One of the greatest conservation success stories has been the increase in protected land globally in this era of massive loss in biodiversity (Jones et al. 2018), with more than 260 000 protected areas (PAs) now covering over 15.3% of the world’s terrestrial area (UNEP-WCMC & IUCN 2024). Conservation of biodiversity is also conducted through Transfrontier Parks that encompass areas that overlap political boundaries between sovereign states, including one or more PAs (Sandwith et al. 2001). The principle of creating Transfrontier Conservation Areas (TFCAs) focuses on dissolving political and economic borders to re-establish traditional migratory routes for wildlife (Barquet 2015). Most of these PAs represent conservation ‘fortresses’ that keep wildlife in and humans out (Huang et al. 2024). However, extensive human activity, including grazing, logging, and poaching, poses threats to protected areas, undermining their role as primary defence against biodiversity loss (Rai et al. 2021).

Maintaining connectivity between PAs is important for the long-term viability of wildlife populations. Conservation corridors are widely acknowledged as effective links between habitats at a large scale (Riggio & Caro 2017). African savanna elephants (Loxodonta africana) play a significant role in shaping the conservation management objectives of protected areas (Ferreira, Greaver & Simms 2017). However, human pressures from communities living near PAs represent just one aspect of the broader anthropogenic impact on endangered wildlife at the landscape scale. Multiple threats, such as habitat loss, poaching, unsustainable extraction of natural resources, and the degradation of natural habitats (including corridors) between PAs, detrimentally affect species populations (Neelakantan, DeFries & Krishnamurthy 2019). The use of non-protected wildlife corridors by humans often diminishes connectivity between PAs, subsequently threatening endangered species (Tscharntke et al. 2012). Several of these species risk genetic inbreeding if barriers to dispersal persist (Natesh et al. 2017). Ecological corridors remain a viable management tool to maintain biodiversity at large scales and to allow species and ecological processes to track climate change (Gregory et al. 2021).

Elephants leaving protected areas threaten the lives and livelihoods of rural people living alongside these areas (Osborn & Parker 2003). Human populations bordering PAs or living in wildlife migratory corridors interact with wild animals. These interactions are mostly negative because wild animals are killed, or humans lose their lives, crops, and infrastructure (Hariohay & Røskaft 2015). Elephants avoid densely populated areas (Douglas-Hamilton, Krink &Vollrath 2005; Graham et al. 2009). Among the major conflicts associated with human landscapes is the destruction of crops.

Protected areas contain approximately 48% of southern Africa’s elephants, with well-protected and connected areas providing the best solutions for the conservation of savanna elephants and their landscapes (Huang et al. 2024). The rest of the southern Africa’s elephants are in landscapes outside of protected areas, some of which are occupied by humans. Protected area managers typically seek to conserve biodiversity, and elephants act as ecological engineers that alter several aspects of biological diversity, including microhabitat availability (Valeix et al. 2011). To ensure long-term stability of Africa’s elephants, conservation activities must recognise the importance of connectivity and availability of space for these animals (Van Aarde & Jackson 2007). Savanna elephants across the continent are listed as endangered, given the variety of threats they face (Gobush et al. 2021); however, elephants in southern Africa have shown a stable growth rate for the past quarter century (Huang et al. 2024).

Where dispersal is limited, the density of elephants may exacerbate the effects they have on biodiversity. The way elephants use landscapes depends on the distribution of water (Chamaillé-Jammes, Valeix & Fritz 2007; Loarie, Van Aarde & Pimm 2009), availability of forage (Codron et al. 2006), shade (Kinahan, Pimm & Van Aarde 2007), and resources that elephants select for at different intensities during different seasons (Young, Ferreira & Van Aarde 2009).

In southern Africa, seasonal rainfall and surface water availability strongly influence elephant movement behaviour (Bohrer et al. 2014; Loarie et al. 2009). At a landscape scale, movement beyond PA boundaries may further be influenced by corridor linkages (Douglas-Hamilton et al. 2005), avoidance of direct and indirect conflict with humans (Graham et al. 2009), and age and sex differences (Stokke & Du Toit 2002). Home ranges of bull elephants may be larger and more overlapping than breeding herds, and the movement rules for adult males may be dictated by water availability, male-to-male competition, and searching for receptive females (Smit, Grant & Whyte 2007). Vegetation parameters may also drive localised movement. In Kruger National Park, for example, Ferguson (2017) found a seasonal peak in elephant ‘excursions’ that was caused by the marula (Sclerocarya birrea) fruiting season.

Gonarezhou National Park (GNP) is surrounded by local communities within Zimbabwe and borders private properties on the eastern border with Mozambique. It is separated from Kruger National Park by a landscape settled with people and interspersed patches of wilderness areas that are woodlands undisturbed by humans. Most of the northern and western boundaries of GNP have an electrified fixed knot fence, erected primarily to control the spread of foot and mouth disease. This leaves the eastern boundary and part of the southern boundary accessible for migrations by elephants. The influence of elephants on the decline of tree species and the degradation of vegetation types in GNP has been recorded (Gandiwa et al. 2012; Kupika et al. 2014; O’Connor et al. 2024; Tafangenyasha 2001). The dispersal of elephants from GNP to neighbouring PAs such as Zinave and Banhine in Mozambique is important to avoid further habitat degradation. Currently, the elephant population in GNP has a growth rate of about 6% per annum (Dunham 2022). Considering the global conservation status of African savanna elephants, dispersal of these elephants to other less populated areas would be ideal and the best option for managing this elephant population. This study investigated the extent to which male and female elephants used landscapes outside GNP, focusing on the effects of season and land cover types on the likelihood of dispersal outside the protected area. It also examined the impact of time of day on elephant activity and landscape use in Mozambique and Sengwe communal lands that form the landscape between GNP and Kruger National Park. In addition, the study evaluated behavioural differences among elephants when ranging in non-protected areas with varying human densities.

Dispersal of study elephants out of the GNP boundary was limited during the first 3 years of the study period (Figure 2) but increased post-2020, with most movement being across the eastern boundary of GNP, into Mozambique. Results indicated increased elephant presence in the Mozambique landscape adjacent to GNP compared to the Sengwe communal area. Elephant activity outside GNP was higher in the adjacent Mozambique landscape (68.17%) than in the Sengwe communal area (31.83%). The ANOVA, using Poisson regression with a log link, showed that time of day (F_1,384 = 9750, p < 0.05), landscape type (F_1,1456 = 8293, p < 0.05), and the combined effect of both time of day and location (F_1,162 = 8132, p < 0.05) (Table 1) significantly influenced the use of the Mozambique and Sengwe landscapes by the study individuals. Chi-squared tests showed that there were significant differences in the land cover types that were utilised by elephants when they were inside versus outside the Gonarezhou (χ² = 5712, df = 7, p < 0.05). The multinomial regression established that, outside of GNP, males predominantly used Open Forest (deciduous broadleaf) habitats and reduced their use of shrublands (Table 2), whereas females continued to utilise shrub-dominated areas similar to those they favoured inside of GNP (Figure 3). The log-linear regression model showed that distances travelled by the elephants were significantly different (F_5,10770 = 133.1, p < 0.05) between all seasons for both males and females. Males travelled significantly longer distances than females during all three seasons (F_5,10770 = 133.1, p < 0.05) (Table 3), with females remaining within a 15 km radius of the GNP boundary and males travelling up to 60 km (Figure 4). Female elephants travelled significantly farther outside the GNP during the CD season (mean distance 1299.70 m, range: 1147.86 m – 1471.64 m), less during the HD season (mean distance 791.40 m, range: 564.07 m – 1110.33 m), and the least during the HW season (mean distance 502.23 m, range: 355.96 m – 708.59 m). Male elephants instead travelled approximately the same amount during the CD season (mean distance 2795.26 m, range: 2157.42 m – 3621.70 m) and the HD season (mean distance 2930.53 m, range: 1455.91 m–5898.73 m) but showed more variability during the latter. During the HW season, males travelled the least away from the GNP (mean distance 1763.37 m, range: 864.81 m – 3595.52 m).

Within the GLTFCA, GNP represents a formally protected (core) area surrounded by land use systems offering varying degrees of protection. These land use systems pose various threats to wildlife dispersal, such as poaching, with some offering no opportunities for movement because of linear infrastructure such as fences. This isolates the GNP elephant population, likely contributing to the consistently high growth rate (Huang et al. 2024).

The increased use of Mozambique by elephants post-2020 could have been influenced by the provision of artificial water (Tshipa et al. 2017), increased law enforcement activities (Demeke 2003) in some private concessions in Mozambique, and a reduction in trophy hunting during the coronavirus disease 2019 (COVID-19) pandemic (Tucker et al. 2023). Elephants spent more time in Mozambique than in the Sengwe region, based on location data across all years and seasons combined (Figure 2). The study elephants mostly dispersed to Mozambique through wildlife concessions. Some of these elephants ranged farther into the Massangena and Chicualacuala districts, both having a human population density of approximately 2 persons per square kilometre (see Online Appendix, Table A1-3). Use of the Sengwe area by elephants was proportionally lower (31.83%), possibly because of human presence in the region. In the Sebungwe region, Hoare and Du Toit (1999) found that elephant density was unrelated to human density until a threshold of human density of about 15.6 persons per km² was reached, after which resident elephants effectively disappeared from the area. The Sengwe communal area is part of the wildlife corridor that connects Gonarezhou NP and Kruger NP, and the movement of elephants between these two protected areas was also recorded by Henley et al. (2023). Our results suggest that the elephants under study avoid areas inhabited by subsistence farmers, and the majority of elephant location data in Sengwe were recorded at night (Figure 5). However, when elephants do move into communal lands (typically at night), it suggests the possibility of crop-raiding behaviour.

Graham et al. (2009) found that elephants had a preference for communal areas with croplands at night, rather than during the day. Past studies show that savanna elephants use the cover of the night to move onto cropland where they raid crops (Bell 1984; Hoare 1995). We were unable to demonstrate this during our study, as the elephants spent relatively little time in areas dominated by crop fields (Figure 5). Nevertheless, the risk-avoidance strategy of preferred nocturnal movement by elephants in the Sengwe communal areas (Figure 5) appears to be consistent with other research carried out on temporal patterns of elephant movement relative to unprotected and protected areas (Ihwagi et al. 2018; Wittemyer et al. 2007).

Our results showed that males had a higher probability of dispersing outside the protected area and travelled farther than females (Ngene et al. 2010). This suggests that there may be an opportunity for elephant dispersal between large PAs in the region because the long-distance movements can show potential migration routes (Huang et al. 2022). The GNP has several linear barriers, which include fences and roads that surround the park that could be limiting female elephants because they travel in breeding herds with calves that may be unable to cross barriers (Naidoo et al. 2022). Male elephants in other parts of Africa are known to break fences more frequently than females (Slotow 2011), suggesting that their movements are less limited by any form of barrier. Movement outside the GNP by study elephants was in areas with no fences, including the 100 km long border with Mozambique and a small section of the south-eastern corner of the GNP.

The distribution patterns of elephants can be influenced by fear landscapes as they avoid areas with threats to their safety (Cook & Henely 2019; Cromsigt et al. 2013). They facultatively alter their behaviour to avoid risk in areas dominated by humans, and that helps them to maintain connectivity in fragmented land use systems, possibly lessening some of the potential negative impacts of fragmentation (Graham et al. 2009). Evidence shows that female elephants, in family units, take fewer risks and do not travel far from PAs (Wall et al. 2021), which suggests that elephants perceive areas farther from core protected zones as increasingly risky because human activity is a major cause of mortality and injury in adult elephants (Obanda et al. 2008). This was not demonstrated in our study. Females that remained close to the PA demonstrated possible risk avoidance behaviour (Mortimer et al. 2021) and considering how complex and dynamic anthropogenic mortality risk is in human-dominated landscapes, varying distances from PAs are perceived differently as the ability to quickly escape into a safe zone diminishes with increasing distance from a PA (Chiyo et al. 2014).

Our results showed that elephants preferred to be in the park during the HD and HW seasons, and as the seasons progressed from CD to HD to HW (cool to hot), the probability of elephants dispersing outside the park significantly decreased, indicating how the PA was used as refugia during the hot months. A study of elephants from northern Kruger also found that elephants moved to communal areas in Mozambique and Zimbabwe more during the dry season than the wet season (Cook, Henely & Parrini 2015). The findings of our study suggest some habitat drivers; however, overall movement outside the GNP appears to be driven by the need to access forage, water, and potential mates (for bulls) (Bohrer et al. 2014; Owen-Smith et al. 2020). Our study found that, outside the GNP, male elephants preferred forested areas with deciduous broadleaf vegetation, likely as a strategy to access forage and possibly for concealment (Graham et al. 2009).

Elephants are sensitive to seasonal variation in potential anthropogenic mortality risk, as hostile human-elephant interactions tend to be more frequent in the dry season compared to the wet season (Kangwana, Moss & Croze 2011). Contrary to these findings, elephants in our study were unlikely responding to increased potential anthropogenic mortality risk in the dry season, as it is the season when they utilised areas outside the GNP the most. It may also be that surface water was less available in GNP during the dry season, as there was no artificial water provision in GNP. The probability of our study elephants leaving the GNP was lowest during the wet season, likely because the benefits of using areas with livestock and human activity are minimal when water and forage are abundant within the PA (Chiyo et al. 2014).

Study elephants moved into wildlife concessions in Mozambique to access forage and provisioned water, and anecdotal reports from personnel managing the wildlife concessions in Mozambique neighbouring GNP indicated increased use of their water points by elephants from GNP during the dry season (Werno Drinkwater, pers. comm). Alternatively, they may have been responding to dry season resource distribution, which is likely to consist of a few discrete foraging patches, for example, around swamps, water sources, and patchily distributed Senegalia and Vachellia browse (Stokke & Du Toit 2002). This explanation is consistent with the behaviour of an elephant bull that travelled about 60 km to the same patch of land in Mozambique close to Banhine National Park for three consecutive years, returning to GNP at the beginning of the HD season. This elephant bull might have been attracted by marula fruits, as similar seasonal movements were recorded in the GLTFCA during the marula (Sclerocarya birrea) fruiting season (Ferguson 2017).

The diurnal and nocturnal behaviour of the study elephants in Sengwe demonstrated risk-avoidance behaviour that moderates connectivity by limiting the movement of elephants in the area to mostly at night. Ensuring that the corridor between Gonarezhou and Kruger NP is functional could enhance the persistence of elephants in the Sengwe landscape, but our data suggest that most elephants will avoid this area until a corridor allows for movement. Disturbance of wildlife associated with human population density remains relatively high across much of the Sengwe landscape, although not as high on the Mozambique side. Intensification of agriculture in the southern regions outside of GNP is leading to fragmentation and associated isolation effects; however, two collared elephant bulls did travel from GNP to Kruger National Park and then returned, but such movements may represent a small proportion of the elephant population resident in GNP. Moreover, those elephants that dispersed into Mozambique did not manage to connect with either Zinave or Banhine National Parks. Consequently, the identification and conservation of appropriate wildlife corridors between GNP and KNP should be a priority for the GLTFCA. Our results aid with the understanding of corridor uses by elephants that could influence elephant persistence in both the protected and non-protected areas within the GLTFCA landscape.

This study illustrated how elephants from GNP interact with landscapes outside of the PA. The only remaining open areas that allow the movement of elephants from GNP into the GLTFCA landscape are to the east, where the GNP shares the 100-km boundary with Mozambique and a section of the south-eastern corner of the GNP, which gives elephants access to the Sengwe Tshipise Wildlife Corridor. Some of the study elephants travelled into Kruger NP, but most of the observed movement was into Mozambique, although none of these connected with any of the PAs there. Conservation planning for functional connectivity within this region of the GLTFCA should consider ways of encouraging the movement of elephants from GNP to Zinave and Banhine National Parks in Mozambique if this landscape is to remain connected. A section of the communal areas adjacent to the GNP that is being cleared of land mines could offer a potential functional corridor through the Sengwe. To improve connectivity, there is a need to consider practical steps that include non-lethal deterrent strategies for mitigating human–wildlife conflict (in consultation with local communities) and also addressing socio-political barriers such as fences and settlement patterns that limit the movement of wildlife. The landscape could benefit from further studies that explore the long-term impact of creating corridors, the role of artificial water provision within non-protected areas, and strategies to reduce fear landscapes.