Origins, Functions, and Persistence of Crop Biodiversity in the Betsileo Highlands, Madagascar

Origins, Functions, and Persistence of Crop Biodiversityin the Betsileo Highlands, Madagascar1

T. RADANIELINA2, S. M. CARRIÈRE

3, AND G. SERPANTIÉ*,3

2Département de Biologie et Ecologie Végétales, Université d’Antananarivo, BP 906,Antananarivo 101, Madagascar

3Institut de Recherche pour le Développement (IRD), Gouvernance Risque EnvironnementDéveloppement (GRED), Montpellier, France*Corresponding author; e-mail: [email protected]

Origins, Functions, and Persistence of Crop Biodiversity in the Betsileo Highlands,Madagascar In order to assess Malagasy crop biodiversity (CB) richness and stability on aregional scale, an analysis at both species and variety levels was carried out in the Betsileoarea of the Madagascar Highlands. The study zone was a transect between the humid forestzone and the drier savanna. Peasant agriculture is faced with the challenges of a fast–growingresident population, government development and forest conservation-based policies.Sampling was done at two scales: five villages and 45 farms representative of three mainfarm types (small and poor, medium–sized, and richer farms). Farm surveys were conductedto collect data on several aspects of CB: taxonomic richness, functions, origin, date ofintroduction, rarity, and threats. Forty–five species were recorded altogether. The number ofspecies in each village varies from 27 (a forest village) to 41 (a savanna village). Sixteenspecies are omnipresent in the area: they constitute the common base of Betsileo agriculture.On the farm scale, species diversity is high everywhere: each farm grows half the speciespresent in its village, but only several varieties thereof. Yet the six main species include 57varieties in all, of which 46% are ancient and 76% are hardy landraces. The bigger farmsgrow significantly fewer varieties than smaller, poorer farms. Successful varietal introductionshave not reduced local varietal diversity. Regional and local varietal richness of a species isclosely linked to the number of varieties per farm, depending on local importance, ecologicalsuitability, and risk management, and is increased by seed exchanges. Betsileo agricultureappears to be simultaneously dynamic and faithful to its risk strategies, way of life, andvalues, allowing conservation of its technical, genetic, and cognitive capital. Currentindividual practices of conservation of rare varieties through small collections and seedexchanges should be supplemented by collective actions.

Origines, fonctions, et persistance de la biodiversité culturale sur les Hautes-TerresBetsileo, Madagascar Afin d’évaluer la biodiversité culturale Malgache (CB) et sa stabilité àune échelle régionale, une analyse de la diversité spécifique et variétale a été réalisée en paysBetsileo, au Sud des Hautes Terres de Madagascar. La zone d’étude représente un transectentre la zone forestière humide et une zone de savane plus sèche. Les agriculteurs y sontsoumis à la pression démographique, aux mesures de conservation et aux projets dedéveloppement. Cinq villages et 45 exploitations représentatives de trois types majeursd’exploitation (petites pauvres, moyenne, riche) ont été échantillonnés et enquêtés. Desinformations ont été collectées sur les différents aspects de la diversité des plantes cultivées:taxonomie, fonctions, origine, ancienneté, propriétés, importance quantitative et qualitative

1 Received 18 August 2013; accepted 2 April 2014;published online 14 May 2014.

Electronic supplementary material The online ver-sion of this article (doi:10.1007/s12231-014-9266-7)contains supplementary material, which is available toauthorized users.

Economic Botany, 68(2), 2014, pp. 123–136© 2014, by The New York Botanical Garden Press, Bronx, NY 10458-5126 U.S.A.

http://dx.doi.org/10.1007/s12231-014-9266-7

de l’espèce et de la variété. Au total, 45 espèces ont été enregistrées. Le nombre d’espècespar village varie de 27 (forêt) à 41 (savane). Seize espèces sont omniprésentes dans la zone,constituant la base de l’agriculture betsileo. Au niveau de l’exploitation, le nombre d’espècesmaintenues est élevé: chaque exploitation cultive environ la moitié des espèces du village. Parcontre le nombre de variétés par exploitation est modéré. Les six principales espèces cultivéescontiennent pourtant au total 57 variétés dont 46% sont anciennes et 76% rustiques. Larichesse variétale d’une espèce au niveau du village dépend d’abord du nombre de variétésmaintenues par exploitation en relation avec l’importance locale de la culture, son adaptationécologique et la gestion du risque. Elle est accrue par les pratiques d’échanges de semences.Les grandes exploitations maintiennent moins de variétés que les petites. L’introduction devariétés de la recherche ne conduit pas à réduire l’utilisation des variétés locales rustiques ettoujours appréciées. L’agriculture betsileo parait être à la fois dynamique et fidèle à sonpatrimoine, les choix et pratiques individuelles de conservation et d’échanges permettant à lafois l’évolution et la conservation des ressources génétiques et les valeurs culturelles. Lespratiques individuelles de conservation et d’échanges des variétés rares pourraient êtrecomplétées par des actions plus collectives.

Key Words: Madagascar, Betsileo, Agrobiodiversity, Conservation, Forest, Highlands,Ethnobotany.

IntroductionAgrobiodiversity constitutes the biodiversity of

agroecosystems, and crop biodiversity (CB) representsthe “planned” aspect of this as opposed to its “associated”aspect—pests, weeds, diseases, companion species, andsymbiotic and auxiliary species (Jackson et al. 2007). CBrepresents not only technical and cultural capital(Emperaire 2004), but also a genetic resource forcreating varieties in anticipation of future adaptationrequirements (Bazile et al. 2008). The biodiversity of acrop results from a balance between two dynamicprocesses: diversification by selection, introductions,and gene exchange with wild relatives; and impover-ishment by gene exchange, migration, genetic driftthrough multiplication, and abandonment (Louette etal. 1997). It follows that the gene pool is potentiallythreatened by erosion as a result of many factors:conversion to varieties introduced by the GreenRevolution, standardization of farming systems andconsumer models, destruction of wild relatives’ habitats,forgotten know–how, and competition between species(Caillon 2005; Hammer et al. 2003).There have been studies of biodiversity in Madagas-

car, a megadiversity hotspot where the issues ofconservation incentives, poverty alleviation, and adap-tation to climate change are current political concerns.Research has focused especially on wild native biodi-versity, given its richness and well–known endemicvalue (Andrianoelina et al. 2006). Ongoing studies ofcultivated material focus on genetic diversity in a smallnumber of model crops: rice, due to its preponderancein Malagasy agriculture (Ahmadi et al. 1988; LeBourdiec 1974; Radanielina et al. 2013), and yamsand coffee, due to the existence of many endemic wild

species (Campa et al. 2008; Tostain and Rejo–Fienena2010). However, there is a genuine knowledge gapoutside these two research fields. Regional CB atecosystem, species, and variety levels is little known,and it is not known whether local varieties with theirbroad genetic base are being maintained. There areindeed potential threats. In the Malagasy Highlands, anumber of factors are rapidly changing farmingsystems, including the pressure of population onirrigable land and forest and the pressure ofdevelopment projects and forest conservation (Brookset al. 2009; Serpantié et al. 2007).The first goal of this paper is to assess crop

biodiversity—richness, origins, functions, and threatsof loss—in the Malagasy Highlands on a regionalscale. Second, it aims to identify potential actions foroptimum in situ conservation, particularly in a zoneprioritized for wild biodiversity conservation. Boththese purposes call for an analysis of links betweenregional CB and farming practices at more detailedscales: village and farm.

Study AreaThe middle Betsileo area, around Fianarantsoa city,

covers the southern part of the Madagascar Highlands(Fig. 1).This area comprises large basins with numerous

bottomlands and higher upland areas. The naturalvegetation is a steppe savanna. A narrow band of naturalforest marks its eastern boundary. Its medium altitude(1,000–1,300 m) moderates the tropical climate. Thesoil consists of poor ferralsols on the slopes, andsomewhat peaty hydromorphic soil in the bottomlands(fluvisols and cambisols).

124 ECONOMIC BOTANY [VOL 68

The Betsileo area is densely populated by smallfamilies; together, they practice agriculture, ani-mal husbandry, and forestry. Betsileo farmers arewell known for being hard workers, farming allyear long. Paddy rice is the staple food and theheart of social exchanges (Le Bourdiec 1974). Inthe Haute Matsiatra region in the rural district ofFianarantsoa II (105 hab/km2), irrigated lowlandpaddies and terraces (rice and counter–seasoncrops) make up 6% of the land and 8% is rainfed(Ministère de l’Agriculture 2003). The rest isdevoted to livestock, fallow land, reforestation,and conservation of natural forests. In addition tothe challenges of rural development and foodsecurity, an intensification policy was reinforcedafter 1995 to meet the environmental challenge ofbiodiversity conservation of natural forests aroundthe “forest corridor of Ranomafana–Andringitra”(Serpantié et al. 2007).

Five villages of about 50 households wereselected along a 45 km climatic transect perpen-

dicular to the forest edge (Fig. 1, Table 1).Ambalamarina is isolated in the driest savanna, tothe west. Amparihilava and Igodona lie insavanna land in the middle of the transect.Ambendrana (on the forest edge) and Amindrabe(inside the forest) have fresher, more humid,conditions.

The socio–economic situations differ accordingto distance to markets and other characteristics(Table 2).

MethodsIn this representative zone of the Highlands,

we examined the importance of CB (exceptingforages) at different scales (regional, village, andfarm) and its variation between forest zones withhigh environmental management pressure andnon–forest zones with high population pressure.

To characterize and assess CB dynamically,several dimensions were taken into account,taking the farmers’ point of view and using their

Fig. 1. Study site and location of villages studied.

Table 1. MAIN ENVIRONMENTS IN STUDY SITES.

Villages Ambalamarina Amparihilava Igodona Ambendrana Amindrabe

Climates Drier, warmer Highland climate Highland climate More humid Fresher, morehumid

Vegetationtype

Steppe–savanna Steppe–savanna Steppe–savanna Forest edge Forest

Rainfedland–use

Pasture, rainfedcrops (RC)

Pasture, RC,eucalyptusplantations

Pasture, RC,pineplantations

Pasture, RC,shrub fallows,peak forests

Mature Forest +fire regrowth,RC, shrubfallows

Quality ofpaddies

Good (mineral soils) Good (mineral soils) Various Various Poor (peatyorganic soil)

125RADANIELINA ET AL: CROP BIODIVERSITY IN MADAGASCAR2014]

own criteria: (1) Taxonomic richness was de-scribed through the farmers’ identifications ofvarieties, on the assumptions that different varietynames correspond to different genomes and thatthe same name used in two different locationswithin a same area indicates a close geneticrelationship; (2) The different species and varietyfunctions fulfilled were classified in terms ofeconomic and agronomic criteria and propertiesof the product; (3) Date and origin of introduc-tion; and (4) Rarity and exclusiveness. Theselatter indicators enabled us to assess the risk ofextinction of a local variety.

A typology of households was established onthe basis of household size and means ofproduction (Table 3).

Type 1 comprises small farms without cattlerunning a small area of rented paddies. They areeither new farms with young households (30years old) with off–farm incomes, or poorfarmers. Type 2 comprises medium–sized farms.These have more paddies and only one animal;they are mostly owner–occupied but not suffi-cient to meet higher needs. Type 3 comprises thelargest, richest farms. They have sufficient cattlefor manure, tillage, and trampling the riceplantation; paddy area is sufficient, and familylabor is supplemented by hired or “invited”laborers. We selected 10 farms per village(sampling rate of roughly 20%) except inAmindrabe (N = 5), in proportions reflectingthe distribution of the three farm types within thevillage.

On each farm, a semi–structured interviewwith the household heads (generally the father,frequently accompanied by the mother) wascarried out first, to obtain general data.

The survey determined how the farmersassigned species or varieties, using an in situinventory in humid season and inquiry aboutother seasons. Some plots on different “landscapeunits” (i.e., different soil types, types of water andfertility management, and distance from thedwelling [Blanc–Pamard and Milleville 1985])were visited. The granaries were also inventoried.

Later talks with farm heads provided detailedinformation on the roles of the observed diversity,the reasons for the choice of each species andvariety, the sources of seed, the date of introduc-tion, the cropped area, and the local criteria forclassifying the varieties and their properties, inagronomic, economic, or culinary terms. Forexample, hardiness of the variety is an important

Table2.

MAIN

LIVELIHOODSAND

SOCIO

–ECONOMIC

CONDITIO

NSIN

STUDYSITES.

Villages

Ambalamarina

Amparihilava

Igodona

Ambend

rana

Amindrabe

Historicalkingdoms

Isandra

Lalangina

Lalangina

Lalangina

Lalangina

Padd

yavailability

(100

m2/inh

ab)

<55

1010

5

Maincommercial

resources

Dry

cassava

Eucalyptusandpine

products

Rice,pine,cassava

Rice,naturalforestproducts

Naturalforestproducts,livestock

Weeklymarket

3km

10km

10km

10km

15km

Accessibility

Goodtrail

Goodtrail

Goodtrail

Poor

trail

Footpath

Constraints

Livestocktheft,insect

pests,drought

Cassava

theft

—Vertebratepests(rats,birds,boars)

Vertebratepests,cold,moisture

Localprojects

AHI,19

99Farm

ingsystem

research,

agroforestry,irrigatio

n

—ANAE,20

00Com

batin

gerosion

LDI–ERI,20

00–20

06Fo

restcommun

itymanagem

ent,

diversificatio

n,intensificatio

n

LDI–ERI,20

02–20

06Fo

restcommun

itymanagem

ent,

diversificatio

n

Legend

:AHI=African

HighlandInitiative,

aninternationalR

&D

project;ANAE=NationalAssociatio

nforRural

andEnvironmentalPlanning;LD

I–ERI=Land

scape

DevelopmentInitiativeandEco

RegionalInitiative.


property for risk management that characterizes“landraces” (i.e., ancient varieties), but also someintroduced more recently from other areas.

Samples of plants and seeds (rice, bean, andBambara groundnut) were collected and placed ina herbarium to check the relationship betweenthe vernacular name and morphological features,which might differ between villages and farms.

For each species, a CB richness value wasassigned on the basis of the number of varietiesper village and per farm. The rarity of a varietywas assessed by classifying each variety as “ubiq-uitous” (present in at least four villages out offive), “frequent” (two or three villages), “rare”(seen in only one village), “exclusive” (found ononly one farm in a given village), and “very rare”(found on only one of the 45 farms).

Date of introduction according to farmers wasclassed as ancient (attributed to their ancestors;i.e., before 1960), recent (introduced between1960 and 1990), and new (since 1990). Bycomparing rarity levels with dates of introduction,we assessed a possible threat of loss: varieties thatare both very rare and ancient are assumed to bethreatened varieties.

To draw up a typology of the varietalcombinations practiced by the farmers, a corre-spondence analysis was carried out on a binarytable for farms and varieties (XLstat™). Varianceanalysis (XLstat™) was used to model the distri-bution of the variance in varietal richnessaccording to the farms’ parameters.

ResultsREGIONAL AND VILLAGE SCALE APPROACH

CB taxonomic richness was evaluated in itsvarious dimensions at both species and variety levels.

Taxonomic Richness at Species Level

On the five sites, 45 species were recordedaltogether (Appendix 1—Electronic Supplemen-

tary Material [ESM]). Sixteen species are omni-present in the area, as a common base of Betsileoagriculture. Among them, the main crops are rice,cassava, sweet potato, taro, Bambara groundnut,bean, corn, and groundnut, cultivated on plots ofmore than 200 m2. Of these eight, only rice,cassava, and sweet potato are cultivated on all thefarms, except in Ambalamarina (sweet potato,40% of farms). Fifteen other species are grown ingardens (less than 200 m2) or orchards, andsometimes associated with the main crops. Ofthese, only potatoes, sugar cane, banana tree, andsome leaf vegetables are common to all thevillages.

Finally, very small stands of 22 other species(11 fruit trees, 3 spices, 5 vegetables, 3 medicinalplants) are grown.

On average, 36.4 ± 2.4 species are cropped ineach savanna village, which is similar to regionalrichness (Table 4); the forest village has the fewestspecies, with a total of 27.

The importance of each species, measured asthe percentage of farmers producing it, varies. Inthe west, sweet potato (40%) and taro (50%) arerare as the climate is too dry. In or near the forest,the importance of taro increases (>80%) whilethat of cassava decreases (only 80%).

Taxonomic Richness at Variety Level

For the six main species (rice, cassava, sweetpotato, taro, bean, and Bambara groundnut), 57varieties were counted in all: 18 for sweet potato,12 for rice, 11 for cassava, 6 for taro, and 5 eachf o r B am b a r a g r o u n d n u t a n d b e a n(Appendix 2—ESM).

For these six species combined, each village has28.4 ± 2.0 varieties, only half the numberrecorded for the region as a whole—a lesserproportion than at species level.

For any given species, varietal richness per farmis low (1.81 to 2.40) (Fig. 2). It depends

Table 3. CHARACTERISTICS OF MAIN FARM TYPES: MEAN AND STANDARD DEVIATION.

Type of farm 1 2 3

Oxen (N) 0 1 (0) 2.53 (0.77)People (N > 15years old) 2.4 (0.6) 3.3 (1.0) 3.8 (0.7)Age of head (father or widow) 42.4 (9.4) 43.3 (8.6) 42.3 (6.7)Paddies (N) 2.0 (0.9) 2.4 (0.6) 4.1 (0.6)Land holding status of paddies Most rented Rented and owned Most owned


significantly on the species (p = 0.001). Compar-ison of means gives three significantly distinctgroups: a) sweet potato (3.6 per farm); b) cassava(2.4) and rice (2.2); and c) taro (1.6), Bambaragroundnut (1.6), and bean (1.4).Rice, cassava, and sweet potato, which occupy

the largest areas, show greater varietal richness atall scales than the other crops. The rank of thecrop in the farming system appears to be the mainfactor in diversification. Sweet potato seems to bea particular case, with 8 varieties per village and3.7 per farm. In the west, where only 40% of thefarms grow it, the ecological conditions areunfavorable and its varietal richness falls to 3 pervillage and 2.5 per farm.For secondary crops, village–scale varietal

richness varies little and does not depend on thepercentage of farms growing the species. Eachspecies has its peak of diversity in a particularsection of the transect. Taro is highly diversified

in the forest, the ecological zone that suits it best(6 varieties per village; 3.0 per farm) (Fig. 3).Taro, bean, and sweet potato are all at their leastdiverse in the west. The main factor in varietalrichness in the secondary crops, at both villageand farm scales, thus seems to be the ecologicalfactor.The species rank is in the same order at all

scales (farm, village, and region) (Fig. 2). There issignificant correlation between scales of varietalrichness at consecutive spatial scales (Fig. 4).Richness at village scale is twice that at farm scale(Fig. 4a), regardless of species. Significant corre-lation also exists between village scale and regionalscale, albeit of a different kind (Fig. 4b). Second-ary species diversify a little (x 1.5), while the mainspecies that are already diversified at the villagescale have much higher diversity regionally (x2.5). This regional diversification thus dependson species rank and farm diversification.

Table 4. COMPARISON OF CROP BIODIVERSITY AT VILLAGE SCALE.

Village Ambalamarina Amparihilava Igodona Ambendrana Amindrabe Mean

Species Total 38 41 37 39 27 36.4Per farm 17 (±3) 14 (±2) 16 (±3) 16 (±2) 13 (±4) 15.2

Varieties of 6mainspecies

Total 25 31 27 35 24 28.4Per farm 7.8 a 11.3 b 12.4 b 12.5 b 10.4 b 10.9Per species & farm 1.81 a 2.36 bc 2.18 b 2.40 c 2.36 bc 2.22

Exclusivevarieties

Total 8 3 1 9% farms with

at least one50 30 10 70

Landraces % of total varieties 66 a 74 b 77 b 81 b 83 b 76

Legend: Different letters = significant differences (NK test at p = 0.05)

1

6

11

16

rice (100%)

cassava (98%)

sweet potato (87%)

bambara groundnut

(76%)

taro (71%) bean (62%)

mea

n nb

of v

arie

ties per farm

per village

per region

Fig. 2. Variety number per species for each study scale. (% of farms cropping the species + standard error).


Levels of Rarity and Origin

Some varieties are extensively cropped; forexample, the newly introduced rice variety X265and the ancient landrace vary mena. Others areencountered only in one village (Table 5). For thesix main species, these rare varieties are notuniformly distributed: six in Ambalamarina, fivein Amparihilava, one in Igodona, four inAmbendrana, and zero in the Amindrabe forest.Six very rare varieties (grown only by one farmer)were found: three in Ambalamarina, one inAmparihilava, and two in Ambendrana.

“New” varieties were introduced very recently,since 1990 (35%). “Recent” ones were intro-duced between 1960 and 1990 (19%). Others are“ancient” varieties (46%). For rice, the percentageof ancient varieties varies between villages but is

never less than one–third: Ambalamarina 50%,Amparihilava 43%, Igodona 33%, Ambendrana60%, and Amindrabe 33%. These ancestralvarieties have a broader genetic base than thevarieties selected by research and are generallyhardy landraces.

The sources of the introduced varieties arethemselves various. Some varieties come fromMalagasy agronomic research in relation withinternational research (X265 and Zato andro rice);others come from other parts of the Betsileoregion (rice variety Isandra madio). Others againhave been brought from the paddies of Marovoayon the west coast by the Betsileo migrant workers’network (rice variety Ambaniravina).

There are many rare varieties of cassava andsweet potato, but these are mainly recentlyintroduced varieties that have become or remain

1

1.5

2

2.5

3

3.5

4

4.5

rice cassava sweet potato

bambara groundnut

taro bean

varie

ties

per

farm

cro

ppin

g th

e sp

ecie

s Ambalamarina

Amparihilava

Igodona

Ambendrana

Amindrabe

Fig. 3. Varietal richness per farm for each main species. (+standard error).

R² = 0,93*1

2

3

4

5

6

7

8

1 2 3 4 5 6 7 8

vari

etie

s p

er v

illag

e

varieties per farm

R² = 0,99*13579

1113151719

1 3 5 7 9 11 13 15 17 19

vari

etie

s p

er r

egio

n

varieties per village

a b

Fig 4. Relationship between varietal richness at two scales. (one point = one species).


marginal. Except in the west zone, the old bittercassava varieties Kajaha mainty, Fitoravina, andBalila are frequently listed but often representedonly by a few plants mixed with pure stands ofintroduced Mita or Valga. They are thus “relicvarieties,” preserved by accident or a spirit ofconservation rather than economic interest.Only one old variety of sweet potatoes is

“rare”: Vinanitelo was only found in Ambendrana,a village rich in varieties. However, it is cultivatedthere by half of the farmers surveyed, whichmeans that it is not in serious danger ofextinction.The question of the threat of loss caused by

introduced varieties arises for an old rice varietyfound alongside well–adopted introduced varie-ties: Vary vory or “round rice” in Ambalamarina.Being cultivated by 4 farmers out of 10, it isprobably cultivated more widely in the west andtherefore not inevitably threatened.There are two very rare old varieties of

Bambara groundnut: Mainty and Petina. Theyare not threatened by varietal introductions fromresearch, which does not target this secondarycrop, but they could be by other factors, such ascultural change. For example the Mainty variety isused in a ritual against sorcery. This also indicatesthat special CB may be maintained at individualscale, for example for rituals (soothsayers) or otherspecial purposes (healers), but our inquiry did nottarget these “tradition keepers.”By our criteria, all taro varieties are old:

ancestral taro diversity in Amindrabe inside theforest may be a marker of an ancient link betweenman and this forest corridor—seen as “pristine”by conservation projects. No rare varieties of tarowere found, but some are cropped only in verysmall stands, and so may be threatened byextinction. Our under–sampling in the forestzone (only five farms, only one village) supposes

there to be many other taro varieties in the forestbelt, on the basis of scale relationships in the caseof main crops (Fig. 4).

FARM SCALE APPROACH

Taxonomic Richness

Each farm grows many species, but only half ofthose are present in its village. There are morespecies per farm in Ambalamarina, in the driestzone (17 ± 3) and fewer in the forest byAmindrabe (13 ± 4).However, all the farms grow only a few

varieties of each species at any given time, exceptfor sweet potato (Fig. 2). ANOVA confirms twosignificant factors in varietal richness per farm:the village (p = 0.006) and the species (p <0.0001), with a significant interaction betweenthese two factors (p < 0.0001).For the eastern group of villages, variance in

varietal richness per farm for the six main specieswas analyzed. The variables “village,” “family size”(number of adults), “age of head of farm,”“numbers of paddies,” and “number of tambinaplots” have no significant effect, while “farmtype” and “mode of tenure” do (p < 0.05). Thericher farms that own paddies (Type 3) cultivatesignificantly fewer varieties (10.6) than small,poor farms with rented paddies (Type 1: 12.4) orexpanding farms (Type 2: 12.0).

Taxonomic Composition

Initial correspondence analysis (Fig. 5a) of thevarietal composition of the farms shows weakinertia on the first two axes (18.3%). This weakinertia indicates the highly variable nature of thevarietal composition decided on by each farmer.The F1 axis (9.4% of inertia) isolates three

Table 5. NUMBER OF VARIETIES BY SPECIES, RARITY, AND DATE OF INTRODUCTION.

Frequency Ubiquitous Frequent Rare Very rare

Date of introduction Ancient New Ancient New Ancient New Ancient NewRice 2 2 1 1 1 1Cassava 1 3 3 1 3Sweet potato 2 2 5 1 2Bean 2 1 1Bambara groundnut 3 2Taro 2 4


varieties grown on the majority of farms inAmbalamarina. This western village, with itsparticular historical and ecological context, hasseveral specific varieties, particularly rice (R5, R8,R10) and a rare variety of Bambara groundnut(P3).

A second analysis (Fig. 5b), carried out on thegroup of villages in the east, gives an inertia of21.2% distributed over the first two axes. Itisolates three groups of varieties that correspondprecisely to three villages (Vi2, Vi3, Vi4) includedin the analysis as additional variables (Fig. 5c).The closer the farms are to each other, the morevarieties they have in common, including locallyspecific varieties, revealing local practices of seedexchange or shared preferences. Alongside these“village marker” varieties, some varieties play nopart in the inertia on the first two axes. This is thecase for rice R1, R2, R3, and R4, bean H1 andH2, cassava M1 and M2, Bambara groundnutP1, and sweet potato A8. These varieties areubiquitous and many of them are varietiesintroduced region–wide by public developmentprograms and local projects.

There is no correlation between farm type,included as an additional variable, and these firsttwo factors of inertia. As regards varietalcomposition, the village effect is stronger thanfarm type.

Functional Diversity

Farm (and village) varietal diversity variesprimarily in line with the economic importanceand ecological suitability of the species. Indeed,diversity availability allows the farmer to crop theright variety at the right place and time. In thecase of rice, each variety is also characterized by its

suitability for a certain level of fertilization (3levels) and a particular transplanting date (3dates) (Table 6). The agronomic properties ofthe species therefore play a major role. At anygiven time, farmers use the number of varietiesthey need. However, they also rotate varieties ontheir field, using seeds from relatives or bought atthe market.

The rate of landraces is always high butsignificantly lower on Type 3 farms, especiallyfor cassava (NK test, p = 0.05; Table 7): the lossof cassava landraces would be a possibility in theevent of farming concentration, but this is notcurrently the case.

Other material, social, and symbolic criteriaalso play a role in variety choice. For example,with beans, the tasty Ambalalava is grown forfamily meals, while Mena has better nutritionalvalue and is grown for the meals of the hiredlaborers.

Level of Rarity

In some villages, exclusive varieties (only onefarmer in the village) are few (Igodona 10%). Inothers, most farmers (Ambendrana 70%) grow atleast one exclusive variety and some have up tothree, reflecting a collector’s behavior at farm andvillage scale. These cases testify to villages andfarmers with the motivation and know–how forconservation and perhaps for producing newplant material.

Active individual conservation practices wereobserved, for example small stands of ancientcassava varieties. In two villages in or near theforest, two farmers grow a few plants of the tarovariety Saonjo fotsy so that it will not disappear. Inthe current context, its growth cycle is considered

R1

R2

R3

R4

R5

R6

R7 R8

R9

R10

R11

M1M2M3

M4

M5

M7

M8

M9

T1 T2

T3

T4T5

T6 H1

H2

H3H4

P1

P2

P4

A1A2

A3

A5A6

A7

A8

A9

A10

A11A12

A13

A14

A15

A16

A17

-2

-1.5

-1

-0.5

0

0.5

1

1.5

2

-2 -1 0 1 2 3

F2

(8,9

%)

F1 (9,4 %)

a AFC1, active vars

R1

R2 R3

R4R6

R7

R9

R11

M1

M2M3

M4

M5

M7

M8

M9

T1

T2

T3

T4T5

T6

H1H2

H3

H4

P1P2

P4

A1

A2

A3

A5

A6A7

A8

A9

A10

A12

A13

A14

A15

A16

A17

-2.5

-1.5

-0.5

0.5

1.5

2.5

3.5

-3 -2 -1 0 1 2 3F

2 (9

,6 %

) F1 (11,6 %)

b AFC2, active vars

VI2

VI3

VI4

VI5

TY1

TY2TY3

-2.5

-1.5

-0.5

0.5

1.5

2.5

3.5

-3 -2 -1 0 1 2 3

F2

(9,6

%)

F1 (11,6 %)

c AFC2, supplementary vars

Fig. 5. Correspondence analysis on a binary table. (farm vs. variety) (VI = village, TY = farm type).


too long (more than a year, compared to eightmonths for the other varieties). However, itproduces white, aromatic tubers that are highlyappreciated by these families.

Discussion and ConclusionLIMITS OF THE METHOD

In each village, the maximum number ofvarieties was reached after surveying between onlysix and nine farms. Our sample of 10 farms wasthus sufficient to assess the CB of a village.Statistical modeling of the marginal productivityof varieties (y) for each new farm surveyed as afunction of the cumulative number (x) of surveysgives the following formula:

y ¼ 15:04 e0:425 x

This predicts an average detection of 0.35 newvarieties per five additional surveys. In Amindrabe,where only five farms were surveyed, we presumablycovered only 80% of the varieties in that village.Taking the group of four eastern villages, the

number of varieties reaches 100% after threevillages. It follows that the western and forestzones were under–sampled: the single village

surveyed in each provides only 60% of the CBof its zone. A more systematic assessment shouldsurvey at least three villages per historical orecological zone.

NEW INSIGHTS

The global literature on CB is already abun-dant but usually focuses on sites rich in biodiver-sity such as wild relatives and agroforestry systems(Baco et al. 2008; Emperaire 2004; Perreault-Archambault and Coomes 2008; Tostain andRejo–Fienena 2010). By contrast, our analysis is acontribution towards an assessment of “ordinary”CB that is nonetheless essential in terms of riskregulation, “provisioning,” and “cultural” ecosys-tem services. The crucial point of this type ofdiagnosis was to establish whether there is a threatof local CB erosion and, if so, to examine itscauses, particularly in a zone already prioritizedfor wild biodiversity conservation

Farm Strategy Remains the Primary Sourceof Diversity

Greater crop diversity than expected in thisancient rice civilization has been found. The firstsource of regional CB is diversity within the farm.

Table 6. RICE VARIETIES, FERTILIZATION DOSE, AND TRANSPLANTING SEASON.

Only green vegetation reactsto fertilization.

Does not exhaust the soil withoutfertilization

Yield reacts to fertilization

Low dose only(organic fertilizer)

High dose of mineral + organicfertilizer necessary to maintain

fertility

Early transplanting Vary mena—

ZatoandroSego

Vary lavaJaponais

Main transplanting Vary menaPiritikaVary vory

AmbaniravinaIsandra madio

Tsipala

X265Japonais

—Late transplanting Piritika

Vary voryChine—

X265—

Table 7. USE OF LANDRACES AND RESEARCH CULTIVARS BY FARM TYPE.

Farm type 1 (n = 17) 2 (n = 15) 3 (n = 13)

Landraces (% of all varieties, all species) 75.6 b 72.1 b 63.2 a% farms using only rice landrace 71 53 31

research 12 13 54cassava landrace 29 27 0

research 29 0 85


All Betsileo farmers grow many species, includingbetween six and eight principal food crops.Betsileo agriculture is thus a mixed–farmingsystem rather than rice farming, even though riceconstitutes the symbolic benchmark crop, with allother species being secondary crops in thesymbolic hierarchy (Beaujard 1981).

A fundamental factor of this diversity is thefarming strategy—adapting to the local climateand diversity of soils, the diversity of the seasons,the many landscape units, and the slow nature ofmanual work. Each farmer has access to differentecosystems, from rainfed land use systems toirrigated rice fields. To make full use of cold anddry periods (with irrigation or not), farmers needa range of species with differing cycle lengths andwater and heat requirements. Rice and vegetablesare thus cultivated on bottomlands and onterraces; cassava, Bambara groundnut, corn, andsweet potato are cultivated on the slopes; and awide range of spices, leafy vegetables, andmedicinal plants are cultivated on tambinagardens around the paddies.

In contrast to this high species diversity,farms grow only a few varieties of each crop atthe same time (although never only one).Varieties are indeed complementary in agro-nomic terms. Some varieties are better adaptedto a particular local climate, planting date,mode of fertility management, and availablelandscape units, due to their cycle, watertolerance, and level of hardiness. The Menabean, an old variety, is well–suited to the rainfedtambina fields, whereas the introduced Fotsybean is sensitive to rain and is cultivated onlyin the dry season in the bottomlands. For eachagronomic criterion (manure requirements, dateof plantation, cold tolerance) there are severalsubstitutable varieties that constitute genuinefunctional groups. Elsewhere, it has beenestablished that varietal diversity at farm scaledepends on risk management originating fromhuman or natural factors (Pinton and L.Emperaire 2001) and on ecosystem diversity(Rana et al. 2007). Diversification of plantmaterial on the farm thus reflects a strategy formaking best use of complementary situations, asin many so–called “traditional” farming systems(Bazile et al. 2008; Thurston et al. 1999). Thisstrategy reflects careful, skillful husbandry, butalso reveals a farming system with limitedresources, as richer farms were found to be lesslikely to diversify crops.

Varietal richness per farm or village is highlydependent on the relative importance of thespecies and its ecological suitability: essentialcrops are consistently more diversified thansecondary crops. In the Vakinankaratra moun-tains, Radanielina et al. (2013) also found lessrice diversity in colder zones. However, sweetpotato is an exception. It is the most diversifiedeven though it is grown on less land area thanrice or cassava. Different degrees of precocity(cycles from three to seven months) make itpossible to stagger plantation and harvestingaccording to need. But the principal factor maybe diversity in the field. Each plot, even if it issmall, is planted with a mix of varieties. Forsome, this contributes to risk management. Forothers, mixing varieties systematically increasesoutput. It has been proved in China thatgrowing several varieties of rice together on thesame plot provides several benefits including on-farm crop biodiversity conservation (Zhu etal. 2003).

Indeed, agronomic criteria are not the onlysources of diversity for each farm. The differentfood values of the varieties may also help satisfythe diversity of nutritional needs and dietarypreferences of the farmer’s family or workers. Thelocal classification of sweet potatoes use tastecriteria, increasing the many ways a sweet potatophenotype can be described: taste of leaves, colorand taste of the tuber flesh. Tevololona is famousfor its leaves, Ebokely for its taste, Mbizo karaotyfor its color and particular taste, appreciated bysome more than others.

The Scaling Effect

Farm strategy is not the only source ofconsiderable CB richness being maintained inthe Betsileo region. Particular individual skillsmay be a factor in maintaining rare varieties (suchas varieties used in rituals), but our inquiry didnot especially target these specialists. This indi-vidual source of diversification and conservationwas only incidentally found and should beinvestigated further in terms of social criteria(gender, age, special functions, etc.).

Varietal CB increases strongly moving fromfarm to village and regional scales. The diversityof farms, and the common practices of alternatingvarieties on the same plot and testing new seedsto maintain or increase yield, require exchangesbetween neighbors and relatives or market trade.


Migrations and introductions by developmentprojects promote the construction of a varietaldiversity at the village and regional scale. Ulti-mately, however, village diversity depends mostlyon ecological adaptation and the economicimportance of each crop.Partly due to under–sampling, more research is

needed to confirm the apparently poorer CB inthe forest and western zones. If confirmed, thiscould be related primarily to the relative rarity ofspecies and varieties that are well adapted to theclimate and soils of these margins. Many rice andcassava varieties in these zones have been intro-duced recently (Chinese, Japanese, Malagasyresearch). There are very few natural foodresources (fruits, spices, honey) in this forest, sofood gathering is not the cause of the current lowCB. Historical reasons may be also suggested.Human settlement in the forest is ancient butsporadic. The forested historical margin ofBetsileo was used more for pastoralism, goldmining, and gathering forest products than foragriculture.

CB is Probably Not Currently Endangeredin the Betsileo

Overall, CB in the Betsileo region is beingmaintained, thanks to farmers adopting intro-duced varieties without giving up their ancestralvarieties. Most of the old varieties of rice areubiquitous and frequently grown, which suggeststhey are not threatened in the current context inwhich Type 3 farms, which are less loyal tolandraces, are not increasing. The genetic base oflandraces is broad and they are common andwidespread. In savanna areas, research varieties areunlikely to threaten local CB, which is foundedon other needs and other cultural norms.Conformity to regional preferences thus leads toold varieties being maintained. The red ricevarieties (Vary mena, Vary piritika, Vary vory)—allancestral ones for which there is market demandto supply the city of Fianarantsoa—are also hardylandraces; they do not need mineral fertilizingand they do not exhaust the soil. The identityissue (Perreault-Archambault and Coomes 2008)may explain the Betsileo people’s liking for redrice (the red color symbolizes nobility), whereasthe international standard in agricultural researchis fertilizer–responsive and generally white rice.Socio–cultural and ecological diversity (farm,

landscape, village, history, climate) are other

significant factors in favor of maintenance.Diversity of social position, skills, gender, andages are other hypothetical factors that should betested for.This is a new example of unplanned or de facto

conservation (Brush 1991). The persistence ofCB has been confirmed in several other situationsin spite of threats such as serious drought(Bezançon et al. 2009), agricultural moderniza-tion (Barry et al. 2007), increasing marketintegration (Perreault-Archambault and Coomes2008), and demographic pressure (Blanco et al.2013). Local farming systems have their ownmechanisms for producing and conserving plantmaterial (Thurston et al. 1999). Betsileo agricul-ture exhibits this de facto CB conservation,featuring diversification, loyalty to local norms,and a low level of development.We also found individual conservation–orient-

ed living practices. Many rare varieties arecropped in small stands, thanks to an individualspirit of collection or conservation on the part ofsome farmers. But a potential threat of loss ofdiversity remains, because these preserved popu-lations are very small and dispersed.Better in situ conservation of this common

agricultural heritage and better valorization forlocal development should be carried out on acollective level. The first step in this is a CBassessment. This paper gives a basis of method-ology for a more systematic regional assessment ofcrop biodiversity. It should be followed bycomparisons among regions. The second stepinvolves monitoring. Producer organizations, ag-ronomic research, and forest corridor conserva-tion projects should play a role in assessment andmonitoring of CB and the creation of a seedexchange system. They should also mobilizespecial markets with geographical or ethical labelsfor urban, tourist, or international markets.

Links between CB and Wild BiodiversityConservation

Paradoxically, research varieties introduced forcropping intensification by public policies or localintegrated conservation and development projects(ICDP) represent potential threats of loss for bothwild biodiversity and local crop diversity in forestzones. Rice and cassava varietal introductions inHighlands have targeted criteria such as high yieldpotential, highland rainfed rice, and early cassava.All these criteria are of particular interest for the


richest farmers who fertilize their soils, but alsofor forest farmers on fertile forest soils. Obviouslyrice introductions may fulfill the yield potential inintensively fertilized, irrigated rice systems inlowlands and so help to conserve forest onhillslopes (Serpantié and Rakotondramanana2013). However, non–hardy varieties of cassavaand rainfed rice may represent a new threat toforest habitat because they are especially suitablefor fertile and wet forest hillslopes. These newcrops could also replace taro as the main forestcrop. Conservation of the two sources ofbiodiversity—natural and domesticated—is there-fore linked in the forest corridor.

AcknowledgmentsThis research was financed by Institut de

Recherche pour le Développement (IRD) andthe Malagasy CNRE (GEREM Fianarantsoaproject). Our grateful thanks go to Betsileofarmers for their knowledge, to S. Caillon and S.Tostain for their generous advice, and to allEconomic Botany reviewers.

Literature CitedAhmadi, N., T. Becquer, C. Larroque, and M.

Arnaud. 1988. Variabilité génétique du riz(Oryza sativa L.) à Madagascar. L’AgronomieTropicale 43:209–221.

Andrianoelina, O., H. Rakotondraoelina, L.Ramamonjisoa, J. Maley, P. Danthu, andJ.-M. Bouvet. 2006. Genetic diversity ofDalbergia monticola (Fabaceae) an endangeredtree species in the fragmented oriental forest ofMadagascar. Biodiversity and Conservation15:1109–1128.

Baco, M. N., G. Biaou, J.-L. Pham, and J.-P.Lescure. 2008. Facteurs géographiques etsociaux de la diversité des ignames cultivées auNord Bénin. Cahiers Agricultures 17:172–177.

Barry, M. B., J. L. Pham, S. Béavogui, A.Ghesquière, and N. Ahmadi. 2007. Diachron-ic (1979–2003) analysis of rice genetic diver-sity in Guinea did not reveal genetic erosion.Genetic Resources and Crop Evolution55:723–733.

Bazile, D., S. Dembélé, M. Soumaré, and D.Dembélé. 2008. Utilisation de la diversitévariétale du sorgho pour valoriser la diversitédes sols au Mali. Cahiers d’études et derecherches francophones / Agricultures17:86–94.

Beaujard, P. 1981. Hiérarchie végétale, hiérarchiesociale à Madagascar, La place symbolique destubercules et du riz et leurs origines à traversles mythes et les contes. ASEMI XII:157–191.

Bezançon, G., J.-L. Pham, M. Deu, Y.Vigouroux, F. Sagnard, C. Mariac, I. Kapran,A. Mamadou, B. Gérard, J. Ndjeunga, and J.Chantereau. 2009. Changes in the diversityand geographic distribution of cultivatedmillet (Pennisetum glaucum [L.] R. Br.) andsorghum (Sorghum bicolor [L.] Moench) vari-eties in Niger between 1976 and 2003.Genetic Resources and Crop Evolution56:223–236.

Blanc–Pamard, C. and P. Milleville. 1985.Pratiques paysannes, perception du milieu etsystème agraire. Pages 101–138 in Dynamiquedes systèmes agraires, A travers champsAgronomes et Géographes. ORSTOM, Paris.

Blanco, J . , L . Pasca l , L . Ramon, H.Vandenbroucke, and S. M. Carrière. 2013.Agrobiodiversity performance in contrastingisland environments: The case of shiftingcultivation in Vanuatu, Pacific. Agriculture,Ecosystems & Environment 174:28–39.

Brooks, C. P., C. Holmes, K. Kramer, B. Barnett,and T. H. Keitt. 2009. The role of demogra-phy and markets in determining deforestationrates near Ranomafana National Park, Mada-gascar. PLoS ONE 4(6):e5783.

Brush, S. B. 1991. A farmer–based approach toconserving crop germplasm. Economic Botany45:153–165.

Caillon, S. 2005. Les taros du Vanuatu: queconserver et comment? Natures SciencesSociétés 13:306–310.

Campa, C., J.-J. Rakotomalala, A. de Kochko,and S. Hamon. 2008. Chlorogenic acids:Diversity in green beans of wild coffee species.Advances in Plant Physiology 10:421–437.

Emperaire, L. 2004. La biodiversité agricole enAmazonie brés i l i enne . Ressource e tpatrimoine. JATBA, Revue d’ethnobiologie42:113–126.

Hammer, K., N. Arrowsmith, and T. Gladis.2003. Agrobiodiversity with emphasis on plantgenetic resources. Naturwissenschaften90:241–250.

Jackson, L. E., U. Pascual, and T. Hodgkin.2 0 0 7 . U t i l i z i n g a n d c o n s e r v i n gagrobiodiversity in agricultural landscapes.Agriculture, Ecosystems & Environment121:196–210.


Le Bourdiec, F. 1974. Hommes et paysage du rizà Madagascar, Etude de géographie humaine.FTM, Antananarivo, Madagascar.

Louette, D., A. Charrier, and J. Berthaud. 1997.In situ conservation of maize in Mexico:Genetic diversity and maize seed managementin a traditional community. Economic Botany51:20–38.

Ministère de l’Agriculture, de l’Elevage et de laPêche, Repoblikan’i Madagasikara. 2003.Monographie de la région Haute-Matsiatra.http://www.maep.gov.mg/hautematsiatra.pdf(19 March 2014).

Perreault-Archambault, M. and O. T. Coomes.2008. Distribution of agrobiodiversity in homegardens along the Corrientes River, PeruvianAmazon. Economic Botany 62:109–126.

Pinton, F. and L. Emperaire. 2001. Le manioc enAmazonie brésilienne: diversité variétale etmarché. In: Connaissance et gestion des ressourcesgénétiques (Current knowledge and managementof genetic resources). Genetics Selection Evolution33(Supplement 1):S491–S512.

Radanielina, T., A. Ramanantsoanirina, L.-M.Raboin, J. Frouin, X. Perrier, P. Brabant, andN. Ahmadi. 2013. The original features of rice(Oryza sativa L.) genetic diversity and theimportance of within–variety diversity in thehighlands of Madagascar build a strong casefor in situ conservation. Genetic Resources andCrop Evolution 60:311–323.

Rana, R. B., C. J. Garforth, B. R. Sthapit, A.Subedi, P. Chaudhary, and D. I. Jarvis.2007. On–farm management of rice geneticdiversity: Understanding farmers’ knowledgeon rice ecosystems and varietal deployment.Plant Genetic Resources Newsletter152:58–64.

Serpantié, G. and M. Rakotondramanana. 2013.L’intensification de la riziculture malgache, enpratiques. Cahiers Agriculture 22:401–10.

Serpantié, G., Rasolofoharinoro, and S. Carrière.2007. Transitions agraires, dynamiquesécologiques et conservation: le corridorRanomafana–Andringitra. IRD–CITE, Parisand Antananarivo, Madagascar.

Thurston, H. D., J. Salick, M. E. Smith, P.Trutmann, J. L. Pham, and R. McDowell.1999 . T r ad i t i on a l manag emen t o fagrobiodiversity. Pages 211–243 in D. Woodand J. M. Lenné, eds., Agrobiodiversity:Characterization, Utilization and Manage-ment. CAB International, New York.

Tostain, S. and F. E. Rejo–Fienena. 2010. Lesignames malgaches: une ressource à préserveret à valoriser, Actes du colloque international,July 29–30, 2009. Université de Toliara,Toliara, Madagascar.

Zhu, Y., Y. Wang, H. Chen, and B.-R. Lu. 2003.Conserving traditional rice varieties throughmanagement for crop diversity. Bioscience53:158–162.


http://www.maep.gov.mg/hautematsiatra.pdf

Documents

Origins, Functions, and Persistence of Crop Biodiversity in the Betsileo Highlands, Madagascar