The study was conducted in the Eastern flank of Mount Cameroon, covering Buea Municipality, located in the South West Region of Cameroon. Geographically, the flank spans from 04˚8.036’N to 04˚12.627’N and 009˚13.104’E to 009˚18.675’E. ([11]). The area is mountainous with thick and evergreen forest vegetation at the lower slope of the mountain and there are transitional changes along altitudinal gradient ([21]). The area covers approximately 5695.5 km² and has an estimated population of 800,000 inhabitants by 2023 ([20]). The study area is characterized by fertile soil and receives annual rainfall ranging from 4000 to 6000 mm and has an equatorial climate with a rainy season (March to October) and a dry season (November to February). The region supports a diverse range of ecological zones from lowland rainforest to subalpine grassland. Given the high soil fertility, the legendary hospitality of the indigenes, the high educational hub and commercial development, there is a consequential increase in human population and pressure on land resources, making it necessary to study land use and land cover changes, consequences and to ensuring sustainable land resource use.

Change detection, a component of land monitoring, Remote Sensing (RS) and Geographic Information Systems (GIS) were employed to detect, classify, and quantify LULC changes from 2010 to 2025. The classification workflow method was adapted from [14]. Satellite images were acquired from the United States Geological Survey (USGS) through NASA’s Earth Explorer platform. Landsat 7 ETM+ (was used for the years 2010 and 2015), Landsat 8 OLI (2020) and Landsat (2025) scenes were selected with less than 10% cloud cover during dry season months (Table 1). Cloud-free images ensure unobstructed visibility of the Earth’s surface, allowing for precise identification and classification of land features ([26]). Images were subjected to radiometric correction, pan sharpening, and layer stacking to improve classification accuracy ([7]). A hybrid classification approach combining unsupervised (k-means) and supervised (maximum likelihood (MLC) and support vector machine (SVM)) methods were applied ([3]; [28]). The MLC method enhances classification accuracy by incorporating the statistical distribution of training data for each category, producing a precise LULC map ([30]). The LULC of the study area was classified into six classes, namely, Dense Forest, Secondary Forest, Agricultural Land, Grassland, Lava/Bare Rocks, and Buildups/Bare Soils (Table 1). Three satellite images were employed for this research. The satellite image was captured at intervals between 2010-2015, 2015-2020, and 2020-2025. Accuracy was assessed using error matrices and the Kappa index ([34]). Kappa coefficient (Kappa) was measured to determine the interdependence between the classes. Change detection was performed using the cross-tabulation module in ArcGIS and different LULC were used in the study (Table 2).

Table 1.Remote sensing inputs, path/row, acquisition date and Spatial resolution.

Table 2.Description of land use and land cover classes.

The rate of change in the different LULCC classes were calculated using the equation;

where R is rate of change in the different LULCC;

P is LULCC class of the last year and Q is baseline year.

The annual change (A) was calculated using the difference between baseline year (2010) and the recent year (2025), considering 15 years in between the years as shown in Equation (2);

A qualitative research design was adopted from [1]. Semi-structured questionnaires and interviews were administered to residents in ten villages within Buea municipality (Muea, Mile 16, Bonakanda, Bokova, Ndongo, Molyko, Bokwaongo, Boteva, Bwitingi, and Bova) presenting 20 questionnaires per village. A purposive sampling technique was used to select 200 participants aged 30 years and above who had lived in Buea for at least 20 years. The questionnaire was partitioned into three sections: demographic information, drivers of deforestation and perceived consequences of deforestation. Occupation of respondents, drivers of change and perceived consequences had multiple responses. Informed consent was obtained from all participants prior to questionnaire administration.

LULC change data were analyzed using ArcGIS cross-tabulation tools to compute change matrices. Community survey data were coded and entered into Microsoft Excel and analyzed using SPSS version 25. Descriptive statistics (frequencies, percentages, mean, and standard deviation) were computed. Chi-square tests were used to assess associations between categorical variables. Qualitative responses were analyzed through thematic coding. Results are presented using tables and figures.

The area of the classified polygon was calculated in ArcGIS from the pixel count of the classified raster multiplied by the pixel (cell) area from the total area of 5695.5 km² of the municipality. Satellite image analysis revealed substantial LULC changes on the Eastern flank of Mount Cameroon over the 15-year study period (Figure 1).

There was a substantial variation in the LULC in the study with Agriculture increasing (from 1373.09 ha to 12414.44 ha) in area while dense forest decreases from 16228.34 to 3453.78 ha (Table 3).

The most dramatic change occurred between 2010 and 2015, when dense forest declined by 6362.9 ha (39.21%) and agricultural land increased by 5194.50 ha (378.31%). Between 2020 and 2025, a further sharp decline in dense forest was recorded: a loss of 5642.20 ha (62.03%) accompanied by an agricultural land gain of 4892.60 ha (65.04%) over the study period. The annual change in LULCC was equally higher in agricultural land as more land was gained from either dense forest and or secondary forest. Equally Buildups increased from 1115.83 (ha) in 2010 to 2655.50 (ha) in 2025, with an annual change of +736.09 (ha) and +102.64 (ha) for agricultural land and buildups respectively (Table 4).

The accuracy assessment of the classification matrices demonstrated the reliability of the LULC data across all four time periods, with overall accuracy values and Kappa coefficients meeting the minimum threshold for acceptable classification ([34]). Table 5shows the accuracy assessment of the classification for each of the 4 different years classified, showing the overall accuracies and their corresponding kappa coefficients. The combine classification approach was useful and produced high classification accuracies.

Figure 1. LULC the different years; (a) 2010, (b) 2015, (c) 2020, (d) 2025.

Table 3.LULCC between 2010-2025 in the eastern flank of mount Cameroon.

Table 4.Rate of change in LULC categories between 2010-2025.

Table 5.Accuracy assessment of the classification matrices in the study period.

A total of 200 respondents participated in the community survey. The majority were between 46 and 50 years old (56.0%), with 33.5% aged over 50 years. More than half were male (54.5%), while females constituted 45.5% (χ² = 62.11, df = 2, p < 0.001). Farming was the dominant occupation (85.0%), followed by teaching (18.5%) and business (5.5%) (χ² = 200.7, df = 2, p < 0.001). The majority resided in Muea (33.5%) and Bonakanda (24.5%). Most respondents were married (89.0%) and had 2 - 4 children (67.0%). Regarding education, 47.0% had attained primary-level education, followed by secondary (34.0%), vocational (13.0%), and no formal education (6.0%).

Table 6.Selected variables to drivers of deforestation.

The perceived drivers of deforestation on the eastern flank of Mount Cameroon were dominated by agricultural expansion (39%), followed by urban expansion (31%) and infrastructure development (14%). Fuelwood and charcoal production and population growth were each reported by 5% of respondents. Logging (legal and illegal) was cited by 4% and fire by 1%. The distribution of drivers was statistically significant (χ² = 95.09, df = 6, p < 0.001). The majority of respondents (89%) affirmed that agricultural expansion has contributed to deforestation in their area (χ² = 150.1, df = 1, p < 0.001). The dominant farming type was mixed farming (83.5%), followed by subsistence farming (31.5%) and cash crop farming (7.5%). The most commonly cultivated crops were yam (64.5%), plantain (58.5%), and cocoyam (55.0%), with farms most frequently located in Bonakanda (43.5%) and Bova (20.5%). Tree removal was common, with 82.5% of respondents reporting that trees were present when they first acquired their farmland. Most trees were cut down (53.5%) or burned (36.5%). The trees most affected included Milicia excelsa (38%), Cecropiapeltata (33.5%), Swietenia macrophylla (23%), and Perseaamericana (17%). Fuelwood was the primary use of harvested trees (67%), followed by house construction (10%), timber (8%), and medicine (6.5%). Primary drivers of deforestation identified were farm expansion (90.5%), fuelwood harvesting (81%), and bush burning (76.5%) (χ² = 194.9, df = 4, p < 0.001). Most respondents (68%) had farms in multiple locations, and 61% reported higher yields from recently cleared forested land compared to other areas. The main reasons for farming in distant locations were fertile soil (61.5%) and land availability (50.5%). About 67.5% of respondents were aware of protected areas such as the national park, and 87.5% reported conflict with wildlife from protected areas (χ² = 120.3, df = 1, p < 0.001), with antelopes (80%), monkeys (29%), and squirrels (18.5%) being the most common crop-raiding animals (Table 6).

The most frequently reported consequence of deforestation was water scarcity (36%), followed by biodiversity loss (24%), human-wildlife conflict (12%), climate variability (10%), and soil erosion and decline in non-timber forest products (NTFPs) (7% each). Disease outbreak was mentioned by 4% of respondents. The distribution of consequences was statistically significant (χ² = 56.10, df = 6, p < 0.001).

A total of 78% of respondents reported that deforestation had negatively impacted water availability in their communities while 84.5% noted a decrease in fuelwood availability. The main water sources were boreholes (68%), rainwater (48.5%), and streams (43.5%). Additionally, 89% of respondents stated that deforestation had worsened road conditions through soil erosion and flooding (χ² = 124.16, df = 1, p < 0.001), and 83% reported over 7 hours of daily rainfall in the past 24 years (χ² = 93.52, df = 1, p < 0.001). Most respondents (62.5%) indicated no traditional practices supporting forest conservation in their community (Table 7).

Table 7. Perceived consequences of deforestation in study.