integrated indicators for performance assessment of traditional agroforestry systems in south west...

Integrated indicators for performance assessmentof traditional agroforestry systems in South West Cameroon

Geraldo Stachetti Rodrigues Inacio de Barros Euge`ne Ejolle Ehabe Patrick Sama Lang Frank Enjalric

Received: 13 February 2008 / Accepted: 29 May 2009 / Published online: 13 June 2009

Springer Science+Business Media B.V. 2009

Abstract Farming Systems developed in Humid

Tropical Zones are frequently characterized by a

combination of perennial and annual plants, inter-

mixed in complex tree-crop associations. The

productive functioning, the agronomic and economic

performances, and the sustainability of these crop

associations remain poorly understood. To improve the

management capacity of these complex agroforestry

systems, adequate indicators must be developed and

integrated in assessment systems. These may then be

used to aid farmers, assisted by their extension agents,

in making decisions regarding management practices.

The present study focused on the agroforestry systems

developed by 38 farmers in the South West Region of

Cameroon, which were surveyed for a large set of

variables, aiming at formulating a Traditional Agro-

forestry Performance Indicators System (TAPIS).

Analyses of the relationships among indicators in

TAPIS allowed an improved understanding of agro-

ecological and agro-economic performances in the

studied plots, revealed tradeoffs regarding plant stand,

income generation, food production, input demands

and work requirements; and may contribute to the

sustainability assessment of agroforestry systems.

Keywords Agroforestry systems (AFS) Sustainability assessment Integrated indicators Traditional agriculture

Introduction

Although highly varied, typical traditional farming

systems in Humid Tropical Zones (HTZ) are most

G. S. Rodrigues (&)Embrapa Labex Europe, Montpellier, France

e-mail: [email protected]

URL: http://www.agropolis.fr/international/labex.html

Present Address:G. S. Rodrigues

Unite Propre de Recherche Performance des syste`mes de

culture de plantes perennesCIRAD-PerSyst, Avenue

Agropolis, 34398 Montpellier, France

I. de Barros

INRA, Unite de Recherche Agropedoclimatique da la

Zone Carabe, Domaine Duclos, 97170 Petit-Bourg

(Guadeloupe), France


E. E. Ehabe

Latex Programme, Institute of Agricultural Research for

Development (IRAD), Ekona Regional Research Centre,

PMB 25, Buea, Cameroon


P. S. Lang

CARBAP (Centre Africain de Recherche sur la Banane et

le Plantain), Douala, Cameroon


F. Enjalric

Unite Mixte de Recherche Syste`me, CIRAD Cultures

Perennes, 2 Place Viala, 34000 Montpellier, France


123

Agroforest Syst (2009) 77:922

DOI 10.1007/s10457-009-9237-7

commonly diversified, plurispecific and multi-layered

associations of perennial and annual plants, coexis-

ting in long-lasting, complex and ever evolving

cropping stands. Even if usually managed with low

levels of inputs and technology application, these

farming systems tend to have acceptable productive

and economic performance, while being less suscep-

tible to climatic risks and showing excellent social

acceptability. Although confronted with low land

availability and low soil fertility, shortening of fallow

periods and market insertion difficulties, such Agro-

forestry Systems (AFS) continue to ensure the

livelihood of large portions of rural populations in

HTZ. Given the current value placed on preserving

the ways of life of traditional peoples, these AFS

represent much more than a simple subsistence

alternative, but a contribution to conservation of

biodiversity and to sustainability (Nair 1998).

The continued contribution of traditional AFS to

these aforementioned objectives depends upon a

better understanding of their agro-ecological and

socio-economic performances (Nair 2001; Schroth

and Sinclair 2003). However, methods and indicators

usually enlisted for the performance evaluation of

conventional (intensive, mono-cropping) farming

systems are inappropriate for these AFS (Kumar

and Nair 2004), given the essential role played by

issues such as family income and food security, work

productivity, harvest diversification, input indepen-

dence, judicious use of natural resources and man-

agement of beneficial and adventitious plants (Izac

2003). Even simpler attributes such as biomass

production and yields in these spatially stratified

and temporally heterogeneous systems are structur-

ally different from those observed in mono-cropped

areas (Rao and Coe 1991).

Studies with detailed characterization of tree-crop

biodiversity trends and interactions in traditional AFS

have been widely carried out, e.g., in the Vanuatu

archipelago (Barrau 1955; Bonnemaison 1996; Stro-

bel 1998; Walter and Leplaideur 1999; Greindl 2000;

Allen 2001; Seremele 2002; Morelli 2003; Lamanda

2005). The descriptive approach of these studies,

however, hampers their application to support farm-

ers management decisions, or even to carry out

performance assessments, based on meaningful eco-

logical, agronomic and economic indicators. None-

theless, the successful design and sustainable

management of AFS depends on the ability to

harness a very diverse set of biophysical, economic

and social data, and organize them into synthetic,

understandable recommendations (Ellis et al. 2004)

that account for and elucidate the relationships and

tradeoffs among concurrent indicators.

The adoption of such tools in the management of

farming systems can greatly favour the sustainability

of AFS, by improving farmers decision making

regarding, for example, the planning of land parti-

tioning and sequencing of perennial and annual crops;

the selection of appropriate practices for weed and

pest control; and the allocation of inputs, resources,

investment capacity and workforce and product

destination. These issues constitute priorities in the

appraisal of challenges and opportunities for the

agricultural sector of developing regions. This is

especially true in the HTZ, where diversified agro-

forestry systems are usually practised in small family

plots much in need of technical and managerial

support (Tollens 2003), and where sustainability

issues are becoming more evident in policy making,

as in Central and West Africa (Duguma et al. 2001).

Hence, appropriate indicators have been eagerly

sought to allow for performance assessment and the

ensuing recommendation of management practices

for AFS.

The present study focused on the performance

assessment of AFS in the South West Region of

Cameroon, aiming at (1) proposing an integrated

indicator system that may aid farmers, assisted by

their extension agents, to decide on management

practices and (2) contributing toward sustainability

evaluations of traditional agroforestry systems.

Choice of approach for indicator definition

and expression

Sustainability assessment and its consideration for

farm management consist of an exercise of translat-

ing concepts, ideas and paradigms into locally valid

value judgments, according to locally defined objec-

tives, systematized into practical measurement pro-

cedures and meaningful expression units (Bosshard

2000). Once the objectives have been defined, namely

to improve farmers management capacity (essen-

tially a biophysical efficiency attribute) and to foster

sustainability of landholdings (essentially a socio-

economic adequacy attribute), it is possible to list and

select appropriate field measurements to produce

10 Agroforest Syst (2009) 77:922

123

coherent indicators (Bockstaller et al. 1997; Girardin

et al. 1999; Lewandowski et al. 1999; Rodrigues

2004).

These aforementioned objectives for sustainability

assessment of AFS provide the basis for grading all

selected field measurement variables according to

improved or worsened performance, allowing the

ranking of plots and combined production practices

into normalized and aggregated indicators (Andreoli

and Tellarini 2000). The advantage of opting for

normalization of data sets and indicators (instead of,

for example, utility valuation or benchmarking) is the

consistency obtained for the ranking baseline, and the

meaning of the information conveyed by the indica-

tors pertaining to the local reality (Hardi and

DeSouza-Huletey 2000).

The objective of devising an assessment system to

aid farmers and extension agents in management and

decision making requires that the aggregation and

expression of indicators should favour prompt under-

standing, preferably with clear visualization of per-

formance levels and tradeoffs among sets of

indicators. The expression of assessment results in a

normalized scale eases comparison among different

indicators without the need for weighting factors,

thereby facilitating integration and clear graphic

expression, as accomplished for example with the

now widely used sustainability polygons (Hani et al.

2003; Rodrigues and Moreira-Vinas 2007).

These premises have been carefully observed in the

present choice of format for organizing an AFS

sustainability and performance assessment frame-

work, based on relative ranking and indexing proce-

dures (Liebig et al. 2004), so as to provide a single

unifying measure (non-dimensional) for the multiple

monetary and non-monetary units measured in the field

(Hajkowicz 2005). Also, in order to build consistency

for the ranking baseline, variables and indicators must

be equally valid and similarly surveyed, with the

advantage of making the information conveyed by the

indicators to the farmers more meaningful within the

locally encompassed reality, well represented by the

mean performance [that is, the composite output

indicator (Bockstaller and Girardin 2003)]. It was

with the aim of satisfying these requisites and choice of

approach that the Traditional Agroforestry Perfor-

mance Indicators System (TAPIS) described in the

following sections has been devised and validated in

the present case study.

Research setting and methodology

Field observations were carried out on 38 typical

agroforestry plots owned by farmers who agreed to

participate in the study, distributed around Kumba and

extending to the Bombe-Malende zones (42504800 N and 92509350 E) in the South West Regionof Cameroon. These regions fall within the rainforest

area (mean precipitation 2,852 mm year-1), have a

marked rainy season (March to October) and high

mean annual temperatures (*23C). Soils are ferral-litic with patches of fertile volcanic areas, and altitudes

varying from 25 to 400 m toward the North. The land

use systems in the area are typically characterized as

agroforestry (Sinclair 1999), permanently occupied

(no fallow) small areas integrating main perennials

(cocoa, oil palm and rubber trees), food crops (plantain,

cassava, yams, maize, banana, etc.) and native trees (as

well as ornamentals and medicinal plants not consid-

ered in the surveys). The main crops, their develop-

ment stages and basic terrain characteristics of the

studied plots can be seen in Table 1.

A comprehensive questionnaire and a field survey

datasheet were developed and filled out for the 38

selected plots. These contained information on the

variables derived from researchers proposed check-

lists and previous inventory procedures: identification

of farmers (name, gender and age, origin, education,

main occupation, sources of income, family compo-

sition, land property status, availability of working

tools, etc.), plot location and land use history

(cultivated area, crop stands, crop development stages

and densities, native trees and adventitious plants

presence, etc.) and plot economics (value of produc-

tion per crop, sales and self-consumption, input

acquisition, expenses, costs, revenue, etc.).

Data collection involved a first visit to interview

the farmer, locate the plot and note general charac-

teristics and fill out the household and production/

economics information sheet (all data gathered for

expression as ha-1 year-1). Subsequent visits (num-

ber conditioned by complexity of plot composition)

were dedicated to sampling plants for identification

as needed and to describing plant stands (horizontal

and vertical typology, based on complete observation

or randomly demarcated 100 m2 replicated plots) and

the chronosequence based on production stage and

characterization of agricultural practices and deter-

minants. Verification of the interview data for

Agroforest Syst (2009) 77:922 11

123

consistency with plot history and described stand was

also carried out during the field visits.

All basic variables collected were then combined

with additional external information, such as produce

prices and workforce costs, to produce intermediate

measures such as total plant stand diversity and

density, adventitious and spontaneous plants diversity

and densities, partial and total input costs, total

Table 1 Main crops and development stages, basic terrain characteristics and mean indicator indices of studied plots

Parcel

number

Main

perennial crop

Main crop

development stage

Altitude

(m)

Slope

(%)

Clay

(%)

Mean agroeconomic

performance

Mean agroecological

performance

1 Cocoa Immature 219 1 35 0.375 0.371

2 Cocoa Immature 220 0 50 0.280 0.230

3 Cocoa Immature 229 10 30 0.332 0.334

4 Cocoa Immature 178 2 40 0.393 0.243

5 Cocoa In production 216 0 40 0.341 0.232







12 Oil palm Immature 207 7 30 0.376 0.236



15 Oil palm In production 199 1 35 0.423 0.276




19 Rubber tree Immature 219 2 35 0.265 0.351

20 Rubber tree In production 209 3 40 0.323 0.226

21 Cocoa Immature 45 3 30 0.459 0.347

22 Cocoa Immature 67 5 35 0.434 0.403

23 Cocoa Immature 89 3 40 0.429 0.279

24 Cocoa Immature 38 3 30 0.525 0.463




28 Oil palm Immature 87 12.5 30 0.505 0.335












123

expenditures and revenues, etc. Finally, indicators

were created by the aggregation of variables, and

normalized for all farmers into indices of relative

performance by linear transformation based on the

indicator score for each farmer and the maximum

indicator score across all plots.

Two sustainability dimensions, agro-economic and

agro-ecological, were defined for plot performance

ranking, each comprised of a set of eight indicators,

as follows:

The composition of these locally meaningful

indicators resulted from (1) a regression significance

analysis of the broad set of field variables surveyed,

(2) the experience attained by contact with the

farmers and the local reality, and (3) a review of

integrated indicators systems for environmental farm

(and AFS) management. Accordingly, agro-economic

indicators were devised to assess the essential socio-

economic adequacy attributes of cash flow, work

dedication, expenses and profitability (Hani et al.

2003; Monteiro and Rodrigues 2006). Agro-ecolog-

ical indicators were, for their part, devised to cover

the essential biophysical efficiency attributes of

productivity, land use, productive diversity and weed

competition (Liebman 1988; Akinnifesi et al. 1999;

Schroth 1999; Schroth et al. 2001; Hani et al. 2003).

The rationale underlying the choice of indicators and

intervening variables were defined as shown below.

Agro-economic dimension indicators

Income score for each plot was estimated as the sum

of income from all crops, normalized by maximum

obtained income across all plots:

iIni IniInmax

and Ini Xn

c1Cc 10000

Si1

where iIni was the income score of plot i (no unit); Inithe total crop income per hectare (in FCFA); Inmaxthe maximum crop income obtained across all plots

(in FCFA); Cc refers to the income provided by the

crop c and Si the area of the plot (in m2).

Input expenses score was the sum of expenses for

all inputs, normalized by maximum input expenses

across all plots:

iExi ExiExmax

and Exi Xn

e1Iee 10000

Si2

where iExi was the input expenses score of plot i (no

unit); Exi the total expense for all inputs per hectare

(in FCFA); Exmax the maximum input expenses

obtained across all plots (in FCFA); Iee refers to

the expense for the input e and Si the area of the plot

(in m2).

Pesticide independence score was the balance of

the sum of expenses on herbicides and pesticides,

normalized by the maximum pesticide expenses

across all plots:

iPesti 1 PestiPestmax

and

Pesti Xn

h1Chh

Xm

p1Cpp

! 10000

Si

3

where iPesti was the pesticide independence of plot i

(no unit); Pesti the total expense for herbicides and

pesticides per hectare (in FCFA); Pestmax the max-

imum expenses for herbicides and pesticides obtained

across all plots (in FCFA per hectare); Chh refers to

the expense for herbicide h, Cpp refers to the expense

for pesticide p and Si the area of the plot (in m2).

Hired workforce independence score was the

balance of total cost of hired workers per hectare,

normalized by maximum cost of labour across all

plots:

Agro-economic dimension

indicators

Agro-ecological dimension

indicators

(1) Income (1) Harvest

(2) Input expenses (2) Area equivalence index

(3) Pesticide independence (3) Soil resource use index

(4) Hired workforce

independence

(4) Productive diversity

(5) Family workforce

engagement

(5) Diversity of associated

arboreal species

(6) Total workforce

independence

(6) Adventitious plants

controllability

(7) Internal gross added

value

(7) Beneficial adventitious

plants

(8) Total gross added value (8) Adventitious plants

infestation control


123

iWfi 1 WfiWfmax

and Wfi Wfti 10000Si

4

where iWfi was the hired workforce independence

index of plot i (no unit); Wfi was the total cost of

hired workforce per hectare (in FCFA per hectare);

Wfmax the maximum observed cost of hired work-

force per hectare across all plots (in FCFA per

hectare); Wfti refers to the total costs of hired

workforce in the plot and Si the area of the plot (in

m2).

Family workforce engagement score for each plot

was the total estimated value of family work per

hectare, normalized by maximum family work value

across all plots:

iFwei FweiFwemax

and Fwei Fwvi 10000Si

5

where iFwei was the family workforce engagement

index for plot i (no unit); Fwei the estimated value of

the family workforce (in FCFA per hectare); Fwemaxthe maximum estimated value of the family work-

force per hectare across all plots (in FCFA); Fwvirefers to the total estimated value of the family

workforce per hectare in the plot and Si the area of the

plot (in m2).

Total workforce independence score for each plot

was the balance of total cost and value of (family and

hired) workers per hectare, normalized by workforce

independence across all plots:

iTwfi TwfiTwfmax

and Twfi 1 Wfi Fwei6

where iTwfi was the total workforce independence

index for plot i (no unit); Twfi the total cost and value

of (family and hired) workers per hectare (in FCFA).

Internal gross added value score for each plot was

the sum of income from all crops minus the sum of all

expenses, per hectare, normalized by internal gross

added value across all plots:

iGavi GaviGavmax

and Gavi Ini Exi 7

where iGavi was the internal gross added value index

for plot i (no unit) and Gavi the internal gross added

value per hectare (in FCFA).

Total gross added value score for each plot was the

sum of income from all crops minus the sum of all

expenses excluding the value of family work, per

hectare, normalized by total gross added value across

all plots:

iTgai Tgai

Tgamaxand Tgai Ini Exi Fwei

8where iTgai was the total gross added value index for

plot i (no unit) and Tgai the total gross added value

per hectare (in FCFA).

Agro-ecological dimension indicators

Total Harvest score per hectare for each plot was

estimated as the sum of production from all crops,

normalized by maximum obtained production across

all plots:

iHi HiHmax

and Hi Xn

c1Pcc

! 10000

Si9

where iHi was the harvest score for each plot; Hi the

total crop harvest per hectare (in kg) and Pcc the

harvested mass of each crop c.

Area equivalence index (Aei) score for each plot

was the sum of the ratios of each given crop density

and the standard density for each given crop in

monoculture (Liebman 1988), normalized by maxi-

mum Aei across all plots:

iAeii AeiiAeimax

and Aeii Xn

c1

Cdc

Sdc10

where iAeii was the area equivalence index score (no

unit); Aeii the area equivalence index (in ha ha-1);

Aeimax the maximum observed Aei observed across

all plots; Cd observed planting density (in plants

ha-1) and Sd the standard density for crop c in

monoculture (in plants ha-1). The Sd values used in

this study are presented in Table 2.

Soil resources use index (Sui) score for each plot

was the AEIi, weighted with a (0.1) discount for

annuals and stage one perennials, normalized by

maximum SUI across all plots:

iSuii SuiiSuimax

and Suii Xn

c1Succ

While


123

Succ CdcSdc if c perennial stage [ 1Succ 0:1 CdcSdc if c annual or perennial stage 1:

11where iSuii was the soil resources use index score (no

unit); Suii the soil resources use index (no unit);

Suimax, the maximum observed Sui across all plots

and Suc the soil resource use coefficient (no unit).

Productive diversity score for each plot was the

ShannonWiener Index (H0) of the proportion ofincome from all crops:

Pdi Xn

c1

Cc 10000SiIni

ln Cc 10000

Si

Ini

!12

where Pdi was the productive diversity index.

Diversity of associated arboreal species score for

each plot was the ShannonWiener Index (H0) of theproportion of spontaneous tree species conserved in

the plot:

Dari Xn

a1

Nta

Nti ln Nta

Nti

!13

where Dari was the Diversity of associated arboreal

species score for plot i; Nta the number of spontaneous

trees of species a and Ntt the total number of

spontaneous trees.

Adventitious plants controllability score for each

plot was the balance of the ShannonWienner Index

(H0) of the proportion of adventitious plants infestingthe plot:

Davi 1 Xn

v1

Advv

Advi ln Advv

Advi

!14

where Davi was the adventitious plant (weed)

controllability score; Advv the area covered by the

adventitious species v (in % area) and Advt the sum

of area covered by all adventitious plants (in %

area).

Beneficial adventitious plants score for each plot

was the sum of the ratios of cover (% cover) for the

selected non-weed adventitious plants (especially

N-fixing legumes (Schroth et al. 2001)) and the sum

of cover for all weeds in the plot:

iBai BaiBamax

and Bai Pn

a1 NadaAdvi

15

where iBai was the beneficial adventitious plants

score and Nad the area covered by selected non-weed

adventitious plants a.

Finally, adventitious plant infestation control score

for each plot was the balance of the product of the

sum of each adventitious species cover (including

over cover) and the total adventitious surface cover,

normalized by the maximum observed value across

all plots:

iAdci 1 AdciAdcmax

and Adci

Pna1 Ai;aAi

16where iAdci was the adventitious plant infestation

control score, Adci the ratio between the sum of the

estimated coverage for each adventitious species Ai, aand the estimated field coverage by adventitious

species Ai. This indicator accounts for the over-

coverage between the different species, as two or

more adventitious plants may occupy the same tri-

dimensional space over the field surface.

With these formulations, all indicators fit within a

relative performance index from 0 to 1 (Hajkowicz

2005), allowing straightforward aggregation and

integrated ranking of all plots under the two sustain-

ability dimension axes, related to agro-economic and

Table 2 Standard densities for the main crops in mono-cropping (in plants ha-1) used for Area equivalence index

calculations

Crop Latin name Standard density

(plants ha-1)

Cocoa Theobroma cacao 1,300

Oil palm Elaeis guineensis 143

Rubber tree Hevea brasiliensis 550

Coffee Coffea spp. 1,500

Plantain Musa spp. 1,600

Cassava Manihot esculenta 10,000

Cocoyam Xanthosoma sagittigolium 10,000

Maize Zea mays 25,000

Yam Dioscorea spp. 10,000

Pineapple Ananas commosus 30,000

Egusi Citrullus lanatus 2,500

Peanut Arachis hypogea 50,000

Huckleberry Solanum scabrum 150,000

Eggplant Solanum melongea 10,000

Avocado Persea americana 120

Source: CIRADGRET 2002


123

agro-ecological performances. The integrated analy-

sis of the assessment results consisted of graphic

examination of performance indices for the 16

indicators, individually for each plot and for the

mean of all plots. Linear regression analyses were

also carried out for interpretation of convergences

and tradeoffs in the full set of indicator indices,

across all plots.

To make the integrated system more interactive, a

blank open line was included in the TAPIS spread-

sheets to allow farmers and extension agents to use it

for the assessment of new plots, or for the periodic

reassessment of previously studied plots to check for

any performance changes as crops mature, or the

efficacy of management improvements carried out, in

relation to the database currently available within the

system.

Results

General data regarding plot sizes, distribution accord-

ing to production stage for the three different main

crops, associated annual crops, presence of sponta-

neous arboreal species and basic economics are

displayed in Table 3. Although some of these data

integrate information that comprise certain of the

indicators in TAPIS, their expression as raw values,

share of the gross income distribution and contrasts

according to plot situation provided in Table 3 offer a

complementary understanding of the local agrofor-

estry productive arrangement. With sizes ranging

from just 1,000 m2 up to 4.0 ha, irrespective of main

crops and production stages, all plots are densely

packed with perennial seedlings and a diversity of

annuals in the implantation phases, progressing to

still dense plant stands even when main crops reach

production; with the exception of rubber tree plots,

which tend to almost exclude annuals after onset of

latex extraction. The small number of spontaneous

trees in the plots confirms their relatively intensive

management, while not showing a definite tendency

according to the production stage or type of main

crop.

Gross incomes vary from approximately 640

1,120 US$ ha-1 year-1, with oil palm bringing the

larger amounts. When accounting for expenditures,

pesticide uses tend to reach a maximum of 18% for

cocoa and 13% for rubber trees (essentially ethylene),

both in their productive stages, while not used in oil

palm plots. The sum of other inputs (fertilizers, plant

multiplication material and transportation) tend to be

most important in the immature stages of the main

crops, reaching 56% in oil palm and up to 50% in

cocoa plots, essentially for acquisition of seedlings.

In several instances, negative cash flows were

observed in this stage, with farmers investing to

establish their perennial crop stands. These expendi-

tures for other inputs were sharply reduced in the

productive stages of the main crops, remaining at

most at just 6.4% in rubber tree plots (Table 3).

An important income item in the studied plots is

production for self consumption (as added value

obtained from associated annuals). While being

predominant in the immature stages of main crops

development, it accounts for approximately 20, 40

and 60% in rubber tree, cocoa, and oil palm plots

respectively. These correspond to essential gains

during the investment phase of plant stand imple-

mentation, when expenditures on inputs tend to be

higher. Costs then shift to workforce demand (be

these hired or familial) for mature stand management

and harvesting, resulting in net incomes ranging from

224 to 731 US$ ha-1 year-1 for oil palm in immature

and production stages respectively (all gains consid-

ered, including self consumption).

The aggregated results for the mean performance

indices across all plots (included in Table 1) show

that no farmer obtained combined agro-economic and

agro-ecological indices ranked within the upper

performance quartile, according to the set of indica-

tors assessed in TAPIS (Fig. 1). Even if one considers

the modest socio-economic situation of the region,

and the typically small, plurispecific character of the

plots represented in the dataset, this result implies, on

the one hand, performance unevenness among farm-

ers within each of the indicators; and on the other,

important tradeoffs among indicators for all plots.

This is reaffirmed with the hypothetical representa-

tion of a new observation expressing the third

quartile level for the field variables evenly across all

plots, which is the mechanism for including new

plots, or re-assessments, in TAPIS.

Observation of the distribution of main crops and

their development stages (whether immature or in

production) shows that there are no evident clusters

determining performance trends, which was con-

firmed in an agglomerative hierarchical clustering


123

Ta

ble

3P

lot

size

s,d

istr

ibu

tio

nac

cord

ing

top

rod

uct

ion

stag

efo

rth

eth

ree

dif

fere

nt

mai

ncr

op

s,as

soci

ated

ann

ual

s,p

rese

nce

of

spo

nta

neo

us

arb

ore

alsp

ecie

s,an

db

asic

eco

no

mic

s

Ch

arac

teri

stic

sM

ain

cro

pC

oco

ap

rod

uct

ion

stag

eO

ilp

alm

pro

du

ctio

nst

age

Ru

bb

ertr

eep

rod

uct

ion

stag

e

Imm

atu

reIn

pro

du

ctio

nIm

mat

ure

Inp

rod

uct

ion

Imm

atu

reIn

pro

du

ctio

n

Plo

tsi

zem

ean

and

(ran

ge,

ha)

1.0

9

(0.0

7

2.0

)

1.3

2

(0.2

4

2.0

)

0.8

1

(0.2

1

.31

)

1.1

0

(0.3

5

2.3

6)

1.1

9

(0.2

2

3.5

)

1.6

5

(0.1

0

4.0

)

Plo

tsac

cord

ing

tom

ain

cro

pp

rod

uct

ion

stag

e:

pro

po

rtio

nw

ith

incr

op

and

(nu

mb

erw

ith

insa

mp

le)

44

(08

)5

6(1

0)

46

(06

)5

4(0

7)

60

(04

)4

0(0

3)

Mai

ncr

op

den

sity

(pla

nts

ha-

1)

1,7

00

1,2

28

22

32

06

81

58

20

Mai

nas

soci

ated

ann

ual

cro

ps

den

sity

(pla

nts

ha-

1),

for

plo

tsw

her

ep

rese

nt,

com

bin

atio

ns

var

yin

gw

idel

y

Pla

nta

in(2

9p

lots

)1

,74

22

57

57

32

28

22

00

.0

Cas

sav

a(1

6p

lots

)4

70

45

01

0,4

50

7,8

00

6,6

30

1,2

80

Co

coy

am(X

an

tho

som

a)

(22

plo

ts)

1,7

45

1,7

73

3,3

00

7,0

00

20

0.0

Mai

ze(0

8p

lots

)1

0,4

30

0.0

3,1

30

3,0

00

17

,64

00

.0

Mea

nn

um

ber

of

spo

nta

neo

us

arb

ore

alsp

ecie

s

(act

ual

occ

urr

ence

ha-

1)

89

12

35

13

Eco

no

mic

sac

cord

ing

top

rod

uct

ion

stag

e

Plo

tm

ean

gro

ssin

com

e(U

S$

ha-

1y

ear-

1)

Ex

chan

ge

rate

52

7.3

fcfa

per

US

$(0

1/1

2/0

8)

80

47

80

63

91

,12

29

67

80

0

Plo

tm

ean

net

inco

me

(US

$h

a-1

yea

r-1)

36

85

65

22

47

31

65

76

88

Pro

po

rtio

na

of

inco

me

spen

tw

ith

pes

tici

des

(%)

3.8

18

.10

.00

.06

.51

3.6

Pro

po

rtio

no

fin

com

esp

ent

wit

ho

ther

inp

uts

(%)

50

.61

.15

5.7

1.9

3.5

6.4

Pro

po

rtio

no

fin

com

eas

self

-co

nsu

mp

tio

n(%

)3

9.1

8.8

62

.01

5.0

18

.25

.2

aA

llp

rop

ort

ion

sb

ased

on

tota

lin

com

ep

ery

ear


123

carried out on the dataset. This means that the variety

of crop combinations, associated production stages,

and practices adopted in the different plots were more

important in determining performance, as indicated

by TAPIS indicators, than the main crop alone, while

a significant relationship still exists between the sets

of indicators for the two axes (Fig. 2).

About one-fifth (8 out of 38) of the plots gave

agro-economic mean performance indices above the

0.5 level, with the best performance indices being

Traditional Agroforestry Performance

New observation

0.0

0.5

1.0

0.0 0.5 1.0Agroeconomic Indicators

Agr

oe

co

logi

ca

l In

dic

ato

rs

Fig. 1 Aggregated agro-economic and agro-

ecological performance

indices for 38 plots studied

in south west Cameroon

with the Traditional

Agroforestry Performance

Indicators System (TAPIS).Open symbols representimmature stages of main

crop development, filledsymbols represent maincrops in production stage.

Main crops are represented

as follows: m cocoa, d oil

palm, j rubber tree

00

Mean Agroeconomic Performance

Mea

n A

groe

colo

gica

l Per

form

ance

Observations Predictions Conf. on prediction (95.00%)

0.6

0.5

0.4

0.3

0.2

0.1

0.1 0.2 0.3 0.4 0.5 0.6 0.7

Fig. 2 Linear regression analysis between agro-economic andagro-ecological performance indices for the 38 plots studied

with TAPIS in south west Cameroon. Regression equation for

predicted linear regression: mean agroecological perfor-

mance = 0.198 ? 0.282 9 mean agroeconomic performance

(P-value = 0.046)


123

related to Pesticide independence (mean = 0.80,

measured according to expenses, hence an agro-

economic indicator), Total workforce independence

(mean = 0.78), and Hired workforce independence

(mean = 0.66, Fig. 3). This result means that, on

average, low expenditures were directed toward

pesticide inputs and hired worker recruitment. Inter-

estingly enough, these indicators are inversely and

significantly correlated with the level of Income and

Added value (or profit, both internal and total), which

in turn are directly correlated among themselves,

meaning that those who obtain better incomes tend to

rely on higher investments. These trends are con-

firmed by a very low level of Input expenses (0.08) in

general, and a fairly low level of Family worker

engagement (0.21).

Regarding the agro-ecological indicators, and with

only one exception, all plots ranked in the lower

performance quartile (Fig. 1). Only the Adventitious

plants infestation control indicator reached a mean

value above 0.5 (0.61), indicating a certain equita-

bility among the local farmers, which is interpreted as

a tendency for an adequate management situation for

the indicator (Fig. 3), as suggested by a significant

positive correlation between this indicator and the

adventitious plants controllability. This latter indica-

tor, itself related to a low diversity of weeds, is

significantly but inversely correlated with the pres-

ence of Beneficial adventitious plants. This strategy

seems logical as weeding is a major time consuming

agricultural practice and usually a constraint for

farmers.

The Area equivalence index was the second

highest among the agro-ecological performance

indicators (value = 0.47), being related to a high

level of crop association (max = 4.9 ha ha-1). Not

surprisingly, the plot with the highest level of crop

association showed one of the highest Productive

diversity indices (0.71), but lowest Soil resource use

index (0.22), and lowest Harvest (0.08), given the

clear situation of immature crop stand for the

perennials. This expected correlation between AEI

and Productive diversity (that is, crop variety) is

verified as significant in the general dataset.

Confirming the performance results and the trade-

offs observed for the agro-economic indicators for the

whole group of plots, with mean Income and Added

value indicators being low, the total Harvest indicator

showed the lowest mean agro-ecological performance

index, implying that the majority of the plots had

dense plant stands (high AEI) consisting mostly of

still immature crops, resulting in a low mean Soil

resource use index (0.31). In fact, only 15 of the 38

plots (*40%) already had the main perennial crop inproduction stage. A modest Diversity of associated

arboreal species (0.33) indicated a relatively low

importance of non-crop, spontaneous tree species

conserved in the plots.

Discussion and conclusions

The use of a sampling strategy directed at complex

plurispecific cropping systems fulfilled a two-way

objective: on the one hand it brought homogeneity in

spatial scale and level of management capacity, and

on the other hand it introduced a great heterogeneity

in terms of complexity and temporal dynamics

Income

Input expenses

Pesticide independence

Hired work force independence(men days)

Family work force engagement (mendays)

Total work force independence

Internal Gross Added Value

Total Gross Added Value

Harvest (kg) / ha

Area Equivalent Index

Soil Resources Use Index

Productive diversity

Diversity Associated ArborealSpecies

Adventitious Plants Controllability

Beneficial adventitious plants

Adventitious Plants Infestationcontrol

Fig. 3 Mean results for the agroeconomic and agroecologicaldimension indicators studied in 38 plots in south west

Cameroon with the Traditional Agroforestry Performance

Indicators System (TAPIS). s mean of immature developmentstage and h in production stage of the main crop


123

(cropping stages, volume of production, plant den-

sity, produce self-consumption, etc.), in order to

include very contrasting contexts. The representation

of the actual situation observed at the moment of

sampling and extrapolation to the yearly and hectare

scale levels led to a maximal range in the observed

results, so that the occurrence of outliers favoured the

establishment of extremes in the sample range. This

is why the results show no clear clustering of plots

according to their main characteristics, be these main

perennial crop, crop diversity or total plant stand.

Thus, there is no chronosequence or other clear

grouping order for arranging the plots; nonetheless

the set of integrated indicators generated in TAPIS

still remain applicable and meaningful.

The fact that no plot was ranked within the higher

performance quartile when both dimensions were

considered in TAPIS confirms the important tradeoffs

among concurrent indicators frequently observed in

AFS, especially for the duration of the transition

phase of the implantation of perennial crops, when

low income is simultaneous with high input demands

and work requirements (Schroth et al. 2001). These

kinds of tradeoffs were clearly observed in the

present study among, for instance, crop stand inten-

sification and income generation, given the immature

stage of development (hence low harvest) for the

most densely packed plots.

One important observation resulting from the

integration of indicators in TAPIS, however, is that

these opportunities for tradeoffs are a valuable

attribute of the studied AFS, given their high level

of crop association and diversified productive base,

which offer farmers alternatives for work dedication

and income generation, even under low external input

investment. In other words, under the low investment

and low input regime practised in the studied AFS,

work capacity is a decisive factor in management

adjustment, which drives farmers decisions regard-

ing the geometry and intensification level of their

cropping systems (Feintrenie et al. 2008).

At the same time, the significant correlation

observed in the present study between Income and

hired worker expenditures points to an important role

of AFS for the creation of local employment or

occupation opportunities in the studied group of

farmers. The significant correlation observed for

these indicators when considered for the whole

dataset means that, contrary to an expectation that

savings on inputs and workforce would lead to

improved income and hence superior combined agro-

economic performance, in reality the larger the

income (and the profit), the larger the input-buying

capacity and the need for workers.

A most evident agro-ecological attribute of the

studied plots was the very high level of cropping

association and stand intensification, with half of the

plots showing a combined plant stand at least twice as

high as their mono-cropping standard. The resulting

productive diversity is a favourable attribute provid-

ing options for farmers work dedication and produce

generation, as stressed above. As this high level of

stand intensification implies immature perennials, it

was positively related to the priority given by farmers

to keeping adventitious plants in check, and also

correlated with their controllability (Ewel 1999).

However, as shown by the low beneficial adventitious

plants indicator, farmers have not been able to

exercise selective control of adventitious plants,

failing to take advantage of valuable nitrogen-fixing

legumes (such as Desmodium sp., Mimosa invisa,

Mimosa pudica, Mucuna puriens, Pueraria phaseo-

lides) spontaneously occurring in the plots.

The establishment phase (Schroth et al. 2001)

observed in most of the plots studied, with their

immature perennials and predominance of annuals,

implied less than desirable utilization of the available

soil resources (Schroth 1999), especially in deeper

soil layers, as root distribution can be assumed to be

mostly superficial (Akinnifesi et al. 1999). Also, the

relative paucity of spontaneous tree species con-

served, or the immature stage of development

observed in many plots, does not yet fully provide

the function of capture of leached nutrients and their

return to the top of the soil as litter-fall, which is a

very important nutrient inflow for AFS under the

environmental conditions observed in the studied

region (Kanmegne et al. 2006). As they develop,

however, these trees will reinforce this function, now

mostly dependent on fruit trees and perennial crops.

With this kind of interactive indicator analysis and

interpretation, TAPIS offers farmers, extension

agents and researchers a tool for interpreting and

deciding on management options and resource allo-

cation strategies, as well as an approach for better

understanding tradeoffs in traditional agroforestry

systems. Graphic analyses of relationships between

indicators facilitate improved understanding of the


123

implications of management practices and decisions,

such as crop stand and association, inputs and

workforce allocation, and weed control for income

generation and added value. Such analyses may be

helpful for extension agents in tailoring recom-

mended management practices and resource applica-

tion strategies to the needs of individual farmers. A

strength of the TAPIS is that it can be used in new

areas by making several new plot assessments of the

16 performance indicators proposed in this study to

build a database reflecting local realities.

Acknowledgments The authors are grateful to the fieldpersonnel and local farmers for their time, personal

knowledge, dedication, and active participation in the study.

This paper is an output of a research grant from the Centre deCooperation Internationale en Recherche Agronomique pour leDeveloppement (CIRAD-France), under the ProgrammedThematic Action project Characterization and assessment of

agro-ecological performance of mixed cropping systems in the

humid tropics (ATP-Caresys). The research was carried out in a

partnership with the Institut de Recherche Agricole pour leDeveloppement (IRAD-Cameroon) and the Centre Africain deRecherche sur la Banane et le Plantain (CARBAP). We thankthe journals anonymous reviewers for their comments and

questions that helped to clarify the results and conclusions.

References

Akinnifesi FK, Kang BT, Ladipo DO (1999) Structural root

form and fine root distribution of some woody species

evaluated for agro-forestry systems. Agroforest Syst 42:

121138

Allen MA (2001) Change and continuity: land use and agri-

culture on Malo island, Vanuatu. M.Sc Thesis, The Aus-

tralian University, Canberra

Andreoli M, Tellarini V (2000) Farm sustainability evaluation:

methodology and practice. Agric Ecosyst Environ 77:

4352

Barrau J (1955) Lagriculture vivrie`re Melanesienne. Com-

mission du Pacifique Sud. Noumea, Nouvelle-Caledonie

Bockstaller C, Girardin P (2003) How to validate environ-

mental indicators. Agric Syst 76:639653

Bockstaller C, Girardin P, van der Werf HMG (1997) Use of

agro-ecological indicators for the evaluation of farming

systems. Eur J Agron 7:261270

Bonnemaison J (1996) Gens de pirogue et gens de la terre: les

fondements geographiques dune identite: larchipel du

Vanuatu, essai de geographie culturelle. Livre 1, Paris,

ORSTOM

Bosshard A (2000) A methodology and terminology of sus-

tainability assessment and its perspectives for rural plan-

ning. Agric Ecosyst Environ 77:2941

CIRADGRET, Ministe`re des Affaires etrange`res (2002)

Memento de lagronome. Societe Jouve, Paris

Duguma B, Gockowski J, Bakala J (2001) Smallholder Cacao

(Theobroma cacao Linn.) cultivation in agro-forestry

systems of West and Central Africa: challenges and

opportunities. Agroforest Syst 51:177188

Ellis EA, Bentrup G, Schoeneberger MM (2004) Computer-

based tools for decision support in agro-forestry: current

state and future needs. Agroforest Syst 61:401421

Ewel JJ (1999) Natural systems as models for the design of

sustainable systems of land use. Agroforest Syst 45:121

Feintrenie L, Enjalric F, Ollivier J (2008) Syste`mes agro-

forestiers a` base de cocotiers et cacaoyers au Vanuatu, une

diversification en adequation avec le cycle de vie familial.

In: Ruf F (ed) Ecologie et Economie de la diversification

agricole: les cultures perennes dans le monde tropical. Ed.

Quae (in press)

Girardin P, Bockstaller C, van der Werf HMG (1999) Indica-

tors, tools to evaluate the environmental impacts of

farming systems. J Sustain Agric 13(4):521

Greindl D (2000) Vivres en ville, des jardins au marche sur

larchipel du Vanuatu. PhD Dissertation, Universite Libre

de Bruxelles, Brussels

Hajkowicz S (2005) Multi-attributed environmental index

construction. Ecol Econ 57(1):122139

Hani F, Braga F, Stampfli A, Keller T, Fischer M, Porsche H

(2003) RISE, a tool for holistic sustainability assessment

at the farm level. Int Food Agribus Manage Rev 6(4):

7890

Hardi P, DeSouza-Huletey JA (2000) Issues in analyzing data

and indicators for sustainable development. Ecol Model

130:5965

Izac AMN (2003) Economic aspects of soil fertility manage-

ment and agro-forestry practices. In: Schroth G, Sinclair

FL (eds) Trees, crops and soil fertility: concepts and

research methods. CABI Publishing, Cambridge

Kanmegne J, Smaling EMA, Brussaard L, Gansop-Kouomegne

A, Boukong A (2006) Nutrient flows in smallholder pro-

duction systems in the humid forest zone of southern

Cameroon. Nutr Cycling Agroecosyst 76:233248

Kumar BM, Nair PKR (2004) The enigma of tropical home-

gardens. Agroforest Syst 61:135152

Lamanda N (2005) Caracterisation et evaluation agroecologi-

que de syste`mes de culture agroforestiers: une demarche

applique aux syste`mes de culture a` base de cocotiers

(Cocos nucifera L.) sur lle de Malo, Vanuatu. Ph.D.Dissertation, Institut National Agronomique, Paris-

Grignon

Lewandowski I, Hardtlein M, Kaltschmitt M (1999) Sustain-

able crop production: definition and methodological

approach for assessing and implementing sustainability.

Crop Sci 39:184193

Liebig MA, Miller ME, Varvel GE, Doran JW, Hanson JD

(2004) AEPAT: software for assessing agronomic and

environmental performance of management practices

in long-term agroecosystem experiments. Agron J 96:

109115

Liebman M (1988) Ecological suppression of weeds in inter-

cropping systems: a review. In: Altieri MA, Liebman M

(eds) Weed Management in Agroecosystems: ecological

Approaches. CRC Press, Boca Raton

Monteiro RC, Rodrigues GS (2006) A system of integrated

indicators for socio-environmental assessment and


123

eco-certification in agricultureAmbitec-Agro. J Technol

Manag Innov 1:4759

Morelli C (2003) Evaluation des performances agronomiques

des jardins Melanesiens: proposition dintensification par

association aux cocoteraies. Memoire de fin detudes,

Ecole Nationale Superieure Agronomique de Rennes

Nair PKR (1998) Directions in tropical agro-forestry research:

past, present, and future. Agroforest Syst 38:223245

Nair PKR (2001) Do tropical home gardens elude science, or is

it the other way around? Agroforest Syst 53:239245

Rao MR, Coe RD (1991) Measuring crop yields in on-farm

agro-forestry studies. Agroforest Syst 15:275289

Rodrigues GS (2004) Indicadores Sociais, Economicos e

Ambientais de SAFs. In: Embrapa forests: annals of the V

Brazilian congress of agro-forestry systems

Rodrigues GS, Moreira-Vinas A (2007) An environmental

impact assessment system for responsible rural production

in Uruguay. J Technol Manag Innov 2:4254

Schroth G (1999) A review of belowground interactions in

agro-forestry, focusing on mechanisms and management

options. Agroforest Syst 43:534

Schroth G, Sinclair FL (2003) Impacts of trees on the fertility

of agricultural soils. In: Schroth G, Sinclair FL (eds)

Trees, crops and soil fertility: concepts and research

methods. CABI Publishing, Cambridge

Schroth G, Lehmann J, Rodrigues MRL, Barros E, Macedo

JLV (2001) Plant-soil interactions in multistrata agro-

forestry in the humid tropics. Agroforest Syst 53:85102

Seremele TS (2002) Analyse des Syste`mes de culture racines

et tubercules: pratiques et strategies paysannes dans la

conduite des cultures et la mise en marche des produits

(Efate, Vanuatu). Memoire du CNEARC, Montpellier

Sinclair FL (1999) A general classification of agro-forestry

practice. Agroforest Syst 46:161180

Strobel F (1998) Etude pluridisciplinaire dun syste`me agraire en

transition (Wusiroro), Ile de Espiritu Santo, Archipel du

Vanuatu. Memoire de fin detudes ISTOM, Cergy Pontoise

Tollens E (2003) Estimation of production of cassava in

Bandundu (19871988) and Bas Congo (19881989)

regions by an intensive yearlong, weekly visit, production

and marketing survey, on a large sample, as compared to

official R D Congo statistics. In: Proceedings of the expert

consultation on root crop statistics, FAO 2003, Harare

Walter A, Leplaideur A (eds) (1999) Syste`me de culture tra-

ditionnels et pratiques paysannes nouvelles en Oceanie.

Rev dethnobiol 41(1):230


123

Integrated indicators for performance assessment of traditional agroforestry systems in South West CameroonAbstractIntroductionChoice of approach for indicator definition and expression

Research setting and methodologyAgro-economic dimension indicatorsAgro-ecological dimension indicators

ResultsDiscussion and conclusionsAcknowledgmentsReferences

/ColorImageDict > /JPEG2000ColorACSImageDict > /JPEG2000ColorImageDict > /AntiAliasGrayImages false /DownsampleGrayImages true /GrayImageDownsampleType /Bicubic /GrayImageResolution 150 /GrayImageDepth -1 /GrayImageDownsampleThreshold 1.50000 /EncodeGrayImages true /GrayImageFilter /DCTEncode /AutoFilterGrayImages true /GrayImageAutoFilterStrategy /JPEG /GrayACSImageDict > /GrayImageDict > /JPEG2000GrayACSImageDict > /JPEG2000GrayImageDict > /AntiAliasMonoImages false /DownsampleMonoImages true /MonoImageDownsampleType /Bicubic /MonoImageResolution 600 /MonoImageDepth -1 /MonoImageDownsampleThreshold 1.50000 /EncodeMonoImages true /MonoImageFilter /CCITTFaxEncode /MonoImageDict > /AllowPSXObjects false /PDFX1aCheck false /PDFX3Check false /PDFXCompliantPDFOnly false /PDFXNoTrimBoxError true /PDFXTrimBoxToMediaBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXSetBleedBoxToMediaBox true /PDFXBleedBoxToTrimBoxOffset [ 0.00000 0.00000 0.00000 0.00000 ] /PDFXOutputIntentProfile (None) /PDFXOutputCondition () /PDFXRegistryName (http://www.color.org?) /PDFXTrapped /False

/Description >>> setdistillerparams> setpagedevice

integrated indicators for performance assessment of traditional agroforestry systems in south west...

Documents