INTRODUCTION
Quinoa (Chenopodium quinoa Willd.), an Andean grain crop, has recently gained worldwide attention because of its nutritional value and functional features (Bhargava et al. Reference Bhargava, Shukla and Ohri2006; Hirose et al. Reference Hirose, Fujita, Ishii and Ueno2010; Vega-Gálvez et al. Reference Vega-Gálvez, Miranda, Vergara, Uribe, Puente and Martínez2010), as well as its ability to grow under conditions of soil salinity, drought, frost and a wide pH range (Jacobsen et al. Reference Jacobsen, Mujica and Jensen2003; Fuentes & Bhargava Reference Fuentes and Bhargava2011). Because of these characteristics, quinoa has been introduced to new areas outside its native region, especially in Europe and the subtropical regions of the world, where it has provided good yields and so demonstrated the potential of quinoa as a grain and also as a fodder crop (Mujica et al. Reference Mujica, Jacobsen, Izquierdo and Marathee2001; Casini Reference Casini2002; Jacobsen Reference Jacobsen2003; Bhargava et al. Reference Bhargava, Shukla and Ohri2007a ; Pulvento et al. Reference Pulvento, Riccardi, Lavini, D'andria, Iafelice and Marconi2010).
Quinoa has been cultivated for c. 7000 years in the Andes of South America (Jacobsen Reference Jacobsen2003; Mujica Reference Mujica, Sepúlveda, Thomet I., Palazuelos F. and Mujica2004). Its current distribution is from Nariño (1°39′N; Colombia) to the Salares of southern Bolivia (21°15′S), including countries such as Peru, Ecuador, northern Argentina (Jujuy and Salta provinces), northern Chile (Tarapaca and Antofagasta regions) and the southern extreme of the Andes in the Chilean lowlands (39°48′S) (Wilson Reference Wilson1990; Bhargava et al. Reference Bhargava, Shukla and Ohri2006; Fuentes et al. Reference Fuentes, Martínez, Hinrichsen, Jellen and Maughan2009c ), with an altitudinal range from sea level to 4300 m asl (Cancosa, Chile). Its diversity, at a continental scale, is associated with five main ecotypes: Highlands, Inter-Andean valley, Salares (salt flats in the Andean High plateau), Yungas (a warm, rainy and humid stretch of forest along the eastern slope of the Andes Mountains, transitional between the Andean highlands and the eastern forests; stretching from central Bolivia to south-eastern Peru) and Coastal/Lowlands; each of these are associated with sub-centres of diversity that originated around Lake Titicaca (Risi & Galwey Reference Risi and Galwey1984).
Thus, the genetic diversity of quinoa probably evolved as ancient societies along the Andean range tested new soils and climates through cycles of seed exchange and domestication processes (Wood & Lenné Reference Wood and Lenné1997; Tagle & Planella Reference Tagle and Planella2002; Badstue Reference Badstue2006; de Haan Reference de Haan2009). It has been proposed that at least three genetic events could have created ‘bottlenecks’ and affected this biodiversity (Jellen et al. Reference Jellen, Kolano, Sederberg, Bonifacio, Maughan and Kole2011). The first, and potentially most severe, event could have occurred at the very beginning when the two diploid ancestry of quinoa hybridized. The second event occurred when quinoa was domesticated from its wild tetraploid relatives. The third can be considered a political bottleneck, starting more than 400 years ago during the Spanish conquest period and has continued until the present time, reflecting the different national histories of current South American countries. During this 400-year period, quinoa was excluded from the production process due to its importance in the indigenous social and religious beliefs (Cusack Reference Cusack1984; National Research Council 1989; Fuentes et al. Reference Fuentes, Maughan and Jellen2009a ). A new, fourth bottleneck event can be added viz. the migration from rural zones of the Andes to modern urban centres, resulting in abandoned villages and farms (Bazile et al. Reference Bazile, Carrié, Vidal and Negrete2011). In Chile, the last national census (INE 2004) revealed that 13·5 million people of 15·6 million, i.e. 0·87 of the population, live in urban areas. This social change has exposed quinoa to a risk of genetic erosion, particularly considering that germplasm diversity is conserved mostly on farm or in situ within zones where there are no massive intervention programmes of agricultural modernization (Wale Reference Wale2008; Martínez et al. Reference Martínez, Bazile, Thomet, Delatorre, Salazar, León-Lobos, Von Baer, Nuñez, Coudel, Devautour, Soulard and Hubert2010). This agrobiodiversity is held by indigenous farmers – often women – at very small spatial scales, especially in Chile. This genetic diversity reflects the wide range of quinoa ecotypes adapted to different agroecological zones in the Andean area (Fuentes et al. Reference Fuentes, Espinoza, Carevic, Von Baer, Soliai, Jellen, Maughan, Seguel Benítez, León-Lobos, Piñera Vargas, Muñoz Vera and Avendaño Fuentes2009b ), where farmers have used the grain as a staple food source because of its nutritional significance (Rana et al. Reference Rana, Narzary and Ohri2010; Fuentes & Bhargava Reference Fuentes and Bhargava2011). In this context, the Andean farmers might be considered as resource persons (Bazile & Abrami Reference Bazile and Abrami2008) due to the fact that: (i) they conserve the genetic diversity of quinoa in their fields as in situ collections (Cleveland et al. Reference Cleveland, Soleri and Smith1994); (ii) they know the agronomic behaviour of each quinoa archetype; and (iii) they know the complex networks that explain the seed fluxes in their territories (Bazile & Weltzien Reference Bazile and Weltzien2008; Aleman et al. Reference Aleman, Thomet, Bazile, Pham, Coudel, Devautour, Soulard and Hubert2010). However, although farmers have a broad knowledge of quinoa archetypes, they do not appreciate the diversity at much larger spatial scales, ignoring most of these diverse varieties. Hence, it is necessary to share the traditional knowledge more widely, to recognize and thus to avoid the risks of genetic erosion (Brookfield et al. Reference Brookfield, Padoch, Parsons and Stocking2002; Bazile & Negrete Reference Bazile and Negrete2009). The lack of systematic assessment of seed systems on-farm, as a key gap in seed security, is a longstanding issue (Sperling et al. Reference Sperling, Cooper and Remington2008). Nevertheless, it has been possible in recent years to relate the major factors (e.g. natural disasters and farmer's seed exchanges/selection) shaping population dynamics at micro-regional levels with the genetic diversity pattern by using a molecular marker-based approach (Pressoir & Berthaud Reference Pressoir and Berthaud2004; Ferguson et al. Reference Ferguson, Jones, Bramel, Dominguez, Torre Do Vale and Han2011).
Polymorphic DNA markers have been used in diverse applications such as population genetics, conservation biology, studies of evolutionary history and for developing conservation strategies as well as core collections in a important number of crop plants (Matus & Hayes Reference Matus and Hayes2002; Woodhead et al. Reference Woodhead, Russell, Squirrell, Hollingsworth, Mackenzie, Gibby and Powell2005; Orabi et al. Reference Orabi, Backes, Wolday, Yahyaoui and Jahoor2007; Angioi et al. Reference Angioi, Rau, Attene, Nanni, Bellucci, Logozzo, Negri, Zeuli and Papa2010; Vargas et al. Reference Vargas, Elzinga, Rojas-Beltran, Bonifacio, Geary, Stevens, Jellen and Maughan2010; Bellucci et al. Reference Bellucci, Nanni, Bitocchi, Rossi and Papa2011). Among them, the microsatellites or simple sequence repeats (SSRs) have become a powerful tool for diversity analysis since they are highly polymorphic, multi-allelic, frequently co-dominant, highly reproducible, and randomly and widely distributed in the genome (Bhargava & Fuentes Reference Bhargava and Fuentes2010). In quinoa, the study of genetic diversity has been carried out using morphological data (Rojas Reference Rojas2003; Bhargava et al. Reference Bhargava, Shukla, Rajan and Ohri2007b ; Fuentes & Bhargava Reference Fuentes and Bhargava2011) as well as by using molecular tools (Christensen et al. Reference Christensen, Pratt, Pratt, Nelson, Stevens, Jellen, Coleman, Fairbanks, Bonifacio and Maughan2007; Fuentes et al. Reference Fuentes, Martínez, Hinrichsen, Jellen and Maughan2009c ; Rana et al. Reference Rana, Narzary and Ohri2010), resulting in important advances in describing the relationships among quinoa ecotypes at regional spatial scales. This information has also been useful in planning actions for conservation and use of germplasm applied at a wider scale (regional and national). However, these approaches to the diversity analysis of quinoa do not consider, for example, the dynamics of seed distribution through on-farm exchanges, as an anti-risk strategy for food security using the diversity of the agro-ecological sites (Subedi et al. Reference Subedi, Chaudhary, Baniya, Rana, Tiwari, Rijal, Sthapit and Jarvis2003; Chevassus-au-Louis & Bazile Reference Chevassus-Au-louis and Bazile2008).
The goals of the present study were to relate the agro-ecosystem factors, the farmers’ seed exchanges and their local production practices on-farm with the emerging genetic pattern of quinoa at national county scale in Chile using microsatellites molecular markers.
MATERIALS AND METHODS
Study area
The study was based in the three main quinoa growing areas of Chile, which exemplify a wide range of geographic areas, ecosystems and cultural groupings from the north to the south of the country between 18 and 39°S. In the northern zone (Tarapaca region, between 18 and 22°S), quinoa is cultivated as a main crop by indigenous communities (Aymara culture). The crop is grown on small farms under highland conditions, with a mean altitude of 3500 m asl, on saline soils and an annual precipitation of 100–200 mm from December to February (Fuentes Reference Fuentes2008). The characteristic dry climate leads farmers to cultivate the same field only once every 2 years, permitting water accumulation in the upland soils (endorheic geographical component) for the fallow period. The central zone (O'Higgins and Maule regions) is characterized by a temperate climate with dry summers and winters, having precipitation c. 500–800 mm per year (DMCH 2011). The agriculture is based mainly on vineyard, tree fruits and cereals: quinoa is cultivated without irrigation in marginal areas as an alternative crop at an altitude of c. 100–200 m between 34 and 36°S. The southern zone (Araucania region) has a rainy oceanic climate with annual average precipitation reaching up to 2000 mm per year (DMCH 2011); the principal agricultural activities are cultivation of forages, cereals and vegetables. Here quinoa is cultivated by indigenous communities (Mapuche culture) and other smallholders at various altitudes from 50 to 600 m between 37 and 39°S.
Fieldwork
In 2008, a field survey was carried out to characterize the diversity of agroecosystems of quinoa production in the three areas. Initially, a participatory rapid appraisal (PRA) with 10–15 people in each site was conducted, to present the objectives of the work, to understand the general agriculture dynamics in the zone and to obtain precise information about quinoa diversity production and management in this sector. Then, semi-directed interviews of farmers were conducted to establish the importance of quinoa cultivation in the farms and the management of local varieties. The sample size was 21 farmers in the north, 13 in the centre and 5 in the south, reflecting the relative importance of quinoa crop revealed in the latest available national agricultural census (INE 2007). A qualitative treatment of data allowed descriptions of production systems, management of local varieties and external variables affecting the farming system. In 2009, a second round of field surveys was carried out in the same areas. In order to identify the details of practices of local varieties management undertaken by farmers and to assess the key elements of quinoa cultivation and the roles of different actors (farmers and institutions), semi-constructed and qualitative interviews with farmers and institutions (three or four villages from each zone) were conducted. In total, 92 interviews were conducted: 31, 26 and 35 in the northern, central and southern zones, respectively. The database constructed following these interviews allowed determination of who provided seeds to whom through local networks, and analysis of the level of diversity of local varieties and the diffusion of associated knowledge regarding crop adaptation and seed management. The analysis of information was performed through multiple factor analysis using Statistica 6.0® Software (StatSoft 2001).
Genetic materials
Genetic analysis was performed on 34 quinoa accessions, which were provided by diverse institutional seed banks as seed pools collected from farmers (Table 1). Twenty-six accessions were geographically representative of the three main quinoa growing areas in Chile (northern, central and southern zones) (Fuentes et al. Reference Fuentes, Martínez, Hinrichsen, Jellen and Maughan2009c ) and the remaining eight were included as controls from the Andean zone outside of Chile (Christensen et al. Reference Christensen, Pratt, Pratt, Nelson, Stevens, Jellen, Coleman, Fairbanks, Bonifacio and Maughan2007).
Table 1. Passport data of quinoa accessions utilized in the present study
* UNAP (Arturo Prat University, Chile), Baer (Seeds Baer Company, Chile), INIA (Institute of Agricultural Research, Chile), CIP-FAO (International Potato Center and Food and Agriculture Organization).
DNA extraction and microsatellite analysis
All quinoa accessions were sown and grown in pots at 25 °C and 12 h photoperiod in a growth chamber at Arturo Prat University (Iquique, Chile) until four true leaves formed. DNA was extracted from bulked leaf material from three individual plants per accession (c. 0·1 g from each), according to protocol described by Lodhi et al. (Reference Lodhi, Yen, Weeden and Reisch1994).
Twenty di-/tri-nucleotide loci microsatellites utilized by Fuentes et al. (Reference Fuentes, Martínez, Hinrichsen, Jellen and Maughan2009c ) were reviewed and chosen in accordance with their reproducibility and clear amplification of bands (Table 2). The polymerase chain reaction (PCR) mix was performed with 60 ng of genomic quinoa DNA (3 μl), 1 mM of Cresol Red (1·5 μl), 2 mM of deoxynucleotide triphosphates (dNTPs) (1·5 μl), 25 mM of magnesium chloride (MgCl2) (1·5 μl), 10 times of PCR buffer (1·5 μl), 10 μM of each primer (1·5 μl), and 1 unit of Taq polymerase (Bioline USA Inc.) (0·5 μl), for a total volume per reaction of 15 μl.
Table 2. Name, motif, primer sequences, GenBank accession number, number of alleles, H value modified and Wright's F ST-statistics value for each locus microsatellites used in the present study
* Nei (Reference Nei1978).
† Wright's F ST statistics value with 0·95 confidence interval for groups identified (Wright Reference Wright1951).
The amplification parameters of DNA were as follows: 18 cycles at 94 °C for 60 s, 67 °C for 30 s and 72 °C for 60 s, followed by 20 cycles at 58 °C as annealing temperature with a final extension cycle of 72 °C for 10 min on a thermocycler Eppendorf Mastercycler Gradient® (Hamburg, Germany).
The amplicons were visualized and photo-documented under a UV transilluminator Digi Doc-It System (UVP, BioImaging system, Upland, CA, USA) using 0·03 (w/v) Metaphor® agarose gel (CambrexBio Science, East Rutherford, NJ, USA) stained with ethidium bromide. The size fragments were measured using the Gel-Pro Analyser 6.0 software (Media Cybernetics, Bethesda, MD, USA). A standard ladder of 100 bp (Bioline USA Inc., Boston, MA, USA) and a set of fragments of known size reported by Fuentes et al. (Reference Fuentes, Martínez, Hinrichsen, Jellen and Maughan2009c ) were utilized as reference for size fragment measuring.
Molecular data analysis
Molecular data were converted into a binary matrix, which registered the presence (1) or absence (0) of alleles at each microsatellite locus per accession. A pairwise analysis of molecular data was performed using Jaccard's similarity coefficient using the FreeTree® program (Pavlicek et al. Reference Pavlicek, Hrda and Flegr1999). An unweighted pair group method with arithmetic mean (UPGMA) cladogram was computed after 500 replicates for a bootstrap test and constructed with help of Tree View (Win32) ver. 1.6.6. software (Page Reference Page2001). Levels of heterozygosity (H) per locus and Nei's (Reference Nei1972, Reference Nei1978) identities/distances among populations, as well as Wright's F ST-statistics value (Wright Reference Wright1951) were calculated using TFPGA software, version 1.3 (Miller Reference Miller1997). Comparison of polymorphic nucleotide motifs were made using unpaired Student's t test; P ⩽ 0·05 with INFOSTAT (Infostat 2008) statistical software.
RESULTS
Microsatellite analysis
The molecular analysis yielded 118 polymorphic markers for all quinoa accessions assessed, with a mean value of alleles per locus of 5·9. The QAAT76 locus had the highest record for alleles number (n = 11) and QCA88 locus the lowest record (n = 2) (Table 2). In general, it was observed that loci with tri-nucleotide motifs were significantly more polymorphic than those loci with two-nucleotide motifs (t-test; P ⩽ 0·05). The H for all quinoa accessions ranged between 0·12 (QGA17) and 0·87 (QAAT76) and a mean value of 0·65 (Table 2).
The UPGMA analysis using the Jaccard coefficient identified two major groups, which were subdivided into five populations. Population I contained nine accessions representative of the northern zone of Chile, population II seven accessions of the central zone, population III included nine accessions from the southern zone and only one from the central zone (B042), population IV contained six accessions from the highlands of Peru, Bolivia and Argentina, and finally population V contained two accession from Ecuador and Colombia (Fig. 1).
Fig. 1. UPGMA cladogram based on Jaccard's similarity coefficient of 34 quinoa accessions performed after 500 replicates for bootstrap test (percentage number between each node). Grey rectangles represent the main growing areas of quinoa in Chile. (Population I: Salares/Northern Chile; Population II: Lowland/Central Chile; Population III: Lowland/Southern Chile; Population IV: Highland/Peru–Bolivia–Argentina; Population V: Inter-Andean Valley/Ecuador–Colombia).
The identification of populations allowed the comparison of allelic diversity patterns (Table 3); the highest values found for average H and proportion of polymorphic loci were observed in populations III (lowland) and IV (highlands outside Chile). Additionally, population III had the highest values for mean alleles per locus and total alleles, in spite of having less geographic dispersion than population IV (Peru, Bolivia and Argentina) (Fig. 2). Population I (northern zone of Chile) exhibited the maximum value for unique alleles, followed by populations III (southern zone of Chile) and IV (highlands outside Chile). Finally, populations II and V yielded lower values for all variables that described the allelic diversity pattern (Table 3).
Fig. 2. Model of biodiversity dynamics of quinoa associated with populations identified in the study (Population I: Salares/Northern Chile; Population II: Lowland/Central Chile; Population III: Lowland/Southern Chile; Population IV: Highland/Peru-Bolivia-Argentina; Population V: Inter-Andean Valley/Ecuador–Colombia). Arrows show most likely seed migration routes, as inferred from genetic similitude and from ancestral people's interactions and cultural exchanges.
Table 3. Pattern of total alleles, mean alleles per locus, unique alleles, average heterozygosity (Nei Reference Nei1978), proportion of polymorphic loci (PPL) and Nei's (Reference Nei1972, Reference Nei1978) identities/distances of populations identified in the study*
* Population I: northern zone of Chile; population II: central zone of Chile; population III: southern zone of Chile; population IV: Peru, Bolivia and Argentina highlands; population V: Inter Andean valleys of Ecuador and Colombia.
Nei's identities/distances values among populations (Nei Reference Nei1972, Reference Nei1978) were consistent with their geographical origin. Thus, quinoa populations from highlands (I and IV) and lowlands (II and III) had the lowest genetic distances. Interestingly, quinoa from the Inter-Andean Valley was shown to be genetically closer to quinoa from lowlands (Table 3).
Through the analysis of Wright's F ST values, it was possible to describe a new sub-set of informative loci for population discrimination (Table 2). Thus, within all SSR markers the loci with higher F ST values, which were the most biologically significant, stood out: KGA16, QCA24, QCA48 and QCA57. In contrast, the loci QCA88 and QAAT76 were not significant, having F ST value <0·1.
Assessment of fieldwork surveys
In the northern zone, it was possible to detect a high level of variability in phenotypically selected cultivated quinoa, which gives farmers a degree of security against climate change (drought and frost). The size of these farms ranged from 1 to 4 ha per farmer, with quinoa grown mostly for self-consumption. The terms variety or landrace are used by farmers to describe a quinoa line that is adapted to a specific locality and recognized by phenotypic characteristics, principally by panicle colour. Thus, farmers had on average 2·4 varieties each; however 0·32 of farmers have only one variety and 0·15 of them have four or more varieties (Table 4). Among the communities assessed, 0·90 of farmers use the same landrace, called ‘white’ because of its grain colour and its suitability for cooking. Those farmers who manage two or more other varieties as well as the dominant white one always use this variety in association with the others, i.e. biodiverse farmers tend to adopt this strategy of maintaining more than one variety through different cultivation periods.
Table 4 Principal agronomic and social variables of quinoa in the three ancestral production zones of Chile
i, ii, iii and iv: information from 2007 National Agricultural Census (INE 2007) completed and corrected by field works.
* Minimum estimation.
In the central zone, the diversity of soils is limited to sandy type with or without salt. Cultivation area ranges from 1 to 5 ha per farmer. Farmers throughout the area make reference to a single landrace (Table 4), called white, independently of conditions and properties affecting the farm system (soil texture, salinity, etc.). However, although they only talk about one landrace, there was evidence of synonyms for the same kind of seed (e.g. white, golden and yellow).
The interviews in the southern zone revealed a total of eight landraces, with 0·32 of the farmers having a single landrace, 0·49 two landraces and 0·18 can distinguish three landraces using various criteria such as grain colour before and after saponin extraction, or the colour of leaves and panicles. These farms were the smallest in the present study, with an area of c. 0·1 ha per farmer (Table 4).
In all of the regions, farmers classified varieties of quinoa based on colour, whether of the flower or grain, which implies also different type of uses (cultural or economic). Each farmer manages diversity via different management practices. This contributes to the diversity of the production system within each area through the growing of different crop or horticultural species and also generates more diversity within the quinoa species, through growing different ecotypes or landraces.
Clearly, even though several scientific or social projects have been carried out in Chile, the farmers themselves are the main actors in biodiversity conservation (Table 4). Also, social and professional networks are important: two types of exchange take place in the communities and represent the main access to seeds. The first is an individual exchange (between individuals or families) inside the community or between close communities. The second is a collective exchange that is organized during events such as local markets or traditional ceremonies. Each type of exchange uses the same main participants. These are largely women in the south (0·85), and have been termed seed curators (Pérez Reference Pérez2005), in contrast with 0·95 of men in the north and centre. These farmers typically maintain high levels of crop diversity and have knowledge regarding crop adaptations and seed management, and can be categorized as nodal farmers because they play an essential role in seed exchanges. These nodal farmers participate in all the formal exchanges organized in their regions, thus giving a dynamic to the exchange and confidence to the various farmers they collaborate with.
DISCUSSION
Microsatellite analysis
The significant difference (t test; P ⩽ 0·05) of polymorphism between two- and three-nucleotide motifs confirms the observation in other studies of microsatellites, also carried out in quinoa, by Mason et al. (Reference Mason, Stevens, Jellen, Bonifacio, Fairbanks, Coleman, Mccarty, Rasmussen and Maughan2005), and recently in Amaranthus by Mallory et al. (Reference Mallory, Hall, Mcnabb, Pratt, Jellen and Maughan2008), where it was concluded that the development of highly polymorphic microsatellites markers should be focused to use tri-nucleotide motifs with repeat of >20 bp.
Previous assessments of microsatellites markers in multi-origin quinoa germplasm have reported H values of 0·57 (Mason et al. Reference Mason, Stevens, Jellen, Bonifacio, Fairbanks, Coleman, Mccarty, Rasmussen and Maughan2005), 0·75 (Christensen et al. Reference Christensen, Pratt, Pratt, Nelson, Stevens, Jellen, Coleman, Fairbanks, Bonifacio and Maughan2007), 0·57 (Jarvis et al. Reference Jarvis, Kopp, Jellen, Mallory, Pattee, Bonifacio, Coleman, Stevens, Fairbanks and Maughan2008) and 0·65 (Fuentes et al. Reference Fuentes, Martínez, Hinrichsen, Jellen and Maughan2009c ). Comparison with the H-value of 0·65 (ranging from 0·12 to 0·87) found in the present study suggests that there was a wide genetic diversity in the quinoa samples of the present study, and also confirmed the highly informative quality of the markers used. 0·60 (n = 12) of them were classified as highly polymorphic (H ⩾ 0·7) according to the classification established by Ott (Reference Ott1992).
The geographic origin of accessions and the populations identified by UPGMA analysis were consistent with the ecotype classifications described by Risi & Galwey (Reference Risi and Galwey1984) and the production zones focused in the present study. In this sense, population I was representative of Salares (North of Chile), population II and III of Coastal/Lowlands quinoa (South of Chile), population IV of Highlands (Peru, Bolivia and Argentina) and population V of the northernmost Inter-Andean Valleys (Ecuador and Colombia) (Fig. 2).
The pattern of high genetic diversity within Coastal/Lowlands ecotypes has been described in previous studies of southern Chile ecotypes using dominant AFLP markers (Anabalón Rodriguez & Thomet Isla Reference Anabalón Rodriguiez and Thomet Isla2009) and co-dominant SSR markers (Fuentes et al. Reference Fuentes, Martínez, Hinrichsen, Jellen and Maughan2009c ). It has been suggested that Coastal/Lowlands ecotypes are representative of active crop/weed complexes having a monophyletic coevolving behaviour (Rana et al. Reference Rana, Narzary and Ohri2010). This might explain why lowland breeders find it difficult to obtain pure quinoa cultivars, because of its close affinity to weed populations of C. album and/or C. hircinum (Fuentes et al. Reference Fuentes, Martínez, Hinrichsen, Jellen and Maughan2009c ). This is also suggested by the large number of alleles recorded within population III, in spite of its smaller geographic dispersion than population IV (representative of a broad geographic area surrounding the centre of origin of quinoa). Finally, the close genetic relationship among lowland and Inter-Andean Valley representations found in the present study supports the hypothesis proposed by Fuentes et al. (Reference Fuentes, Martínez, Hinrichsen, Jellen and Maughan2009c ), which is based on the artificial introduction of alien germplasm to lowland areas. An example of this is the case of the Regalona cultivar from the Baer Seed Company, obtained by artificial hybridization between parents from Araucania region (Baer II) and Ecuador (I. von Baer, personal communication).
Values for average H and proportion of polymorphic loci in population II (central zone) were lower than those recorded for population III (southern zone). Thus, the diversity for the central zone revealed a sub-pattern of a geographic bottleneck within lowland/coastal quinoa. This pattern matches well with the isolation among farmers (Table 4), resulting in a fragmented pattern of diversity among these two geographic areas (Fig. 2). Meanwhile, the lowest H value, observed for population V, was limited by the small size of the sample considered. This situation has also been reported in previous quinoa diversity analyses, because of its poor representation in seed banks (Christensen et al. Reference Christensen, Pratt, Pratt, Nelson, Stevens, Jellen, Coleman, Fairbanks, Bonifacio and Maughan2007).
The Wright's F ST values of each microsatellite locus reported interesting features related to the detected polymorphism at population level (Table 2). This information may make the partial use of some microsatellite loci in faster diversity assessing or marker-assisted selection purposes possible in the future (Forapani et al. Reference Forapani, Carboni, Paoletti, Moliterni, Ranalli and Mandolino2001; Skøt et al. Reference Skøt, Humphreys, Armstead, Heywood, Skøt, Sanderson, Thomas, Chorlton and Sackville Hamilton2005).
Dynamics of seed exchanges
Even the most cautious interpretation of the observed molecular diversity with respect to geographic, edaphic, climatic, social and cultural groupings, correlates well with the reported information. Thus, Salares and Highlands ecotypes, even if they are geographically separated, do remain close in the dendogram because they are in similar edaphic and climatic contexts (populations I and IV, respectively) (Fig. 2). They share the high Andes climatic conditions and also similar photoperiods and soils (Rojas Reference Rojas2003; Martínez et al. Reference Martínez, Veas, Jorquera, San Martín and Jara2009a ). The Aymara–Quechua people currently occupy areas in Peru, Bolivia and Chile and lowland accessions form a clear distinct grouping (Fig. 1), which correlates well with differences in the geographic/edaphic/climatic and cultural contexts.
In Chile, quinoa is mainly used for the farmers’ own consumption and it has remained a marginal crop, different from others such as maize, potatoes or wheat, which are sold in the domestic markets. Quinoa is recognized for its culinary diversity and its properties as a high-quality food. It is also grown as an old habit acquired through the knowledge of their parents, who cultivated it since their childhood, associating it with family memories in different ancient cultures. The generation changes are already affecting rural areas; indeed, one consequence of the mass migration from rural areas into cities in Chile is that agro-industries hire immigrants from the neighbouring countries of Peru and Bolivia for harvesting and other rural activities, due to the lack of local people. This is a global phenomenon (Taran & Geronimi Reference Taran and Geronimi2004) from which Latin America, and particularly women, is not exempt (Staab Reference Staab2003). However, quinoa is still maintained in the three major areas of its cultivation in Chile as well as in the rest of the Andes (Bhargava et al. Reference Bhargava, Shukla and Ohri2006; Fuentes et al. Reference Fuentes, Martínez, Hinrichsen, Jellen and Maughan2009c ), being an essential part of the intangible cultural rural heritage and identity (Nuñez et al. Reference Núñez, Bazile, Chia, Hocdé, Negrete, Martinez and Ceorgudis Maya2010).
In the southern Mapuche communities, it is still possible to identify the key elements of the quinoa history associated with the role of the different modern actors (farmers and institutions) involved in the biodiversity dynamics of this crop. The crop almost disappeared following the Spanish conquest, replaced by rice and wheat in the Mapuche diet first and then throughout the Chilean people. Nevertheless, some families never stopped growing quinoa and conserved their family population varieties as local collections of landraces. According to Anabalón Rodriguez & Thomet Isla (Reference Anabalón Rodriguiez and Thomet Isla2009), they even maintained low levels of seed exchange between Coastal/Lowlands and Andean piedmont regions, separated by not more than 150 km (this is also supported by genetic analysis). However, the important role played by curators at intra- and extra-regional scales (Pérez 2005; Bazile et al. Reference Bazile, Thomet, Aleman, Pham and Copa2010a ) could change the future dynamic of seed exchanges, and it will be necessary to determine the impact of these public exchanges on biodiversity dynamics.
This particular situation at the Coastal/Lowlands also exists in other regions of quinoa production. In the highlands, the altitude and climatic constraints determine a natural limit to wider seed exchanges. However, a better understanding of the Aymara crop system and the extent of trans-border trading could reveal the real diversity between Chile, Bolivia and Peru zones in the highlands (Fuentes et al. Reference Fuentes, Martínez, Hinrichsen, Jellen and Maughan2009c ; Bazile et al. Reference Bazile, Sánchez, Espinoza, Delatorre and Copa2010c ). The diversity of local varieties of quinoa has the same ranking among the people of the northern highlands of Chile (Villages of Colchane and Cancosa, near Bolivia) and south (Village of Socaire, near Argentina). The present results show that farmers define different varieties by agro-morphological characteristics such as the colour of the panicle of quinoa. There is also a sub-classification of plants by size and panicles for each colour. In the present surveys, farmers identified more than 18 landraces, conserved at the village scale, including four to five colours, as was also observed in Bolivian diversity of the Quinoa Real type. Thus, for example in the village of Colchane (northernmost region of quinoa production for Chile), 0·90 of the farmers sow the white quinoa, then comes a second landrace in the crop system: the yellow, red and pink, used by 0·30–0·40 of farmers. Besides these four landraces there is another one which is used by less than 0·10 of the farmers. A future study on the dynamics of what farmers sow every year (number of farmers sowing a particular landrace and sown areas), and under what criteria they have saved certain seed types (and within what conditions of conservation) would be interesting.
The information provided by farmers in the central zone revealed the existence of a single landrace. However, the analysis of practices showed that sowing and harvest periods can vary within this landrace, with harvest being February to April. However, the existence of many differences in sowing to flowering and harvesting dates means that it is not possible to validate that all the farmers have the same landrace. The practices give evidence of two groups of landraces, one harvested early and one later (4 and 7 months growth duration, respectively), and each group has different coefficients of photoperiodism to justify the heterogeneity of practices, as shown in Table 4. Thus, the diverse handling of the production system by farmers in the central zone has generated a high biodiversity of quinoa types as revealed by the relationships within population II (Fig. 1). It is also true that some farmers do select for higher tolerance to salt stress by sowing on lands naturally invaded by brackish coastal waters close to river outlets (Ruiz-Carrasco et al. Reference Ruiz-Carrasco, Antognoni, Coulibaly, Lizardi, Covarrubias, Martínez, Molina-Montenegro, Biondi and Zurita-Silva2011).
However, the strengthening of international markets and export incentives for other products in the area could cause a loss of seed diversity in the future (Marti & Pimbert Reference Marti and Pimbert2007). This could be caused by changes in land use or by homogenizing the seed throughout the zone, to respond to increasing market demands for quinoa (Martínez et al. Reference Martínez, Bazile, Thomet, Delatorre, Salazar, León-Lobos, Von Baer, Nuñez, Coudel, Devautour, Soulard and Hubert2010). This highlights the importance of linking sociological and agronomic studies to further a deeper analysis with molecular markers in order to validate different hypotheses of diversity levels such as, for instance, whether there is a hidden higher diversity within a single landrace, white quinoa, in the area of central Chile. In that case, even if a great diversity of practices is observed along with high genetic diversity, the genetic erosion risk farms is very high, particularly due to small scale (Martínez et al. Reference Martínez, Jorquera, Veas and Chia2009b ; Bazile et al. Reference Bazile, Olguín, Núñez, Negrete and Copa2010b ). Thus, this agricultural diversity needs to be measured and protected before the risk of it disappearing becomes too great.
Even today, farmers who inherited the traditions of quinoa consumption and associated culture, but belong to different highland and lowland regions, do not share quinoa germplasm. This is also due to different sowing and harvesting seasons, derived from very different photoperiods (Bertero et al. Reference Bertero, King and Hall1999; Bertero Reference Bertero2001), a limit that could be established around 25°S (Fig. 2). In previous experiments where quinoa ecotypes from highland and lowland seed sources were sown at 30°S (Coquimbo region), greater production was achieved by southern ecotypes than by high Andes ones (Martínez et al. Reference Martínez, Delatorre and Von Bae2007, Reference Martínez, Veas, Jorquera, San Martín and Jara2009a , Reference Martínez, Jorquera, Veas and Chia b ). It is considered that 3–5 years are needed to adapt a landrace transferred between zones from a particular local region and 10 years for landraces from other regions further away (Aleman Reference Aleman2009).
Farming practices have a significant role in the distribution and conservation of genetic variation, thus increasing the resilience of the systems to production constraints and environmental changes (Sperling et al. Reference Sperling, Cooper and Remington2008; Ferguson et al. Reference Ferguson, Jones, Bramel, Dominguez, Torre Do Vale and Han2011). Many plant genetic resources such as quinoa are increasingly being lost from traditional agricultural landscapes through external factors (e.g. population migration and commercial pressures). The utility of ex situ gene banks to conserve germplasm is not enough for two reasons: firstly, it is very difficult to capture or collect all the cultivated diversity with so many different landraces and variants, and secondly even if that were possible, the ex situ conservation cannot integrate other key elements associated with the genetic resources when conserved in situ, e.g. the agricultural knowledge and practices, and the associated culture, which are specific to the diverse array of agrobiodiversity conservation systems. To maintain this co-evolutionary adaptation and selection between the genetic resources and their in situ users, the central point is to conserve the characteristics of the agrobiodiversity dynamics, considering the whole associated social context.
The main conclusion from the present work is that the highly polymorphic microsatellite alleles present in the three main growing areas of quinoa in Chile (as well as of the rest of the Andean region) suggests the existence of a large amount of genetic diversity and confirms the status of previously recognized ecotypes. The genetic information allows the detection of variation among and within the populations identified, which matches well with natural geographical–edaphic–climatic constraints to the expansion of quinoa biodiversity. This grouping also correlates well with the social–linguistic context of ancient people inhabiting the Andes region, where agronomic and cultural traditions that have survived until the present time are very different. Risks to biodiversity in this species are postulated due to export incentives for agricultural products that might favour none or fewer quinoa genotypes. Better incomes from export systems are normally obtained in a short time scale, although the highest profit is generally achieved by bigger land owners, whose greater numbers of hectares are cultivated with much higher investment. In such circumstances, small-scale farmers migrate or become employees of the agro-industry, and risk losing the culture and the agro-biodiversity of their landscapes.
