Climate-Related Risks and Agricultural Yield Assessment in the Senegalese Groundnut Basin

Abstract

Climate change and variability pose significant threats to agricultural production, particularly in regions heavily dependent on rainfed agriculture like Senegal. The problem addressed in this study revolves around the impact of climate-related risks on agricultural yields in the Senegalese Groundnut Basin as a key agricultural region. Daily rainfall, temperatures, and yield over 1991–2020 were used. The data were analyzed using multiple regression, trend analysis, and correlation approaches. The results indicate that the overall seasonal precipitation increases over time (98 mm in the north and 103 mm in the south). However, we found that the south Groundnut Basin has a much slower seasonal precipitation rate than the northern zone. Our results also show that the northern zone exhibits a more consistent and predictable growing season, with onset and offset, in contrast with the southern zone, which shows higher variability. The analysis further reveals that both the northern and southern zones are experiencing a warming trend, with the southern zone showing a more pronounced increase in maximum temperatures (+0.7 °C) than to the northern zone (+0.4 °C). Estimates from the regression analysis revealed that total seasonal precipitation and maximum temperature positively and significantly influence groundnut, millet, and maize yields in the northern and southern zones. All the other weather-related parameters have different influences depending on the zone. These findings highlight the heterogeneous nature of the study area and the significant role climatic factors play in crop yield variability in the Groundnut Basin. Understanding these influences is crucial for developing targeted agricultural strategies and climate adaptation measures to mitigate risks and enhance regional productivity. The study provides valuable insights for policymakers and farmers aiming to improve crop resilience and sustain agricultural outputs amidst changing climatic conditions.

1. Introduction

Agriculture sustains and defines our modern lives through its importance to human nutrition and the ability of complex societies. As the world’s population grows, it is crucial to have an agricultural system that can continuously provide food and other resources necessary for human existence [1]. However, many problems threaten agriculture’s stability to meet current and future human needs, not least climate change.

Climate change is expected to increase the frequency and magnitude of hydro-climatic and extreme events [2,3]. More floods and droughts are expected, impacting economies, livelihoods, and the environment, particularly in vulnerable areas, such as in some African countries [4]. West Africa is experiencing drastic impacts from climate change, as evidenced by increasing variability in rainfall and rainy season features, as well as an increase in extreme events [5]. The frequency, magnitude, and duration of adverse weather conditions are changing [6] and more drastically so in the Sahel region. The most important weather risk is water stress caused by the late onset of rain, erratic rainfall, the early cessation of rain, or prolonged drought [7]. Agricultural calendars are continually being disrupted by these climatic upheavals, adversely affecting agricultural production. These effects have been observed in Senegal and present serious threats to food security and the health status of the farming communities that depend on agriculture for their livelihood [2,8]. According to Ouikotan et al., climatic factors such as rainfall and temperature are direct inputs for production [9]. Any change and variability in these indicators are predicted to significantly affect production, causing yield losses, which, in turn, lead to food insecurity and escalating famine [10,11,12].

Senegalese agriculture is primarily rainfed, so wetter years are generally associated with higher food production, while dry years are associated with a higher risk of severe food shortages. Indeed, as the country experiences more climate risks, such phenomena have become pervasive features of the agricultural landscape and an increasingly important issue. The decrease in production among staple crops is largely attributed to the decline in total annual rainfall and increasing inter-annual variability, soil degradation, and higher demographic pressure. According to Arce et al., it has been shown that more than 40% of the variation in national crop yields can be attributed simply to variations in annual rainfall amounts [7]. Recent climate trends in the central part of the country show marked incidence of worsening climate variability, with evidence of exacerbating food insecurity [13]. Huang et al. assert that areas with unpredictable rainy season patterns or an uneven start to the rainy season and/or protracted dry spells experience a constant decrease in crop productivity [14]. As a result, climate is one of the biggest challenges in increasing or at least stabilizing food production [15,16]. Any adverse change in climate-related variables will influence crop growth and yield, input supply, and other elements of farming system management [17]. In fact, in Senegal, 60% of the farms are in the Groundnut Basin, which is the predominant area for cash crops, and about half of arable land is cropped with millet and groundnut cultivable land [15]. Most of these cultivated crops serve a dual purpose: first, as a staple crop for household consumption, and more importantly, as cash crops sold on the domestic market that also drive economic activities of numerous value chain actors.

While general climatic parameters are often studied, the impact of localized microclimates within the Senegalese Groundnut Basin has not been thoroughly investigated. The hypothesis of this study is that climate-related risks, particularly variations in rainfall and temperature, significantly affect agricultural yields in the Senegalese Groundnut Basin, with distinct impacts across the northern and southern zones. The current study seeks to contribute to the evolving literature by investigating the impact of these climatic variables at localized levels in the Groundnut Basin of Senegal. Variations in temperature, humidity, and rainfall at a more granular level could significantly affect yield, and these nuances might not be captured in broader studies. The overall objective of this study is to contribute to a better understanding of climate-related risks and their impact on yield in the Senegalese Groundnut Basin. Specifically, we seek to characterize the climate-related risks in the form of rainfall and temperature variability using satellite data (CHIRPS and ERA5) and analyze crop yield data in the Groundnut Basin.

2. Methodology

2.1. Study Area

The Groundnut Basin of Senegal, as displayed in Figure 1 below, covers the western and central administrative regions of the country, which comprises the regions of Thies, Diourbel, Fatick, and Kaolack in addition to the departments of Louga, Kebemer, and Koumpentoum based on the delineation made by the Ecological Monitoring Centre of Senegal [18].

The zone’s administrative departments are divided into two main agricultural zones: the northern and the southern. The distribution of departments in the different agricultural zones is based on studies [19] where the Groundnut Basin is split into two zones based on the rainfall gradient. With latitudes 13–15° 30 north and longitudes 13–16° 30 east, it is in the semi-arid part of the country. Due to disparity and low rainfall levels that rise from the south to the north, the climate is classified as Sahelian in the north and Sahelo-Sudanian in the south [20]. The Groundnut Basin of 49,500 km² is densely populated and subject to ecosystem degradation and depletion of land resources (soil fertility and timber resources). It has a human population of 8,640,894, representing about 47% of the national population, according to ANSD [21]. Rainfall in the Groundnut Basin’s north is often in the range of 300 to 600 mm. Sahelian agro-pastoral production systems, which rely on dry farming and customary animal keeping, dominate it. Agriculture is predominantly rainfed, with crops such as groundnut, millet, and maize being the most commonly cultivated [22]. However, in the Thies region, cultivation of horticultural crops in the form of market gardening and arboriculture are quite widespread. Rainfall in the Groundnut Basin to the south ranges from 500 to 800 mm. Other crops of economic and consumption importance are sorghum, cotton, and cowpeas.

2.2. Model Estimation

Given the nature of the dataset having a time and cross-sectional dimension, the paper adopted a model by Cohen et al., Wooldridge et al., and Belford et al. [23,24,25]. We used a pooled ordinary least squares multiple regression technique to assess the overall relationship between our outcome of interest (i.e., crop yields) and various explanatory variables, mostly agro-climatic variables (rainfall and temperature). As shown by Cohen et al. [23], the model is written in the following form:

$$Y_{ij}=a{X}_{1j}+b{X}_{2j}+\dots +z{X}_{ij}+{\epsilon }_{ij}$$

(1)

where $Y_{ij}$ is the observed variable (crop yields), $X_i$ represents the predictor variables, and a, b, and z are the estimated values of the parameters associated with climate-related variables. It must be emphasized that j, as used in the model specification, represents a given department on which the data are collected.

Table 1. Presentation of agro-climatic variables used.

Variables	Definition	Unit	Expected Sign
Agro-climatic	Total seasonal precipitation (TSP) Seasonal total precipitation in wet days (PR ≥ 1 mm)	mm	+
	Simple daily rainfall intensity index (SDII) Average daily PR intensity = annual total PR divided by the number of wet days (when total PR ≥ 1.0 mm)	mm	+
	Max 5-day precipitation (RX5 days) Maximum 5-day PR total	mm	+
	Number of very heavy rain days (R20 mm) Number of days when PR ≥ 20 mm	day	+
	Consecutive wet day (CWD) Maximum seasonal number of consecutive wet days with PR > 1 mm	day	+
	Consecutive dry days (CDDs) or length of dry days Maximum seasonal number of consecutive dry days with PR < 1 mm	day	-
	Seasonal maximum mean temperature (Tmax)	°C	-
	Seasonal minimum mean temperature (Tmin)	°C	+
	Seasonal start date (onset)	Julien day	-
	Seasonal end date (offset)	Julien day	-
Agricultural	Crop yields (maize, millet, and groundnut)	kg/ha

below shows the different agro-climatic variables that were calculated and analyzed for each zone and their prior expectations of crop yields.

A stepwise multiple regression model is used because it involves comparing several models to determine which model performed the best. When the number of predictors was small, manually comparing the models to select the most appropriate one was easy. However, when the number of independent variables increased, this approach quickly became unusable, and it was useful to use an appropriate goodness-of-fit measure. Step-by-step selection based on the Akaike Information Criterion (AIC) was one of the approaches used. With this AIC goodness-of-fit measure, the model with the lowest AIC value was chosen. The following is how Akaike et al. [26] demonstrate how to compute the AIC:

AIC = 2∗ln (RSS) + 2KAIC = 2∗ln (RSS) + 2K

(2)

where K is the number of independent variables, n is the number of observations, and RSS is the sum of squared residuals. This measure favors the accuracy of predictions and penalizes complexity. When models are compared by the AIC, the model with the lowest AIC is the preferred model.

2.3. Data Collection

The various data (climate variables and crop yields) were acquired from 1981 to 2020. However, due to gaps in the yield series collected by the Directorate of Analysis, Forecasting and Agricultural Statistics (DAPSA) of Senegal for the various crops, this study focused on 1991–2020. This study period ensured consistency in the data and allowed us to identify potential changes that have been observed over the last three decades. Moreover, the period from 1991 to 2020, which corresponds to the latest climatological normal of the World Meteorological Organization (WMO), also has the advantage of coinciding with the so-called rainfall return period in the West African Sahel after the two dry decades (1970–1990). This period seems, therefore, more consistent in climatic terms for assessing agro-climatic characteristics and their relationship to the yields of various crops considered in this study. The data were computed using MATLAB and R version 4.2.2 software.

Table 2. Description of climate data sources used in the overall analysis.

Products Name	Variable	Temporal Resolution	Spatial Resolution	Source	Time Coverage
CHIRPS	Precipitation	Daily	0.05°	CHC-UCSB	1991–2020
ERA5	Temperature	Hourly	0.31°	Climate Data Store (CDS)	1991–2020

briefly highlights the sources of climatic data used for the current analysis.

The agro-climatic characterization of the seasons in the study area was based on the discretizing of the seasons by calculating different agro-climatic variables or indicators. Various statistical parameters, such as the mean, standard deviation, and annual coefficient of variation, were calculated for the different agro-climatic variables in each zone to assess the potential variations observed over the study period.

To analyze trends in the annual series of agro-climatic variables at the level of each agricultural zone, the statistical approach based on the Mann–Kendall test (1975) was applied to the average annual data from 1991 to 2020. This test is widely used in the literature to detect trends in agro-environmental data [27].

The Pettitt test [28] was chosen for the homogeneity test. It is widely used and is suitable for determining a break in a time series. The Pettitt test is a rank-based nonparametric statistical test used to detect change points present in a data series derived from the Mann–Whitney test. Its normalization purpose is to detect a break in a time series. Based on this, the null hypothesis of the test states that there is no break in the series. For time, t, varying from 1 to N (N being the number in the series), is the following is assumed:

The series $\left(x_i\right)$, i = 1 to t, and $\left(x_i\right)$, i = t + 1 to N, belong to the same population.

Or $D_{ij}=sgn\left(X_i-X_j\right)$, with sign (x) = 1 if x > 0

0 if x = 0

−1 if x < 0

We consider the variable $u_{t,N}$ with the following:

$$u_{t,N}={\displaystyle \sum }_{i=1}^t{\displaystyle \sum }_{j=t+1}^N{D}_{ij}$$

(3)

To assess the potential relationship of climate on yields of different crops (maize, millet, and groundnut), a modeling approach based on the analysis of linear relationships between climatic parameters characterizing agricultural seasons and crop yields was applied in each agricultural zone. We normalized the variable of the dataset to have an average of 0. The following formula was used to normalize the data of the Groundnut basin (north and south):

$$I_z={\displaystyle \frac{x_i-{\overline{x}}_i}{\sigma }}$$

(4)

where $I_z$ is the normalized value of the parameter in the zone; x_i is the ith value in the dataset; x is the sample mean; and $\sigma$ is the standard deviation of the sample.

The correlation between the various agro-climatic variables and crop yields was analyzed by calculating the Pearson correlation coefficient. When a change in one variable (i.e., independent) is correlated with a proportionate change in the other (i.e., dependent), the relationship is said to be linear. Comparing the p-value of the estimated coefficient to the significance level gives an indication of whether the estimated relationship truly exists (with a certain level of confidence) or if it is purely due to chance. For most scientific research, the widely accepted significance level of 0.05 indicates that there is a 5% chance of drawing an incorrect conclusion where correlations appear but do not actually exist. This parameter measures the link between two quantitative variables. This step enabled us to assess which variables are strongly related to each other or vice versa. It is described by the following formula:

$$r={\displaystyle \frac{{\displaystyle \sum \left(x_i-\overline{x}\right)}-\left(y_i-\overline{y}\right)}{{\displaystyle \sum {\left(x_i-\overline{x}\right)}^2-{\left(y_i-\overline{y}\right)}^2}}}$$

(5)

where r = correlation coefficient.

• $x_i$ = value of the x-variable in a sample.
• $\overline{x}$ = mean of the value of the x-variable.
• $y_i$ = values of the y-variable in a sample.
• $\overline{y}$ = mean of the values of the y-variable.

3. Results and Discussion

3.1. Analyses of Spatial and Inter-Annual Rainfall Variabilities

The total rainfall index is shown on the spatial rainfall distribution map, with north-south gradient trends that are upward in the Groundnut Basin, as shown in Figure 2. The Groundnut Basin recorded fairly variable rainfall patterns over the 1991–2020 period, with cumulative rainfall ranging from 300 mm in the extreme north to 800 mm in the extreme south.

However, Figure 3, on both sides, indicates a positive trend in total seasonal precipitation, with the average ranging from 433.9 mm to 598.8 mm in the northern and southern zones, respectively. This suggests that the overall seasonal precipitation increases over time for both the southern and northern zones. But seasonal precipitation increases at a slower rate in the southern zone compared to the northern. This increase in rainfall is consistent with the findings of [25,26], which conclude that cumulative rainfall in the Sahel has increased since the 2000s.

The shift from the less rainy period to the rainier one is highlighted by the Pettit test. It shows a break in homogeneity in 1998 in both zones, with a mean rainfall (1991–1998) of 362 mm and mean rainfall (1999–2020) of 460 mm in the northern zone and mean rainfall (1991–1998) of 501 mm and mean rainfall (1999–2020) of 604 mm in the southern zone (cf. Figure A1 in Appendix A). A comparison of the two averages using Student’s t-test revealed an increase in rainfall of 98 mm in the north and 103 mm in the south for the period 1999 to 2020, meaning that the study area had two major periods: lower rainfall from 1991 to 1998 and a significant increase in rainfall from 1999 to 2020. These observations are similar to those of [29,30,31] who showed an improvement in rainfall from 1999 onwards based on a study carried out on the 1961–2010 series. Similarly, Bodian also identified 1999 as the year in which the rains began to return to Senegal [32].

Both northern and southern zones show an upward trend in RX5 days, suggesting an increase in the intensity of rainfall events over time. Maximum cumulative rainfall over 5 days fluctuated between 20 and 120 mm per year in the different zones. In the northern zone, RX5 days averaged 76.9 mm (SD of 14.6 mm) and 79.4 mm in the southern zone (SD of 11.9 mm) (cf.

Table A1. Mann–Kendall results for climatic variables.

Groundnut Basin		R5 × Days	TSP	SDII	CDD	CWD	R20 mm	Offset	Onset	Tmax	Tmin
North	Mean	76.9	433.9	11.4	22	5	4	271	199	32.2	21.6
	Standard deviation (SD)	14.6	88.3	1.5	5	1.3	2	6.5	12.5	0.44	0.36
	Tau of Mann–Kendall	0.12	0.28	0.17	0.13	0.07	0.21	0.09	−0.10	0.34	0.09
	p-value	0.33	0.02	0.18	0.3	0.56	0.10	0.46	0.42	0.008	0.45
South	Mean	79.4	568.8	11.7	14	6	6	278	184	33.12	22.7
	Standard deviation (SD)	11.9	91.2	1.19	4	1.08	2	6	9.6	0.48	0.32
	Tau of Mann–Kendall	0.08	0.17	0.11	0.11	0.13	0.14	−0.04	0.04	0.51	0.11
	p-value	0.49	0.18	0.39	0.37	0.32	0.26	0.72	0.72	6.432 × 10−5	0.37

in Appendix A). The slopes of the linear regression lines showed that the rate of increase in both RX5 days and TSP is higher in the northern zone compared to the southern zone. This means that the northern zone experiences a more pronounced increase in both the intensity of heavy rainfall events and total seasonal precipitation.

3.2. Variability of Standardized Precipitation Index (SPI)

Figure 4 shows the evolution of the standardized precipitation index (SPI). The SPI values indicate alternating wet and dry years in both the north and south of the Groundnut Basin. Over the considered 30 years, there have been 17 dry years (57%) in the north compared with 15 dry years (50%) in the south, with varying degrees of drought from one year to the next. We notice that significant periods of drought (SPI < 0) can be observed, particularly during 1991–1993 and 1995–1997 for the north and during 1994–1997 and 2013–2014 for the south. There were only 13 wet years (43%) in the north, with varying degrees of humidity from one year to the next. The remarkable surplus year for the Groundnut Basin corresponds to 1999 in the south and 2012 in the north of the Groundnut Basin, with indices above (1.88 and 2.17, respectively). But, in general, the Groundnut Basin (northern and southern zones) exhibits a pattern of wetter conditions in the 2000s and early 2010s. These findings were confirmed by Mbaye et al., who affirmed that in Senegal, the earlier 2000s were marked by a “return of wet conditions” with a multitude of positive rainfall anomalies [33].

3.3. Evolution of the Simple Daily Intensity Index (SDII)

Figure 5 shows that the variation in SDII is between 8 and 15 mm/day in the Groundnut Basin. In the northern and southern zones, respectively, the average SDII is 11.4 mm/day (SD of 1.5 mm/day) and 11.7 mm (SD of 1.19 mm/day) (cf. Table A1 in Appendix A). Both zones show a positive trend in SDII, indicating increasing average daily rainfall intensity over time. However, the Mann–Kendall test shows that this increasing trend is not significant. This finding is similar to Mbaye et al., who analyzed past extreme precipitation and evapotranspiration indices over Sub-Saharan countries and noticed no significant trend in rainfall intensity when considering the whole period of 1981–2019 [33]. Both zones experience similar fluctuations in SDII, with peaks and troughs occurring around the same years. The variations in SDII are more extreme in the northern zone compared to the southern zone, indicating higher variability in daily rainfall intensity in the north.

3.4. Evolution of Occurrences of Rainy and Dry Days

Figure 6 shows that R20 mm varies between 1 and 10 days in the season depending on the zone, with an average of 4 and 6 days, respectively, in the northern and southern zones. Concerning consecutive dry days, they varied between 5 and 35 days in the year. In the north, there was an average of 22 CDD (SD of 5 days) compared with 14 days (SD of 4 days) in the south (cf. Table A1 in Appendix A). The average number of consecutive wet days was 5 to 6 in the north and south, respectively. The overall statistical trend in the occurrence of wet and dry days increased over time, reflecting a trend toward an increase in these parameters. The northern zone is characterized by longer periods of consecutive dry days (CDDs) compared to the southern zone. Both zones exhibit similar patterns for consecutive wet days (CWDs) and days with heavy rainfall (R20 mm), suggesting that these factors have a uniform impact on the agricultural conditions across both zones. These findings are similar to Didi et al., who affirmed an increase in precipitation for Senegal’s results, including increase in precipitation frequency as well as the maximum length of wet spells (an increase in CWD) [34]. The consistency in wet days and heavy rainfall days between the zones indicates that the difference in crop yields is more likely influenced by the length of dry spells rather than the frequency of wet days or heavy rainfall events.

3.5. Evolution of Start Dates (Onset) and End Dates (Offset) of Seasons

Figure 7 and Figure 8 provide additional information on the trends in the growing season for the north and south Groundnut Basin. Figure 7 illustrates the trends for the onset and offset of the growing season, while Figure 8 shows the cumulative probability of these onsets and offsets. The results in Figure 7 indicate that in both the north and southern zones, the onset and offset of the rainy season exhibit weak and insignificant trends based on the p-values > 0.05 (cf. Table A1 in Appendix A). This suggests that there has not been a clear shift in the timing of the rainy season in recent years, meaning that the start and end of the season have remained relatively stable. Overall, the seasons in the study area start between the first ten days of June and the last ten days of July, depending on the agricultural zone. As for the end of the seasons, they fluctuate between September and October, depending on the agricultural zone. In the northern zone, the seasons begin on average around the end of June each year compared to the end of May and the beginning of June in the southern zone. Similarly, the seasons end between late September and early October in the northern zone and late October in the southern zone. According to Balliet et al., a false start to the growing season or even the loss of a whole crop year might result from a drought that strikes at the beginning or middle of the season [35].

The cumulative probability curves in Figure 8 illustrate the differences in the growing season dynamics between the northern and southern zones. The northern zone’s curve shows a more gradual slope, while the southern zone’s curve shows more steepness, indicating greater variability. The southern zone has greater variability in terms of the onset and offset of rain compared to the north. The northern zone exhibits more consistent patterns for both the onset and offset of the growing season, as indicated by the gradual slopes of cumulative probability curves. The southern zone shows more variability in both onset and offset, as indicated by the steeper slopes of the curves.

3.6. Evolution of Temperatures

The average maximum temperatures (T_max) varied between 31 and 34 °C depending on the zone (as shown in Figure 9). The average T_max in the northern zone was 32.2 °C, compared with 33.1 °C in the southern zone. The average minimum temperature (T_min) varied between 21 and 23.5 °C. The increase in Tmax is statistically significant in both zones (North (0.008) and South (6.432 × 10⁻⁵)), with the South showing a more pronounced trend. T_min does not show any statistically significant trend in either zone (p-value > 0.05), suggesting relative stability in minimum temperatures. Despite the fluctuations in yearly data points, the trend lines of T_max indicate a consistent rise in temperature, suggesting potential long-term climate changes in these regions.

The analysis suggests that both the northern and southern zones are experiencing a warming trend. According to Schlenker et al., the continuing temperature rise has a greater impact on the inter-annual variations in yield [36]. The potential for decreased yields is due to evaporation from water bodies, evapotranspiration, and soil drying up in both locations; these hazards can also emphasize heat stress, fail to fertilize, require more water, and shorten the cycle [37].

3.7. Inter-Annual Variation in Crop Yields in the Groundnut Basin (1991–2020 Period)

Figure 10 shows that maize yields fluctuated between 400 and 2000 kg/ha, with an average of 982 and 1153 kg/ha in the northern and southern zones, respectively. Groundnut yields fluctuated between 200 and 1500 kg/ha. On average, yields were 591 and 952 kg/ha in the northern and southern zones, respectively. Finally, millet yields fluctuated between less than 100 kg and 1300 kg/ha, with averages of 489 and 813 kg/ha in the northern and southern zones, respectively.

Table 3. Mann–Kendall results.

Groundnut Basin		Maize	Groundnut	Millet
North	Mean	982	591	489.3
	Standard error (SE)	52.5	42.4	30.8
	Tau Mann–Kendall	0.51	0.31	0.32
	p-value	6.432 × 10−5	0.01	0.01
South	Mean	1153	952	813.2
	Standard error (SE)	41.5	44.6	30.4
	Tau Mann–Kendall	0.08	0.22	0.33
	p-value	0.49	0.08	0.09

shows a strong positive trend for maize in the northern zone (0.51) with a highly significant p-value < 0.001, indicating that the increasing yield trend is not due to chance. The southern zone shows a weak positive and non-significant trend (0.08), indicating no clear trend. For groundnut, the northern zone has a moderate positive and significant trend (0.31) with a significant p-value (0.01). The southern zone shows a weak positive and significant trend (0.22). However, for millet, both zones exhibit moderate positive trends (north: 0.32, south: 0.33) with a significant p-value in the north (0.01) and a more precise estimate with a lower (SE 30) in both zones. The trends are generally significant in the northern zone and nearly significant in the south.

Maize in the northern zone performs significantly better in terms of the mean yield but has higher variability. Concerning groundnut and millet, the southern zone outperforms the northern zone in terms of mean yields for both crops, though variability is comparable. Over time, maize in the northern zone shows a significant upward trend, suggesting improvements or favorable conditions over time. The southern zone shows no significant trend. Both zones show positive trends for maize, groundnut, and millet, but these trends are more pronounced and statistically significant in the northern zone. The data suggest that agricultural yields for maize, millet, and groundnut have generally improved over the past three decades in both the north and southern zones, with some variations in the rate of improvement between the zones.

Figure 11 indicates that, for all three crops (maize, millet, and groundnut), the median yield is higher in the southern zone compared to the north. However, the southern zone also has more outliers, indicating greater variability. The southern zone has a higher and less variable maize yield, indicating more stable production compared to the northern zone, with more outliers. For millet, the southern zone also shows a higher median yield but with some outliers above 1000 kg/ha, suggesting exceptionally high production in certain cases. Concerning groundnut, the southern zone again shows a higher median yield with some outliers below 500 kg/ha, suggesting exceptionally low production. The interquartile range (IQR) for all crops is wider in the southern zone compared to the northern zone, indicating more variability in southern zone yields. The southern zone has more outliers for millet and peanut, indicating occasional extreme yield values. The northern zone has fewer outliers, suggesting more stable yield patterns.

The boxplots reinforce the observation that millet and groundnut crop yields are generally higher and less stable in the southern zone compared to the northern zone. This could be due to better agricultural practices, climatic conditions, soil fertility, or other factors that favor crop production in the southern zone. The higher variability and frequent outliers in the southern zone suggest that yields are less predictable and more susceptible to fluctuations.

3.7.1. Analysis of Relationships between Climatic Variables and Crop Yields

Figure 12 and Figure 13 are correlation matrices displaying the relationship between agro-climatic variables and crop yields in the northern and southern zones, respectively.

In the northern zone, the values in Figure 12 show the Pearson correlation coefficient between pairs of variables, the number of heavy precipitation days (R20 mm), and daily rainfall intensity (SDII), indicating a strong positive correlation (0.92). The maximum (T_max) and minimum (T_min) temperatures have a high positive correlation (0.59). The heavy precipitation days (R20 mm) and total seasonal precipitation TSP indicate a strong positive correlation (0.8). The rainfall intensity (SDII) and consecutive wet days (R×5 days) show a strong correlation (0.78).

Concerning precipitation metrics, TSP, R20 mm, and SDII generally show positive correlations with crop yields, indicating the importance of adequate rainfall for crop production. Maize does not show strong correlations with any specific variable in the provided matrix, suggesting it may be influenced by a combination of factors rather than any single variable. However, groundnut yields are positively correlated with T_min and T_max, implying that warmer temperatures are beneficial. Millets yields are positively correlated with SDII, suggesting consistent and intense rainfall benefits millet yield. This is also correlated with T_max, indicating a preference for higher maximum temperatures. This result is in line with [38], who found that rainfall indicators significantly impact on crop yields in the West African Sahel.

In the southern zone, R20 mm and SDII (0.9) have a very strong positive correlation between the number of heavy precipitation days and daily rainfall intensity. The maximum (T_max) and minimum (T_min) temperatures have a moderate positive correlation (0.56). The heavy precipitation days (R20 mm) and total seasonal precipitation (TSP) correlate strongly and positively (0.88). The total seasonal precipitation (TSP) and number of rainy days (RX5 days) have a moderate correlation (0.42).

Both temperatures (T_max and T_min) show moderate positive correlations with crop yields, suggesting that higher temperatures might favor yield increases within the observed range. Precipitation metrics (TSP, R20 mm, SDII) generally show strong positive correlations with each other, indicating the importance of adequate rainfall for crop production. For maize, it shows a moderate positive correlation with T_min, indicating that minimum temperatures may be beneficial. Groundnut has a positive correlation with T_min, suggesting that minimal temperatures are favorable. Millet is also positively correlated with SDII and T_max, indicating that consistent and intense rainfall and higher maximum temperatures benefit millet yield.

The moderate positive correlations between crop yields and T_min, as well as SDII, indicate that both temperature and rainfall intensity are critical factors for successful crop yields in the southern zone.

A comparative analysis between the northern and southern zones (Figure 12 and Figure 13) reveals that while both zones share similar patterns in terms of relationships between agro-climatic variables and crop yields, there are notable regional differences. The southern zone tends to have more moderate correlations with Tmin, impacting maize yields, while the northern zone shows stronger inter-correlations among precipitation metrics, and TSP has a notable impact on millet yield. In fact, maize yield in the southern zone shows a moderate positive correlation with Tmin, whereas no strong correlations are observed in the northern zone. Millet yields in both zones also show a positive correlation with R20 mm and SDII, emphasizing the importance of rainfall intensity. However, the onset and offset of the rainy season in both zones show weak correlations with crop yields, indicating that the timing of the rainy season may not be as critical as the total precipitation and its distribution. The northern zone exhibits a stronger positive correlation between CWD and R×5 days compared to the southern zone. The southern zone shows a moderate correlation between maize yield and Tmin, which is not seen in the northern zone. The relationship between total seasonal precipitation and maize in the northern zone shows that there is a positive relationship, which means an increase in precipitation will lead to an increase in maize yield, as can be seen in the scatterplot in Figure 14. Thus, to reinforce the above correlation figures between the northern and southern zones, scatterplots for some yields and selected climate variables with a regression line are illustrated in Figure 14 below.

However, it should be pointed out that climatic parameters alone are not sufficient to determine crop yields [39]. According to Noufe et al. [40], the guarantee of good yield also depends on conditions linked to the proper coordination of cultivation operations (tillage, sowing, weeding, fertilization, use of improved varieties, etc.). They have shown that knowledge of rainfall patterns alone is not enough to explain the yield of a crop such as rainfed maize since a low yield results from either insufficient or excess water. Other parameters, such as soil fertility, seed quality, disease, and pests, also need to be taken into account, all of which play a decisive role [41].

3.7.2. Analysis of the Impact of Climate on Yields

Table 4. Statistical relationship between climate and crop yield.

	Northern Zone			Southern Zone
	Groundnut	Millet	Maize	Groundnut	Millet	Maize
TSP	0.941 *** (0.251)	0.655 *** (0.170)	-	0.329 (0.207)	0.422 ** (0.153)	0.857 *** (0.186)
RX5 days	0.381 * (0.227)	-	0.482 (0.299)	0.292 (0.189)	-	-
CWD	-	-	−0.272 (0.204)		-	−0.224 (0.156)
CDD	0.420 *** (0.158)	-	-	0.386 (0.243)	0.501 *** (0.189)	-
SDII	-	-	−1.253 ** (0.549)	-	-	-
R20 mm	−0.647 ** (0.279)	-	0.999 * (0.504)	-	0.273 * (0.154)	-
Onset	-	-	-	−0.436 * (0.235)	−0.465 ** (0.184)	0.385 ** (0.168)
Offset	-	-	-	-	0.180 (0.135)	−0.381 ** (0.164)
Tmx	0.499 *** (0.176)	0.466 ** (0.193)	-	0.398 * (0.208)	0.601 *** (0.165)	0.318 ** (0.146)
Tmn	−0.369 * (0.199)	−0.444 ** (0.210)	-	−0.269 (0.207)	−0.222 (0.164)	-
R-squared	0.563	0.404	0.219	0.388	0.642	0.513
Adjusted R-squared	0.449	0.335	0.094	0.228	0.529	0.411
p-value	0.002	0.003	0.169	0.058	0.001	0.003
Number of parameters	6	3	4	6	7	5

Significance levels are indicated as follows: * p < 0.1, ** p < 0.05, *** p < 0.01.

below compares the statistical performance of models predicting crop yields for groundnut, millet, and maize between the northern and the southern zones of the Groundnut Basin, focusing on the implications of climate variables on agricultural production. Precipitation (TSP) has a positive and significant effect on all crop yields, except for maize, which is positive but statistically insignificant in the northern zone. The maximum temperature (Tmx) is also found to positively and significantly impact all crop yields in both zones. On the contrary, minimum temperature (Tmn) negatively impacts yields in both zones where it is considered. Days with heavy rainfall (R20 mm) have mixed effects: negative for groundnut in the north, positive for maize in the north, and positive for millet in the south. The start and end of the rainy season (onset and offset) significantly influences yield in the southern zone, especially for millet and maize. In the northern zone, the groundnut model demonstrates a better ability to explain yield based on climate variables, as evidenced by a relatively high R-squared value of 0.56 and a significant p-value (0.002). Total seasonal precipitation (TSP) and maximum temperature (Tmx) have strong and significant positive impacts on yield, indicating that both rainfall and temperature play crucial roles in influencing groundnut production. The results from the model fitted to millet also perform adequately in terms of its explanatory ability, but that fitted to maize in the northern zone is notably weak explanatory power, with low R-squared. In the southern zone, the models show generally stronger explanatory power, particularly for millet, which has the highest R-squared value (0.64) and highest statistical significance (p-value 0.001), suggesting a robust relationship between climate variables and crop yield. While the model fitted to maize has better explanatory value in the north, it shows somewhat moderate predictive power, and the groundnut model is less statistically significant than the north. Overall, the effectiveness of these models highlights the critical role of climate factors, such as temperature, rainfall patterns, and seasonal shifts, in agricultural productivity. These results highlight the complexity of climate–yield relationships and suggest that different crops and regions require tailored approaches to improve agricultural productivity.

4. Conclusions and Policy Implications

The study of the characterization of climatic risks and the impact of climatic parameters on agricultural yields is still a topical issue in the Sahelian regions. We used climate data to investigate the effects of climate-related variables on yield in the Groundnut Basin of Senegal. Regarding changes in rainfall, both the northern and southern zones showed an increasing trend in TSP, with the northern zone experiencing a more pronounced increase. This suggests a general rise in seasonal rainfall over the past three decades. The Pettitt test has enabled the identification of a break that occurred in the Groundnut Basin in 1999 for both zones. The SPI analysis reveals alternating wet and dry periods in both zones, with increased wet conditions in the 2000s and early 2010s and significant droughts in the early 1990s, late 1990s, early 2000s, and late 2010s. The variability shown in SPI, with alternating periods of drought and wet conditions, indicates a high level of climate variability. This unpredictability of climate poses a significant risk to agricultural planning and crop yields. Both zones exhibit a slight increase in RX5 days, indicating more intense rainfall events within short periods. The northern zone shows a steeper increase compared to the south. There is a slight increase in SDII in both zones, with the northern zone experiencing a more pronounced increase. This indicates a trend toward more intense daily rainfall. Both the north and southern zones exhibit an upward trend in maximum temperatures, indicating a general warming pattern over the study period.

The analysis of climatic trends in the Senegalese Groundnut Basin reveals a complex interplay of increasing rainfall intensity, significant variability, and periods of extreme drought and wet conditions. These findings suggest that yields are correlated with rainfall variability, uneven spatiotemporal distribution, and the frequency of wet and dry spells. These climate risks pose substantial challenges to groundnut, millet, and maize yields, necessitating comprehensive adaptation strategies to ensure agricultural sustainability and resilience in changing climate conditions. Implementing effective water management, erosion control, and resilient agricultural practices will be critical in mitigating the impacts of these climatic trends on agricultural production in the Groundnut Basin. Policymakers should prioritize the development of climate-resilient agricultural systems, including the promotion of drought-tolerant crop varieties, improved water management, and soil conservation techniques. Additionally, investment in climate-smart agricultural infrastructure, such as rainwater harvesting systems and irrigation, is crucial to enhance farmers’ ability to cope with increasingly unpredictable rainfall patterns. These findings also underscore the importance of enhancing farmers’ access to climate information services, enabling them to make informed decisions and adjust farming practices accordingly. By integrating these insights into agricultural policy, decision-makers can mitigate the risks posed by climate variability and promote resilience in farming systems. This will help ensure that farmers in the Groundnut Basin can maintain and even improve yields despite changing climate conditions.

Research should also explore how these climate changes affect the broader socio-economic fabric of the Groundnut Basin, including farmer livelihoods, food security, and regional trade.