farmers enetic perceptions regarding their rop...

Economic Botany 55(1) pp. 106–128. 2001q 2001 by The New York Botanical Garden Press, Bronx, NY 10458-5126 U.S.A.

FARMERS’ GENETIC PERCEPTIONS REGARDING THEIR CROP

POPULATIONS: AN EXAMPLE WITH MAIZE IN THE CENTRAL

VALLEYS OF OAXACA, MEXICO1

DANIELA SOLERI AND DAVID A. CLEVELAND

Soleri, Daniela (Arid Lands Resource Sciences Program, University of Arizona, 1955 E 6thSt, Tucson, AZ 85719, USA; Current address: Center for People, Food and Environment, 340South Arboleda, Santa Barbara, CA 93110, USA and Institute for Social, Behavioral and Eco-nomic Research, University of California, Santa Barbara, CA 93106, USA, [email protected]) and David A. Cleveland (Department of Anthropology and EnvironmentalStudies Program, University of California, Santa Barbara, CA 93106-3210, USA, [email protected]). FARMERS’ GENETIC PERCEPTIONS REGARDING THEIR CROP POPULATIONS: AN

EXAMPLE WITH MAIZE IN THE CENTRAL VALLEYS OF OAXACA, MEXICO. Economic Botany 55(1):106–128, 2001. Collaborative plant breeding is an approach to crop improvement that includesclose attention to specific adaptation and interaction between farmers and formal plant breedersto better meet the needs of those farmers. Collegial interaction capable of making best use ofthe knowledge and skills of farmers and breeders will depend upon an understanding of thosein terms that are relevant to each. To facilitate this interaction with the goal of making farmerselection practices more effective, the work described here sought to improve outside research-ers’ understanding of farmers’ fundamental perceptions about their populations, growing en-vironments, and expectations for response to selection. Various methods were used to accom-plish this with a small sample of maize farmers in two communities in the Central Valleys ofOaxaca, Mexico. Farmers’ decisions about maize varietal type repertoires imply assessmentsbased on genetic and environmental variation in the local context. A clear distinction was madebetween traits of high and low heritability and expected response to selection, however, sometraits of interest to farmers such as large seed size may involve considerations other than theirpotential for expression in the progeny generation.

OPINIONES GENETICAS DE LOS GRANJEROS CON RESPECTO A SUS POBLACIONES DE LA COSECHA: UN

EJEMPLO CON MAıZ EN LOS VALLES CENTRALES DE OAXACA, MEXICO. El fitomejoramiento cola-borativo es una forma de mejora de las plantas, que presta especial atencion, a la adaptacionespecıfica y la interaccion entre agricultores y fitomejoradores para un mejor respuesta a lasnecesidades de los primeros. Lo que facilita la interaccion entre agricultores y fitomejoradores,pretendiendo que la seleccion de los agricultores sea mas eficiente. El trabajo describe una viapara el mejor entendimiento de los investigadores en relacion a las percepciones fundamentalesde los agricultores respecto a sus poblaciones cultivadas, sus ambientes de cultivo y sus expec-tativas en relacion con la respuesta a la seleccion. Varios metodos fueron aplicados a unapequena muestra de agricultores en dos comunidades en los Valles Centrales de Oaxaca, Mexico.La decision de los agricultores de escoger sus variedades esta basada en la variacion geneticay ambiental a nivel local. Una clara distincion fue hecha por los agricultores entre los caracteresde alta y baja heredabilidad ası como la respuesta a la seleccion; pero, los resultados sugierenque algunos caracteres de interes para los agricultores como el tamano del grano son importantescomo criterio de seleccion, aun cuando no lo asocian con un efecto genetico.

Key Words: collaborative/participatory plant breeding; farmers’ knowledge; genetic percep-tions; maize; heritability; crop varietal classification; Zea mays; Mexico.

Collaborative plant breeding (CPB) is a relative-ly new approach to crop improvement (CGIAR1997; Eyzaguirre and Iwanaga 1996; Witcombe

1 Received 22 September 1999; accepted 5 Septem-ber 2000.

et al. 1996). Though still under development,CPB involves some form of interaction betweenfarmer-breeders and professional, formallytrained plant breeders (hereafter, farmers andplant breeders, respectively) in crop improve-ment for local use (Cleveland, Soleri, and Smith

2001] 107SOLERI & CLEVELAND: FARMERS’ GENETIC PERCEPTIONS REGARDING MAIZE

2000; Smale et al. 1998; Sperling and Loevin-sohn 1996). In the context of CPB, a broad def-inition of plant breeding is emphasized that in-cludes varietal choice, seed selection, and seedprocurement (Cleveland, Soleri, and Smith2000; CPRO-DLO and WAU 1999). CPB isseen as particularly relevant for agricultural sys-tems in marginal growing environments, andthere is frequently an emphasis on specific ad-aptation to local biophysical and socioculturalfactors (Ceccarelli, Grando, and Booth 1996;Sperling and Scheidegger 1995; Weltzien et al.1998). Marginal environments are characterizedby multiple stress factors with high variance, re-sulting in low average and variable productivity(Falconer 1989). Crop varieties planted in theseenvironments are frequently farmers’ varieties(FVs), farmer managed and selected populationsthat include landraces, and locally adapted prog-eny from crosses between landraces and modernvarieties. FVs are usually assumed to have rel-atively narrow geographical adaptation to mar-ginal growing environments, and high yield sta-bility (low variance across environments includ-ing years and locations) and moderate yield inthose environments (Harlan 1992).

One form of CPB, and likely one of the mostuniversally accessible, will be based on modestrevisions of existing local selection methods toimprove farmers’ own crop populations, part ofwhat has been referred to as ‘‘farmer-led’’ CPB(McGuire, Manicad, and Sperling 1999). Notonly can this approach be economical, it has thepotential to strengthen the extant selection pro-cess because a) it is based in the farming com-munity and is less dependent on external expertsand resources, b) the target environment (wherepopulations will ultimately be grown and used)and selection environments are the same, thuseliminating any problem with lack of correlationbetween performance in selection plots andfarmers’ fields (Atlin and Frey 1989; Ceccarelli1989), and c) it uses as its starting point croppopulations that are locally adapted through ahistory of selection under local conditions (Cec-carelli, Grando, and Impiglia 1998).

The limited number of CPB efforts that in-clude changes in farmers’ selection practicestypically involve a transfer of technology ap-proach, especially regarding biological compo-nents (McGuire, Manicad, and Sperling 1999).For example, a project with maize farmers inHonduras attempted to teach basic genetics and

techniques such as in-field plant selection, de-tasseling, and bagging to control cross pollina-tion (Gomez et al. 1995).

However, it seems likely that understandingfarmers’ plant breeding in terms of the same the-oretical principles that underlie professionalplant breeding would enhance the probability forsuccessful CPB both in terms of empowering lo-cal farmers socially, and of making significantbiological progress in plant breeding. While anunderstanding of farmers’ plant breeding interms of professional plant breeding is of obvi-ous importance for CPB, most research on farm-ers’ plant breeding has focused on listings andclassifications of farmers’ varietal repertoires oron their stated selection criteria (e.g., Soleri andCleveland 1993). One exception is Louette andSmale’s work in Jalisco, Mexico (1998). Theyfound that farmers’ selection served primarily tomaintain broad varietal phenotypes for ear char-acteristics, but that farmers did not undertake se-lection with the idea of changing their maizepopulations.

Not only will CPB likely require greater un-derstanding of farmers’ plant breeding, butmethodological adjustments and innovations inprofessional plant breeding as well (CPRO-DLOand WAU 1999). Professional plant breedinghas typically emphasized the development ofmodern varieties (MVs) with geographicallywide adaptation to optimal (relatively low stress,and uniform) growing environments, and highyield in these environments (Evans 1993; Sim-monds 1979). While there has also been atten-tion to breeding for stress tolerance, this atten-tion has focused on relatively large-scale envi-ronments and commercial farmers who can af-ford to purchase seed and other inputs, not onthe farmers who are the topic of this paper (Ban-ziger, Edmeades, and Lafitte 1999; Heisey andEdmeades 1999). Thus, CPB will need to ad-dress the complexities of interdisciplinary (bothwithin and between the social and biological sci-ences) and intercultural (farmers and research-ers) collaboration, and of plant improvement invariable, stressful environments.

The research we report here is situated withinthe tradition of research on local (or indigenous)biological knowledge in terms of the nature oflocal knowledge, the conceptual level of thisknowledge, and the purpose of understanding it.First, in terms of the basic views of local knowl-edge within ethnobiology we take an alternative

108 [VOL. 55ECONOMIC BOTANY

approach, a middle ground, between utilitarianand intellectualist views (Berlin 1992, Medinand Atran 1999). The utilitarian view is a rela-tivist one, that local knowledge depends on thegoals, theories and beliefs of the local people.As Berlin notes, the utilitarian tradition is oftendominated by economic concerns, or descrip-tions of uses, and this continues to be a strongtradition in economic botany and zoology. Theintellectualist or comparativist tradition suggeststhat categories are recognized rather than con-structed because nature itself is made up of anorganized pattern of units, and there are univer-sals in human cognition, resulting in cross-cul-tural similarities in the ways in which humansview or conceive biological organisms. A moreinductive alternative to these two common ap-proaches is one that unites them by seeking toexplain both similarities and differences betweenlocal groups, including comparisons between lo-cal and scientific knowledge (Medin and Atran1999).

Second, in terms of the level of knowledge,our focus is different than that of most studiesof local knowledge of the biological environ-ment, which have been on classification. Thisstudy contributes to a growing area of researchon understanding more complex local biologicalknowledge systems, of which classification is animportant part, and on the relationship betweenknowledge and practice. For example, Ellen(1999) concluded on the basis of his own andothers’ research on subsistence of rain forestpeoples, that knowledge of general principlesforms the basis for deductive models that func-tion, for example, in connecting observations atthe species level with forest structure and dy-namics. Our focus is on local knowledge aboutcomplex functional relationships between organ-isms and their environments, namely the inter-actions between genotype and environment thatdetermine phenotype in crop plants, and that af-fect varietal classification systems and farmers’plant breeding strategies.

Third, our goal of understanding similaritiesand differences between local and scientificknowledge is a practical one. Thus we are in-terested not only in the question of the extent towhich local and scientific knowledge are similaror different, but also in how each can contributeto achieving farmers’ and plant breeders’ goals.

The research reported here is also based ontwo assumptions. First, the biological model of

plant breeding described below is a valid rep-resentation of biological reality. Second, in CPBthe knowledge and practice of both farmers andbreeders are important, and neither should be as-sumed to be better a priori—their relative meritsin terms of contribution to CPB need to be em-pirically assessed in each situation. This is thereason we have suggested collaborative plantbreeding (CPB) as an alternative to participatoryplant breeding (PPB) (Cleveland and Soleri1997; Cleveland, Soleri, and Smith 2000).

This paper reports results of a small casestudy we carried out with farmers in two com-munities in the Central Valleys of Oaxaca, Mex-ico. We begin by very briefly describing theplace of maize (Zea mays L.) FVs in Mexicoand the study region, then use a basic biologicalmodel to outline relevant points regarding theinteraction of genetic and environmental varia-tion in crop populations, focusing on heritability.Our methods included traditional ones of socialscience, as well as a new method that uses hy-pothetical scenarios based on the biologicalmodel and farmers’ own experiences to explorefarmers’ genetic perceptions—their knowledgeof genetic variation and its relation to environ-mental variation—in terms of two componentsof farmers’ plant breeding. First, farmers’ per-ceptions of intervarietal differences throughtheir descriptions of varieties of the most widelysown class of maize and our measurements oftheir maize populations, and the implications ofthose designations in terms of genetic and en-vironmental variation. Second, farmers’ percep-tions of heritability for two different traits andcorrelation with plant development for one traitwithin their maize populations, through their cri-teria for seed selection, and responses to hypo-thetical scenarios.

For both of these components we asked threefundamental questions: (1) What methodologiescan outside researchers use to understand farm-ers’ plant breeding? (2) What is the nature offarmers’ plant breeding knowledge in relation tothe biological model of professional plant breed-ers? (3) Can this knowledge inform and improveCPB? The findings suggest both similarities anddifferences between the knowledge of farmersand plant breeders that have important implica-tions for CPB.

MAIZE IN MEXICO

Mexico, the center of maize domesticationand diversity, is also the home of the green rev-


olution approach to developing MVs of wheatand maize for increased yield and production. Inthe Third World this approach is characterizedby the application of industrial agriculture basedon MVs and high levels of inputs in the bettergrowing environments. In spite of the release of222 MVs of maize (104 open-pollinated varie-ties, 118 hybrids) between 1966–1997 by theMexican public sector, and 155 private-sectorMVs (5 open-pollinated varieties, 150 hybrids)being available on the market by 1997, approx-imately 80% of the total maize area in 1996 (6.3million ha) was in FVs (compared to 52% forall of Latin America) (Morris and Lopez-Pereira1999).

As with other major grain crops, high yieldingmaize MVs (as compared to FVs) have beenbred for relatively optimal (fairly uniform, lowstress) environments across wide geographic ar-eas, and are relatively lacking in genetic diver-sity—limited work has been done on breedingfor the more stress-prone environments of manysmall-scale farmers where yields are relativelylow (Heisey and Edmeades 1999; Smith andPaliwal 1997). As discussed below, availabledata suggest that most maize MVs are not ap-propriate for small-scale farmers in Mexico,who are responsible for the majority of landplanted to this crop (Garcıa Barrios and GarcıaBarrios 1994).

The low adoption rate for MVs is likely duein part to the highly stress-prone growing con-ditions (drought stress, low fertility, lack of in-puts such as irrigation and fertilizers) of small-scale farmers in Mexico. Many MVs have notbeen targeted for such conditions, and thus pro-duce lower yields than FVs (Aquino 1998; Heis-ey et al. 1998). Even if MVs performed accept-ably in these farmers’ fields, provided they sup-plied the additional inputs that the large-scalecommercial farmers do (commercial fertilizers,pesticides, irrigation), low resource farmers can-not afford to supply these inputs. Yet small-scaleMexican farmers are often assumed to be ‘‘onlydimly aware of the potential benefits of im-proved germplasm and crop management prac-tices,’’ and lacking the education and skillsneeded to manage MVs ‘‘properly’’ (Aquino1998:249), although no data are usually provid-ed to support such statements. To the contrary,the few studies that have been carried out sug-gest that small-scale farmers make decisions inallocating resources to maize MVs v. FVs based

on rational comparisons of performance, includ-ing yield stability (e.g., Perales, Brush, andQualset 1998; Smale, Heisey, and Leathers1995).

While it is difficult to compare yields for FVsand MVs at a regional or national level becausemost data are not disaggregated, some relevantsurvey data do exist. For example, disaggregateddata from 1990 and 1994 surveys of 275 ejidos(communities managing land in common) showthat in the spring-summer season with irrigationand fertilizer, MVs (‘‘improved seeds’’) yieldedmore (2.36 t/ha in 1990, 1.85 in 1994) than FVs(‘‘local seeds’’) with irrigation and fertilizer(1.36 t/ha in 1990, 1.37 in 1994), and yieldedmuch more than local seed that were not fertil-ized or irrigated (0.84 t/ha in 1990 and 0.80 in1994) (de Janvry, Gordillo, and Sadoulet 1997:72). However, improved maize seed was usedby only 4.6% (1990) and 15.5% (1994) of allfarms, and only 3.0% and 7.8% of farms under2 ha, which were almost entirely rain fed, andaccounted for 28.8% of all farms (de Janvry,Gordillo, and Sadoulet 1997:78, 82, 32). Thesedata support the observation that the adoption ofmaize MVs has generally been limited to bettergrowing environments (Heisey and Edmeades1999).

In the southern Mexican state of Oaxaca,92.7% of maize area harvested in 1990 was inFVs, with the remainder in MVs (5.5% in open-pollinated varieties, 1.8% in hybrids) (AragonCuevas 1995). Grain yields in Oaxaca duringthis period were only 0.8 t/ha (INEGI 1996:32),40% of the yield for Mexico as a whole, and20% of the world average. In addition there arehigh rates of emigration from farm communities,and agriculture may often be associated withdegradation of soil and vegetation (M. Reespers. comm. 1998; Stephen 1991).

THE BIOLOGICAL MODEL:HERITABILITY

A central theme of biological scientific theoryand practice, including plant breeding, is the rel-ative contributions of nature (genetic variation,VG) and nurture (nongenetic or environmentalvariation, VE) to individual phenotypes. Varia-tion in population phenotype (VP), on whichchoice and selection are based, is determined bygenetic variation (VG), environmental variation(VE), and variation in genotype-by-environment(G 3 E) interaction (VG3E), (VP 5 VG 1 VE 1


VG3E). VG3E represents the degree to which ge-notypes behave consistently across a number ofenvironments. Environmental variation can bepartitioned into several components: VE 5 VL 1VT 1 VM (VL 5 variance due to location, e.g.,soil and climatic variables; VT 5 variance dueto time, e.g., season or year; and VM 5 variancedue to breeder or farmer management). Lowquantitative G 3 E means relatively little changein performance over environments. High quan-titative G 3 E is characterized by marked chang-es in performance with changes in environmen-tal factors and is associated with reduced stabil-ity of performance (defined as variance acrossenvironments) of an individual genotype. Qual-itative G 3 E between two or more varietiesmeans that they change rank across environ-ments, and this is often referred to as a crossoverbecause the regression lines for yield (or othertraits) cross over at some point.

Broad sense heritability (H) is the proportionof VP due to genetic variance (VG/VP), while nar-row sense heritability (h2) is the proportion ofVP due to additive genetic variance (VA/VP), thatis, the proportion of VG directly transmissiblefrom parents to progeny, and therefore of pri-mary interest in plant breeding. Heritabilities in-fluence not only selection, but also choice ofgerm plasm for planting or crossing. Traits withhigh average heritability vary less with variationin the environment than traits with low averageheritability, and therefore high heritability traitsare theoretically easier to use in classification ofcrop genotypes.

Estimating the heritability of traits in partic-ular environments and populations is of centralinterest to plant breeders (Nyquist 1991; Sim-monds 1979). Traditionally, estimates have re-lied upon methods that attempt to characterizeenvironmental variation through research designand the use of genetically defined materials, e.g.,clones or families of full or half siblings. How-ever, as increasing attention is being given tomore marginal environments, interest in herita-bility estimates for these environments is grow-ing (e.g., Bolanos and Edmeades 1996; Ceccar-elli 1996). Consideration of heritability withinthe context of CPB requires attention to the po-tential for specific adaptation and farmer-breederinteraction. Attention to specific adaptationmeans heritability estimates made under the con-ditions experienced by farmers and their croppopulations (Soleri and Smith n.d.). Attention to

interaction suggests enhancing communicationand understanding of researcher and farmer per-spectives (e.g., Dhamotharan et al. 1997), for ex-ample by trying to understand the extent towhich farmer plant breeding (knowledge, prac-tice and genotypes, and environments) can beunderstood in terms of the basic biological mod-el of plant breeding (Cleveland, Soleri, andSmith 2000).

MATERIALS AND METHODS

This research was undertaken as part of astudy of local maize populations and farmer se-lection in two communities in the Central Val-leys of Oaxaca, Mexico (Soleri 1999). Com-munities in this area are predominantly eitherindigenous Zapotec, Mestizo, or a mix of thesetwo (INEGI 1993:35). While off-farm work isincreasingly important in this region, includingtemporary migration within Mexico and to theUSA (M. Rees pers. comm. 1997; Stephen1994), subsistence agriculture is still predomi-nant and maize production is the foundation ofmost rural households’ economy (INEGI 1993:73). Eighty-eight percent of summer maize pro-duction in the Central Valleys is under rain-fedconditions with most households experiencingharvest failure about one of every four years(Dilley 1993:114). We worked with eight farmfamilies in Santa Maria (pseudonyms are usedfor communities throughout), a community inthe Zimatlan Valley, and with five families inSan Antonio, a community in the Mitla Valley(Table 1). Households were initially selected forparticipation in another component of this re-search concerning quantitative description oftheir crop populations (Soleri and Smith n.d.).Some households were identified through rec-ommendations of fellow community membersand municipal authorities as households knownto be managing diverse maize varieties orknown as respected maize farmers (e.g., hard-working, not implying large-scale). Others werechosen during walking tours of fields in the 1996spring planting season. The sample containedrepresentatives of the main household types ineach of the communities, based on the two mostimportant distinguishing characteristics: genderof household head and wealth. Interviews wereconducted with individuals primarily responsiblefor agriculture, typically a wife and husband, ormother and son, and younger workers who usu-ally deferred to the primary pair. When we use


TABLE 1. CHARACTERISTICS OF STUDY COMMUNITIES IN THE CENTRAL VALLEYS OF OAXACA, MEXICO.

Characteristic Santa Maria San Antonio

Elevation (meters above sea level)a

Average annual precipitation (mm)b

Predominant soil characteristicsc

District average maize yield (t/ha)a

Average sowing rate (seed/ha)d

Population (1995)a,e

Predominant ethnic/linguistic groupf

1490685alluvial, sandy clay0.7647 0002800Mestizo/Spanish

1780468piedmont, gravel0.4540 0002533Zapotec/Zapotec

a INEGI 1996.b Dilley 1993.c Kirkby 1973.d Based on field observations, Soleri 1996–1997.e 1998 estimates for both communities 5 3000, M. Rees personal communication 1998.f INEGI 1993.

the word ‘‘farmer’’ in the rest of this paper, itrefers to a farm household, unless otherwise in-dicated.

The larger study that this work was part ofwas conducted from June–December 1996,June–December 1997, and June–August and Oc-tober 1998, with data collected in Spanishthrough participant observation, informal dis-cussions, formal interviews, and on-farm andexperimental plot research. Questions regardingfarmers’ varietal choices were administered dur-ing formal interviews in 1996 and 1997. Com-munity-level comparisons between farmer esti-mates of maize cycle length (days from sowingto anthesis and harvest) were analyzed using theMann-Whitney one-tailed test of medians withsignificance at P # 0.05.

In this research we investigated farmers’ prac-tices (varietal classification, seed selection) andfarmers’ genetic perceptions. Farmer identifica-tion of ear phenotypes they associate with whitemaize varieties was accomplished using a ran-dom sample of 100 ears from a plot in a farmer’sfield (measured as part of the larger study, Soleri1999) in their community. Ten of each of thewhite maize varieties recognized in their com-munity were identified by farmers. To comparethe two farmer-identified varieties as representedby the 10 ear samples, we analyzed 15 morpho-phenological plant and post-harvest (seed andear) traits using orthogonal contrasts with sig-nificance at P # 0.05 (Soleri, Smith, and Cleve-land n.d.). All traits were measured on an indi-vidual plant basis and included traits measuredpost anthesis in the field: ear height, total plantheight, stalk diameter, ear leaf width, ear leaflength, ear leaf area (width 3 length 3 0.75,

Lafitte and Edmeades 1994), number of primarytassel branches, anthesis silking interval (daysbetween initiation of pollen shed and first silkemergence), days to anthesis from sowing; andtraits measured post harvest: ear diameter, earlength, kernel row number, grain yield, weightof 100 grains, and shelling ratio (grain weight/ear weight).

Farmers’ perceptions regarding heritabilitiesin their populations and environments were as-sessed during formal interviews in 1996, 1997and 1998. We used hypothetical scenarios re-garding traits that typically have high (tassel col-or) or low (ear length) average heritabilities, andtheir expression in both a variable, stressful, typ-ical field of the region and in a hypothetical op-timal field, one that is uniform and in no waylimits plant growth (Table 2). Using the sametwo environmental types, we also asked aboutthe effect of seed size on plant development.Here ‘‘seed’’ refers specifically to maize grain(also called kernels) used by farmers for plant-ing. These scenarios built on farmers’ experi-ence, but also presented some situations unfa-miliar to them, for example an optimal fieldwithout resources limiting plant growth. Ourquestions about the expression of traits in typicaland optimal environments were designed to cre-ate a contrast in the variability present in thegrowing environment and the opportunity to dis-cuss the interaction of this and VG.

RESULTS

Our objectives were to test both a methodo-logical approach for understanding how farmersperceive of abstract concepts such as heritabilityin their maize varieties, and to test hypotheses


TABLE 2. SUMMARY OF FARMER PERCEPTIONS OF HERITABILITY AND GENETIC VARIANCE FOR TWO TRAITS

AND CORRELATION WITH PLANT DEVELOPMENT FOR ONE TRAIT IN MAIZE.

Phenotypic traitselected for

Question based onbiological model

Scenario presentedto farmers

Hypotheses regarding farmers’perceptions

Tassel color How do farmers perceive the h2 ofa trait with high average h2 inenvironments with high and lowVE?

What would be variation in tasselcolor if only seeds from plantswith preferred tassel colorplanted?

Null: low h2

Alternative: high h2

Ear length How do farmers perceive the h2

and VG of a trait with medium/low average h2 in environmentswith high and low VE?

What would be variation in earlength if only seeds fromplants with long ears planted?

Null: low h2

Alternative: high h2

Seed size How do farmers perceive the phe-notypic correlation betweentraits with low heritability (seedsize, plant development) in en-vironments with high and lowVE? That is, would the pheno-typic correlation between seed(endosperm) size and plant de-velopment (VG3E where E 5 in-ternal environment, i.e. endo-sperm) be overridden by fieldvariability (VG3E where E 5field)?

What would be the variation inemergence, seedling size/vigorand grain yield when onlylarge seeds planted?

Null: no phenotypiccorrelation be-tween seed sizeand developmentof the plant

Alternative: pheno-typic correlationbetween seed sizeand developmentof the plant

TABLE 3. MAIZE VARIETY CLASSIFICATION IN TWO

COMMUNITIES IN THE CENTRAL VALLEYS OF OA-XACA, MEXICO.

Name andcolor class

Varietal classificationstated by farmers

Graintype

Cyclelengtha

Community

SantaMaria

SanAntonio

blanco (white) cuadradobolita

——

——

tardonviolento

uu

uu

amarillo (yellow)

negrito (black)

belatove (purple)

—————

tardonviolentoviolento

—violento

uuu

uuu

u

a Determined by farmers’ statements about both days from planting toanthesis and from planting to harvest.

about the similarities and differences betweenfarmers’ perceptions and those of plant breedersin terms of the biological model. These hypoth-eses were designed to provide insights into thenature of farmer knowledge underlying practicesof particular relevance to CPB, for example, thepossibility and nature of a theoretical basis forfarmer knowledge and practices.

VG, VE, AND VARIETAL CLASSIFICATION

Although Santa Maria and San Antonio areonly approximately 65 km apart and both com-munities have good access to major markets,there is a distinct difference in farmers’ namingpractices regarding their white maize varieties inthe two communities (Table 3). In both locationsblanco criollo (local white) is the primary classof maize cultivated. In Santa Maria varieties ofblanco are categorized solely based on featuresobserved in the ear post harvest—particularlykernel/ear type (cuadrado v. bolita), as well aspigmentation of the cob and husk. In contrast,varieties of blanco in San Antonio are catego-rized on the basis of their cycle length (tardon[long cycle] v. violento [short cycle]), as mea-sured by days to anthesis and harvest.

To obtain farmers’ estimates of cycle length,we asked them to tell us the time from plantingto flowering and to maturity for the blanco va-rieties they had experience with. Farmers re-sponded in terms of months and fractions ofmonths, sometimes adjusting this in terms ofweeks or days (Table 4). Farmers in Santa Mariaperceived bolita and cuadrado as having differ-ent cycle lengths, but they disagreed about


TABLE 4. FARMERS’ ESTIMATES OF CYCLE LENGTH FOR BLANCO CRIOLLO MAIZE VARIETIES.

Days from planting until:

Anthesis Ready to harvest

Santa Maria varietiesMeanStandard deviationMedian

Bolita (n 5 8)671060

Cuadrado (n 5 7)69

869

Bolita128

15120

Cuadrado134

22134

San Antonio varietiesMeanStandard deviationMedian

Violento (n 5 5)681360a

Tardon (n 5 5)811390a

Violento105

11105a

Tardon144

8150a

a One-tailed Mann Whitney test of medians, significant at P # 0.05.

which was slower or faster, and the differencesbetween cycle length estimates for the two va-rieties were insignificant. In addition, the lack ofagreement regarding correlation between thebolita and cuadrado ear phenotypes and cyclelength does not support the hypothesis that theseear types are indirect selection criteria for cyclelength. In San Antonio, with a more stressfulgrowing environment due to lower rainfall andpoorer soils, the difference in farmer-declaredcycle lengths was significant. Although farmerswe spoke with in Santa Maria preferred the bol-ita phenotype, all stated that they grew mixedpopulations containing both varieties, and thatkeeping them separate was impossible becauseof cross-pollination.

Hibrido (hybrid) blanco is the only blancomaize currently distinguished by cycle length inSanta Maria. One household we worked with inthat community grew it occasionally and pur-chased seed from suppliers in Oaxaca City,though a shop owner in Santa Maria also soldhibrido seed. Hibrido was universally known inboth communities for its long cycle and greaterwater requirements as compared to local blancovarieties, and as a maize of foreign origin is notconsidered a criollo variety and was not includ-ed in this study. Color classes of maize otherthan blanco are grown in both communities—amarillo, negrito, and belatove in San Antonio,amarillo and negrito in Santa Maria. However,in both communities the non-blanco classes aretypically assumed to be of short cycle lengthwith the exception of long cycle amarillo pop-ulations reported in San Antonio and negrito ofcomparable cycle length as the blanco popula-tions in Santa Maria (Table 3). None of the non-blanco color classes were described by farmers

in either community as having consistent vari-ants for ear or kernel type as was true of blancoin Santa Maria.

In addition to asking farmers what traits char-acterize their blanco maize varieties, we alsoasked farmers to identify ears belonging to dif-ferent varieties from among the 100 ear sampleswe presented to them (Table 5). Of the 15 mor-phophenological traits measured on these earsand the plants that produced them, some within-community patterns are suggested by farmers’identification, though they were not always con-sistent and are not conclusive. Although only afew (n 5 3) of the identification exercises withfarmers from San Antonio used samples forwhich cycle length data were available, none ofthose comparisons showed significant differenc-es between farmer identified violento and tardongroups of individual plants, as represented bytheir ears, for days to anthesis. Of all of the iden-tification exercises conducted with farminghouseholds in that community (n 5 8 house-holds), six had at least one significant trait con-trast between varietal groups. Of the total of 12significant trait contrasts from that community,nine (75%) portrayed violento plants/ears/ker-nels as being smaller than those of tardon. Thesignificant contrast for kernel row numbershowed violento with a higher row number.However, the most common significantly differ-ent trait (n 5 3) from the exercises in that com-munity was ear diameter, where results of twoidentification exercises showed a violento groupwith means greater than the tardon, while theother exercise had the opposite result.

Of a total of eleven identification exercisesconducted in Santa Maria, seven contained va-rietal identifications with significantly different


TA

BL

E5.

SIG

NIF

ICA

NT

OR

TH

OG

ON

AL

CO

NT

RA

ST

Sa

BE

TW

EE

NB

LA

NC

OC

RIO

LL

OV

AR

IET

IES

AS

RE

PR

ES

EN

TE

DB

YF

AR

ME

RS’

CH

OIC

ES

INID

EN

TIF

ICA

TIO

NE

XE

RC

ISE

FR

OM

100

EA

RR

AN

DO

MS

AM

PL

Eb .

Com

mun

ity&

exer

cise

no.

Trai

tm

eans

for

farm

er-i

dent

ified

vari

etal

repr

esen

tativ

es,s

igni

fican

tco

ntra

sts

only

Tota

lpl

ant

heig

ht(c

m)

Flag

leaf

wid

th(c

m)

Flag

leaf

leng

th(c

m)

18ta

ssel

bran

ches

(cou

nt)

Stal

kdi

amet

er(m

m)

Ear

diam

eter

(mm

)

Ear

leng

th(c

m)

Ker

nel

row

num

ber

(cou

nt)

Gra

inw

t.(g

m)

100

grai

nw

eigh

t(g

m)

Shel

ling

ratio

(gra

inw

t/ear

wt)

SA

nton

ioc

1 2 3 4 5 68.

1/9.

2

86.3

/97.

5

67.5

/76.

8

4.8/

4.3

4.8/

4.4

4.2/

5.0

12.9

/15.

4

12.4

/13.

8

11.2

/9.8

77.7

/125

.0

28.9

/38.

2

32.5

/44.

9S

taM

aria

c

1 2 3 4 5 6 7

266.

9/24

3.4

296.

6/26

4.7

8.0/

10.2

82.5

/73.

6

11.3

/16.

3

1.7/

2.0

4.3/

4.7

10.6

/12.

9

9.2/

11.4

65.5

/95.

5

92.6

/109

.4

0.82

/0.8

5

0.85

/0.8

8

0.84

/0.8

6

aSi

gnifi

canc

eat

P#

0.05

.b

Bas

edon

iden

tifica

tion

of10

ears

tore

pres

ent

each

vari

ety.

Mea

nsof

vari

etie

sid

entifi

edby

farm

ers

inSa

nA

nton

io:

viol

ento

/tar

don.

Mea

nsof

vari

etie

sid

entifi

edby

farm

ers

inSa

nta

Mar

ia;

boli

ta/c

uadr

ado.

cTo

tal

num

ber

ofid

entifi

catio

nex

erci

ses

cond

ucte

din

San

Ant

onio

:n

58,

inSa

nta

Mar

ia:

n5

11.


means for one or more traits (n 5 14 significanttrait contrasts). Of these significant contrasts,79% (n 5 11) represented bolita as havingsmaller plant or ear characteristics as comparedto cuadrado. In the comparison of farmer-iden-tified varieties in this community, the most fre-quent (n 5 3 households) significantly differenttrait was shelling ratio.

HERITABILITY, SEED SIZE, AND

INTRAPOPULATION SELECTION

The scenarios presented to farmers made useof traits with high and low average heritabilitythat were familiar and of interest to them, aswell as both familiar and unfamiliar growing en-vironments (see Table 2 above). Our purposewas to create hypothetical situations to facilitatediscussion of the abstract concept of heritability.Scenarios regarding seed size were intended toclarify farmers’ perceptions of the significanceof that trait in selection.

Tassel Color. Tassel, glume, and anther color(including yellow, red and purple), is an aes-thetic trait that farmers in both communitiespointed out to us. The pleasure of looking acrossa field of green plants with purple tassels wasthe reason one household in San Antonio soughtout a yellow maize population known to havetassels of that color. In Santa Maria a householdwas growing a bolita blanco population recentlydeveloped by a family member to have predom-inantly purple tassels, cobs, and husks; purplehusks transfer their color to tamales steamed inthem, a desired effect. Tassel color is highly her-itable and as such is among the pigmentationtraits used to identify genotypes in experimentalresearch (Coe, Neuffer, and Hoisington 1988:135).

These scenarios were designed to improve un-derstanding of how farmers perceive the influ-ence of VG and VE on expression of tassel color.The potential role of VG was represented by therelationship between phenotypes of maternaland progeny generations. The potential role ofVE was represented by the contrasting growingenvironments. The null hypothesis was thatfarmers see a relatively small contribution by VG

to total VP—low heritability—saying that seedsfrom plants with a given tassel color would pro-duce plants with a diversity of tassel colorswhen planted in a typical environment, andmostly tassels of the given color when plantedin an optimal environment, attributing VP pre-

dominantly to VE and VG3E. The alternative hy-pothesis was that farmers see tassel color pri-marily determined by VG, that the tassel color ofthe progeny plant would be the same as that ofthe parent regardless of the environment. Ourhypotheses did not include the effects of the pol-len parent or of segregation in the formation ofprogeny phenotypes, although some farmers didmention this.

Using photographs from a local population ofmaize that included plants with both purple andyellow tassels, we asked farmers what tassel col-or would result if seed were only taken fromplants with purple tassels and those seed wereplanted in 1) a typical field, and, 2) an optimalfield (Fig. 1). The majority of responses to thesescenarios stated that tassel color would be purplein either field; that is, it will not be affected bythe growing environment (Table 6). The remain-der stated that there would be a mixture of col-ors, and that after five years of isolation fromcross pollination with other populations and con-tinued selection for that color, the populationwould have all purple tassels.

Ear Length. Ear length is one of farmers’ cen-tral selection criteria in both communities (So-leri, Smith, and Cleveland n.d.), and is a traitwith medium to low average heritability (,0.50)(Hallauer and Miranda 1988). Again, our sce-narios were designed to elicit farmers’ percep-tions about the relative influence of genetic andenvironmental sources of variation on VP of thistrait. The null hypothesis was that farmers see arelatively small contribution by VG to total VP,saying seeds from long ears would produceplants with a diversity of ear lengths when plant-ed in a typical field and mostly long ears whenplanted in an optimal field. The alternative hy-pothesis was that farmers see ear length primar-ily determined by VG, with progeny phenotypefor the most part the same as that of the parent,regardless of the environment. As with tasselcolor, our hypotheses did not include the effectsof the pollen parent or of segregation in progenyphenotypes although these were noted by somefarmers.

We asked what would be the length of the earsproduced in a typical field as compared to thoseproduced in a optimal field, if they planted onlyseed from the long ears from a typical harvestof variable sized ears (Fig. 2). The farmers stat-ed that the typical field would produce a harvestof variable ear lengths while the harvest from


Fig. 1. Genetic perceptions: Responses to tassel color scenario. (P 5 purple, Y 5 yellow).

the optimal field would consist of uniformlylong ears, and gave environmental reasons whenasked why this would occur (Table 7). Onefarmer noted that there would always be somevariation present in any environment.

Seed Size. Some studies have found a signif-icant relationship between seed size and earlyplant growth (e.g., Revilla et al. 1999). Partici-pant observation, discussions and formal inter-

views during our first field season (1996) clearlyindicated that, as with ear length, farmers foundlarge seed size desirable but completely depen-dent on the environment in which the plantgrew. To avoid questions that would seem re-dundant to the farmers, we did not present sce-narios regarding heritability for seed size as wedid for tassel color and ear length. Instead, wewanted to ascertain why large seed size was


TA

BL

E6.

FA

RM

ER

S’

RE

SP

ON

SE

ST

OG

EN

ET

ICP

ER

CE

PT

ION

SS

CE

NA

RIO

SR

EG

AR

DIN

GH

ER

ITA

BIL

ITY

AN

DG

EN

ET

ICV

AR

IAN

CE

OF

TA

SS

EL

CO

LO

R.

Sce

nari

o/qu

esti

onN

o.of

hous

ehol

dsR

espo

nse

Tot

alS

ta.

Mar

iaS

.A

nton

io

Wha

tw

ill

beth

eta

ssel

colo

rof

plan

tsgr

own

from

seed

sele

cted

only

from

seed

pare

nts

wit

hon

eta

s-se

lco

lor

(eit

her

purp

leor

yell

ow),

whe

ngr

own

ina(

n):

Typ

ical

fiel

d?

Opt

imal

fiel

d?

13 13

All

sam

eco

lor

asse

edpa

rent

Mos

t,bu

tno

tal

lof

seed

pare

nt’s

colo

rA

mix

ture

ofco

lors

All

sam

eco

lor

asse

edpa

rent

Mos

t,bu

tno

tal

lof

seed

pare

nt’s

colo

rA

mix

ture

ofco

lors

9 1 3 9 1 3

5 0 3 5 0 3

4 1 0 4 1 0W

ill

ther

ebe

mor

eof

that

colo

rif

isol

ated

and

sele

ct-

edfo

r5

year

s?(i

fdi

dno

tan

swer

‘‘al

lsa

me

colo

ras

seed

pare

nt’’

abov

e)

4Y

esN

o4 0

3 01 0

sought out for planting. Therefore, in subsequentinterviews we focused on discerning farmers’perceptions of differences between large andsmall seeds that might explain their preferencesfor large seed.

Our question was, Do farmers perceive a cor-relation between seed phenotype (size) and otherphenotypic traits of the plant that grows fromthat seed, such as seedling vigor? The null hy-pothesis was that farmers would see no corre-lation, saying that later developmental stageswould be the same for big seeds and small seedswhen planted in an optimal field, and similarly,they would be the same in a typical field. Thealternative hypothesis was that later develop-mental stages would not be the same for bigseeds and small seeds, that a correlation exists.Effects of seed size on subsequent stages withinone generation can be a genetic effect due topleiotropy, linkage, or epistatic effects, or maybe entirely due to seed size per se.

Two comparably large ears, one with smallseeds and the other with large seeds, but other-wise similar, were used to demonstrate thesescenarios. We asked farmers to imagine that inboth a typical field, and an optimal field one rowof each seed type is planted with identical spac-ing, one seed per hill, and only ears producedby single ear plants considered. We also askedfarmers more specifically what major problemsthey have in the first month after planting, andreferred to those answers in explaining our ques-tion about the effects of seed size in a typicalfield. We asked if there would be any differencesin a typical field between large and small seeds.The majority of farmers (10) answered that therewould be no differences, with a minority (3)stating that larger seeds emerge better and pro-duce larger seedlings in the typical field they arefamiliar with (Table 8).

Since more farmers initially provided moredetailed answers about differences in an optimalfield, we asked farmers more specifically if therewould be differences in emergence/seedling sizeand vigor, plant size, or grain harvested whenlarge and small seeds were planted in an optimalfield. The majority of responses suggest thatthese farmers see seed size as having no con-sequence for plant performance, implying nophenotypic or genetic correlation between thesetraits. A total of three households stated thatthere would be differences in some aspect ofseedling/plant performance associated with dif-


Fig. 2. Genetic perceptions: responses to ear length scenarios.

ferent seed sizes: in emergence (two), seedlingsize/vigor (two), and yield (one).

Altogether, four households said that therewould be differences between large and smallseeds in either a typical or optimal field. Re-sponses that differences in seed size will resultin differences in seedling/plant performance in

both types of fields (two households) can be in-terpreted as suggesting a genotypic correlationbetween these that is evident regardless of en-vironmental variation. The response of onehousehold that there was a difference only in anoptimal but not in a typical field suggests a ge-netic correlation may exist but is masked by the


TA

BL

E7.

FA

RM

ER

S’

RE

SP

ON

SE

ST

OG

EN

ET

ICP

ER

CE

PT

ION

SS

CE

NA

RIO

SR

EG

AR

DIN

GH

ER

ITA

BIL

ITY

AN

DG

EN

ET

ICV

AR

IAN

CE

OF

EA

RL

EN

GT

H.

Sce

nari

o/qu

esti

onN

o.of

hous

ehol

dsR

espo

nse

Tot

alS

ta.

Mar

iaS

.A

nton

io

Ifon

lyse

edfr

omlo

ngea

rsar

ese

lect

ed,

wha

tsi

zew

ill

the

ears

beth

atar

epr

oduc

edw

hen

thos

ese

eds

are

sow

nin

a

Typ

ical

fiel

d?

Why

?

13 10

All

long

All

shor

tM

ixed

size

sE

nvir

onm

ent

dete

rmin

esev

eryt

hing

Wat

erde

term

ines

Not

hing

lim

itin

ggr

owth

Soi

lde

term

ines

0 0 13 8 1 0 1

0 0 8 4 1 0 1

0 0 5 4 0 0 0If

only

seed

from

long

ears

are

sele

cted

,w

hat

size

wil

lth

eea

rsbe

that

are

prod

uced

whe

nth

ose

seed

sar

eso

wn

inan

Opt

imal

fiel

d?

Why

?

13 10

All

long

All

shor

tM

ixed

size

sE

nvir

onm

ent

dete

rmin

esev

eryt

hing

Wat

erde

term

ines

Not

hing

lim

itin

ggr

owth

Soi

lde

term

ines

The

reis

alw

ays

som

eva

riat

ion

12 0 1 7 0 1 1 1

7 0 1 3 0 1 1 1

5 0 0 4 0 0 0 0

variation present in a typical field, and thereforenot reflected in phenotypic correlation betweenseed size and seedling/plant performance wherethere is high VE and VG3E. The response of oneother household that there is a difference onlyin typical fields suggests that the advantages oflarge seeds are only evident under stress. How-ever, due both to the small sample size and thecomplex nature of the relationships involved, thefindings are only suggestive and further researchis required. One farmer provided the followingexplanation: This is why we (farmers here) donot use small seeds, this is what we think willhappen (poorer seedling vigor), but we do notknow this for certain because we never plantsmall seeds because that is not our custom here.To be sure of what the results would be wewould need to try it for a while.

Finally, since the majority of farmers said thatthere were either no differences in performancebetween large and small seeds, or that any dif-ferences in early plant development did not re-sult in differences at harvest, we asked themwhy they select/purchase large seeds for plant-ing, mentioning that they cost more, and providefewer seeds per unit volume (maize is sold byvolume). Most households (nine) offered no rea-son except custom for selecting large seeds.However, three households (the three who saidthat larger seeds produced larger and/or morevigorous seedlings in an optimal field) reiteratedthat a reason for planting larger seeds is becausethey produce larger seedlings. Finally, the onehousehold responding that there would be a dif-ference in a typical field, stated a possible re-duction in seed size over generations if largeseed were not selected, implying recognition ofthe possibility of a genetic component to seedsize and that selecting large seed size maintainsthis characteristic in a population, rather thanchanging the population.

DISCUSSION AND CONCLUSIONS

The results of this study address the three fun-damental questions we posed at the beginning ofthis paper. First, the genetic perceptions scenar-ios proved to be a useful method for outsidersto communicate with farmers about, and to un-derstand the abstract conceptual bases of, farm-ers’ plant breeding. Second, the results providetheoretical and empirical insights into farmerplant breeding knowledge and practice in termsof the biological model of plant breeding. Third,


TA

BL

E8.

FA

RM

ER

S’

RE

SP

ON

SE

ST

OG

EN

ET

ICP

ER

CE

PT

ION

SS

CE

NA

RIO

SR

EG

AR

DIN

GT

HE

EF

FE

CT

OF

SE

ED

SIZ

E.

Que

stio

nH

ouse

hold

sA

nsw

erTo

tal

Sta.

Mar

iaS.

Ant

onio

Wha

tar

em

ajor

prob

lem

sin

firs

tm

onth

afte

rpl

ant-

ing

mai

zese

ed(m

ore

than

one

resp

onse

poss

ible

)?13

Lac

kof

wat

erIn

sect

larv

aeA

nts

Bir

dsW

eeds

Rat

s,li

zard

sD

isea

se

12 6 5 3 2 2 1

7 5 3 2 2 0 1

5 1 2 1 0 2 0W

ith

larg

ev.

smal

lse

eds

sow

nin

aty

pica

lfi

eld,

are

ther

ean

ydi

ffer

ence

s?13

Yes

(lar

ger

seed

emer

ges

bett

er,

give

sla

rger

seed

ling

)N

o3 10

3 50 5

Wit

hla

rge

v.sm

all

seed

sso

wn

inan

opti

mal

fiel

d,ar

eth

ere

diff

eren

ces

ina :

All

diff

eren

ces

13Y

es3

30

Em

erge

nce?

See

dlin

gsi

ze/v

igor

?

Am

ount

ofgr

ain

harv

este

d?

13 13 7

No

Yes

No

Yes

No

Yes

No

10 2 11 2 11 1 6

5 2 6 2 6 1 3

5 0 5 0 5 0 3W

hyse

lect

/pur

chas

ela

rge

seed

for

plan

ting

?A

nsw

ers

ofho

useh

olds

that

neve

rsa

idla

rge

seed

sdi

ffer

ent

than

smal

lon

es

Ans

wer

sof

hous

ehol

dsth

atsa

idla

rge

seed

sdi

ffer

ent

than

smal

lon

esin

aty

pica

lor

anop

tim

alfi

eld

orbo

th

9 4

Who

know

s?O

urcu

stom

No

diff

eren

ce,

sam

ety

peof

seed

Ifw

edo

n’t,

seed

sw

ill

beco

me

smal

ler

over

the

year

sL

arge

rse

eds

prod

uce

larg

erse

edli

ngs

1 2 6 1 3

1 1 2 1 3

0 1 4 0 0

aSo

me

hous

ehol

dsre

spon

ded

posi

tivel

yto

mor

eth

anon

eco

nseq

uenc

eof

diff

eren

ces

inse

edsi

ze.


these insights appear to have potential for in-forming and improving CPB.

In the remainder of this section we discuss theresults regarding farmers’ varietal classificationand intrapopulation selection, and the usefulnessof the insights this information provides forCPB.

VG, VE, AND VARIETAL CLASSIFICATION

The objectives, methods and results for theinvestigation of varietal classification withinblanco criollo are presented in Table 9. Farmersindicated that they recognize a concept parallelto VG by their use of distinctly named varieties.In the case of Santa Maria, both varieties as con-ceived by farmers may be present within onepopulation. We hypothesize that the contrast inblanco criollo varieties between Santa Mariaand San Antonio reflects, in part, differences be-tween their growing environments. In Santa Ma-ria, aside from hibrido, no distinct blanco vari-eties were identified as being maintained for al-location to particular environments determinedby variation between locations, years or otherfactors among fields cultivated by that commu-nity. San Antonio farmers, on the other hand,say that they maintain blanco varieties based ontheir different performances in response to VE,specifically year-to-year variation in timing andamount of precipitation. Attention to cyclelength in San Antonio may be one reason thatdespite gene flow through seed exchange andsubsequent pollen movement there is a signifi-cant difference in days to anthesis betweenwhite maize populations evaluated from thesetwo communities (Soleri 1999).

These findings suggest that intrafield VE ap-pears greater to farmers in Santa Maria than VE

between fields/years and, therefore, maintainingseparate varieties (distinct sets of VG) for differ-ent fields/years is not worth their effort. This isnot the case in the eyes of San Antonio farmers.Rather, the findings suggest the hypothesis thatone of the factors contributing to farmers’ main-tenance of distinct varieties of a class of maize(blanco in this case), is those farmers’ assess-ment of the magnitude of VE among their grow-ing environments and the costs and benefits tothem of maintaining each variety. Evidence fromtwo communities in the Sierra Juarez de Oaxacasupports this hypothesis as well (Soleri et al.1998). Those communities are both located at2500 m above sea level, their maize fields are

distributed across a 300 m range of elevationsand farmers maintain distinct local white maizevarieties specifically for two or even three dif-ferent classes of field elevations. Whether or notthis classification and use of varieties accordingto specific environments is based on genetic dif-ferences, and if it occurs simultaneously with thepost-harvest oriented distinctions reported inSanta Maria (cuadrado v. bolita), cannot be de-termined from this study but would have impli-cations for population VG.

The result that farmer categorization of vari-eties by duration in the ear identification exer-cise in San Antonio was not supported by phe-nological measurements of plants from whichselected ears came complicates interpretation offarmers’ comments and practices, makes draw-ing conclusions difficult. The simplest conclu-sion regarding the San Antonio identifications isthat these cycle length categories are the resultof poor observation by the farmers. However,the consistent interest in cycle length in San An-tonio, the efforts made to seek out planting ma-terial based on this, and other evidence of farm-ers’ astute observations do not support this ar-gument. The findings suggest a number of otherexplanations of what may be occurring. It couldbe that the ears in the identification exercise sim-ply did not present adequate or familiar variationfor distinctions to be made. It may also be thatwhen they enter the community, violento varie-ties may be identified by ear or kernel pheno-types as is done in the market place. However,after years of cultivation (and cross-pollination)in local fields, those ear and kernel phenotypesare no longer so obvious and distinct. Instead,the sorting by cycle length may be occurring infarmers’ fields intentionally sown to a particularvariety that experiences conditions eliminatingor reducing the presence of individuals of a dif-ferent cycle type. This is particularly plausibleif we assume that alleles contributing to ear andkernel phenotypes and cycle length are segre-gating independently of one another. This hy-pothesis is also supported by field data showinga significant difference in days to anthesis be-tween a farmer-identified violento populationand others identified as tardon (Soleri 1999). Fi-nally, a different standard for kernel phenotypesof newly acquired seed of different cycle vari-eties as compared to seed saved on farm forthese same varieties may also help explain theresults of this study.


TA

BL

E9.

SU

MM

AR

YO

FO

BJE

CT

IVE

S,

ME

TH

OD

SA

ND

RE

SU

LT

Sa

FO

RT

HE

INV

ES

TIG

AT

ION

OF

VA

RIE

TA

LC

LA

SS

IFIC

AT

ION

OF

BL

AN

CO

CR

IOL

LO

MA

IZE

VA

RIE

TIE

S.

Que

stio

nM

etho

d

Res

ults

Sta.

Mar

iaS.

Ant

onio

Far

mer

s’st

ated

asse

ssm

ents

Wha

tar

efa

rmer

s’de

clar

edva

riet

ies

oflo

-ca

lbl

anco

mai

ze?

For

mal

inte

rvie

ws

Two

vari

etie

sba

sed

onea

ran

dke

rnel

mor

phol

ogy.

Cua

dra-

doan

dbo

lita

.

Two

vari

etie

sba

sed

oncy

cle

leng

th.

Vio

lent

oan

dta

r-do

n.W

hat

are

farm

ers’

esti

mat

ions

ofcy

cle

leng

thof

thei

rlo

cal

blan

com

aize

vari

e-ti

es?

For

mal

inte

rvie

ws

Cyc

lele

ngth

esti

mat

esfo

rcu

adra

doan

dbo

lita

are

not

sign

ifica

ntly

diff

eren

t.

Cyc

lele

ngth

esti

mat

esfo

rvi

-ol

ento

and

tard

onar

esi

g-ni

fica

ntly

diff

eren

t.F

arm

ercl

assi

ficat

ions

anal

yzed

infie

ldtr

ials

Are

the

ears

(and

the

plan

tsth

eyca

me

from

)id

enti

fied

byfa

rmer

sin

post

har-

vest

sele

ctio

nas

repr

esen

ting

diff

eren

tbl

anco

vari

etie

sac

tual

lydi

ffer

ent?

Ear

iden

tifi

cati

onex

erci

se.

Far

mer

iden

tifi

-ca

tion

ofbl

anco

vari

etie

sba

sed

onea

ran

dke

rnel

mor

phol

ogy.

10%

iden

tifi

ca-

tion

from

100

ear

sam

ples

.P

lant

and

ear

trai

tsm

easu

red

onfa

rman

dpo

stha

rves

t.

85%

ofsi

gnifi

cant

cont

rast

ssh

owed

cuad

rado

asha

ving

larg

erea

ror

plan

tch

arac

ter-

isti

csco

mpa

red

tobo

lita

.

No

iden

tifi

cati

ons

show

edsi

gnifi

cant

diff

eren

ces

be-

twee

nvi

olen

toan

dta

rdon

for

cycl

ele

ngth

.

Isth

ere

asi

gnifi

cant

diff

eren

cein

repr

oduc

-ti

veph

enol

ogy

betw

een

farm

er-d

ecla

red

long

and

shor

tcy

cle

blan

cova

riet

ies

inS

.A

nton

io?

Com

pari

son

wit

hor

thog

onal

cont

rast

ofda

ysto

anth

esis

ofva

riet

ies

infa

rmer

s’fi

elds

.

Not

appr

opri

ate.

The

viol

ento

popu

lati

ons

(n5

2)ha

dsi

gnifi

cant

lysh

orte

rda

ysto

anth

esis

than

did

the

tard

onpo

pula

tion

s(n

52)

.

aSi

gnifi

canc

eat

P#

0.05

.


TABLE 10. SUMMARY OF FARMER PERCEPTIONS OF HERITABILITY FOR TWO TRAITS AND CORRELATION

WITH PLANT DEVELOPMENT FOR ONE TRAIT IN MAIZE.

Farmers’ responses:Observed and expected

frequencies based on nullhypotheses about phenotypevariation (tassel color, ear

length) or correlation (seedsize) in field that is

Phenotypic traitselected for

Observed;expected

Typical

NoneSome/much

Optimal

NoneSome/much Conclusions

Tassel color ObservedExpected (null hypothe-

sis: low heritability)

9a

04a

139

1340

Most farmers perceive VG of traits withhigh h2 even in environments withhigh VE. A minority sees some vari-ability due to segregation.

Ear length ObservedExpected (null hypothe-

sis: low heritability)

00

1313

1213

10

Most farmers do not perceive VG oftraits with low h2; VG is swamped byVE. A minority does see VG.

Seed size ObservedExpected (null hypothe-

sis: no positive phe-notypic correlationbetween seed size anddevelopment of theplant)

1013

30

1013

30

Most farmers do not explicitly perceiveany phenotypic correlation betweenseed size and plant development. Aminority does perceive this, and it isa reason for selecting large seeds forplanting.

a Chi square significant at P # 0.05.

HERITABILITY, SEED SIZE, AND

INTRAPOPULATION SELECTION

The hypotheses, results, and conclusions forthe investigation of intrapopulation variation arepresented in Table 10. Farmers’ responses to thegenetic perceptions scenarios showed generalagreement among farmers regarding high andlow heritability traits. Genetic variation and thecapacity to select from it were clearly recog-nized for the high heritability trait, tassel color.Here, farmers see phenotypic variation consis-tently expressed despite contrasting environ-ments, and they attribute this variation to a non-environmental source. In contrast, based on thedesign of our scenarios, farmer responses indi-cated that for traits with medium to low averageheritability most farmers perceived no VG, e.g.,an optimum environment will produce uniformphenotypes. These responses can be interpretedin two other ways. First, perhaps there is no VG

for these traits in these populations. This seemsunlikely based on seed procurement practices inthe region (Smale, Aguirre, and Bellon 1998),potential for cross pollination, and Oaxaca’s lo-cation in the region of origin and diversificationof maize (Doebley 1990). Second, farmers’ cat-egorization of all phenotypes being the same,

’’van a estar todos iguales, parecidos’’ (theywill all be the same, identical) may actually in-clude a certain level of variation that an outsidermight categorize as being different. We tried toaddress this possibility by pointedly questioningrespondents in this regard, ‘‘¿Es decir, exacta-mente parecidos en todos sus aspectos?’’ (Thatis to say, exactly the same in all ways?). Giventhe findings of this small study neither of theseexplanations appear likely and it seems best tolimit our interpretation to the original one out-lined above, that farmers perceive of no VG forlow h2 traits.

In the Central Valleys of Oaxaca, where var-iation among years and soils and moisture avail-ability even within fields is substantial (Dilley1993; Kirkby 1973), farmers’ responses areoverwhelmingly that ‘‘the environment is every-thing’’ for some traits of interest to them suchas ear length. They clearly distinguished be-tween two traits of low average heritability (earlength, seed size) and a trait with high averageheritability (tassel color), and their expectationsfor response to directional selection reflect this.Still, this research was only a beginning step inunderstanding the complexities of these farmers’selection. For example, it is difficult to ascertain


Fig. 3. Graphic representation of the hypothesis ofexperience limiting perceptions and theory: farmers’experience of VE obscuring contribution of VG to VP*.

Fig. 4. Graphic representation of hypothesis of ex-perience limiting perceptions and theory: range of VE

experienced by plant breeders limiting their anticipa-tion of G 3 E*.

farmers’ motivation for seeking large seed sizeeven in the face of greater costs to themselvesas was the case in the 1997 summer plantingseason, when the price of maize seed for plant-ing was approximately 30% greater than that ofmaize grain of the same variety for eating at thetwo markets nearest Santa Maria and San An-tonio. Most households (9, see Table 6) did notgive a reason for selecting/purchasing largeseeds or said that it was a custom, despite wide-spread recognition that it has no consequencesin terms of changing population traits. It is notclear whether this preference for large seeds isbased largely on unarticulated recognition oftheir physiological superiority, or is based oncustom or aesthetics. Our planned experimentson the effect of seed size on seed viability andseedling vigor for farmers’ varieties should helpto illuminate the biological situation. However,while determining the original motivation for acontemporary practice would be difficult, thisshould not preclude the possibility that a con-cern for seed viability and seedling vigor was afactor in its origin.

Three farmers said that they select large seedbecause of the larger seedlings they produce, butonly one of them implied a potential for herita-bility of seed size. Therefore, it seems appropri-ate to consider the following hypothesis: al-though farmers consider seed size to be a traitrelated to seed and seedling performance, mostconsider seed size the result of the maternal

plant’s growing environment and not inherited,making their low expectations for genetic re-sponse irrelevant in determining how and whythey conduct their selection (Louette and Smale1998 had similar findings). Still, care must betaken to avoid rationalizing farmer practices be-yond what can be convincingly tested simply tosatisfy researchers’ desires for a mechanisticlogic underlying those practices (Richards1995).

These findings indicate that the limits of farm-ers’ theory must be understood in context. Aswith formally trained researchers, it appears thatmost farmers base their understanding of VG andh2 on their own experiences. As such, farmers’responses may not so much deny the presenceof VG in their maize populations for traits of lowaverage h2, but reflect their unfamiliarity withoptimal growing environments and indicate theoverwhelming influence of VE in local fields, ob-scuring VG in low h2 traits (Fig. 3). Similarly, ithas been suggested that the theory underlyingsome plant breeders’ practices reflects their ex-periences (Ceccarelli 1989, 1996; Clevelandn.d.). For example, contrasting assumptionsamong plant breeders regarding appropriate se-lection environments for highly stress-prone tar-get environments have been attributed to con-trasting experiences with range and type of VE,affecting the likelihood of anticipating the ge-notype 3 environment interactions that mightoccur in the marginal fields of many farmers(Fig. 4).


When challenged with these imaginary situa-tions, some of the components of which they arefamiliar with, some farmers made sophisticatedanalyses of the determinants of VP suggesting anunderstanding similar to that of plant breeders.For example, two of the households we workedwith pointed out that even after five cycles ofisolation and selection for a highly heritable traitsuch as tassel color, occasional nonselected phe-notypes will still occur—a few yellow tasselsamong the population selected for purple tas-sels—a result that outside researchers would at-tribute to crossing and segregation in a hetero-geneous population.

Overall, the variation in farmers’ responses isnot surprising, given the variability in the distri-bution of expertise and inquisitiveness that is afrequent finding of researchers, for example inregard to genetic resources (Friis-Hansen 1996)or propagule selection (Boster 1996; for a gen-eral review see Berlin 1992: Chapter 5).

IMPLICATIONS FOR COLLABORATIVE PLANT

BREEDING

These genetic perceptions discussions withfarmers were not undertaken as tests of theirknowledge, nor was their knowledge being com-pared against a correct or scientific template. Werecognize that many other factors that lie beyondthe realm of this work contribute to farmers’knowledge and practices regarding their crops,including sociocultural, economic, and individ-ual variables (Berlin 1992).

A genetic perceptions-style approach attemptsto neutralize the realm of practice—in this caseselection and crop improvement—to the extentthat the dichotomy between scientific and non-scientific practice is abandoned and the commonelements contributing to farmer and breederpractice, such as theory, are recognized. Thegreatest obstacle to this has typically been thehierarchy of knowledge implicit in many con-ventional approaches to agricultural researchand extension, particularly in the context of lowresource, small-scale agriculture (Chambers1993). Though we are well aware that our ap-proach is still clearly grounded in the westernscientific paradigm, the hope is that attemptingto be cognizant of the limits of that paradigmserves to ameliorate the bias of that grounding.

This research provides early empirical evidencethat farmers’ knowledge and practice concerningvarietal choices, including seed procurement pat-

terns, and selection strategies and practices are atleast partially based on theory—fundamental per-ceptions about their crop populations and growingenvironments and the interaction between them.Thus, while documentation of the specific patternsand practices themselves is valuable, it seems like-ly that identifying and understanding the percep-tions that underlie them may ultimately providemore versatile tools for the development of CPB.

It may be that a more profound understandingof farmers’ genetic perceptions could contributeto more appropriate educational efforts that gobeyond a hierarchical transfer of technologiessuch as experimental design, stratified selectionor pollination control, and provide farmers withconceptual tools they can use to adapt or devel-op their own innovations to best meet theirneeds (Bentley 1989; Cleveland and Soleri1991). Finally, giving plant breeders and otherresearchers an appreciation for the reality of thechallenges facing farmers, the theory contribut-ing to how farmers address those challenges, andthe situation of theory among other factors inthe formation of both farmers’ and researchers’practices may facilitate real collaboration andthe benefits it has to offer.

ACKNOWLEDGMENTS

We thank the farming households in Oaxaca who taught us much andbrought great patience and good humor to our work together, and themunicipal authorities of the study communities for permission to conductthis work. Thanks to Estudios Rurales y Asesorıa (ERA) and Union deComunidades Forestales Zapoteco–Chinantecas (UZACHI) of Oaxaca forsharing information and discussions regarding the Sierra Juarez com-munities. We also thank M. Smale and M. E. Smith for valuable discus-sions and advice relating to this work; Humberto Rios Labrada for trans-lation of the Spanish abstract; three anonymous reviewers for EconomicBotany, one of who provided particularly valuable suggestions for im-provement. Special thanks to S. E. Smith for stimulating discussions andassistance in many forms with this work. This study was supported inpart by grants to DS from the American Association for University Wom-en, Association for Women in Science; Research Training Grant for theStudy of Biological Diversity, University of Arizona; Sigma Xi; the US–Mexico Fulbright Commission; and the USAID–CGIAR Linkage Pro-gram; grants to DAC from the Committee on Research, Faculty Senate,UC Santa Barbara; and financial and logistic support of the EconomicsProgram of CIMMYT through M. Smale, and the Maize Program ofCIMMYT through S. Taba.

LITERATURE CITED

Aquino, P. 1998. Mexico. Pages 231–250 in M. L.Morris, ed., Maize seed industries in developingcountries. Lynne Rienner and CIMMYT, Boulder,Colorado and Mexico, D.F.

Aragon Cuevas, F. 1995. La produccion de maız enOaxaca, situacion actual y perspectivas futuras.(Report for Instituto Nacional de InvestigacionesForestales y Agropecuarias (INIFAP), Centro de In-


vestigacion Regional del Pacifico–Sur, Febrero1995).

Atlin, G. N., and K. J. Frey. 1989. Breeding cropvarieties for low-input agriculture. American Jour-nal of Alternative Agriculture 4:53–58.

Banziger, M., G. O. Edmeades, and H. R. Lafitte.1999. Selection for drought tolerance increasesmaize yields across a range of nitrogen levels. CropScience 39:1035–1040.

Bentley, J. 1989. What farmers don’t know can’t helpthem: the strengths and weaknesses of indigenoustechnical knowledge in Honduras. Agriculture andHuman Values 6:25–31.

Berlin, B. 1992. Ethnobiological classification: prin-ciples of categorization of plants and animals intraditional societies. Princeton University Press,Princeton, New Jersey.

Bolanos, J., and G. O. Edmeades. 1996. The impor-tance of anthesis-silking interval in breeding fordrought tolerance in tropical maize. Field CropsResearch 48:65–80.

Boster, J. 1996. Human cognition as a product andagent of evolution. Pages 269–289 in R. Ellen andK. Fukui, eds., Redefining nature: ecology, cultureand domestication. Berg, Oxford, UK.

Ceccarelli, S. 1989. Wide adaptation: How wide? Eu-phytica 40:197–205.

. 1996. Adaptation to low/high input cultiva-tion. Euphytica 92:203–214.

, S. Grando, and R. H. Booth. 1996. Inter-national breeding programmes and resource–poorfarmers: crop improvement in difficult environ-ments. Pages 99–116 in P. Eyzaguirre and M. Iwan-aga, eds., Participatory Plant Breeding. Proceedingsof a workshop on participatory plant breeding, 26–29 July 1995, Wageningen, The Netherlands. Inter-national Plant Genetic Resources Institute, Rome.

,———, and A. Impiglia. 1998. Choice of se-lection strategy in breeding barley for stress envi-ronments. Euphytica 103:307–318.

CGIAR. 1997. New frontiers in participatory researchand gender analysis: proceedings of the interna-tional seminar on participatory research and genderanalysis for technology development. CGIAR SWP,Cali, Columbia.

Chambers, R. 1993. Challenging the professions:frontiers for rural development. Intermediate Tech-nology Publications, London.

Cleveland, D. A. n.d. (Accepted) Plant breeding sci-ence as objective truth and social construction: thecase of yield stability. Agriculture and Human Val-ues.

, and D. Soleri. 1991. Food from dryland gar-dens: an ecological, nutritional, and social approachto small–scale household food production. Centerfor People, Food and Environment (with UNICEF),Tucson.

, and ———. 1997. Posting to the Farmer-breed-

ing list serve ,[email protected]. sponsored by the CGIAR’s SystemwideProgram on Participatory Research and Gender Anal-ysis, http://www.prgaprogram.org/prga/, September27, 1997

,—-, and Smith, S. E. 2000. A biologicalframework for understanding farmers’ plant breed-ing. Economic Botany 54(3):377-394.

Coe, E. H., Jr., M. G. Neuffer, and D. A. Hoising-ton. 1988. The genetics of corn. Pages 81–258 inG. F. Sprague and J. W. Dudley, eds., Corn and cornimprovement, Third Edition. ASA, CSSA, SSSA,Madison, Wisconsin.

CPRO-DLO and WAU, eds. 1999. Summary Pro-ceedings: Workshop on ‘Breeding for low–inputconditions, and consequences for participatoryplant breeding.’ 26 and 27 January. CPRO–DLO,Wageningen, The Netherlands.

de Janvry, A., G. Gordillo, and E. Sadoulet. 1997.Mexico’s second agrarian reform: household andcommunity responses. Ejido Reform Research Pro-ject, Center for U.S.–Mexican Studies, Universityof California, San Diego, San Diego.

Dhamotharan, M., E. Weltzien R., M. L. Whitaker,H. F. W. Rattunde, M. M. Anders, L. C. Tyagi,V. K. Manga, and K. L. Vyas. 1997. IntegratedSystems Project Progress Report no. 9: Seed man-agement strategies of farmers in western Rajasthanin their social and environmental contexts: resultsfrom a workshop using new communication tech-niques for a dialog between farmers and scientists.5–8 Feb 1996, Digadi Village, Jodhpur District,Rajasthan, India. ICRISAT and University of Hoh-enheim, Pantancheru, Andhra Pradesh, India, andStuttgart, Germany.

Dilley, F. B. 1993. Climate change and agriculturaltransformation in the Oaxaca Valley, Mexico. Ph.D.dissertation, The Pennsylvania State University.

Doebley, J. 1990. Molecular evidence and the evolu-tion of maize. Economic Botany 44 supplement:6–27.

Ellen, R. 1999. Models of subsistence and ethnobiol-ogical knowledge: between extraction and cultiva-tion in Southeast Asia. Pages 91–117 in D. L. Med-in and S. Atran, eds., Folkbiology. MIT Press,Cambridge, Massachusetts.

Evans, L. T. 1993. Crop evolution, adaptation andyield. Cambridge University Press, Cambridge.

Eyzaguirre, P., and M. Iwanaga, eds. 1996. Partici-patory Plant Breeding. Proceedings of a workshopon participatory plant breeding, 26–29 July 1995,Wageningen, The Netherlands. International PlantGenetic Resources Institute, Rome.

Falconer, D. S. 1989. Introduction to quantitative ge-netics, Third Edition. Longman Scientific & Tech-nical, Essex, UK.

Friis-Hansen, E. 1996. The role of local plant geneticresource management in participatory breeding.


Pages 66–76 in P. Eyzaguirre and M. Iwanaga, eds.,Participatory Plant Breeding. Proceedings of aworkshop on participatory plant breeding, 26–29July 1995, Wageningen, The Netherlands. Interna-tional Plant Genetic Resources Institute, Rome.

Garcıa Barrios, R., and L. Garcıa Barrios. 1994.The remnants of community: migration, corn sup-ply and social transformation in the Mixteca Altaof Oaxaca. Pages 99–118 in C. Hewitt de Alcantra,ed., Transformation of Rural Mexico, Number 2:Economic restructuring and rural subsistence inMexico: Corn and the crisis of the 1980s. EjidoReform Research Project, Center for U.S.–MexicanStudies, University of California, San Diego, andUnited Nations Research Institute for Social De-velopment (UNRISD), San Diego, California, USAand Geneva, Switzerland.

Gomez, F., F. Bueso, R. Reconco, P. Hughes-Hal-lett, J. Bentley, and M. Smith. 1995. Manual demejoramiento y conservacion del maız criollo conpequeno agricultores. Escuela Agrıcola Panameri-cana Zamorano, Departamento de Agronomıa,Centro Internacional de Tecnologıa de Semillas yGranos, Honduras.

Hallauer, A. R., and J. B. Miranda. 1988. Quanti-tative genetics in maize breeding, second Edition.Iowa State University, Ames, IA.

Harlan, J. R. 1992. Crops and man, Second Edition.American Society of Agronomy, Inc. and Crop Sci-ence Society of America, Inc., Madison, Wiscon-sin.

Heisey, P. W., and G. O. Edmeades. 1999. Part 1.Maize production in drought–stressed environ-ments: technical options and research resource al-location. Pages 1–36 in CIMMYT, ed., Worldmaize facts and trends 1997/98. CIMMYT, Mexico,D.F.

Heisey, P. W., M. L. Morris, D. Byerlee, and M. A.Lopez-Pereira. 1998. Economics of hybrid maizeadoption. Pages 143–158 in M.L. Morris, ed.,Maize seed industries in developing countries. Lyn-ne Rienner; CIMMYT, Boulder, Colorado; Mexico,D.F.

INEGI. 1993. Valles Centrales de Oaxaca, perfil so-ciodemografico, IX censo general de poblacion yvivienda, 1990. INEGI, Aguascalientes, Aguasca-lientes, Mexico.

. 1996. La agicultura en Oaxaca. INEGI,Aguascalientes, Aguascalientes, Mexico.

Kirkby, A. V. T. 1973. Prehistory and human ecologyof the Valley of Oaxaca, Volume 1: The use of landand water resources in the past and present Valleyof Oaxaca, Mexico. Museum of Anthropology,University of Michigan, Ann Arbor, Michigan,Memoirs of the Museum of Anthropology, Univer-sity of Michigan, Number 5.

Lafitte, H. R., and G. O. Edmeades. 1994. Improve-ment for tolerance to low nitrogen in tropical

maize. I. Selection criteria. Field Crops Research39:1–14.

Louette, D., and M. Smale. 1998. Economics Work-ing Paper 98–04: Farmers’ seed selection practicesand maize variety characteristics in a traditionalMexican community. CIMMYT, Mexico, D.F.

McGuire, S., G. Manicad, and L. Sperling. 1999.Technical and institutional issues in participatoryplant breeding—Done from the perspective offarmer plant breeding: a global analysis of issuesand of current experience. CGIAR SystemwideProgram on Participatory Research and GenderAnalysis for Technology Development and Insti-tutional Innovation. Working Document No. 2,March 1999, Cali, Columbia.

Medin, D. L., and S. Atran. 1999. Introduction. Pages1–15 in D. L. Medin and S. Atran, eds., Folkbiol-ogy. MIT Press, Cambridge, Massachusetts.

Morris, M. L., and M. A. Lopez Pereira. 1999. Im-pacts of maize breeding research in Latin America.CIMMYT, Mexico, D.F.

Nyquist, W. E. 1991. Estimation of heritability andprediction of selection response in plant popula-tions. Critical Reviews in Plant Sciences 10:235–322.

Perales, H., S. B. Brush, and C. O. Qualset. 1998.Agronomic and economic competitiveness ofmaize landraces and in situ conservation in Mexi-co. Pages 109–126 in M. Smale, ed., Farmers, genebanks, and crop breeding: economic analyses of di-versity in wheat, maize, and rice. Kluwer Academ-ic, Norwell, Massachusetts.

Revilla, P., A. Butron, R. A. Malvar, and A. Ordas.1999. Relationships among kernel weight, earlyvigor, and growth in maize. Crop Science 39:654–658.

Richards, P. 1995. Farmer knowledge and plant ge-netic resource management. Pages 52–58 in J. M.M. Engels, ed., In situ conservation and sustainableuse of plant genetic resources for food and agri-culture in developing countries. International PlantGenetic Resources Institute, Rome.

Simmonds, N. W. 1979. Principles of crop improve-ment. Longman Group Ltd, London, UK.

Smale, M., A. Aguirre, and M. Bellon. 1998. Farmermanagement of maize diversity in the Central Val-leys of Oaxaca, Mexico: CIMMYT/INIFAP Base-line socioeconomic survey. CIMMYT: El Batan,Mexico.

, P. Heisey, and H. D. Leathers. 1995. Maizeof the ancestors and modern varieties: The micro-economics of high-yielding variety adoption in Ma-lawi. Economic Development and Cultural Change43:351–368.

, D. Soleri, D. A. Cleveland, D. Louette, E.Rice, J. L. Blanco, and A. Aguirre. 1998. Collab-orative plant breeding as an incentive for on-farmconservation of genetic resources: economic issues


from studies in Mexico. Pages 239–257 in M. Smale,ed., Farmers, gene banks, and crop breeding: econom-ic analyses of diversity in wheat, maize, and rice. Klu-wer Academic, Norwell, Massachusetts.

Smith, M. E., and R. L. Paliwal. 1997. Contributionsof genetic resources and biotechnology to sustain-able productivity increases in maize. In K. Watan-abe and E. Pehu, eds., Plant biotechnology andplant genetic resources for sustainability and pro-ductivity. R.G. Landes Company.

Soleri, D. 1999. Developing methodologies to under-stand farmer-managed maize folk varieties andfarmer seed selection in the Central Valleys of Oa-xaca, Mexico. Ph.D. dissertation, University of Ar-izona, Tucson.

, and D. A. Cleveland. 1993. Hopi crop di-versity and change. Journal of Ethnobiology 13:203–231.

, and S. E. Smith. n.d. (Submitted) Broad-sense heritability of farmer-managed maize popu-lations in the Central Valleys of Oaxaca, Mexico,and implications for improvement.

,—-, and D. A. Cleveland. 2000. Evaluatingthe potential for farmer and plant breeder collabo-ration: a case study of farmer maize selection inOaxaca, Mexico. Euphytica, p. 103.

, D. A. Cleveland, S. E. Smith, F. Chapela,and C. Vilchis. 1998. Development of two meth-odologies useful to PPB. Proposal funded by theCGIAR Systemwide Program on Participatory Re-

search and Gender Analysis: Plant breeding pro-gram. CIAT: Cali, Columbia. 8 pp.

Sperling, L., and M. E. Loevinsohn, eds. 1996. Usingdiversity: enhancing and maintaining genetic re-sources on-farm. International Development Re-search Centre, New Delhi, India.

, and U. Scheidegger. 1995. Gatekeeper Se-ries, No. 51: Participatory selection of beans inRwanda: results, methods and institutional issues.International Institute for Environment and Devel-opment, London.

Stephen, L. 1991. Zapotec women. University of Tex-as Press, Austin, Texas.

. 1994. Viva Zapata!: generation, gender andhistorical consciousness in the reception of ejidoreform in Oaxaca. Center for US-Mexican Studies,UCSD, San Diego, CA, USA. (Transformation ofrural Mexico, Number 6, Ejido Reform ResearchProject)

Weltzien R., E., M. L. Whitaker, H. F. W. Rattun-de, M. Dhamotharan, and M. M. Anders. 1998.Participatory approaches in pearl millet breeding.Pages 143–170 in J. Witcombe, D. Virk, and J. Far-rington, eds., Seeds of choice. Intermediate Tech-nology Publications, London.

Witcombe, J. R., A. Joshi, K. D. Joshi, and B. R.Sthapit. 1996. Farmer participatory crop improve-ment. I. Varietal selection and breeding methodsand their impact on biodiversity. Experimental Ag-riculture 32:445–460.

farmers enetic perceptions regarding their rop...

Documents