GENETIC EROSION OF CROP POPULATIONS IN CENTERS OF DIVERSITY: A REVISION
Stephen Brush
University of California, Davis

This appraisal of genetic erosion is written from the perspective of a conservationist who emphasizes in situ methods of preserving crop diversity. The appraisal raises issues about the genetic erosion concept that have to do with the ecology of crop populations in centers of diversity. One issue is infraspecific competition between modern and improved varieties and whether an equilibrium is possible. Another issue is the relationship between crop area and diversity and the degree to which decreasing area of traditional crops, or even their �extinction" in particular areas, affects diversity. A third issue concerns the adjustments needed in population biology of crops to accommodate both natural and artificial (conscious) selection. These issues may be of passing interest to ex situ conservation, but they are crucial to the in situ approach. Genetic erosion of crop populations is an ecological concept that draws implicitly on models of ecology and population biology, but this concept has not been theoretically elaborated or empirically tested for crops. The need for theory and data on this topic has been recognized for nearly three decades, but little progress has been made. The concept of genetic erosion of crops can be traced to a period when crop population biology was still in an exploratory stage and before the availability of ecological analysis of crop populations in their centers of diversity. A review and a revision of the concept are long overdue, especially because the concept has important implications for conservation policy. In reviewing the concept of genetic erosion, this paper has three objectives:

1. Evaluate the concept of genetic erosion of crop genetic resources in light of ecological models.
2. Review empirical evidence of genetic erosion.
3. Suggest further research relating to the problem.

The paper begins with a review of the genetic erosion concept as it was developed for crops in centers of diversity. Section II reviews research methods and findings on the three hypotheses of the genetic erosion concept. Section III presents two general models from ecology � niche theory and metapopulation analysis � that bear directly on genetic erosion. The emphasis in Sections II and III is on issues of competition and population maintenance rather than on genetic diversity per se. In developing a theoretical basis and organizing research on genetic erosion, it is helpful to pose the population issues of genetic erosion � competition, fragmentation, migration � as separate from genetic issues � gene flow, minimum viable population, genetic structure of populations. Ultimately, these two issues can be merged into a single model. The research methods and findings and the theoretical discussion that follow in this paper give greater emphasis to the first issue � competition and replacement between landraces and modern crop varieties. The paper concludes with recommendations for further research on the theme of genetic erosion.

1. The Genetic Erosion Concept

Genetic erosion in crops is the loss of variability from crop populations. Variability refers to heterogeneity of alleles and genotypes with their attendant morphotypes and phenotypes. Population can be identified at different spatial levels from a locality to the crop�s global distribution. Genetic erosion implies that the normal addition and disappearance of genetic variability in a population is altered so that the net change in diversity is negative. The manifest cause of genetic erosion is the diffusion of modern varieties from crop improvement programs. Latent causes include population growth, poverty, markets, and cultural change. Genetic erosion affected crops in Europe and North America, and agricultural changes occurring elsewhere in the world are believed to follow a similar trajectory. Key factors in the genetic erosion model are the segmented nature of diversity in crop populations and the negative effect of shrinking the area of local crop populations.

The American plant explorers H.V. Harlan and M. L. Martini (1936) are credited with first recognizing the problem of genetic erosion in crops. A striking thing about Harlan and Martini�s observation is that it occurred relatively early in the deployment of crop science to improve and distribute modern varieties and relatively early in the science of crop population biology. The concept emerged forcefully between 1965 and 1970, in a period when crop improvement had clearly demonstrated its power to transform local crop populations in industrialized countries and in certain less developed regions. Of particular importance was the rapid diffusion of semi-dwarf wheat and rice varieties in Asia, from Turkey to the Philippines, between 1965 and 1970. The definitive statement of the problem of genetic erosion for crop populations was the volume Genetic Resources in Plants (Frankel and Bennett 1970a), resulting from the 1967 FAO Technical Meeting on plant exploration and introduction. Frankel�s (Frankel 1970a, 2) preface lays the groundwork for the subsequent chapters by asserting that �it is now generally recognized (that) many of the ancient genetic reservoirs are rapidly disappearing". Frankel (1985) indicates that he had come to this conclusion in the preparatory phase of the 1967 meetings. Frankel (1985) singles out the contribution of H. Kuckuck as having particular relevance to the emergence of the genetic erosion concept. Ironically, neither Kuckuck�s comments at the 1967 meeting (Bennett 1967) nor his published contributions in Frankel and Bennett (Kuckuck 1970) address genetic erosion.

In fact, the record of the 1967 meeting (Bennett 1967) and Frankel and Bennett (1970a) volume are distinguished by a lack of data and/or analysis of specific case studies of genetic erosion. In the subsequent volume, Crop Genetic Resources for Today and Tomorrow (Frankel and Hawkes 1975), two contributions partly make up for the absence of data. The first is Frankel�s (1973) summary of the 1970/71 FAO Survey of Crop Genetic Resources in their Centres of Diversity. This survey is described as being (of necessity) �somewhat superficial and incomplete" (Frankel 1973, x), relying mostly on the anecdotal observations of plant explorers. The one systematic research effort on genetic erosion reported in the 1970 survey failed to confirm the genetic erosion hypothesis (Hernandez X., 1973). Nevertheless, Frankel (1975) drew the general conclusion that genetic erosion was demonstrated in the survey. The second contribution to present data on genetic erosion was Ochoa�s (1975) report on potato collecting in Chile, Peru and Bolivia. Ochoa confirms reports by others of declines in the number of primitive potatoes found by collectors. On the Chilean island of Chiloé, for instance, collectors found nearly the same number of primitive potato varieties (200) in 1928 and 1938 but �not much more than half" that number in 1948, �even fewer" in 1958, and only 35-40 in 1969 (Ochoa 1975, 167-168). Similar results are reported in northern Peru, where collection in one village yielded 25 native samples in 1955 and none in 1970 (Ochoa 1975, 169).

Efforts to estimate genetic erosion have been stymied by the lack of longitudinal (time-series) data and different taxonomic approaches. Sporadic efforts since 1970 to document genetic erosion have yielded mixed results. Hammer et al. (1995) compared collections in Albania and Italy made at different time periods and concluded that genetic erosion had occurred at similar rates (~72%) in both places. In contrast, cross-sectional research in a number of different centers of diversity suggests that genetic erosion is a problematic hypothesis (Brush 1995). The most frequently cited evidence for genetic erosion is indirect � the diffusion of modern and high yielding varieties into areas once known for crop diversity (Hawkes 1983).

After 30 years of concern for genetic erosion of crop populations in centers of diversity, the concept remains more a presumption of what is likely to occur than a demonstrated fact. Re-reading the Frankel and Bennett (1970a) volume, it is possible to see major unresolved issues that pose serious problems for the concept of genetic erosion. An instance is the taxonomic problems of describing and analyzing crop populations and species, reviewed by Baker (1970) and Harlan (1970) in separate chapters. Nevertheless, the presence of these issues did not lead to discussion of the validity of the genetic erosion concept.

The concept of genetic erosion of primitive crop varieties is elaborated by Frankel (1970a), but it relies strongly on a model of crop population structure in centers of diversity elaborated by others, notably Bennett (1970). Frankel�s (1970) definition rests on five principles:

1. Diversity in crops exists because of adaptation by localized populations.
2. Pre-modern agriculture in centers of diversity is stable.
3. Introduction of modern (exotic) agricultural technology, including modern varieties, is a recent phenomenon and leads to instability.
4. Competition between local (diverse) and introduced varieties results in displacement of local varieties.
5. Displacement of local varieties reduces the genetic variabity of the local crop population.

Bennett (1970) suggests that crop diversity and adaptation are linked in two ways. First, adaptation results in �ecotype" formation and stabilization � the creation of �biotypes possessing a number of adaptive features which are genetically linked" (Bennett 1970, 170). Bennett (1970, 127) labels this �adaptive differentiation," using the same term as Harlan (1970). The differentiation of crop populations reflects the clinal and mosaic patterns of genetic variability in wild plants described by Allard (1970). Second, Bennett suggests that variability exists in cultivated populations because it confers adaptability to the population � an allusion to the diversity/stability concept.

The idea of stability of pre-modern crop populations derives from the supposition of adaptation. Harlan�s (1975: 188) classic definition of landraces describes them as �balanced populations � variable, in equilibrium with both environment and pathogens and genetically dynamic�the result of millennia of natural and artificial selections." The image here is of locally adapted populations that remain relatively stable as long as the local natural and human environments remain stable. The concept of balance repeats the definition given by Frankel and Bennett (1970b, 7) � �land races are crop populations in balance with their environment and remain relatively stable over long periods of time."

In this model, the stability of primitive agriculture remains largely undisturbed, albeit with changes resulting from migration, gene flow and adaptation, until the 20^th Century. The watershed was the rise of modern biology and industrial societies in Europe and North America. The rapid transformation of European and North American agriculture presaged transformations in less developed and genetically more diverse regions that Harlan and Martini (1936) detected. The diffusion of modern crop varieties is repeatedly identified as a critical, destabilizing element in centers of diversity although changes such as improved soil fertility are also recognized.

Frankel�s (1970a) fourth principle of genetic erosion recognizes that rivalry between indigenous and modern crops strongly favors the latter and that these two types are mutually exclusive. Exclusivity results from the fact that indigenous crops are adapted to the conditions of less developed agriculture such as �crude land preparation �(and) low soil fertility" (Harlan 1975, 188). As these conditions change with improved traction and fertilizer, the existing adaptation of landraces turns from asset to liability. Modern cultivars, in contrast, were specifically adapted to higher fertility. The speed of modern cultivar diffusion indicates the determinant role of yield advantage.

Frankel�s (1970a) final principal is a corollary of the first four � that displacement of indigenous crop varieties leads to loss of genetic varibility � or genetic erosion. Local adaptation and exclusivity of type are especially important, so that reference to the presence of modern varieties in a farming system is taken as prime facie evidence of genetic erosion. Indeed, much of the evidence for genetic erosion presented in the 1970/71 FAO survey (Frankel 1973) is data on the diffusion of modern cultivars rather than on the loss of local material (e.g. Kjellqvist 1973).

The concept of genetic erosion of crop genetic resources outlined above was developed by scientists engaged primarily in plant collection for crop breeding and secondarily in research on the botany and systematics of crop species. The observation of genetic erosion was largely anecdotal and only rarely as a result of a focused research effort. The work of Hernandez X. (1973) on maize in Mexico is an exception that disproved genetic erosion. The sense of urgency that imbues Frankel and Bennett (1970a) explains why focused research on the topic was not an early priority, but the collection of genetic resources that occurred after 1967 undermines this explanation as a reason for the subsequent lack of genetic erosion.

One reason for the lack of research on genetic erosion is the difficult demands for data implied in the concept. The concept describes processes that take place over time and which require measurements of biological variability. Time series data is not generally available for crops or agriculture in centers of diversity and different or even incompatible measures of biological variability are used at different times. Data on pre-modern crop populations are rare because the question posed by genetic erosion arose after these populations had been affected by technological change. Agricultural data, for instance, on the extent of modern varieties in centers of diversity was virtually non-existent in 1970, and remains spotty and unreliable. Measures of crop diversity are costly and difficult to acquire, and they are still being developed. Baker (1970) anticipated the difficulty of measuring variability in his discussion of the need for better taxonomy of cultivated plants based on the species concept.

In light of these difficulties in a formal assessment of genetic erosion, two steps can help. First, alternative methods to the direct, longitudinal observation of genetic erosion are needed. Second, a theoretical basis to the concept must be formulated. Both of these steps are attainable, and research to date suggests that the concept of genetic erosion requires revision.

 

II. Cross-Sectional Methods and Findings for Assessing Genetic Erosion

The lack of time-series data on crop populations and agricultural systems is a serious but not insurmountable problem in research on genetic erosion. After all, great progress has been made in understanding crop domestication and evolution without direct observation of the process. The key method is cross-sectional analysis that examines the purported causes of genetic erosion through comparison between farms, villages or regions. For instance, the effect of using modern varieties on the status of indigenous varieties or on their variability can be estimated by comparing farms that use modern varieties at different rates. Obviously, a limitation of cross-sectional research is that it can only be applied in regions where indigenous crop populations still exist. Moreover, this method of investigating genetic erosion works best in situations where both modern and indigenous crops are present. It is not appropriate in places where indigenous crop varieties have disappeared or where modern types are absent:

Cross-sectional research is dependent on a testable model of the processes of agricultural change and technology choice as they relate to crop populations and diversity. The genetic erosion model described above provides three hypotheses that can be tested through cross-sectional comparison.

1. Indigenous crop varieties have limited distribution and slow turnover.
2. Use of purchased inputs and modern crop varieties is inversely correlated with the cultivation of indigenous crop varieties.
3. Adoption of modern crop varieties is inversely correlated with crop population diversity.

These hypotheses are similar to a large class of propositions that social scientists have examined in relation to agricultural development and the diffusion of new technology. Major themes in explaining the patterns of agricultural change include the impact of environmental heterogeneity, social constraints to adoption of technology, the role of risk, and the effect of markets and missing markets. This social science research provides a set of key variables relating to genetic erosion resulting from agricultural development and technology diffusion. These variables are useful in estimating the extent of technology diffusion and the contexts that might limit it. Four classes of social and environmental variables are useful to analyzing the diffusion of new agricultural technology and its impact on local crop populations:

1. farm characteristics [e.g., farm size, parcel number, soil heterogeneity, geographic location (e.g., altitude), and irrigation];
2. household characteristics (e.g., education, wealth, credit availability, age, and labor availability);
3. economic strategies (e.g., commercialization, use of purchased inputs, and off-farm employment);
4. village characteristics (e.g., proximity to markets, presence of agricultural extension agents).

The three hypotheses of genetic erosion listed above have been researched with cross-sectional methods in several centers of crop diversity. Without exception, they are problematic.

1. Indigenous crop varieties have limited distribution and slow turnover.

Ecological research in centers of diversity has found that the crop populations of farms and villages are comprised of varieties that are often widely distributed across different regions and show moderately rapid turnover. In the Peruvian Andes, potato farmers in widely dispersed agro-ecological zones in general geographic regions periodically acquire seed from a limited number of localities in the region (Brush et al. 1981). In central Peru, farms on the dryer western range of the Andes traditionally acquired seed potatoes from a seed market on the wetter eastern range. Brush (1992) and Zimmerer (1991) found that individual native potato varieties are widely distributed in the regions studied. Three of the six species of cultivated potatoes studied showed adaptation to specific altitude zones, but the most common Andean subspecies, Solanum tubersoum subsp. andigena, is best described as a versatile generalist that does well across contrasting agro-ecological zones (Zimmerer 1991). Zimmerer�s (1991) experiments with cultivars of the same species showed that they were only weakly adaptated to elevational environments, contradicting the idea that diversity in this crop is a result of finely tuned adaptation. Native potatoes are frequently moved in mixed lots between fields with different soils, drainage, exposure and altitudes. A significant number of the potato varieties found in one village are found in neighboring villages, and certain native varieties are highly cosmopolitan. At the genetic level, there appears to be a high rate of migration of both alleles and genotypes (Brush et al. 1995). In fact, the migration of alleles among farms was found to be infinite. The plasticity of Andean potato cultivars is suited to the common practices of moving cultivars from field to field with different micro-environments and acquiring seed from distant locations.

Dennis (1987) found that frequent variety turnover was an important part of traditional rice agriculture in Thailand. He reports that indigenous rice varieties are regularly acquired from distant locations. Dennis reports that farmers sow 1.7 varieties per farm and replace them on average every three years. Comparisons of rice accessions found in six administrative districts in the period 1950-61 to those found in 1982-83 reveal the extent of turnover. Of 89 varieties found in 1950-61, only 15 were found in 1982-83, and 82 varieties were found in 1982-83 in these districts that were not present in 1950-61. The rates of variety turnover described by Dennis (1987) for traditional rice agriculture in Thailand are comparable to the rate of turnover of wheat varieties in modern agricultural systems (Brennan and Byerlee 1991).

The Mexican national collection and the CIMMYT international collection each store more than 10,000 accessions of Mexican maize (FAO 1996). This diversity represents local variation within a much smaller number of races and inter-racial populations with broad regional and environmental definitions (Sanchez and Goodman 1992). Louette et al. (1997) conducted intensive research on the distribution and movement of maize in the village of Cuzalapa in the west-central state of Jalisco. Twenty-six varieties are grown in the village, nine �major" and 17 �minor." Six of the major varieties are �local" because they have been grown locally for at least 30 years, and three are �foreign" � recent introductions. All of the minor varieties are �foreign." The six local varieties accounted for 69.4% of the maize area. Table 1 summarizes the data from Louette et al. (1997) relating to the frequency of seed exchange.

The striking finding in Cuzalpa is the relatively high rate of the use of seed lots from other communities (Louette et al. 1997). Fifteen per cent of the seed lots of the predominant local variety were acquired from other communities, and 11% of all seed lots are reported to have come from other regions (Louette et al. 1997). Seed exchange, both within and outside Cuzalapa�s valley, is so important that the authors conclude that this �traditional" community is an open system. They also conclude that seed flow is high enough to mean that no farmer is planting seed stock bequeathed from parents.

Table 1. Origin of maize seed in Cuzalapa

	Own seed	Seed acquired in Cuzalpa	Seed acquired from another community
Local varieties	57.9	34.2	7.9
Foreign varieties (major)	41.2	52.6	5.3
All seed lots	52.9	35.7	11.4
Area	44.9	39.9	15.1

Source: Louette et al. 1997

These studies of seed exchange affirm the fact that �traditional" farming systems are more dynamic and open than allowed in the conventional model of genetic erosion of crop resources. They also suggest that crops in centers of diversity are not assemblages of locally endemic or relatively static populations. Rather, they indicate that landraces are often made up of relatively plastic, �general purpose" (Baker 1965) genotypes whose adaptation is wider than local.

 

2. Use of purchased inputs and modern crop varieties is inversely correlated with the cultivation of indigenous crop varieties.

The genetic erosion model is predicated on the broad adaptability and fertilizer response advantage of modern varieties and the lack thereof in indigenous varieties. However, two common aspects of farming systems in centers of diversity complicate the competitiveness of modern varieties and simple association between use of purchased inputs and decline of indigenous varieties. First, yield is but one of several selection criteria. Second, the weight of fertilizer/yield response may be judged differently by farmers who operate in heterogeneous farming systems with different parcels devoted to the same crop.

Quality and stability have been identified as criteria that influence selection decisions in addition to yield (Bellon 1996b). Quality characteristics include storage, cooking aspects, and demand in the market or in non-market exchange. Numerous ways of using and consuming staple crops exist and may involve complex selection criteria. In Peru, for instance, potatoes are evaluated not only according to production criteria but also on whether they are suited for watia (baking in field ovens), steaming, soups, frying, drying, or freeze-drying (Brush 1992). Chalo potatoes, diverse mixtures of indigenous varieties, are valued as gift items and used as in-kind payments to attract workers at times of peak labor demand. In Mexico, the dietary staple is maize in the form of tortillas. There is no direct link between tortilla quality and maize diversity (Perales 1998), but maize has many more uses in Mexican cuisine than simply tortillas. Hernandez X. (1985) observed that the diversity of maize races within and between regions can be explained partly by use. In his regional study of maize in northern Mexico, Hernandez X (1985) found dents (e.g. Tabloncillo) used for tortillas and fermentation; flints (e.g. Amarillo Cristalino) used for fresh maize, sweetened flour and animal feed; floury types (e.g. Blanco) used for cookies and parched maize soup (pozole); and popcorns (e.g. Chapalote) used for sweetened confections and popping. Maize is ubiquitous in Mexican cuisine, and not merely as tortillas, as shown by the listing of 605 maize dishes by the Mexican National Museum of popular Culture (1982).

A key consideration in weighing the importance of quality characteristics in selection is the ability of markets to provide for the different qualities that concern farmers in centers of diversity. Theoretically a Mexican maize farmer who wants maize for pozole as well as for tortillas can grow only the most profitable type and acquire the other type in the market, thus using his land and labor most efficiently. However, the farmer must be confident that the other type will be available in the market and that the cost of locating and acquiring it will be reasonable. In the farming villages of centers of diversity, such smoothly operating markets are frequently missing or deficient (De Janvry et al. 1994). The appropriate selection decision under such conditions is to trade-off yield for other quality concerns. The existence of a market for purchased in-puts such as fertilizer does not insure that a local market for specific varieties or types of varieties will exist. The result of missing markets is to maintain diverse varieties for household consumption.

The hypothesis that purchased in-puts are inversely related to maintenance of indigenous crops falters on the heterogeneity and fragmentation of farming systems in centers of diversity. Bellon (1996b) points out that farmers have a set of concerns that help explain infraspecific diversity. He divides these concerns into five general categories: (1) environmental heterogeneity (soils, temperature, rainfall, etc.); (2) pests and pathogens; (3) risk management (drought, lodging, frost); (4) culture and ritual; and (5) diet. The first three of these categories are concerned with production conditions in places with steep environmental gradients. Mountainous terrain, lack of inputs to mitigate production constrains (e.g., poor soils, water deficits), and an abundance of co-evolved pests and pathogens contribute to steep environmental gradients in centers of diversity. A result of the steep environmental gradient is the division of local farming systems into different production zones. This is most obvious in mountains, where agriculture is divided into altitude belts or zones with different crops, production activities and intensity of use (Brush 1977; Zimmerer 1996).

A common response to steep environmental gradients is to fragment the farming system into numerous parcels at the farm level. Such fragmentation might follow the contours of production zones in a mountainous landscape, but it also is found where the physiographic factors are less obvious. A consequence of fragmentation is that technology adopted for increasing yields in one zone or parcel may be less effective in others, thus providing a limit to the impact of purchased inputs on selection.

Peru epitomizes a mountainous landscape divided into production zones. The Andes provide a gradient of moisture, temperature, evapo-transpiration, soils, and vegetation that is observed at different scales, from the macro-scale of a trans-Andean transect to the micro-scale of production zones for one crop in a single valley. At the scale of single, village � based farming systems, general production zones (e.g., for cropping and pasture) are usually evident, and a single zone may be sub-divided into specific zones for certain crops or crop varieties. The potato crop is particularly affected by this fragmentation. One common division is by altitude with levels that are worked in different cycles, with different types of the tubers and different treatments (Brush 1992, Zimmerer 1996). A low to moderate altitude potato zone (2,500m � 3,000m) is traditionally planted the short cycle cropping period (August � March), with emphasis traditionally on diploids (e.g. S. phureja) but recently on modern varieties of S. tuberosum and relatively high amounts of fertilizer and pesticides. A mid-altitude potato zone (3,000m � 3,600m) is planted in the long cycle cropping period (August � June), with emphasis on S. tuberosum subsp. andigena and lower amounts of fertilizer and pesticides. A high zone (3,700m � 4,100m) is planted in the long cycle period with emphasis on bitter species (e.g. S. juzepczukii), very little fertilizer or pesticides. As Zimmerer�s (1996) detailed study of one valley in southern Peru shows, these general zones can be subdivided into even smaller areas, depending on micro-environmental characteristics such as slope, drainage and soils.

The Peruvian peasant�s response to this heterogeneous agricultural environment is to farm parcels in different production zones and to use several parcels for the same crop. Typical farm families may have as many as 8 to 12 different parcels for potatoes alone, and in some areas the number is much greater (Brush et al. 1992). The number of parcels per family is partly dictated by the Andean custom of sectoral fallow managed by the community (Orlove and Godoy 1986).

While Peru�s natural heterogeneity provides an environment where farming systems are easily fragmented into production zones and numerous parcels, similar patterns exist in places with moderate environmental gradients. In Mexico, Bellon has shown that soil differences are significant factors in the selection of different types of maize (Bellon and Taylor 1993). His study site in Chiapas has less than 100m difference in elevation and a single climate zone. Nevertheless, the farming system is subdivided into two different sub-systems, arado and pedregal, that are managed differently (Bellon 1996a). Six soil classes are recognized by farmers and the communal land management system is organized so as to distribute good and poor soils evenly among the farm households. Families cultivate an average of 4 maize parcels, distributed between arado and pedregal and among various soil classes. As in Peru, different parcels receive different amounts of purchased inputs and are judged to be more suitable for indigenous or modern maize varieties. Purchased inputs have a high rate of use, for instance fertilizer is used on all arado fields and virtually all (98.7%) pedregal fields. Indigenous varieties persist in this area because they are better adapted to the marginal soils of particular parcels and because they are more tolerant of poorly timed weeding and fertilization that characterizes poorer households (Bellon 1991). Indigenous varieties are described as aguantadora (tough, resistant) because of their ability to thrive on poorer soils and to compete with weeds.

In sum, field research in centers of diversity has shown that purchased inputs do influence selection decisions, but not to the degree suggested in the genetic erosion hypothesis. Selection is influenced by the heterogeneity of a farming system � natural, social, and economic. Yield is an important criteria, but only one of several that are weighed in choosing crops and varieties. The genetic erosion hypotheses fails to anticipate this heterogeneity in farming systems, selection criteria, and market conditions. This failure limits the hypothesis� ability to foresee limits to the diffusion of modern varieties.

 

 

3. Adoption of modern crop varieties is inversely correlated with crop population diversity.

The operative variable in this hypothesis is the decrease in area devoted to indigenous crop varieties as modern ones are adopted. One footing of the genetic erosion hypothesis is that modern varieties are spatially competitive with indigenous ones. In preceding paragraphs, this paper has argued that the heterogeneity of farming systems in centers of diversity limits the diffusion of modern varieties and maintains production spaces for indigenous varieties. Nevertheless, modern varieties have diffused into centers of diversity and caused declines in area devoted to indigenous varieties. A second footing of genetic erosion is that declining area reduces diversity. The genetic erosion hypothesis takes a simple and direct approach to this relationship � smaller area in traditional crops reduces diversity (e.g. Hawkes 1983). However, as long as some areas continue to be planted in indigenous varieties, the relationship between area and diversity is complicated by the population structure of landraces and by the role of conscious (artificial) selection.

Assessing the area/diversity relationship requires judgements on (1) the significance of diversity that might be found and (2) the definition of area. Theoretically, the amount of diversity of a crop population increases to the limits of the population because each individual potentially is a unique genotype. Moreover, we have expanded the measurement of diversity to include morphological characterization, biochemical characteristics (e.g., isozymes), and molecular markers, and this growth offers numerous ways to increase the amount of variation that can be identified in a population. Measuring diversity depends not only on the method used but also on the unit of analysis � population, individual, genome, locus, DNA base sequence (Kresovich and McFerson 1992).

While increased area will always provide new genetic variation, the question is whether adding diversity by increasing area is significant from the perspective of farmers, conservationists, or plant biologists. There are no easy or definitive answers to this question, but the practice of farmers, conservationists, and plant biologists is to limit the amount of diversity that is considered relevant. Farmers may overlook diversity within populations of field crops in applying a single folk name to the population. Conservationists recognize that some diversity is trivial or otherwise not worth capturing (Marshall and Brown 1975), and plant biologists focus primarily on species level differences and above. In other words, expanding area will add diversity, but this diversity may not add significantly to crop germplasm stocks.

The definition of area is likewise troublesome. Crop population area is comprised of nested and scaled units � farms, villages, micro-regions, nations, macro-regions. All but the smallest of these is vague, and all are potentially arbitrary. Environmental, political, cultural and economic criteria are obviously pertinent to defining crop population area, but these criteria do not coincide. In many instances, environments and cultures cross political boundaries, while in others political entities comprise distinct cultures and environments. We might focus on effective crop population area as the proper unit of analysis, defined by isolation from other populations and linked to broadly defined agricultural environments. Isolation and agricultural environment involve physical and ecological parameters, and social factors, such as cultural affiliation, may also count. This type of approach was recommended as �ecogeographic" survey and analysis (IBPGR 1985), but no crop has been thoroughly studied in this manner.

The lack of analysis of effective crop population areas leaves us without an empirical basis to evaluate the impact of declining area on diversity. Farm level studies offer insights, but when populations are parts of active seed flows, diversity measurements may be as much related to sampling as to area. This sampling problem relates to conscious selection and is illustrated by case studies of cassava, sorghum, potatoes, and maize.

Researching the indigenous diversity of cassava (Manihot esculenta Cranz) among the Amuesha people of the Peruvian Amazon, Salick et al. (1997) report a mean of 5.9 cassava varieties per field among 27 fields samples, but one field belonging to the shaman had 53 varieties. Variations of diversity among Amuesha fields may be unimportant to Amuesha farmers as long as the shaman maintains the pool of diversity in his/her field that can be accessed through exchange. Similarly, distribution of diversity of sorghum (Sorghum bicolor) landraces in Ethiopia reported by Teshome et al. (1999) shows that diversity of the smallest fields approaches that of the largest ones. Likewise, potato diversity in the Peruvian Andes is not distributed by field size so much as type of potato according to the farmers� classification (Brush 1992). Andean farmers divide their fields and potatoes into different categories, and some categories and fields are relatively uniform, while others are purposefully diverse. The overall size of a farm is not significantly correlated with diversity because of conscious selection and management of diversity in designated parcels that are a small part of the total potato area (Brush et al. 1992).

The role of conscious selection and crop management on diversity is also evident in maize. Research in Mexico (Louette et al. 1997, Perales 1998) shows that maize diversity (counting varieties) is most concentrated in minor varieties that are grown in kitchen gardens. In Cuzalpa (Jalisco), for instance 14% of the seed lots account for 65% of the maize varieties in the village. Likewise in central Mexico, Perales (1998) reports that one or two white maize varieties are dominant, although farmers commonly keep up to a dozen colored varieties in kitchen gardens. Sampling the normal field area would miss most of the diversity. There is no evidence that diversity was more evenly dispersed in earlier times.

The cassava, sorghum, potato, and maize examples discussed above show that the area/diversity relationship in crops is complicated by conscious selection and management of crop populations. In these cases, seed is handled individually in horticultural fashion rather than in bulk. Horticultural management of seed may increase the effectiveness of conscious selection in maintaining diversity in spite of decreasing area. The effect of conscious selection and management may be less noticeable in field crops in which seed is handled in bulk. Wheat and rice, for instance, may reflect a more direct relationship between area and diversity than the crops previously mentioned. However, measuring field crop diversity at the farm level is hampered by the fact that few varieties are grown on individual farms.

Compiling data on rice diversity in SE Asia, Bellon et al. (1998) report that the average number of rice varieties per farmer is low � usually below two and never higher than three or four. Similarly, research in Turkey�s Transitional Zone revealed that 85% of the households that planted traditional wheat varieties planted only one of them, and households kept an average of less than two wheat varieties while farming an average of seven wheat parcels (Meng 1997). In field crops, therefore, another measure besides variety number is needed to capture intravariety diversity. In our Turkey research, for instance, qualitative characteristics of the populations of wheat were used to generate a Shannon index for each randomly sampled field. Another diversity index using the coefficients of variation of quantitative characteristics was also generated. However, neither index was significantly correlated with the area of the sampled field.

 

4. The Ecology of Genetic Erosion of Crop Populations

The complex concept of genetic erosion appears deceptively simple. Interactions between social and environmental factors and the action of both natural and conscious selection affect the processes of conservation and replacement. It has been assumed that broadly adapted, disease resistant, and high yielding varieties have a natural and irresistible ability to replace locally adapted, disease prone and low yielding varieties, but this assumption overlooks the heterogeneity of farming systems and the many unexpected factors that are important to specific farming people. Historical experience and fieldwork in different cropping systems seems to suggest that there is no definitive pattern of loss. Replacement has occurred in some areas but not in others. Storage, ritual uses, special preparations, cultural identity, market limitations, and many other factors give farmers reasons to conserve local crop varieties in the face of competition from improved varieties. However, this apparent conservation may be deceptive, and the conditions of genetic erosion may be in place in centers of diversity. In the face of this complexity, we need to formulate a more explicit model of genetic erosion that refers to both natural and social processes of selection.

Population models from ecology can be useful to our assessment of genetic erosion of crops because these models also address the issues of diversity, competition, and habitat change. Two approaches to explaining patterns of species diversity � niche theory and metapopulation analysis � are especially relevant to modeling genetic erosion. Niche theory is useful to modeling the conditions that induce genetic erosion, and metapopulation analysis can be used to describe the structure and maintenance of crop populations.

 

III.a. Niche theory

One approach to diversity is found in niche theory that directs us to environmental heterogeneity as a source of diversity (Whittaker and Levin 1975). Niche is defined by the status of the species in its community, for instance by the size of an animal and its food habits (Elton 1927). In more contemporary terms, a niche is the multidimensional space that is unique and exclusive to every species (Hutchinson 1957). This space defines the limiting factors that determine a species� distribution and range. Competition between species in single environments is a major concern of the theory. The Lotka-Volterra competition model developed in the 1920s and 1930s formalized Gause�s principle of competitive exclusion and was summarized by Hardin (1960, 1292) as �complete competitors cannot coexist" (see Whittaker and Levin 1975). The competitive exclusion principle rests on the possibility of occupying the same niche within the same geographic territory (sympatric existence). A secondary condition is that the two species have unequal reproduction rates (Hardin 1960). The formal model of competitive exclusion as having three assumptions (Gotelli 1998):

1. Resources are in limited supply.

1. Competition coefficients and carrying capacities are constants.
2. Density dependence is linear.

Although competitive exclusion implies a zero-sum solution, equilibrium outcomes are also possible.

Niche theory was fully elaborated at the time when concern arose over genetic erosion of crop populations, and the theory is a logical source of the ideas that framed the genetic erosion concept. Niche theory�s stress on competition and exclusion is strongly echoed in the genetic erosion concept. Crops and crop varieties in farming systems are easily treated as analogies to species and natural habitats. The production space of a particular crop variety in a farming system resembles a niche with specific environmental limits and carrying capacity. The introduction of competitive crops or varieties into this space will negatively impact the existing varieties.

However, niche theory and its attendant assumptions and conditions are not formally recognized in the literature on genetic erosion, and examination of these assumptions and conditions alerts us to possible limitations of the niche model for analyzing population processes in crops. The competitive exclusion model was developed for interspecies interaction where natural selection but not artificial selection operates. Competition over essential resources and space affects resource availability to competing species, their reproduction and mortality rates, population densities, and carrying capacities (Tilman 1982). Conscious selection fundamentally alters these variables. For crops, competition is not for essential resources per se but for space in the farming system that is allocated by farmers. In other words, competition among crops is for the farmer�s favor. Niche theory is meant to address interspecies competition and not intraspecies competition. Indeed, intraspecific interaction may reduce the competitive exclusion principle to a carrying capacity model (Gotelli 1998). Nevertheless, applying the analogies of species to crop varieties (e.g., high yielding, traditional) and farm habitat to environmental niche allows us to use niche theory and to predict genetic erosion caused by the appearance of a superior competitor in a variety�s production space within a farming system.

Application of niche theory to competitive interaction between a local variety and a modern variety in the same farm habitat gives four possible solutions:

1. The local variety wins the competition and no genetic erosion occurs.
2. The modern variety wins the competition and the local variety disappears from the niche (local extinction).
3. The two varieties establish as stable equilibrium with uncertain genetic erosion results.
4. The two varieties establish an unstable equilibrium leading to the eventual exclusion of one.

The solution to competition between two species depends on the carrying capacity of each species and the competition effect of each species on the other. If the isoclines that define the combination of abundances of each species interacting with the other are symmetrical (parallel), then the first solutions pertain. If the isoclines cross, then the third and fourth solutions are possible.

Recent research on the concept of niche and its relation to biological diversity has disputed the earlier emphasis on competitive exclusion. Equilibrium solutions to competition require a set of assumptions about the even distribution of resources within an environment. Including tradeoffs between competitive ability, stress tolerance, and reproductive performance in niche models solves them in favor of coexistence of different species rather than exclusion (Grace 1990). Building on the work of others, Tilman (1982) suggests that the classical niche theory of the Lotka-Volterra model should be reformulated to account for resource requirements of different species. Tilman�s (1982) model predicts that the species with the lowest minimum resource requirement will be the superior competitor. For crops, this might be either use of resources, such as soil nutrients, or per unit cost of production. If use resources is adopted, traditional varieties might be superior competitors. If per unit cost is adopted, modern varieties would be superior. Competition can, however, be defined differently � growth capacity, reproduction, dispersal. Moreover, competition is not limited to single resources and is influenced by non-resource conditions such as disturbance (Grace 1990).

Tilman (1994) further elaborates his resource based competition model to account for spatially structured habitats, and this elaboration predicts coexistence rather than competitive exclusion. In other words, spatial structure challenges the predicitons of nonspatial models such as Lotka-Volterra. Spatial structure caused by habitat subdivision can occur in any environment, even though underlying physical conditions are similar (Tilman 1994). Habitat subdivision is inevitable because of random colonization and mortality. Coexistence of numerous species in the same general environment results from inevitable trade-offs between allocation of plant energy to roots and reproduction. Tilman (1994) observes that superior competitors for limited soil resources are poorer colonizers. The spatially structured model reveals a set of simplifying assumptions of the competitive exclusion models � �that resource supply rates and physical factors are spatially homogeneous, that each organism is spread uniformly throughout the environment, that resources do not fluctuate, that localized mortality does not occur, and that higher trophic levels are unimportant" (Tilman and Pacala 1993: 19). Violation of even a few of these assumptions and accepting trade-offs in response to resource limits and reproduction lead to models that predict the coexistence of unlimited species (Tilman and Pacala 1993). The surprising conclusion of spatially structured models is to turn niche theory on its head and to allow for numerous species in single environments.

Both the older and more recent versions of niche theory provide useful insights to the diversity of crop varieties in centers of crop diversity and to competition between local and improved varieties in these centers. The classical view of niches as mulitdimensional spaces where competitive exclusion is probable explains the diversity of crop varieties. As discussed above, local farming systems in centers of diversity often contain different niches in the form of micro-environments and production zones. Crop niches are distinguished by physical and agronomic factors � soils, water availability, slope, temperature, and exposure to stresses such as wind, frost, and disease. Competitive advantage of particular varieties within the multidimensional space of each agricultural niche appears as yield potential and yield stability. The fact that different varieties have yield advantages in different niches is recognized and exploited by farmers, resulting in diversity. It is possible to add social dimensions to this definition of niche as Bellon (1996b) has done in his discussion of �farmers� concerns." Concern for ritual or other cultural characteristics of the crop may, for instance, be factored into the calculation of yield, or this concern may be added as an additional niche dimension to make a new niche.

The likelihood of genetic erosion also follows readily from the application of classical niche theory to explain crop diversity in centers of origin. The availability of broadly adapted modern varieties and inputs that smooth out micro-environmental differences in a farming system conform to classical niche theory. Modern varieties are superior competitors to local varieties, especially in an environment that is homogenized by the application of external inputs. If a crop is planted to its maximum density or carrying capacity in a uniform environment, then the competitive advantage of modern varieties will give them an isocline above that of the local variety, resulting in the competitive exclusion of the local type.

Historical change in crops and crop diversity in developed countries and in certain underdeveloped regions appears to conform to the predictions of classical niche theory. Given this experience, it is reasonable to apply niche theory to other underdeveloped regions, such as centers of diversity. However, both theoretical reasons and experience suggest that direct application of classical niche theory to centers of diversity is inappropriate. As noted above, niche theory involves a number of assumptions. Most importantly, it assumes that competition coefficients and carrying capacities of competing species are constant. Tilman (1982) observed that the constant competition coefficient of classical niche theory leads to models in which resource availability is also constant, and he adjusted the model to account for reduction of resources as they are used (Grace 1990). Moreover, the introduction of trade-offs in the model � between competitive growth advantage, stress tolerance, and reproduction � opens the door to competition coefficients and carrying capacities that vary in structured environments. Tilman�s (1994) later and more extensive assessment of classical niche theory shows the near certainty of violating the assumptions of the Lotka-Volterra competitive exclusion model of classical niche theory.

The probability that farming systems will likewise violate the assumptions of niche theory must be weighed. A set of assumptions from classical niche theory is implied in predictions of genetic erosion in crops. These assumptions are similar to the ones noted by Tilman and Pacala (1993):

• resource supply rates and physical factors in the modernized farm system will be spatially homogeneous;
• resources in the modern farm system do not fluctuate;
• localized mortality of crop varieties does not occur;
• higher trophic levels (e.g. human consumption or predation) are unimportant.

Following Tilman and Pacala�s (1993) logic, violation of several of these leads us to expect coexistence of different crop varieties (diversity) rather than competitive exclusion.

Field research on the ecology of landraces in centers of diversity where modern cultivars are also present provides ample evidence that the conditions of competitive exclusion are, indeed, violated. Habitats are highly structured, as evidenced by the fragmentation of fields and the practice of cultivating numerous fields in the same crop. Physical heterogeneity is not eliminated by the application of modern inputs, for instance in hill land and dry land agriculture. Resource supply rates to different fields are inevitably variable. Resources fluctuate within and between years according to the availability and timing of modern inputs, especially in locations that are removed from major transportation and input supply systems. Crop failures (localized mortality) under stress and in marginal environments are well known to farmers. Finally, human consumption is a critical element in crop selection, particularly in subsistence production. Storage, taste, cooking qualities, and use as fodder may not be more important than yield, but they are relevant to selection. In sum, the assumptions of the classical niche model are easily violated in farming systems in centers of diversity, so that we should expect coexistence rather than competitive exclusion to prevail. The fieldwork cited above tends to confirm this expectation.

 

III.b. Metapopulation analysis

The reduction of area and fragmentation of populations threaten species� survival and biological diversity in many different environments. Understanding the effect of area reduction and fragmentation on species survival was a principal objective of island biogeography (MacArthur and Wilson 1967). Island biogeography turns our attention to species in degraded environments. In particular, this approach was designed to understand patches that remain in disturbed environments, some so small as to be islands. Others are large enough to be thought of as �mainland" � areas that resemble habitats before degradation. Island biogeography would appear to offer much to modeling crop populations in centers of diversity that are experiencing agricultural development. In this analogy, habitat degradation is the diffusion of modern crop varieties into systems that previously were fully planted in landraces. Villages and farms that maintain landraces are analagous to islands surrounded by a sea of modern varieties. A problem with this analogy, however, is defining what might constitute a �mainland." Like niche theory, there seems to be no direct application of this theoretical framework to describing crop populations, although the concept is mentioned in reference to establishing biological preserves for wild crop relatives (Frankel et al. 1995).

After three decades of research on island biogeography, ecologists have developed a new approach � metapopulation analysis (Levins 1969; Hanski and Simberloff 1997). This approach seems to be particularly useful to examining crop populations and the issue of genetic erosion. A metapopulation is defined as �a �population� of unstable local populations, inhabiting discrete habitat patches" (Hanski 1998: 41). Each local population is extinction-prone, but migration from other local populations allows the species to re-establish itself where it has suffered extinction. Extinction is a recurrent event many patches that is moderated by colonization to establish an equilibrium for the metapopulation. This approach has been utilized to study species in fragmented and degraded landscapes and in competition with other species. An advantage to metapopulation analysis is its treatment of the concept of mainland. From the vantage point of any one island, the mainland is merely the sum of other islands that might provide colonizing populations.

The usefulness of metapopulation analysis to understanding crop population dynamics has been noted by crop ecologists (Zimmerer 1998; Louette et al. 1997). The analogy of the crop population of a farm or village to local population is intuitively appropriate. Farmers may keep seed from year to year, but loss of seed and its replacement through exchange with other farms and across different regions is a recurrent event. In other words, individual farms experience local extinction of seed but local extinction is balanced by seed exchange (migration) among farms. The habitat of seed production and exchange is degraded by the adoption of modern varieties in that the number of sources of seed (patches, islands) are reduced and their connectivity is decreased. In other words, landrace populations are appropriately described as metapopulations. While the extinction of landraces in a single farm or village may not threaten the entire landrace, the extinction of a metapopulation of the landrace is possible as the habitat of the landrace is degraded by modernization. Dispersal rates may be part of the nature of landraces to replenish seed stocks of traditional farmers, but these dispersal rates may not be able to adjust or adapt to the modernization of agriculture.

A potential advantage of metapopulation analysis is to help bridge different approaches to genetic erosion. Hanski (1998) notes that in large-scale spatial ecology, metapopulation ecology occupies an intermediate position between theoretical ecology and landscape ecology. Theoretical ecology has demonstrated the complexity and spatial patterns of population processes that occur in idealized, homogeneous environments. At the other extreme, landscape ecology has described the complex nature of real environments but without a well-developed theoretical framework (Hanski 1998). Metapopulation analysis provides a means to link these two approaches by providing a theoretical framework to analyze population dynamics in heterogeneous environments.

For crop ecology relating to genetic erosion, the original models of genetic erosion presented in the beginning of this paper are analogous to theoretical ecology in that they are conceptualized without environmental heterogeneity. They do not allow for the mitigation of bio-physical and social factors in the process of genetic erosion. Ecological research on real farming systems where landraces and modern varieties compete is analogous to landscape ecology. This research has shown that the theoretical models of genetic erosion are difficult to apply and may be wrong, but this ecological fieldwork has not provided a convincing theoretical framework. What is now needed is a concerted effort to develop the middle ground approach suggested by metapopulation analysis.

Two premises underlie the metapopulation concept (Hanski and Simberloff 1997):

1. Populations are spatially structured into assemblages of local breeding populations.
2. Migration among local populations affects local dynamics, including reestablishment after local extinction.

Metapopulation analysis expands on earlier research on niches and competition by adding local extinction and colonization to the population dynamics of a species. Metapopulation analysis serves to describe populations in both pristine and degraded habitats. It is based on the fact that a species is comprised of numerous distinct breeding populations. Environmental patches experience periodic local extinction and re-colonization. Whereas island biogeography is envisioned as the interaction of island and mainland populations, metapopulation analysis is based on the possibility that the entire population of a species exists on islands scattered throughout the landscape. In this approach, persistence of regional population and possibilities of migration and re-colonization are major concerns rather than population size at a specific site. Critical variables are the fraction of population sites that are occupied, the extinction rate across patches, and the migration rate among them.

Metapopulation analysis reorients our perspective on competition and local extinction in two ways. First, multiple patches of a population spread the risk of extinction to the species. The mathematics of metapopulations are such that relatively high probabilities of extinction at the local level may be greatly reduced at the regional level. In Gotelli�s (1998) model, for instance, a 70% probability of extinction of a local population, if shared evenly among 10 different populations, is reduced to a 97% chance of survival of at least one population. As more patches are added, the probability of regional persistence increases rapidly (Gotelli 1998).

The second reorientation of metapopulation analysis is, however, far less optimistic. The ability of a species to survive depends on the persistence of a minimum number of patches. Although individual patches may persist, if their number is less than the �eradication threshold" the species is bound for extinction. Population size at a particular site is not a critical variable to a metapopulation�s survival, but filling a minimum number of patches is required for the population to persist. This minimum is defined as the number of unoccupied niches when the metapopulation is at �a dynamical equilibrium between colonization and extinction" (Nee et al. 1997: 125). Nee (1994) defines this minimum threshold as the unused amount of patches of a population�s suitable habitat. Failure to fill the minimum number of patches establishes an �extinction debt" (Tilman et al. 1994) that must eventually be paid.

Hanski (1998) summarizes three general conclusions about metapopulation response to habitat destruction:

1. Metapopulations respond to habitat destruction in a non-linear fashion because habitat destruction is itself non-linear as are its effects on migration.
2. A time lag (extinction debt) occurs between the time of habitat destruction and metapopulation decline.
3. The amount of empty habitat in a landscape before habitat destruction equals the extinction threshold, or the minimum amount of habitat required for metapopulation persistence.

These conclusions direct us to landscape heterogeneity, to population patterns (e.g., replacement) over different environments, and to how populations are connected. Habitat destruction is non-linear because habitats are heterogeneous. The survival of local populations does not mean that metapopulations will survive. Metapopulation survival may be estimated from understanding the connections between local populations and the degree to which those connections have been maintained or disrupted.

Researchers have noted that metapopulation analysis is appropriate for understanding crop population processes in centers of diversity (e.g., Zimmerer 1998; Louette et al. 1997), but the approach has not been formally adopted in any field study or theoretical exercise relating to crop dynamics. Landraces appear to conform to a metapopulation model in that they are naturally fragmented into unstable local populations that are connected through seed exchange among farmers, villages, and regions. Local extinction and re-establishment appears to be common in crop populations. In Mexico, for instance, farmers of maize landraces don�t worry about seed loss or conservation because they believe that seed will be available from neighbors or in markets (Perales 1998). Landraces exist as heterogeneous populations rather than as pure-line varieties, and this population heterogeneity may be due to frequent exchange and incorporation of new seed into existing stocks. Ecological research has also demonstrated that environmental heterogeneity is important in population dynamics of landraces, including habitat destruction caused by the introduction of modern crop varieties. Landrace populations are stable in some environments because habitat destruction does not approach the threshold of the minimum habitat needed to maintain metapopulations. Maize in Mexico (Perales 1998; Louette et al. 1997) seems to conform to this description. In other cropping systems, such as native Andean potato production and Anatolian wheat production, habitat destruction of landraces � measured by the diffusion of modern potato and wheat varieties � may be at or beyond the threshold where metapopulations of local crops can survive. One research objective defined by a metapopulation analysis of landraces would be to account for the patchy nature of their undisturbed habitats in order to determine the equilibrium level of the landrace metapopulation � their extinction threshold. In other words, research on landrace survival in habitats where modern varieties are present will benefit from research on agricultural habitats where only landraces exist.

IV. Conclusion

An assessment of genetic erosion of crop populations in centers of crop diversity is at an impasse. On one hand, general but under-theorized models of agricultural change present a logical argument for genetic erosion. This argument, however, has not been rigorously tested, and it does not contend with environmental heterogeneity � either social or bio-physical � in farming systems. On the other hand, specific but equally under-theorized descriptions of landraces that compete with modern varieties suggest that survival of crop diversity is plausible in certain environments. This impasse between contradicting general models and actual field data can be solved by developing a better theoretical framework for the ecology of crops in centers of diversity and genetic erosion. This framework needs to be explicitly interdisciplinary, drawing on theory, methods, and data from both crop and social science.

This paper has presented ecological models that might serve to generate a more robust crop ecology. Niche theory and metapopulation analysis offer numerous insights and advantages to efforts to understand genetic erosion. One insight is that general population processes are affected by environmental heterogeneity. An advantage of these approaches is that they incorporate this heterogeneity. Both niche theory and metapopulation analysis offer a middle ground between general theory and site specificity.

A limitation of both niche theory and metapopulation analysis for crop ecology is that neither was conceived for conscious (artificial) selection. Another limitation is that both were designed to deal with interspecific competition and processes. Numerous questions surround the use of ecological models for crops. Is it appropriate, for instance, to treat types of crop varieties (e.g., landraces and modern varieties) as though they are species? Should we look at modern varieties as though they are invasive species? Is agricultural modernization equivalent to habitat destruction?

The application of formal population models to crops presents daunting obstacles in the way of defining key variables and relationships. The need to include both biological and social variables and functional relationships is particularly difficult to satisfy. Social science offers insights and methods that are needed in any general ecology of crop populations in centers of diversity, such as models of the diffusion of modern crop varieties. However, these insights and methods were not conceived for ecological research. Thus, there is a need to re-work both ecology and social science theory and methods for studying crop populations.

 

5. Acknowledgments

I wish to thank Ecology Graduate students at the University of California � Davis who contributed to this paper. Ben Shouse noted the applicability of several models from population biology, and Dawit Tadessee suggested the metapopulation approach before it appeared in the crop ecology literature. My gratitude to Margaret Brush for comments on the paper. The U. S. National Science Foundation supported research in Peru, Mexico and Turkey that is reported here.

VI. References

Allard, R. W. 1970. Population structure and sampling methods. . In Genetic Resources in Plants - Their Exploration and Conservation. O. H. Frankel and E. Bennett, eds. Oxford: International Biological Programme Handbook No, 11. Blackwell Scientific Publications. Pages 97-113.

Baker, H. G. 1970. Taxonomy and the biological species concept in cultivated plants. In Genetic Resources in Plants - Their Exploration and Conservation. O. H. Frankel and E. Bennett, eds. Oxford: International Biological Programme Handbook No, 11. Blackwell Scientific Publications. Pages 49-68.

Baker, H. G. 1965. Characteristics and modes of origins of weeds. In Genetics of colonizing species. G. L. Stebbins, ed. New York: Academic Press. Pages 17-168.

Bellon, M. R., 1991. The ethnoecology of maize variety management: a case study from Mexico. Human Ecology 19: 389-418.

Bellon, M. R. 1996a. Landholding fragmentation: are folk soil taxonomy and equity important? A case study. Human Ecology 24: 373-393.

Bellon, M. R. 1996b. The dynamics of crop infraspecific diversity: A conceptual framework at the farmer level. Economic Botany 50: 26-39.

Bellon, M. R., Brar, D. S., Lu, B. R., and Pham, J. L. 1998. Rive genetic resources. In Sustainability of Rice in the Global Food System. N. G. Dowling, S. M. Greenfield, and K. S. Fischer, eds. Davis, CA: Pacific Basin Study Center and Manila: International Rice Research Institute. Pages 251-283.

Bellon M. R. and Taylor J. E. 1993. Farmer soil taxonomy and technology adoption. Economic Development and Cultural Change 41: 764-786.

Bennett, E., ed. 1967. Record of the FAO Technical Conference on the Exploration, Utilization and Conservation of Plant Genetic Resources. Rome, Food and Agricultural Organization of the U. N.

Bennett, E. 1970. Adaptation in wild and cultivated plant populations. In Genetic Resources in Plants - Their Exploration and Conservation. O. H. Frankel and E. Bennett, eds. Oxford: International Biological Programme Handbook No, 11. Blackwell Scientific Publications. Pages 115-129.

Brennan, J. P. and Byerlee, D. 1991. The rate of crop varietal replacement on farms: measures and empirical results for wheat. Plant Varieties and Seeds 4: 99-106.

Brush, S. B. 1977. Mountain, Field and Family: The Economy and Human Ecology of an Andean Valley. Philadelphia: The University of Pennsylvania Press.

Brush S. B. 1992. Ethnoecology, biodiversity, and modernization in Andean potato agriculture. Journal of Ethnobiology 12: 161-85.

Brush, S. B. 1995. In situ conservation of landraces in centers of crop diversity. Crop Science 35: 346-354.

Brush, S. B., Carney, H. J., and Huamán, Z. 1981. Dynamics of Andean Potato Agriculture. Economic Botany 35: 70-85.

Brush, S. B., Taylor, J. E. and Bellon, M. R. 1992. Biological diversity and technology adoption in Andean potato agriculture. Journal of Development Economics 39: 365-387.

Brush S. B., Kesseli, R., Ortega, R., Cisneros, P., Zimmerer, K. S., Quiros, C. 1995. Potato diversity in the Andean center of crop domestication. Conservation Biology 9: 1189-98.

DeJanvry A., Fefchamps, M., Sadoulet, E., 1991. Peasant household behavior with missing markets - Some paradoxes explained. Economic Journal 101: 1400-1417.

Dennis, J. V. 1987. Farmer management of rice variety diversity in northern Thailand. Unpublished Ph.D. dissertation, Cornell University. UMI No. 8725764. Ann Arbor MI: University Microfilms International.

Elton, C. 1927. Animal Ecology. London: Sidgwick & Jackson.

FAO. 1996. Report on the State of the World�s Plant Genetic Resources. Rome: Food and Agricultural Organization of the U. N.

Frankel, O. H., 1970a. Genetic conservation in perspective. In Genetic Resources in Plants - Their Exploration and Conservation. O. H. Frankel and E. Bennett, eds. Oxford: International Biological Programme Handbook No, 11. Blackwell Scientific Publications. Pages 469-489.

Frankel, O. H., 1970b. Preface. In Genetic Resources in Plants - Their Exploration and Conservation. O. H. Frankel and E. Bennett, eds. Oxford: International Biological Programme Handbook No, 11. Blackwell Scientific Publications. Pages 1-4.

Frankel, O. H., ed. 1973. Survey of Crop Genetic Resources in Their Centres of Diversity, First Report. Rome: Food and Agricultural Organization of the U. N.

Frankel. O. H. 1985. Genetic resources: the founding years. Diversity Issue number 7 (Fall 1985): 26-29.

Frankel, O. H. and Bennett E., eds. 1970a. Genetic Resources in Plants - Their Exploration and Conservation. Oxford: International Biological Programme Handbook No, 11. Blackwell Scientific Publications.

Frankel, O. H. and E. Bennett, 1970b. Genetic resources � introduction. In Genetic Resources in Plants - Their Exploration and Conservation. O. H. Frankel and E. Bennett, eds. Oxford: International Biological Programme Handbook No, 11. Blackwell Scientific Publications. Pages 7-17.

Frankel, O. H. Brown, A. H. D., and Burdon, J. J. 1995. The Conservation of Plant Biodiversity. Cambridge: Cambridge Univeristy Press.

Frankel, O. H. and Hawkes, J. G., eds. 1975. Crop Genetic Resources for Today and Tomorrow. International Biological Programme 2. Cambridge: Cambridge University Press.

Gotelli, N. J. 1998. A Primer of Ecology, Second Edition. Sunderland, MA: Sinaur Associates, Inc.

Grace, J. B. 1990. On the relationship between plant traits and competitive ability. In Perspectives on Plant Competition. J. B. Grace and D. Tilman, eds. San Diego: Academic Press. Pages 51-65.

Hammer, K., Knupffer, H., Xhuveli, L., and Perrino, P. 1996. Estimating genetic erosion in landraces - two case studies. Genetic Resources and Crop Evoution 43: 329-336.

Hanski, I. A. 1998. Metapopulation dynamics. Nature 396: 41-49.

Hanski, I. A. and Gilpin, M. E., eds. 1997. Metapopulation Biology: Ecology, Genetics, and Evolution. San Diego: Academic Press.

Hanski, I. and Simberloff, D. 1997. The metapopulation approach, its history, conceptual domain, and application to conservation. In Metapopulation Biology: Ecology, Genetics, and Evolution. I. Hanski and D. Simberloff, eds. San Diego: Academic Press. pages 5-26.

Hardin, G. 1960. The competitive exclusion principle. Science 131: 1292-1297.

Harlan, H. R. and Martini, M. L. 1936. Problems and results of barley breeding. USDA Yearabook of Agriculture. Washington DC: US Government Printing Office. Pages. 303-346.

Harlan, J. R. 1970. Evolution of cultivated plants. In Genetic Resources in Plants - Their Exploration and Conservation. O. H. Frankel and E. Bennett, eds. Oxford: International Biological Programme Handbook No, 11. Blackwell Scientific Publications. Pages 19-32.

Harlan, J. R. 1975. Our vanishing genetic resources. Science 188: 618-621.

Hawkes, J. G. 1983. The Diversity of Crop Plants. Cambridge, MA: Harvard University Press.

Hernandez X., E. 1973. Genetic resources of primitive varieties of Mesoamerica, Zea spp., Phaseolus spp., Capsicum spp., Cucurbita spp. In. Survey of Crop Genetic Resources in Their Centres of Diversity, First Report. Frankel, O. H., ed. Rome: Food and Agricultural Organization of the U. N. Pages 76-86.

Hernandez X., E. 1985. Maize and man in the greater Southwest. Economic Botany 39: 416-30.

Hutchinson, G. E. 1957. Concluding remarks. Cold Spring Harbor Symposium on Quantitative Biology 22: 415-427.

International Board for Plant Genetic Resources (IBPGR). 1985. Ecogeographical Surveying and In Situ Conservation of Crop Relatives. Rome: IBPGR Secretariat.

Kjellqvist, E. 1973. Near East � Cereals, Turkey, Syria, Iraq, Iran, Pakistan. In Survey of Crop Genetic Resources in Their Centres of Diversity, First Report. Frankel, O. H., ed. Rome: Food and Agricultural Organization of the U. N. Pages 9-21.

Kresovich, S. and J. R. McFerson. 1994. Assessment and management of plant genetic diversity: considerations of intra- and interspecific variation. Field Crops Research 29: 185-204.

Kuckuck, H. 1970. Primitive wheats. In Genetic Resources in Plants - Their Exploration and Conservation. O. H. Frankel and E. Bennett, eds. Oxford: International Biological Programme Handbook No, 11. Blackwell Scientific Publications. Pages 249-266.

Levins, R. 1969 Some demographic and genetic consequences of environmental heterogeneity for biological control. Bulletin of the Entomological Society of America 15: 237-240.

Louette D., Charrier A., and Berthaud J. 1997 In situ conservation of maize in Mexico, genetic diversity and maize seed management in a traditional community. Economic Botany 51: 20-39.

Marshall, D. R. and Brown, A. H. D. 1975. Optimum sampling strategies in genetic conservation. In Crop Genetic Resources for Today and Tomorrow, O. H. Frankel and J. G. Hawkes. eds. International Biological Programme 2. Cambridge: Cambridge University Press. Pages 53-80.

McArthur, R. H. and E. O. Wilson. 1967. The Theory of Island Biogeography. Princeton NJ: Princeton University Press.

Meng, E. 1997 Land Allocation Decisions and In situ Conservation of Crop Genetic Resources: The Case of Wheat Landraces in Turkey. Unpublished Ph.D. Dissertation, University of California, Davis. University Microfilms, Ann Arbor, Michigan.

Museo Nacional de Culturas Populares. 1982. El Maiz, Fundamento de la Cultural Popular Mexicana. Mexico, DF: Museo Nacional de Culturas Populares.

Nee, S. 1994. How populations persist. Nature 367: 123-124.

Nee, S., May, R. M., and Hassell, M. P. 1997. Two-species metapopulation models. In Metapopulation Biology: Ecology, Genetics, and Evolution. I. Hanski and D. Simberloff, eds. San Diego: Academic Press. pages 123-147.

Ochoa, C. 1975. Potato collecting expeditions in Chile, Bolivia and Peru, and the genetic erosion of indigenous cultivars. In Crop Genetic Resources for Today and Tomorrow. International Biological Programme 2. O. H. Frankel and J. G. Hawkes, eds. Cambridge: Cambridge University Press. Pages167-173.

Orlove, B. S. and Godoy, R. 1986. Sectoral fallow systems in the central Andes. Journal of Ethnobiology 6: 169-204.

Perales, H. 1998. Conservation and Evolution of Maize in the Valleys of Amecameca and Cuautla, Mexico. Unpublished Ph.D. Dissertation. University of California, Davis. University Microfilms, Ann Arbor, Michigan.

Salick, J., Celinese N., and Knapp, S. 1997. Indigenous diversity of cassava: generation, maintenance, use and loss among the Amuesha, Peruvian Upper Amazon. Economic Botany 51: 6-19.

Sanchez J. J. and Goodman, M. M. 1992. Relationships among Mexican races of maize. Economic Botany 46: 72-85.

Teshome, A., Fahrig, L., Torrance, J. K., Lambert, J. D., Arnason, T. J., and Baum, B. R. 1999. Maintenance of sorghum (Sorghum bicolor, Poaceae) Landrace diversity by farmers� selection in Ethiopia. Economic Botany 53: 79-88.

Tilman, D. 1982. Resource Competition and Community Structure. Princeton NJ: Princeton University Press.

Tilman, D. 1994. Competition and biodiversity in spatially structured habitats. Ecology 75: 2-16.

Tilman, D., May, R. M., Lehman, C. L., and Nowak, M. A. 1994. Habitat destruction and the extinction debt. Nature 371: 65-66.

Tilman, D. and S. Pacala 1993. The maintennce of species richness in plant communities. In Species Diversity in Ecological Communities: Historical and Geographical Perspectives. Chicago: University of Chicago Press. Pages 13-25.

Whittaker, R. H. and Levin, S., eds. 1975. Niche Theory and Application. Benchmark Papers in Ecology 3. Stroudsburg, PA: Dowden, Hutchinson & Ross, Inc.

Zimmerer, K. S. 1991. The regional biogeography of native potato cultivars in highland Peru. Journal of Biogeography 18: 165-178.

Zimmerer, K. S. 1996. Changing Fortunes: Biodiversity and Peasant Livelihood in the Andes. Berkeley: University of California Press.

Zimmerer K. S. 1998: The ecogeography of Andean potatoes. BioScience 48: 445-454.