This work has benefited from state funding managed by the Agence Nationale de la Recherche under (i) the Investissements d'avenir program with reference ANR-10-EQPX-17 (Centre d'accès sécurisé aux données - CASD) and the PEPR Systèmes Alimentaires, Microbiome et Santé (PLAT4TERFOOD project). We would like to thank the reviewers of this article for their invaluable advice, which helped us improve the article. We would also like to thank Laura Sayre for translating this paper.

Introduction

1In the decades since the Second World War, French and European agricultural systems have undergone an array of technical, economic, and social changes, as amply described by previous research (e.g., Servolin, 1989; Malassis, 1997; Allaire and Daviron, 2017). These include the rationalization of agricultural field systems, increased dependence on fossil fuels and synthetic inputs, the large-scale substitution of capital for labor, a massive decline in the agricultural population, the economic concentration of the means of production, a simplification and standardization of production techniques, and the integration of agricultural products into increasingly internationalized processing and marketing chains. Many of these features of commodity agriculture (Lyson and Guptill, 2004) have in turn generated significant negative externalities for society, the economy, and the environment (removal of hedgerows, straightening of water ways, etc.; Croix, 1999; Preux et al., 2015; Paysant et al., 2021). In response to these negative impacts, a variety of public policies have been developed to constrain and/or incentivize farmers to better protect environmental resources (Horlings and Marsden, 2011; Duru et al., 2015).

2Today, faced with the need to decarbonize agriculture, preserve biodiversity and address the interrelated challenges of economic viability and food security, European policies are showing increased support for a profound transformation of agricultural and food systems. Nevertheless, the Common Agricultural Policy (CAP) 2023-2027 assigns only a limited role to agroecology (Charbonneau and Thivet, 2022), and a recent analysis suggests that a large number of French farms could qualify for the new CAP eco-scheme without any change to their current practices (Lassalas et al., 2023). Other research (Bryngelsson et al., 2016; Willett et al., 2019; Aubert et al., 2021) converges around the idea that a sustainable transition of food systems in developed countries involves, at least in theory, four transformation levers: a shift toward plant proteins and the reduced production and consumption of meat; curtailing losses and waste in the food chain; shortening the distance between links in the food supply chain (farmers, processors, distributors, consumers); and, finally, a transformation of agricultural production methods at the field level via a diversification of systems and a massive reduction in the use of synthetic inputs. This transition of the agricultural and food system must be compatible with social justice issues, guaranteeing not only access for all to healthy, affordable food, but also maintaining jobs and incomes for all players in the sector. Various transition strategies have prompted concerns as to their ability to reconcile biophysical and socio-economic issues, given the transformations they entail.

3This article contributes to ongoing work studying agro-ecological transitions. Agronomists, sociologists, and economists, for example, have all described the diversity of forms of sustainable agriculture (Féret and Douglets, 2001; Laurent et al., 2003; Therond et al., 2017; Plumecocq et al., 2018) and sought to assess their performance (Midler et al., 2019). Although early geographical and agronomic studies considered the question of agricultural change at the macro-regional (Mignolet, 2005) or even national (Guermond, 1976) levels, more recent work has looked more specifically at agri-food transitions (Bermond et al., 2019) or agro-ecological transitions (Arnaud de Sartre et al., 2019), supported by field studies suggesting that agro-ecological transitions respond to differentiated territorial logics (Loudiyi S., Cerdan C., 2021; Lamine, 2021). Driven by various factors whose combination in time and space can vary, these agricultural transitions require an updated geostatistical examination of their spatial dimension at a detailed level.

4As a hypothesis, we take up an idea that emerged in the early 2000s during debates on the links between agronomy, geography, and territory: how farms function is partly determined by their insertion into different "systems of interaction with other activities and [the] occupation of space" (Boiffin, 2004). In other words, how do farms’ location influence the forms of agricultural transition? The aim of this article is to characterize the evolutionary dynamics of transitional farming between 2010 and 2020 on a detailed spatial scale in mainland France. We seek to contribute to the study of the spatial dimension of agricultural transitions across diverse French regional systems, including the effects of specific local contexts (e.g., agritourism, peri-urban agriculture). This geography is based on a typology of farms combining modes of agricultural production (conventional or organic) with modes of agricultural marketing (selling into long or short distribution channels) using data from the 2010 (Bermond et al., 2019) and 2020 agricultural censuses.

5The first part sets out the challenges of a geographical approach to agricultural transitions, and recalls the structuring dimensions of farm sustainability. Following a method for defining a statistical field of farms engaged in different forms of transition, the second part presents a geostatistical analysis of eight cantonal trajectories of emergence or resistance to agricultural transitions between 2010 and 2020. Caught between stability and major change, the contrasting and shifting geography thus uncovered gives rise, in the third part of the article, to a discussion of the sectoral and contextual factors conducive to the emergence of one form of transition or another in a given location.

Farming in transition: analytical framework and approach

Transitions in agricultural systems: geographies at the crossroads of rural studies and food studies

6Understanding the geographic logics of agricultural transitions has been a preoccupation of Anglophone rural studies for at least two decades. As early as 2004, Lyson and Guptill in the United States (USA) described the coexistence of commodity agriculture with what they termed "civic agriculture," i.e., small and medium-sized farms managing to survive via the development of local and regional marketing strategies. Drawing on USDA (United States Department of Agriculture) agricultural census data, Lyson and Guptill focused on farms engaged in direct sales to consumers, showing that such farms could coexist with large-scale commodity farms, but that they tended to be over-represented in urban or urban-adjacent counties, and more generally in the Northeast, where they showed the strongest growth (compared with the rest of the USA) between 1992 and 1997. Conversely, commodity agriculture predominated in counties characterized by the presence of agribusiness processors, and more generally in the Midwest and South (Lyson and Guptill, 2004: 379-380). Lyson and Guptill associated sectoral factors with these patterns, noting that commodity agriculture was more strongly represented in counties with a high percentage of tenant operators, where highly capital-intensive farms prevail. They noted that direct sales tended to be statistically associated with market-gardening and orchards. Other, non-sectoral factors included the over-representation of direct sales in counties with higher median family incomes (Id.) The authors concluded by predicting a likely future development of direct sales in (peri-)urban contexts. In the late 2010s, however, Amy Guptill and other co-authors reassessed the development and location of direct-sales farms in the USA (Guptill et al., 2018). While direct sales had tripled in twenty years (from $404m in 1992 to $1,200m in 2012), and were more associated with organic farming than conventional farming, Guptill et al. questioned the role of affluent urban consumers as a determining factor. From the perspective of the 2018 article, the narrative of direct-sales farms located in peri-urban settings seemed over-simplified (Ibid.: 79). While the importance of certain sectoral and non-sectoral factors was confirmed, the authors’ new analyses found these to be positively correlated with market-garden production and high household income. Conversely, the correlation between population growth and the number of direct-sales farms was negative. There was no significant positive correlation between the share of direct agricultural sales or associated farms and the increase in median income. The authors concluded that the “civic agriculture” model was more complex than a simple peri-urban phenomenon reflecting high incomes in the Northeast USA.

8These Anglophone geographies are ultimately more rooted in food studies. They give less attention to organic agriculture (OA) and quality and origin labeling (QOL) and more attention to indicators that are described as part of a movement toward reterritorialization (Rieutort, 2009) in France. The latter is likewise attracting considerable scholarly investigation, mostly qualitative and local in focus (Perrin and Soulard, 2014; Rouget et al., 2016; Duvernoy, 2018; Darrot et al., 2022). It is in this context that typological and cartographic approaches addressing regional or national agricultural dynamics, especially in terms of the product quality regimes of three Mediterranean countries (Spain, Italy, and France; Dedeire et al. & Hirczak et al., 2013), have emerged. These approaches have taken as indicators: (1) the percentage of OA area relative to total agricultural area; and (2) the numbers of Protected Designation of Origin (PDO) or Protected Geographical Indication (PGI) products present at the level of the Unité Territoriale Statistique NUTS 3, corresponding to the French départements. A typology of this kind, crossed with that of agricultural trajectories (changes in structure, types of employment, technical-economic orientations, and productivity), confirms first of all a situation of coexistence, in which the quality system does not replace the dominant, conventional system. The development of quality schemes is, however, correlated "with stabilized, dynamic and diversified agricultures,” and, particularly for OA, with local contexts characterized by agritourism (Hirczak et al., 2013).

9The identification of geographical factors for “local food” on the one hand (food studies) and organic farming on the other seems to be converging. However, it remains to formalize an analysis articulating quality farming systems (around organic farming, as a strong ecological indicator; Hirczak, 2011) and farming systems reconnected to local food networks (direct sales, but also local or regional indirect sales). With the release of the 2020 French agricultural census, it now seems possible to analyze geographical trajectories simultaneously along these two dimensions.

What structural dimensions of transitions are visible in the Census of Agriculture?

10Faced with the necessary transition of agricultural systems, a great deal of research over the past three decades has interrogated the concept of agricultural sustainability (Von Wirén-Lehr, 2001; Zahm, 2019). The scientific literature has schematically identified two main paths toward sustainable agriculture—corresponding to the paradigms of weak or strong sustainability (Hill, 1998; Horling and Marsden, 2011)—based on two different conceptions of the Human-Nature relationship. The first aims for ecological efficiency through technological innovation; the second aims for the development of ecosystem services by restoring the ecological functionalities of agrosystems. While this conceptual dichotomy has heuristic value, it does not allow us to describe the real diversity of types of sustainable agriculture observed in Western societies. Typological approaches, based on empirical observations to establish distinctions based on precise criteria (Van der Ploeg, 1996; Gliessman, 2007, Therond et al., 2017), can complement this dichotomy. In 2018, in the journal Economie rurale, Plumecocq et al. presented a typological synthesis of forms of sustainable agriculture structured around two major dimensions: the biotechnical functioning of production systems (more or less dependent on external inputs vs. ecosystem services); and the level of insertion of the production system within the food system (globalized vs. localized). This typology enables us to think about the relational and geographical proximities between different forms of agriculture, and to approach them through a paradigm of the “coexistence and confrontation of agricultural and food models” (Gasselin et al., 2021). To do so, we understand the call for "quantitative, statistical, and cartographic tools" capable of shedding light on the spatial foundations of the agricultural dynamics of relocalization (Frayssignes et al., 2021) as articulated with the dynamics of agricultural ecologization.

11The question of a geostatistical approach is all the more pressing in light of recent work suggesting that we cannot understand agricultural transition processes without due attention to their socio-spatial anchoring (Truffer et al., 2015; Loudiyi and Cerdan, 2021). The coexistence—meaning the co-presence in a single territory—of different agricultural models, “alternative” and “conventional,” produces "processes of hybridization, but also of critique and controversy that contribute over time to the affirmation and legitimization of more ecological models, and therefore more broadly to processes of ecological transition" (Lamine, 2021).

12Developing a spatialized approach to agricultural transitions requires identifying the available and appropriate databases, indicators, and variables. This study is based on data from the 2010 and 2020 French Censuses of Agriculture at the disaggregated level (the farm level), i.e., non-anonymized thanks to an access authorization from the Comité du Secret Statistique (CSS) via the Centre d'Accès Sécurisé à Distance (CASD). Although it is sometimes criticized as a statistical tool tailored to a productivist vision of agricultural development (Rémy, 1990), the French agricultural census has been updated since 2010 to better reflect some criteria linked to agro-ecological transitions—even if many blind spots remain (input management, biocontrol technologies, precision farming technologies, etc.). Exhaustive and providing information at a very granular geographic scale (the commune), the French Census of Agriculture includes many technical and economic variables for describing farms (land area, revenue, production strategies, etc.). Among the various agricultural databases available (others include the Mutualité Sociale Agricole's database of non-salaried contributors, the Réseau d'Information Comptable Agricole, and the Registre Parcellaire Graphique), only the Census of Agriculture brings together detailed data on the two variables that are emblematic of agricultural transitions: mode of production—organic or conventional—and marketing strategy—selling into short or long distribution channels. We used these two variables to define a statistical field of farms engaged in agro-ecological and agri-food transitions (the latter meaning the reconnection of agriculture and food via local or regional sales; Baysse-Lainé and Guillemin, 2021; Baysse-Lainé, 2021). This is an undeniable simplification of the diversity of forms of agricultural transition, understood as “the gradual transition of agriculture from the industrial period to the plurality of forms of multifunctional agriculture established to provide appropriate responses to society's demands” (Delfosse and Vaudois, 2000). For example, there are forms of agriculture that are not certified as organic but which can nevertheless develop ecosystem services by reducing the use of external inputs: including conservation agriculture, agroforestry, and self-sufficient and low-input livestock systems. Organic farming itself can cover a wide range of practices (Allaire and Bellon, 2014), with sometimes uneven environmental effects. Similarly, the level of integration of farms into food systems cannot be summed up as an opposition between distant or local sales: the two strategies sometimes coexist within the same production system (Chiffoleau, 2019), and “hybrid” food systems combining the advantages of local supply chains with the logistical efficiency of long supply chains are emerging (Chazoule et al., 2015). The relocalization of agricultural economies based on geographical proximity between production and consumption areas is unlikely, on its own, to meet the challenges of a large-scale transition. Nonetheless, the two indicators selected here are recognized as central to the agro-ecological transition. Their presence in certain areas can create a ripple effect of similar situations, and likewise serve as a marker of broader and more diversified forms of transition. We contend that organic farming and local marketing can be catalysts for agricultural transitions (Benoit et al., 2017).

Farms in transition: a diachronic, geostatistical approach based on the agricultural census

14Group CAL: conventional farms marketing exclusively through long distribution channels, representing 71.3% of farms surveyed in 2020, compared with 84.2% in 2010.

15Group CAS: conventional farming operations using mainly short distribution channels, the proportion of which is rising significantly (16.7% of all farms in 2020, compared with 12% in 2010).

16Group OAL: organic farms marketing exclusively through long distribution channels, whose weight increased slightly between 2010 and 2020 (5.6% in 2020 vs. 2% in 2010).

17Group OAS: organic farms using mainly short distribution channels, numbers of which have also grown significantly between 2010 and 2020 in both absolute and relative terms (6.4% of total farms in 2020, compared with 1.6% in 2010).

Characterizing farms in transition

18Farms in the CAS, OAL and OAS groups will be considered here as those delimiting a field of farms in transition. A comparative analysis of the evolution of the four groups over the last intercensal period shows a decline in conventional farms selling into long distribution channels, a clear increase in organic farms and a significant rise in farms selling locally (Table 1).

19This division differentiates farms on a number of socio-demographic variables, including the age pyramid of farmers and their level of general education (Table 1). The proportion of farm operators under the age of 40 increases with organic certification and local sales (30% for the OAS group vs. 16% for the CAL group). The level of general education also remains higher among organic farmers than among conventional farmers, including conventional farmers selling locally (Mahé and Lerbourg, 2012; Leroux, 2013). In 2020, the proportion of higher-education graduates reached 36% among organic farmers selling locally, compared with 12% among conventional farmers selling into distant markets. There is also a slightly higher proportion of women among farm operators in organic farming associated with short distribution channels (Laisney, 2012): 25% versus 22% on average overall.

20Production categories also differentiate farm groups (Table 2). Like their conventional counterparts, organic farmers selling into long distribution channels remain over-represented in arable farming, dairy farming and viticulture. They occupy farms and operate farm businesses that are, on average, larger than those of the conventional farmers in the ACL group: 50% of farms were classified in 2020 as "medium" or "large" according to the agricultural census’ economic size categories; and the share of micro-businesses was reduced to 20% in 2020 within this OAL group (versus 30% within the CAL group). On the other hand, a stronger commitment to short distribution channels is reflected in the implementation of production systems of smaller economic size, especially in organic farming (OAS), with production types more oriented towards market gardening, horticulture, arboriculture, viticulture or mixed farming/livestock. This commitment goes hand in hand with a higher level of employment than on conventional farms: the number of full-time equivalents (FTE) per farm—a ratio traditionally used to measure labor intensity, where 1 is equal to the annual full-time work of one person—is at least 2.5 FTE for farms selling locally, compared with 1.4 FTE for conventional farms selling into commodity markets. Organic farms selling locally (OAS) make more use of labor per unit of farmland (6.1 FTE vs. 1.9 FTE per 100 ha for conventional farms selling into long distribution channels), confirming the idea that farms in transition do more to maintain agricultural employment (Massis and Hild, 2016; Midler et al., 2019).

Spatializing agricultural transition trajectories: materials and methods

21How are these four farm groups distributed across mainland France? How has this geographical distribution changed between 2010 and 2020? Figures 1, 2, 3 and 4 provide some initial answers to these questions, showing the share of each group in the total number of farms at the level of INSEE 2010 pseudo-cantons for 2010 and 2020.

22A comparative reading of the different maps shows that the vast geographical block formed by conventional farms in 2010 had begun to break up by 2020 (Figure 1). The geography of conventional farms selling into local markets (Figure 2) expands preferentially from areas where these farms were already well represented in 2010 (Alpes-Maritimes, Savoie, Cévennes, Hautes Vosges, Val de Loire orléanais, Monts du Lyonnais). Organic farms using long distribution channels (Figure 3), marginal in 2010, had expanded out from their historical areas (Diois, Larzac, Bergeracois) to emerge (western France), or even to assert themselves in areas dominated by conventional agriculture (Châtillonnais, Avalonnais, Pays d'Auch). Organic farms using short distribution channels (Figure 4) are expanding around their historical strongholds (Vercors, Drôme provençale, Cévennes, Conflent, Haut-Jura, Pays de Sault), but are also appearing along the Atlantic and Mediterranean coasts, in rural tourist areas (Perche Ornais, Puisaye-Forterre, Val de Loire tourangeau, Pays Basque), in urban areas (Lyon, Marseille, Toulouse, Bordeaux, Nice, Nantes, Rennes), and on the outskirts of small towns (Laval, Amiens, Saintes, Bergerac, Pau, Bourges, Troyes, Soissons, etc.).

23However, this comparative reading of the maps has two limitations. First, it treats the different farm groups separately, making it sometimes difficult to interpret the geographical combinations. One assumes that agricultural transitions are not carried out everywhere by the same combinations of groups. Second, it is not entirely satisfactory to use the temporal dimension to assess the trajectory between two dates of the same spatial unit described by different variables (in this case, farm groups). This is why we made use of multivariate analysis methods.

24Using the "story analysis" method (Benzecri, 1973), the aim was to track changes in the spatial distribution of the four farm groups between the two editions of the Agricultural Census. The decision was taken to construct a multi-date typology of cantons that summarized the principal trajectories observed in the statistical structure of the four farm groups (Figure 5).

25First, principal component analysis was performed on a table describing the distribution of the four farm groups (columns) by canton (rows) at the two agricultural census dates. This geographic information table had a particular design feature: each canton is present as many times in a row as there are dates to describe its profile (here, twice: 2010 and 2020). This method is all the more relevant as the distribution structure of farms by group is not radically different between the two dates. A new geographic information table was then constructed in which each canton is uniquely described by its factorial coordinates at the two census dates (columns). Next, we classified the cantons on the basis of their factorial coordinates at each date, using the dynamic clustering method. The Euclidean distance is used as the dissimilarity index. The optimization criterion chosen here was the Determinant(W), which eliminates the effects of variable magnitude. The number of iterations was set at 500. Figure 6 shows a graphical representation of the eight types of agricultural development trajectories for cantons between 2010 and 2020 (number of classes defined using the Elbow method—8-class inflection point). The classification results are mapped to track spatial dynamics using an established methodology (Piron, Dureau and Millon, 2006).

Contrasting cantonal trajectories: from resistance to emergence

27In the first sub-group, four types of cantonal trajectory have been identified (Figures 8 and 9). Cluster 1 (1166 cantons, or 43.5% of total farms in mainland France in 2020) includes a large group of cantons where the share of conventional commodity-market farms remains overwhelming, at 85% on average in 2020, with little change between 2010 and 2020. Cluster 2 (591 cantons; 14.1% of farms) illustrates a first variant, identifying a group of cantons where since 2010 there has been a stable cluster of conventional, short-distribution-channel farms (around 20%) in a local context dominated by intensive agriculture. Organic farms are expanding slowly but remain rare, rising from 2.4% to 8% of total farms between 2010 and 2020, for the most part selling into short distribution channels. The trajectory of Cluster 3 cantons (631 cantons; 22% of farms) is marked by the emergence of organic farms, primarily using long-distance marketing channels, in an overall context of conventional agriculture: from 2.8% in 2010, the share of organic farms using long-distribution channels rose to 9.8% in 2020. Local-marketing organic farms also began to appear (6.3% in 2020). Cluster 4 (156 cantons; 4.2% of farms) gathers together cantons where conversion to organic farming continued at a steady pace between 2010 and 2020: from 10% in 2010, the share of organic farms reached an average of 33.5% in 2020. This increase is largely driven by farms marketing non-locally (2 out of 3). Organic farms using short distribution channels expanded from almost non-existent in 2010 to reach the national average (12%), while conventional farms using long distribution channels now represent only one farm in two.

28In the second sub-group, where conventional commodity-market farms are in sharp decline, four types of cantonal trajectory have also been identified (Figures 10 and 11). Cluster 5 (117 cantons; 3.2% of farms) is seeing a sharp rise in organic local-marketing farms (from 11% in 2010 to an average of 32% in 2020), in a context where local-marketing farms were already well established in 2010 (45% of such farms were predominantly conventional). Cluster 6 (375 cantons; 10.3% of farms) illustrates a trajectory of strong growth in short distribution channels from 2010 to 2020, especially among conventional producers (from 16% in 2010 to 29% in 2020), but also among organic producers (3% in 2010 to 16% in 2020). Organic farms selling into long distribution channels remain very marginal (5.8% in 2020 vs. 3.3% in 2010). On average, conventional producers selling into long distribution channels represent less than 50% of all farms. Cluster 7 (260 cantons; 1.4% of farms) is marked by the consolidation of a binary marketing structure among its almost exclusively conventional farms: in 2020, 51% of farms sold into long distribution channels and 41% into short distribution channels (compared to 60% and 38%, respectively, in 2010). The Cluster 8 (106 cantons; 1.1% of farms) trajectory is made up almost entirely of cantons in Corsica or in the area around Nice. Since 2010, almost two-thirds of farms have been selling conventional products through short distribution channels, primarily sheep/goat (17%) or mixed livestock (16%) production. The only change over the intercensal period has been the emergence of small-scale organic farms in short distribution channels (up from 4% in 2010 to 15% in 2020) in in a limited number of cantons. The following section looks at the determinants of these locations.

Weakness or coexistence of French agricultural transitions: geography and contextual effects

29The map of cantonal agricultural trajectories shows different configurations of agri-food transitions or, conversely, areas where agriculture remains predominantly associated with the model inherited from productivist modernization. The results we have presented above suggest sectoral determinations linked to agricultural production systems and/or agri-food chains that have been developed in the past or more recently. However, other, non-sectoral factors also appear to play a decisive role, such as local residential and tourist economies, residential dynamics (heliotropism), and/or territorial dynamics (e.g., the valorization of heritage features). Before discussing how these two sets of factors work in combination, we will consider the role of each in the spatial differentiation of agricultural transition dynamics.

Sectoral factors: production, sectors, terroirs

Table 3: Breakdown of organic farms in mainland France in 2020 by year of conversion to organic, year of farm start-up, and farm operator age at start-up

	Farm start-up before 2010	Farm start-up after 2010	All organic farms
Conversion to organic before 2010	19.8%	5.3%	25.1%
Farmer age at start-up < 40	17.4%	3.7%	21.2%
Farmer age at start-up > 40	2.4%	1.6%	3.9%
Conversion to organic between 2010 and 2020	38.8%	36.1%	74.9%
Farmer age at start-up < 40	35.7%	24.8%	60.5%
Farmer age at start-up > 40	3.2%	11.3%	14.4%
All organic farms	58.6%	41.4%	100%

Note: In 2020, 38.8% of organic farms in mainland France were operated by farmers who began farming before 2010 and converted to organic farming after 2010. 5.3% of farms had already been converted to organic before a new operator arrived (takeover of an existing organic farm).

Source: Agricultural censuses 2010 and 2020 – CASD; authors presentation, 2024

Table 4: Distribution of organic farms in cluster 5 cantons by year of conversion to organic, year of farm start-up, and farm operator age at start-up

	Farm start-up before 2010	Farm start-up after 2010	All organic farms
Conversion to organic before 2010	25.1%	6.7%	31.8%
Farmer age at start-up < 40	22.1%	5.0%	27.1%
Farmer age at start-up > 40	3.1%	1.7%	4.8%
Conversion to organic between 2010 and 2020	25.0%	43.2%	68.2%
Farmer age at start-up < 40	22.8%	28.5%	51.3%
Farmer age at start-up > 40	2.2%	14.7%	16.9%
All organic farms	50.1%	49.9%	100%

Note: In 2020, 43.2% of organic farms in mainland France were operated by farmers who began farming after 2010 and converted to organic farming after 2010. 6.7% of farms had already been converted to organic before a new operator arrived (takeover of an existing organic farm).

Source: Agricultural censuses 2010 and 2020 – CASD; authors presentation, 2024

The spatial dimension of non-sectoral factors: peri-urban location, social morphology, residential and/or tourist economies

33While the geography of the typology of cantons presented above borrows explanatory elements from the sectoral context specific to agriculture, it also suggests that the nature of the socio-territorial (non-agricultural) environment in which farms are embedded influences the orientation and intensity of forms of agricultural transition. Rural areas strongly marked by the legacies of productivism, and that are the most demographically and economically fragile, remain the most resistant to the emergence of transitional forms. Aube, Aisne, Eure-et-Loir, Indre, Marne and Cantal fall almost entirely within Cluster 1. In the case of the Grand Est, low population densities are seen as "the reflection of a particularly high-performance, ultra-modern agro-industrial system" (Monot, 2017). These systems are geared towards exports, as are the grass-based cow-calf systems of the Cantal, which export cattle to southern Europe for fattening (Rieutort et al., 2014). Does this mean that, in areas with limited local markets, farmers and production chains can only be maintained by continuing to focus on exports? Not necessarily, since diversification into short distribution channels has emerged within some productivist contexts, thanks to rapidly expanding residential and tourist economies. Clusters of conventional farms selling into short distribution channels in some areas belonging to Cluster 2 can be interpreted as a response to local markets supported by more affluent populations: for example, in seaside holiday resorts (Le Touquet, Honfleur, Dinard, Auray...), in rural areas popular for retirement migration and second homes (Landes coast, Sologne, Royannais, Castres country...), and even in areas of peri-urbanization around major cities (Paris, Lille, Lyon).

34Similarly, the growing importance of the tourist economy in certain mountain areas covered by Cluster 7 (Northern Alps, Pyrenees) explains the significant development of short distribution channels alongside quality-label regimes (Northern Alps and Pyrenees). A closer look at the map also reveals that the strong development of short distribution channels between 2010 and 2020 in Cluster 6 cantons is not limited to wine-producing or fruit-growing areas; it also characterizes many urban peripheries (Paris, Lille, Strasbourg, Marseille, Bordeaux, Lyon, Nantes, Grenoble, Dijon, Orléans, Tours, Angers, Limoges). There is continuity here with various factors favoring transitions, notably the presence (Bermond et al., 2019) and/or arrival through socio-residential recomposition (Bermond et al., 2024) of more affluent and/or highly educated resident populations seeking supplies of local products. These determinations are not confined to France, as evidenced by the structuring of new forms of distribution for the food transition in Wallonia. Vonthron and Devillet (2023) identify one such form linked to direct agricultural and artisanal sales, which they locate "in the more affluent sectors, where the level of education is highest, but where income inequalities are also most marked."

Conclusion

36The 2020 French agricultural census confirms that organic production and short distribution channels have been gaining ground since 2010. Our spatio-temporal analysis at the pseudo-canton level, however, reveals strong geographical disparities, with eight typical cantonal trajectories identified according to the dynamics of organic farming and short distribution channels. While a third of cantons (Type 1; mostly located in the least dynamic rural areas and those most strongly marked by the legacies of the productivist model) show little change in their farm distribution profile with regard to the two indicators, 38% are following a trajectory marked by profound changes in their profile (Types 3, 4, 5, 6). Another 30% of cantons (Types 2, 7, 8), meanwhile, show profiles that maintain or consolidate transitions already underway in 2010.

37These eight trajectory types reveal a contrasting national geography in which sectoral factors (adaptation of farming systems to pedoclimatic conditions, territorialized production chains, specific agricultural policies, the history of agricultural social movements, etc.) interact with non-sectoral factors linked to the socio-geographical environment (urban proximity and urban growth, tourist areas, rural gentrification, neo-rural migration, standard of living, etc.). In general, local transformations in farming systems are more readily sustained in areas where non-sectoral factors (favorable socio-demographic profile, a dynamic tourist economy, significant road traffic, etc.) are supported by sectoral factors (quality label products, regions popular among back-to-the-landers, vineyard and market-gardening areas), creating synergistic effects.

38The main limitation of this work is that it restricts the identification of farms “in transition” to the two indicators of organic farming and short distribution channels. Future work will seek to further subdivide the group of farms practicing conventional agriculture and selling into long distribution channels so as to identify other potential signs of transition visible within the agricultural census data (e.g., farms engaged in intra-parcel agroforestry, precision agriculture, soil conservation agriculture, grassland livestock systems, diversified systems integrating mixed crops and livestock, low-input systems, etc.). In any case, the initial results obtained here argue in favor of paying closer attention to socio-territorial contexts in the crafting of public policies to support agricultural and food transitions, since geographical disparities can further accentuate inequalities. The cross-referencing of this typology with data from the French general population census constitutes another prospect for future research.