40.9

5.104.606.562.06

2.22.12.01.7



169

mucuna because farmers had to buy expensive pesti-cides and other chemicals to control the infestations.Hence, researchers need to find better suited legumesand develop appropriate methods to process herbagefrom mucuna before it is fed to livestock.

Most farmers complained that mucuna smothersmaize when it is intercropped, causing severe yieldreductions. Researchers need to identify less aggres-sive varieties of this legume and/or devise appro-priate intercrop planting methods that will favour thegrain crop.

The shortage of seeds of lablab and mucuna wasalso said to be a major limitation to the adoption oflegume-based technologies on a large scale.

Shortage of resourcesFarmers indicated that they were having problems

in getting enough money to buy fertilisers that couldbe applied to planted legumes. They mentioned thattheir first priority is to find enough fertilisers for cashcrops. Many said that the project should set up acredit scheme to provide loans for fertilisers and limeto be used in pasture development. Normally, thefarmers do not have enough money to buy these fer-tilisers.

At both Dendenyore and Zana II, more than 87%of the farmers suggested that the project should starta livestock finance scheme to provide loans tofarmers who wish to buy better breeds of dairy orbeef cattle. Many said that local (indigenous) cattledo fully capture the advantages of using cultivatedpastures because of their low milk production poten-tial, small body weights and slow growth rates.Farmers wanted better breeds of cattle, especiallyexotics, which they would cross with local breeds.

A Dutch-sponsored project (BTZ) operating inWedza District bought motorised chopper–grindersfor farmers under its program. Farmers on theACIAR project wanted similar machines to chopforages for ensilage and to mill grain or legume seedfor use in mixed rations. Local business people do notallow the farmers to process legume seed or foragesin their hammer mills.

Level of participation by farmers in the projectFarmers expressed the opinion that, in some

instances during the current phase of the project,researchers designed experiments and farmers onlyparticipated in evaluating the promising technologies(consultative participation). The project staffresponded that there was a need to balance the twoapproaches, because it is not possible to have all

experiments in the full collaborative participationmode. This issue requires further discussion.

Conclusion

The impact of the project was demonstrated byimproved household incomes from the sale of milk,beef cattle and seed, and by improved crop yields.Widespread adoption of legume-based technologieson this project confirmed that participatory actionresearch and the better practices process are powerfultools that should be used in developing appropriatetechnologies in smallholder farming communities. Inprojects of this nature, scientists will have to placeincreased importance on developing and runningresearch programs in partnership with the end-usersof the technology.

Forage legumes fit well into the integrated crop–livestock systems that prevail on smallholder farmsin Natural Region II of Zimbabwe. Farmers mustattempt to produce seed of the legumes at a commer-cial scale in order to satisfy local demands andexternal markets. When seed of the appropriatelegumes is readily available at an acceptable price,the rate of adoption of the technologies is likely to beeven greater. Farmers’ comments at the end of thesurvey indicated that the project still has a lot toaccomplish before it ends. For an even bigger impact,the project should help the farmers to obtain cropinputs and better breeds of livestock.

References

Alemseged, Y.B., King, G.W., Coppock, V.L. and Tothill,J.C. 1991. Maize-legume intercropping in a semi-aridarea of Sidamo region, Ethiopia. 1-Maize Response. EastAfrican Agricultural and Forestry Journal, 56(4), 77–84.

Clark, R.A. and Timms, J. 2000. Enabling and achievingcontinuous improvement and innovation. The betterpractices process-focussed action for impact onperformance. Lawes, Queensland, The Rural ExtensionCentre, Gatton College, University of Queensland.

Francis, J. and Sibanda, S. 2000. Economic performance ofsmallholder enterprises in Nharira-Lancashire inZimbabwe. In: Sibanda, S. and Kusina, N.T., ed.,Integrated crop–livestock production systems in thesmallholder farming systems in Zimbabwe. Proceedingsof a review workshop held in Harare, Zimbabwe, 10–13 January 2000.

McCartney, A., Rea, E., Clark, R. and Robinson, E. 1998.Best practices-extension that makes a difference topractices and performance. In: Clark, R. and Timms, J.,



170

ed., Enabling continuous improvement and innovation —the better practices process. Selected readings. Brisbane,Department of Primary Industries, 39–46.

Manyawu, G.J., Muchadeyi, R. and Chakoma, I. 1995.Agronomic evaluation of Lablab purpureus accessionsintercropped with maize (Zea mays) for grain and forageproduction. In: Annual Report 1995, Department of

Research and Specialist Services, Division of Livestockand Pasture. Harare, Zimbabwe, DR&SS, 101–104.

Murungweni, C., Mabuku, O. and Manyawu, G.J. 2004.Mucuna, lablab and paprika calyx as substitutes forcommercial protein sources used in dairy and pen-fattening diets by smallholder farmers of Zimbabwe.These proceedings.



171

Using the Agricultural Simulation Model APSIM with Smallholder Farmers in Zimbabwe to Improve

Farming Practices

A.M. Whitbread,* A. Braun,† J. Alumira§ and J. Rusike§

Abstract

Computer simulation models can serve as a platform for achieving change in a farming system. Such modelsrepresent crop and animal production processes to assist farmers, researchers and agricultural managers. Thispaper presents examples of using a simulation model, APSIM, with farmers to analyse farm decision making,to assess the impact of climatic risk on productivity and profitability, and to set priorities for participatoryresearch. At the Linking Logics II workshop held at ICRISAT, Bulawayo, Zimbabwe in October 2001, a teamconsisting of a modeller/soil scientist, a sociologist, an economist, a participatory methods expert and localNARES (national agricultural research and extension systems) extension officers spent two days with morethan 20 local farmers at Zimuto Siding near Masvingo in Zimbabwe. The scenarios and simulations developedduring these meetings are presented. Important fertiliser, manure and crop management choices were exploredfor poorer and wealthier groups of farmers, and as a result, the farmers were able to consider a broader rangeof management decisions for the next season.

Resource-poor farmers are the focus of manyresearch and development projects in the developingworld. Poor adoption of technological innovationsfrom many agricultural research projects has resultedin greater emphasis by funding agencies on farmeradoption and adaptation of technologies. This has ledmany researchers to operate in a participatoryresearch context, recognising that the inputs andoutputs of the biophysical ‘production system’ ofcrops, pastures, animals, soil and climate operatestogether with the ‘management system’, which ismade up of people, values, goals, knowledge,resources, monitoring opportunities and decision

making (Keating and McCown 2001). Both the bio-physical and the human systems are characterised bycomplexity, diversity and dynamism (McDougalland Braun 2003).

In the semi-arid regions of southern Africa, small-holder farmers face serious social challenges tomaintaining food security, exacerbated by low soilfertility and highly variable rainfall. In these environ-ments, water and nutrient use efficiency are low(Mapfumo and Giller 2001), and technical optionsfor improving soil fertility and production are limitedby a range of factors, which often include the poorresource endowment of farmers and their aversion torisk. The approaches used by farmers to manage bio-physical risk include utilising the spatial variabilityof their landscape and on-farm resources (Giller et al.2004), adapting the timing of operations such asplanting and fertilising (Harrington and Grace 1998),diversification, keeping reserves, and a range of other

* CSIRO Sustainable Ecosystems, 306 Carmody Road, StLucia, Queensland 4067 Australia.

† Paideia Resources PO Box 462, Nelson, New Zealand.§ International Crop Research Institute for the Semiarid

Tropics, PO Box 776, Bulawayo, Zimbabwe.



172

marketing and financial strategies outlined by McCo-nnell and Dillon (1997) and Anderson and Dillon(1992). Simulation models such as the AgriculturalProduction Systems Simulator (APSIM) are provinguseful for capturing the interactions between climaticconditions, soil types and nutrient dynamics incereal-based farming systems in Africa and Aus-tralia. In Africa, the development of the APSIMmodel has been pursued in collaboration with theInternational Maize and Wheat Improvement Center(CIMMYT), the International Crops Research Insti-tute for the Semi-Arid Tropics (ICRISAT), and theirNARES partners since 1985; a number of publica-tions outline this work (Keating et al. 2000, 2003;Shamudzarira et al. 1999; Shamudzarira and Rob-ertson 2002; Carberry et al. 2002b).

Keating and McCown (2001) used APSIM as anexample of how systems simulation models could beused to integrate ‘hard’ biophysical systems and‘soft’ systems approaches to intervention in socialmanagement systems. In this context, APSIM canenable scientists, extension officers and farmers toexplore the consequences in the farming system (e.g.grain yield, gross margin, soil fertility) of manage-ment actions and strategies and of seasonal and long-term climatic conditions. This can lead to a series of‘what if’ scenario analysis sessions and informationon trade-offs.

Keating and McCown (2001) summarised the evo-lution of the use of computer-aided farming systemsanalysis and intervention over the past 40 years.Many modelling efforts have continued because oftheir ‘potential’ benefits in aiding agricultural devel-opments, but models have not been widely used tohelp make land management decisions. Theseauthors criticised many modelling efforts (decisionsupport systems, expert systems) for their lack of asystems perspective, and especially for not engagingthe end user. Doyle (1990) found that ‘the failure ofsystems researchers to liaise with farm decisionmakers has meant that [farmers who have beenexposed to modelling] are rightly suspicious of com-puter generated predictions of optimal resource use’.

That the use of simulation can be relevant and sig-nificant to farmers in certain situations is found in theAustralian FARMSCAPE experience reported byCarberry et al. (2002a). FARMSCAPE (Farmers,Advisors, Researchers, Monitoring, Simulation,Communication and Performance Evaluation) is aprogram of participatory research with ‘elite’ Aus-

tralian farmers and their advisors, who work withresearchers in their own farming operations.

Can this use of simulation with elite farmers in thedeveloped Australian farming environment be rele-vant to smallholder farmers in southern Africa?While modelling has been extensively applied inAfrica to interpret on-farm experiments, assess theclimatic risk of technologies, analyse trade-offs andfind best-bet options, Dimes (2002) and Carberry etal. (2004) describe a process of engaging smallholderfarmers directly with simulation models in semi-aridZimbabwe. They conclude that, in an action researchframework, intervention strategies can successfullyincorporate aspects of simulation.

The objective of this paper is to present our expe-rience with a small group of smallholder farmers inZimbabwe, using the APSIM modelling system and amultidisciplinary group of scientists, as was part of alarger workshop approach called Linking Logics II.The Linking Logics workshop was a joint venture ofthe Consultative Group on International AgriculturalResearch (CGIAR) system-wide programs on Partic-ipatory Research and Gender Analysis and on Soil,Water and Nutrient Management; the CGIARresearch centres; ICRISAT; and CIMMYT. The jointventure workshop explored complementaritiesbetween farmer participatory research approachesand computer-based simulation modelling in soil fer-tility management at the smallholder level. Carberryet al. (2004) describe their experiences at Tsholotsho,Zimbabwe, and this paper outlines a similar exercisewith a farming community at Zimuto in Zimbabwe.

Background

During the Linking Logics II workshop, a team con-taining a modeller/soil scientist, a sociologist, aneconomist, a participatory methods expert and localNARES extension officers spent two days withsmallholder farmers from the village of ZimutoSiding. Most of the 20 farmers who attended themeetings were participants in a farmer research andlearning group, which had just been established in thevillage by ICRISAT with the objective of testing soilmanagement technologies and improving farm man-agement decisions and performance. The objective ofthe exercise was to explore how APSIM could beused as an educational and decision-making tool.Previous work by ICRISAT in the region gave confi-dence in the ability to simulate the actual yields ofmaize achieved by farmers under various manage-



173

ment systems (Keating et al. 2000; Shamudzarira etal. 1999). The Zimuto Siding farmers had no priorexposure to simulation modelling, but some of themhad been involved in previous on-farm research anddevelopment work with ICRISAT and the Depart-ment of Agriculture.

The climate at Zimuto Siding (19∞51'S, 30∞52'E) issubtropical with a unimodal rainfall pattern fromOctober/November to March, when most of the rainfalls as sporadic heavy convectional storms, and along dry spell from April to September. The meanannual rainfall for the Zimuto area (derived from1961–98 rainfall records at nearby MakoholiResearch Station) is 631 mm, with a range of200 mm to 1200 mm.

The soils are largely derived from granite, withfour major categories recognised on the basis of soilmoisture-holding capacity, drainage, waterlogging,weed burden, and soil fertility on a typical catena. Allthese parameters increase downslope from drytopland granitic sands (Lithic Ustirhents/Lithic orUmbric Leptosols) of the upland ridges and valleyslopes (Typic Kandiustalfs/Haplic Lixisols), to thehydromorphic Vlei-margin and valley bottom Vleisoils (Typic to Aquic Ustipsamment/Luic Aarenosolsand Plinthic Lixsols) adjacent to watercourses. Sizeand texture of soil particles range from fine/medium-grained sands over medium to coarse-grained sandyloams at the Topland sites, to medium to coarse-grained sandy loams for the vleis and vlei margins.The vleis and vlei margin are significantly moreacidic than the topland soils, and this increase inacidity appears to be associated with significantincreases in organic carbon in the top 15 cm of aprofile (J. Dimes, pers. comm.)

The APSIM model, described most recently inKeating et al. (2003), was used to run the scenariosdeveloped during discussions with farmers. Dailyweather records (1961–98) from Makoholi ResearchStation were used as the climate files for the APSIMruns. In all simulations, the soil described in Table 1was tilled on 20 October and again at sowing. Farmer

tillage practice varies with the availability of draughtpower, and tillage usually takes place in October orNovember. The units of measurement used werethose chosen by the farmers (bags/acre, where onebag is equivalent to 50 kg of hulled grain maize).

APSIM initialisation

The soil type in Table 1 was used for all simula-tions, and represents an infertile, sandy profiletypical of both the field sites discussed above. Thesoil type is a shallow sandy soil (1 m deep) with plantavailable water capacity (PAWC) of 59 mm. The pHis 6.0, increasing to 6.5 at 0.75 m depth. Specificdetails of the lower limit (LL15), drained upper limit(DUL) and organic carbon for the soil profile are inTable 1. The crop lower limit (LL15) measures theamount of water left in the soil at a suction of 15 barand represents the lowest limit at which plant rootscan remove soil water. The DUL is the amount ofwater the soil holds after drainage has ceased. Thedifference between the DUL and the LL15 is the the-oretical plant available water held by the soil. Furtherdetails of the characterisation methods can be foundin Dalgliesh and Foale (1988).

In order to trigger the sowing event (dates ofsowing specific to each simulation), rainfall of atleast 10 mm over 10 days and a stored PAWC of50 mm or more was required. A grass weed was alsosown at the same time as the maize to mimic theeffect of weeds on the maize crop. The weeds wereremoved by tillage at 20 days after sowing. Densityof maize was 5 plants/m2, and weeds were sown at4 plants/m2.

Results of the Zimuto Siding meetings

Five men and 16 women attended the two-dayworkshop, held in a school building at ZimutoSiding. All the men and nine of the 16 women wereheads of households. The farmers collectively identi-fied and ranked the main agricultural problems theyfaced (Table 2).

Table 1. Lower limit (LL15) and drained upper limit (DUL) expressed as volumetric water percentage, andorganic carbon (O.C.) of the soil used for the Zimuto APSIM scenarios.

Depth (cm) 0–15 15–30 30–45 45–60 60–75 75–100

LL15 DULO.C. (%)

0.040.141.0

0.070.150.9

0.130.200.7

0.130.220.5

0.180.220.5

0.220.240.4



174

All farmers at the meeting used a maize-basedfarming system. Food produced was usually enoughto feed the family, but rarely was there a surplus tosell. Only four of the farmers had no cattle, which areimportant for manure, draught power, milk and meat,and for paying for funerals, weddings etc. Some ofthose without cattle collected leaf litter from forestedareas, or ant-hill soil, to fertilise their fields, butlabour constraints (often due to HIV/AIDS-relatedillness) made this practice less common. All farmerssaid that labour was limiting.

The richer farmers sometimes had surplus maize tosell or exchange for labour, and those with draughtcattle were able to exchange draught power forlabour. The poorer farmers without their own draughtpower were forced to use the draught animals of other

farmers, and late planting was often a consequence.Lack of implements was also noted as a seriousproblem. Overall, the farmers felt that if a householdpossessed cattle and implements the other problemswere much less important.

Wealth ranking

Through discussion, it quickly became clear thatthere were important differences in the relativewealth of the farmers present. Through a series ofparticipatory wealth ranking exercises, the farmerscategorised households in the area into three general-ised resources/wealth groups (Table 3). At no timewere farmers asked to say which wealth group theybelonged to. Indicators of wealth identified byfarmers included livestock ownership (numbers andtypes), amount of land cultivated, farm implements,draught power, and labour availability.

Poor households. Resource-poor farmers struggleto remain self-sufficient in maize in all but very goodseasons with high rainfall. These farmers are forced tosupplement their income by selling their labour(herding cattle, planting, weeding and harvesting) andother income-earning enterprises (trapping termites,making clay pots and wares). The average householdsize was 5–11 people (with 3–9 children). The childrenusually attended school, but paying fees was cited as

Table 2. Assessment of the main agriculturalproblems facing the farmers at Zimuto.

Problem Rank (out of 10)

Lack of draught powerPoor soil fertility — sandy soilsExpensive inputs (seeds and fertiliser)Very little available manure for use as a fertiliser

621

1

Table 3. Resources available to poor, average and rich farmers at Zimuto Siding, Zimbabwe.

Poor Average Rich

Farmed land (contoursa) Total Farmed/season

1–20.5

5–101

>103–4

Draft power _ Shared Owned

Implements WheelbarrowsScotch cartsPloughs/harrows

___

__

Shared

___

Animals CattleDonkeysGoatsSheepChickens

00000

2–40

1–50–1

7–10

5–404–6

6–104–10

15–50

Labour FamilyHired

__

__

__

Agricultural inputs SeedFertiliser

_b

_b__

__

Timeliness CultivationPlanting

LateLate

Sometimes lateSometimes late

On timeOn time

a One contour = ~ 1 acre.b Small quantities can be purchased at a local shop.



175

an important problem. Most of these families wereheaded by females (usually a widowed mother orgrandmother). If there was a husband at home, he wassaid to be sick, elderly or irresponsible. Low maizeproduction was caused by factors mostly out of thecontrol of the poorer farmers (a small amount of landavailable for planting, inadequate or inappropriateinputs, hired draught power that resulted in lateploughing and delayed planting, labour constraints atcritical times such as weeding and harvest).

Average households. The farmers in the averagecategory are also forced to supplement their incomefrom the farm by selling labour and other products.Since this group commonly owns poultry and othersmall animals, the sale of eggs and meat providesimportant income. The average family size is similarto that of the poor farmers, and about 60% of house-holds are headed by females. Some ‘average’ fami-lies receive remittances from family members whowork in cities, either in Zimbabwe or abroad.

Rich households. Rich farmers have more diverseincome sources, ranging from primary production(livestock and surplus grain) to off-farm income fromoutside activities (salaries, remittances, pensions,businesses). Labour for farming activities comesfrom the extended family or is bought. Rich house-holds are considerably smaller than those of poor andaverage farmers, with 4–6 people. Some rich farmersare able to purchase Vlei land areas. Commonly, it isthe richer farmers that own the majority of the cattle.As a result, some of the communal grazing areas arebeing fenced to contain their stock and excludeothers’ stock. Some farmers are buying more culti-vable land in order to increase their grain harvests.

Development of APSIM scenarios

Farmers were divided into three groups. Eachgroup was asked to represent one of the wealthgroups that had been characterised by the wholegroup. The six men attending the meeting repre-sented the rich farmer category, and the 16 womenwere divided randomly into two groups to representthe poor and average farmers. The wealth grouping afarmer belonged to for this exercise did not neces-sarily reflect their actual wealth, but each was askedto act as if they belonged to that category. This was toavoid embarrassment, and to encourage frankexchange. The groups further discussed farming-related problems and drew resource maps of typicalpoor, average and rich households (Figure 1). Only

the poor and rich scenarios developed are reportedhere. The average scenario was discarded because ofits similarity to Agritex (Department of AgriculturalExtension) recommendations for maize production inthis district and because it did not seem to reflect anauthentic view of local farmers. The farmers consid-ered that a minimum yearly maize requirement of12 bags (600 kg) per family was the minimum foodsecurity threshold.

Scenarios for poor farmersThe major constraints to maize production and

food security identified by the poor were related tolabour and fertiliser. Since no cash was available topurchase inorganic fertiliser, the poor farmers wereforced to collect leaf litter and manure from the com-munal forested areas as a fertiliser source for maize.This required substantial labour inputs. The farmerswanted to investigate the trade-off between col-

Figure 1. Resource map drawn by farmersrepresenting the poorest group in ZimutoSiding, Zimbabwe, and illustrating a‘poor’ farmer’s enterprise with homebuildings, compost heap, well, fields,firewood and people (drawn and labelledin English and Shona, the local language).





176

lecting leaf litter as a low-quality fertiliser and usingcash to buy fertiliser. They estimated that in the timeit took to make 1 t of low-quality manure from leaflitter, approximately 15 kg N fertiliser/ha could bepurchased. The farmers also wanted to know theeffect of sowing date (1 or 15 November) on thesescenarios. The following scenarios were then simu-lated before the next day’s meeting.1. Low-quality manure, 1 t/ha, applied 15 October

– Sowing window starting 1 or 15 November– P, N and C in applied manure (dry weight

basis) = 0.76%, 0.5% and 25% respectively(C:N = 50).

– PO4-P, NH4-N, NO3-N in applied manure = 10,10, 10 ppm

2. 15 kg N/ha fertiliser applied as NH4NO3 at sowing– No manure applied.– Sowing window starting 1 or 15 November.The application of 15 kg N/ha at sowing

(1 November – 15 January) produced higher grainyields than the manure treatment in 13 out of the 18years (Figure 2). There was little difference in thegrain produced between the two sowing times, andonly the results for the earlier sowing time are pre-sented in Figure 2. In 1986 and 1998, the applicationof manure produced slightly higher grain yields, butin 1983 and 1992 there was no grain productionbecause of lack of rainfall, regardless of the treat-ment. When climatic conditions were good, the treat-ment receiving 15 kg N/ha produced much moregrain than the low-quality manure treatment.

The poor farmers generally have no animals andhave to collect manure from the field and combine itwith many types of low-quality residues, such as leaflitter. The composition of this manure is usually verypoor (C:N ratio = 50), resulting in very little N for thecrop and often in the immobilisation of any soilnitrate, further depriving the maize crop. As the pur-chase of inorganic N fertiliser requires cash, thefarmers decided that it was better to work for a richerfarmer for cash or fertiliser, rather than spend timecollecting manure. Overall, the small application of15 kg N/ha resulted in food security in 10 of the sea-sons, compared with only three seasons in which themanure was applied (Table 4).

Scenarios for rich farmersThe farmers representing the rich group were most

concerned with the effect of manure and its quality onmaize yield. They said that large amounts of manurewere available each season because of the largenumbers of cattle owned. This group wanted to knowwhether applying manure as collected from the fieldor kraal (low-quality C:N ratio = 50) was better thancomposting this manure to form a higher qualitymanure (high-quality C:N ratio = 25). They insistedthat 5 t manure/ha be applied, because they believedthat such a large amount of manure would be avail-able from the many cattle owned by rich farmers.However, based on local extension officer andresearcher experience, it is unlikely that so muchwould be available.

0

5

10

15

20

25

30

35

40

Yie

ld (b

ags/

acre

)

1 November sowing, 15 kg/N ha

1 November sowing, 1 kg/N ha low-quality manure

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

Figure 2. Maize grain yield of the scenario planted in the sowing window from 1November to 15 January with 15 kgN/ha or 1 t/ha of low-quality manureapplied at planting. Note: horizontal line represents minimum yearly maizerequirement for a family.



177

1. Low-quality manure, 5 t/ha, applied 15 October– P, N and C in applied manure (dry weight

basis) = 0.76%, 0.5% and 25% (C:N = 50)– PO4-P, NH4-N, NO3-N in applied manure = 10,

10, 10 ppm– Sowing window starting 1 or 15 November.

2. High-quality manure, 5 t/ha, applied 15 October– P, N and C in applied manure (dry weight

basis) = 0.76%, 1.0% and 20% (C:N = 20)– PO4-P, NH4-N, NO3-N in applied manure = 10,

10, 10 ppm– Sowing window starting 1 or 15 November.The higher quality manure generally resulted in

higher maize yield. Grain yield exceeded the familythreshold of 12 bags/acre in 10 of the seasons wherethe high-quality manure was used (Table 4). If the

low-quality manure was applied annually, there wereonly three occasions when this threshold was satis-fied (Figure 3). Total grain yield over the 18 seasonswas slightly higher at the earlier sowing time(1 November) regardless of the manure quality(Table 4), and only the yields for this sowing time areshown in Figure 3.

Soil fertility was positively affected by both treat-ments. Plant available soil nitrate nitrogen increasedfrom a negligible amount at the beginning of the sim-ulation to 99 kg nitrate nitrogen/ha on the high-quality manure treatments and 20 kg nitrate nitrogen/ha on the low-quality manure. As expected with largeyearly applications of manure, the soil organiccarbon improved from the starting value of 1% to1.2%, regardless of manure quality and sowing date.

Good-quality manure

Low-quality manure

0

5

10

15

20

25

30

35

40

Yie

ld (b

ags/

acre

)

1981

1982

1983

1984

1985

1986

1987

1988

1989

1990

1991

1992

1993

1994

1995

1996

1997

1998

Table 4. Seasons between 1981 and 1998 that maize production exceeded food security minimum (12 bags/acre), and cumulative grain produced during this period.

Sowing date Number of years >>>>12 bags/acre Total grain (1981–1998)(bags/acre)

1 Nov–15 Jan 15 Nov–15 Jan 1 Nov–15 Jan 15 Nov–15 Jan

Poor farmers 1 t/ha low-quality manure 15 kg N/ha

38

29

100243

83251

Rich farmers 5 t/ha low-quality manure 5 t/ha high-quality manure

310

310

161266

149257

Figure 3. Maize grain yield of the scenario planted in the sowing window1 November to 15 January, with 5 t/ha of good–quality and low-qualitymanure applied yearly.



178

Presentations of the simulations to farmersBefore the farmers were shown the simulation out-

puts, they were asked to identify the years in whichthe best and worst climatic seasonal conditionsresulted in the highest and lowest maize yields. Thefarmers indicated that the lowest yields had occurredduring 1983 and 1992, which were serious droughtyears. There was general agreement that 1993 was agood season with high yields. These best and worstyears corresponded well with the simulations, and therange of yield levels we presented was within thefarmers’ own experiences of maize production. Wepresented the results of the simulations as hand-drawn histograms on flip charts.

There is very little difference in total grain produc-tion between the poor farmers’ treatment, in which15 kg N/ha was applied at planting, and the richfarmers’ treatment with 5 t/ha of high-qualitymanure. Obtaining 5 t/ha of manure would require avery large input of labour and, because the collectedmanure would probably be of low quality, a period ofcomposting to narrow the C:N ratio to 20 would belikely. This would be an additional labour cost. Overthe long term, the increasing organic matter contentof the soil receiving the manure applications mayleave the system more fertile, but this should beweighed against the labour costs.

Several poverty alleviation strategies emergedfrom the scenario analysis:• Farmers who were currently selling labour should

pay more attention to their own fields to enabletimely planting. Some suggested the use ofminimum tillage to avoid ploughing delays.

• Farmers representing the poor group suggestedthat labour should be sold to meet immediate needsand fund small fertiliser purchases.

• All groups suggested that taking productionwindfalls from good years to invest in assets, suchas animals and education, is a good long-termstrategy.

Conclusions and Recommendations

The use of simulation modelling in the field withfarmers is a relatively new concept, and one theauthors believe can facilitate the integration of hardand soft sciences to improve farm management inter-vention. In Australia, the APSIM model has beenused extensively with commercial farmers throughthe FARMSCAPE projects (Keating and McCown2001; McCown et al. 2002). However, its use in par-

ticipative research with smallholder farmers inAfrica has been limited. In the experience describedin this paper, the APSIM model was able to achievea number of outcomes that would not have emergedfrom discussion of field trial results, local knowledgeand farmers’ experiences alone.

During a series of facilitated discussions, theZimuto farmers identified labour and resources askey constraints on soil fertility management and theimprovement of maize yields (Table 3). Scenariosdeveloped for each of the wealth rankings (poor andrich) and simulated using APSIM helped farmersmake decisions about investment in soil fertility toachieve household food security through increasedmaize harvests. The farmers were then able to discussways to more profitably direct their labour efforts.The choice between collecting poor-quality manuresand using this labour to buy small amounts of ferti-liser could be weighed up in terms of potential maizeyields. The fertiliser option was clearly superior.

The value of sharing hard data, such as long-termweather records and simulated grain yields, withfarmers as an input to discussions of trade-offs andclimatic risk management was also evident. Suchdata can be effectively presented in tabular or histo-gram form, depending on the researchers’ ability tosummarise data concisely. The farmer’s memory ofthe best and worst seasons is a powerful starting pointfor discussing the implications of such data.

Simulation models such as APSIM are unlikely tobe used by farmers directly. An expert user (con-sultant, scientist, researcher) might use the models inconsultation with the farmer as one tool for farmanalysis. Using simulation modelling to conduct pre-liminary investigations of issues that require researchcould also lead to more-focused field trials.

Whether the exercise at Zimuto resulted in anypractice change by the farmers was not further inves-tigated. Other work, reported in Pengelly et al. (2003)and Dimes et al. (2003), demonstrates that on-farmparticipative experimentation can change practices.The use of simulation tools in on-farm experimenta-tion and participatory research is suggested as a sen-sible progression of this work.

In using simulation models in the field, the authorslearnt the following lessons:• Good preparation is essential, but there is also a

need to be flexible and adapt the process on thespot.

• Target options to different groups (for example,farmers at different wealth levels).



179

• Keep the simulations simple enough to be runquickly by the modeller and understood by thefarmers.

• Where long-term climatic records are available,the element of risk can be brought into discussionsof profitability and production.

• Establishing credibility in the simulation output (inthe eyes both of the farmers and of the researchers)requires some field validation. Even the leastcompetent models can represent nil yields in dryyears, so it is important to ensure that thesimulations represent reasonable and sensibleresponses to changes in inputs.

Acknowledgments

Placedes Muzozviona and Mike Dhliwayo, extensionofficers at Agritex/DRES, Masvingo Agritex Office;Dishon Odero Raga, research officer of the KenyaAgricultural Research Institute; and Richard Foti forassisting in the field interviews at Zimuto.

References

Anderson, J.R. and Dillon, J.L. 1992. Risk analysis indryland farming systems 2. Rome, Food and AgricultureOrganization of the United Nations, 9.

Carberry, P., Gladwin, C. and Twomlow, S. 2004. Linkingsimulation modelling to participatory research insmallholder farming systems. In: Delve, R. and Probert,M.E., ed., Modelling nutrient management in tropicalcropping systems. ACIAR Proceedings No. 114, 32–46.

Carberry, P.S., Hochman, Z., McCown, R.L., Dalgliesh,N.P., Foale, M.A., Poulton, P.L., Hargreaves, J.N.G.,Hargreaves, D.M.G., Cawthray, S., Hillcoat, N.,Robertson, M.J. 2002a. The FARMSCAPE approach todecision support: farmers, advisors researchersmonitoring, simulation, communication and performanceevaluation. Agricultural Systems 74, 141–177.

Carberry, P.S., Probert, M.E., Dimes, J.P., Keating, B.A.,McCown, R.L. 2002b. Role of modelling in improvingnutrient efficiency in cropping systems. Plant and Soil,245, 193–203.

Dalgliesh, N. and Foale, M. 1998. Soil matters: monitoringsoil water and nutrients in dryland farming. CSIROAustralia, 71.Dimes, J., ed., 2002. Linking logics: farmerand researcher interactions on farm management andsimulation modelling in smallholder farming systems.ICRISAT, Bulawayo, Zimbabwe.

Dimes, J., Twomlow, S. and Carberry, P. 2003. Applicationof new tools: exploring the synergies between simulationmodels and participatory research in smallholder farmingsystems. Global theme 3: Water, soil and agrodiversity

management for ecosystem resilence, Report no. 5. POBox 776, Bulawayo, Zimbabwe, ICRISAT and PO Box102, Toowoomba, Queensland 4350, Australia, CSIROSustainable Ecosystems/APSRU, 20p.

Doyle, C. 1990. Application of systems theory to farmplanning and control: modelling resource allocation. In:Jones, J.G.W., Street, P.R., ed., Systems theory applied toagriculture and the food chain. London, Elsevier AppliedScience, 89–112.

Giller, K., Rowe, E., de Ridder, N. and van Keulen, H. 2004.Resource use dynamics and interactions in the tropics:moving to the scale of the ‘livelihood.’ AgriculturalSystems, in press.

Harrington, L. and Grace, P. 1998. Research on soil fertilityin southern Africa: ten awkward questions. In:Waddington, S.R., Muwira, H.K., Kumwenda, J.D.T.,Hikwa, D. and Tagwira, F., ed., Soil fertility research formaize-based farming systems in Malawi and Zimbabwe.Proceedings of the Soil Fert Net Results and PlanningWorkshop held from 7 to 11 July 1997 at AfricaUniversity, Mutare, Zimbabwe. Harare, Zimbabwe, SoilFert Net and CIMMYT-Zimbabwe, 3–11.

Keating, B.A., Carberry, P.S., Hammer, G.L., Probert, M.E.,Robertson, M.J., Holzworth, D., Huth, N.I., Hargreaves,J.N.G., Meinke, H., Hochman, Z., McLean, G., Verburg,K., Snow, V., Dimes, J.P., Silburn, M., Wang, E., Brown,S., Bristow, K.L., Asseng, S., Chapman, S., McCown,R.L., Freebairn, D.M., Smith, C.J. 2003. An overview ofAPSIM, a model designed for farming systemssimulation. European Journal of Agronomy, 18, 267–288.

Keating, B.A. and McCown, R.L. 2001. Advances infarming systems analysis and intervention. AgriculturalSystems, 70, 555–579.

Keating, B.A., Waddington, S., Grace, P., Rohrbach, D.,Dimes, J., Shamudzarira, Z., Carberry, P. and Robertson,M. 2000. Exploring farmer options for maize productionstrategies via scenario analyses using the APSIM model— an example of the approach. Risk ManagementWorking Paper Series, No. 2000/02. SOILFERTNET/CIMMYT.

Mapfumo, P. and Giller, K.E. 2001. Soil fertility andmanagement strategies and practices by smallholderfarmers in semi-arid areas of Zimbabwe. ICRASAT andFood and Agriculture Organization of the United Nations.

McConnell, D.J. and Dillon, J.L. 1997. Farm managementfor Asia: a systems approach. FAO Farm SystemsManagement Series 13. Rome, Food and AgricultureOrganization of the United Nations, 272.

McCown, R.L., Keating, B.A., Carberry, P.S., Hochman, Z.and Hargreaves, D. 2002. The co-evolution of theagricultural systems simulator (APSIM) and its use inAustralian dryland cropping research and farmmanagement intervention. In: Ahuja, L.R., Ma, L. and



180

Howell, T.A., Agricultural system models in fieldresearch and technology transfer. Lewis Publishers.

McDougall, C. and Braun, A. 2003. Navigating complexity,diversity and dynamism: reflections on research fornatural resource management. In: Pound, B., Snapp, S.,McDougall, C. and Braun, A., ed., Managing naturalresources for sustainable livelihoods: uniting science andparticipation. Earthscan/IDRC, 20–47.

Pengelly, B.C., Whitbread, A., Mazaiwana, P.R. andMukombe, N. 2003. Tropical forage research for thefuture — better use of research resources to deliveradoption and benefits to farmers. Tropical Grasslands, 37,207–216.

Shamudzarira, Z. and Robertson, M.J. 2002. Simulatingresponse of maize to N fertiliser in semi-arid Zimbabwe.Experimental Agriculture, 38, 79–96.

Shamudzarira, Z., Robertson, M.J., Mushayi, P.T., Keating,B.A., Waddington, S., Chiduza, C. and Grace, P. 1999.Simulating N fertilizer response in low-input farmingsystems. 1. Fertilizer recovery and crop response. ESASymposium on Modelling Cropping Systems, Spain,June 1999.



tropical legumes for sustainable farming systems in ...aciar.gov.au/files/node/555/proc115.pdf ·...

Documents