The climate of West Africa is characterized by a strong north-south rainfall gradient and a unimodal monsoon season that provides the vast majority of the region’s annual precipitation. This system is inherently variable, with a history of profound multi-decadal fluctuations, most notably the severe Sahelian droughts of the 1970s and 1980s ([29]). Since the mid-20th century, however, a clear and consistent warming trend has been superimposed upon this natural variability. The IPCC’s Sixth Assessment Report (AR6) states with high confidence that surface temperatures across Africa have increased at a faster rate than the global average ([17]). For West Africa specifically, observations show an average temperature increase of 1˚C - 3˚C since the mid-1970s, with the most pronounced warming occurring in the interior Sahelian zones. This warming has been accompanied by a marked increase in the frequency and intensity of heatwaves and a rise in the number of tropical nights ([17]).

While temperatures have risen, rainfall patterns have exhibited a more complex evolution. Following the extended drought period, there has been a partial and spatially heterogeneous recovery of rainfall since the mid-1990s, a phenomenon sometimes referred to as the “Sahel greening” ([21]). This recovery has not been a simple return to previous conditions. The character of the rainfall has changed, with evidence suggesting a trend towards fewer, but more intense, rainfall events and a higher frequency of extreme rainfall leading to flooding ([17]). This shift towards more intense, erratic rainfall, coupled with rising temperatures that increase evaporative demand, creates a complex and challenging environment for agriculture.

Future climate projections for the region, while consistent in forecasting continued and significant warming, exhibit considerable uncertainty regarding precipitation. This uncertainty is a major challenge for impact modeling and adaptation planning. The projections from the Coupled Model Intercomparison Project Phase 6 (CMIP6), which represent the latest generation of global climate models, suggest a different potential future for West Africa compared to the previous CMIP5 ensemble. A key study by [30] found that CMIP6 models, on average, project a wetter and slightly cooler future for the Sahel by mid-century compared to the hotter and drier scenarios common in CMIP5. This divergence is substantial, as it significantly alters the potential impacts on crop yields and water resources. Nonetheless, under all emission scenarios, temperatures are projected to continue to rise, with West Africa warming at a rate higher than the global average. Under an intermediate emissions scenario (SSP2-4.5), West Africa is projected to experience warming of approximately 2˚C by mid-century (2041-2060), while under a high emissions scenario (SSP5-8.5), this could reach 2.4˚C in the same period, and as high as 4.4˚C by the end of the century ([17]). Such temperature increases, irrespective of rainfall changes, will inevitably intensify evapotranspiration and place additional stress on crops.

The Gambian economy and the livelihoods of a majority of its population are intrinsically linked to the performance of its agricultural sector. The sector is dominated by smallholder farmers practicing subsistence and semi-commercial farming, with a heavy reliance on direct rainfall. This dependence on a brief and variable monsoon season is the central element of the country’s vulnerability to climate change ([7]). The agricultural system is characterized by low levels of mechanization, limited use of modern inputs such as improved seeds and fertilizers, and most importantly (within the context of this study), a near-total absence of irrigation infrastructure. According to [10], only about 1.12% of the total cultivated area in The Gambia is equipped for irrigation, leaving the remaining 99% entirely exposed to the vagaries of the rainy season. This situation is precarious. Any delay in the onset of the rains, a mid-season dry spell, or a premature end to the monsoon can have devastating consequences for crop yields and rural incomes. [4] demonstrated that household vulnerability to climate variability in Ghana is shaped by multiple interacting factors, including livelihood strategies, access to resources, and institutional support, findings that are applicable to similar contexts across West Africa, including The Gambia.

The country’s crop production is concentrated on a few key commodities. Groundnut is the traditional cash crop and a major source of export earnings, occupying a substantial portion of the arable land. The main food crops are coarse grains such as millet, maize, and sorghum, which form the bedrock of household food consumption, particularly in rural areas ([41]). Rice is also a major staple, but domestic production falls far short of national demand, leading to a heavy reliance on imports. It is estimated that The Gambia imports between 80% and 90% of its rice, making the national food supply highly susceptible to fluctuations in international market prices and supply chain disruptions ([15]). This structural food deficit highlights the importance of strengthening domestic production, a goal that is directly threatened by the impacts of climate change. The socioeconomic context further compounds this vulnerability. High rates of poverty, particularly in rural areas, limit the capacity of farmers to invest in adaptive measures. Access to credit, extension services, and markets remains a challenge for many smallholders. Furthermore, land degradation, driven by a combination of factors including population pressure and unsustainable farming practices, has reduced the productive capacity of the soil in many parts of the country. These baseline stressors mean that even minor climatic shocks can push vulnerable households into food insecurity. As noted by [6], under current trends of population growth and agricultural productivity, The Gambia will not be able to meet its domestic food demand by 2050, a gap that climate change is expected to widen.

The physiological processes of crops are intricately linked to climatic conditions, primarily temperature and water availability. Climate change impacts agricultural productivity by altering these fundamental drivers. The most direct impact of rising temperatures is an increase in atmospheric evaporative demand, a phenomenon described by the Penman-Monteith equation, which forms the basis for estimating reference evapotranspiration (ETo). As temperatures rise, the rate at which water evaporates from the soil surface and transpires from plants increases, meaning that crops require more water to maintain optimal growth and produce a given yield ([12]). Therefore, even if rainfall were to remain constant or increase slightly, the higher evaporative demand caused by warming could lead to increased water stress for crops.

Numerous studies have projected the consequences of this for crop water requirements (CWR) and irrigation water requirements (IWR) across Africa. [31], using high-resolution regional climate models, projected that under a 2˚C global warming scenario, crop water demand and irrigation needs would increase across West Africa. [12], in a study on the Nile Delta, predicted that agricultural water requirements could increase by as much as 9.7% to 18.2% under a high emissions scenario (RCP 8.5) by the end of the century. These findings are consistent with the general understanding that a warmer world will be a thirstier world, particularly for agriculture. This increased water demand, when set against a backdrop of potentially more erratic rainfall, translates directly into negative impacts on crop yields, especially in rain-fed systems. A meta-analysis by [24] found that projected impacts on crop yields in several African countries were predominantly negative, with average losses ranging from −10% to −6%.

The IPCC AR6 reinforces this conclusion, stating with high confidence that climate change has already reduced agricultural productivity across Africa. The report highlights that for West Africa, maize and wheat yields have already seen average reductions of 5.8% and 2.3%, respectively, due to historical climate change. Looking forward, the IPCC projects further declines; for instance, at 1.5˚C of global warming, sorghum yields are expected to decline, and maize yields could fall by 9% if adaptation measures are limited ([17]). [23] identified three overlapping adaptation phases: incremental, preparatory, and transformational, and demonstrated that many regions in sub-Saharan Africa will require transformational adaptation within the next few decades as current cropping systems become unviable. [2], on the other hand, reviewed common agricultural adaptation strategies across Africa, including drought-resistant crop varieties, crop diversification, changes in cropping patterns and planting calendars, and improved irrigation efficiency, noting that strengthening human capital through education and extension services is fundamental to enhancing adaptive capacity.

The impact of climate change on crop yields and production in Africa has been keenly examined. A meta-analysis by [18] projected mean yield changes of −17% for wheat, −5% for maize, −15% for sorghum, and −10% for millet across Africa by the 2050s, underscoring the severity of climate threats to food production in the region. The work of [30] provides a more nuanced perspective by comparing CMIP5 and CMIP6 projections. While their findings suggest that the CMIP6 ensemble projects less severe negative impacts for sorghum and millet in the Sahel compared to CMIP5, largely due to the wetter projections in CMIP6, the impact on maize yields remains negative. This indicates that the impact is crop-specific and that temperature-induced stress remains a dominant factor. The study also highlights the potential mitigating effect of elevated atmospheric CO₂ concentrations, which can improve water use efficiency in some crops (C3 plants like rice and groundnut). The magnitude of this CO₂ fertilization effect in real-world field conditions remains a subject of scientific debate and is often insufficient to offset the negative impacts of heat and water stress.

For The Gambia, these regional findings imply a significant future challenge. The projected increases in temperature will drive up the CWR for all major crops. For rain-fed crops like groundnut, millet, and sorghum, this will increase the risk of yield failure during dry spells. For rice, which is often grown in lowland areas that retain water, changes in the timing and intensity of rainfall could disrupt the growing cycle, while increased temperatures will still elevate water demand. Quantifying these changes in CWR is the first and most important step in assessing climate risk to Gambian agriculture and in formulating viable adaptation pathways.

The Gambia is physically defined by the River Gambia (Figure 1), which bisects the country and creates distinct agro-ecological zones that shape national farming systems. The river’s floodplain supports extensive rice cultivation, the country’s staple crop, particularly in the freshwater zones of the Central River Region, while the upland plateaus are dominated by rain-fed groundnut and coarse-grain production. This riverine geography dictates a gradient of agricultural practices, where tidal irrigation potential along the river banks contrasts with the drought-prone sandy soils of the uplands, making the entire agricultural sector highly sensitive to hydrological variability.

The agroclimatology of The Gambia is defined by a pronounced dry season and a concentrated wet season that creates a narrow window for crop production. It is characterized by a short, unimodal rainy season from June to October, which governs its predominantly rain-fed agricultural system (Figure 3). The country’s economy is heavily reliant on agriculture, which employs approximately 70% of the labor force and is dominated by smallholder farming. The primary crops include groundnut (the main cash crop) and staple cereals such as rice, maize, sorghum, and millet. With very limited irrigation infrastructure, the agricultural sector is especially vulnerable to variations in rainfall and temperature, making it a valid case for studying climate change impacts and adaptation.

Figure 1. Land cover distribution of The Gambia derived from the Environmental Systems Research Institute’s annual 10-meter global land use/land cover maps. These maps are based on the ESA Sentinel-2 imagery at 10 m resolution. It illustrates the dominance of agricultural land use along the River Gambia corridor (flooded vegetation) and the distinct transition between riverine rice cultivation zones and upland cropping systems.

The first stage of the analysis focuses on quantifying the direct impact of climate change on the water needs of key crops. This is achieved using the CROPWAT 8.0 software, a decision-support tool developed by the Land and Water Division of the Food and Agriculture Organization (FAO) of the United Nations ([9]).

3.2.1. Reference Evapotranspiration (ETo)

The foundation of the crop water requirement calculation is the reference evapotranspiration (ETo), which represents the evaporative demand of the atmosphere. ETo was calculated using the FAO Penman-Monteith equation (Equation 1), which is recommended as the standard method for its accuracy and ability to account for all major climatic parameters ([3]). The equation is given as:

where: ETo is the reference evapotranspiration [mm day⁻¹], Rₙ is the net radiation at the crop surface [MJ m⁻² day⁻¹], G is the soil heat flux density [MJ m⁻² day⁻¹], T is the mean daily air temperature at 2 m height [˚C], u₂ is the wind speed at 2 m height [m s⁻¹], eₛ is the saturation vapor pressure [kPa], e_a is the actual vapor pressure [kPa], e_s − e_a is the saturation vapor pressure deficit [kPa], Δ is the slope of the vapor pressure curve [kPa ˚C⁻¹], γ is the psychrometric constant [kPa ˚C⁻¹].

The parameter values used and the assumptions made are described in Appendix.

3.2.2. Climate Data and Scenarios

Climate data required for the Penman-Monteith equation (minimum and maximum temperature, humidity, wind speed, and sunshine hours) were sourced for two distinct periods. A historical baseline period (1995-2014) was established using data from the FAO CLIMWAT 2.0 database, which provides long-term monthly averages for over 5000 stations worldwide ([9]). For future projections, downscaled climate data from the Coupled Model Intercomparison Project Phase 6 (CMIP6) were obtained from the World Bank’s Climate Change Knowledge Portal. The ensemble mean of available models is selected for a mid-century period (2041-2060) under two Shared Socioeconomic Pathways (SSPs): SSP2-4.5, an intermediate emissions scenario, and SSP5-8.5, a high-emissions, fossil-fuel-intensive development scenario. These scenarios were chosen to bound a range of plausible future climate conditions.

3.2.3. Crop Parameters and Crop Water Requirement (CWR)

Five crops of central importance to The Gambia’s food security and economy were selected for analysis (groundnut, rice, maize, millet, and sorghum). The crop-specific water requirement (ETc), also known as crop water requirement (CWR), was calculated by adjusting the reference evapotranspiration (ETo) based on the specific characteristics of each crop, using the crop coefficient (Kc) approach (Equation (2)):

Table 1.Crop parameters used in CROPWAT analysis.

Source: Adapted from [3] and [1].

The crop coefficient (Kc) varies over the growing season, reflecting changes in crop canopy and physiological development (Table 1). Stage-specific Kc values (initial, mid-season, and late-season) and the length of each growth stage were derived from FAO-56 guidelines and supplemented with data from regional studies in West Africa to ensure local relevance ([3]; [1]). The parameters used are detailed in Table 1.

3.2.4. Net Irrigation Requirement (NIR)

The net irrigation requirement (NIR) was calculated as the difference between the crop water requirement (ETc) and the effective rainfall (Peff), which is the portion of total rainfall that is available for crop use. Effective rainfall is calculated using the USDA Soil Conservation Service method ([35]):

If P ≤ 250 mm/month:

Peff = P × (125 − 0.2 × P)/125

If P > 250 mm/month:

Peff = 125 + 0.1 × P (3)

where: Peff = Effective rainfall [mm/month]; P = Total monthly rainfall [mm/month].

This calculation identifies the water deficit that needs to be supplied by irrigation to avoid crop water stress (seeAppendix for how the irrigation water supply and the water stress factor are implemented in the model).

System dynamics (SD) is used to analyze the systemic implications of the changes in water requirements calculated in Part 1. This approach allows for the exploration of feedback mechanisms, long-term trends, and the potential impacts of policy interventions ([11]; [28]). The model was implemented in Python 3.11, utilizing the NumPy library for numerical calculations and Matplotlib for generating visualizations. This approach was chosen to allow for direct integration with the CROPWAT data processing scripts and to ensure full programmatic control over the simulation experiments and scenario analysis. The complete model specification, including equations and parameters, is provided in Appendix.

3.3.1. Conceptual Model Development

Key relationships in the climate and food production dynamics in The Gambia are characterized as a coupled socio-hydrological and land-use system where climatic forcing shapes crop water stress and yields, and where yield shocks propagate into food production, livelihoods, and demographic change (Figure 2). Temperature-driven increases in atmospheric demand and rainfall-mediated effective rainfall alter water stress, which controls realized crop yield and, through potential yield and irrigation coverage, the scope for buffering drought impacts. Within a system dynamics framing, these linkages are naturally interpreted through reinforcing and balancing feedback structures that help explain non-linear responses and threshold behaviour in food systems ([28]; [26]). [36] demonstrated how system dynamics modelling can integrate qualitative and quantitative data to explore the systemic interaction of drivers affecting economic, environmental, and social sustainability in agricultural production, providing a methodological foundation for the approach adopted in this study.

The first balancing mechanism (B1) operates as a distress-driven regulator. As the food security gap widens, rural livelihoods deteriorate, raising out-migration and/or mortality and thereby reducing population and food demand, which can narrow the gap (Figure 2). This loop stabilizes the accounting of food security at the expense of welfare and human capital, consistent with work highlighting the importance and potential erosion of stabilizing feedbacks in social-ecological systems ([20]).

Figure 2. Conceptual presentation of the relationship between climate, food production, and land in the systems analysis modeling.

The various loops in the system are described as follows:

B1. Food insecurity-demographic buffering. A widening food security gap undermines rural livelihoods, increasing mortality and/or out-migration, which reduces population and food demand and partially dampens the original gap (a stabilizing mechanism with high welfare costs).

B2. Degradation-expansion land compensation. Loss of arable land and rising demand trigger land expansion to compensate for degradation-driven constraints, partially restoring arable land while shifting pressure across the land system.

R1. Investment-irrigation yield intensification. Higher yields raise production and income, strengthening investment capacity and technology adoption that expand irrigation coverage and potential yield, reinforcing crop yield gains.

R2. Production-population-land expansion. Higher production improves food security and livelihoods, supporting population growth and food demand, which raise land demand and expansion, increasing arable land and reinforcing production.

R3. Investment-land stewardship(sustainable intensification). Production-driven income supports investment and adoption of improved practices; better land management can reduce degradation and stabilize arable land, reinforcing production through a maintained land base.

The land subsystem mediates longer-run trajectories through a tension between intensification and degradation. Reinforcing loop R1 describes a productivity-led pathway: higher crop yields increase food production and farmer income, which strengthen investment capacity and technology adoption, expand irrigation coverage, and raise potential yield, pushing yields upward. R2 adds a demand-mediated growth channel, where improved food security supports livelihoods and population growth, increasing food demand, driving demand for agricultural land and land expansion, and increasing arable land to sustain production. R3 represents a second investment pathway in which capital and improved practices raise land management intensity; when management quality is high, this can reduce land degradation and maintain arable land, aligning with sustainable intensification principles that seek higher output from existing land while conserving the resource base ([8]). The countervailing balancing loop B2 reflects land constraints. Here, degradation reduces arable land and can depress production, but the resulting pressure induces expansion to compensate, risking a trap in which short-term gains are offset by further degradation. [32] demonstrated that continued cropping without sufficient inputs of nutrients and organic matter leads to soil degradation and renders many soils in a non-responsive state, creating a chronic poverty trap where smallholder farmers cannot benefit from increased inputs of fertiliser and labour. These dynamics echo broader assessments that land degradation and climate pressures interact with food security and livelihoods, and that managing cross-sectoral linkages is necessary for robust policy design ([16]; [13]).

3.3.2. Model Simulation and Policy Analysis

Once calibrated using historical data on crop production and population, as well as recent studies on farmer incomes in Gambia’s rural agricultural landscapes, such as [41], the model is used to simulate the system’s behavior from the baseline period through 2060 under the SSP2-4.5 and SSP5-8.5 climate scenarios. The primary outputs for analysis will be national food security (measured as the self-sufficiency ratio), average farmer income, and groundwater sustainability.

Three policy scenarios will be simulated to assess their effectiveness as adaptation strategies:

Business-as-Usual(BAU). No significant new investment in agricultural water management. Infrastructure investment. A phased investment in expanding the area equipped for irrigation by 10% over two decades. Crop diversification. A program that incentivizes a gradual shift of 15% of land from water-intensive crops to more drought-tolerant traditional grains (millet and sorghum).

The results of these simulations will be compared to identify high-leverage interventions and potential trade-offs, providing a dynamic and policy-relevant extension to the static CROPWAT analysis.

The system dynamics model was validated against historical data from 2010 to 2020 to ensure its structural and behavioral realism, using data from the World Bank and FAOSTAT (see Appendix). The results showed a structural bias of −5.5% for population and 66.7% for total crop production. The normalized root mean square error (NRMSE) was 5.5% for population and 68.0% for total production, indicating the model’s simulated population figures were closely aligned with historical data, while simulated crop production showed a significant overestimation compared to observed figures.

The analysis began by quantifying the projected changes in key climate variables that directly influence crop water needs. Using downscaled CMIP6 data for The Gambia, the reference evapotranspiration (ETo) for the mid-century period (2040-2059) is calculated under two shared socioeconomic pathways, SSP2-4.5 (medium emissions) and SSP5-8.5 (high emissions), and compared to the historical baseline (1995-2014).

4.1.1. Projected Changes in Reference Evapotranspiration (ETo)

Figure 4.Monthly Reference Evapotranspiration (ETo) under different climate scenarios.

The analysis shows a consistent and significant increase in ETo across all months under both future scenarios, driven primarily by rising temperatures (Figure 4). For the baseline period, the total annual ETo was approximately 1870 mm. Under the SSP2-4.5 scenario, this is projected to increase by 6.3% to 1988 mm annually. The high-emissions SSP5-8.5 scenario results in a more pronounced increase of 7.4%, bringing the total annual ETo to 2009 mm. Figure 4illustrates the monthly ETo values for the baseline and the two future scenarios, highlighting that the largest absolute increases are projected to occur during the dry season (January to April), when temperatures are highest.

4.1.2. Crop-Specific Water Requirements (CWR)

The increased atmospheric water demand, reflected in higher ETo, translates directly to higher crop water requirements (CWR) for all five major crops analyzed. As shown in Figure 5 and detailed in Table 2, rice consistently has the highest water requirement, while millet has the lowest. Under the SSP5-8.5 scenario, the seasonal CWR for rice is projected to increase by 4.3%, from 714 mm to 745 mm. Similar increases are projected for the other crops, with groundnut, maize, millet, and sorghum showing CWR increases of 4.3% under the same high-emissions scenario.

Figure 5.Seasonal crop water requirements under different climate scenarios.

The direct biophysical consequence of rising temperatures and changing climatic conditions is a consistent increase in the seasonal Crop Water Requirement (CWR) for all major crops in The Gambia. As illustrated in Figure 6, by the mid-century period (2040-2059), all five crops studied—groundnut, rice, maize, millet, and sorghum—are projected to experience a uniform increase in water demand. Under the moderate emissions scenario (SSP2-4.5), CWR increases by approximately 3.3% - 3.4% relative to the 1995-2014 baseline. Under the high emissions scenario (SSP5-8.5), this increase is more pronounced, with all crops requiring approximately 4.3% more water over their growing season. This uniform rise in CWR across different crop types highlights the systemic nature of the impact, driven primarily by the higher evaporative demand (ETo) resulting from increased temperatures.

Figure 6. Projected change in water requirements for all crops in 2040-2059, relative to the baseline period of 1995-2014.

4.1.3. Net Irrigation Requirements

The combination of higher CWR and a slight projected decrease in effective rainfall during the growing season results in a substantial increase in the NIR, which represents the water deficit that must be supplied by irrigation to avoid crop water stress. For the baseline period, only rice, with its high water needs, showed a significant irrigation requirement of 227 mm per season (Figure 7). For all other crops, the effective rainfall was sufficient to meet the CWR. While the other rain-fed crops still do not show a seasonal water deficit in this aggregated model, the shrinking gap between water supply and demand indicates a rising vulnerability to mid-season droughts, a factor explored in the system dynamics model.

Table 2 presents a comparative summary of the seasonal Crop Water Requirement (CWR), effective rainfall (Peff), and Net Irrigation Requirement (NIR) for the five major crops under baseline and projected climate scenarios. The results reveal a consistent pattern across all crops. CWR increases while Peff decreases under both future scenarios. For the rain-fed crops groundnut, maize, millet, and sorghum, the effective rainfall during the growing season exceeds the CWR under all scenarios, resulting in a NIR of zero. This indicates that, on a seasonal aggregate basis, these crops can theoretically meet their water needs from rainfall alone, though this does not account for intra-seasonal dry spells that may cause periodic water stress. Rice, however, presents a markedly different situation. Its higher water demand (714.2 mm baseline CWR) already exceeds effective rainfall (487.5 mm) under current conditions, creating a baseline irrigation deficit of 226.7 mm per season. Under climate change, this deficit widens substantially. The NIR increases to 260.5 mm (+14.9%) under SSP2-4.5 and to 276.3 mm (+21.9%) under SSP5-8.5, driven by both rising CWR and declining effective rainfall. This finding highlights the acute vulnerability of irrigated rice production to climate change in The Gambia.

Figure 7. Net irrigation requirements under different climate scenarios.

Table 2.Comparative summary of CWR and NIR (mm/season).

While all crops exhibit increased water demand, the most severe impact is observed in the Net Irrigation Requirement (NIR) for rice, the only crop in this study cultivated under irrigated conditions. Figure 8 reveals a dramatic projected increase in the irrigation water needed to supplement rainfall and avoid yield loss for rice. Under the SSP2-4.5 scenario, the NIR for rice is projected to increase by a substantial 14.9% compared to the baseline period.

Figure 8. Projected change in net irrigation requirement in 2040-2059, relative to the base period of 1995-2014.

This increases to a 21.9% increase under the more severe SSP5-8.5 scenario. For the other four crops (groundnut, maize, millet, and sorghum), which are modeled as purely rain-fed, the NIR remains zero. This result highlights the acute vulnerability of The Gambia’s irrigated rice production, as the widening gap between crop water needs and available effective rainfall will place immense strain on the country’s already limited water and irrigation infrastructure.

The system dynamics model simulated the long-term (2020-2060) behavior of The Gambia’s agricultural system, integrating the biophysical impacts from the CROPWAT analysis with socioeconomic variables. Three policy scenarios were tested against the SSP2-4.5 and SSP5-8.5 climate futures: Business-as-Usual (BAU), Infrastructure Investment, and Crop Diversification.

4.2.1. Food Security Projections

Under the Business-as-Usual (BAU) policy scenario, The Gambia’s food security index (defined as the food self-sufficiency ratio) is projected to decline significantly under both climate scenarios. As shown in Figure 9, the index drops from a baseline of 0.65 in 2020 to approximately 0.35 by 2060. This decline is driven by a combination of climate-induced yield stagnation and rapid population growth. The “Infrastructure” and “Diversification” policy interventions show a marginal ability to slow this decline, but neither is sufficient to reverse the trend. The Food Security Index under these improved policy scenarios is only slightly higher, reaching approximately 0.37 by 2060. This indicates that while these interventions help, they are insufficient to counteract the overwhelming pressures of climate change and population growth on the national food system.

Figure 9.Projected food security index (2020-2060).

4.2.2. Crop Yield and Farmer Income

Average crop yields show a slight decline under the BAU scenario, driven by increasing water and heat stress. The policy interventions, particularly the “Infrastructure” scenario, which promotes irrigation, help to stabilize and slightly increase yields over the long term (Figure 10). By 2060, under the SSP5-8.5 climate scenario, the average yield in the BAU case is 1.15 t/ha, whereas in the Infrastructure and Diversification scenarios, it improves to 1.22 t/ha.

This modest yield improvement does not translate into significant gains in farmer income. Average farmer income is projected to decline across all scenarios, falling from a baseline of around $145 USD/year to below $68 - $75 per year by 2060 (Figure 11). The high costs of production and irrigation, coupled with low crop prices and a growing rural population, trap farmers in a cycle of low profitability, even when yields are stabilized.

Figure 10.Projected crop yields (2020-2060).

Figure 11.Projected farmer income (2020-2060).

4.2.3. Effectiveness of Policy Interventions

Figure 12. Comparison of key outcome indicators in 2060 under three policy scenarios (Business-as-Usual, Infrastructure Investment, and Crop Diversification) for SSP2-4.5 and SSP5-8.5 climate pathways.

The simulation results highlight the differential impacts of the policy scenarios. The “Infrastructure” scenario, which focuses on expanding irrigation, is the most effective at stabilizing crop yields by mitigating water stress (Figure 12). Its high investment cost limits its impact on farmers’ income. The “Diversification” scenario shows similar modest benefits to yield but has a lower implementation cost.

Figure 12 shows that while both adaptation policies offer an improvement over the BAU trajectory, the magnitude of this improvement is small. For instance, under the high-emissions SSP5-8.5 scenario, the Infrastructure policy only improves the food security index from 0.34 to 0.36 compared to BAU. This emphasizes the profound challenge facing The Gambia, where even proactive adaptation measures struggle to keep pace with the combined pressures of climate change and socioeconomic trends.

The CROPWAT analysis confirms a clear trend. Rising temperatures are set to significantly increase atmospheric water demand (ETo) in The Gambia, by as much as 7.4% under a high-emissions scenario (SSP5-8.5) by mid-century. This finding is consistent with the fundamental principles of the Penman-Monteith equation and aligns with regional projections for West Africa ([29]; [31]). The calculated increase in crop-specific water requirements (CWR) of 3% - 4% is a direct consequence of this heightened ETo. While this percentage may seem modest, it represents a substantial new stress on a system already operating at its margins.

The most important finding from this part of the analysis is the projected 21.9% increase in the Net Irrigation Requirement (NIR) for rice under the SSP5-8.5 scenario. This represents a significant new water deficit that, if unmet, will directly translate into yield losses. This result echoes the findings of [12], whose work in a different semi-arid region also projected substantial increases in irrigation demand under high-emissions futures. Given that The Gambia currently irrigates just over 1% of its cultivated land ([10]), and that rice is the primary staple for which the country is already 80% - 90% import-dependent, this escalating water deficit for the one crop that is commonly irrigated presents a formidable challenge to national food security goals.

For the other major rain-fed crops (groundnut, maize, millet, and sorghum), the model did not project a seasonal water deficit, even under future scenarios. This result, however, should be interpreted with caution. It reflects the fact that, on an aggregated monthly basis, total seasonal rainfall still exceeds the total seasonal CWR. This masks the threat of intra-seasonal drought. The West African monsoon is characterized by high variability, and IPCC AR6 highlights that the region is experiencing fewer but more intense rainfall events, alongside an increase in the frequency of meteorological and agricultural droughts ([17]). Therefore, while the total volume of water may be sufficient, its temporal distribution is becoming less reliable. A dry spell of just a few weeks during a growth stage, such as flowering, can devastate rain-fed crop yields, a nuance that a monthly water balance model like CROPWAT cannot fully capture but which the system dynamics model begins to address through its yield-stress functions.

The system dynamics simulation gave rise to some important results in relation to food security in The Gambia. The model projects a steep decline in The Gambia’s food security index, from 0.65 in 2020 to around 0.35 by 2060, across all policy and climate scenarios. This demonstrates that the combined pressures of rapid population growth (projected at over 2% annually for the coming decades) and climate-induced stress on agricultural productivity are set to overwhelm the system’s capacity to feed itself. This finding provides a dynamic and quantitative validation of the concerns raised by [6], who concluded that The Gambia’s current cropland will not meet future food demand.

Another important contribution of the system dynamics model simulation is its revelation of a “policy resistance”. The simulations showed that even proactive, well-intentioned adaptation strategies, such as the “Infrastructure Investment” and “Crop Diversification” scenarios, had only a marginal impact on the long-term food security trajectory. They slowed the decline but could not reverse it. This suggests that the system is caught in a reinforcing feedback loop, or an “adaptation trap”. Climate change suppresses yields and incomes, which in turn limits the capital available for farmers to invest in the very adaptation measures (like irrigation or improved inputs) that could break the cycle. The model shows that farmer income is projected to decline even in the adaptation scenarios, because the high costs of inputs and the small gains in yield are not enough to create a pathway to prosperity. [19] found that techno-informational, economic, and environmental barriers are strongly and inversely correlated with the adoption of adaptation measures, suggesting that agricultural development policies must consider adaptation measures tailored to specific agroecosystems to effectively alleviate barriers and promote smallholder resilience.

This result challenges the prevailing narrative that simply implementing a checklist of climate-smart agriculture (CSA) practices will be sufficient. The model outputs in this study align with the findings of [25], who systematically identified 24 potential adaptation options for The Gambia, but also noted the paucity of evidence for their effectiveness in the local context and the importance of considering barriers to adoption. Results of the simulation in this study suggest that the primary barrier is systemic; the socioeconomic context itself prevents the widespread adoption and success of these technical solutions. The problem is not a lack of knowledge about what to do, but a lack of capacity to do it at the scale and speed required.

The findings of this study have profound implications for policymakers in The Gambia and for its international development partners. The results strongly suggest that the country’s current adaptation planning, as outlined in its National Adaptation Programme of Action (NAPA) and subsequent strategies, may be necessary but is likely insufficient ([14]). These plans rightly focus on options like water management, crop diversification, and climate information services. Model outputs show that these are helpful, but they are incremental adjustments to a system that is facing an existential threat.

A business-as-usual approach, as simulated in the BAU scenario, leads to a severe decline in national food security and rural welfare. The policy debate, therefore, must shift from a focus on isolated, project-based interventions to one centered on impactful, systemic change. What might this look like? The model points to several high-leverage areas:

De-linking livelihoods from rain-fed agriculture. With over 90% of agriculture being rain-fed, the entire economy is hostage to the increasingly erratic monsoon. A national strategy focused on creating large-scale employment opportunities outside of traditional agriculture could be the most effective long-term adaptation strategy, as it would reduce the population’s significant direct dependence on rain-fed agricultural systems that are vulnerable to a changing climate. Key sectors for labor absorption include the services sector, which already employs over half the workforce ([39]), and the tourism industry, which contributes approximately 20% to GDP and is a growing source of employment ([34]). Further opportunities exist in the coastal fisheries sector, which provides direct and indirect employment to an estimated 25,000 - 30,000 people and is a significant source of nutrition and export earnings ([33]), and the emerging digital economy, which offers potential for youth employment and entrepreneurship through initiatives such as the National Youth and Women Development and Empowerment Through ICTs Strategy ([22]). High levels of state-led irrigation infrastructure. The model showed that farmer-led, income-dependent irrigation expansion is too slow. Only a large-scale, nationally coordinated program to take advantage of the Gambia River as a key resource for irrigation could potentially alter the trajectory of water availability for agriculture. Structural economic transformation. The low profitability of farming is an essential part of the trap. Policies that address the structural issues, such as input costs, market access, post-harvest losses, and low farm-gate prices, could be more impactful than policies that focus solely on climate resilience. Without a viable economic foundation, climate adaptation becomes an unaffordable luxury for subsistence farmers.

While this study provides valuable insights, it is essential to acknowledge its limitations. The CROPWAT model uses a monthly time step, which can mask the impact of short-term dry spells that are critical for rain-fed agriculture. A daily-step crop model could provide a more granular assessment of yield impacts. Second, the system dynamics model is an aggregation and simplification of a highly complex reality. The parameters used are based on the best available data, but they carry uncertainties. For example, the coefficients for farmer decision-making, technology adoption rates, and the precise impact of policies are based on literature averages and would benefit from in-country empirical validation. Third, the model does not explicitly account for the impacts of extreme weather events, such as floods or severe storms, which are also projected to increase in frequency and intensity. The policy scenarios tested are stylized representations; real-world policy implementation is far more complex and subject to political, social, and institutional factors not captured in the model.

Notwithstanding these limitations, the study serves its primary purpose, to move beyond a simple impact assessment and explore the dynamic behavior of the entire socio-ecological system. The strength of the model lies not in predicting the exact value of the food security index in 2060, but in revealing the underlying structures and feedback loops that are likely to govern its trajectory.

The SD model integrates five subsystems: Climate, Water Resources, Agricultural Production, Socioeconomic, and Policy. These subsystems interact through feedback loops that govern the long-term behavior of The Gambia’s agro-ecological system over a 40-year simulation period (2020-2060).

The model operates on an annual time step (dt = 1 year). At each time step, the Climate Subsystem provides temperature and rainfall values based on the selected climate scenario. These values drive the Water Resources and Agricultural Production subsystems, which calculate water availability and crop yields. The Socioeconomic Subsystem then computes population, food security, and farmer income based on agricultural output. The Policy Subsystem modifies parameters within the other subsystems according to the selected intervention scenario.

The Climate Subsystem serves as the primary external driver of the model. It does not contain internal feedback dynamics; rather, it reads projected values of temperature and rainfall from lookup tables based on the selected climate scenario (SSP2-4.5 or SSP5-8.5). These projections are derived from CMIP6 multi-model ensemble means, downscaled to the national level for The Gambia.

ReferenceEvapotranspiration(ETo)

For the SD model, a simplified temperature-based approximation is used:

ETo_annual = ETo_baseline × (1 + 0.04 × ΔT)

where: ETo_baseline = 1825 mm/year (baseline annual ETo for The Gambia), ΔT = Temperature change from baseline [˚C], 0.04 = Sensitivity coefficient (4% increase in ETo per 1˚C warming).

ParameterValues

Assumptions

Climate projections follow a linear interpolation between baseline (2020) and mid-century (2050) values.Inter-annual variability is not modeled; the simulation uses smoothed trend values.Wind speed, humidity, and solar radiation are assumed constant at baseline values due to limited projection data.The 4% ETo sensitivity to temperature is based on regional studies ([31]).

DataSources

Baseline climate data: FAO CLIMWAT 2.0 database for Yundum station, The Gambia (1995-2014 average).Future climate projections: World Bank Climate Change Knowledge Portal, CMIP6 ensemble mean for The Gambia, SSP2-4.5 and SSP5-8.5 scenarios ([37]).

EffectiveRainfall(Peff)

Effective rainfall is calculated using the USDA Soil Conservation Service method ([35]):

If P ≤ 250 mm/month:

Peff = P × (125 − 0.2 × P)/125

If P > 250 mm/month:

Peff = 125 + 0.1 × P

where: Peff = Effective rainfall [mm/month], P = Total monthly rainfall [mm/month].

IrrigationWater Supply

where:

Irrigation_Coverage = Fraction of agricultural land with irrigation infrastructure [dimensionless];

Irrigated_Area = Total irrigated area [ha];

Irrigation_Efficiency = 0.60 (typical for surface irrigation);

Water_Availability_Factor = MIN (1, Available_Water/Water_Demand).

Water stressFactor

Water_Stress = 1 – MIN (1, (Peff + Irrigation_Supply)/CWR)

StockEquation

Assumptions

The model uses a single, aggregated national water balance; spatial variations are not modeled.Groundwater resources are assumed to be negligible for large-scale irrigation, consistent with current infrastructure limitations.Irrigation efficiency is fixed at 60%, representing typical surface irrigation systems.Water allocation for non-agricultural uses is not explicitly modeled.Rice cultivation is modeled under a supplemental irrigation regime, where rainfall provides the primary water input and irrigation supplements water deficits during dry periods; full water control (paddy flooding) is not assumed, reflecting the limited extent of formal irrigation infrastructure in The Gambia.

Data Sources

Irrigation coverage: [10], approximately 1% of cultivated land.Irrigation efficiency: FAO Irrigation and Drainage Paper No. 56.

The Agricultural Production Subsystem simulates crop water requirements, calculates actual yields based on water stress, and aggregates total national food production. It is the core biophysical component of the model.

CropWater Requirement(CWR)

CWR = Σ (Kc_i × ETo × Growing_Days_i)

where: CWR = Seasonal crop water requirement [mm/season]; Kc_i = Crop coefficient for growth stage i [dimensionless]; ETo = Reference evapotranspiration [mm/day]; Growing_Days_i = Duration of growth stage i [days].

ActualYield(Ya)

The FAO 33 water-yield relationship ([27]):

Ya = Yp × (1 − Ky × (1 − ETa/ETc))

where: Ya = Actual yield [t/ha]; Yp = Potential yield [t/ha]; Ky = Yield response factor [dimensionless]; ETa = Actual crop evapotranspiration [mm]; ETc = Crop evapotranspiration under no stress (= CWR) [mm].

Implementation

Ya = Yp × (1 − Ky × Water_Stress) × Technology_Index

TotalFood Production

Total_Production = Σ (Ya_crop × Area_crop)

CropParameters

Stock andFlow Structure

StockEquations

Arable_Land (t) = Arable_Land (t − dt) − Land_Degradation × Arable_Land (t − dt) × dt

Assumptions

Crop coefficients (Kc) and yield response factors (Ky) are representative of varieties commonly grown in The Gambia.The relationship between water stress and yield loss is linear (FAO 33 methodology).Pests, diseases, and nutrient limitations are implicitly captured in the Potential Yield parameter.Land degradation occurs at a constant rate of 0.5% per year under BAU.Technology improvements increase potential yield linearly.

DataSources

Crop coefficients (Kc): FAO Irrigation and Drainage Paper No. 56 ([3]).Yield response factors (Ky): FAO Irrigation and Drainage Paper No. 66 ([27]).Baseline yields and cultivated areas: FAOSTAT database (2018-2022 average).

PopulationDynamics

Population (t) = Population (t − dt) + (Births − Deaths) × dt

Births = Population × Birth_Rate

Deaths = Population × Death_Rate

where: Birth_Rate = Base_Fertility × Fertility_Modifier/1000 [per year]; Death_Rate = Base_Mortality × Mortality_Modifier/1000 [per year]; Fertility_Modifier = f (Food_Security) decreases with food insecurity; Mortality_Modifier = f (Food_Security) increases with food insecurity.

FoodSecurity Index

Food_Security_Index = MIN (1, Total_Food_Supply/Total_Food_Demand);

Total_Food_Demand = Population × Per_Capita_Requirement;

Per_Capita_Requirement = 180 kg grain equivalent/person/year.

FarmerIncome

Farmer_Income = (Total_Production × Crop_Price − Production_Costs)/ Farm_Households

InvestmentCapacity

Investment_Capacity = MAX (0, Farmer_Income − Subsistence_Threshold) × Savings_Rate

Parameter values

Stock andFlow Structure

Assumptions

Population growth follows modified exponential growth, with fertility and mortality rates influenced by food security.Per capita food demand is constant at FAO minimum requirements (2,100 kcal/day ≈ 180 kg grain equivalent/year).Crop prices are exogenous and based on 5-year historical averages; market dynamics are not modeled.The savings rate (10%) represents the fraction of income above subsistence available for investment.

Data sources

Population data: World Bank World Development Indicators (2020).Fertility and mortality rates: World Bank WDI (2020).Crop prices: FAOSTAT producer prices (2018-2022 average).

PolicyScenario Definitions

Scenario 1. Business-as-Usual (BAU).

Scenario 2. Infrastructure investment.

Scenario 3. Crop diversification.

Assumptions

Policy parameters are implemented consistently throughout the simulation period.The model does not account for political, social, or financial barriers to policy implementation.Policy effects are additive and do not interact with each other.Infrastructure investments begin immediately in 2020 and continue at a constant rate.

ValidationApproach:The system dynamics model was validated to ensure its structural integrity, parameter credibility, and behavioral realism in relation to its intended purpose. In line with established best practices in the field of system dynamics ([28]; [5]), the validation process was multifaceted. The primary goal was not to achieve point-by-point historical replication, which is inconsistent with the model’s design as a long-term strategic tool, but rather to assess its ability to endogenously generate behaviors that are broadly consistent with historical patterns and to transparently evaluate its structural limitations.

Data Sources for Validation:Historical data for the validation period (2010-2020) were obtained from two primary sources. Data for the national population were sourced from the World Bank’s World Development Indicators (WDI) database ([40]). Time-series data for area harvested and production for the five specific crops included in the model (maize, millet, groundnut, rice, and sorghum) were obtained from the Food and Agriculture Organization’s Statistical Database (FAOSTAT) ([10]).

Behavioral Validation Results:The behavioral validation was conducted by running the model from 2010 to 2020 using the baseline parameters documented in the main appendix, without fitting or calibration to the historical data. Simulated production was calculated by multiplying the observed area harvested for each year (from FAOSTAT) by the model’s potential yield (Yp) parameters. This approach provides an objective assessment of the model’s structural performance. The results for population and total crop production are presented in Figure S1.

Figure S1.Comparison of simulated model output against observed historical data (2010-2020).

Summary of statistical performance metrics for the validation period (2010-2020).

Definition ofMetricsandInterpretation

Bias measures the systematic difference between the mean of the simulated values and the mean of the observed values, expressed as a percentage of the observed mean. A positive bias indicates the model, on average, overestimates the variable. It is calculated as:

Bias (%) = (Mean (Simulated) – Mean (Observed))/Mean (Observed) * 100

Normalized Root Mean Square Error (NRMSE) measures the magnitude of the error relative to the mean of the observed values, providing a standardized measure of the model’s predictive accuracy. It is calculated as:

NRMSE (%) = [sqrt (mean ((Simulated − Observed)²))]/Mean (Observed) * 100

The validation results highlight the model’s capabilities and limitations. The demographic module performs well, with a low bias (−5.5%), indicating that the assumed fertility and mortality rates provide a reasonable basis for long-term population projections. In contrast, the agricultural module shows a significant positive bias (+66.7%) for total production. This discrepancy is an expected outcome of the model’s design. The simulated production is based on the model’s potential yields (Yp) multiplied by observed harvested area, whereas observed production is consistently lower and more volatile. This is due to a host of factors not explicitly modeled, including intra-seasonal droughts, pest outbreaks, and constraints on access to inputs. The model’s purpose is not to simulate these high-frequency, stochastic events but to analyze the long-term impact of structural changes.