Agricultural shock resilience strategies and their effects on family farms in rural Senegal

The health crisis caused by the COVID-19 pandemic came on top of several preexisting agricultural shocks (rising prices of inputs, crop pests, crop diseases, inter alia) and climatic hazards (flooding, extended seasons, inter alia) that threaten rural farm households. Agriculture contributes between 30 and 40% of gross domestic product (GDP) in African countries, and small-scale farm households are the fundamental basis for food security and the prosperity of rural communities (Asfaw et al., 2018; Call and Sellers, 2019; Makate et al., 2019). Consequently, agricultural shocks and extreme weather conditions that negatively influence farmers’ welfare and farm productivity also threaten rural economies overall (IPCC, 2021). In response to perceived vulnerability, farms may adopt resilience strategies (also known as adaptation, resistance and prevention strategies) to mitigate future shocks (Fatemi et al., 2017; Aryal et al., 2016). Such resilience strategies might include adjustment of agronomic practices (changes in the agricultural calendar, soil conservation and irrigation), changes in agricultural processes (crop diversification), capital investments (income diversification), among others (Devkota et al., 2017; Niles et al., 2016; Zhang et al., 2015; Tagang Tene et al., 2025).

Agriculture in Senegal has declined as a share of GDP, from 30% in the 1960s to around 10% as of 2021 (World Bank, 2021), yet it remains the primary source of income and employment for rural populations, with 95% of producers cultivating less than two hectares. Most farmers practice subsistence agriculture, live below the poverty line, and are exposed to agricultural and climatic shocks that limit asset accumulation and human capital development (World Bank, 2006). Vulnerabilities are exacerbated by natural disasters, economic fluctuations, health crises, such as COVID-19, and climate-related challenges, including irregular rainfall, rising temperatures and worsening pest and disease pressures. Projected climate changes—temperature rises of 2–4°C, rainfall reductions of 5–25%, aridification in northern Senegal, coastal erosion, and sea level rise—could reduce cereal productivity by up to 50% and increase food insecurity by 17%, affecting nearly half the population (Lacroix et al., 2021; CIRAD-GRET, 2021; FAOSTAT, 2017).

Despite these challenges, agriculture remains a key driver of economic growth, contributing 15.32% to GDP in 2021 and providing 30% of employment (World Bank, 2023). The sector is central to the Emergent Senegal Plan (PSE) that has sustained economic growth of 6.6% on average from 2014 to 2018, with projections of 10.2% growth from 2019 to 2023 (AfDB, 2023). However, poverty remains highest in rural areas (53.6% versus 19.8% in urban areas) (ANSD, 2018). Senegalese agriculture is largely small-scale, low-input and rain-fed, with growing seasons of up to three months across agroecological zones such as the Groundnut Basin, Casamance, River Valley, Niayes and Silvo-pastoral zones. These systems are highly vulnerable to soil, water and forest degradation, as well as climate change impacts, which reduce productivity and household incomes, undermining sectoral growth and exacerbating rural poverty (Cisse and Diop, 2022).

To address the challenges of climate change and agricultural vulnerability, the Government of Senegal has invested in the sector through the Senegalese Agriculture Acceleration Program (PRACAS), under the Emerging Senegal Plan (PSE). PRACAS focuses on promoting family farming, fostering rural entrepreneurship, involving youth and women, and strengthening the resilience of vulnerable populations (FAO, 2023). To support these objectives, the Government has established several institutions, including Polyvalent Rural Expansion Centers (CERP), Regional Development Assistance Centers (CRAD), the Agricultural Marketing Office (OCA) and the Senegalese Development Bank, whose primary mission is to assist rural farms, businesses and communities. Alongside these institutions, 4.3% of the national budget has been allocated to the agricultural sector (CEA, 2017). Despite this state investment, farmers still face production constraints related to desertification (20–25% depending on the region), water limitations (14%), phytosanitary challenges (8%), steep slopes (11.2%) and water and wind erosion (4%) (EAA, 2018). Consequently, resilience strategies remain necessary to reduce the vulnerability of farms to agricultural and climatic shocks (Gregory et al., 2005). In line with these goals, PRACAS emphasizes sustainable land management to curb land degradation and enhance agricultural productivity across agro-ecological zones. To this end, a variety of techniques are implemented, including assisted natural regeneration (RNA), erosion control, agroforestry practices for restoring degraded land, the use of organic and mineral amendments and water-saving measures through improved production practices.

Resilience is a complex phenomenon, comprising different strategies that can play important roles in improving the living conditions of farmers. Broadly defined, resilience is the adjustment of natural or human systems to respond to current or expected situations, to moderate or adapt to negative consequences and to take advantage of opportunities that changing conditions may present (Di Falco et al., 2011; IPCC, 2001). Thus, the resilience of farms in Senegal consists of the adoption of different strategies depending on the perception of risks, local knowledge of adaptation strategies and resources available to adapt. Senegalese farmers use a range of strategies including crop diversification (37.5%), use of traditional practices, knowledge and heritage (30%), use of seeds adapted to local conditions and stresses (22.9%), sale of animals (27.6%), diversification of farm and household activities (27.5%), sale of crops (22.2%), irrigation (7.6%) and drawing on government support (20.8%) (EAA, 2021). This study examines the impacts of resilience strategies designed to mitigate agricultural shocks on family farms in rural Senegal. It focuses on two main questions: first, what factors limit or promote the adoption of these strategies among rural farm households, and second, what effects do such strategies have on farm productivity and household income? By addressing these issues, the research contributes to the literature on agricultural adaptations to economic and climate-related shocks, offering new insights into the dynamics of resilience in rural farming systems. While a considerable literature has examined agricultural shocks and climate variability, existing studies have predominantly emphasized the relationship between climate change adaptation strategies and agricultural yields (Khanal et al., 2018a; Quan et al., 2019; Tagang Tene, 2022; Basse et al., 2022). In Sahelian countries with agroecological conditions comparable to those of Senegal, research has largely addressed climate change adaptation in general terms, without isolating the effects of specific strategies (Cissé and Khalifa, 2022; Basse et al., 2022; Diallo et al., 2022). To date, to the best of knowledge, no study has investigated the impacts of individual adaptation strategies in the Senegalese context. This study, therefore, builds on recent contributions (Tagang Tene, 2022; Biru et al., 2020; Affoh et al., 2024) that seek to identify the most effective strategies for mitigating the adverse impacts of climate change on agricultural productivity in developing countries, by analyzing their individual effects. Evidence from prior work (Stoneman and Toivanen, 1996; Biru et al., 2020) indicates that disaggregated analyses yield more robust insights than approaches treating adaptation as a single construct, though results may vary depending on the type of strategy and the geographic context.

This study contributes to the existing literature by drawing on the context of Sahelian African countries, with particular focus on the case of Senegal, analyzing the effects of individual adaptation strategies on the country’s main cereal crop (millet) in terms of yield. From a policy perspective, the findings of this study will provide valuable insights for strengthening the agricultural sector. They will support the development of robust, evidence-based adaptation strategies and inform public policy aimed at enhancing their effectiveness for smallholder farmers. A clearer understanding of the impacts of these strategies will also help address the shortcomings of previous policies. To achieve these objectives, the study hypothesizes that adaptation strategies such as the use Dikes, crop rotation and certified seed; contribute by increasing agricultural productivity for farmers in Senegal. This work is organized as follows: Section 2 presents the literature review, and Section 3 summarizes the methodological approach. Section 4 presents results and discussions, and Section 5 discloses robustness tests. The final section concludes.

Many studies have been conducted globally to investigate the factors influencing the adoption of climate change adaptation in smallholder farming systems in Africa. There exist two strands of literature. The first strand of literature considers a single climate change adaptation, such as Dikes, crop rotation and certified seed (Tagang Tene et al., 2025; Asante et al., 2024; Guo et al., 2022; Mossie, 2022). For example, Tagang Tene et al. (2025) found that socioeconomic, farm-level and institutional factors influence a farmer’s decision to adopt improved sorghum in Cameroon. Asante et al. (2024) investigated the factors influencing drought-resistant seed in Ghana. They found that household heads’ education levels, family labor, farm size, membership in training and association positively and significantly impact drought-resistant-seed adoption. According to Martey et al. (2020), the main factors influencing farmers’ adoption of improved sees in Ghana are access to seeds and extension, gender, labor availability and location. Kimathi et al. (2021) found that access to information, quality seeds, training, group membership and variations in agroecological zones are the most important factors influencing farmers’ decisions to adopt climate-resilient potato varieties.

The second stand of literature has investigated the impact of climate change adaptation on the specific impacts of individual adaptation strategies on agricultural production. Khanal et al. (2018b), using a simultaneous equation model and an endogenous switching regression model, showed that in Nepal, farmers’ adoption of climate-smart agricultural practices and technologies led to significant increases in food production. Among the strategies analyzed, soil and water management had the most substantial impact, followed by adjustments to agricultural calendars and changes in crops and varieties. Paddy Mveng et al. (2023), using the propensity score matching method, found that in Cameroon, the use of improved seeds significantly enhanced household food productivity. Zegeye et al. (2022), using an endogenous switching regression model, demonstrated that, in Ethiopia, the use of organic and inorganic fertilizers, along with herbicides, in response to climate change significantly improved food production. Similarly, Musafiri et al. (2023) highlighted the role of inorganic fertilizers in improving agricultural yields under climate stress in Kenya. In Nigeria, Onyeneke (2020), using a multivariate Probit model, instrumental variable models and an endogenous treatment effect model, found that rice farmers’ adaptation actions, including minimum tillage, bunds, drainage systems, fertilizers, livelihood and crop diversification, improved rice varieties, pesticides, nurseries and adjustments in sowing and harvesting dates; significantly improved yields.

Regarding cereal, the main staple crop for households in Senegal, Musafiri et al. (2022), using an endogenous switching regression model, found that in Kenya, improved soil management practices such as minimum tillage are essential for cereal production in areas highly vulnerable to climate change. They found that minimum tillage significantly increases yields by improving soil organic matter, structure, fertility and reducing erosion. Getachew et al. (2021), through crop modeling in Ethiopia, showed that full irrigation and early sowing enhance cereal yields. However, full irrigation after late sowing fails to fully offset the negative effects of climate change. Tagang Tene (2022), using an endogenous switching regression model and matching method, found that, in Cameroon, livelihood diversification and early sowing significantly improved cereal productivity among adopting households. They reported that adaptation increased yields by 21.99% for adopters and that yields would decline by 11.22% without adaptation.

In Senegal, research highlights the importance of agricultural adaptation strategies. Basse et al. (2022) show that adopting good agricultural practices in the cashew sector significantly increases yields, with particularly notable gains among women, while Cissé and Khalifa (2022) indicate that farmers in Louga also use a variety of adaptation strategies. Diallo et al. (2022) reveal that rainfall variability in the peanut basin of Thiès reduces both yields and cultivated areas, emphasizing the importance of improving farming practices. Although several studies have assessed the effectiveness of adaptation strategies for enhancing agricultural productivity in Senegal and other Sahelian countries with similar agroecological characteristics, very few have conducted a comparative analysis of the individual effects of these strategies across different agroecological zones. This study contributes to this effort by performing a comparative analysis of the impact of selected adaptation strategies on agricultural yields, focusing particularly on cereal-producing households, the country’s main staple crop.

This work is based on resilience theory, which emphasizes the ability of systems to adapt and transform in the face of shocks. This theory provides a dynamic and holistic framework for understanding not only the challenges faced by farms but also evaluating the effectiveness of strategies put in place to overcome them. It allows us to analyze the strategies adopted by family farms in response to climate disruptions and variations. In rural contexts, where agriculture is predominant and often subject to climatic, economic, and sociopolitical shocks, this theory offers critical insights (Gunderson, 2000).

Resilience makes it possible to assess how farms can not only survive shocks but also emerge from them by developing adaptation strategies (Walker et al., 2004). By developing these practices, farmers demonstrate their ability to adapt to changing conditions while transitioning to sustainable agricultural approaches that respond to immediate shocks.

Thus, drawing on the importance of social networks and community relationships, this theory of resilience shows that solidarity among farmers and the sharing of resources and knowledge play a crucial role in collective resilience. Community initiatives that target the restoration of natural resources and cooperation to access markets strengthen social cohesion and promote a collective response to crises.

Building on the previous theory and literature, in this paper we conceptualize the behavior of farms faced with the perceived risk of an agricultural and climatic shock as a function of the individual characteristics of farmers (sex, age, education, well-being), the characteristics of the farms (cultivated area, etc.), the societal and geographic environment (region, environment, urban/rural), the perceived risk of consequences (economic, security, environmental) of shocks and finally state intervention (Figure 1). Although there are several resilience strategies possible, we focus on the adoption of preventive measures recommended by the WHO and emphasized in previous research, including the diversification of on-farm and off-farm income-generating activities, infrastructure investments such as irrigation or erosion control measures, improved seeds and others. In many cases, farms that adopt such resilience measures not only protect themselves against agricultural and climatic shocks, but also enhance resilience to economic and social stressors (as seen through the COVID-19 pandemic) that can undermine agricultural systems and affect the production of farms and the profitability of farmers.

The impact of resilience strategies seeking to mitigate agricultural shocks on farms in sub-Saharan Africa is poorly documented, particularly in Senegal. This study aims to fill this gap through a combination of econometric approaches, including multivariate Probit models and an endogenous switching regression to better understand the impacts of three types of resilience strategies: the construction of dikes, crop rotation, and use of certified seeds. This empirical approach makes it possible to have more robust results by comparing differences in outcomes realized between farmers who have adopted resilience strategies and those similar farmers who have not adopted them. We are thus able to test the effectiveness of resilience strategies to agricultural shocks for enhancing productivity and incomes for family farms in rural Senegal.

This study relies on a combination of household-level agricultural data and climate information to examine the determinants of climate resilience and their impact on agricultural production in Senegal. The primary source of data is the 2018 Senegal Annual Agricultural Survey (ASS), conducted by the National Agency of Statistics and Demography (ANSD). This data set is cross-sectional in nature, capturing information collected at a single point in time. The survey gathered detailed agricultural data from 5888 households, providing a nationally representative sample that covers all 45 departments of Senegal. The data set has a hierarchical structure, integrating information at the household, plot and community levels. Household-level variables capture demographic and socioeconomic attributes, including education, household size and access to productive assets, as mentioned in Table 1. Plot-level data describe agricultural practices, crop rotation, certified seed, land area, input utilization and soil management techniques. Regional identifiers also make it possible to link households to their corresponding agro-ecological zones, thereby facilitating spatial analysis. Overall, the cross-sectional nature of the data provides a comprehensive overview of agricultural production systems and adaptation practices across Senegal’s diverse ecological contexts. To complement the household survey, we merged the AAS data with climate records obtained from Senegal’s network of meteorological stations. These records include monthly averages of temperature and precipitation for 2017 and 2018, covering both the rainy and dry seasons. However, a common challenge when using cross-sectional data for micro-level climate analysis lies in the limited spatial variation of key climatic indicators, particularly in developing countries where weather stations are sparsely distributed (Koudjom, 2022). To mitigate this limitation, we computed seasonally disaggregated averages of temperature and rainfall for each agro-ecological zone using a spatial interpolation method (Wahba, 1990). The interpolation procedure was implemented in QGIS software, using multiple reference points to enhance spatial precision. We acknowledge that uncertainties may arise from the interpolation method, measurement errors and local variations in topographic characteristics such as elevation and slope (Dandonougbo, 2021). In line with recommended best practices (Hutchinson and Webster, 1998), we increased the density of reference points and concluded cross-validation to assess the robustness of the interpolated estimates. This approach is consistent with previous studies employing similar procedures in climate-agriculture research (McKenney-Easterling et al., 2000; Di Falco et al., 2011).

First, we identify key resilience strategies of farms in the face of agricultural and climatic shocks. In the context of this research, resilience strategies are measures taken by farmers to adapt, resist or prevent future vulnerability to agricultural shocks (plant disease, variation in input prices, high rates of crop pests, fire, etc.) and climatic shocks (floods, drought, irregular rains, landslides, soil erosion, etc.). Farms in rural Senegal have adopted several strategies; we focus on three, including the construction of dikes (C), crop rotation (R) and the use of certified seeds (S). Resilience is a capacity for anticipation, resistance and adaptation to maintain production in the face of a shock, farmers can draw on these measures to maintain activity by mitigating the shock, resisting or adapting after the shock.

After identifying key agricultural resilience strategies, we then examine factors that limit or promote adoption of these strategies using a multivariate Probit model. Faced with a shock, each farm may adopt nothing as a resilience strategy or adopt one or more strategies at a time. We are aware of whether a technique was used at each farm, but we are unaware of how often it was used. Let us look at the three indicator variables Ci, Ri and Si. If the farm has used a particular resilience approach, then the value will be 1, and if not, it will be 0. Each of these indicator variables’ values is derived from a latent variable, $Ci*,Ri*,Si*$ which is unobserved but whose components may be broken down into a vector of parameters β_C, β_R, β_S, a vector of explanatory variables (X), and a stochastic component that is represented by an error ε_C, ε_R, ε_S. It is assumed that the joint distribution of these stochastic variables follows a normal distribution.

Formally, we obtain a system of three equations as follows:

where:

For identification purposes, we assume that var(ε_ij) = 1(j = C, R, S).

The equation defines a multivariate Probit that predicts the probabilities of adopting the different agricultural resilience strategies. The simultaneous structure of equation (1) jointly decides whether to implement any resilience measures. The three latent variables that represent unobserved qualities for farms can be correlated, if any, according to this definition. Griffith and Peres-Neto (2006) demonstrated that a univariate approach would overlook potentially non-zero non-diagonal parts of the variance–covariance matrix. When there is a correlation between the error terms, this would result in estimates that are inconsistent (Maddala, 1983). Under the block specification of equation (1), the probability of a family farm adopting one of the three resilience strategies is given by equation (2):

with c, r, s = 0, 1

The parameters β and ρ_ij are estimated by maximum likelihood, with $Φ4(.,∑˜)$ the Gaussian cumulative function of dimension 3 with $∑˜$ the associated variance–covariance matrix. This model can be estimated consistently and efficiently by maximizing the log-likelihood function:

According to Cappellari and Jenkins (2003), the GHK (Geweke–Hajivassiliou–Keane) approach simulates the maximum likelihood method, which is used to estimate equation (3). The GHK approach takes advantage of the fact that a multivariate normal distribution function can be written as the product of simple and precisely estimable one-dimensional sequentially conditioned functions of the normal distribution.

We next evaluate the impact of the adoption of the same three resilience strategies [construction of dikes (C), crop rotation (R) and the use of certified seeds (S)] on farm productivity and the income of rural farm households using endogenous switching regression, which takes into account observed and unobserved variables to correct for endogeneity and self-selection in the adoption of agricultural resilience strategies. Relying on the framework of random utility, we assume a farm chooses whether or not to adopt a resilience strategy depending on the characteristics of the farmers, the farm and the context. Adoption (A_i) is a visible manifestation of the unobservable latent variable (⁠$Ai*$⁠) of the adoption decision as shown in equation (4):

where Z_i = (1, z_i1, z_i2, z_i3, …, z_ik) is a vector of explanatory variables, α a vector of parameters to be estimated and ε_i is random error distributed according to a normal law. Current climatic factors, perceptions of production shocks, access to financing, information on available agricultural innovations, access to agricultural extension, farmer characteristics and family farm characteristics can all have an impact on the adoption of a resilience strategy (⁠$Ai=1 or Ai∗>0$⁠).

HC is the vector of household characteristics, LI is the vector of agricultural labor inputs, EZ is the vector of ecological zones, SO is the origin of seeds, CV is the vector of climatic variables, CP is the vector of production constraints, and IS is Vectors related to information on agricultural shocks, such as radio broadcasts, climate information and perceptions of production shocks.

We investigate numerous functional forms to model the effect of implementing resilience techniques on family farm productivity. The simplest way to investigate the impact of resilience strategies would be to include a dummy variable (A) equal to 1 if the farmer has implemented resilience measures in response to agricultural shocks and then use ordinary least squares (OLS) to predict productivity or income. However, this approach may produce skewed estimates since it implies that the adoption of an agricultural resilience strategy is exogenously influenced whereas it could be endogenous. The decision to adopt an agricultural resilience strategy or not may be based on self-selection, which means that farms that have implemented agricultural resilience methods may have systematically different features than those that have not.

In addition, farmers may have chosen agricultural resilience techniques based on predicted yields. Unobservable characteristics of farms and their operators can influence both the decision to use resilience methods and agricultural performance, posing the risk of inaccurate assessments of the impact of resilience strategies on agricultural output. Taking into consideration the potential endogeneity of the decision to adopt resilience methods, we estimate a model of simultaneous equations of adoption and agricultural production using a full information maximum likelihood endogenous switching regression method. Unlike studies that employ the fitted values created automatically by the non-linearity of the selection model to account for endogeneity (Fu et al., 2018; Aboal and Tacsir, 2017), we use an exclusion constraint to identify the model (Abdulai and Huffman, 2014; Takam-Fongang et al., 2019; Tsambou and Tagang Tene, 2024). This constraint is required when some variables directly affect the selection variable (adoption of resilience techniques) but not the outcome variable, which is approximated by agricultural productivity (Coromaldi et al., 2015).

Thus, we use as selection instruments for the agricultural productivity function variables related to information on other agricultural shocks, including the perception of sand/silting, and the use of seeds from personal reserves as well as seeds purchased on the local market. The admissibility of these instruments is effective after a sample homogeneity test and a tampering test (Di Falco et al., 2011). Therefore, if the selection instruments are valid, they will have an impact on the decision to adopt agricultural resilience strategies, but not on the productivity of farms that have not adopted them. To account for selection bias, we use an endogenous switching regression model of agricultural productivity in which farms face two regimes: adopting agricultural resilience strategies (regime 1) and not adopting them (regime 2). The regression model for yield is defined in equation (5a):

With the estimated model in equation (5b):

where HC is the vector of household characteristics, LI is the vector of agricultural labor inputs, EZ is the vector of ecological zones, SO is the origin of seeds, CV is the vector of climatic variables, CP is the vector of production constraints and $A⌣$ is the instrumental variables vector.

Where A is the probability of adoption of agricultural resilience strategies in equation 5a, Y_i represents the productivity within regimes 1 and 2, X_i represents the vector of the explanatory variables. The variables β₁ and β₂ are the vectors of the parameters to be estimated, and the error terms ε_i, μ_1i and μ_2i in the selection equation (1) and yield (productivity) equation (3) are assumed to have a trivariate normal distribution with mean 0 and covariance matrix ∑ i.e., (ε, μ₁, μ₂): N(0, ∑):

where $σε2$ are the variances of the error terms in the yield functions (5a) and (5b), σ_μ1ε and σ_μ2ε represent the covariance of ε_i, μ_1i and μ_2i and ε where is the variance of the error in the selection equation (4), which can be assumed equal to 1 since the coefficients can only be estimated up to a scale factor (Maddala, 1983). Since Y_1i and Y_2i are not observed concurrently, there is no definition for the covariance between μ_1i and μ_2i. The error structure has a significant meaning in that the error terms of the productivity functions (5a) and (5b) relate to the error term of the selection equation (6). In the event of sample selection, the anticipated values of μ_1i and μ_2i are zero:

where $λ1i=ϕ(Ziα)Φ(Ziα)$⁠, $λ2i=ϕ(Ziα)1−Φ(Ziα)$⁠, ϕ(.) is the standard normal probability density function and Φ(.) is the normal cumulative density function. Thus, we have in equation (7):

where: $θji=(Ziα+ρjμji/σj)1−ρj2$⁠, J = 1,2 with $ρ1=σμ1ε2σμ1σε$ and $ρ2=σμ2ε2σμ2σε$ meaning the correlation coefficient between the error term ε_i of selection equation (4) and the error terms μ_ji of equations (5a) and (5b), respectively. Equation (8) illustrates the significance of this regression model, which is that it enables post-estimate analysis to compare the yield in terms of expected production across family farms that have implemented agricultural resilience strategies (a) and those that have not (b). Furthermore, in the case of the hypothetical counterfactuals (c) for the farms that adopted resilience strategies if they had not, as well as for the farms that did not adopt these resilience strategies (d) if they had, one can assess the return in terms of expected production. In each of the four scenarios, these conditional agricultural yield expectations are described as follows:

The sample’s real expectations are shown in cases (a) and (b). Cases (c) and (d) show what the counterfactuals’ anticipated results would be. Additionally, we compute the treatment effect of implementing an agricultural resilience strategy on the treated (ATT) in accordance with Heckman and Vytlacil (2001) as the difference between (a) and (c), which denotes the impact of implementing agricultural innovations on the productivity of farmers who genuinely implemented a resilience strategy for their family farm in equation (9):

Likewise, the difference between (d) and (b) can be used to compute the treatment’s impact on untreated farmers (ATU) who did not implement resilience mechanisms for their farms in equation (10):

Productivity is the ratio of the quantity produced (in kg) and the area (in hectares) of the farm. The survey provides information on the structure of production, including the quantities produced and the areas of agricultural holdings. This allows us to calculate agricultural productivity, which is different from total factor productivity, which would further incorporate measures of the efficiency of labor and the productivity of capital in the production process.

We use ESR instead of Propensity Score Matching (PSM) for more than one reason. First, ESR (endogenous switching regression) offers greater flexibility for modeling complex relationships between resilience strategies and agricultural productivity (Paddy Mveng et al., 2023). It allows for the incorporation of interactions between covariates and the control of unobserved effects, which are often more difficult to achieve with PSM. Second, ESR uses all available data, avoiding the loss of observations that can occur during matching. This ensures that even unmatched individuals contribute to the analysis as covariates, allowing for better use of the data. In addition, ESR directly provides estimates of variable coefficients, making it easier to assess effects on the outcome of interest and perform statistical tests on these coefficients. Finally, ESR is more robust to selection bias by explicitly modeling causal relationships.

Table A1 presents the mean and difference test of the means of the socioeconomic variables by resilience strategy adoption status. The main dependent variable of interest – farmer productivity as proxied by crop yield per hectare – is measured in two ways: the minimum productivity reported by the farmer for a given plot and the maximum productivity reported for that same plot.

Farmers who built dikes report lower productivity on average than those who did not. The minimum productivity of farmers who built dikes is 212.28 kg/ha lower than that of farmers who did not (p < 0.050), with no significant difference in maximum productivity. Farmers who built dikes are less likely to be female-headed households, and more likely to be married. They are also younger on average, and more likely to have no formal education (p < 0.050). Farmers who built dikes on average have greater farm area than those who did not, with a significant difference of 0.33 hectares (p < 0.001). In terms of farm management practices, farmers who have built dikes are more likely to purchase seed on the local market, and less likely to purchase with subsidies with partners or from specialized firms. There are also significant differences between farmers who built dikes and those who did not use manual work (more common on farms with dikes), animal-drawn work (less common with dikes) and use of mineral inputs (less common with dikes). Finally, the construction of dikes also varies by agroecological zone.

Farmers who have adopted crop rotation similarly report lower productivity than those who have not adopted this resilience strategy. Specifically, the minimum productivity of farmers who have implemented crop rotation is 194.26 kg/ha lower (p < 0.001) and the maximum productivity is 331.36 kg/ha lower (p < 0.001). Farmer with a crop rotation system have fewer members in their household on average compared to those who have not, with no meaningful differences across gender of the household head. On average, the age of farmers who have set up a crop rotation system is roughly 1 year higher than that of farmers who have not (p < 0.001), and although most farmers are married, those engaged in crop rotation are on average more likely to be married. In contrast to dikes which were more common among those farmers with no formal education, those farmers engaged in crop rotation are more likely to have at least some primary- or secondary-level education and also more likely to have received training in agriculture. In terms of plot size, farmers who have set up a crop rotation system on average have more land than those who have not, with a significant difference of 0.55 hectares (p < 0.001). Patterns for farm management strategies, largely mirror the patterns observed for adopters versus non-adopters of dikes – however, we note that farmers using crop rotation are more likely than non-adopters to use organic fertilizers and mineral fertilizers, and more likely to purchase seed with support of state subsidies. Adoption of crop rotation also varies significantly by agroecological zone.

Patterns among farmers who adopted certified seeds versus non-adopters differ notably from the other resilience strategies examined. First, adopters have higher average minimum and maximum crop yields than those who did not adopt certified seeds, although the differences are not statistically significant. Farmers who adopted certified seed are also more likely to be in female-headed households (p < 0.100) and less likely to be married (p < 0.100). We again see significant differences in the use of this resilience strategy by education, with farmers with no formal education much less likely to adopt certified seed (p < 0.001). Speaking Wolof is also positively associated with adopting certified seed (although this was not significant for either of the other resilience strategies). Among farm characteristics, farmers who adopted certified seed had larger plots on average than those who did not (p < 0.001); they were also more likely to use motorized equipment (p < 0.001) and mineral fertilizers (p < 0.001), perhaps reflecting certified seed’s tendency to be used as part of a “package” of modern inputs. Not surprisingly, there was a significant difference in seed purchasing between farmers who adopted certified seeds and those who did not, with farmers using certified seed less likely to report sourcing seed from personal reserves or the local market (p < 0.001), and more likely to make use of seed subsidies or to purchase from specialized firms (p < 0.001). As with the two previous resilience strategies, adoption of certified seed also varies significantly by agroecological zone.

Figure 2 shows adoption rates for the three specific resilience strategies selected for the analysis – construction of dikes, crop rotation and use of certified seed – across agroecological zones. Farmers who have constructed dikes are most found in the Groundnut Basin (35.2%) as are farmers who have implemented crop rotation (32.7%). Farmers who have adopted certified seeds are most common in the Casamance area (33.7%) and the Groundnut Basin (28.8%), with smaller numbers (6–9%) in other zones.

Table 2 summarizes factors associated with the decision to adopt a resilience strategy in the face of agricultural shocks. Many recent studies in sub–Saharan Africa suggest that farmers perceive significant changes in climate, including rising temperatures, changes in humidity and reduced precipitation. Findings in Table 2 suggest that, in the rainy season, farmers may choose to build dikes, rotate crops or use certified seeds in the hope of improving or maintaining the yield of their crop. In the rainy season, family farms in rural areas adopt assorted strategies to cope with climate variability such as variation in humidity, temperature and precipitation. Similarly, in the dry season, the probability of building dikes, rotating crops and using certified seeds decreases in the presence of humidity, while it increases in the presence of rainfall. Farmers being rational, they consider the perception of climatic shocks in their agricultural decision-making. These results confirm those of Bryan et al. (2009) in the Limpopo basin in South Africa and of Koguia et al. (2021) which show that adaptation to climate change is an appropriate response to perceptions of climate shocks.

The probability of adopting the crop rotation strategy increases significantly with household size, as household size is often considered an indicator of farm household labor. In Senegal, family farms mostly use family labor, which is why the large size of the household would be a source of increased probability of adopting resilience strategies. These results are consistent with those of Diallo et al. (2020) in Mali and those of Onyeneke (2020) in Nigeria, which show that household size is a relevant explanatory factor for resilience to agricultural shocks. The female sex, the age of the head of the household and the level of education of the head of the farmer significantly increase the probability of rotting the crops, which shows that women farmers are more involved in agricultural activities in rural areas. Farmers acquire knowledge and experience with age, which allows them to confidently implement resilience strategies. Also, education is not generally a statistically significant determinant of crop rotation adoption. Only higher education shows a weak significance level (1%), and, interestingly, the coefficient is negative. This suggests that, in our sample, farmers with tertiary education are slightly less likely to adopt crop rotation practices compared to those without formal education. A similar pattern is observed for the adoption of certified seeds, where higher education is negatively and significantly (10%) associated with this practice. This finding may suggest that more educated farmers tend to diversify their income sources or engage more in non-farm activities, which may reduce their direct involvement in specific agricultural practices such as crop rotation or the use of certified seeds. These results contrast with the expected positive relationship often found in the literature (e.g. Khanal et al., 2018b), which emphasizes that education improves farmers’ analytical capacity and their ability to adopt adaptive and productivity-enhancing practices. In the Senegalese context, however, higher education may instead lead to a gradual disengagement from traditional farming activities. In addition, the mastery of the local language (Wolof) is necessary for the use of certified seeds because the extension bodies generally use the Wolof language to sensitize local farmers to reach not only a large layer, but also to draw their attention to climate variability and agricultural shocks.

The likelihood of adopting resilience strategies depends on the agro-ecological zones. The agro-sylvo-pastoral zone, groundnut basin, eastern Senegal and Casamance are positively favorable to the construction of dikes and crop rotation. While the agro-sylvo-pastoral zone and groundnut basin are negatively linked to the use of certified seeds. These results are in line with those found by Diallo et al. (2020) showing the importance of location in the adoption of adaptation strategies to climate shocks.

The correlation coefficients between the different resilience strategies taken two by two are positively significant in two out of three cases, showing a positive interaction between some resilience strategies. The construction of dikes has a positive and significant association with crop rotation (rho 21). Similarly, the construction of dikes has a negative and significant association with the adoption of certified seeds (rho 31). Overall, farmers are able to combine the two strategies of building dikes and other types of resilience strategies (crop rotation and improved seeds). The correlation between crop rotation and other types of resilience strategies is positive and significant. This shows that crop rotation promotes the adoption of certified seeds (rho 32). Thus, in the presence of agricultural and climatic hazards, farmers in rural areas in Senegal adopt the crop rotation strategy combined with the use of certified seeds as resilience strategies.

We used the endogenous switching regression model to account for the issue of endogeneity while analyzing the effect of resilience tactics on agricultural productivity to assess the impact of agricultural shocks on productivity. productivity in agriculture. The explanation for the endogenous switching of production functions can be found in Table 3. An intriguing outcome consists of the signs and significances of the covariance terms ρ1 and ρ2. The findings demonstrate that, in the majority of models, the covariance terms for nonresilient individuals are statistically significant, suggesting that self-selection took place within the context of resilience. Therefore, even if nonadopters decide to adopt resilience techniques, the adoption of such strategies to lessen the impact of agricultural shocks does not have the same effect on them. Furthermore, there may be heterogeneity in the sample based on the variations in the productivity equations’ coefficients between the farmers who used them and those who did not.

In line with the economic literature, farmers who speak most of the time in the Wolof language significantly reduce the agricultural productivity of those who have respectively built dikes, rotated crops and adopted certified seeds by 13%, 19% and 25%, respectively. They still reduce the agricultural productivity of those who have not terminated. Despite the awareness and popularization of certain resilience strategies, farmers are still attached to their cultures and ancestral methods and, therefore, are reluctant to adopt certain resilience strategies. Manual labor or hiring labor reduced the agricultural productivity of those who built dikes by 7%, by 17% for those who rotated crops and by 118% for those who adopted seeds certified. Thus, the hiring of unskilled labor contributes to the reduction of productivity due to the non-mastery by these workers of good farming practices. These results contradict those found by Onyeneke (2020), Diallo et al. (2020) and Khanal et al. (2018b) who show that the labor factor significantly increases agricultural productivity in Nigeria, Mali and Nepal, respectively.

The use of agricultural inputs by farmers contributes to the increase in agricultural productivity by 22% for those who have built dikes, 38% for those who have rotated crops and 31% for those who have adopted certified seeds. These results are consistent with the work of Khanal et al. (2018b), which shows that the use of fertilizers increases the agricultural productivity of all farmers who adopt adaptation strategies. Moreover, the increase in productivity in this case is justified by the fact that the quantities of fertilizer dosage in the fields have been respected by the farmers. Farmers living in Casamance contribute significantly to the reduction of agricultural productivity by 76%, 105% and 198% for those who have built dikes, rotated crops and adopted certified seeds, respectively. Thus, farmers in Casamance are less reluctant to adopt resilience strategies. This is justified by a more favorable environment for agriculture compared to other areas.

Climatic and agricultural shocks drastically impact agricultural production. In this regard, the silting up of fields contributes significantly to the increase in agricultural productivity of those who have built dikes, rotated crops and adopted certified seeds by 11%, 21% and 37%, respectively. As for soil salinity, it considerably increases the agricultural productivity of those who have built dams, rotated crops and adopted certified seeds by 39%, 6% and 118%, respectively. Thus, in the face of climatic shocks, farmers are led to adopt resilience strategies to cope with these various shocks.

Table 4 presents predicted agricultural productivity under real and counterfactual conditions. Cells (a) and (b) for each resilience strategy represent the observed agricultural productivity of the sample. Cell (c) represents the predicted agricultural productivity per hectare of the different resilience strategies if farmers had decided not to adopt them and cell (d) represents the predicted agricultural productivity per hectare of the non-resilient of the different resilience practices if they had decided to adopt them. The last column shows the effect of various resilience strategies on agricultural productivity, which is calculated as the difference between columns 3 and 4. Thus, we find that the adoption of various resilience strategies to agricultural shocks has a statistically significant impact on agricultural productivity; that is, the construction of dikes, crop rotation and adoption of certified seeds are statistically significant for agricultural productivity.

Farmers who built dikes produced 248.68 kg/ha compared to 364.98 kg/ha for farmers who did not. However, this simple comparison can be misleading and lead researchers to conclude that, on average, those who built dikes produced 31.86% (or 116.3 kg/ha) less than those who did not. do. Rainwater retention bunds help prevent field flooding, erosion, leaching and soil nutrient depletion. These results have not yet been discussed in the literature. Farmers who rotated crops produced 258.56 kg/ha compared to 255.61 kg/ha for farmers who did not. Farmers who did rotate crops produced 2.95 kg/ha (or 1.15%) more than those who did not rotate crops. These results are consistent with those obtained by Onyeneke (2020), Diallo et al. (2020) and Khanal et al. (2018b) in their respective countries, which show that farmers who used resilience methods such as crop rotation produced 7% more than those who did not.

Similarly, the agricultural productivity of farmers who have adopted certified seeds is about 421.65 kg/ha against 244.99 kg/ha for farmers who have not adopted this variety of seeds. Farmers who adopted certified seeds produced about 176.68 kg/ha, an increase of 72.11% compared to those who did not. To increase yields in the face of the scarcity of rainfall, farmers are forced to sow improved (certified) seeds. These results confirm those of Onyeneke (2020) in Nigeria and Diallo et al. (2020) in southern Mali, who found that the use of seeds with short germination times increases agricultural productivity and household food security. In addition, farmers who built dikes would have produced about 148,135.6 kg/ha more if they had not built dikes. These results are in line with those found by Quan et al. (2019) who conclude that certain strategies such as soil and water conservation and irrigation have a significantly negative impact on crop yields in China. All this is explained through the poor adaptation of some farmers.

Similarly, those who rotated the crops would have produced 169.58 kg/ha less if they had not rotated the crops. Finally, farmers who adopted certified seed would have produced 227.66 kg/ha less if they had not adopted certified seed. On the other hand, farmers who did not build dikes would have produced 188.03 kg/ha less if they had built dikes. Also, those who did not rotate crops would have produced 26.48 kg/ha less if they had. On the other hand, those who did not adopt certified seeds would have produced 21,656.93 kg/ha more if they had adopted it. The results of this study are consistent with those of many other studies (Di Falco et al., 2011; Khanal et al., 2018b; Diallo et al., 2020; Dessalegn et al., 2022; Tagang Tene, 2022). They also show that if households do not adapt, they will lose 20% of their production. And a 35% increase in the production of non-adaptive households if they adapt to climate change.

Table 5 presents the predicted agricultural productivity under real and counterfactual conditions in the different agro-ecological zones of Senegal.

To this end, farmers in the Senegal Valley who built dikes produced 373.27 kg/ha against 1464.91 kg/ha for farmers who did not. However, this simple comparison can be misleading and lead researchers to conclude that, on average, those who built dikes produced 1091.64 kg/ha, or 74.51% less than those who did not. As for farmers in the Niayes area, they produced 254.08 kg/ha against 182.007 kg/ha for those who did not. Thus, farmers in the Niayes area who actually built dikes produced 72.073 kg/ha, or 39.59% more than those who did not. For farmers in the Sylvio-pastoral zone who built dikes, they produced 281.14 kg/ha against 118.04 kg/ha for farmers who did not. Farmers in the Sylvio-pastoral zone who actually built dikes produced 163.1 kg/ha, or 138.09% more than those who did not. Concerning the Groundnut Basin farmers who built dikes, they produced 279.38 kg/ha against 229.64 kg/ha for the farmers who did not. Farmers in the Groundnut Basin who actually built dikes produced 49.74 kg/ha, or 21.65% more than those who did not. As for farmers in the eastern Senegal zone, they produced 109.18 g/ha against 134.63 kg/ha for farmers who did not build dikes. Farmers in the eastern Senegal zone who built dikes actually produced 25.45 kg/ha, or 18.9% less than those who did not. Similarly, farmers in Casamance who built dikes produced 165.31 kg/ha against 411.12 kg/ha for those who did not build dikes. Thus, farmers in Casamance who actually built dikes produced 245.81 kg/ha, or 59.79% less than those who did not. Rainwater retention bunds help prevent field flooding, erosion, leaching and soil nutrient depletion. These results have not yet been discussed in the literature.

In addition, farmers in the Senegal Valley who built dikes would have produced about 239,845.1 kg/ha more if they had not built dikes. Those in the Niayes area who built dikes would have produced 113,968.7 kg/ha more if they had not built dikes. Similarly, farmers in the Sylvio-pastoral zone who built dikes would have produced 269,466.7 kg/ha more if they had not built dikes. For farmers in the Groundnut Basin who built dikes, they would have produced 164,986.5 kg/ha more if they had not built dikes. As for farmers in Eastern Senegal who built dikes, they would have produced 19,192.02 kg/ha more if they had not built dikes. Finally, farmers in Casamance who built dikes would have produced 44,746.55 kg/ha more if they had not built dikes. These results are in line with those found by Quan et al. (2019) who conclude that certain strategies such as soil and water conservation and irrigation have a significantly negative impact on crop yields in China. All this is explained through the poor adaptation of some farmers.

On the other hand, farmers in the Senegal Valley who did not build a dike would have produced 1092.91 kg/ha less if they had built dikes. Also, those in the Niayes area who did not build a dike would have produced 69.18 kg/ha less if they had built dikes. For farmers in the Sylvio-pastoral zone who did not build a dike, they would have produced 15.69 kg/ha less if they had built dikes. Unlike the Groundnut basin farmers who did not build a dike, they would have produced 18.97 kg/ha more if they had built dikes. Regarding farmers in Eastern Senegal who did not build a dike, they would have produced 30.03 kg/ha less if they had built dikes. For farmers in Casamance who did not build a dike, they would have produced 254.65 kg/ha less if they had built dikes. The results of this study are consistent with those of many other studies (Di Falco et al., 2011; Khanal et al., 2018b; Diallo et al., 2020; Dessalegn et al., 2022; Tsambou and Tagang Tene, 2024). They also show that if households do not adapt, they will lose 20% of their production. And a 35% increase in the production of non-adaptive households if they adapt to climate change.

Table 5 presents the predicted agricultural productivity under real and counterfactual conditions in the different agro-ecological zones of Senegal. It should also be noted that crop rotation has a significant impact on agricultural productivity.

To this end, farmers in the Senegal Valley who rotated crops produced 452.3 kg/ha against 387.3 kg/ha for farmers who did not. However, this simple comparison can be misleading and lead researchers to conclude that, on average, those who did indeed burp crops produced 65 kg/ha, or 16.78% more than those who did not. As for the farmers in the Niayes area who rotated the crops, they produced 346.07 kg/ha against 153.9 kg/ha for those who did not. Thus, farmers in the Niayes area who actually rotated crops produced 192.17 kg/ha, or 124.86% more than those who did not. For farmers in the Sylvio-pastoral zone who rotated crops, they produced 274.79 kg/ha against 310.72 kg/ha for farmers who did not. Farmers in the Sylvio-pastoral zone who did rotate crops produced 35.93 kg/ha, or 11.56% less than those who did not. Concerning the Groundnut Basin farmers who rotated the crops, they produced 259.94 kg/ha against 311.12 kg/ha for the farmers who did not. Farmers in the Groundnut Basin who effectively rotated crops produced 51.18 kg/ha, or 16.45% less than those who did not. As for the farmers in the eastern Senegal zone who rotated the crops, they produced 118.04 g/ha against 107.67 kg/ha for the farmers who did not rotate the crops. Farmers in the eastern Senegal zone who rotated crops actually produced 10.37 kg/ha, or 9.63% more than those who did not. Similarly, farmers in Casamance who rotated crops produced 190.7 kg/ha compared to 183.32 kg/ha for those who did not rotate crops. Thus, farmers in Casamance who actually rotated crops produced 7.38 kg/ha, or 4.02% more than those who did not. These results are consistent with those obtained by Onyeneke (2020), Diallo et al. (2020) and Khanal et al. (2018b) in their respective countries, which show that farmers who used resilience methods such as crop rotation produced 7% more than those who did not.

In addition, farmers in the Senegal Valley who rotated the crops would have produced about 314.83 kg/ha less if they had not rotated the crops. Those in the Niayes area who rotated the crops would have produced 245.61 kg/ha less if they had not rotated the crops. Similarly, farmers in the Sylvio-pastoral zone who rotated the crops would have produced 163.47 kg/ha less if they had not rotated the crops. For farmers in the Groundnut Basin who rotated the crops, they would have produced 162.97 kg/ha less if they had not rotated the crops. As for the farmers in Eastern Senegal who rotated the crops, they would have produced 84.19 kg/ha less if they had not rotated the crops. Finally, farmers in Casamance who rotated the crops would have produced 132.61 kg/ha less if they had not rotated the crops.

On the other hand, farmers in the Senegal Valley who did not rot the crops would have produced 0.42 kg/ha more if they had rotted the crops. Also, those in the Niayes area who did not rotate the crops would have produced 7.38 kg/ha more if they had rotated the crops. Unlike farmers in the Sylvio-pastoral zone who did not rotate the crops, they would have produced 120.95 kg/ha less if they had rotated the crops. Similarly, farmers in the Groundnut Basin who did not rot the crops would have produced 51.86 kg/ha less if they had rotted the crops. For farmers in Eastern Senegal who did not rot the crops, they would have produced 1.72 kg/ha more if they had rotted the crops. For farmers in Casamance who did not rot the crops, they would have produced 15.92 kg/ha more if they had rotted the crops. The results of this study are consistent with those of many other studies (Di Falco et al., 2011; Khanal et al., 2018b; Diallo et al., 2020; Dessalegn et al., 2022; Tagang Tene, 2022). They also show that, if households do not adapt, they will lose 20% of their production. And a 35% increase in the production of non-adaptive households if they adapt to climate change.

Table 5 presents the predicted agricultural productivity under real and counterfactual conditions in the different agro-ecological zones of Senegal. It should also be noted that the adoption of certified seeds has a significant impact on agricultural productivity.

To this end, farmers in the Senegal Valley who adopted certified crops produced 609.45 kg/ha against 426.23 kg/ha for farmers who did not. However, this simple comparison can be misleading and lead researchers to conclude that, on average, those who actually adopted the certified crops produced 183.22 kg/ha, or 42.98% more than those who did not. As for farmers in the Niayes area who adopted certified crops, they produced 500.58 kg/ha against 231.09 kg/ha for those who did not. Thus, farmers in the Niayes area who actually adopted certified crops produced 269.49 kg/ha, or 116.61% more than those who did not. For farmers in the Sylvio-pastoral zone who adopted certified crops, they produced 203.4 kg/ha against 274.07 kg/ha for farmers who did not. Farmers in the Sylvio-pastoral zone who actually adopted certified crops produced 70.67 kg/ha, or 25.78% less than those who did not. Concerning the Groundnut Basin farmers who adopted certified crops, they produced 592.81 kg/ha against 260.42 kg/ha for farmers who did not. Groundnut Basin farmers who actually adopted certified crops produced 332.39 kg/ha, or 127.63% more than those who did not. As for farmers in the eastern Senegal zone who adopted certified crops, they produced 120.6 kg/ha against 113.89 kg/ha for farmers who did not adopt certified crops. Farmers in the eastern Senegal zone who actually adopted certified crops produced 6.71 kg/ha, or 5.89% more than those who did not. Similarly, farmers in Casamance who adopted certified crops produced 325.86 kg/ha against 171.76 kg/ha for those who did not adopt certified crops. Thus, farmers in Casamance who actually adopted certified crops produced 154.1 kg/ha, or 89.71% more than those who did not. These results are consistent with those obtained by Onyeneke (2020), Diallo et al. (2020) and Khanal et al. (2018b) in their respective countries, which show that farmers who used resilience methods such as crop rotation produced 7% more than those who did not.

In addition, farmers in the Senegal Valley who adopted certified crops would have produced about 278.87 kg/ha less if they had not adopted certified crops. Those in the Niayes area who adopted certified crops would have produced 274.4 kg/ha less if they had not adopted certified crops. Similarly, farmers in the Sylvio-pastoral zone who adopted certified crops would have produced 11.37 kg/ha more if they had not adopted certified crops. For farmers in the Groundnut Basin who have adopted certified crops, they would have produced 358.78 kg/ha less if they had not adopted certified crops. As for farmers in Eastern Senegal who adopted certified crops, they would have produced 0.61 kg/ha more if they had not adopted certified crops. Finally, farmers in Casamance who adopted certified crops would have produced 189.88 kg/ha less if they had not adopted certified crops.

In addition, farmers in the Senegal Valley who did not adopt certified crops would have produced 48,880.99 kg/ha more if they had adopted certified crops. Also, those in the Niayes area who did not adopt certified crops would have produced 10,235.24 kg/ha more if they had adopted certified crops. Unlike farmers in the Sylvio-pastoral zone who did not adopt certified crops, they would have produced 14,138.54 kg/ha less if they had adopted certified crops. As for farmers in the Groundnut Basin who did not adopt certified crops, they would have produced 33,967.61 kg/ha more if they had adopted certified crops. Regarding farmers in Eastern Senegal who did not adopt certified crops, they would have produced 3869.54 kg/ha more if they had adopted certified crops. For farmers in Casamance who did not adopt certified crops, they would have produced 6538.6 kg/ha more if they had adopted certified crops.

The results of this study are consistent with those of many other studies (Di Falco et al., 2011; Khanal et al., 2018b; Diallo et al., 2020; Dessalegn et al., 2022). They also show that if households do not adapt, they will lose 20% of their production. And a 35% increase in the production of non-adaptive households if they adapt to climate change.

The objective of this work has been to analyze how family farms are resilient and how such strategies of resilience influence the productivity of family farms. To do this, we first identified relevant resilience strategies used by farms in rural Senegal in the 2018/2019 Annual Agricultural Survey. Then, we have examined factors associated with the adoption of alternative strategies of resilience by using a multivariate Probit framework. Finally, the study assessed how engaged strategies of resilience influence how considered family farms are productive.

We focus on three practices, crop rotation, the construction of dikes and the use of certified seeds as rural farms’ resilience strategies in the face of climatic and agricultural shocks. It is apparent from the multivariate analysis that climate variables (humidity, temperature, precipitation), farm owner’s education level, the size of the agricultural household and the agroecological zone are substantially linked to the adoption of resilience strategies by family farms. The impact evaluation shows that attendant strategies of resilience in rural areas positively impact family farms’ productivity.

Ultimately, these results provide a solid basis for policies intended to facilitate the promotion of agriculture, for the design or redesign of agricultural programs or for the decision to invest in the agricultural sector in rural areas. These results constitute a methodological guide for the march toward the achievement of SDG 2 on zero hunger. More policy implications are discussed in what follows, especially as it pertains to the relevance of the findings in the improvement of resilient strategies of family farms linked to established climatic factors (i.e. humidity, temperature and precipitation), farmer’s educational level, the agroecological zone and the size of the agricultural household.

First, practical targeted extension services can be considered by policy makers in providing critical information and the relevant training to farmers in the light of the contingencies in the level of education, household features and attendant agroecological zones of the farmers. Second, policy makers should also design practical measures aimed at promoting climate-smart agriculture, especially as it pertains to encouraging the wide adoption and implementation of practical strategies such as improved livestock breeds, agroforestry, drought-resistant crops and water conservation. Third, extant farmer associations should be consolidated and if such associations are not apparent, policy makers should form such associations with the aim of facilitating the sharing of knowledge as well as collective actions among farmers with a view of sharing effective farming practices that enable crop and livestock resilience to the negative effects of climate change.

Fourth, improved access to resources should also be encouraged by ensuring that farmers have access to insurance, credit facilities and other valuable resources that are relevant in implementing the considered resilience strategies. Fifth, policy makers should also support research and development that are designed to improve current knowledge on climate resilient strategies among farmers, especially among family farmers. A step in this direction could entail, investment in research that is designed to develop and distribute agricultural technologies and crop varieties that are resilient to climate change.

By implementing targeted interventions, policy makers can enhance the adoption of resilience strategies and by extension, build more resilient family farms. The suggested climate-resilient farming practices are closely linked to several sustainable development goals (SDGs), especially those aimed at eliminating hunger, enhancing water resource management and encouraging responsible land use. By implementing methods such as water-saving irrigation, drought-resistant crops and soil preservation, farmers can better adapt to the effects of climate change, secure food supplies and lessen their vulnerability to climate-induced disasters. These approaches also support broader sustainability targets by improving living standards, promoting biodiversity and decreasing greenhouse gas emissions which are relevant in the achievement of most SDGs of the United Nations.

The study obviously leaves room for future research especially in view of considering whether the results withstand empirical scrutiny in the other African countries and panel of countries for which data are available. It also worthwhile to consider the problem statement within the remit of other SDGs and Agenda 2063 of the African Union pertaining to sustainable development.

This work was technically and financially supported by the International Fund for Agricultural Development (IFAD) under the 50x2030 initiative to close the agricultural data gap. The authors thank members of the IFAD research network who served as sources for valuable comments at various stages of the study. Our thanks also go to our mentor Professor Travis Reynolds who has used all his time to read and reread this work in order to offer us comments and constructive criticism to improve the quality of this work. The analyzes and opinions expressed in this document are the authors’ own. The authors are indebted to the editor and reviewers for constructive comments. The authors are indebted to the editor and reviewers for constructive comments.

All precautions related to scientific ethics have been taken into account from the conception to the writing of this article, as well as the submission to the said journal.

All authors consent to the publication of this article in this journal.

The codes or commands used for the analysis and processing of the data are available and can be made available to any reader for any purpose.

This document was coauthored by four authors, each of whom participated fully.