1 Introduction

Understanding wildlife home range and habitat utilization is critical for species conservation and management, especially with the increase in human activities (i.e., habitat encroachment) and escalating impacts of climate change [1, 2]. Home range is the area in which an animal lives and moves on a continuous basis. This area is a fundamental concept of ecology which is used to characterize space use patterns of different species. Home range is also defined as the total area required to meet nutritional and reproductive needs throughout the year [1, 3]. Migration information, habitat preference and interactions may also aid in evaluating population dynamics in both spatial and temporal contexts [4, 5]. Derived information on movements of large herbivores such as African savanna elephants (Loxodonta africana) can also be applied in identifying disease interface for potential transmission between wildlife and domesticated animals [5], human wildlife conflict zones [6] and migration corridors [5, 7]. Such information can also be important in ecological restoration efforts [4]. In sub-Saharan Africa, valuable habitats for elephants have been impacted by the increasing human activities coupled with effects of climate change, inter alia, extreme weather events characterized by droughts and floods. These have led to habitat fragmentation and loss, resulting in extirpation of wildlife populations [8, 9].

Large herbivores, such as elephants, identified as keystone species can be used to improve wildlife monitoring and management [5, 10]. Home range for large species is usually indicative of the spatial extent of the minimum area required to be under protection as these species transverse vast areas to meet their habitat requirements such as food, mates, and water for biological fitness [2, 11]. Threats, such as poaching, illegal wildlife trade and human encroachment, make it urgent for species conservation. According to the International Union for Conservation of Nature (IUCN) Red List, the population for African elephants declined by 118,000 between 2007 and 2016 [1], as such, this species is classified as near threatened (https://www.iucnredlist.org/). However, African elephants remain important to tourism, cultural and socioeconomic spheres in sub–Saharan Africa.

A variety of factors have been shown to best characterize variations in home range size in different species [1, 12]. Variations in home range at the population and individual level have been attributed to intrinsic (e.g., age, sex) and extrinsic factors (e.g., resource availability, climate, season, terrain) [13, 14]. The age of elephants significantly influences movement patterns, with older individuals leading herds and guiding younger ones through migration routes, habitat selection, and social interactions; while sex also influences movement, with females leading and remaining in their territory, while males often disperse and travel longer distances [14]. The availability of food, water, and other essential resources significantly influences the movement patterns of elephants, driving them to migrate between habitats and adjust their routes in search of sufficient sustenance and suitable living conditions [13]. The terrain and topography, including slope, elevation, and vegetation density, can significantly impact the movement patterns of elephants, with them often preferring to follow established pathways and avoiding areas with difficult or impassable terrain. Viana [15] and Alfred [16] highlighted that among these factors, seasonality has a large influence on home range selection and size as it influences the availability of resources. Thus, the movement of elephants will differ depending on a particular season as the amount and distribution of food, breeding space, and water depend on season. In addition, human driven activities such as farming, hunting, and wildfire outbreaks also depend on seasons [6, 9, 17]. Understanding how various factors impact home range size and space utilization is important for informing conservation and developing management strategies for vulnerable species such as elephants which move long distances in a landscape [5, 6].

Detailed studies employing Geographic Positioning System (GPS) tracking to understand home range and habitat selection for elephants are generally lacking in Zambia. Chibeya [18] studied African elephants resource utilization in Sioma Ngwezi National Park (NP) with a particular focus on habitat connectivity, while Squarre [19] used GPS tracking to understand the spread of trypanosomes in the Kafue Ecosystem. Additionally, O’connor [20] focused on geographic range for giraffes across Africa in which datasets from Zambia were also incorporated. It is clear from this background that there is still a dearth of information on home range size and resources utilization of elephants that employed GPS tracking systems in Zambia.

The objectives of this study are to: (1) establish seasonal home range sizes for elephants, and (2) assess seasonal variation on habitat selection using different factors. It is important to note that these objectives were addressed by using data generated from two collared elephants and these were sub-adult bulls which were linked to the maternal herd. The results from this study will bring new insights that can be used by wildlife managers and policymakers to improve conservation of elephants in West Lunga NP. Here, we hypothesize that the home range size for the hot dry season would be larger than that of the wet season and the cold dry seasons. It is expected that the elephants cover long distances in search of water and food during the hot dry season compared to the wet season.

2 Methods and materials

2.1 Study site

This study was conducted in the Kabompo Landscape (total area = 9532 km2) of North-Western Zambia (Fig. 1). Located at central geographic coordinate of 12.78°S and 24.75°E, this area is mainly characterized by the West Lunga NP and four Game Management Areas (GMAs); Chibwika-Ntambu, Lukwakwa, Musele-Matebo and Chizera. The study area is found in Zambia’s Agro-ecological Zone III which is characterized by rainfall ranging from 1000 to 1500 mm per annum and the soils are sandy and well drained [12]. The mean annual temperature is about 22 ℃ (Fig. 2), with coldest temperatures dropping to 19 ̊C, while temperature can reach 40 ℃ in hot season [22]. The Kabompo Landscape is mainly characterized by the evergreen Crystosepalum forests which form the largest evergreen forest outside the equatorial region [23]. Patches of miombo woodland and grasslands also exist in the southern and western side of the park. The population around the park is relatively low (< 7 people/km2) and these people practice subsistence farming, small scale fishing and traditional hunting [24].

Fig. 1
figure 1

Location of the study area, the Kabompo Landscape, in the North-Western Zambia. The study area is characterized by the West Lunga NP and four GMAs. This region has limited tourism activities because of inadequate facilities as this area is about 300 km from major towns such as Solwezi. The management of the park has been enhanced by the concerted efforts from the new mines (Lumwana, Kansanshi and Kalumbila) which have been opened in the province. The major conservation goal is to improve the wildlife population for this national park which was once depleted

Full size image
Fig. 2
figure 2

Weather pattern based on mean monthly rainfall and temperature in the study area, Kabompo Landscape [22]. The study area is in Agro-ecological Zone III of Zambia which receives the highest levels of rainfall (above 1500 mmm per annum). Rainfall occurs between November and April, with the peak period being between December and February. The highest temperatures are normally in October, while temperature drop to its lowest levels in July

Full size image

The animal species found in this area include elephants, hippopotamus (Hippopotamus amphibius), buffaloes (Syncerus caffer), yellow-backed duikers (Cephalophus silvicultor), and lions (Panthera leo). The park is bounded by the Kabompo River to the east and south, and by the West Lunga River to the west. The national park was depleted of major wildlife species in the past due to poaching; however, it is currently undergoing species restoration. Between 2007 and 2013 the population of elephants was estimated to be between 600 and 500; however, due to poaching and encroachment the population declined to less than 100 in 2016 [21]. Due to different conservation activities which are currently being implemented in West-Lunga National Park, the population is expected to increase [9].

2.2 Methods

2.2.1 Datasets from collared elephants

Two male elephants from the same maternal herd were captured and fitted with GPS transmitters on 1st September 2018. These elephants were from the sub-adult class with estimated age range of 10–15 years as such these elephants were found among the maternal herd and their movement were not influenced by musthing as they were not fully mature. Thus, these two elephants were used as proxies of the maternal herd to understand the home range for the whole range. The Africa Wildlife Tracking GPS collars (Pretoria, South Africa) that uses the Inmarsat satellite system to upload GPS fixes to an internet site was used in this study (see Supplementary Material). All the capturing and collaring were performed by veterinarians from the Department of National Parks and Wildlife Service (DNPW) and staff from the West Lunga NP. Spatio-temporal data were recorded at an average interval of one hour and the dataset included GPS locations recorded from 1st September 2018 until 25th March 2021. Additional information including temperature, altitude (elevation above sea level), speed, date and time were recorded for each location. The location data were originally in World Geodetic System 84 (WGS 84) which was then projected to Universal Transverse Mercator Zone 35 (UTM 35). In our analysis, we separated the data into three seasons; hot dry (August to November), cold dry (May to July) and wet (December to April) seasons as these are the main seasons experienced in Zambia [25].

2.2.2 Auxiliary datasets

In this study, we also used other datasets for exploring factors that influence home range size and habitat utilization (Table 1). This dataset includes Shuttle Radar Topography Mission (STRM), digital elevation models (DEMs), Sentinel 2 satellite images [26], layers for rivers and official boundaries for the NP and GMAs. Additionally, we also used MODIS active fire product (https://firms.modaps.eosdis.nasa.gov/map/) to understand the wildfire in the study area. The DEMs were used to derive topographic variables, such as slope and elevation, while Sentinel 2 image was used to generate a land cover map and normalized difference vegetation index (NDVI). We used vector layers for rivers and roads to derive the Euclidean distances to estimate access to water sources and proximity for development, respectively.

Table 1 Details of the factors associated with home range selection that were assessed in this study
Full size table

To extract the spatial dataset for these factors, we used ArcGIS Pro’s (Esri, Redlands, California, US) spatial analysis tool to generate 200 random points. A minimum distance of 500 m between points was maintained, preventing clustering. The points were then allocated proportionally to the three home range categories: 20% for the core area (KDE50%), 30% for the KDE95%, and 50% for MCP100%. Finally, we extracted the actual dataset by assigning the corresponding values to each point using the Extract Values to Points tool under Spatial Analyst. Prior to the analysis, we employed the Global Moran’s Index [27] in ArcGIS Pro to test if the dataset exhibits signs of spatial autocorrelation. The Moran’s Index was less than zero, whereas the p-value was greater than 0.05, indicating that the data were randomly distributed. Although the data did not show some autocorrelation, clustering occurred; however, the dataset was a combination of both random and clustered data as such this could be cancelling the net effects of clustering.

Data preparation included land cover classification using Sentinel 2 images, extracting spectral indices, deriving Euclidean distances for rivers and roads, extracting slope and elevation from DEMs. We employed Google Earth Engine (GEE) to conduct the land cover classification for six land cover classes including primary (Crystosepalum) forest, secondary forest, grassland, wetland, water, and cropland. In GEE, the analysis of all available remotely sensed images is done on the web-based code editor without downloading these datasets. For the MODIS fire dataset, we employed point density data processing to interpolate the probability of wildfires occurring for specific areas within the study area. In addition to GEE, we utilized ArcMap10.7 (ESRI, Redlands, California, USA) in processing home range data.

2.2.3 Home range analysis

It is important to note that during analysis, we combined the data for the two datasets because the two elephants were from the same herd. In addition, after conducting a visual assessment on the dataset, the movements of the two elephants were similar. We employed both the Minimum Convex Polygon (MCP) and the Kernel Density Estimator (KDE) approaches to establish the seasonal home range size in ArcMap10.7 [16, 31]. Although the MCP method is thought to overestimate space use [5], it is useful here to enable comparison with other studies and to estimate the maximum (100%) area needed for the conservation of elephants. We also used Kernel-based area estimators to determine home range and core area sizes because the kernel method produced more detailed results [31]. For home range, we calculated the 95% probability contour using the least-squares cross validation procedure to determine the smoothing parameter for each kernel distribution, while 50% probability contour of the kernel method was used to calculate core areas.

2.2.4 Home range selection—multinomial logistic regression

We used a multinomial logistic regression [3, 32] to test the influence of different factors on home range selection. The major assumptions for logistic regression which were tested include linearity, outliers, independence, and multicollinearity. The analysis was based on the three seasons and different home range sizes including KDE50%, KDE95%, and KDE100%. The factors used here include elevation, NDVI, land cover, distance to rivers and fire occurrence. All the statistical analysis (see the code under Supplementary Materials) were conducted in R statistical software using the ‘nnet’ packages [33].

3 Results

3.1 Home range size

The MCP spatial extent for the movement of the two elephants showed the selective use of the home range. Figure 3 shows that the two elephants were mostly found on the southern part of West-Lunga National Park. The distribution also showed variation on the seasonal home range size and location (Fig. 3). During the cold-dry season, the elephants restricted their movement more towards the southern side of the park, while during hot-dry and wet season, they extended their movement to northern side of the park. It is important to note that the location points were only confined within West Lunga NP, indicating that the two elephants utilized resources within the national park regardless of the season. The results presented here were combined from the two elephants so that information of the movements of the maternal herd could be indirectly assessed.

Fig. 3
figure 3

The seasonal home range of elephants in West Lunga NP between 2018 and 2021. The red boundaries indicate the MCP-derived home range. The elephants utilized the southern part of the park during the cold dry season, while this area extended towards the north during the hot and wet seasons. Generally, the elephants exploited open areas which include grasslands and wetlands more than they accessed the Cryptosepalum evergreen forests

Full size image

For both the MCP and KDE method, the total area utilized by the elephants varied among the three seasons (see Figs. 3 and 4). The MCP showed that the elephants used a large area in the wet season, and the area was smaller for the cold dry season. For the KDE, these two elephants also used the largest area during the wet season; however, this was not different from the area for the hot dry season. The area estimated by MCP was three times larger than the area estimated by the KDE, emphasizing the large disparities between these methods.

Fig. 4
figure 4

The home range size based on the KDE approach indicates the 95% area in yellow, and the KDE 50% core areas in red. The elephants utilized a larger space during the wet season which is mainly in the western apart, while the home range extended towards the southern part of the park during the dry and cold seasons. During the wet season, the KDE 50% core areas were spread apart compared to the clustering which was exhibited during the other seasons

Full size image

The wet season had the largest core areas compared to the dry and the cold dry season. The core areas for the wet season (79.5 km2) and the hot dry season (62.1 km2) were two times larger than the core for the cold dry season (30.5 km2). This indicates that during the cold dry season the elephants are more confined to small areas than the other seasons. This was in line with our hypothesis that during the wet season, elephant movements increased due to the available resources which they can exploit on a broader scale. The spatial location of the core areas also varies; during the cold dry and the hot dry season have many locations in the central part of the park, while the core areas for the wet season were stretched on the western part of the park (Fig. 4). Figure 5 illustrates the area for the home range based on the MCP 100%, KDE 95% area and the KDE 50% (core area).

Fig. 5
figure 5

The area for the home range based on MCP 100%, KDE 95%, and KDE 50% (core area). The cold season has the smallest home range coverages for both the MCP and KDE, while the elephants utilized large areas during the wet season

Full size image

3.2 Factors affecting home range selection

Table 2 illustrates the effects of different factors on home range selection based on three categories of the home range: (1) core areas (KDE 50%), KDE 95%, and MCP 100%. It is important to note that the results are present in two forms: (1) factors with continuous values are presented in Table 2A, and (2) a categorical factor—land cover is presented in Table 2B as it required multi-level analysis. For all these categories, elevation and land cover types were significantly affecting home range selection in all the three seasons (p-value < 0.05, Table 2). NDVI was a key factor in the cold and wet season as the elephants prefers certain vegetation cover during these seasons, while fire intensity had much influence during the hot season, perhaps because of increasing fire activities. Besides the distance to rivers, movements of the large elephants in the hot dry season were also significantly affected by the distance to roads and the fire intensity.

Table 2 Results for the multinomial logistic models which were applied on different factors based on a single level factor (Table 2A) and multi-level (Table 2B)
Full size table

Elevation exhibited a seasonal influence on home range selection, positively impacting choices during the wet season, but negatively during the cold and dry seasons (see Table 2A). Similarly, slope had a positive effect during the wet season, as elephants sought higher ground. In contrast, distance to rivers showed a positive influence during the cold and dry seasons, but not during the wet season. Fire intensity and distance to roads negatively impacted home range selection during the hot dry season, while NDVI had a consistently positive influence throughout the year, likely due to its correlation with habitat suitability. At the multi-level factors level (Table 2B), primary forests had a high negative impact during the cold and hot seasons, possibly due to elephants moving towards the north, characterized by high elevation and dense forests, during the wet season. Grasslands had a positive influence throughout the year, except during the wet season, whereas croplands only attracted elephants during the wet season.

4 Discussion

In this study, we report the first findings on home range size for elephants in Zambia’s Kabompo Landscape. Although this study is only based on two collared elephants which makes it difficult to generalize the results [2, 32]; important lessons can be drawn. First, the visual analysis showed that the collared elephants were mainly restricted to the park because the West-Lunga National Park is located in between two major rivers (maximum width 1 km) that are creating a natural barrier to the elephants. The rivers are covering about 165 km around the park which is 80% of the boundary such that only the northern part is not bordered by the river (see Supplementary Material). Thomas [5] indicated that natural and artificial barriers inhibited the movements of elephants in Kruger National Park. Secondly, the home range size for the wet season was larger than that for the cold and hot dry seasons, signifying that these elephants were more mobile during the wet season than in the hot and cold season—perhaps because the exploited the grassland and other habits which become suitable during the wet season.

The increasing movement during the wet season can also be associated with increasing vegetation during the rainy season such that these elephants access areas which are seemingly degraded during the cold and hold seasons. During the wet season, the home range was stretched to the northern part where these elephants exploited part of the evergreen Cryptosepalum forests; however, the elephants utilized the grasslands in the wetland move than the evergreen forest in the north. Thomas [5] showed that elephants utilized grassland and open areas in the wet season more than the closed canopy areas in Krugar National Park of South Africa. Generally, the home range size and habitat utilization vary with seasons [5, 34] and it is also affected by different factors, such as levels of disturbance, access to water and vegetation type. However, it is a common phenomenon that home range for the wet season are larger than that for the cold and dry seasons due to increasing availability of resources (e.g., food and water) and reduction in human induced impacts [13, 14].

Generally, the findings from this study are key in improving the management strategies such as routine monitoring and patrols which are usually integrative—regardless of a species. Whyte [35] indicated that movements of elephants tend to follow the same corridors over the years. Hence, it is recommended to establish the home range size based on the data collected for more than two seasons. The MCP over the three seasons ranged from 900 to 1700 km2 with minimal movements in the cold season and large home range during the wet season. Mlambo [36] indicated a smaller home range of between 176 and 2400 km2 in dry woodland of Hwange National Park in Zimbabwe– similarly the elephants had the largest home range during the wet season. In his study, an adaptive local convex-hull approach was used on two bulls. In a separate study, Ngene [31] reported a larger home range of up to 10,000 km2 using the KDE approach in Amboseli ecosystem in Kenya which is dominated by miombo woodlands. Here, 9 bulls were collared. The variation in home range size in past studies [5, 13, 36] is mainly attributed to the different approaches used in recording location and other environmental factor which are specific to each study area.

The MCP approach was employed to understand the extent of the outer boundaries of the home range; however, interesting results from this study were highlighted from the KDE approach [1, 8]. It was clear from the analysis that the home range size for the cold dry and hot dry seasons was located on the southern part of the park and was smaller than that of the wet season. The southern part is generally characterized by grassland or open areas which is usually flooded during the wet season. Chibeya [18] indicated that water availability is one of the top factors for animal movements and home range selection. In our study, core areas for the elephants are located near water sources, especially during the dry season leading to minimal movements of elephants. The KDE analysis also indicated that the elephants were moving up north during the wet season and utilized part of the ever-green forest. This pattern of movement can be attributed to the increased elevation which is associated with the northern region and the possibility of less flooding during this period. These findings are in line with Chibeya [18] who indicated that the elephants uses more of the grassland during the wet season in the Sioma Ngwezi Landscape and they exhibited increasing movements. In our study area, the grasslands are seasonal wetlands as such it is expected that the elephants utilize most of these grasslands across the seasons.

The results from this study indicated that animal movements and habitat selection depend on several factors, such as elevation, land cover type, slope and distance to water sources. It is important to note that although this study had a focus on male elephants, these were sub-adults which were closely linked to the maternal herd. Hence the findings here can be associated with female elephants. Similar to Ngene [31] elevation and distance to rivers was the common factors determining home range size and selection across the season. It is common ecological behavior for elephants to avoid flooded areas; however, areas in the seasonally flooded grassland are commonly utilized during the dry season. Elephants also avoid getting too close to settlements and roads, perhaps because of the increasing human disturbances, high risks of poaching and the outbreaks of seasonal fires, mainly in the cold dry season. In this study, fire intensity and NDVI were significant in the cold dry season indicating that home range is dependent on areas with minimal wildfire activities and will open forest canopy. During the hot dry season, distance to rivers, fire intensity and distance to roads were the major factors for home range. Interestingly, during the wet season topographic factors (elevation and slope) and distance to rivers were the most influential factors. This can be attributed to the fact that these large elephants tend to occupy high ground which are located away from rivers [2, 31].

Land cover type determines the type of resources elephants use during each season. The findings from this study indicated that land cover type was closely associated with habitat selection in cold and hot dry seasons. It was expected that the land cover types would not have much effect during the wet season as most of the vegetation flourished during this season and thus food is usually in abundance for elephants [5]. Our study indicated that during the cold and the hot dry season, the home range was dominated by grassland or open forest land cover types, while during the wet season these elephants were retreating to the thick forests on the northern part of the park. Depending on the season, the movements of elephants differ [35, 37]. In the dry season it is expected that they will retreat and explore areas close to the rivers; during the wet season when vegetation grows the elephants could even explore areas close to human settlement or even feed on crops [7, 36].

The study also revealed that environmental factors influence elephant home range selection both positively and negatively. Elevation and slope positively impact selection during the wet season, while distance to rivers is more important during cold and dry seasons. Fire intensity and proximity to roads negatively impact selection during hot dry seasons [18, 31]. NDVI, a measure of habitat suitability, consistently positively influences selection. On a broader scale, primary forests are avoided during cold and hot seasons, possibly due to elephants seeking higher elevations and denser forests to the north [5]. In contrast, grasslands are preferred year-round, except during the wet season, and croplands attract elephants only during the wet season. These findings highlight the complex interplay between environmental factors and elephant habitat selection, which varies across seasons and spatial scales [13, 36].

This study makes several contributions towards the management of elephants in the Kabompo Landscape and other areas with similar environmental conditions as it highlights the seasonal movements of large elephants. The findings give a clear picture for planning, monitoring, and wildlife protection across this landscape. The Kabompo Landscape, like other conservation areas, is under threat from poaching and impacts of climate change [9, 18]. Thus, our results present opportunities on where to focus monitoring and protection efforts for the elephants in the study area. Although there is no clear indication for these elephants exploiting the areas outside the national park, anecdotal evidence confirms that the elephants occasionally move across the GMA, especially during the wet season when agricultural crops are available. Insights from this study has the potential of reducing the human-wildlife conflict [6, 9] as there would be increased awareness among managers and the local communities about the corridors for the elephants.

The findings from this study need to be applied by taking into consideration the limitations encountered during the research process. Firstly, only male elephants from the sub-adult class were studied without considering the female and mature bulls. As such our finding are closely linked to female than mature bulls which are influenced by musthing. Secondly, a small sample size of two elephants were collared in this study which was combined. Therefore, this study gives a snapshot, yet important information, on what to expect on the movement and resource utilization of elephants. We recommend collecting data from larger sample sizes to understand the overall movements in the park and focusing on the home range for individual elephants. Third, expanding the study period beyond the three-year period used in our study may provide more insights on the periodical movements and their associated variations of these elephants. Finally, a detailed study on habitat selection is required to give a clear profile on habitat preference by the elephants in the Kabompo Landscape.

5 Conclusions

This study shows that the elephants were mainly confined to West-Lunga National Park, with minimal movements crossing the boundaries. This confinement pattern was largely affected by the natural barriers (Kambompo and West-Lunga River) which define the boundaries of the West-Lunga National Park. Both approaches showed that the home range size and resource utilization varied across season, with the largest home range being in the wet season, while these elephants were confined to small patches during the cold dry season as the home range compressed. The large home range during the wet season can be attributed to the increasing distances these elephants traverse due to increasing suitable habitat during the wet season as there is increasing vegetation growth, while the smaller home range size during the cold and hot seasons can be attributed to reduction in available resources, increasing human and wildfire threats. The findings also showed that the home range stretched up north towards the evergreen forests during the wet season. This is attributed to the high elevation and low slopes on the upper part of the park which is characterized by the occurrence of less wetlands.

It is important for wildlife managers to develop an understanding of elephant behaviour and their use of space based on season to improve monitoring. As conservation of these elephants involves understanding their corridors, establishing the seasonal locations of these corridors is key in accurate monitoring of these important resources. As this study covered more than two seasons, we were able to provide insights for within and between seasonal movements. Decision-makers must explicitly understand the spatial–temporal dynamics for the movements of large elephants to develop accurate strategic management strategies, including restoration projects in wildlife depleted areas. Wildlife managers and local communities are faced with different challenges, such as human-wildlife conflict, and as such, home range-related information can be applied to reduce these conflicts.