Unravelling the Paradoxical Seasonal Food Scarcity in a Peasant Microregion of Mexico

Abstract

Seasonal food scarcity during pre-harvest months is, widely, considered to be the principal manifestation of food insecurity, for some 600 million members of smallholder families, who rely on a variety of coping strategies. This paper analyses both the peasant-economy variables that explain the presence and intensity of seasonal food scarcity, and the coping strategies of 120 rural households in a microregion of southern Mexico. We, also, examine how supply networks for six archetypical foods of the peasant diet express robustness or vulnerability during seasons of abundance and scarcity. The method combines surveys, ethnographic fieldwork, statistical models and social network analyses. Results show that 74% of households experience at least one month of food scarcity annually, and 34% of shortages last more than six months. In total, 29% of affected households gather wild foods, and 14% use intense coping strategies, such as international migration, taking out rural loans, and parental food buffering. During scarce seasons, self-sufficiency networks for maize and beans contract, but still maintain the food supply of peasant households, while cash-consumption networks such as those of beef become accessible only to a small sector of economically differentiated households. In contrast to the vast majority of research, which simply reports the presence of seasonal food shortages and describes the coping strategies of rural households, this paper provides an in-depth analysis—based upon a novel methodological integration—of the socioeconomic, agrifood, and land tenure conditions that may determine why many peasant territories in the Global South face the “farmer–food-scarcity paradox”.

1. Introduction

The vast majority of impoverished people inhabit rural areas, and depend, mostly, upon subsistence agriculture for survival. During pre-harvest months, when the previous year’s harvest has been, largely or fully, exhausted, rural inhabitants become particularly vulnerable to seasonal food scarcity [1,2], which is one of the most common forms of food insecurity for an estimated 600 million people [3]. Many of the affected are members of peasant households (PHs), which are understood as family farms, generally smaller than five hectares, mostly dedicated to food production for self-sufficiency and highly dependent on the labour of their members [4]. Seasonal food scarcity among PHs has been studied extensively in Sub-Saharan Africa and Southeast Asia, using predominantly economic or nutritional approaches, combined with analyses of national surveys [5,6]. However, knowledge gaps exist—particularly from an agrifood perspective—regarding seasonal food scarcity in other peasant regions around the world, including Latin America [7,8].

The subjects treated in the literature on seasonal food scarcity fall into four major categories: associated factors, multidimensional effects, PH coping strategies and gaps in specific public policies [1,9,10]. Research on associated factors, in addition to studying the increasing vulnerability to climate change, has addressed impoverished soils, technological gaps in rain-fed peasant agriculture, periodic increases in regional market prices during pre-harvest seasons, inadequate infrastructure for community food storage, a lack of availability of loans, and accumulation of high levels of debt. Other associated factors include a lack of negotiating power by peasant families and organizations, in establishing market prices for their cash crops, diseases that undermine the family labour force and global financial crises that affect international food prices. Consequently, there are rural micro-economies, in which people often spend, approximately, three-quarters of their income on food [11,12,13,14].

Seasonal food scarcity has many dire social and economic consequences for PHs because it is harmful to public health, local economies, production capacity, and overall human development. During the scarce season, some PH members are forced to reduce the diversity and quantity of food they consume, thereby suffering macronutrient and micronutrient deficiencies, high levels of anxiety and stress, and susceptibility to serious illnesses [15,16]. These effects of food scarcity are exacerbated by the need to migrate to other locations to find wage work, when demand for agricultural labour is high, thus reducing the following season’s production [8]. In addition, PHs often take out high-interest loans, or mortgage their land to pay off their debts [16]. Solidarity among PHs may decrease during periods of scarcity because most families are equally affected by seasonal hardships [17,18]. Children and youths are, often, forced to leave school, either to participate in farming the family’s holdings, or to migrate to generate income for the household. Peasant psychology is, undoubtedly, influenced by the stigmatization of scarcity and the poverty that results from the confluence and cyclical nature of these factors [19,20].

Peasant households cope with scarcity via a range of direct and indirect food-based strategies. The most common direct strategies described in the literature consist of limiting food portions, eating fewer preferred foods, parental buffering (i.e., adults refrain from eating for several days in order to feed children), hunting and gathering wild foods, and diversifying or temporarily alternating crops [21,22]. Indirect strategies include migration, taking out loans, increasing rural wages, reciprocity among PHs (generally within extended families) or dyadic contracts, raising livestock that can be sold during times of scarcity, and, even, community aid [1].

Meanwhile, many actors with a variety of agendas—including government agencies, international organizations, academics, activists, and community-based organizations—attempt to alleviate rural seasonal food scarcity through a wide range of interventions. Some interventions aim to increase agricultural production, mostly by providing access to irrigation systems, seed and fertiliser, community infrastructure for grain storage, and the training and technical support that are needed to increase yields, by either conventional or agroecological means [8,23]. Other interventions emphasise economic and social protection via seasonal employment programs, pensions for the elderly, guaranteed prices, food pantries, direct cash transfers to women for purchasing food, and school breakfast programs [1,12,23,24]. However, it is extremely difficult for such support systems to fully counteract seasonal food scarcity because many PHs are stuck in structural agrifood insecurity traps [2,25,26].

Although the literature on seasonal food scarcity in indigenous and peasant regions is extensive, we identify two important gaps. One is the need for analyses—on a microregional level—of PHs’ coping strategies, plus the roles of land tenancy, socioeconomic conditions, public programs, and other external agendas of territorial intervention. A second gap is the need for a microregional understanding of the functioning of peasant food-supply networks, including the roles of certain key foods during periods of seasonal scarcity.

This paper addresses both gaps by studying an indigenous peasant microregion in the Sierra Madre of Chiapas. To a considerable degree, agrifood conditions in this microregion are representative of those in Mexico’s migrant rural territories: one finds both a high level of dependence upon government subsidies and welfare programs and interventions by external actors, who try to impose their respective local-development agendas. Our study involved PHs located in rural communities (ejidos) in the upper and lower parts of an agricultural valley that forms part of a MAB-UNESCO biosphere reserve. We expected to find differences in scarcity conditions between the upper part (the core zone of the reserve) and the lower one (buffer zone) because environmental organizations have been implementing green-economy type agroforestry projects in the upper part for almost two decades. In contrast, ejidos in the lower part focus on agriculture.

Our three research questions were: (i) According to the PHs, how many months of food scarcity do they experience per year; during what time span; and what types of coping strategies do the PHs deploy? (ii) What land-tenancy and socioeconomic conditions—both endogenous and exogenous—may be determining seasonal food scarcity and the severity of coping? and (iii) How do agrifood supply networks perform during scarce seasons, as compared to during abundant seasons? Particularly with regard to the second question, we propose the following hypotheses:

A.

Greater production of maize and beans (which are the bases of the peasant diet) reduces the duration of scarcity and the intensity of coping strategies.

B.

Households with more-precarious land tenancy face longer periods of scarcity, and adopt the most intense coping strategies.

C.

International migration—although not national or regional migration—significantly reduces the duration of food scarcity and the intensity of coping strategies.

D.

Means of production are positively correlated to government support, which determines fewer months of food scarcity, as well as less-intense coping strategies.

E.

Significant differences exist in the duration of food scarcity and intensity of coping strategies between the communities that focus on agroforestry versus those that focus on agriculture.

2. Methods

2.1. Study Area and Research Sample

The upper watershed of the El Tablon River (Cuenca Alta del Río El Tablón, or CART) is a mountainous neotropical area of approximately 24,000 ha, in the northwestern portion of the Sierra Madre de Chiapas, which lies in southeastern Mexico (Figure 1). The CART’s climate varies sharply with altitude, which ranges from 800 to 2550 masl. Some of Mexico’s richest biodiversity is found in the CART’s six types of forest, which grow within a dense network of permanent and seasonal drainages [25,27]. The CART has a dynamic, conflictual socioenvironmental history. During the past 70 years, settlements in the microregion transitioned from predominantly forested private properties and large-scale extensive cattle farms—or fincas—to a form of collective, communal peasant land tenure, known as the ejido [28]. The CART participated in the nation´s agricultural boom of the 1970s and 1980s, then suffered during the agrifood crash of the 1990s that followed Mexico’s entry into the North American Free Trade Agreement [29]. During that same decade, the CART underwent a socio-productive reorganization that prioritised livestock raising (financed by small rural banks) over small-scale agriculture [30]. In 2004, the La Sepultura Biosphere Reserve (REBISE), which had been developed by federal decree, was incorporated into the UNESCO MAB Program. Since then, numerous domestic and international NGOs have sought to promote “green economy” agroforestry projects in the CART [31], involving cultivation of shaded coffee varieties, extraction of Pinus oocarpa resin, and sustainable cultivation of the Chamaedora quezalteca palm [32,33,34].

The CART is currently the most populated portion of REBISE’s buffer zones, with approximately 6000 inhabitants, spanning four generations [35], and comprising approximately 1500 PHs in 12 ejidos. Approximately 30% of the households belong to members of indigenous groups: mainly Tzotziles, Tseltales, and Zoques (all three are Mayans). The other householders are either mestizos from Chiapas or migrants from other states. The domestic group (DG) is the basic economic entity of social reproduction, although several peasants´ organizations also exist. PHs generally cultivate maize and beans for family subsistence, along with small- and medium-scale cattle ranching and/or agroforestry, depending upon the PH’s family composition, means of production, and available land [17,36,37]. Other significant income sources for PHs include federal (Mexican) subsidies and remittances from the United States [38].

For the present study, based on sampling capacity, we selected a convenience sample of 120 PHs in six of the CART’s 12 ejidos. The sampling strategy aimed to capture contrasts in the CART, and to be as representative as possible of the microregion’s mosaic of peasant economies. We divided the sub-basin of the CART into the upper section, or headwaters, and the lower section, or central valleys. The headwaters region is located near the core area of the REBISE, where agroforestry is practiced, according to a forest-transition policy. The central valleys are located in the REBISE general-use area, where PHs cultivate and raise livestock. Using the snowball method, we selected 20 households in three ejidos of the headwaters area, and in three ejidos from the central valleys. The selections were based upon economic profiles. In the headwaters region, we selected 20 households that specialised in coffee production; 20 from a community specializing in pine-resin extraction; and 20 from a community specializing in palm extraction. We applied the same method to a community that, predominantly, raises cattle; to a crop-raising community; and to a community in the central valleys, in which most of the population works as farm labour.

2.2. Ethnographic Fieldwork

From 2017 to 2019, we conducted extended and short-period ethnographic fieldwork in the CART. Through systematic, non-intrusive immersion in ejido communitarian life and PH domestic life, we gathered information related to the interconnections between the peasant economy and seasonal food scarcity. Our ethnographic approach may be classified in terms of microregional spatiality and second-order scope; that is, we were interested in observing specific aspects of our study topic, without carrying out a complete ethnographic study of community life as a whole [39,40]. Fieldwork consisted of guided tours in agriculture plots, visits to food-supply points, and informal conversations with key community and commercial actors in the microregion. Ethnographic observations were recorded systematically in a field diary. Some conversations were audio-recorded and later transcribed for analysis. Throughout this article, when making ethnographic observations and capturing central narratives, we use an emic approach to recover the voices of the peasants.

2.3. Statistical Analyses

Following the initial stage of ethnographic fieldwork and informal interviews, we designed a structured survey to be applied to each PH. The survey contained two sections. The first addressed the DG’s peasant economy [41,42], while the second focused on the origin, quantity, seasonality, and cost of supplying different foods in the microregion. Both sections referred to the 2017–2018 growing season. We tested the predicted associations between food scarcity and the 15 primary peasant-economy variables addressed in the survey. The testing methods depended upon the type of variable (

Table 1. Description of peasant-economy variables used in statistical analyses of food scarcity and coping strategies (n = 120).

Variable	Description	Type	Coding or Mean (% or Range)
Type of ejido membership	1. Ejidatario (landholders with voice and vote), 2. Poblador (settlers with voice but no vote), or 3. Avecindado (neighbours with neither voice nor vote)	Categorical	1 (45%), 2 (21%), 3 (34%)
Stage of household development cycle	1. Expansion (newly constituted), 2. dispersal (with children already leaving), or 3. replacement (old)	Categorical	1 (46%), 2 (26%), 3 (28%)
Household members	Total number of members in the household	Continuous	4.4 (2–11)
Agricultural labour force	Household members whose principal occupation is agriculture	Continuous	1.5 (0–4)
Economic activities	Number of agricultural and non-agricultural practiced by the household	Continuous	2.8 (1–6)
Type of migration	1. No migration, 2. regional, 3. national, or 4. international	Categorical	1 (68%), 2 (7%), 3 (8%), 4 (17%)
Total labour-days worked	Days worked by all household members	Continuous	46.2 (0–300)
Total labour-days paid	Days paid for all economic activities	Continuous	121.2 (0–1910)
Total agricultural hectares	Hectares used for agriculture	Continuous	18.1 (0.5–139)
Maize production	Tons of total maize yield (for self-supply, animal feed or sale)	Continuous	4654.6 (0–30,000)
Bean production	Tons of total bean yield (for self-supply or sale)	Continuous	285.9 (0–4000)
Months of food scarcity	Number of months the household reported shortages	Continuous	2.9 (0–12)
Eating wild foods	1. Whether or 2. not the household consumed wild foods	Categorical	1 (57%), 2 (43%)
Government assistance	Amount of government assistance the household received	Continuous	16,045 (0–62,000)
Income	Total income of the household	Continuous	56,806 (0–300,000)
Livestock	Total head of cattle	Continuous	6.8 (0–80)
Location in the watershed	1. Upper (agroforestry in the headwaters) or 2. lower (agropecuary in the central valleys)	Categorical	1 (50%), 2 (50%)
Cluster identity	1 and 2 (social reproduction in accumulation), 3 and 4 (social reproduction squeeze)	Categorical	1 (4%), 2 (32%), 3 (43%), 4 (21%)

). When both variables were continuous, we used Spearman’s correlation coefficients. Due to the presence of ties, we report approximate p-values. When both variables were categorical, we used a chi-squared test for contingency tables. Since many of the predicted values in tables were <5, p-values were calculated based upon a distribution that was approximated via Monte Carlo resampling, using a chi-squared test from the R package “coin” [43]. We calculated standardised residuals for the contingency table and interpreted a residual value >2 in a given cell as supporting the rejection of the null hypothesis, and, thus, as indicating an association between the two factors whose intersection defines the cell [44].

We tested the predicted differences in the number of months of food scarcity across different levels of categorical variables (e.g., “type of ejido membership”) via analyses of variance (ANOVA) or (when the normality assumption was not met) Kruskal–Wallis (KW) tests. When ANOVA was used, we carried out Tukey post-hoc tests to compare across levels. When KW was applied, we implemented post-hoc multiple comparisons using the R function “npmc”. When the categorical variable had only two levels, we used t-tests for comparisons. All statistical analyses were carried out in R v. 3.5.3 [45].

2.4. Social Network Analysis and Visualizations

Social Network Analysis (SNA) is a reticular methodological formalism that allows for establishing analytical relationships between two primary classes of entities (ties and edges), with diacritical indicators for type, weight, and directionality. SNA allows researchers to examine practically any object of study in these terms [46,47]. In agrifood research, studies tend to focus on “agrifood networks” of production, transformation, distribution, marketing, and consumption, which are rarely formalised through SNA, despite a myriad of academic discussions.

The present study formalises an analysis and visualization of supply networks of six foods (corn, beans, sugar, tomatoes, beef, and fish) that are important to the microregional diet, and in which our ethnography work had found a variety of supply logics. We expected that this data could point to the overarching patterns and differentiated logics of agrifood supply in the microregion. Data used to construct the networks were collected from the second section of the above-mentioned structured survey regarding the peasant economy. To visualise the networks, we employed the hive-plot method (combed networks: http://www.hiveplot.com/, accessed on 28 March 2022), which allows for simple graphic interpretations of the overarching patterns of networks, by assigning nodes to radially distributed linear axes [48]. To reveal structural features in terms of robustness and vulnerability, we used parallel coordinate visualization methods to reveal how metric dimensions of food networks changed between abundant and scarce seasons. (For the metrics used and their interpretation in the context of the study, see

Table 2. Description of the metrics used in the social network analysis.

Metric	Level	Description	Interpretation
Nodes	Global	Total number of basic data structure units	Shows changes in size, cohesion and dissociation of the supply by food type between abundant and scarce seasons
Edges	Global	Total number of links
Density	Global	Total number of existing links, divided by total links possible in the network
Average degree	Global	Measurement of the number of edges versus the number of nodes
Average in degree centrality	Global	Average connectivity per node
Global efficiency	Global	Efficiency of a node pair in a graph is the multiplicative inverse of the shortest path distance between nodes. Average global efficiency of a graph is the average efficiency of all node pairs	Enables observation of changes in the food-supply flow between abundant and scarce seasons
Average page rank	Global	A centrality metric that results from an algorithm, which measures the network ranking on a scale of 1–10, based on the weight, the quantity, and the quality of links	Enables the observation of changes in the robustness of the supply by food type, between seasons
Average node connectivity	Global	The minimum number of nodes that must be removed to break all paths from source to target	Enables observation of supply vulnerability by food type between seasons
Connected components	Global	Number of subgraphs in which any two vertices are connected to each other by paths and which are connected to no additional vertices in the supergraph	Enables observation of changes in food supply groupings between seasons
Average out-degree centrality	Global	The out-degree centrality for a node is the fraction of nodes that its outgoing edges are connected to	Captures the centrality of a provider in the supply for all households
Local efficiency	Local	The local efficiency of a node in the graph is the average global efficiency of the subgraph, induced by the neighbours of the node. The average local efficiency is the average of the local efficiencies of each node	Allows observation of changes in the food supply flows at node level, between abundant and scarce seasons
Average node strength	Local	The sum of link weights connected to all nodes	Reports changes in the food flows that the nodes hold within the network
Self-loops	Local	Nodes connected to themselves	Shows directly the importance of self-sufficiency, for each food and season

). In parallel coordinates, each network is represented as a segmented line with vertices on parallel axes displayed vertically. Horizontal lines suggest a positive correlation between the two contiguous dimensions, while overlapping/crossed lines in the form of an X suggest a negative correlation. In order to show/indicate variables of drastically different magnitudes from side to side, it is necessary to rescale them by converting their values to decimals from 0 to 1, while also conserving their proportions.

Agrifood networks were given an identity as ego-networks, based upon a previous cluster analysis, which we used in this study to colour each node (representing a PH) of the network, according to the socioeconomic subgroup to which the PH belonged. From a previous K-means cluster analysis [14], we concluded that four types of PHs existed within the 120 that we sampled: Cluster 1 (comprising 4% of PHs) includes those that have accumulated the highest levels of means of production; Cluster 2 (32%) includes those with an intermediate level of means of production and labour force; Cluster 3 (the poorest 43%) includes those with few means of production and little labour force; and Cluster 4 (21%) includes those with marginal means of production but with several members who carry out agricultural labour. The visualizations of networks allowed for the synthetic exercise of simply recognizing differential supply patterns and logic, for example, evident centralities; price increases among suppliers; expansion and contraction of agrifood networks between abundant and scarce seasons; and endogenous and exogenous supply patterns.

3. Results

3.1. Temporality and Acuity in Seasonal Food Scarcity and Coping Strategies

PHs defined seasonal food scarcity as those lean months, in which they have neither a sufficient quantity and/or diversity of self-supplied food nor the means to purchase food. PHs in the CART report that seasonal food scarcity occurs during the growth phase of the agricultural cycle—i.e., from March to October (Figure 2). PHs that experience several months of seasonal scarcity are forced to implement one or more of the following coping strategies, as classified in order of intensity: Strategy S1—reducing the amount or diversity of food consumed (23% of PHs); Strategy S2—gathering wild foods (29%); Strategy S3—sale of backyard animals or pursuit of agricultural wage labour (8%); Strategy S4—temporary migration or high-interest-rate loans (9%); and Strategy S%—parental buffering (5%) (see Appendix A, for detailed descriptions of the strategies). The remaining 26% of PHs did not report seasonal food scarcity (Strategy S0).

In the agrifood section of the structured survey, we asked PHs what they considered to be the best strategies for countering seasonal food scarcity. As shown in Figure 3, 89% of PHs recommended economic strategies (e.g., creation of steady jobs, maintaining savings, and encouraging solidarity loans), while only 11% recommended agriculture-related strategies.

3.2. Land Tenancy and Socioeconomic Determinants of Seasonal Food Scarcity and Coping Strategies

Regarding our Hypothesis A (higher production of maize and beans reduces the duration of scarcity and the intensity of coping strategies),

Table 3. Spearman correlations between continuous peasant-economy variables, related to Hypothesis A (i.e., that higher production of maize and beans reduces the duration of scarcity and the intensity of coping strategies).

Variables	Seasonal Food-Scarcity Months	Livestock	Government Assistance
Total maize production (kg)	−0.38 ***
Maize production for self-sufficiency	−0.12 ns
Proportion of maize production for self-sufficiency	0.56 ***
Maize production for sale	−0.01 ns
Proportion of maize production for sale	−0.01 ns
Maize production for livestock	−0.59 ***
Proportion of maize production for livestock	−0.68 ***
Total bean production	−0.10 ns
Bean production for self-sufficiency	0.03 ns
Proportion of bean production for self-sufficiency	0.38 ***
Bean production for sale	−0.20 *
Proportion of bean production for sale	−0.31 ***
Total agricultural hectares	−0.64 ***	0.84 ***	0.53 ***
Livestock	−0.65 ***	-	0.51 ***
Government assistance	−0.39 ***	0.51 ***	-
Economic activities	−0.10 ns
Number of household members	0.15 ns
Agricultural labour force	0.20 *

*** p < 0.005, * p < 0.05, ns: no significant.

and Figure 4 show that a greater total maize production shortens the duration of seasonal scarcity significantly. In contrast, the correlation between total bean production and duration of scarcity is not significant. The PHs’ purposes for producing maize were also important: the proportion of maize cultivated for self-sufficiency was positively associated with the number of months of food scarcity, whereas the opposite was observed for maize that was intended for livestock (

Table 4. Frequency of coping strategies carried out across the different variables outlined in the hypotheses. In parentheses are the standardised residuals (differences between the observed frequencies and the frequencies expected under the null hypothesis). Standardised residuals >2 (in boldface) indicate a lack of fit of the null hypothesis for that cell.

	Coping Strategies
Type of ejido membership	1	2	3	4	5
Avecindado	4 (−3.70)	14 (−0.06)	8 (3.04)	6 (0.93)	5 (2.15)
Poblador	7 (−0.09)	12 (1.82)	0 (−1.81)	2 (−0.53)	1 (0.47)
Ejidatario	18 (3.94)	8 (−1.60)	1 (−1.51)	3 (−0.48)	0 (−1.81)
Type of migration	1	2	3	4	5
International	2 (0.76)	0 (−1.61)	0 (−0.69)	2 (2.34)	0 (−0.55)
National	3 (−0.19)	0 (2.64)	0 (−1.13)	7 (5.88)	0 (−0.90)
No migration	20 (−0.96)	34 (4.25)	8 (1.00)	0 (−6.18)	5 (0.47)
Regional	4(1.10)	0 (−2.33)	1 (0.23)	2 (1.14)	1 (0.68)
Location in the watershed	1	2	3	4	5
Upper	24 (3.51)	16 (−1.36)	2 (−2.17)	5 (−0.77)	3 (−0.32)
Lower	5 (−3.51)	18 (1.36)	7 (2.17)	6 (0.77)	3 (0.32)

). Similarly, the proportion of beans cultivated for self-sufficiency was positively associated with the duration of seasonal food scarcity, whereas the proportion of beans for sale was associated with shorter periods of scarcity.

Hypothesis B is that households with fewer land-tenancy rights face longer periods of scarcity, so they adopt the more-intense coping strategies. More specifically, we predicted that the duration and intensity increase in the order of ejidatario < poblador < avecindado. Figure 5 and Table 4 support this hypothesis (KW = 27.79, 2 d.f., p < 0.001). Non-parametric multiple comparisons indicated that avecindados and pobladores experience more months of food scarcity than ejidatarios (p < 0.01), without significant differences between avecindados and pobladores (p = 0.16). Coping strategies are significantly associated with the type of ejido membership, as indicated by the chi-squared test (chi²: 28.79, 8 d.f., p < 0.001). For example, standardised residuals in Table 4 show that S1 (reducing the diversity and/or quantity of food consumed) was more frequent than expected by chance among ejidatarios, and less frequent than expected among avecindados. By contrast, parental buffering (S5) was higher than expected by chance among avecindados.

Contrary to Hypothesis C (i.e., that international migration—but not national or regional migration—significantly reduces the duration of food scarcity and the intensity of coping strategies), the duration of seasonal food scarcity did not vary across types of migration (KW = 6.14, 3 d.f., p = 0.11, Figure 5). However, coping strategies were, indeed, significantly associated with type of migration (chi²: 56.31, 12 d.f., p < 0.001). Standardised residuals (Table 4) show that a high level of reliance on wild foods (S2) was more frequent than expected by chance among PHs that either did not report migration or that reported regional or national migration. In contrast, a high level of dependence on wild foods was less frequent than expected among PHs with international migration. Regarding Hypothesis D (i.e., that the means of production are positively related to governmental support, which determines the absence of food scarcity as well as less-intense coping strategies), Table 3 and Figure 6 show a positive correlation among cattle-raising, total agricultural hectares possessed, and receiving monetary transfers from the government. These same variables are negatively correlated with the number of food-scarcity months. Total agricultural hectares varied across coping strategies (KW = 27.05, 4 d.f., p < 0.001), with S1 (reduction in food diversity and/or amount of consumption) being significantly more frequent than the Strategies S2 (high dependence upon wild food), S4 (temporary migration), and S5 (parental buffering; p < 0.05, Figure 6). Government assistance, too, varied across coping strategies (F_4,84 = 3.97, p < 0.01), with S5 differing significantly from S1, S2, and S4 (p < 0.05, Figure 6).

Hypothesis E (that agroforestry and agricultural communities differ in their durations of food scarcity and intensity of coping strategies) is both supported and not supported by our results. Contrary to our hypothesis, the number months of food scarcity did not vary between high-watershed (i.e., agroforestry) locations and low-watershed (agricultural) locations (t = −0.36, 104.9 d.f., p = 0.72, Figure 5). In contrast, coping strategies were significantly correlated with location, as indicated by the chi-squared test (chi² = 14.29, 4 d.f., p < 0.005). Standardised residuals in

Table 5. Wild foods gathered by peasant households during scarce seasons.

Food Type	Local Name	Scientific Name	Gathering Unit
Herbs and vegetables	Hierba mora	Solanum nigrum	Polycrop
	Chipilín	Crotalaria longirostrata	Polycrop
	Orejita	Porophylum ruderale	Polycrop
	Bledo	Amaranthus blitoides	Polycrop
	Canutillo	Hamelia patens	Polycrop
	Hierba santa	Piper auritum	Bush
	Berro	Nasturtium officinale	Polycrop
	Chaya	Cnidoscolus aconitifolius	Bush
	Verdolaga	Portulaca oleracea	Bush
	Nabito	Brassica napus	Polycrop
	Ejote de monte	Phaseolus vulgaris	Bush
	Cilantro de monte	Peperomia peltilimba	Bush
Inflorescences, fruits, meristems, and roots	Flor y puntas de Calabaza	Cucurbita lundelliana	Polycrop
	Puntas de chayote	Sechium chinantlense	Bush
	Pacaya	Chamaedorea tepejilote	Bush
	Anona	Annona reticulata	Bush
	Cuscuta	Amphipterygium adstringens	Bush
	Tomatillo de monte	Lycopersico esculentum	Bush
Tubercles	Madre maíz	Mirabilis sp.	Bush
Stalks	Nopal espinero	Opuntia	Bush
	Nopal masetero	Opuntia	Home garden
Mushrooms	Hongo blanco	Tremelloscypha gelatinosa	Bush
	Hongo orejita de cochi	Hypomyces lactifluorum	Bush
Insects	Hormiga chicatana	Atta mexicana	Bush
	Gusano zats	Arsenura armida	Bush
	Chapulín	Sphenarium purpurascens	Polycrop
Reptiles	Iguana verde	Iguana iguana	Bush
	Iguana negra	Ctenosaura pectinata	Bush
	Turipache	Corytophanes percarinatus	Bush
Fish	Macabí	Albula vulpes	River
	Mojarra de río	Diplodus vulgaris	River
	Charal de río	Chirostoma sp.	River
	Carpa de río	Cyprinus carpio	River
Crustaceans	Cangrejos de río	Astacoidea and Parastacoidea	River
Snails	Caracoles chuti	Pachychilidae	River

show that S1 (reduction in food diversity and/or amount of consumption) was more frequent than chance among PHs of the upper watershed, but less frequent among PHs of the lower watershed. The opposite trend was observed for S3 (sale of backyard animals and/or pursuit of agricultural wage labour).

3.3. Peasant Agrifood Supply Networks during Scarce and Abundant Seasons

Ethnographic fieldwork found that the CART’s foodscape (Figure 7) consists of approximately 150 food products, 45 of which are highly industrialised packaged foods, 35 are gathered wild foods (Table 5), 33 are fresh fruits and vegetables, 7 are cultivated staple crops (i.e., maize, beans, fava beans), 6 are meats (i.e., beef, pork, chicken), and 5 are fish and seafood products, plus there are other dairy products (i.e., salt, sugar, oil). As is the case for the vast majority of peasant and indigenous communities in Mesoamerica, maize and beans are the cornerstones of rural food culture. These, and to a lesser degree squash, are produced for family subsistence, with little sale among PHs. Wild foods are, also, collected on a household level. The remaining products are purchased (variously) from a government-run retail store (Diconsa), local grocery stores, itinerant salespeople, or regional or nation-wide chain stores in a nearby city.

To answer the third research question (“How do agrifood supply networks perform during scarce seasons, as compared to abundant seasons?”), we selected six foods: maize, beans, sugar, tomatoes, fish, and beef. This selection was based upon PHs’ reports about variations in the sources of supply and upon our previous observation that such networks had archetypical patterns of supply, which would help us to understand the functioning of the regional agrifood system. Data for the social-network analysis and visualization were obtained via structured questions (in the survey that was given to PHs), regarding the origin, volume, and cost of each of the six foods during both seasons. The hive plots in Figure 8 capture the total supply flow of these foods for an entire year (network weight). The plots represent oriented bipartite networks, in which the A axis of the nodes represents the PHs, identified by colour according to the socioeconomic cluster to which they belong. Green nodes belong to cluster 1; orange nodes to cluster 2; pink nodes to cluster 3; and blue nodes to cluster 4. The C axis of the nodes, which represents food providers in the microregion and wider regions, shows an internal–external gradient from the vortex axis outward. A third axis, B (which is a mirror of the A axis), was required to represent food self-sufficiency and the exchange or sale of food among households.

In the case of corn and beans, the pattern of self-supply is evident, as is the internal trade network of beef: a single PH manages the microregion’s only butcher shop. Food-supply networks fall into two categories, according to their degree of centrality: (1) star networks, which are exogenous for fish, tomatoes, and sugar, but mixed endogenous–exogenous for beef; and (2) endogenous disconnected or self-looping networks (for maize and beans).

The parallel coordinate displays (spectral signatures) in Figure 9 variations in the supply-network metrics (described in Table 2) between periods of abundance and scarcity for maize, beef, and sugar. The inter-seasonal stabilities of these networks’ respective structures differed substantially. The spectral signatures of the sugar network and tomato network indicate that the structures and functions of these networks do not change diachronically. This result is reasonable because both foods are basic ingredients in the peasant dietary culture, and are consumed continually throughout the year. In contrast, maize and beans are self-supplied foods that are rationed across abundant and scarce seasons, with lower consumption during the latter. Furthermore, it is difficult for households to purchase these foods—not only because households lack money, but because of the social stigma that respondents mentioned during our ethnography work. According to PHs, the ability to produce these crops is what defines someone as a peasant. The case of beef and fish is quite different. Since these foods are expensive, only higher-income PHs (mainly clusters 1 and 2) can purchase them as a strategy for coping with scarcity. As illustrated in Figure 9, the sugar network maintains its full structure, and the maize network’s average node strength decreases only during the scarce season. The beef network´s nodes and edges decrease, while the following features increase: density, local efficiency, average node connectivity, average page rank, average in-degree centrality, and average out-degree centrality. Therefore, the maize network shows changes only in the food flow of the nodes within the network. In contrast, the beef network becomes more cohesive and more efficient both locally and globally, although the beef network increases its centrality and excludes a large number of elements from the network.

4. Discussion

4.1. Scarcity and Coping-Strategies in the CART, as Compared to Other Peasant Contexts in Central America

Our findings show that 74% of the 120 PHs that we studied experience at least one month of seasonal food scarcity each year. For 34% of the PHs, this scarcity lasts six months or more. These results indicate an intermediate level of seasonal scarcity, as compared to the findings of the few existing similar studies in other Central American indigenous and peasant regions. For example, [6] reports that 97% of peasant coffee households studied in El Salvador experience “lean months”; [49] documented lean months for 69% of coffee-growing households in Nicaragua; and [50] found that 36% of agricultural households in Honduras experience food scarcity. It is important to note that in both the Nicaraguan and Honduran case studies, farmers harvest two cycles of maize and beans annually, which, undoubtedly, explains the lower rates of food shortage, as compared to our study and that of El Salvador. Therefore, the most valid approach is to compare our findings with the latter case.

The coping strategies that we documented, and those reported for 29 coffee farmers in El Salvador, show certain similarities. Strategies found in both regions include eating less, modifying the diet, selling livestock, and borrowing money. However, numerous strategies documented in El Salvador (such as borrowing food and using earnings from coffee cultivation to meet food supply requirements) were not reported by the PHs in our study. Conversely, we documented strategies that were not reported in El Salvador, such as parental buffering, temporary migration, consumption of wild foods, and seeking paid agricultural work. The three main strategies used by households in the El Salvador case were borrowing food (25%), eating less (25%), and eating different foods based upon availability and price (25%). By contrast, the primary strategies in our study included a high level of consumption of wild foods (29%), reduction in diversity and/or quantity of food consumed (23%), and temporary migration or taking out rural loans (9%).

A brief comparison of these two peasant regions in Central America immediately reveals some contrasts in seasonal scarcity and coping strategies due to the particularities of the peasant economy of each territory. In the Tacuba region of El Salvador, borrowing food or money from family members is pervasive among PHs, and so is the remarkable seasonality in savings from coffee sales. Meanwhile, in the CART, loans from family members are rare because they come with very high interest rates, and in some cases involve mortgaging means of production, such as agricultural plots [16]. Two other differential strategies reported in our case study are migration to the United States, and seeking paid agricultural work within the microregion.

It is striking that in the Salvadoran context, PHs that produce coffee do not refer to wild-food collection as a coping strategy. Several studies in Central America as well as other parts of the world have demonstrated the importance of gathered species for peasant households facing dietary and economic constraints [51,52]. Such foods are, generally, most available precisely during cycles of agriculture production renewal, making them central to the diet during the lean season. In general, bromatological analyses have shown that these foods are highly nutritious—especially leafy greens and healthy protein sources such as insects, fungi, fish, crustaceans, and gastropods [53,54]. Management practices associated with these species are, usually, quite regenerative, and, therefore, do not compromise the biological stock of the resources [55].

4.2. Breaking down Seasonal Food Scarcity in an Attempt to Counteract It in a Targeted Manner

As mentioned in the Introduction, the existence of seasonal food scarcity and deployment of coping strategies in rural areas are, relatively, well documented in some regions of the world [56]. However, the central knowledge gap that this article helps to fill concerns internal and external socioeconomic factors that determine both food scarcity and coping strategies. For example, numerous findings show that staple crops are deeply related to reduction in food vulnerability within peasant communities [57]. In the CART, this is the case for maize production, but not for bean production. Therefore, the cultural value of a food does not, necessarily, imply sufficiency of its supply. Another important aspect of seasonal food scarcity that has not been sufficiently addressed is land tenancy, which impacts access to community decision-making [2]. Within the Mexican context of agrarian communities (the ejidos), we show that access to land and community decision-making processes by ejidatarios, pobladores, and avecindados is a variable that explains both the number of lean months and the intensity of coping strategies. Ejidatario PHs (with full access to land and rights) experienced significantly fewer lean months, and carried out less-severe coping strategies than the other two groups. Furthermore, type of land tenancy plays an important role, in determining the amount of cropland and livestock that PHs can utilise or own, which, in turn, influences the amount of government transfers they receive. The most significant economic assistance provided by Mexico´s federal government consists of cash transfers for crop, livestock, and forestry production, which are conditional upon possession of means of production, such as land or livestock. The interconnection between land tenancy and government support allows us to understand the dynamics of internal differentiation that are accentuated in peasant regions, and which cause some subgroups to experience prolonged periods of scarcity each year, while other groups do not [14].

Migration to the United States is a crucial strategy for overcoming the internal peasant differentiation in the microregion. Our statistical analyses indicate that international migration can mitigate seasonal food scarcity and, consequently, reduce the intensity of coping strategies. However, net income from regional and national migration are comparable to wages from agricultural labour within the communities, given the high cost of living in the regions to which peasants migrate [6]. Furthermore, these jobs, often, have extremely precarious working conditions [58].

A central finding of this study is the lack of statistically significant differences between seasonal food scarcities experienced by PHs of the upper (agroforestry) and lower (agricultural) regions of the watershed. This finding points to the effects of “win-win” agroforestry projects that have been promoted for over a decade by NGOs and the biosphere reserve’s government administration. These projects focus, mainly, upon biological conservation, and the promoters’ priorities include participating in payment for environmental services; advancing forest frontiers, while halting agricultural frontiers; and promoting organic certification of products destined for elite markets (e.g., “conservation coffee”). However, these projects contribute very little to improving rural livelihoods [17] and practically ignore the dietary conditions of peasant families [14]. Thus, our findings align with those of other studies that are critical of so-called “green economies” [30,59], which imply a new form of division of labour and generate marginalization or an “eco-precariat” [60,61]. Like those studies, ours found that peasants find themselves producing cheap inputs for “green” agroindustry (e.g., pine resin, palms, and parchment coffee), while earning next to nothing with which to purchase high-priced processed foods and agricultural inputs.

Another novel approach of this study was to ask PH members about agrifood innovations that they believe could counteract seasonal food scarcity. It is noteworthy that households expressed greater interest in social economy practices and generation of steady jobs than in agricultural production. Many mentioned local and federal government programs of the 1970s, which fomented cooperative models of production, marketing, and consumption, as well as savings banks, trust loans, price regulation, an efficient state purchase network, and the public procurement of rural supply [59]. PHs argue that one of the best ways to combat seasonal food scarcity is by providing steady rural jobs that offer “a minimal secure source of income during the shortage”. Such jobs are completely absent in the CART, as well as in most of rural Mexico. The relatively low emphasis upon agricultural alternatives is related to the peasants’ acute awareness of the growing risks, attrition, and production vulnerabilities that they face. Many PHs are already at the limit of their maize- and bean-planting capacity [40], and realise that their transaction costs and shadow prices are not taken into account in the markets in which they participate, in an asymmetrical, marginal manner [62]. Tittonell and Giller [25] explain that it is common for peasant households to seek to overcome the food-insecurity poverty trap by “lowering the trap barrier”, rather than via a “push over the barrier” (the usual objective of government assistance), or by a “transformation of the system” (advocated by agroecologists), which would radically reorient agricultural and economic models [63].

4.3. Seasonal Vulnerability and Robustness in Peasant Food-Supply Networks

Although the term “network” is widely used in agrifood research [64,65], few authors have formalised metric analyses and network visualizations, as we have done in this study (see related work on network interventions to catalyse agricultural innovations [66]; collective action in local agrifood systems [67]; complexity of the global agrifood sector [68]; and historical changes in global agricultural trade networks, due to climate change [69]). Another important contribution of this study is the peasant identity of the nodes, within the network visualizations, and the specific level of microregional resolution in the metric analysis. Upon interpreting our findings, we note that the literature [70] contains reflections upon the difficulties and risks of attempting to explain the structure and function of other types of networks—such as ecological, computational, and epidemiological networks—in specific social networks, such the agrifood networks analysed in this study. These authors [70] argue that supply networks—by the very nature of the processes they entail—tend to be well represented as directed bipartite networks. Similarly, [71] argues that metric attributes, which inform structural robustness and vulnerability, should be, specifically or contextually, defined beyond the general parameters for all types of networks.

Heeding that advice, we note that in the CART, the beef network’s robustness during the scarce season derives from the network’s improved structural attributes (density, local efficiency, average node connectivity, average page rank, average in-degree centrality and average out-degree centrality; refer to Table 2 for interpretation). However, in ethnographic terms, the network becomes highly exclusive, and a large majority of PHs in the lower socio-economic clusters do not have enough money to consume beef. By contrast, the maize supply network appears to be structurally vulnerable because it is highly disconnected and dissociated, but empirically it is a robust self-sufficiency network that maintains metric attributes across both seasons, and provides for PHs’ maize consumption. Indeed, structural-network analyses can identify potential community or government actions, aimed at making supply networks more robust—for example, by activating solidarity-economy-type community arrangements, which could strategically link the excess of locally produced maize and beans with the supply needs of households that are unable to self-supply these dietary staples throughout the year.

Network studies certainly provide valuable lessons that can guide the theoretical design of robust and efficient networks [72,73]. However, agrifood networks, such as the ones documented here, cannot be “engineered” as if they were composed of anonymous agents: the power relationships and external forcing that have shaped the networks have to be taken into account, in order to aspire to fair and long-lasting structural changes of supply networks. Thus, in agrifood research, it will be important to generate theoretical networks that make it possible to analytically deepen the study of empirical food networks in multiple contexts, on the basis of well-documented networks and in dialogue with local knowledge. In subsequent works, we will further develop the conceptualizations necessary to reach this goal.

5. Conclusions

Approximately 4.5 billion people, mainly belonging to peasant and indigenous communities, and using less than a quarter of the land, water, and fuel used in global food production, provide the majority of the food for approximately 70% of the world’s population. We support approaches that politically defend the role of these small-scale producers, but at the same time, we emphasise that many of these indigenous and peasant families of the Global South face a “farmer–food-scarcity paradox”, of recurrent annual seasonal food shortage that forces them to deploy intense coping strategies. Therefore, we believe that progressive governments, the global peasant movement, human rights organizations, and international agrifood-decision-making spaces should begin discussions to guide efforts to counteract this indigenous and peasant reality. In the calls to acknowledge and respond to seasonal food scarcity as one of the main expressions of food insecurity among indigenous and peasant communities worldwide, it will be important to carry out studies that analyse the socioeconomic and land-tenancy conditions that determine the extent and severity of shortages. Studies based upon research methods, such as those reported here, may contribute to designing public policies, local development plans, and community-support programs focused on counteracting specific seasonal food needs. As demonstrated in this article, rather than assuming that improving peasant household’s food supply requires imminent boosting of agriculture capacities, it is important to consult indigenous and peasant families regarding their needs. This study indicates that, rather than seeking more agricultural incentives, many PHs seek the generation of stable employment, market arrangements that allow them to receive higher prices for their products, associative figures to promote generating savings and social consumption of food, and cash subsidies that correlate with their periods of scarcity. Such initiatives should be promoted on the international, national, and local levels, to address the farmer–food-scarcity paradox.