A pan-African spatial assessment of human conflicts with lions and elephants

The study addresses escalating human–wildlife conflict in Africa, particularly involving African lions and elephants, set against unprecedented biodiversity loss, habitat conversion to agriculture and livestock, and shrinking wildlife ranges. Human–wildlife conflict results in human injury/death and significant economic losses to crops and livestock, reducing local support for conservation and prompting retaliatory killings. While multiple mitigation tactics exist (e.g., deterrents, improved husbandry, compensation), their large-scale effectiveness is mixed. There is a contentious debate about fencing: although criticized for disrupting movements and connectivity (particularly veterinary fences in southern Africa), properly deployed mitigation fences can reduce conflicts and support wildlife persistence where budgets allow. The research aims to: (1) map where conflict risk is highest along the perimeters of current and extendable ranges of lions and elephants across Africa using human population, cropland, and cattle density; (2) evaluate socio-economic and political correlates of population sizes; (3) assess ecological risks (fragmentation) for migratory mammals; and (4) estimate the economic return on investment of constructing and maintaining high-standard mitigation fences to reduce cattle loss, crop damage, and human injury/death. The goal is to identify where fencing could be a financially sustainable and ecologically responsible strategy to mitigate conflict while minimizing negative conservation impacts.

The paper synthesizes prior work indicating: rapid declines and range contractions of large mammals due to habitat loss, unsustainable hunting, and conflict; mixed effectiveness of non-fencing mitigation strategies; evidence that adequate funding and high-standard fencing can stabilize or increase populations in some areas; and concerns that poorly planned fences (notably veterinary fences) have historically fragmented habitats and blocked migrations. It highlights the need to shift from generic pro/anti-fence positions to identifying where mitigation fencing provides net benefits to people and wildlife while limiting ecological costs.

Data and preprocessing: All spatial data were converted to vectors and managed in PostgreSQL 11.4 with PostGIS 2.5; Python 3.7 and R 3.6.0 were used for analyses. Human pressures included: GPWv4 human population density (30 arcsecond, extrapolated to 2020); Gridded Livestock of the World (2010) cattle density (0.083333°); and Copernicus Global Land Cover cropland fraction (100 m). Species distributions: Updated IUCN range maps and site-based population sizes for lions (2018) and elephants (2016). Additional IUCN Red List ranges for terrestrial mammals (Cetartiodactyla, Perissodactyla, Primates, Carnivora) were used for fragmentation assessment. Protected areas: WDPA (May 2019), merged to avoid overlaps, limited to IUCN categories Ia–IV. Extended ranges and conflict risk mapping: Protected areas containing or adjacent to lion/elephant ranges were merged with species ranges to create extended range layers to reflect potential restoration areas and avoid unnecessary fragmentation. Human pressure rasters were thresholded to their top decile values per variable, producing a conservative map of high human population, cropland, and cattle densities. Conflict risk classes were defined where these highest decile layers overlap: severe (all three: human, crop, cattle), high (human plus either crop or cattle), moderate (crop plus cattle), and low (only one of the three). Conflict risk lines were identified along extended range perimeters and areas within 10 km of the edge, reflecting wide-ranging behavior and known decay of conflict with distance from protected boundaries. Sensitivity and certainty: Robustness to commission/omission errors was tested by Latin hypercube sampling, randomly varying distances between conflict lines and human pressures within ±10% across buffers of 10, 20, and 30 km. Certainty of lion/elephant presence was inferred where range perimeters overlapped protected areas. Statistical analyses: Generalized linear mixed models with negative binomial error and log link examined factors affecting site-level lion (n=77) and elephant (n=191) population sizes. Country was a random intercept. Fixed-effect predictors (from Supplementary Table 1/8) were screened to avoid multicollinearity (retain variables with pairwise r<0.7). Model selection used Bayesian Information Criterion; predictor importance was derived by summing weights across the six best models. Multi-model averaging used the R package glmulti. Predictive performance was assessed via 10-fold cross-validation with 1000 bootstrap iterations. Range fragmentation analyses: For 20 migratory mammal species whose ranges intersect severe-risk areas, perimeter length-to-area ratios were computed before and after introducing proposed mitigation fences (with a 20 m buffer for maintenance clearances). Lower ratios indicate more compact, connected ranges; increases imply fragmentation risk. Economic analyses: Economic return was estimated using Equivalent Annual Annuity (EAA), computed from Net Present Value (NPV) over a 10-year horizon with country-specific discount rates. Benefits comprised reduced cattle loss due to lions and reduced crop damage due to elephants in 10 km buffers adjacent to severe/high-risk segments. Assumptions: 1% cattle loss rate; regional average adult cattle weights (West 262 kg, Central 281 kg, East 283 kg, Southern 339 kg); producer meat prices (FAOSTAT). Crop damage assumed 1% of cropland area in the buffer for the most extensive elephant-targeted crop per country (cassava, maize, millet, banana, sorghum, groundnuts), using country-specific yields and producer prices (FAOSTAT). Costs included fence construction (median USD 9522/km) and annual maintenance (median USD 487/km) from Pekor et al. (2019). Latin hypercube sampling (100 partitions) varied all model parameters by ±10% to propagate uncertainty; resulting EAA distributions by country are presented.

- Estimated populations: ~25,125 (±549) lions and 415,428 (±20,112) elephants across Africa. - Predictors of abundance: Human population density was the strongest negative predictor of both lion and elephant numbers at site level. Nationally, lion populations were higher where conservation expenditures were higher; elephant numbers were higher with greater GDP per capita. - Exposure to human pressures: 82% of all sites (protected or other conservation areas) containing lions and elephants are adjacent to substantial human pressures. - Perimeter risk: About 60% of the extended-range perimeter is adjacent to high densities of human population, crops, or cattle. Nine percent (≈10,000–12,000 km) is at severe risk (overlap of all three pressures), spread across 18 countries that host ~74% of lions and 41% of elephants. An additional 10% of perimeter is at high risk (human plus either crops or cattle), across 26 countries. Countries with severe/high risk contain ~95% of Africa’s lions and ~66% of elephants. - Robustness: The identification of countries with severe risk was stable across buffer distances (10–30 km) and random distance perturbations; severe-risk segments had higher certainty of species presence. - Fragmentation risk: Proposed mitigation fences in severe-risk areas would not substantially increase range fragmentation for most migratory mammals; slight increases were noted for Grévy's zebra (Equus grevyi) and Thomson's gazelle (Eudorcas thomsonii). - Economic returns: Building and maintaining high-standard mitigation fences around severe-risk areas likely yields positive net return on investment (EAA > 0) in 17 of 18 countries, with South Sudan as the exception. The largest absolute returns are expected in Tanzania, Ethiopia, and Kenya. In contrast, fencing in high-risk (non-severe) areas would seldom be cost-effective relative to benefits; alternative mitigation strategies may be preferable there. On a per capita basis, potential benefits are highest for Benin, South Africa, and Zambia. - Demographic context: Most countries with severe/high conflict risk are projected to have substantial human population growth by 2100, potentially expanding severe conflict zones without intervention.

The study directly identifies where human–lion–elephant conflicts are most likely along range perimeters and quantifies the financial and ecological implications of deploying mitigation fences. Findings indicate lions face greater conflict risk than elephants and that severe-risk zones coincide with a large fraction of remaining lion populations. Given strong negative associations between human density and wildlife abundance, and projected population growth in many affected countries, conflict pressures are likely to intensify, threatening conservation outcomes and tourism-dependent economies. Economic analyses show that strategically located, high-standard fences around severe-risk segments can be financially justified in most countries, primarily by reducing cattle losses and crop damage, thereby potentially increasing local tolerance for wildlife. However, fences should be implemented to mitigate conflict rather than impede natural movements; fine-scale movement data and safeguards (e.g., gates, wildlife passages) are needed to avoid blocking migrations, particularly for species like Grévy’s zebra and Thomson’s gazelle. In areas with lower conflict levels, non-fencing strategies (human-dimensions approaches, husbandry improvements, compensation/performance payments) may be more cost-effective, provided they are effective and sustainable. The analysis also notes unquantified benefits such as reduced human injury/death and psychological stress, and potential local costs (e.g., access restrictions) that require participatory design (gates, permits) and community engagement. The South African context suggests opportunities to reduce fencing where conflict risk is low to restore ecosystem processes, and to reassess legacy veterinary fences in Botswana and Namibia. Overall, targeted fencing in severe-risk zones can help reconcile human safety, livelihoods, and large-mammal conservation if carefully planned to minimize ecological disruption and maximize community benefits.

This pan-African spatial assessment maps where conflicts with lions and elephants are most likely and evaluates the cost-effectiveness of mitigation fencing at continental scale. Severe-risk perimeter segments are concentrated in 18 countries that host the majority of remaining lions and a substantial portion of elephants. High-standard mitigation fences in these severe-risk areas are generally economically viable and could reduce livestock losses and crop damage, supporting both human wellbeing and wildlife persistence. The study provides a decision-support framework to prioritize where fencing is justified, emphasizes the need to align fencing with conflict mitigation rather than movement restriction, and highlights contexts where alternative strategies may be preferable. Future work should include: fine-scale collar-based movement analyses to inform fence placement and wildlife-friendly designs; assessments of impacts on additional taxa and ecological processes; incorporation of non-material human wellbeing costs/benefits; and participatory planning with local communities to balance access and conservation outcomes.

- Spatial resolution: Species-range maps represent coarse distribution boundaries rather than fine-grained habitat edges; mitigated by incorporating site-level population sizes. - Cost heterogeneity: Construction and maintenance costs likely vary geographically; cost data are unavailable for some countries lacking existing fences. Median costs from literature were applied continent-wide. - Benefit assumptions: Conservative, uniform assumptions (1% cattle loss; 1% crop damage; most extensive elephant-targeted crop per country) may not capture local variation; crop composition within buffers was simplified due to data limitations. - Benefit distribution: Economic analyses assume benefits accrue to local stakeholders, which may differ from donor investment structures. - Scope: Results provide a continental screening, not local implementation blueprints; on-the-ground validation, local socio-political considerations, and community consultations are essential. - Potential local costs: Social costs (e.g., restricted access to resources) were not monetized; ecological costs for some migratory species may occur without safeguards. - Unmonetized benefits: Reductions in human injury/death and psychological stress were not included, potentially underestimating benefits.