Analysis of Land Cover Dynamics in Gashaka-Gumti National Park, North-East, Nigeria

*Corresponding author: Favour N. Eze Copyright © 2025 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution License 4.0. Musa W. R 1, G. O. Enaruvbe 2, M. A. Oyinloye 1, Favour N. Eze 1, *, Sarki H. L. A 3 and Reinhard Oni 2 1 Federal University of Technology, Akure, Nigeria. 2 African Regional Institute for Geospatial Information Science & Technology (AFRIGIST), Ile-Ife, Nigeria. 3 National Institute for Hospitality and Tourism (NIHOTOUR), Lafia Campus Nasarawa State, Nigeria. World Journal of Advanced Research and Reviews, 2025, 28(02), 1624–1637 Publication history: Received on 09 October 2025; revised on 17 November 2025; accepted on 19 November 2025 Article DOI: https://doi.org/10.30574/wjarr.2025.28.2.3853 Abstract Land use change, driven by human population growth, poses significant environmental degradation and challenges in nature conservation planning. This study aimed to assess land use and land cover change patterns in Gashaka-Gumti National Park, focusing on land cover types. Landsat images from 1987, 2003, and 2021 were processed and classified into five classes: high forest, woodland, grassland, barren land, and open land. The intensity of change was determined using three increasing details of land change. The results showed an accuracy of over 85%, indicating acceptable precision. The rate of land change decreased by 2.6% between 1987 and 2003 and further decreased by 2.0% between 2003 and 2021. During the initial period, grassland and barren land expanded and decreased, while open land and grassland increased. During the initial and subsequent epochs, grassland increased across all categories except for high forest and woodland. The transformation of dense forest and woodland into grassland and other land use types indicates the potential impact of human activities, posing a significant challenge to conservation efforts. The study's findings indicate that forest and woodland ecosystems are facing significant risks due to grassland encroachment, primarily attributed to anthropogenic activities. Therefore, intensifying efforts to preserve forests and woodlands is crucial for achieving conservation goals. Keywords: Land Change; High Forest; Exponential Growth; Gashaka-Gumti National Park; 1. Introduction Land Use and Land Cover Change (LULCC) reflect human activity's impact on the Earth's surface, driven by rapid population growth, particularly in developing regions such as sub-Saharan Africa, Southeast Asia, and Latin America (Nyamekye et al., 2020; Kourosh et al., 2019). This phenomenon contributes to both global and local environmental changes, including urban expansion and significant deforestation. Since the 1960s, over half of global forests have been lost (IUCN, 2017), with 178 million hectares disappearing between 1990 and 2000 alone (FAO, 2020). Tropical forests, which experienced over 10 million hectares of loss between 2000 and 2020, are particularly affected, with Africa accounting for much of this decline (Venkatappa et al., 2020; World Bank, 2014). This forest loss endangers 80% of terrestrial biodiversity and critical ecosystem services, including air quality, water regulation, and climate stabilization (IUCN, 2017; Petersen et al., 2018). Furthermore, deforestation is the second-largest contributor to global greenhouse gas emissions after fossil fuel combustion ((Mitchell et al., 2017; Stocker, 2014; Engdaw, 2020). Advancements in remote sensing (RS) and geographic information systems (GIS) have facilitated the integration of multi-source satellite data, enabling precise and cost-effective LULCC analysis. Techniques involving Landsat data are particularly prevalent in developing countries due to their accessibility, medium-resolution observations, and open data Analysis of Land Cover Dynamics in Gashaka-Gumti National Park, NorthEast, Nigeria World Journal of Advanced Research and Reviews, 2025, 28(02), 1624–1637 1625 policy since 2008 (Hirschmugl et al., 2017; Banskota et al., 2014; Herold et al., 2011). Protected areas (PAs), such as Nigeria's national parks, game reserves, wilderness areas and sanctuaries (Oyinloye and Ado, 2019; Ejidike and Ajayi, 2013) aim to conserve biodiversity (Dudley et al., 2010). However, their effectiveness is debatable, with studies indicating higher forest loss within some PAs compared to areas outside them (Leberger et al., 2020). Gashaka-Gumti National Park (GGNP), established in 1991, is Nigeria's largest protected area and a biodiversity hotspot (Aina et al., 2018; Umar et al., 2019). Despite its status, GGNP faces threats from illegal logging, agricultural expansion, and poaching, driven by population growth and poverty (Ejidike & Ajayi, 2013). Previous studies on GGNP's LULCC have been constrained by data limitations and inadequate methodologies. For instance, Gumnior & Sommer (2011) encountered inadequete cloud-free Landsat coverage, while Aina et al. (2018) used NDVI data that failed to fully capture land cover dynamics. To address these gaps, this study applies intensity analysis (Aldwaik & Pontius, 2012), a proven method for examining LULCC patterns (Sun et al., 2020; Enaruvbe et al., 2019; Enaruvbe & Atafo, 2019; Huang et al., 2018; Yang et al., 2017). The study examines LULCC in GGNP from 1987 to 2021, aiming to inform sustainable management strategies to mitigate ecosystem degradation and preserve biodiversity. 2. Materials and Methods Figure 1 River Gashaka Catchment in Gashaka-Gumti National Park, Nigeria. The River Gashaka Catchment, covering an area of 1531 square kilometers, is located between latitudes 60 94′ N to 70 39′ N and longitudes 110 39’ E to 110 88′ E within the southern sector of the Gashaka-Gumti National Park (GGNP), as depicted in Figure 1. This region provides suitable habitats for various important species, including the rare NigerianCameroonian chimpanzee, black-and-white colobus, Putty-nosed monkey, Tantalus monkey, Olive baboon, and others. Additionally, it encompasses significant park facilities such as the research center, the Gashaka Primate Project site, and popular tourist destinations like the Gangirwal or Chappal Wade (known as the mountain of death), which is the highest peak in West Africa, as well as the Selbe/Hendu highlands. Gumnior and Sommern (2012) reported these details. The area is characterized by undulating high lands, mountains, and riparian plains, with leptosols, ferrasols and acresols soils (Mubi & Tukur, 2012). Its elevation ranges from 240 m/asl. in Gashaka plains to 2,419 m/asl on Gangirwal (Gumnior & Sommern, 2012; Mubi &Tukur, 2012). The area is drained by three main streams: Mayo (River) Ngetti, Gam-gam, Gashaka and their tributaries into Mayo Kam, which empties into River Taraba, a major tributary to River Benue (Gumnior & Sommern, 2012). The river valleys are characterized by gallery forest surrounded by grass savanna mixed with herbs and shrubs (Adeonipekun et al., 2018). The mean annual rainfall in the area is about 2033 mm. the World Journal of Advanced Research and Reviews, 2025, 28(02), 1624–1637 1626 wet season lasts between April and September. The temperature ranges from a mean annual minimum of 200 C to a mean annual maximum of 31.7°C, but the higher altitudes may sometimes drop below 5°C around December, while daytime temperatures may also rise to 40°C in March (Oruonye et al., 2018). There are many enclaves of human settlements comprising various ethnic groups within the GGNP. The population in the area mostly engages in subsistence farming, animal husbandry, vocational jobs, among other occupations (Oruonye et al., 2018). The HausaFulani Jihad of the 19th century has also led to the expansion of grazing enclaves in the area, with the consequences of a growing human populations within the park, which pose an increasing threat to the ecosystem (Gumnior & Sommern, 2012). 2.1. Sources of data Table 1 Data collection and their sources Landsat(30m) Acquisition date Path/Raw Landsat 5 TM 22/01/1987 186/55 Landsat 7 ETM+ 10/01/2003 186/55 Landsat 8 OLI 19/01/2021 186/55 Table 1 shows the datasets used in this study and their sources. The shape file of Gashaka-Gumti National Park, which cuts across Path//Row 186/055, 186/054 and 185/055, was downloaded from the World Database on Protected Areas (WDPA)’s website, www.protectedplanet.net, from which River Gashaka catchment area was delimited for this study. Cloud-free, dry season Landsat images of January 22, 1987, January 10, 2003, and January 19, 2021 were acquired from the archives of the United States Geological Survey (USGS). The choice of all the images was to avoid phenological differences in land cover. GPS coordinates for various land covers were randomly collected across the study area. Some locations were, however, inaccessible because of the complex nature of the terrain in the area. Ground information for these inaccessible areas was collected using high resolution Google Earth and Planet images, complemented by existing topographical maps of the study area. The research and ICT unit of Gashaka-Gumti National Park also provided relevant base maps and other referenced materials. 2.2. Data analysis The shape file for the River Gashaka Catchment was extracted from the Gashaka-Gumti National Park through hydrological modelling of the area. This was overlaid to extract the study area from the image stacks (bands 4, 3 & 2 for TM 1987 & ETM+ 2003 and 5, 4 & 3 for OLI 2021). The field reference data, local knowledge of the study area, and Google Earth were used together to identify 5 land cover classes described in Table 2, following the classification scheme of Anderson et al., (1976). Training Signatures were sampled to represent the 5 land cover classes for each of the Landsat images, and then the Maximum likelihood algorithm was applied to classify each image into high forest, woodland, grassland, barren land, and open land. The accuracy of classification was validated by comparing the features classified on the maps with the ground-referenced data and Google Earth historical images in ERDAS Imaging 2014 then, the LULC maps were smoothened using the generalization extensions of the spatial analyst tools in ArcGIS 10.7. Table 2. LULC categories and their descriptions LULC class Description High forest Rain forest mixed with dense woodland, montane & sub-montane forest (>75 trees per ha, a minimum height of 5 m at maturity) Woodland Woodland (<75 trees per ha) with a mixture of shrub and scattered grasslands Grassland Treeless open canopy shrub savanna & grasslands Barren land Rocks, unexposed bare land/fire scars Open land Exposed soils/rocks, with active human interferences such as Settlements, roads, crop lands & fallows lands World Journal of Advanced Research and Reviews, 2025, 28(02), 1624–1637 1627 2.3. Intensity Analysis Intensity analysis is a quantitative method for examining the interactions between categorical variables from a general to a more detailed level (Kourosh et al., 2019l). Change in the study area was analyzed in terms of size and intensity using the transition matrices generated from the cross-tabulation of the time intervals, 1987-2003 and 2003-2021. The analysis revealed three levels of information, i.e., time interval, category, and transition levels. interval level reveals how size and the overall annual rate of change vary across the time intervals. It shows which time interval is relatively faster or slower when compared with the hypothesized uniform rate of change (represented by a uniform line). The rate of change is faster when it rises above the uniform intensity line and slower when it falls below it. The pattern of change is considered stationary when it remains either above the uniform line across the time interval or below it throughout the different time intervals (Aldwaik & Pontius, 2012). The category level revealed how the size and intensity of annual gross gains and losses vary across the space for each category. Annual gross gains and losses were examined and compared with the uniform intensity of change to see which category is relatively dormant or active in a given time interval. When the category’s gross gain or gross loss is either greater or less than the uniform intensity at all the different intervals then, the pattern of change is considered to be stationary (John et al., 2013; Aldwaik & Pontius, 2012). The transition level reveals the variation in size and intensity of the transition among the categories available for that transition. It shows which categories are intensively avoided or targeted when compared with the annual uniform intensity. When a gain in category m, for example, either targets category n at all intervals or avoids it at all intervals, then the transition from n to m is considered stationary, giving the gain of m. Similarly, when a loss in category m is either targeted at or avoided by n at all intervals, then the transition from m to n is considered to be stationary given the loss of m (Fuwen et al., 2019; John et al., 2013; Aldwaik & Pontius, 2012). 3. Results 3.1. Accuracy assessment of land use and land cover The accuracy of maps derived from remote sensing satellite image data processing is an important consideration when such maps need to be used as input for decision-making. The level of accuracy depends on a number of factors, including the quality of the data, the level of processing and the expertise of the analyst (Enaruvbe et al., 2019; Shao & Wu, 2008). The overall classification accuracy of the maps in this study is: 88% for 1987, 89% for 2003 and 88% for 2021 (Table 3). Both the producer’s and user’s accuracies were generally high. Table 3 Accuracy of land cover classification 1987 2003 2021 Producer’s Accuracy (%) User’s Accuracy (%) Producer’s Accuracy (%) User’s Accuracy (%) Producer's Accuracy (%) User’s Accuracy (%) High forest 100.00 100.00 100.00 93.30 96.40 96.40 Woodland 100.00 66.70 77.80 80.80 63.60 77.80 Grassland 58.30 100.00 100.00 88.90 40.00 100.00 Barren land 72.70 88.90 61.50 88.90 100.00 75.00 Open land 100.00 61.60 100.00 100.0 100.00 71.40 Overall accuracy: 1987 = 88%, 2003 = 89%, 2021 = 88% 3.2. Pattern of land cover change over River Gashaka catchment The pattern of land use and land cover change is shown in Figure 3. The figure shows that the River Gashaka catchment is dominated by high forest. Table 4 indicates an overall increase in high forest, grassland and barren lands area during the period of this study. In contrast, however, woodland and open land recorded declines during the period of this study. World Journal of Advanced Research and Reviews, 2025, 28(02), 1624–1637 1628 Figure 2 Patterns of land cover change over River Gashaka catchment Table 4 Land cover distribution over RGC, in square kilometers (km2) and percentage (%) Category 1987 2003 2021 km2 (%) km2 (%) km2 (%) High forest 797 52 621 41 943 67 Woodland 406 27 418 27 230 15 Grassland 80 5 257 17 115 8 Barren land 114 8 176 12 190 12 Open land 128 8 52 3 45 3 3.3. Intensity of land use and land cover change The transition matrices in Table 5 show the changes that occurred among the land cover classes in the two time intervals. The values in the diagonal cells remain unchanged during the time intervals, which signifies persistence, while the values that appear off-diagonal show the amount of changes from the earlier to the later year for each category (John et al., 2013). 3.3.1. Interval Level Intensity The observed rates of annual change for each time interval are shown in Figure 4 and compared with the hypothesized uniform annual change of 2.32%, represented by the red dashed line. The intensity of land cover change during the first epoch (1987-2003) was above the uniform annual change while the intensity of change for the second epoch (20032021) was below the uniform annual change. This implies that although the rate of annual land use change was fast in the first epoch, it slowed down during the second epoch. World Journal of Advanced Research and Reviews, 2025, 28(02), 1624–1637 1629 Table 5 Transition matrix for the LULC classes (sq. km.) 2003 1987 High forest Woodland Grassland Barren land Open land High forest 16790.7 4698.3 1716.0 511.8 193.4 Woodland 478.8 6998 2274.4 2184.4 194.1 Grassland 26.8 117.5 1031.0 2184.4 137.8 Barren land 568.3 485.1 1109.0 1224.6 26.4 Open land 741.4 242.8 1549.3 269.2 1006.6 2003 2021 High forest Woodland Grassland Barren land Open land High forest 6546.7 4860.5 181.5 735.9 218.6 Woodland 2515.8 1440.8 1813.7 1364.3 553.2 Grassland 701.7 214.8 1071.0 3251.0 210.0 Barren land 397.4 288.4 186.4 187.0 4910.0 Open land 6546.6 4860.5 181.5 735.9 218.6 Figure 3 The interval intensity of land use and land cover change in RGC World Journal of Advanced Research and Reviews, 2025, 28(02), 1624–1637 1630 3.3.2. Category Level Intensity Figure 4 Categorical Intensity of LULC change in RGC Figure 5 shows the category level intensity; grassland and barren land were the most active gaining and losing categories during the first epoch. The rate of land use change during the 1987-2003 epoch was 2.63%. The annual uniform intensity of land use change, however, declined slightly to 2.03% during the second epoch. We observe that open land, barren land, grassland, and woodland have been actively losing land since 1987. In contrast, barren land and grassland have been actively gaining since 1987, but while open land was dormant in the first epoch, it became active in the 2002-2021 epoch, and woodland, which was active during the first epoch, was dormant in the second (Figure 5). 3.3.3. Transition Level Intensity Figure 6-10 presents the transition level intensities. During the first epoch, there was a significant loss of high forest, mainly targeted by woodland, whose change rate was above the annual uniform intensity of 1.58% meaning that woodland was actually taking over land from high forest. But during the second epoch, the annual uniform loss for high forest became as low as 0.15% targeted by grassland and open land, while woodland dropped drastically below the uniform line. On the other hand, high forest gained from open land and barren land at a low annual uniform gain of 0.50% during the first epoch, while in the second epoch, woodland became the main target as high forest picked up to 2.05% annual uniform gains. World Journal of Advanced Research and Reviews, 2025, 28(02), 1624–1637 1631 Figure 5 Transition intensity of high forest During the first epoch, woodland loss was mainly targeted by barren and grassland, whose intensity bars surpassed the annual uniform rate of 1.94%. Meanwhile, woodland also made active gains from high forest at an annual uniform rate of 1.00%. In the second epoch, woodland was losing at a uniform rate of 1.07% mainly to high forest, but was slowly recovering from grassland and open land at an annual uniform rate of 0.33%. World Journal of Advanced Research and Reviews, 2025, 28(02), 1624–1637 1632 Figure 6 Transition intensity of woodland Figure 7 Transition intensity of grassland During the first epoch, grassland was losing mainly to barren land and slightly to open land but was highly gaining from open land, barren and woodland, with an annual uniform gain of 0.96%. In the second epoch, the annual uniform loss for grass land rose to 0.75% but was mainly targeted by open land, barren land, and woodland, while the increase in