Agricultural Systems 211 (2023) 103750 Contents lists available at ScienceDirect Agricultural Systems journal homepage: www.elsevier.com/locate/agsy Crop biocultural traits shape seed networks: Implications for social-ecological resilience in south eastern Senegal Anna Porcuna-Ferrer a, b, c, *, Vanesse Labeyrie b, c, Santiago Alvarez-Fernandez a, Laura Calvet-Mir a, d, e, Ndèye Fatou Faye f, Sarah Ouadah g, Victoria Reyes-García a, h, i a Institut de Ciència i Tecnologia Ambientals (ICTA), Universitat Autònoma de Barcelona, Cerdanyola del Vallès, Barcelona, Spain b CIRAD, UMR SENS, F-34398 Montpellier, France c SENS, Univ Montpellier, CIRAD, IRD, UPVM, Montpellier, France d Institut Metròpoli, Universitat Autònoma de Barcelona, Cerdanyola del Vallès, Barcelona, Spain e TURBA Lab, Internet Interdisciplinary Institute (IN3), Universitat Oberta de Catalunya, Av. Carl Friedrich Gauss, 5, Castelldefels, 08860 Barcelona, Spain f Institut Sénégalais de Recherches Agricoles, Bureau d’Analyses Macroéconomiques (ISRA-BAME), Route des Hydrocarbures, BP 3120 Dakar, Senegal g Université Paris-Saclay, AgroParisTech, INRAE, UMR MIA Paris-Saclay, France h ICREA, Institució Catalana de Recerca i Estudis Avançats, Barcelona, Spain i Departament d’Antropologia Social i Cultural, Universitat Autònoma de Barcelona, Cerdanyola del Vallès, Barcelona, Spain H I G H L I G H T S G R A P H I C A L A B S T R A C T • Seed circulation networks mediate farmers’ access to crop diversity and agroecosystems’ social-ecological resilience. • We compared the factors mediating seed circulation for six staple crops in a Bas­ sari community, south-eastern Senegal. • Crop biocultural traits and network ac­ tors shape seed circulation, causing dif­ ferential access to crops and variety types. • Multi-centric, participatory in­ terventions might strengthen locally adapted crop diversity and seed sys­ tems’ resilience. A R T I C L E I N F O A B S T R A C T Editor: Laurens Klerkx CONTEXT: Agroecosystems’ social-ecological resilience largely depends on the crop diversity generated and maintained by farmers, which provides insurance against changing environmental and socio-economic condi­ Keywords: tions. In turn, crop diversity generation, maintenance, and distribution is influenced by seed circulation net­ Crop diversity works. Thus, patterns of seed circulation can support or constrain households’ access to crop diversity, affecting Local seed systems on-farm crop diversity. Seed access OBJECTIVE: We aimed at understanding the mechanisms shaping seed circulation and farmers’ access to crop Seed exchange networks Social network analysis diversity by: 1) assessing how crop biocultural traits influence patterns of seed circulation; 2) exploring the West Africa connections between household position in the seed circulation network and on-farm crop diversity for different crops. * Corresponding author at: Institut de Ciència i Tecnologia Ambientals (ICTA), Universitat Autònoma de Barcelona, Cerdanyola del Vallès, Barcelona, Spain. E-mail address: anna.porcuna@uab.cat (A. Porcuna-Ferrer). https://doi.org/10.1016/j.agsy.2023.103750 Received 29 April 2022; Received in revised form 3 June 2023; Accepted 21 August 2023 Available online 31 August 2023 0308-521X/© 2023 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by- nc-nd/4.0/). A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 METHODS: We conducted research in south-eastern Senegal applying crop diversity inventories and a survey to document seed acquisitions for the six local staple crops, which differ in biocultural traits. Household’s varietal diversity and household- and community-level network measures calculated for each crop were used to compare seed circulation patterns among crops. Then, we analyzed the association between households’ position in the seed circulation networks and households’ on-farm crop diversity using generalized linear models. RESULTS AND CONCLUSIONS: Our research advances two main findings about the importance of seed circu­ lation networks for farmers’ access to crop diversity. First, several seed circulation networks operate in the same community and at the same time. Each species circulated differently, which can be explained by crop’s biocultural traits. Socio-cultural traits, like the cultural relevance of a crop, and biological traits, like crop’s functional group (e.g., legumes, cereals), affect the patterns of seed circulation. Seed circulation networks that involved external actors, like agricultural extension projects or NGOs, were more centralized than seed circulation networks in which these actors were absent. Second, household’s centrality in the network of seed circulation (indegree and betweenness) was generally associated with higher on-farm varietal diversity. However, the factors that determined household’s access to seeds differed among crops and variety types. SIGNIFICANCE: Farmer-to-farmer seed circulation networks are instrumental for the maintenance and distri­ bution of agrobiodiversity and catalyze the introduction of new diversity in the agricultural system. However, tensions exist between traditional and new (e.g., interventions) mechanisms of seed sharing, resulting in centralized and unidirectional seed distribution, which might affect the social-ecological resilience of the system. 1. Introduction affects seed production and viability and might thus affect seed circu­ lation (Leclerc and Coppens d’Eeckenbrugge, 2012; McGuire and Crop diversity contributes to farming systems’ social-ecological Sperling, 2016). Reproduction type might affect the amount of propa­ resilience1 (Cabell and Oelofse, 2012) by providing response diversity gative material that farmers’ need or can share – e.g., crops that to disturbance through a pool of possible adaptations (Altieri and reproduce vegetatively present a lower multiplication rate and their Nicholls, 2017; Labeyrie et al., 2021b; Renard and Tilman, 2019). At the planting material is less easy to store and transport (McKey et al., 2010); plot and landscape levels, cultivating different species and varieties pollination-type might affect out-farm seed acquisition, as gene flows simultaneously and over time fosters ecological heterogeneity, which may challenge maintaining the identity of out-crossing varieties over helps responding to ecological, social, and economic shocks (Cabell and time – e.g., (Allinne et al., 2008) for millet - compared with self- Oelofse, 2012). Importantly, crop diversity is a direct outcome of bio- pollinating crops– e.g., rice (Nuijten and Almekinders, 2008). Seed cultural interactions,2 including farmers’ knowledge and practices that storability can also affect on-farm seed availability (McGuire and allow the generation and maintenance of a constantly evolving biodi­ Sperling, 2011; Meikle et al., 2002), as seeds sensitive to weather, pests, versity contributing to farms’ long-term social-ecological resilience or other hazards might need faster renewal, which might boost seed (Berkes et al., 2000; Folke, 2004; Reyes-García et al., 2014, 2013). circulation. For example, to minimize seed losses during storage, Researchers have investigated the mechanisms affecting the richness farmers in Haiti sell legume seeds after harvest and buy new ones at and distribution of crop diversity and farmers’ access to this diversity sowing time (McGuire and Sperling, 2016). In the same line, the seed (Jarvis et al., 2008; Leclerc and Coppens d’Eeckenbrugge, 2012; Zim­ stocks of some crops are more susceptible to be used as food in case of merer et al., 2019). Among the lines explored, scholars have examined shortage than the seeds of other crops. Other factors, such as commercial the role of social networks in the generation, maintenance, and diffusion value, dietary relevance, and the customary exchange value might also of crop diversity and associated knowledge (Calvet-Mir and Salpeteur, be relevant to explain the way different crops circulate (Delêtre et al., 2016; Labeyrie et al., 2021a; Pautasso et al., 2013). Seed circulation 2011; McGuire and Sperling, 2016; Thomas and Caillon, 2016). How­ networks are shaped by social forms of organization (Labeyrie et al., ever, with the exception of a study carried out in the Vanuatu archi­ 2014b; Leclerc and Coppens d’Eeckenbrugge, 2012) and network pelago (South Pacific) (Thomas and Caillon, 2016), we know of no structure can support or constrain households’ access to crop diversity, analysis considering the potential simultaneous influence of crops bio­ increasing or decreasing on-farm agrobiodiversity. logical and socio-cultural traits in explaining seed circulation patterns. Previous research has tested the existence of a relationship between A different research line has focused on customary rules guiding seed farmers’ position in the seed circulation network and on-farm crop di­ flows. This body of research shows that seed circulation is limited by versity levels following two main research lines- e.g., (Abizaid et al., social factors such as ethnolinguistic boundaries in Kenya (Labeyrie 2016; Calvet-Mir et al., 2012; Díaz-Reviriego et al., 2016; Kawa et al., et al., 2016), marriage prohibitions in Gabon (Delêtre et al., 2011), or 2013). Most studies have documented general flows of seeds and in­ kinship in the Amazonia (Abizaid et al., 2016). While informative, this formation, not allowing to discern differences between crop species and literature neglects the role of non-community actors. Seeds and related variety types e.g., Calvet-Mir et al. (2012); Díaz-Reviriego et al. (2016); information circulate through networks involving farmers, but also Abizaid et al. (2016), and Kawa et al. (2013). Few studies have other actors such as governmental and non-governmental organizations, considered whether crops and/or varieties might circulate differently local markets, the private sector, or local, national, or international in­ depending on their biocultural properties. For example, crop ecology stitutions (Coomes et al., 2015; McGuire and Sperling, 2016; Pautasso et al., 2013). Research in other fields has shown that the type of actors and the relations between actors impact the resilience and sustainability 1 We adapt the definition of social-ecological resilience - “the capacity of a of social-ecological systems (Bodin, 2017; Bodin et al., 2016), a relation system to absorb disturbance and reorganize while undergoing change so as to still that has yet to be tested in relation to the impact of actor composition on retain essentially the same function, structure, and identity” (Walker et al., 2004, p. seed exchange networks. 3) to agroecosystems by defining agroecosystem resilience as the capacity to Our work combines research insights from these two research lines to produce food despite shocks (Cabell and Oelofse, 2012). explore how crops biological and socio-cultural traits relate to farmers’ 2 Biocultural interactions refer to the dynamic relationship between biolog­ access to seeds. Building on the work of Thomas and Caillon (2016), we ical processes and cultural practices, which are deeply intertwined and mutu­ conceptualize crops as biocultural objects because they embody the ally influential. For crops, these interactions encompass selection, cultivation, interconnectedness of biological and cultural systems. We hypothesize consumption and valuation within specific cultural contexts (Maffi, 2012). 2 A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 Fig. 1. Map of the study area. that different crops might circulate differently, with different seed cir­ Cotton is the main cash-crop, and its cultivation is mostly undertaken culation networks operating within the same community. We further through contract-farming arrangements with Sodefitex, a former hypothesize that the different seed circulation networks can be (at least state-owned company that has been increasingly privatized. Most partly) explained by crops’ biocultural traits, including the constellation households are largely self-sufficient foodwise, although there is an of actors that intervene in their seed circulation. increasing dependence on commercial rice. Beyond farming, most We conducted research in south-eastern Senegal, a region where the households also take part in other income generating and subsistence agricultural system is mostly rain-fed and biodiversity-based. We start activities such as wage labour, artisanal work, hunting and gathering, by comparing the types of actors involved in seed exchanges and the palm wine processing, and harvesting of wild edible plants. Livestock seed circulation patterns of the six staple crops grown in the area. Then, keeping is an important way of capitalization. we analyze the association between households’ position in the seed Bassari farmers adjust to variable soil conditions by diversifying circulation networks and households’ crop diversity for each staple crop crops and using different management strategies. Oxenga are fields in and their varieties. Finally, we discuss the implications of our results for the hills where farmers mostly rotate peanut, Bambara groundnut, sor­ the social-ecological resilience of smallholder farms. ghum, cotton, and fonio, using few or no chemical inputs (except for cotton). A typical crop rotation4 lasts 5–6 years, which requires access to 2. Case-study different plots. Eden are fields in the floodplains, more fertile and intensively cultivated than in oxenga. In eden there are no crop rotations, 2.1. Study site fallow periods are short, and farmers use chemical inputs to sustain production. Farmers’ cultivate rice in eden fields that get inundated Data were collected in a Bassari village3 in south-eastern Senegal during the rainy season, and maize in the rest. In the past, cultural ta­ (Fig. 1). The altitude in the region varies from 115 to 380 m.a.s.l. and the boos prohibited agriculture next to the river basins, however cultivation climate is characterized by a rainy season (approx. May to September) in eden is now growing. Contrasting with up-hill lands, land in the and a longer dry season (ANACIM, 2020). The main administrative lowlands starts to be considered scarce by the Bassari. Enam are fields centre – Kédougou – and the closest market town – Salemata – are around the houses (generally in the hills), managed quite intensively, located respectively 90 and 11 km away from the study site. A paved without rotations, and maintained largely by using organic fertilizers. road connects Salemata with Kédougou, but until 2020 access to the They are mostly cultivated with maize and minor crops. administrative city was limited or completely closed during the rainy As in other locations worldwide (Garine et al., 2018), seed circula­ season. Bassari are the largest ethnic group in the area, where there are tion among the Bassari follows traditional social norms, cultural values, also other ethnic groups (e.g., Fula, Coniagui, Dialanké, Malenke). In and economic rationales expressed through gender, residence, kinship, 2012, UNESCO declared part of the area as a World Heritage site (UNESCO, 2012), called “Bassari territory” in Fig. 1. The main livelihood activity in the area is rain-fed farming, and the 4 A traditional crop rotation would be: a newly-cut field—Field A—is planted bulk of agricultural activities are concentrated during the rainy season with Bambara groundnut and/or peanut. The following year, sorghum is and the beginning of the dry season. In the light of climate change, the planted in Field A, and a second field— Field B—is planted with Bambara rainy season is shortening and the frequency and intensity of dry spells groundnut and/or peanut. In the third year, Field A reverts to Bambara are increasing, challenging local farming activities (Porcuna-Ferrer groundnut and/or peanut and Field B to sorghum. This alternation continues et al., 2023). Agriculture mostly consists of subsistence-oriented culti­ until the fields are exhausted, then, new fields are opened and the old fields are vation of cereals (sorghum, rice, maize, fonio) and legumes (peanut, cultivated with fonio. Nowadays, in rotations, sorghum is being replaced by Bambara groundnut). Households also sell excess crops in local markets. cotton and maize, and Bambara groundnut by peanut. 3 The name of the case-study village is not mentioned to respect anonymity. 3 A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 and age-class5 relations. The Bassari society is generally characterized life (Gessain, 1996). by sex-opposed spheres of ritual and economic activity (Nolan, 1986). Bambara groundnut (Vigna subterranea [L.] Verdc.), an annual Men and women independently carry out agricultural activities (e.g., legume with a high nutritional value and tolerance to adverse envi­ seed selection and storage), although they might help each other during ronmental conditions (Mayes et al., 2019), has also been cultivated by specific moments of the crop cycle. Most crops are gender-specific: the Bassari for a long time and has a strong symbolic and ceremonial cotton and sorghum are mostly cultivated by men, and peanut, importance. Indeed, sorghum beer and Bambara groundnut seeds are the Bambara groundnut, fonio, and rice are mostly cultivated by women. two only products explicitly forbidden to sell by Bassari tradition. Maize is cultivated by men and women. Bambara groundnut and sor­ Bambara groundnut is mostly cultivated by women, who renew ghum are ‘the’ gendered crops par excellence and the basis of ‘enap’, frequently the seed lots due to the difficult storage and high frequency of Bassari staple porridge, which combines women’s and men’s harvest. pest attacks. Like for sorghum, households maintain the same Bambara Residence is patrilocal (women move to their husband’s village), groundnut varieties over decades. resulting in geographically extensive matrimonial networks. Village Fonio (Digitaria exilis Stapf) is a fast-growing cereal well adapted to residency is a socially structuring factor: age-class system ritual activ­ poor soils and semi-arid conditions (Abrouk et al., 2020). For the Bas­ ities (e.g., initiation ceremonies) are performed by groups of neigh­ sari, fonio is a women’s crop, introduced at the end of the crop rotation bouring villages and the organization of communal labour exchange is cycle, when the land is exhausted. Fonio is mostly consumed during structured along neighbourhood residency.6 Indeed, in Bassari lan­ festivities and, although its sale is not culturally prohibited, economic guage, neighbourhoods are called andyana, which translates as “those transactions are rare. Many households only cultivate fonio during the with whom I work” (Nolan, 1986). Kinship and matrimonial networks cropping season preceding an important festivity. As a result, seed lots are also important in explaining seed circulation: upon a person’s death, are not stored for a long time within the households. Women typically the Bassari would traditionally transmit seeds to the niece/nephew from preserve their seeds by sharing them with other farmers at sowing time sister-side (Gessain, 1975; Nolan, 1986), although this is now changing and receiving a portion of their harvest afterwards. towards a system in which men inherit their father/grandfather’ seeds Maize (Zea mays L.) was introduced in the Bassari territory in the and women inherit their mother/ grandmother’s seeds. Finally, until early 1900s by a neighbouring ethnic group (Gessain, 1975). Nowadays, recently, the age-class was important in structuring Bassari social re­ it is strongly promoted by agricultural development projects that supply lationships, tasks, rituals, and communal agricultural work (Nolan, high yield short-cycle maize varieties and chemical fertilizers. Maize can 1986), thus potentially influencing seed circulation. be cultivated by both men and women, who sow it in fertile and deep Besides gender, residence, kinship, and age-class relations, local soils. Maize is mostly used for household consumption and seems to be markets and agricultural extension projects might also shape Bassari replacing sorghum in the Bassari food system, e.g., it has become the seed access. Local markets are mostly attended by women to sell grain or main staple in many meals, and its beer is increasingly substituting other agricultural or wild-plant products and to buy supplies. Bassari sorghum beer. Maize seeds are typically stored at home, with granaries farmers frequent the weekly market in Salemata, and rarely the bigger predominantly managed by men. and permanent market in Kedougou. None of these markets have Asian rice (Oryza sativa L.) has higher water and nutrient re­ specialized seed vendors, but seeds are sold by other farmers and re- quirements than other local staple crops. Asian rice most likely arrived sellers who bring seeds from bigger markets. Within the study village, to the Bassari territory in the 1900s and was initially cultivated in small three small shops also sell local peanut seeds during the sowing season. areas, but its cultivation has increased in the last 20 years, with the In the last decades, there have also been several NGO- or government- arrival of new high yielding varieties initially distributed by the state based development projects promoting maize, rice, and peanut culti­ and later by development projects, which also distribute herbicides and vation through the provision of seeds, chemical fertilizers, and pesti­ fertilizers. Rice is mostly cultivated by women and nowadays occupies cides (Porcuna-Ferrer et al. in press). most floodplains in the case-study village. Although women tend to store their own rice seeds, agricultural extension projects are an increasingly 2.2. Biocultural traits and seed management of Bassari staple crops important rice seed source. Peanut (Arachis hypogaea L.) is a legume domesticated in South Bassari current staple crops have different agronomic characteristics, America, initially introduced in the Gambia river basin by Portuguese cultivation histories and cultural and symbolic functions in Bassari traders (16th century) and cultivated in small quantities until its large- farming system (Porcuna-Ferrer et al., under review) (Table 1). scale commercial production was promoted by the French colonial Sorghum (Sorghum bicolor [L.] Moench) is well adapted to semi-arid government (mid-19th century) and later by post-colonial governments. agronomic conditions, being resistant to drought and heat (Smith and This promotion entailed the creation of state cooperatives and the dis­ Frederiksen, 2000). Sorghum is one of the oldest crops cultivated by the tribution of subsidized seeds (Bernards, 2019). Nowadays, although the Bassari, who rarely renew their seed lot, with most households main­ Senegalese government subsidizes peanut seeds, the Bassari prefer to taining the same varieties over decades. Exclusively cultivated by men, cultivate their own. Peanuts are mostly cultivated by women for sorghum is mostly used for household consumption and particularly for household consumption. Seed lot renewal takes place often, as seeds are the preparation of sorghum beer, a central product in Bassari ceremonial very vulnerable to pest attacks. Peanut seeds have important market value and can even be used as an alternative currency. 5 Age-class (also referred to in literature as “age-grade”) is a highly-structured 3. Methods hierarchic system in which power is held by the elders. Progression in the age- class system entails conducting a series of tasks and rituals, including agricul­ 3.1. Sampling tural common working days (Nolan, 1986; Gessain, 1975). Traditionally, age- class rituals marked the main agricultural seasons and age-class labour pool­ Data were collected during November 2019 – April 2020 and ing was important to sustain the work-intensive Bassari agricultural system September 2020–June 2021. Fieldwork was interrupted by the Covid-19 (Nolan, 1986). Age-class progression is gendered and there are rigid rules about pandemic. As Bassari agricultural activities were not strongly impacted men/women’s roles in ritual and economic activities. 6 by Covid-19, our data can be considered representative of a ‘typical Bassari neighbourhoods are semi-independent units geographically sepa­ rated from each other and generally organized according to patrilineal clans. year’. The first author lived in the case-study village during the two Neighbourhoods have a certain level of independence; each neighbourhood has a chief and neighbourhood boundaries are usually more significant and precise than village boundaries (Nolan, 1986). 4 A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 Table 1 Crops’ characteristics. Sorghum Bambara groundnut Fonio Maize Rice Peanut Scientific name Sorghum bicolor Vigna subterranea Digitaria exilis Zea mays Oryza sativa Arachis hypogaea Biological characteristics Functional group Cereal Legume Cereal Cereal Cereal Legume Type of reproduction Partially outcrossing Predominantly self- Predominantly self- Outcrossing Predominantly self- Predominantly self- pollinating pollinating pollinating pollinating Storage capacity High Low High High High Low Socio-cultural characteristics Time in the local Traditional Traditional Traditional Introduced Introduced (recently) Introduced (formerly) farming system (recently) Food use Staple, traditional dish Staple, traditional dish Staple, mostly for Staple, modern Staple, modern dish Sauce, snack (“Enap”) and sorghum beer (“Enap”) festivities dish Gendered crop Only men Only women Mostly women Women and Mostly women Mostly women management men Marketability* Medium Low High Medium Medium High External support Low Low Low High High Medium * All crops are mostly cultivated for self-consumption. periods of data collection, which allowed combining qualitative ethno­ between variety types (Calvet-Mir et al., 2011; Tardío et al., 2018). We graphic research with systematic data collection. classified varieties in three groups based on farmers’ reports about the Our sampling unit was the household,7 as some agricultural activities period and actors introducing them: “farmer varieties”, “introduced are conducted at that level. We defined the limits of the network as that farmer varieties”, and “non-farmer varieties”. Farmer varieties are vari­ of the studied community and initially included in our sample all Bassari eties selected, reproduced, and kept by farmers and which have been in households in our case-study village. We also included households’ in­ the local farming system for >30 years (also referred to as “landrace” or teractions with actors outside the village, including other households, “heirloom varieties”8). Introduced farmer varieties are varieties intro­ agricultural extension projects and NGOs, and market vendors. duced through farmer-to-farmer seed exchange within the last 30 years. We visited each household twice, first to conduct a crop diversity Non-farmer varieties refer to varieties developed by professional plant inventory and later to conduct a seed exchange network survey. Both breeders, which have recently arrived to the community through agri­ interviews were conducted with all available household members who cultural extension projects. For each variety, during the focus group had cultivated an independent plot during the previous cropping season. discussions, we documented the names most frequently used by farmers, Some households were not available for interviews, so our final sample the existence of any synonyms, and varieties’ characteristics (e.g., includes 117 households (or 95% of a total of 123 households in the maturity cycle, colour, size, uses). village). To obtain household’s crop diversity inventories, we asked farmers to list all the staple crops they had cultivated during the previous growing season. For each staple crop, we asked them to estimate the 3.2. Data collection surface cultivated in ‘cordes’ (1 corde = 0.25 ha) and to provide the local names of all the cultivated varieties of each crop. We insisted that re­ 3.2.1. Crop diversity inventories of local staple crops spondents listed all varieties grown, including those represented by a To characterize crop diversity at the village- and household-levels for very limited number of individual plants. To ensure consistency in the six staple crops, we conducted two village focus group discussions, naming, we asked follow-up questions about the variety characteristics, one with men and one with women, and household crop diversity which we then compared to our reference lists elaborated during focus inventories. groups. When farmers finished listing, we used the reference lists to Focus group discussions unravelled the local taxonomy of the staple ensure that we captured all crops or varieties grown in their fields. crops and helped us establish a reference list of variety names known by farmers. Because local farmers often replace or mix their seed stock with 3.2.2. Seed network survey same-variety seeds from other farmers, we decided to work at two levels: To capture the full diversity of local staple crops cultivated by the seed lot and variety. Following Louette and Smale, 2000, we use the household, we conducted a survey with all male and female adults living term “seed lot” to refer to a particular physical batch of seeds of a variety in the household who had cultivated a plot during the cropping season that farmers’ maintain through time without mixing it and that they use before fieldwork. For the analyses, individual data (n = 258) were to produce next season’s crops. We use the term ‘variety’ to refer to the aggregated at household level (n = 117). emic categories identified by farmers as a management unit composed The survey had three sections: (1) Farmers’ individual information by seed lots of the same kind, corresponding to plants with similar (i.e., age, gender); (2) household information (i.e., number of adults in phenotypic characteristics according to farmers’ perspective (Louette the household, age and gender of household head(s), cultivated area for and Smale, 2000). While acknowledging that introduction of crop va­ each staple crop, number of market assets owned by the household), and rieties is a dynamic process and categorization not always exclusive, we (3) source of each cultivated seed lot for: i) the most recent cropping adopted the 30 years threshold proposed in the literature to distinguish season, ii) the most recent external acquisition (i.e., different than self- produced), and iii) the very first external acquisition (i.e., first time they obtained a particular variety). For each seed lot and variety cultivated, 7 We define a household as a group of people (normally belonging to the we asked for the number of years since the first acquisition. same extended kin group) pooling resources, including exchanging labour For each seed transaction, we also documented: i) its nature (i.e., without payment and “eating from the same pot”. Among the Bassari, there are exchange, purchase, credit, gift/inheritance); ii) seed giver type (i.e., monogamous and polygamous households and it is common that several gen­ erations live together. Household members can exchange or sell part of their own harvest, but they have to provide part of the harvest for cooking common 8 daily meals. Intra-household seed exchanges are a common way of seed We use the term ‘farmer varieties’ instead of the most common term provisioning. ‘landraces’, to emphasize that our classification is based on farmers’ reports. 5 A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 Table 2 Network-level and node-level measures calculated to describe the seed circulation networks of each of the local staple crops. Measure Definition Network-level Size Total number of nodes measures Density Ratio between the number of existing versus possible links Reciprocity The proportion of mutual connections in a directed graph Modularity The share of internal links in the subgroups from the expected number of links if the distribution was random. Members of the same subgroup (also called ‘communities’ in social networks’ literature) exchange more with other members of the same subgroup than with no-members (compared to what we would expect if seed circulation was random). Number of independent Number of connected subgraphs where all actors are directly or indirectly connected with other actors of the same subgraph components but not with actors belonging to another subgraph. Indegree centralization The sum of differences between each node’s indegree and the one having the maximal indegree (normalized by its theoretical version in the case of an in-star graph) Outdegree centralization The sum of differences between each node’s outdegree and the one having the maximal outdegree (normalized by its theoretical version in the case of an out-star graph) Node-level Indegree The number of incoming ties, which in our case represents the number of seed transactions in which the node was considered as measures a seed receiver. We classify indegree according to giver type (i.e., seed transactions received from households within the village, households from outside the village, projects/NGOs, and market vendors from outside the village). Indegree is not weighted. Outdegree The number of outgoing ties, which in our case represents the number of seed transactions in which a node was mentioned as seed giver. Outdegree is not weighted. Betweenness Number of times a node is at the shortest path in transactions between two other nodes. The shortest path is calculated based on the number of nodes that separates the different actors. Betweenness is not weighted. household within the village, household from outside the village, proj­ farmer varieties, and (iv) all varieties (i.e., the sum of all types). ect/NGO, market vendors from outside the village, other). Market The correlation between a households’ varietal diversity and posi­ vendors from the village were considered as households within the tion in the seed circulation network was only calculated for households village and market vendors from outside the village were classified as within the case-study village. To estimate the association, we fitted market vendors; iii) the location of the seed giver vs. the receiver (i.e., several Poisson generalized linear models (GLM) with a logarithmic link within neighbourhood, different neighbourhood, outside village, and using the “stats” package version 4.1.2 (R Core Team, 2021). Since outside Bassari territory); iv) existing social relation(s) between the seed centrality measures (i.e., indegree, outdegree, betweenness) are not in­ giver and receiver (i.e., age-class, neighbours, friends, kinship, other). dependent, their individual effects might be masked if added in the same model. Therefore, we ran different models for each staple crop (6), va­ riety type (4), and centrality measure (3), adding up to 63 models as 3.3. Data analysis sorghum, Bambara groundnut, and fonio did not have ‘non-farmer va­ rieties’. In our models, we sequentially used the four varietal diversity 3.3.1. Seed circulation networks of local staple crops variables (i.e., all varieties, farmer, introduced, and non-farmer vari­ Seed transaction events recorded in the seed network survey were eties) as outcome and network measures (i.e., indegree, outdegree, aggregated by constructing six seed circulation networks from edge lists, betweenness) as explanatory variables. We ran distinct models for one per crop species. Each edge list contained as many rows as docu­ indegree, outdegree and betweenness, while considering as controls mented seed transactions and two columns: the household that received several additional factors previously suggested to affect the level of crop the seed and the actor who gave the seed. Since our interest was on seed diversity grown by the household - i.e., number of adults, percentage of acquisitions outside the household, seed self-production and intra- female adults, age of the household head, cultivated surface per main household seed transactions were excluded from the edge lists. staple crop, and household’s economic status (Supplementary material For each network, we calculated seven network-level measures and 1). Households with missing data in some explanatory variables were three node-level centrality measures (Borgatti et al., 2018). The inde­ removed from the corresponding models and therefore the sample sizes gree metric was split into four different metrics according to seed giver considered for the final models were smaller than the number of type (see Table 2 for details). Data analysis was done using R version households’ actually growing each staple crop. 4.1.2 (R Core Team, 2021). All networks were represented and network We conducted a backward model selection process to assess the best measures calculated using the R package “igraph” version 1.2.7 (Csardi models according to the Akaike information criterion (AIC). For each and Nepusz, 2006). For network visualization, we considered three staple crop and variety type, we built 3 complete models, each one types of nodes: households, projects/NGOs, and market vendors. Ties including a different centrality measure and all considered explanatory between the nodes are directional, not weighted, and represent the variables (number of varieties and control variables - see Supplementary different seed transactions. To compare the networks across different material 1). The backwards model selection process removed explana­ crops, we conducted an independence test (Pearson’s chi-square test and tory variables one by one, minimizing AIC values, until we got the best when not possible, Fisher’s exact test) concerning crop species and seed fitted model. Even though initial complete models have the same pre­ transaction characteristics (i.e., social relationship, actor type, distance dictors except the centrality measures, the models after the backwards to seed sources, and seed transaction type), and post-hoc tests to assess selection process present different relevant predictors, as all variables individual frequency deviations. For the social relationship involved in without a statistically significant relation with the varietal diversity the seed transaction, when seed giver and receiver shared several ties, maintained by the households (p > 0.1) were removed from the final we only considered the strongest one according to our ethnographic model. Final models only include statistically significant relations. understanding of the case-study area. Analyses were carried out using the package “stats” version 4.21. 4. Results 3.3.2. Relation between household centrality and varietal diversity 4.1. Diversity of local staple crops in Bassari fields We calculated households’ varietal diversity for each main crop using data from the crop diversity inventories. For each household and Crop diversity richness and distribution largely varied across crop, we generated four variables capturing the total number of culti­ households. Peanut, maize, and rice were the most cultivated crops by vated (i) farmer varieties; (ii) introduced farmer varieties; (iii) non- 6 A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 the 117 households interviewed, often occupying large surfaces. Fonio We found differences in transaction types among crop species. Gifts/ and sorghum were also cultivated in relatively large surfaces, but by inheritance and exchanges were the predominant ways of acquiring fewer households. Bambara groundnut was cultivated by most house­ seeds for all crops. Seed purchases were reported sporadically (approx. holds, but in very small surfaces (Table 3). 5% of seed acquisitions) for all crops except for Bambara groundnut, for At the village level, we identified 12 varieties of peanut, 10 of maize, which no seed acquisitions involving monetary transactions were re­ 9 of rice, 6 of sorghum, 4 of fonio, and 4 of Bambara groundnut. On ported. Conversely, peanut was the only crop for which purchase (from average, households maintained 1–2 varieties for each staple crop, both market vendors and other households) represented 25.7% of seed except for peanut, for which most households kept more than two va­ acquisitions (p < 0.05). Maize and rice were the only crops for which rieties. For sorghum, Bambara groundnut, and fonio, farmer varieties credit from projects/NGOs was a common way of sourcing seeds (in predominated and non-farmer varieties were not reported. Introduced 11.36% and 30.6% of seed acquisitions, respectively) (Fig. 2D). farmer varieties and non-farmer varieties predominated for maize and Seed lot renewal rate differed between crop species. Bambara rice. Peanut non-farmer varieties were rare, and households mostly grew groundnut, peanut, rice, and fonio had the highest seed lot renewal rates peanut introduced farmer varieties and farmer varieties (Table 3). For (mean seed lot age < 6 years). However, the difference in seed lot age each crop, few varieties were cultivated by most households and most was only statistically significant for Bambara groundnut (mean seed lot varieties were cultivated by only one or two households. age = 2 years; p < 0.05). Sorghum seed lots were renewed less often than seed lots from other crops (mean seed lot age = 11 years; p < 0.05). Sorghum and Bambara groundnut varieties were the ones kept in the 4.2. Networks of seed circulation of local staple crops households for the longest time (mean variety age = 28 years; p < 0.05) and rice varieties were the most recently acquired ones (mean variety 4.2.1. Socio-demographic characteristics of seed circulation age = 7 years; p < 0.05) (Fig. 3). We found statistically significant differences among seed transaction Finally, we found gendered differences in seed circulation. Women characteristics per crop: distance to the seed source (χ-squared = were mostly involved in Bambara groundnut, fonio, rice, and peanut 151.16, df = 30, p < 0.05), social relationship (χ-squared = 145.77, df = seed circulation, and men played a more prominent role in sorghum and 35, p < 0.05), seed transaction type (χ-squared = 445.26, df = 20, p < maize seed circulation (Table 4). 0.05), and type of actor mobilized to acquire seeds (χ-squared = 288.05, df = 15, p < 0.05). Seed lot and variety age also differed between crop 4.2.2. Seed network structure and composition species (p < 0.05). We found different seed circulation patterns among different crops. For all crops, most seed acquisitions took place within the case-study In terms of network composition, seeds were mostly acquired from village (Fig. 2C). Sorghum seed acquisitions had the lowest geographical households. Projects/NGOs only played an important role in the maize spread, with most seed acquisitions taking place within the neighbour­ and rice networks. Market vendors were relevant for peanut seed ac­ hood (70.59%, p < 0.05) (Supplementary material 2). Bambara quisitions (Fig. 4). groundnut and peanut seed acquisitions had the widest geographical Seed circulation networks varied in size among crops, with peanut spread (p < 0.05), with seeds flowing in from villages within the Bassari showing the largest (210 nodes) and fonio the smallest network (61 territory, mostly within Senegal (30%), but also from Guinea (7%). nodes). All networks had low densities, i.e., low number of connections Peanut seed acquisitions within the village (but outside the neighbour­ or ties (from 0.007 for Bambara groundnut to 0.016 for fonio), and low hood) were larger than for other crops (p < 0.05), and acquisitions from reciprocity (from 0 for maize to 0.063 for peanut), meaning that farmers outside the Bassari territory were also more frequent than for other crops giving or selling seeds rarely got back seed of the same crop from the (3.04% taking place outside the Bassari territory), but this difference same person (Table 4). was not statistically significant (p > 0.05). All networks presented relatively low indegree centralization indices Most households mobilized kinship relationships to acquire seeds. (from 0.027 for sorghum to 0.053 for peanut), meaning that there was However, kinship relations were less important for acquiring maize not a single actor concentrating most seed acquisitions. Outdegree (46.7% of seeds acquisitions) and rice seeds (39.9%) than for other crops centralization indices showed a higher variation among crops (from (>60% of seeds acquisitions, p < 0.05). For all crops the second and 0.046 for sorghum to 0.216 for rice), suggesting that seed sourcing is third most important ties mobilized for seed acquisition were neigh­ more concentrated for certain crops than for others. For maize, rice, and bours and members of the same age-class. Other types of ties, such as peanut, projects/NGOs and market vendors were central actors in terms friendship or same group membership (e.g., church, sports, or women of seed sourcing, whereas for sorghum, Bambara groundnut, and fonio, groups), were rarely mentioned (Fig. 2A). seed sourcing was less concentrated (Table 4). Farmers acquired seeds from a diversity of actors. Irrespectively of In all the networks, we found positive modularity scores (from 0.566 the crops, most seeds were acquired from other farmers. Other farmers for peanut to 0.832 for sorghum), indicating the presence of subgroups were the predominant providers of Bambara groundnut (100%) and within the networks. Modularity scores were higher for the sorghum, sorghum (96%) seeds, and less dominant in rice seed acquisitions Bambara groundnut, fonio, and maize networks than for the rice and (65.7%, p < 0.05). For maize and rice, projects/NGOs played a more peanut networks (Table 4). Concerning independent components, rice significant role as seed providers compared to other crops (19.5% and and peanut networks had fewer independent components than networks 33.1% respectively, p < 0.05). Peanut was the only network where seed from other crops, implying that most actors were connected with each acquisitions from market vendors were important (11.5%, p < 0.05) other (for both crops the main component comprised >90% of actors). (Fig. 2B). Table 3 Crop diversity maintained in Bassari fields, per crop. Sorghum Bambara groundnut Fonio Maize Rice Peanut Number of households in the case-study village that cultivate the crop (total, %) 79 (67.52) 98 (83.76) 50 (42.73) 115 (98.29) 102 (87.18) 116 (99.14) Mean surface cultivated per household (in cordes) 0.99 0.52 1.07 1.84 1.55 2.14 Number of varieties (total in the community, mean per household) 6 (1.33) 4 (1.32) 4 (1.02) 10 (1.58) 9 (1.34) 12 (2.53) Farmer varieties (total, % of households that grow them) 5 (100) 2 (82.98) 3 (85.37) 3 (20) 3 (1.01) 3 (87.93) Farmer introduced varieties (total, % of households that grow them) 1 (2.74) 2 (37.23) 1 (17.1) 5 (68.69) 4 (69.7) 8 (91.38) Non-farmer varieties (total, % of households that grow them) 0 (0) 0 (0) 0 (0) 2 (43.47) 2 (55.6) 1 (6.03) 7 A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 Fig. 2. Seed transactions characteristics (in %), per crop. In A, when giver and receiver shared several ties, only the strongest one was considered. From higher to lower strength: kinship, age-class, friends, neighbours, project beneficiaries, members of the same group, client. Fig. 3. Number of years that each seed lot and variety has been kept in the household. Calculated based on the most recent seed transaction that each household did to renew a seed lot or to acquire a new variety. 8 A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 Table 4 Descriptive characteristics of seed circulation networks, per crop. Sample sizes: sorghum (n = 79 households), Bambara groundnut (n = 98), fonio (n = 50), maize (n = 115), rice (n = 102), peanut (n = 116). Sorghum Bambara groundnut Fonio Maize Rice Peanut Socio-demographic Receivers’ age (mean, min-max) 50 (26–76) 45 (16–78) 46 (19–71) 48 (24–86)) 43 (12–86) 44 (14–86) Receivers’ gender (% of women) 27.73 99 79.31 39.82 86.33 93.39 Seed circulation network measures Size 107 173 61 167 119 210 Density 0.010 0.007 0.016 0.008 0.013 0.010 Reciprocity 0.055 0.010 0.036 0 0.011 0.063 Modularity 0.832 0.767 0.809 0.719 0.630 0.566 Number of independent components 17 17 11 17 3 3 Indegree centralization 0.027 0.034 0.034 0.034 0.038 0.053 Outdegree centralization 0.046 0.063 0.068 0.155 0.216 0.105 The most fragmented network was that of fonio, with 11 independent indegree from households within and outside the village (but not inde­ components, the main one concentrating only 31.15% of network actors, gree from Projects/NGOs) were significantly related to sorghum, fonio, followed by sorghum, Bambara groundnut, and maize, with 17 inde­ and peanut varietal diversity. Contrastingly, connections outside the pendent components each, the main one containing 55.14%, 71.67%, village did not have an important contribution to Bambara groundnut, and 79.64% of actors, respectively. maize, and rice household’s varietal diversity. For these crops, diversity was only related to the indegree from households within the village (Table 5). 4.3. Households’ centrality and varietal diversity 4.3.2. Farmer varieties Overall, households’ varietal diversity was associated to different We found a positive and statistically significant association between centrality measures, depending on the crop species and variety type. the diversity of farmer varieties grown by the household and indegree Across models, household indegree and betweenness most consistently from households from inside the village for sorghum, Bambara showed statistically significant associations with varietal diversity. groundnut, and fonio. Only for sorghum the same association was found Household’s outdegree does not appear in the final models because it between households’ diversity of farmer varieties and indegree from was not associated in a statistically significant way to varietal diversity households outside the village (Table 5). We did not find any statistically for any crop. Cultivated surface, age of the household head, households’ significant association between the number of farmer varieties grown economic status, and the number of adults in the household were posi­ and household’s indegree in the maize, rice, and peanut networks. tively associated with on-farm varietal diversity (Table 5). 4.3.3. Introduced farmer varieties 4.3.1. All varieties Households’ diversity of maize, rice, and peanut introduced farmer Household’s indegree showed a positive and statistically significant varieties was associated with household’s indegree and/or betweenness. association with the total number of varieties cultivated by a household This association was significant when considering indegree from (‘all varieties’, Table 5; Supplementary material 3). Household’s Fig. 4. Seed circulation networks per crop (Fruchterman-Reingold representation). A) Sorghum, B) Bambara groundnut, C) fonio, D) maize, E) rice, F) peanut. 9 A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 Table 5 Generalized linear model results. Associations between households’ centrality in different crop networks and the household’s varietal diversity. Only associations statistically significant are shown. Sample sizes: sorghum (n = 79 households), Bambara groundnut (n = 98), fonio (n = 50), maize (n = 114), rice (n = 102), peanut (n = 116). Crop Explanatory variables Outcome variables All varieties Farmer Introduced farmer Non-farmer varieties varieties varieties Sorghum IN households’ village 7.29e-05 *** 0.000215 *** IN households’ outside village 0.0502 + 0.080072 + Bambara groundnut IN households’ village 0.0747 + 0.0423 * Cultivated surface 0.0174 * Fonio IN households’ village 0.00706 ** 0.0258 * IN households’ outside village 0.06935+ Cultivated surface 0.043488 * Indegree (IN), final models Maize IN households’ village 3.99e-05*** 3.83e-05*** IN projects / NGOs 0.08508 + 4.59e-05 *** Household head age group 0.0248 * Rice IN households’ village 0.0521+ 0.04485 * IN projects / NGOs 0.00315 ** Cultivated surface 0.0768+ 0.04092 * Peanut IN households’ village 3.13e-06 *** 2.54e-06 *** IN households’ outside village 0.0946+ 0.0562+ IN projects / NGOs 1.35e-05 *** Bambara Cultivated surface 0.0295 * 0.000818 *** 0.0174 * groundnut Household economic status 0.085400 + Fonio Cultivated surface 0.0145 * 0.0814 + 0.043488 * Maize BET 0.0722 + Betweenness (BET), final Cultivated surface 0.0707 + models Household head age group 0.0394 * Rice BET 0.03969 * Cultivated surface 0.00721 ** 0.022007 * Peanut Cultivated surface 0.0425 * 0.0102 * Household economic status 0.0711 + Household adults 0.0925 + p-values: +, *, **, *** at the 0.1, 0.05, 0.01 and 0.001 levels. households within the village for maize and from households within and taking place at different moments in time and do not consider that the outside the village for peanut. For maize and rice, household’s varietal timings of interactions differ across crops. While this method allows us diversity of introduced farmer varieties was also associated to house­ to capture the origin of seeds cultivated nowadays, it could contribute to hold’s betweenness. mask the contemporary mechanisms driving seed circulation (e.g., hampering the identification of seed donor-hubs). Research considering 4.3.4. Non-farmer varieties only the most recent seed acquisitions could disclose mechanisms at play Household’s diversity of maize, rice, and peanut non-farmer varieties that were blurred by our methodological choice. Future research should was positively and significantly associated to household’s indegree from conduct diachronic analysis to assess how seed circulation changes projects/NGOs. Only for rice, the households’ diversity of non-farmer through time and the possible role of external actors in shaping these varieties was also associated with household’s indegree from other changes. Second, we constructed the networks aggregating individual households within the village. data at household level. However, taking the household as a unit of analysis does not allow to quantitatively assess the importance of some 5. Discussion key biocultural variables, such as gender, for seed access and seed network structure. Previous research has shown the importance of intra- The main result of our work is that several seed circulation networks household dynamics when studying seed access, highlighting diversity operate in the same community and at the same time. This can arguably within the household (Wencélius et al., 2016). In this line, future be explained by differences in the factors determining household’s ac­ research should look at the relationship between individual farmers’ cess to and maintenance of seeds of various crops and varieties. This centrality in the seed circulation network and the agrobiodiversity they result advances two central arguments about the importance of seed manage, which might show important factors now masked by our circulation networks for access to crop diversity and, therefore, for the household analysis. Third, the paper has focused on the effect of social-ecological resilience of farming communities. First, seed circula­ households’ position in seed acquisition networks for cultivated di­ tion is shaped by crops’ and varieties’ biocultural traits and second, versity, leaving aside impacts of other factors (included only as control households’ centrality in the seed circulation network is related to on- variables). We acknowledge that other variables beyond the ones farm crop diversity. Before discussing the main results, we note that considered in this study could also influence households’ position in the our results suffer from several potential shortcomings. network and on-farm diversity – e.g., diversity of source types and dis­ tance to source type. Fourth, our classification of varieties is based on farmers’ reports. Previous research has shown that non-farmer varieties 5.1. Limitations provided by extension services are frequently acquired by farmers through peer-to-peer seed circulation (Labeyrie et al., 2014a; Teeken First, these seed circulation networks aggregate seed transactions 10 A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 et al., 2012). As a result, these varieties often end up being perceived as Among the Bassari, the circulation of seeds of traditional crops is farmer varieties by farmers. In our study, this may have led to an un­ particularly embedded in kin, gender and age-class networks. This result derestimation of the number of non-farmer varieties reported, particu­ is in line with previous research that describes seed circulation as larly concerning peanut. Despite the long history of peanut variety embedded in pre-existing social structures and connected to farmers’ improvement in Senegal, our inventory only documented one non- social identity (Labeyrie et al., 2014b; Leclerc and Coppens d’Eeck­ farmer variety. Fifth, our analysis treats each crop as independent; enbrugge, 2012). For example, as in other small-scale societies (Díaz- however, a more comprehensive approach could adopt a relational Reviriego et al., 2016; Howard, 2006), Bassari women play a more perspective that views crops and seed exchange networks as components important role in household seed provisioning than men, conferring of a complex system. Future research could explore potential links be­ them social status and cultural recognition. They also play an important tween the diffusion of different crops and varieties. This could involve role in the maintenance of communal social relations, household food investigating whether crops that play a similar role in the agroecosystem security, and generally in caregiving. However, there are also gender- (e.g., that occupy a similar position in the crop rotation) circulate related social differences in seed circulation networks. For example, together or through the same networks. sorghum seeds are traditionally considered ‘the’ men crop, and mostly circulate among men. 5.2. Seed circulation is shaped by crops’ biocultural traits The actors involved also influence seed flows and seed circulation network structure. Agricultural development agendas and research Our findings suggest that seeds from different crops do not circulate priorities have historically prioritized the most profitable crops –e.g., in the same way because crops differ on their biocultural traits. Previous peanut production for export during colonial and early post-colonial work has shown that crops’ biological characteristics influence seed times – or cereal crops with high yields, important for the country’s production, viability, and availability (Ellen and Platten, 2011). Seed food security strategy, i.e., rice and maize (Porcuna-Ferrer et al., under circulation is also shaped by social relations, cultural rules, and symbolic review). Consequently, agricultural extension projects, NGOs, and local values associated with crops, which guide farmers’ practices like selec­ markets have also made available seeds of these crops. According to our tion, management, storage, and uses (Delêtre et al., 2011; Labeyrie et al., work, when external actors play a prominent role, seed circulation 2016; Leclerc and Coppens d’Eeckenbrugge, 2012). Moreover, human networks tend to have higher centralization indices. For example, for history might also affect the type of actors involved in seed circulation introduced crops (i.e., maize, rice, and peanut), projects/NGOs and and the geographic distribution of crop diversity – e.g., (van Etten, market vendors play a relevant role, and fewer actors concentrate more 2006). seed sourcing, which risks reinforcing or creating (new) power dynamics Our results add to previous studies highlighting that, even within a and structural inequalities in the local communities. village, seed acquisition networks vary across crop species, arguably In the absence of longitudinal data, it is difficult to assess trends in because of differences in crops’ biocultural traits. For example, Bambara the coexistence of different seed circulation systems. However, based on groundnut’s network is larger, has lower density, lower modularity, and our data, we interpret the effect of agricultural extension projects and wider geographical spread than other traditional Bassari crops, showing NGOs in seed circulation in two complementary ways. First, projects/ that households participate more actively in exchanging, giving, and NGOs-interventions result in centralized seed diffusion models, which acquiring Bambara groundnut than sorghum or fonio seeds. Our might displace (decentralized) traditional mechanisms / institutions, ethnographic understanding suggests that this is at least partly related to like kinship or age-class that have secured access to seeds for genera­ seed lot renewal rate, which is connected to legume’s seed storage tions, and create new social networks, increasing the centrality of qualities. Bambara groundnut (and peanut) seeds are highly vulnerable farmers supported by the projects, NGOs or extension services (Isaac to pest attacks, which forces farmers to frequent seed lot renewal. et al., 2021). The substitution of traditional sources by new, institutional However, crop biology alone does not suffice to explain seed network sources could lead to overcentralized networks, potentially constraining structure, as demonstrated by differences in seed circulation between local communities’ social-ecological resilience (Cretney, 2014; Pelling the two legumes. The peanut seed circulation network was the largest, and Manuel-Navarrete, 2011). best connected, most broadly spread, and most reciprocal of all the seed Broader socio-economic pressures leading to the rapid trans­ networks studied, which might be related with the specific historical formation of smallholder farming systems and to the weakening of circumstances of cultivation. traditional systems of seed sharing also affect how biocultural factors The commercial production of peanuts for export was promoted by shape seed circulation networks. Local cultivation of sorghum, fonio, the French colonial regime in the 19th century and by post-colonial and Bambara groundnut is in recession and farmers increasingly rely on governments after Senegal’s independence (year 1960). Although pea­ projects/NGOs to acquire seeds of newly introduced crops. Market nuts are no longer the main focus of national policies and nowadays integration, which locally started with the expansion of peanut culti­ Bassari mostly cultivate peanuts for self-consumption, peanuts continue vation in the 1900s and has exploded since the 2000s with the arrival of to be largely acquired through trade and are even used as alternative NGOs and agricultural development projects that promoted maize and currency (mostly by women). Peanut abundance in local markets dis­ rice cultivation have largely contributed to traditional crop abandon­ courages farmers from keeping their own seed, as they know that, in case ment (Porcuna-Ferrer et al., under review). We show that the networks of need, they will find seed at the market with relative ease. Being of traditional crops have high levels of fragmentation, small size, and widely adopted by the Bassari in the last century, peanuts are less rooted low densities which reflects a high proportion of isolated farmers and a in the local culture than traditional crops and are therefore less subject small number of possible seed exchanges. This reduced circulation of to cultural rules and norms guiding seed circulation. In contrast, seeds increases network fragility, potentially limiting network’s capac­ Bambara groundnut seeds bear an important cultural and symbolic ity to support crop diversity. value in the local community and circulate in a more restricted way, as A second interpretation of the effect of projects/NGOs in seed cir­ shown by the higher number of subgraphs within the seed circulation culation refers to their integration in traditional networks. Traditional network. Sorghum provides another good example of how social norms networks of seed circulation have a high adaptive capacity to channel and cultural values restrict the circulation of seeds of traditional crops. seeds of new crops. In the theory of change behind the African Green Sorghum is an important crop for Bassari ceremonial life and Bassari Revolution, farmers’ des-centralized seed systems were expected to be men’s identity. Its seeds circulate mostly within men living in the same gradually replaced by ‘formal’ centralized ones – e.g., (Scoones and neighbourhood, which in the case-study village coincides with members Thompson, 2011; Westengen et al., 2023). However, among the Bassari, of the same patrilineage, showing the importance of descent and alliance as it has been the case in several locations worldwide, linkages and in­ for access to sorghum seeds. terdependencies between centralized and des-centralized seed systems 11 A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 have developed (Almekinders and Louwaars, 2002; McGuire and 2020), as farmers capacity to change crop species and varieties is a Sperling, 2016). Despite being relatively ‘new’ crops in the Bassari common response to changing climatic conditions (Ruggieri et al., 2021; farming system and strongly supported by development agendas and the Schlingmann et al., 2021). Building up on social-ecological resilience official seed sector, maize and rice substantially circulate household-to- theory (Cabell and Oelofse, 2012; Walker et al., 2004), farmers’ access household, suggesting that farmers draw on the strengths of the different to crop and varietal diversity should be ensured through a repertoire of seed acquisition systems. flexible responses (i.e., providing farmers with a wide range of source Still, the broader socio-economic dynamics mentioned earlier urge and crop diversity options), accounting for the trade-offs among these for caution when assessing the benefits of co-existing forms of seed responses. For example, while introducing a cash-crop to the on-farm circulation. From an agricultural development perspective, the question crop portfolio can work well to meet market demands and farmers’ is how to improve farmers’ access to high-quality adapted seeds without cash needs, the abandonment of drought-resistant subsistence crops can breaking the tightly connected relationships that have traditionally diminish agroecosystem’s resilience to other stresses, like climate played a pivotal role for the resilience of smallholder farmers (Haider change (McGuire and Sperling, 2013). et al., 2020). In this regard, our data supports previous research that emphasizes the importance of moving beyond the dichotomy of ‘formal/ 6. Conclusion centralized’ versus ‘informal/decentralized’ seed circulation in order to effectively understand and support farmers (McGuire and Sperling, This study shows that farmers’ access to seeds is conditioned by crop 2016; Westengen et al., 2023). It also highlights the need to assess biocultural traits and that farmers’ centrality in the seed circulation existing seed circulation networks when devising any intervention network affects on-farm crop diversity. (Abay et al., 2011). While our findings highlight the instrumental role of farmer-to- farmer seed circulation networks for the maintenance of local crop di­ 5.3. Households’ centrality in the seed circulation network affects access versity and for the introduction of new diversity in the agricultural to crop diversity system, they also indicate that new seed sources, such as local markets, agricultural extension projects or NGOs, can offer access to new seeds Our results show that household centrality in the network of seed with adaptive potential. Considering ongoing climate change impacts in circulation is generally associated with on-farm varietal diversity. Pre­ the Bassari territory, agricultural interventions need to evaluate trade- vious studies have tested this association, finding contrasting results offs between responses and medium- and long- term consequences for (Abizaid et al., 2016; Calvet-Mir et al., 2012; Díaz-Reviriego et al., 2016; farmers’ social-ecological resilience. The challenge remains on how to Kawa et al., 2013). Our results suggest that such contrasting results enable the coexistence of new and old crops and varieties, and of modern might just reflect the fact that there is not a single measure of centrality and traditional ways of accessing seeds. In general, there is a certain that explains on-farm diversity for all the crops and variety types. consensus that horizontal, locally-adapted ways of accessing seeds (e.g., Indeed, owing to the specific crop biocultural traits guiding seed circu­ farmer-to-farmer seed circulation) are more sustainable in the long-run lation, different centrality measures can explain different aspects of than top-down, hierarchical ones (e.g., agricultural extension services), access to crop diversity. but there is also growing agreement that both strategies should be As Kawa et al. (2013) and Abizaid et al. (2016), we did not find an reconciled for more effective crop diversity conservation (Pautasso association between being an important seed provider (i.e., having a et al., 2013; Thomas et al., 2011). high outdegree) and on-farm varietal diversity. However, we found that For the Bassari reconciling both approaches would entail that agri­ the type of actors mobilized for acquiring new seed (indegree types) and cultural extension projects and NGOs shift from the current top-down the level of household intermediation in the seed circulation network approach to multi-centric participatory approaches that situate (betweenness) were differently associated with household’s varietal farmers’ knowledge and practices at the centre. A participatory-based diversity, the association depending on the crop and variety type. This approach would facilitate the broadening of projects’ crop-portfolio, finding suggests that household’s access to different crops and variety including traditional crops like sorghum, Bambara groundnut, and types might depend on its ability to mobilize different types of relations. fonio. This would contribute to strengthening locally adapted crop di­ Access to farmer varieties of traditional crops is best granted through versity and seed systems, thereby bolstering the social-ecological resil­ farmers’ personal network, whereas households with a higher level of ience of smallholder farming communities in the phase of climate and intermediation in the network will hold a more privileged position to global change. access newly arrived varieties, and households that can interact with market-logics and projects/NGOs will probably have better access to Declaration of Competing Interest non-farmer varieties. Our findings also point to the importance of the socio-economic status of the household for accessing crop and varietal The authors declare that there is no conflict of interest. diversity. Specifically, access to land affects cultivated diversity for all crops and most variety types. Household size, age of the household head, Data availability and household’ financial resources all played an important role favouring access to diversity for certain crops and variety types. These The authors do not have permission to share data. results are in line with previous research that point to wealth as a key structuring factor of seed circulation (e.g., Wencélius et al., 2016). Acknowledgements From a social-ecological resilience perspective, farmer varieties are an important reservoir of biocultural memory, as farmers have selected Research leading to this work has received funding from the Euro­ them over generations for their fit to local natural and management pean Research Council (ERC) under an ERC Consolidator Grant (FP7- conditions. Seeds often circulate together with knowledge about their 771056-LICCI).The authors acknowledge the financial support of the characteristics, qualities, and management practices (Calvet-Mir et al., Spanish Ministry of Science, Innovation and Universities, through the 2012; Reyes-García et al., 2013). Both trait diversity and the associated “María de Maeztu” programme for Units of Excellence (CEX2019- knowledge are important legacies to help adapt to new conditions 000940-M). The ideas discussed here have benefited from discussions (Cabell and Oelofse, 2012; Reyes-García et al., 2014). In turn, access to with GDR ReSoDiv, funded by the CNRS. Special thanks to C. Raimond, introduced and non-farmer varieties, particularly to new climate- E. Garine, and J.L. Molina, for their insightful comments and discus­ resilient crops and varieties, can also be an important way for farmers sions, as well as to the two anonymous reviewers whose contributions to cope with or adapt to climate change or other stresses (Acevedo et al., have substantially improved the paper. Thanks as well to B. Klappoth, T. 12 A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 Guillerminet, S. Boubane, A. Bonang, and P. Bindia for their contribu­ Delêtre, M., McKey, D.B., Hodkinson, T.R., 2011. Marriage exchanges, seed exchanges, and the dynamics of manioc diversity. Proc. Natl. Acad. Sci. U. S. A. 108, tion to fieldwork. We are also thankful to the local chiefs in the study 18249–18254. https://doi.org/10.1073/pnas.1106259108. area for opening us the doors of the village. Finally, thanks to all the Díaz-Reviriego, I., González-Segura, L., Fernández-Llamazares, Á., Howard, P.L., Bassari farmers who so generously shared their time and knowledge Molina, J.L., Reyes-García, V., 2016. Social organization influences the exchange with us. and species richness of medicinal plants in amazonian homegardens. Ecol. Soc. 21 https://doi.org/10.5751/ES-07944-210101. Ellen, R., Platten, S., 2011. The social life of seeds: the role of networks of relationships in Appendix A. Supplementary data the dispersal and cultural selection of plant germplasm. J. R. Anthropol. Inst. 17, 563–584. https://doi.org/10.1111/j.1467-9655.2011.01707.x. van Etten, J., 2006. Molding maize: the shaping of a crop diversity landscape in the Supplementary data to this article can be found online at https://doi. western highlands of Guatemala. J. Hist. Geogr. 32, 689–711. https://doi.org/ org/10.1016/j.agsy.2023.103750. 10.1016/j.jhg.2005.12.002. Folke, C., 2004. Traditional knowledge in social–ecological systems. Ecol. Soc. 9. Garine, È., Labeyrie, V., Violon, C., Wencélius, J., Leclerc, C., Raimond, C., 2018. Which References scale to understand seed fluxes in small-scale farming societies? Snapshots of sorghum from Africa. In: Girard, F., Frison, C. (Eds.), The Commons, Plant Breeding and Agricultural Research: Challenges for Food Security and Agrobiodiversity. Abay, F., de Boef, W., Bjørnstad, Å., 2011. Network analysis of barley seed flows in Routledge, London, pp. 159–172. https://doi.org/10.4324/9781315110387. Tigray, Ethiopia: supporting the design of strategies that contribute to on-farm Gessain, M., 1975. Anthropologie écologique des Bassari du Sénégal oriental: Evolution management of plant genetic resources. Plant Genet. Resour. 9, 495–505. https:// du village d’Etyolo depuis 1900. University Paris Pierre et Marie Curie, PhD diss. doi.org/10.1017/S1479262111000773. Gessain, M., 1996. Le sorgho chez des Tenda et des Peul au Sénégal oriental. In: Cuisines: Abizaid, C., Coomes, O.T., Perrault-Archambault, M., 2016. Seed sharing in Amazonian Reflets Des Sociétés. Éditions Karthala, Paris, pp. 97–108. indigenous rain forest communities: a social network analysis in three Achuar Haider, L.J., Boonstra, W.J., Akobirshoeva, A., Schlüter, M., 2020. Effects of villages. Peru. Hum. Ecol. 44, 577–594. https://doi.org/10.1007/s10745-016-9852- development interventions on biocultural diversity: a case study from the Pamir 7. Mountains. Agric. Hum. Values 37, 683–697. https://doi.org/10.1007/s10460-019- Abrouk, M., Ahmed, H.I., Cubry, P., Šimoníková, D., Cauet, S., Pailles, Y., 10005-8. Bettgenhaeuser, J., Gapa, L., Scarcelli, N., Couderc, M., Zekraoui, L., Kathiresan, N., Howard, P.L., 2006. Gender and social dynamics in swidden and homegardens in Latin Čížková, J., Hřibová, E., Doležel, J., Arribat, S., Bergès, H., Wieringa, J.J., Gueye, M., America. In: Kumar, B.M., Nair, P.K.R. (Eds.), Tropical Homegardens. Advances in Kane, N.A., Leclerc, C., Causse, S., Vancoppenolle, S., Billot, C., Wicker, T., Agroforestry, Vol. 3. https://doi.org/10.1007/978-1-4020-4948-4_10. Vigouroux, Y., Barnaud, A., Krattinger, S.G., 2020. Fonio millet genome unlocks Isaac, M.E., Nyantakyi-Frimpong, H., Matouš, P., Dawoe, E., Anglaaere, L.C.N., 2021. African orphan crop diversity for agriculture in a changing climate. Nat. Commun. Farmer networks and agrobiodiversity interventions: the unintended outcomes of 11, 1–13. https://doi.org/10.1038/s41467-020-18329-4. intended change. Ecol. Soc. 26 https://doi.org/10.5751/ES-12734-260412. Acevedo, M., Pixley, K., Zinyengere, N., Meng, S., Tufan, H., Cichy, K., Bizikova, L., Jarvis, D.I., Brown, A.H.D., Pham, H.C., Collado-Panduro, L., Latournerie-Moreno, L., Isaacs, K., Ghezzi-Kopel, K., Porciello, J., 2020. A scoping review of adoption of Gyawali, S., Tanto, T., Sawadogo, M., Mar, I., Sadiki, M., Hue, N.T.N., Arias- climate-resilient crops by small-scale producers in low- and middle-income Reyes, L., Balma, D., Bajracharya, J., Castillo, F., Rijal, D., Belqadi, L., Rana, R., countries. Nat. Plants 6, 1231–1241. https://doi.org/10.1038/s41477-020-00783-z. Saidi, S., Ouedraogo, J., Zangre, R., Rhrib, K., Chavez, J.L., Schoen, D., Sthapit, B., Allinne, C., Mariac, C., Vigouroux, Y., Bezançon, G., Couturon, E., Moussa, D., De Santis, P., Fadda, C., Hodgkin, T., 2008. A global perspective of the richness and Tidjani, M., Pham, J.L., Robert, T., 2008. Role of seed flow on the pattern and evenness of traditional crop-variety diversity maintained by farming communities. dynamics of pearl millet (Pennisetum glaucum [L.] R. Br.) genetic diversity assessed Proc. Natl. Acad. Sci. U. S. A. 105, 5326–5331. https://doi.org/10.1073/ by AFLP markers: a study in South-Western Niger. Genetica 133, 167–178. https:// pnas.0800607105. doi.org/10.1007/s10709-007-9197-7. Kawa, N.C., McCarty, C., Clement, C.R., 2013. Manioc varietal diversity, social networks, Almekinders, C.J.M., Louwaars, N.P., 2002. The importance of the Farmers’ seed Systems and distribution constraints in rural Amazonia. Curr. Anthropol. 54, 764–770. in a Functional National Seed Sector. J. New Seeds 4, 15–33. https://doi.org/ https://doi.org/10.1086/673528. 10.1300/J153v04n01_02. Labeyrie, V., Deu, M., Barnaud, A., Calatayud, C., Buiron, M., Wambugu, P., Manel, S., Altieri, M.A., Nicholls, C.I., 2017. The adaptation and mitigation potential of traditional Glaszmann, J.C., Leclerc, C., 2014a. Influence of ethnolinguistic diversity on the agriculture in a changing climate. Clim. Chang. 140, 33–45. https://doi.org/ sorghum genetic patterns in subsistence farming systems in eastern Kenya. PLoS One 10.1007/s10584-013-0909-y. 9. https://doi.org/10.1371/journal.pone.0092178. ANACIM, 2020. National Weather Data. Agence Nationale de l’Aviation Civile et de la Labeyrie, V., Rono, B., Leclerc, C., 2014b. How social organization shapes crop diversity: Météorologie. an ecological anthropology approach among Tharaka farmers of Mount Kenya. Berkes, F., Colding, J., Folke, C., 2000. Rediscovery of traditional ecological knowledge Agric. Hum. Values 31, 97–107. https://doi.org/10.1007/s10460-013-9451-9. as adaptive management. Ecol. Appl. 10, 1251–1262. https://doi.org/10.1890/ Labeyrie, V., Thomas, M., Muthamia, Z., Leclerc, C., 2016. Seed exchange networks, 1051-0761(2000)010[1251:ROTEKA]2.0.CO;2. ethnicity, and sorghum diversity. Proc. Natl. Acad. Sci. 113, 98–103. https://doi. Bernards, N., 2019. ‘Latent’ surplus populations and colonial histories of drought, org/10.1073/pnas.1513238112. groundnuts, and finance in Senegal. Geoforum 126, 41–450. https://doi.org/ Labeyrie, V., Antona, M., Baudry, J., Bazile, D., Bodin, Ö., Caillon, S., Leclerc, C., Le 10.1016/j.geoforum.2019.10.007. Page, C., Louafi, S., Mariel, J., Massol, F., Thomas, M., 2021a. Networking Bodin, Ö., 2017. Collaborative environmental governance: achieving collective action in agrobiodiversity management to foster biodiversity-based agriculture. A review. social-ecological systems. Science (80) 357. https://doi.org/10.1126/science. Agron. Sustain. Dev. 41 https://doi.org/10.1007/s13593-020-00662-z. aan1114. Labeyrie, V., Renard, D., Aumeeruddy-Thomas, Y., Benyei, P., Caillon, S., Calvet-Mir, L., Bodin, Ö., Robins, G., McAllister, R.R.J., Guerrero, A.M., Crona, B., Tengö, M., Lubell, M., Carrière, M., Demongeot, M., Descamps, E., Braga Junqueira, A., Li, X., 2016. Theorizing benefits and constraints in collaborative environmental Locqueville, J., Mattalia, G., Miñarro, S., Morel, A., Porcuna-Ferrer, A., governance: a transdisciplinary social-ecological network approach for empirical Schlingmann, A., Vieira da Cunha Avila, J., Reyes-García, V., 2021b. The role of crop investigations. Ecol. Soc. 21 https://doi.org/10.5751/ES-08368-210140. diversity in climate change adaptation: insights from local observations to inform Borgatti, S.P., Everett, M.G., Johnson, J.C., 2018. Analyzing Social Networks, 2nd ed. decision making in agriculture. Curr. Opin. Environ. Sustain. 51, 15–23. https://doi. Sage Publications, London. org/10.1016/j.cosust.2021.01.006. Cabell, J.F., Oelofse, M., 2012. An Indicator Framework for Assessing Agroecosystem Leclerc, C., Coppens d’Eeckenbrugge, G., 2012. Social organization of crop genetic Resilience. Ecol. Soc. 17 https://doi.org/10.5751/ES-04666-170118. diversity. The G × E × S interaction model. Diversity 4, 1–32. https://doi.org/ Calvet-Mir, L., Salpeteur, M., 2016. Humans, plants, and networks: a critical review. 10.3390/d4010001. Environ. Soc. Adv. Res. 7, 107–128. https://doi.org/10.3167/ares.2016.070107. Louette, D., Smale, M., 2000. Farmers’ seed selection practices and traditional maize Calvet-Mir, L., Calvet-Mir, M., Vaqué-Nuñez, L., Reyes-Garcia, V., 2011. Landraces In varieties in Cuzalapa, Mexico. Euphytica 113, 25–41. https://doi.org/10.1023/A: Situ Conservation: A Case Study in High-Mountain Home Gardens in Vall Fosca, 1003941615886. Catalan Pyrenees, Iberian Peninsula. Economic Botany 65, 146–157. https://doi. Maffi, L., 2012. What is biocultural diversity?. In: Biocultural Diversity Conservation. org/10.1007/s12231-011-9156-1. Routledge, pp. 3–11. Calvet-Mir, L., Calvet-Mir, M., Molina, J.L., Reyes-García, V., 2012. Seed exchange as an Mayes, S., Ho, W.K., Chai, H.H., Song, B., Chang, Y., Massawe, F., 2019. Bambara agrobiodiversity conservation mechanism: a case study in Vall Fosca, Catalan groundnut (Vigna Subterranea (L) Verdc): A climate smart crop for food and Pyrenees, Iberian Peninsula. Ecol. Soc. 17 https://doi.org/10.5751/ES-04682- nutrition security. In: Kole, C. (Ed.), Genomic Designing of Climate-Smart Pulse 170129. Crops. Springer, Cham, pp. 397–424. https://doi.org/10.1007/978-3-319-96932-9. Coomes, O.T., McGuire, S.J., Garine, E., Caillon, S., McKey, D., Demeulenaere, E., McGuire, S., Sperling, L., 2011. The links between food security and seed security: facts Jarvis, D., Aistara, G., Barnaud, A., Clouvel, P., Emperaire, L., Louafi, S., Martin, P., and fiction that guide response. Dev. Pract. 21, 493–508. https://doi.org/10.1080/ Massol, F., Pautasso, M., Violon, C., Wencélius, J., 2015. Farmer seed networks make 09614524.2011.562485. a limited contribution to agriculture? Four common misconceptions. Food Policy 56, McGuire, S., Sperling, L., 2013. Making seed systems more resilient to stress. Glob. 41–50. https://doi.org/10.1016/j.foodpol.2015.07.008. Environ. Chang. 23, 644–653. https://doi.org/10.1016/j.gloenvcha.2013.02.001. Cretney, R., 2014. Resilience for Whom? Emerging Critical Geographies of Socio- McGuire, S., Sperling, L., 2016. Seed systems smallholder farmers use. Food Secur. 8, ecological Resilience. Geogr. Compass 8, 627–640. https://doi.org/10.1111/ 179–195. https://doi.org/10.1007/s12571-015-0528-8. gec3.12154. Csardi, G., Nepusz, T., 2006. The igraph software package for complex network research. InterJ. Compl. Syst. 1695. 13 A. Porcuna-Ferrer et al. Agricultural Systems 211 (2023) 103750 McKey, D., Elias, M., Pujol, M.E., Duputié, A., 2010. The evolutionary ecology of clonally agricultural knowledge in home gardens of the Iberian Peninsula. Glob. Environ. propagated domesticated plants. New Phytol. 186, 318–332. https://doi.org/ Chang. 24, 223–231. https://doi.org/10.1016/j.gloenvcha.2013.11.022. 10.1111/j.1469-8137.2010.03210.x. Ruggieri, F., Porcuna-Ferrer, A., Gaudin, A., Faye, N.F., Reyes-García, V., Labeyrie, V., Meikle, W.G., Markham, R.H., Nansen, C., Holst, N., Degbey, P., Azoma, K., Korie, S., 2021. Crop diversity management: Sereer Smallholders’ response to climatic 2002. Pest management in traditional maize stores in West Africa: a farmer’s variability in Senegal. J. Ethnobiol. 41, 389–408. https://doi.org/10.2993/0278- perspective. J. Econ. Entomol. 95, 1079–1088. https://doi.org/10.1093/jee/ 0771-41.3.389. 95.5.1079. Schlingmann, A., Graham, S., Benyei, P., Corbera, E., Martinez Sanesteban, I., Nolan, R.W., 1986. Bassari Migrations: The Quiet Revolution. Routledge, New York. Marelle, A., Solemany-Fard, R., Reyes-García, V., 2021. Global patterns of Nuijten, E., Almekinders, C.J.M., 2008. Mechanisms explaining variety naming by adaptation to climate change by indigenous peoples and local communities. A farmers and name consistency of rice varieties in the Gambia. Econ. Bot. 62, systematic review. Curr. Opin. Environ. Sustain. 51, 55–64. https://doi.org/ 148–160. https://doi.org/10.1007/s12231-008-9012-0. 10.1016/j.cosust.2021.03.002. Pautasso, M., Aistara, G., Barnaud, A., Caillon, S., Clouvel, P., Coomes, O.T., Delêtre, M., Scoones, I., Thompson, J., 2011. The politics of seed in Africa’ s green revolution : Demeulenaere, E., De Santis, P., Döring, T., Eloy, L., Emperaire, L., Garine, E., alternative narratives and competing. IDS Bulletin 42 (4), 1–23. https://doi.org/ Goldringer, I., Jarvis, D., Joly, H.I., Leclerc, C., Louafi, S., Martin, P., Massol, F., 10.1111/j.1759-5436.2011.00232.x. McGuire, S., McKey, D., Padoch, C., Soler, C., Thomas, M., Tramontini, S., 2013. Smith, C.W., Frederiksen, R.A., 2000. Sorghum: Origin, History, Technology, and Seed exchange networks for agrobiodiversity conservation. A review. Agron. Sustain. Production, Vol.2. John Wiley & Sons. Dev. 33, 151–175. https://doi.org/10.1007/s13593-012-0089-6. Tardío, J., Aceituno-Mata, L., Molina, M., Morales, R., Pardo-de-Santayana, M., 2018. Pelling, M., Manuel-Navarrete, D., 2011. From Resilience to Transformation: the Inventario Espanol de Conocimientos Tradicionales relativos a la Biodiversidad Adaptive Cycle in Two Mexican Urban Centers. Ecol. Soc. 16 https://doi.org/ Agrícola. Madrid, Spain. 10.5751/ES-04038-160211. Teeken, B., Nuijten, E., Temudo, M.P., Okry, F., Mokuwa, A., Struik, P.C., Richards, P., Porcuna-Ferrer, A., Calvet-Mir, L., Faye, N.F., Klappoth, B., Reyes-García, V., Labeyrie, 2012. Maintaining or abandoning African Rice: lessons for understanding processes V., under review. The Decline of Drought-tolerant Indigenous Crops in A Climate of seed innovation. Hum. Ecol. 40, 879–892. https://doi.org/10.1007/s10745-012- Change Context: a Historical Political Agroecology Account of the Bassari. South- 9528-x. eastern Senegal. Thomas, M., Caillon, S., 2016. Effects of farmer social status and plant biocultural value Porcuna-Ferrer, A., Calvet-Mir, L., Guillerminet, T., Alvarez-Fernandez, S., Labeyrie, V., on seed circulation networks in Vanuatu. Ecol. Soc. 21 https://doi.org/10.5751/ES- Porcuna-Ferrer, E., Reyes-García, V., 2023. "So many things have changed”: Situated 08378-210213. understandings of climate change impacts among the Bassari, south-eastern Senegal. Thomas, M., Dawson, J.C., Goldringer, I., Bonneuil, C., 2011. Seed exchanges, a key to Environmental Science & Policy 148, 10355. https://doi.org/10.1016/j. analyze crop diversity dynamics in farmer-led on-farm conservation. Genet. Resour. envsci.2023.103552. Crop. Evol. 58, 321–338. https://doi.org/10.1007/s10722-011-9662-0. Porcuna-Ferrer, A., Guillerminet, T., Klappoth, B., Schlingmann, A., in press. Agricultural UNESCO, 2012. Decisions Report – 36th Session of the World Heritage Committee. Saint- adaptation to multiple stressors in a climate change context. A case study in south- Petersburg. eastern Senegal. In: Reyes-García and LICCI Team, Routledge Handbook of Climate Walker, B., Holling, C.S., Carpenter, S.R., Kinzig, A., 2004. Resilience, adaptability and Change Impacts on Indigenous Peoples and Local Communities. Routledge. transformability in social-ecological systems. Ecol. Soc. 9 (2) https://doi.org/ R Core Team, 2021. R: A language and environment for statistical computing. 10.5751/ES-00650-090205. Renard, D., Tilman, D., 2019. National food production stabilized by crop diversity. Wencélius, J., Thomas, M., Barbillon, P., Garine, E., 2016. Interhousehold variability and Nature 571, 257–260. https://doi.org/10.1038/s41586-019-1316-y. its effects on seed circulation networks: a case study from northern Cameroon. Ecol. Reyes-García, V., Molina, J.L., Calvet-Mir, L., Aceituno-Mata, L., Lastra, J.J., Soc. 21 https://doi.org/10.5751/ES-08208-210144. Ontillera, R., Parada, M., Pardo-de-Santayana, M., Rigat, M., Vallès, J., Garnatje, T., Westengen, O.T., Paule Dalle, S., Hunduma Mulesa, T., 2023. Navigating Toward 2013. “Tertius Gaudens”: Germplasm Exchange Networks and Agroecological Resilient and Inclusive Seed Systems. Proceedings of the National Academy of Knowledge among Home Gardeners in the Iberian Peninsula. J. Ethnobiol. Sciences 120 (14). https://doi.org/10.1073/pnas.2218777120 1. Ethnomedi. 9 (1), 1–11. https://doi.org/10.1186/1746-4269-9-53. Zimmerer, K.S., de Haan, S., Jones, A.D., Creed-Kanashiro, H., Tello, M., Carrasco, M., Reyes-García, V., Aceituno-Mata, L., Calvet-Mir, L., Garnatje, T., Gómez-Baggethun, E., Meza, K., Plasencia Amaya, F., Cruz-Garcia, G.S., Tubbeh, R., Jiménez Olivencia, Y., Lastra, J.J., Ontillera, R., Parada, M., Rigat, M., Vallès, J., Vila, S., Pardo-de- 2019. The biodiversity of food and agriculture (agrobiodiversity) in the Santayana, M., 2014. Resilience of traditional knowledge systems: the case of anthropocene: research advances and conceptual framework. Anthropocene 25. https://doi.org/10.1016/j.ancene.2019.100192. 14