Hydroclimatic Trends and Land Use Changes in the Continental Part of the Gambia River Basin: Implications for Water Resources

Abstract

Hydrological processes in river systems are changing due to climate variability and human activities, making it crucial to understand and quantify these changes for effective water resource management. This study examines long-term trends in hydroclimate variables (1990–2022) and land use/land cover (LULC) changes (1988, 2002, and 2022) within the Continental Reach of the Gambia River Basin (CGRB). Trend analyses of the Standardized Precipitation-Evapotranspiration Index (SPEI) at 12-month and 24-month scales, along with river discharge at the Simenti station, reveal a shift from dry conditions to wetter phases post-2008, marked by significant increases in rainfall and discharge variability. LULC analysis revealed significant transformations in the basin. LULC analysis highlights significant transformations within the basin. Forest and savanna areas decreased by 20.57 and 4.48%, respectively, between 1988 and 2002, largely due to human activities such as agricultural expansion and deforestation for charcoal production. Post-2002, forest cover recovered from 32.36 to 36.27%, coinciding with the wetter conditions after 2008, suggesting that climatic shifts promoted vegetation regrowth. Spatial analysis further highlights an increase in bowe and steppe areas, especially in the north, indicating land degradation linked to human land use practices. Bowe areas, marked by impermeable laterite outcrops, and steppe areas with sparse herbaceous cover result from overgrazing and soil degradation, exacerbated by the region’s drier phases. A notable decrease in burned areas from 2.03 to 0.23% suggests improvements in fire management practices, reducing fire frequency, which is also supported by wetter conditions post-2008. Agricultural land and bare soils expanded by 14%, from 2.77 to 3.07%, primarily in the northern and central regions, likely driven by both population pressures and climatic shifts. Correlations between precipitation and land cover changes indicate that wetter conditions facilitated forest regrowth, while drier conditions exacerbated land degradation, with human activities such as deforestation and agricultural expansion potentially amplifying the impact of climatic shifts. These results demonstrate that while climatic shifts played a role in driving vegetation recovery, human activities were key in shaping land use patterns, impacting both precipitation and stream discharge, particularly due to agricultural practices and land degradation.

1. Introduction

The Sahel region, often considered the epicenter of global environmental change, has faced disruptions like demographic shifts, intense anthropogenic activity, and severe droughts in the 1970s and 1980s [1,2,3]. These environmental changes have led to land degradation, threatening vegetation growth and food security. It is a paramount example of how land use and land cover change (LULCC) affect climate, hydrology, biodiversity, and land sustainability [4]. Accurate, up-to-date information on LULCC is essential to assess environmental impacts, as it is a major driver of change across spatial and temporal scales [5,6,7,8,9].

Rainfall variability and land use changes in West Africa have significantly affected hydrological and ecological systems. Deforestation, cropland expansion, and urbanization have replaced natural vegetation and caused soil degradation [10,11,12]. In the Black Volta Basin, rainfall trends from 1976 to 2011 showed mixed but statistically insignificant changes, while land use changes between 2000 and 2013 saw a 44.5% decline in grasslands and increases in urban areas, forests, and agriculture. These changes resulted in a 6% rise in dry season discharge, 1% in wet season discharge, and a 27% increase in surface runoff, while groundwater contributions fell by 6% [13]. Similar trends in the Volta, Mono, and Sassandra basins, driven by vegetation loss and reservoir construction, increased discharge variability [14]. In the Senegal River basin, rainfall and flow deficits with declining vegetation worsened water vulnerability [15], and in the Somone River basin, reduced precipitation and higher temperatures accompanied land cover changes [16]. These studies emphasize the complex relationships between climate variability, land use changes, and hydrology, underscoring the need for sustainable management.

Evaluating the spatial distribution and temporal trends of precipitation is challenging due to its complex, non-linear characteristics. However, this is crucial for water resources management in a basin. In West Africa, rainfall is influenced by the south–westerly monsoon and the dry northeasterly trade winds [17]. Most people in the region rely on small-scale, rain-fed agriculture, making them vulnerable to changes in water availability [18]. While rainfall variability may be linked to global phenomena, recognizing regional patterns is key for decision-making.

The growing recognition of the impact of Land Use and Land Cover Change (LULC) on river discharge has prompted this study in the CGRB. While previous research has examined hydroclimatic trends, few have focused on the spatio-temporal dynamics of LULC in the basin. This study aims to fill this gap by analyzing recent hydroclimatic trends and mapping LULC changes, specifically by combining historical hydroclimatic data with remote sensing imagery. The research seeks to explore the interactions between climatic variability, human activities, and land use changes, particularly by mapping spatio-temporal variations in LULC types and quantifying the extent and rate of LULC transformations. It is anticipated that changes in land use, such as deforestation and agricultural expansion, will amplify the effects of climatic variability on hydrological systems, leading to modified rainfall patterns and hydrological processes. These interactions are expected to exhibit spatial variability, with more pronounced impacts in regions undergoing significant land cover changes. This research is intended to provide valuable insights for land use planning, management, and sustainable water resource strategies, thereby aiding in the mitigation of adverse effects arising from land use changes and climate variability.

By combining historical hydroclimatic data with remote sensing imagery, the research seeks to explore the interactions between climatic variability, human activities, and land use changes. It is anticipated that changes in land use, such as deforestation and agricultural expansion, will amplify the effects of climatic variability on hydrological systems, resulting in modified rainfall patterns, reduced water availability, and increased surface runoff. These interactions are expected to exhibit spatial variability, with more pronounced impacts in regions undergoing significant land cover changes. This research is intended to provide valuable insights for land use planning, management, and sustainable water resource strategies, thereby aiding in the mitigation of adverse effects arising from land use changes and climate variability.

2. Data and Methods

2.1. Study Area

The Gambia River Basin spans approximately 77,000 km² and extends, in latitude, from 11°22′ N in Fouta-Djalon to 14°40′ N in southeastern Ferlo; in longitude, from 11°13′ W in Fouta-Djalon to 16°42′ W in Banjul, where the river meets the Atlantic Ocean (Figure 1). Originating in the Fouta Djalon Highlands in Guinea at an altitude of 1125 m, the river crosses three countries: Guinea, which makes up 15.25% of the basin area; Senegal (72.30%); and Gambia (12.45%) [19]. The basin is classified into two sections based on the river’s regime: the upstream basin, extending to the Gouloumbou station in Senegal near the Gambian border, known as the continental basin, and the maritime reach, located downstream of this station. The maritime reach represents the section where the Gambia River is at sea level. The river flows through a tropical climate with a long dry season (November to May) and a short rainy season (June to October). Most of the basin falls within the Sudano-Guinean climatic zone, with rainfall varying between 600 mm in the north of the basin and 1600 mm in the south [20]. The southern part has dense forests, transitioning to savannas and shrub steppes in the north. Human activities such as deforestation, settlement expansion, and population growth exacerbate vegetation loss, soil degradation, and water retention issues, contributing to increased erosion and low-flow periods, particularly in the northern, drier regions [21].

2.2. Data and Methods

2.2.1. Data Sources

Precipitation Data

Precipitation data for this study is sourced from the Climate Hazard Group Infrared Precipitation with Station Data (CHIRPS) version 2 [22], covering the period from 1990 to 2022. The CHIRPS dataset spans from 1981 to 2022 and provides a quasi-global rainfall record with a spatial resolution of 0.05° × 0.05°. By combining satellite imagery with in-situ station data, CHIRPS creates gridded rainfall time series, ideal for trend analysis and drought monitoring. Due to the absence of extensive ground-based data, satellite estimates like CHIRPS are often used as alternatives or supplements. The dataset has been validated against daily rain gauges in the Sahel and Guinea Coast, showing good correlation with seasonal precipitation trends. Data is available at https://data.chc.ucsb.edu/products/CHIRPS-2.0/ (accessed on 23 January 2023).

River Discharge

The discharge data used in this work is obtained from the Directorate of Water Resources Management and Planning (DGPRE) at a daily scale for the period 1990 to 2022, measured at the Simenti station on the Gambia River Basin. The Gouloumbou station could not be used due to tidal influence and because discharge data is only available during the rainy season. This limitation makes the station unsuitable for trend and drought analyses, as it does not provide a complete year-round dataset necessary for accurate assessment of hydrological trends or drought conditions.

2.2.2. Methods

Drought Index

The Standardized Precipitation Evapotranspiration Index (SPEI) is used to assess drought conditions in the study area. SPEI combines precipitation and evapotranspiration, making it more effective in capturing the impact of temperature on water demand [23]. It is particularly useful for monitoring drought patterns in regions where temperatures have increased more than the global average, such as West Africa [24]. SPEI detects changes in both precipitation and temperature variability, offering a comprehensive view of drought severity. The index is calculated on a 12- and 24-month scale using the SPEI package in R.

The potential evapotranspiration (PET) was estimated using the Hargreaves–Samani method [25], which relies on daily minimum and maximum temperatures and latitude. This method is widely recommended for data-scarce regions, as it requires fewer input variables compared with physically based methods like Penman–Monteith, which depend on solar radiation, wind speed, and humidity. Beyond representing atmospheric mean conditions through temperature, the Hargreaves–Samani model indirectly reflects land surface properties, such as soil moisture and surface albedo, through its incorporation of diurnal temperature range, making it more responsive to surface variability. Furthermore, global correlation analyses have confirmed that the Hargreaves–Samani model performs consistently well across diverse climatic regions, including arid, semi-arid, temperate, cold, and polar zones, underscoring the critical role of temperature in PET estimation [26,27,28,29].

Given the reliance on CHIRPS precipitation and ERA5 temperature data to estimate PET using the Hargreaves–Samani method, some uncertainty in SPEI values remains, especially in the absence of solar radiation or wind speed inputs. Future studies should consider sensitivity analysis using alternative PET models (e.g., Penman–Monteith or Thornthwaite) and incorporate satellite-derived variables where possible. Where ground-based evaporation data exist, validation of PET estimates can further strengthen drought assessments. These improvements would enhance the accuracy and confidence in interpreting hydroclimatic trends in data-scarce regions such as the Gambia River Basin.

In addition to SPEI, the Standardized Streamflow Index (SSI) was used to evaluate hydrological drought conditions in the study area. The SSI was calculated from annual mean discharge data, following a process similar to the calculation of the Standardized Precipitation Index, but using discharge data instead of precipitation [30]. This index helps monitor the impact of hydrological conditions on drought severity, offering insights into changes in streamflow over time.

Trend Analyses

Trend analysis employs parametric and non-parametric methods to identify changes in hydroclimatic data. While parametric methods assume normal distribution, non-parametric methods like the Mann–Kendall test are less sensitive to data breaks and do not require normality [31,32]. The Mann–Kendall test detects trends without assuming normality but requires data independence. To address autocorrelation, the modified Mann–Kendall test [31,33] was used in this study, along with Sen’s slope estimator from the modifiedmmk package in R, with significance tested at the 5% level.

Breakpoint Analyses

The objective of breakpoint analysis is to identify the number and location of stepwise shifts in a dataset. This study used the Pettit test [34], a non-parametric method in R, to detect breaks in hydroclimatic time series. The test does not require prior knowledge of data distribution, making it suitable for identifying shifts in the dataset. The rank-based Mann–Whitney test tests for departures from homogeneity, with the null hypothesis assuming independent and identically distributed annual values. The alternative hypothesis suggests a step-wise shift in the mean [35]. The Pettit test is less sensitive to outliers and does not rely on normality, offering an advantage over other tests.

Data Acquisition and Preparation for Land Cover Maps

Land use changes in the CGRB were analyzed using multi-date Landsat satellite images (

Table 1. The Landsat satellite images used.

Date	Acquisition Date	Satellite	Path/Row	Spatial Resolution
1988	1988/12/03 1988/12/10 1988/12/10 1988/12/10 1988/11/15 1988/12/01	Landsat 5	201/052 202/050 202/051 202/052 203/050 203/051	30 m
2002	2002/12/18 2002/12/25 2002/12/25 2022/12/25 2002/12/16 2003/01/01	Landsat 7	201/052 202/050 202/051 202/052 203/050 203/051	30 m
2022	2022/12/25 2022/12/24 2022/12/24 2022/12/24 2022/12/23 2022/12/23	Landsat 9 Landsat 8 Landsat 8 Landsat 8 Landsat 9 Landsat 9	201/052 202/050 202/051 202/052 203/050 203/051	30 m

). Landsat provides free 30 m resolution images, accessible at http://earthexplorer.usgs.gov/. Three dates—1988, 2002, and 2022—were selected based on image availability.

Satellite images were acquired from four sensors: Landsat 5 (TM), 7 (ETM+), 8 (OLI-1), and 9 (OLI-2). For each image, the multispectral bands (Blue, Green, Red, NIR, SWIR1, SWIR2) with a 30 m resolution were selected. A total of six images were needed to cover the entire basin. The images were collected between late November and early January, a period suitable for analysis due to reduced photosynthetic activity and fewer clouds at the end of the rainy season [36]. To account for radiometric differences, the images were obtained within a short time frame and processed using TerrSet software, version 19.

Processing Method

The image processing involved four main steps. First, geometric and radiometric corrections were applied. Geometric correction aligned the images to have consistent rows and columns, enabling integration of images from different sensors and dates. Radiometric correction addressed differences in reflectance values [36,37].

The second step involved creating a mosaic of six images covering the basin and generating a false-color infrared composite based on vegetation’s reflection of near-infrared radiation [38]. This output enabled photo-interpretation to identify land use categories: water, forest (dense, gallery, and woodland), savanna (wooded, tree, and shrubby), bowe and steppe, agricultural areas and bare soils, and burn. Given the sparse presence of built-up areas relative to the overall basin area, these areas were grouped together with agricultural and bare soil areas in the analysis to improve detection. The vegetation classification follows the Yangambi system, adapted for West Africa [39,40,41]. The “burn” category includes areas affected by fires at the time of image acquisition.

The third processing step involved unsupervised hierarchical image classification [36,37,42,43]. This method groups pixels into spectral classes based on their signatures, relying on thematic classes of surfaces being mapped. The process includes initial classification into 10 classes using the k-nearest neighbors (k-NNs) algorithm across all bands from blue to mid-infrared, followed by interpreting the spectral signatures, reclassifying them per selected nomenclature, creating thematic masks, performing hierarchical classification on the masks, and reclassifying the results. The k-NN algorithm [44] involves selecting class numbers, locating centroid numbers, and repositioning them through successive iterations until optimal spectral separation is achieved [45].

The final step involved validating the maps using ground control points (GCP) collected during fieldwork and referencing several studies done in the area [41,46,47]. These GCPs were recorded using handheld GPS devices across representative land cover types, including agricultural land, savanna, burned areas, forest patches, inhabited areas, and bodies of water identified through direct field observations. The points were used to validate the satellite-based land use/land cover (LULC) classification through visual comparison and expert knowledge. In addition to the collection of GPS coordinates and field observations, structured interviews were conducted with local farmers, herders, and community members to gather socio-economic data relevant to land use practices. The survey captured trends in population, crop and livestock farming activities, and the factors behind the perceived trends. These local perspectives provided valuable context for interpreting the socio-economic aspects related to LULC changes.

Land use change maps were created comparing 1988, 2002, and 2022 to analyze changes, including regressions (negative change), progressions (positive change), and stability (no change) based on combined data and statistical analysis.

3. Results and Discussions

3.1. Spatial Precipitation Trends over the Continental Gambia River Basin

Figure 2 presents the spatial trend in annual precipitation across the CGRB between 1990 and 2022, with areas of statistical significance overlaid. The results indicate a clear north–south gradient in precipitation trends. The northern part of the catchment experienced a statistically significant increase in annual precipitation, with changes reaching up to 35%. In contrast, the southern region exhibited a decreasing trend, with reductions of up to 10%, although these declines were generally not statistically significant. The central portion of the basin remained relatively stable or experienced moderate increases.

3.2. SPEI Indices

The SPEI was computed at 12-month and 24-month scales (Figure 3), which are useful in assessing hydrological drought and surface water availability [48]. Several authors [49,50,51,52] confirm SPEI’s effectiveness in evaluating drought impacts across different climates. A notable dry phase occurred in the early 1990s and 2000s, indicated by negative SPEI values, with a breakpoint detected in 2008, marking the onset of rainfall recovery and a shift to wetter conditions.

The causes underlying the detected breakpoints, particularly the notable shift around 2008, merit careful consideration. One potential driver is large-scale climate variability, especially the El Niño–Southern Oscillation (ENSO), which modulates rainfall patterns across West Africa [53]. The warm (El Niño) and cold (La Niña) phases of ENSO can significantly influence moisture transport and the intensity and timing of the West African monsoon, potentially contributing to prolonged dry or wet periods. Similar results were reported by [54], which identified a second phase of rainfall resumption beginning in 2008 across parts of the Sahel, including Senegal and Burkina Faso. This regional synchronicity reinforces the 2008 breakpoint as a key climatic shift toward increased rainfall in the study area. The post-2008 wetting trend aligns with the broader recovery from the 1970s–1980s drought, largely attributed to rising greenhouse gas concentrations and reduced aerosol emissions that altered atmospheric circulation and strengthened the monsoon system [55,56].

In addition to climatic influences, regional land use and land cover (LULC) changes may have played a critical role. For example, [57] documented significant LULC transformations, including forest loss and agricultural expansion, driven by political conflict and population movements in the African Great Lakes Region. In the Niger Basin, widespread land clearance has altered soil properties and infiltration capacity, increasing Hortonian runoff and the formation of gullies and ponds [58]. Conversely, shifts in environmental and land management policies, such as increased reforestation efforts and stricter regulation of slash-and-burn agriculture, may have facilitated forest regrowth and improved vegetation cover post-2008 [59,60]. These changes can enhance soil moisture retention and groundwater recharge, reinforcing the transition toward wetter conditions. Moreover, recent studies indicate that changes in vegetation, particularly the expansion of agricultural land, may have further amplified regional summer rainfall by increasing latent heat flux and moisture availability, suggesting a feedback loop between vegetation and precipitation in the water-limited Sahel [61]. However, despite higher rainfall totals, recent decades have witnessed more intense and erratic rainfall, a phenomenon referred to as the Sahelian paradox, resulting in more frequent floods due to stronger convective systems and increased rainfall concentration in the latter part of the rainy season [55,62].

Modified MK trend analysis revealed significant increasing trends towards wetter conditions for SPEI-12 and SPEI-24, with p-values of 4.77 × 10⁻⁶ and 3.01 × 10⁻⁵, respectively. Despite the recent wet trend, fluctuations and periodic dry spells highlight ongoing variability, consistent with observations across the Sahel and West Africa [63,64,65].

3.3. Discharge Trends at Simenti Station (1990–2022)

Trend analyses for mean annual discharge at the Simenti station reveal a significant increase, with a breakpoint in 2010 (Figure 4). From 1990 to 2010, dry conditions dominated, interrupted by wet years in 1994, 1999, 2003, and 2004. Post-2010, wet conditions prevailed, except in 2014 and 2017. The standardized flow index indicates wetter conditions in the CGRB, with increased discharge variability and more frequent positive anomalies, especially after 2020. This shift reflects regional hydrological recovery consistent with Sahel-wide rainfall trends, raising both opportunities for improved water availability and concerns over flooding risks during extreme wet events.

3.4. Spatial Evolution of Land Use/Land Cover

Spatial analysis (Figure 5) shows vegetation densification from the north to the south of the basin. The northern part of the basin is dominated by savanna, while the southern part is characterized by forest. Bowe and steppe dominate the north and northwest but are mostly scattered in the south. Agricultural areas and bare soils are concentrated in the north, where fire-affected zones were also significant between 1988 and 2002. Water surfaces align with the hydrographic network and ponds, mainly in the central and southern basins.

Between 1988 and 2002, forest areas in the north and center declined significantly, replaced by bowe and steppe, burn, agriculture and bare soil areas, and savanna. By 2022, forests rebounded, especially in the central basin. Savanna areas, dominant in the north and center, decreased due to the expansion of bowe and steppe, particularly in these regions. Burnt areas, evidently in the images of 2002, decreased by 2022, indicating improved fire management. These trends highlight intensified land degradation in the north and diverse land-use changes in the center and south of the study area.

The statistical analyses of the data (

Table 2. Area in hectares and as a percentage of land use in 1988, 2002, and 2022 in the CGRB.

Land Cover	1988		2002		2022
	ha	%	ha	%	ha	%
Water	26,807	0.63	25,337	0.6	11,862	0.28
Forest	1,374,975	32.36	1,092,177	25.7	1,541,035	36.27
Savanna	2,400,358	56.49	2,292,801	53.96	2,019,953	47.54
Bowe and steppe	243,284	5.73	610,811	14.37	536,513	12.63
Agriculture area and bare soils	117,717	2.77	79,453	1.87	130,289	3.07
Burn	86,192	2.03	148,753	3.5	9680	0.23
Total	4,249,333	100	4,249,333	100	4,249,333	100

) show that forest cover experienced a marked decline by 2002 from 32.36% to 25.7%, followed by recovery to 36.27% in 2022, indicating conservation or natural regrowth. Savanna areas declined during the study period from 56.49% in 1988 to 47.54% in 2022, reflecting degradation and agricultural expansion. Water surfaces decreased from 0.63% in 1988 to 0.28% by 2022.

Bowe and steppe tripled by 2002, rising from 5.73% in 1988 to 14.37%, and stabilized at 12.63% in 2022. Agriculture and bare soil areas expanded, by 2022, from 2.77% in 1988 to 3.07%, signifying anthropogenization. Burnt areas peaked at 3.5% in 2002 but reduced dramatically to 0.23% by 2022. These land use trends indicate a phase of degradation (1988–2002), followed by vegetation recovery from 2002 to 2022. The dynamics of vegetation in the basin are primarily influenced by climatic conditions (rainfall and hydrology) and bush fires. Favorable climatic conditions promote increased vegetation density, while unfavorable conditions lead to the reverse conditions.

Figure 6 summarizes the overall percentage change in each land use for the study period from 1988 to 2022. Burns areas reduced significantly by −88.77%, followed by water (−55.75%), and lastly, savannah by −15.85%. The bowe and steppe class, on the other hand, increased significantly by 120.08%, followed by forest and agricultural and bare soil areas, which increased slightly by 12.08 and 10.68%, respectively.

3.5. Changes in Land Use/Land Cover from 1988 to 2002

The spatial map of land use change between 1988 and 2022 is depicted in Figure 7. During this period, dense forests were relatively well-preserved, particularly in the southern and central regions of the basin. Savannah areas are observed to be maintained mostly in the northern and central basin areas. Savanna areas have expanded, especially in the extreme west of the basin. Additionally, cropland areas expanded moderately, primarily in the southern and central regions. Bowe and steppe areas progressed significantly in the north and north–west of the basin.

The statistical representation of change from 1988 to 2002 is presented in

Table 3. Land use change matrix in the Continental Gambia River Basin between 1988 and 2002.

Classes	Water	Forest	Savanna	Bowe and Steppe	Agriculture Area and Bares Soils	Burn	Total	Regression
Water	17,903.5	7491.5	2412.5	253.9	127.0	1523.7	29,712.1	11,808.7
Forest	7872.4	770,864.2	468,283.2	51,551.8	1396.7	68,312.5	1,368,280.7	597,416.6
Savanna	1142.8	290,391.4	1,601,025.6	408,858.9	25,141.0	72,121.7	2,398,681.5	797,655.9
Bowe and steppe	127.0	10,665.9	125,070.2	94,723.2	11,681.7	3936.2	246,204.2	151,481.0
Agriculture area and bares soils	0.0	2539.5	33,140.4	44,441.2	40,124.0	888.8	121,134.0	81,009.9
Burn	127.0	10,031.0	54,345.2	10,792.9	380.9	10,158.0	85,835.0	75,677.0
Total	27,172.6	1,091,983.5	2,284,277.1	610,621.9	78,851.4	156,940.9	4,249,847.5	1,715,049.0
Progression	9269.2	321,119.3	683,251.5	515,898.7	38,727.3	146,782.9	1,715,049.0

. The land cover change matrix reveals significant transformations across various land use types. One of the most notable trends is the substantial reduction in forest cover. Although 770,864.2 ha of forest remained stable, approximately 597,416.6 hectares were lost, with the majority of the lost area, 468,283.2 hectares, being converted to savanna, reflecting extensive degradation. Additionally, 51,551.8 hectares of forest transitioned into bowe and steppe, while 68,312.5 hectares were lost to burn areas, indicating the impact of fire or slash-and-burn agricultural practices. Savanna, on the other hand, emerged as the dominant land cover type, with a total area of 2,398,681.5 hectares by 2002. This expansion was driven by significant gains from forest loss, with 290,391.4 hectares transitioning from forest to savanna, and an additional 408,858.9 hectares gained from bowe and steppe. Despite losing 797,655.9 hectares to other land uses, savanna continued to expand. Similarly, bowe and steppe saw substantial changes, with 125,070.2 hectares transitioning into savanna highlighting an increase in vegetation cover over previously degraded areas. Agricultural and bare soil areas also expanded significantly, gaining 44,441.2 hectares from bowe and steppe and 33,140.4 hectares from savanna, indicating increased land conversion for cultivation. However, agriculture and bare soils lost 81,009.9 hectares to other classes, primarily due to transitions to natural vegetation and burn areas. Water bodies also experienced a notable reduction, with 11,808.7 hectares transitioning into other land cover types, including forest (7491.5 hectares) and savanna (2412.5 hectares). Lastly, burn areas played a significant role in shaping the landscape, with 54,345.2 hectares of savanna and 10,031.0 hectares of forest converted into burn areas. Overall, the period from 1988 to 2002 was characterized by extensive deforestation, significant savanna expansion, and increased agricultural land use.

3.6. Changes in Land Use/Land Cover 2002–2022

Figure 8 highlights land cover changes from 2002 to 2022. This period shows a significant retention of forest areas in the center of the basin, especially the west–central. Savanna areas that remained unchanged are observed throughout the basin but are observed to progress most in the north and east–central of the basin. Agricultural areas and bare soils are mostly unchanged in the north and have progressed in the same areas as well. Bowe and steppe areas remain unchanged, particularly in the north and are observed to progress all over the basin.

The land cover change matrix for the period 2002 to 2022 highlights significant transitions across various land cover types (

Table 4. Land use change matrix (in ha) in the Continental Gambia River Basin between 2002 and 2022.

Classes	Water	Forest	Savanna	Bowe and Steppe	Agriculture Area and Bares Soils	Burn	Total	Regression
Water	8507.3	14,348.2	3936.2	380.9	0.0	0.0	27,172.6	18,665.3
Forest	761.8	811,623.1	211,159.1	60,820.9	4825.0	2793.4	1,091,983.5	280,360.4
Savanna	761.8	580,275.0	1,393,548.7	261,441.2	41,520.8	6729.7	2,284,277.1	890,728.4
Bowe and steppe	0.0	82,914.6	303,215.9	181,701.0	42,663.5	127.0	610,621.9	428,921.0
Agriculture area and bare soils	0.0	1523.7	11,173.8	22,093.6	44,060.3	0.0	78,851.4	34,791.1
Burn	253.9	64,122.3	81,898.8	8888.2	1650.7	127.0	156,940.9	156,813.9
Total	10,285.0	1,554,806.8	2,004,932.5	535,325.9	134,720.3	9777.1	4,249,847.5	1,810,280.1
Progression	1777.6	743,183.7	611,383.8	353,624.9	90,660.0	9650.1	1,810,280.1

). One of the most striking trends is the continued reduction in water bodies, with only 8507.3 hectares remaining stable. A notable 14,348.2 hectares of water were converted into forest, and 3936.2 hectares transitioned into savanna. Overall, water bodies regressed by 18,665.3 hectares, indicating a significant decline over this period. Forest cover exhibited a mixed trend, with 811,623.1 hectares remaining intact, reflecting a relatively high degree of stability. However, forest losses were substantial, with 211,159.1 hectares transitioning into savanna and 60,820.9 hectares converting into bowe and steppe, indicating forest degradation. Additionally, 64,122.3 hectares were lost to burn areas. In total, forest cover regressed by 280,360.4 hectares, continuing the deforestation trend observed in previous years. Savanna emerged as the most dynamic land cover type, with 1,393,548.7 hectares remaining stable and substantial gains from forest (580,275.0 hectares) and bowe and steppe (261,441.2 hectares). Despite these gains, savanna experienced a regression of 890,728.4 hectares, primarily due to conversion to forest (580,275.0 hectares), bowe and steppe (261,441.2 hectares), and agricultural and bare soil areas (41,520.8 hectares). Bowe and steppe, typically associated with degraded land, showed significant changes as well. While 181,701.0 hectares remained stable, 428,921.0 hectares were lost, mainly transitioning into savanna (303,215.9 hectares), forest (82,914.6 hectares), and agricultural land (42,663.5 hectares), reflecting ongoing land conversion. Agricultural areas and bare soils expanded considerably, with 44,060.3 hectares remaining stable and 22,093.6 hectares lost to bowe and steppe. Another 11,173.8 hectares were lost to savanna. Despite the considerable area maintained, agriculture and bare soils lost 34,791.1 hectares in total during the period. Burn areas, characterized by their dynamic nature, saw only 127.0 hectares remain stable. A significant 81,898.8 hectares and 64,122.3 hectares of burn areas were converted to savanna and forest. Nearly all burn areas (156,813.9 hectares) transitioned into other land cover types, emphasizing their transient nature. Overall, the period from 2002 to 2022 was marked by extensive land cover changes, with savanna expansion, deforestation, and agricultural growth as the dominant trends.

3.7. Influence of Precipitation on Land Use Transitions

To evaluate the role of precipitation in shaping land cover dynamics, correlation analyses were conducted between annual precipitation and specific land cover conversions for the periods 1988–2002 and 2002–2022 (

Table 5. Correlation between land use conversions and precipitation.

Land Conversion	Period	Correlation (r)	p-Value
Agricultural Areas and Bare Soil to Bowe and Steppe	1988–2002	0.268	0.034
Forest to Bowe and Steppe	1988–2002	0.225	0.007
Forest to Water	1988–2002	−0.633	0.001
Savanna to Bowe and Steppe	2002–2022	−0.073	0.034
Savanna to Forest	2002–2022	0.058	0.048

).

As previously described, long-term precipitation trends across the CGRB between 1990 and 2022 exhibit a distinct north–south gradient, with increasing rainfall in the northern regions and declining trends toward the south. While this spatial pattern reflects broad climatic shifts over the 33-year period, it is important to note that precipitation conditions during sub-periods such as 1988–2002 may differ from these long-term trends, and regional variability must be considered when interpreting land cover changes.

During the 1988–2002 period, the conversion of agricultural areas and bare soil to bowe and steppe exhibited a weak positive correlation with precipitation (r = 0.268, p = 0.034), suggesting that wetter conditions have favored the spread of these land types. Similarly, the transition from forests to bowe and steppe also showed a weak positive correlation (r = 0.225, p = 0.007). This pattern is unexpected, as increasing precipitation would typically support forest persistence or regrowth rather than degradation into open, sparsely vegetated landscapes of bowe and steppe. This unexpected pattern reflects the overriding role of anthropogenic pressures, such as deforestation, unsustainable land use, frequent burning, and overgrazing, which can alter soil conditions and vegetation dynamics despite favorable climatic conditions [66]. In these contexts, the presence of favorable rainfall is insufficient to reverse or resist degradation processes once they have been set in motion. In other words, anthropogenic pressures may decouple ecological transitions from climatic conditions, leading to land cover changes that would not normally be expected under wetter scenarios.

A contrasting trend is observed in the conversion of forests to water, which showed a strong negative correlation with precipitation (r = −0.633, p = 0.001), suggesting that this transition was more frequent under drier conditions. While reduced precipitation would typically lead to lower surface water availability, studies have shown that human activities, such as deforestation and agricultural expansion, can significantly alter local hydrological conditions, contributing to water accumulation in previously forested areas [67,68].

In the 2002–2022 period, fewer significant correlations were observed, but notable trends include the weak negative correlation between the conversion of savanna to bowe and steppe and precipitation (r = −0.073, p = 0.034), suggesting that this transition was more pronounced in drier conditions. Additionally, the conversion of savanna to forest displayed a weak positive correlation (r = 0.058, p = 0.048), implying that increased precipitation supports forest regrowth in some savanna-dominated areas.

Overall, these results underscore that while precipitation trends do influence land cover change, they do not act in isolation. The interaction between climate variability and human land use decisions, such as farming practices, deforestation, and land abandonment, shapes the landscape in ways that sometimes override the expected climatic responses.

3.8. Local Perceptions of Environmental and Land Use Change

Qualitative data collected through structured interviews with local farmers, herders, and community members in the CGRB provided key insights into land use practices and environmental changes. Most respondents (89%) reported an increase in the local population, primarily attributed to high birth rates (57%). A smaller proportion cited reduced socio-economic opportunities (30%) and migration (13%) as contributing factors. This perceived population growth suggests rising demographic pressure on land and natural resources, potentially intensifying agricultural expansion, grazing demands, and land degradation.

A majority of respondents (55.5%) perceived agricultural activity as declining, while 38% believed it was growing, and only 6.6% reported no change. The main factors cited for these changes were water resource decline (43.8%) and lack of financial and material resources (43.1%). Other notable drivers included population increase (22.8%), rural exodus (22.1%), and the modernization of agriculture (20.34%).

Most respondents (67.5%) perceive livestock farming as declining, while 21.8% see it growing and 10.7% report no change. The main factors driving this trend are poverty (42.4%) and insufficient water (41%), followed by reduced forage availability (26.6%) and bushfires (24.5%). Other less cited factors include population increase (16.9%) and increased forage or sufficient water, which are associated with positive conditions.

4. Discussion

The observed land cover changes between 1988 and 2022 were driven by a combination of human activities and climate variability. The decline in forests and savanna from 1988 to 2002 aligns with a period of relatively drier conditions, as indicated by the discharge trends at Simenti station during this period. These drier conditions may have intensified land use pressures as communities sought alternative livelihoods, such as expanding agricultural lands and increasing deforestation for charcoal production. This pattern is consistent with broader regional trends across West Africa, where prolonged droughts in the Sahel have historically driven land conversion and resource competition, leading to deforestation and land degradation [69,70,71]. The study by [12] similarly found that the annual rate of water body loss in West Africa was greater in the post-drought period than during the drought era, highlighting the long-term effects of past climatic stress on land cover.

Conversely, the increase in precipitation observed after 2008, as indicated by the breakpoints in the SPEI and discharge time series, coincides with a notable expansion of forest areas between 2002 and 2022. This suggests that wetter conditions may have facilitated vegetation regrowth, as supported by the weak positive correlation between savanna-to-forest transitions and precipitation (Table 4). This aligns with studies showing that forest recovery is more likely in areas with increased rainfall [72,73,74]. The observed increase in forest cover after 2002, however, remains somewhat ambiguous and likely reflects the combined influence of climatic, ecological, and human factors. On one hand, the transition to wetter conditions post-2008, as indicated by positive SPEI values and increased streamflow, may have created favorable conditions for natural vegetation regrowth and forest expansion. Enhanced soil moisture availability and reduced fire frequency during wetter periods can facilitate the recovery of degraded areas and support seedling establishment. In addition to climatic drivers, deliberate afforestation and reforestation initiatives, possibly supported by government policies or community-led conservation programs, may have contributed to forest recovery. Senegal and Gambia have implemented tree planting and land restoration projects since the early 2000s in response to concerns over desertification and land degradation. Moreover, natural regrowth following the abandonment of marginal agricultural lands, due to reduced agricultural profitability or outmigration, could also explain part of the forest gain. When farming pressure declines, previously cleared lands can undergo spontaneous succession, leading to shrubland and secondary forest formation. These intertwined processes make it challenging to isolate the exact contribution of each driver without detailed field-based vegetation surveys, socio-economic data, and land management records. Future studies integrating remote sensing time series analysis with ground-truthing and policy evaluation would be essential to disentangle these effects and better understand the mechanisms underlying forest dynamics in the basin.

Additionally, the drastic reduction in burned areas from 3.5% in 2002 to 0.23% in 2022 suggests changes in fire frequency, which could be linked to either wetter conditions reducing fire spread or shifts in land management practices, such as fire suppression efforts or land conversion efforts. Studies in Senegal indicate that fire regimes are predominantly shaped by anthropogenic activities rather than natural causes. Local communities commonly employ fire as a deliberate land management tool tied to livestock grazing, agricultural practices, and clearing of fields, resulting in distinct spatial and temporal fire patterns, particularly concentrated in savannah woodlands and agricultural areas early in the dry season [75,76,77]. The seasonal distribution of fire is marked by two main peaks: the cold dry season (October to January) and the hot dry season (February to March), with the latter accounting for the highest intensity of burned areas. Peak fire activity during February and March coincides with favorable burning conditions such as dry vegetation, high temperatures, Harmattan winds, and agro-pastoral activities like harvesting and transhumance. Fires occurring in April and May are often associated with land preparation activities such as agricultural clearing and brush cutting using fire [78]. Satellite observations combined with field surveys reveal that natural ignition sources like lightning contribute minimally to fire occurrences. This human-driven fire regime significantly influences vegetation dynamics and land degradation, with both beneficial “early fires” used for fuel management and destructive uncontrolled burns observed [79]. Recognizing these socio-ecological drivers of fire is essential for designing effective fire and land management strategies in the CGRB, especially in light of the observed decline in burned areas that may reflect changes in local fire-use practices alongside climatic influences.

Agricultural expansion, in particular, has been encouraged through government initiatives aimed at achieving food self-sufficiency. For instance, the Senegalese state implemented major programs such as the Great Agricultural Offensive for Food and Abundance (GOANA) and the Program to Revive and Accelerate Agriculture in Senegal (PRACAS), targeting an increase in rice production through the expansion of irrigated and rainfed cropland, leading to a >40% increase in cultivated rainfed areas after 2008 [80]. Moreover, large-scale land acquisitions, totaling 844,796 hectares across 40 deals, were facilitated by policy reforms that prioritized private agri-business over traditional land uses [81]. These programs were anchored within broader legal frameworks, such as the Agro-Sylvo-Pastoral Orientation Law (LOASP) and the national land policy strategy adopted by the National Land Reform Commission, which promoted land mobility, modernization of agriculture, and the use of land as collateral to attract private investment [81]. These transformations have occurred most intensively in fertile and well-watered regions, further exacerbating deforestation, degradation, and land pressure. Between 2000 and 2010 alone, 405,000 ha of forest were lost, with 15 million m³ of woody biomass depletion, partly due to agricultural encroachment, charcoal production, and weak governance of natural resources. The decentralization of land management to local councils often lacked adequate regulatory mechanisms or institutional capacity, resulting in informal land sales, uncoordinated land use, and ineffective conservation [81]. These structural shifts in land governance and policy have significantly reshaped land use trajectories in the CGRB and must be factored in when assessing human-induced land use and land cover changes. Although remote sensing data and national agricultural programs confirm a marked expansion of cropland in the region, local perceptions consistently reflect a decline in agricultural activity. This contrast demonstrates that increased cultivated area does not automatically lead to improved productivity or better livelihoods for rural communities. The discrepancy underscores the need to integrate spatial data with local knowledge to fully capture the uneven impacts of land use transformation on local livelihoods.

Beyond agricultural expansion, grazing places significant pressure on land resources in the CGRB and across Senegal. Over 90% of livestock rely on natural pastures, yet forage productivity ranges from just 100 kg/ha in the Sahelian zone to 2000 kg/ha in more humid regions, making pastoral systems highly vulnerable to climate variability [82]. National forage balance assessments from 2009 to 2018 indicate persistent deficits, with available biomass often falling short of livestock needs. These pressures, combined with expanding agriculture, have reduced grazing space and intensified land-use conflicts between pastoralists and farmers [82]. Consequently, overgrazing contributes to vegetation degradation, soil nutrient depletion, and reduced ecosystem resilience.

However, despite these pressures, more recent developments suggest a partial reversal of forest loss trends. This apparent resurgence in forest cover can be attributed to several overlapping drivers. First, national and regional reforestation efforts—particularly under the Great Green Wall (GGW) initiative—have led to the planting and protection of tens of thousands of hectares of forest. Between 2008 and 2021 alone, approximately 57,000 hectares were reforested through GGW-related projects in Senegal, supported by more than 15 million seedlings and reinforced by measures such as fencing and active community engagement [83]. Second, a rise in farmer-managed and private-sector regreening has contributed significantly to landscape restoration. Studies comparing public and private initiatives in the Senegalese Sahel have shown that private plantations, including gum arabic farms, and participatory land rehabilitation programs have produced positive ecological and socio-economic outcomes [84]. Third, Farmer-Managed Natural Regeneration (FMNR) and agroforestry have become increasingly important for restoring degraded lands across the Sahel, including Senegal. Rather than relying solely on tree planting, FMNR promotes the natural regeneration of indigenous woody species through farmer-led protection and selective pruning of stumps and root systems. This approach has led to substantial increases in on-farm tree cover, improved soil fertility, and enhanced resilience to climate variability [84,85,86].

The expansion of cropland in our study area, particularly between 1988 and 2022, mirrors findings from West Africa as a whole. Agricultural expansion, primarily driven by Wolof and Serer settlers from the northwest, led to a significant influx of farmers and herders between the late 1970s and early 2000s [87,88]. Seeking new fertile agricultural land and more favorable climatic conditions, these groups converted wooded savannas into farmland by clearing stumps, shrubs, and trees to enable animal traction. The growing cultivation of cash crops like cotton and peanuts further led to the encroachment on wooded savannas and grazing areas, echoing findings from the regional study that identified cropland expansion as one of the main drivers of forest loss [11,46,87]. Deforestation for charcoal production driven by demand from urban areas such as Dakar, Kaolack, Thiès, Diourbel, and Saint-Louis has also contributed to forest loss, particularly in the south and southwest [89,90]. Bushfires, wildfires, and hunting have further accelerated land degradation, accounting for a significant portion of forest loss [66]. The increase in agricultural lands, bare soils, and burned areas between 1988 and 2002 reflects this pattern, as land clearance for farming often follows tree removal. Additionally, the expansion of bowe and steppe during this period suggests land degradation processes, potentially driven by overgrazing, soil erosion, and declining soil fertility, as documented by [69]. In addition to agricultural expansion and climate variability, socio-economic factors such as migration, urbanization, and decentralization policies have significantly shaped land use dynamics in the CGRB. Rural–urban migration, often driven by limited livelihood opportunities and land access, has led to both land abandonment and shifting cultivation patterns, particularly in southeastern Senegal [91,92]. Migrants and returnees, supported by remittances, have also contributed to land conversion through investment in cash crops and agroforestry [93]. The Casamance conflict, for instance, triggered a rural exodus that reduced land pressure in some zones, allowing forest regrowth, while simultaneously concentrating agricultural activity elsewhere [92]. Moreover, decentralization policies since 2000 have reshaped land governance by promoting agricultural growth through improved rural access to infrastructure and land inputs, though often without sufficient institutional oversight [94]. Urban population growth has further increased charcoal demand and land degradation near urban centers, reinforcing pressures on forest resources [95,96]. These interlinked socio-economic dynamics are crucial to understanding the spatial variability of land cover change and must be considered alongside environmental drivers.

While climate variability has influenced land cover changes, human activities have also played a role in modifying the hydrological system of the CGRB. The expansion of agricultural land and bare soils, particularly in the north, may have increased surface runoff and flash floods due to reduced vegetation cover. While the return to wetter conditions likely played a role in increasing streamflow, land degradation, characterized by loss of savannah and expansion of bowe and steppe, could have amplified runoff efficiency, as suggested by the post-2010 increase in discharge variability and positive anomalies observed at Simenti station. Even though increased discharge improves water availability, it also raises flood risks, posing hazards to local communities. Over the last two decades, frequent floods in West Africa have severely affected millions and caused significant fatalities [97]. This underscores the urgent need for detailed measurements and comprehensive assessments of extreme event severity to enhance flood risk mitigation and disaster management in vulnerable communities. These observations align with studies suggesting that land use changes in the Sahel have influenced river flow recovery by reducing infiltration and increasing surface runoff [98,99,100].

Collectively, these findings underscore the urgency of implementing land and water management strategies in the CGRB. The expansion of agriculture and degradation of savanna and steppe ecosystems underscore the need for integrated land-use planning that balances food production with ecosystem conservation. Reforestation initiatives, including FMNR, should be reinforced as low-cost, community-based approaches for restoring degraded lands and improving soil fertility. Moreover, the observed increases in runoff and discharge variability highlight the importance of watershed-based flood management, including protecting riparian vegetation and enhancing drainage infrastructure in flood-prone areas. As land cover changes influence both water availability and hazard exposure, future management should prioritize sustainable practices that reduce erosion, support groundwater recharge, and mitigate disaster risks while maintaining rural livelihoods.

5. Conclusions

This study analyzed hydroclimatic trends and land use/land cover (LULC) changes in the Continental Gambia River Basin in West Africa. Hydroclimatic analysis using the Standardized Precipitation Evapotranspiration Index (SPEI) and river discharge data revealed a shift from drought conditions (1990–2008) to wetter periods after 2008 for precipitation and after 2010 for river discharge, although variability and drought risks persist. Integrated water resource management (IWRM), flood early warning systems, and increased reservoir capacity are recommended to manage water scarcity and flooding risks. LULC analysis (1988–2022) shows forest expansion in the southern and central basin, while agricultural land and bare soils have increased in the north, driving land degradation. Sustainable land management, such as agroforestry, reforestation, and bushfire control, is essential to counteract these trends. Soil conservation techniques like contour farming and restoring degraded lands are necessary to enhance water retention and reduce erosion.

Finally, the correlations between precipitation and land cover changes indicate that while climatic shifts played a role in driving vegetation recovery, their positive effects were largely counteracted by human activities. Land use changes, driven by agricultural expansion, deforestation, and land degradation, played a dominant role, shaping both precipitation-runoff relationships and stream discharge patterns. This highlights that anthropogenic pressures, rather than climate alone, were the primary force behind landscape transformation during this period. Strengthened land-use policies, community-based resource management, and alternative livelihood programs are critical to balancing development and environmental conservation, ensuring the basin’s long-term resilience.