Socioeconomic factors predict population changes of large carnivores better than climate change or habitat loss

Rapid global change is placing wildlife populations under threat. Identifying drivers of population trends is essential to pinpoint sources of declines and opportunities for recovery, yet the influences are numerous and difficult to measure. While considerable ecological knowledge links local-scale environmental change—such as land-use change and climate change—to population trends, the role of socioeconomic factors is less understood despite indications of strong effects (e.g., increases in wildlife abundance where governance is strong and standards of living are high). This study takes a multidimensional approach to assess how land-use change, climate change, governance (including socioeconomics), and species traits correlate with population trends in large terrestrial carnivores across the globe. By integrating drivers operating at multiple scales and potential interactions, the authors evaluate which factors most strongly influence carnivore population change and project how socioeconomic trajectories over the past 50 years may have shaped carnivore abundances. They find rapid socioeconomic development is associated with carnivore declines, with potential for recovery as development growth slows, highlighting strategies to help bend the biodiversity curve.

Prior research has established strong links between local habitat loss and biodiversity decline, and mounting evidence implicates climate change in population reductions across taxa. However, comparatively few studies have focused on socioeconomic drivers; limited existing work suggests wildlife populations fare better with effective governance and higher human development. Previous analyses reported positive associations between human development/governance and wildlife trends in protected areas, but the broader, global impact of socioeconomic change on large carnivores remains underexplored. The paper situates its contribution within this gap, testing socioeconomics alongside land-use and climate drivers, and considering interactions such as species specialization and protection status modulating climate impacts.

Data: The study compiled 1,123 population trends for large carnivores (order Carnivora; families Canidae, Felidae, Hyaenidae, Ursidae) from CaPTrends and the Living Planet Database. Two response types were used: (1) quantitative annual rates of change (%) derived from time-series or summary change metrics (N=985; 50 species), and (2) qualitative trend categories (Increase/Stable/Decrease; N=138; 21 species). Trends cover 75 countries and span roughly 1970–2015. Deriving quantitative trends: Time-series of abundance/density were modeled via log-linear regressions with a continuous Ornstein–Uhlenbeck autoregressive process to account for temporal autocorrelation. Slope coefficients (instantaneous rate r) were converted to annual percentage rates of change. Integrating qualitative data: Qualitative records were incorporated using a censored-response approach, treating values as partially known within overlapping ranges (Decrease: −50 to 0%; Stable: −5 to 5%; Increase: 0 to 50%), acknowledging categorization uncertainty. Covariates: Sixteen z-transformed covariates across four groups were extracted and matched in space and time to each population’s monitoring period (with tested lags of 0, 5, 10 years; 10-year lag selected for best fit): - Traits: body mass; climatic niche breadth; ecological niche breadth; maximum longevity; reproductive output; population area. - Land-use: primary habitat loss; change in natural land; change in human density. - Governance: war presence; human development (level at start); change in human development (annual change); protected area coverage; governance score. - Climate: change in drought frequency; change in extreme heat frequency. Spatial matching used each population’s reported extent to define a buffer (“population area”), within which covariates were sampled proportionally to area size. Modeling framework: A hierarchical Bayesian linear mixed-effects model (JAGS via R) regressed annual rate of change on 23 effects (16 main covariates + 7 interactions), with random intercepts for phylogeny (species nested within genus) and geography (country within UN sub-region). Core parameters (e.g., change in human density, primary land loss, population area, body mass, change in extreme heat, governance, protected area coverage) were always included; Kuo–Mallick variable selection identified additional main and interaction effects to avoid overparameterization. Lower-quality records (short timeframes/few observations) were downweighted via a modeled error inflation term. Missing data: Trait gaps were imputed (e.g., phylogenetic and MICE approaches), and imputation uncertainty was propagated by treating imputed values as distributions (normal with imputed mean/SD). Model execution and checks: Three MCMC chains (150,000 iterations; 50,000 burn-in; thinning 10). Convergence diagnostics, residual checks for normality/heteroscedasticity, tests for residual spatial (Moran’s I) and phylogenetic (Pagel’s lambda) autocorrelation, and posterior predictive checks were performed. Model fit achieved a conditional R²≈0.4. Counterfactual analyses: Predicted trends under scenarios with (1) no primary habitat loss, (2) no climate change (extreme heat and drought set to zero), and (3) no growth in human development (change set to zero). Differences from observed predictions quantified each driver’s contribution. Socioeconomic pathway simulations: Using the estimated effect of change in human development (median standardized effect size −0.44), the study simulated three within-country development pathways (Slow 1.25%, Moderate 1.5%, Fast 1.75% mean annual change, each decelerating by −0.02% per year) from 1960 HDI=0.2, projecting hypothetical carnivore abundances (baseline 100 in 1960) to 2020 while holding other covariates at mean values.

- Primary habitat loss is associated with carnivore population declines across species, regardless of what replaces the habitat or species’ ecological specialization (no support for predicted interactions). - Climate impacts are complex: increased extreme heat shows no uniform effect overall, but interacts with protected area coverage—declines outside protected areas and rapid growth inside, suggesting a buffering effect of protected areas to climatic extremes; protected areas show no overall marginal effect when other covariates are average. - Contrary to expectations, increased drought frequency correlated positively with carnivore trends, possibly via short-term predator–prey dynamics that increase predator foraging success; mechanisms remain speculative. - Governance score and human development level (at the start of monitoring) showed no consistent effect on trends; however, rapid growth in human development (annual change) was strongly associated with carnivore population declines. - Counterfactual analyses indicate changes in human development explain population changes far more consistently than habitat loss or climate change across continents; habitat/climate impacts were detectable but affected a minority of populations with extreme exposure. - A threshold effect: when annual change in human development falls below approximately 1.2%, carnivore populations can stabilize and increase; faster growth rates are linked to declines. The estimated standardized effect size for change in human development was about −0.44. - Simulations show faster and prolonged development growth yields larger declines, with eventual potential recovery as development growth decelerates, resembling a Kuznets-like nonlinearity. - Dataset: 1,123 population trends across 50 species; 985 quantitative and 138 qualitative records; coverage across 75 countries (approximately 1970–2015); model conditional R²≈0.4; only ~6% of populations experienced high primary habitat loss (≤−5% per year).

The study demonstrates that socioeconomic dynamics—specifically the pace of improvement in human development—are more predictive of large carnivore population changes than habitat loss or climate change, at least within the available data. This finding reframes how drivers of biodiversity change are interpreted: socioeconomics may be the arena in which environmental stressors manifest, modulating tolerance toward wildlife, poaching pressure, and human–wildlife conflict. Climate effects are nuanced: protected areas may buffer extreme heat impacts, potentially via microclimatic refugia and short-term immigration, though their finite capacity limits long-term growth. The positive association between drought frequency and carnivore trends suggests complex, possibly transient ecological responses that warrant further study. Counterfactuals underscore that while habitat and climate change can drive declines, their detectable impacts in this dataset are comparatively less pervasive than the widespread and consistent influence of human development change. The nonlinearity observed—a threshold near 1.2% annual HDI growth where declines transition toward recovery—offers a potential framework akin to a biodiversity Kuznets curve, though mechanisms differ and generality is uncertain. These insights have policy relevance: rapid development may conflict with biodiversity conservation (e.g., SDG trade-offs), suggesting a need to mitigate mechanisms linking development surges to carnivore declines while continuing to improve human well-being.

This work integrates ecological, climatic, land-use, governance, and trait covariates in a global hierarchical model to show that the rate of socioeconomic development is a dominant predictor of large carnivore population change, exceeding detectable impacts of habitat loss or climate change in the compiled data. It highlights complex climate–protection interactions and suggests a thresholded, nonlinear relationship between development pace and carnivore trajectories with potential for recovery as development growth slows. Future research should: (1) test whether similar socioeconomic effects occur across other taxa and ecosystems; (2) identify and quantify the mechanisms (e.g., poaching, conflict, tolerance shifts) linking rapid development to declines; (3) improve data coverage for populations under extreme habitat/climate pressures; (4) refine understanding of temporal lags and nonlinear dynamics; and (5) explore policy strategies to reconcile human development with biodiversity conservation, aiding progress toward multiple SDGs.

- Moderate model fit (conditional R²≈0.4) indicates substantial unexplained variation. - Data biases persist despite integrating qualitative records: more trends originate from high human development countries; underrepresentation likely for populations experiencing extreme habitat loss or climate change, potentially underestimating their effects and prevalence. - Some key drivers are cryptic and poorly measured at global scales (e.g., poaching, persecution, culling, changing tolerance, benefits of flagship status). - Temporal lags between drivers and population responses vary and are not fully known; a 10-year lag improved fit overall but may not suit all drivers; longer lags and slow recoveries could not be fully assessed due to limited long-term data, especially outside the global north. - Qualitative categorization thresholds are somewhat arbitrary (though carefully chosen) and may introduce additional uncertainty; weighting schemes mitigate but do not eliminate issues from short time-series or sparse data. - Space-for-time assumptions in socioeconomic comparisons may be confounded by historical extirpations in highly developed countries, biasing inferences about development transitions elsewhere.