Range-wide population viability analyses reveal high sensitivity to wildflower harvesting in extreme environments.
The ecological effects of harvesting from wild populations are often uncertain, especially since the sensitivity of populations to harvesting can vary across species' geographical ranges. In the Cape Floristic Region (CFR, South Africa) biodiversity hotspot, wildflower harvesting is widespread and economically important, providing an income to many rural communities. However, with very few species studied to date, and without considering range-wide sensitivity to harvesting, there is limited information available to ensure the sustainability of wildflower harvesting. We studied geographical variation in sensitivity to wildflower harvesting for 26 Proteaceae shrubs with fire-driven life cycles using population viability analyses. We developed stochastic, density-dependent population models that were parameterised from individual demographic rates (adult fecundity, seedling recruitment and adult fire survival) and local environmental conditions across the geographical ranges of the study species. We then simulated the effects of harvesting on populations in different environments across species ranges. Our model simulations predicted extinction risk per population, and we derived extinction probabilities over 100 years in response to different harvesting regimes. We used these population-level extinction probabilities to quantify inter- and intraspecific variation in sensitivity to wildflower harvesting, and to explore how geographical variation in sensitivity depends on environmental conditions (climate, soil fertility and fire disturbance). We detected considerable inter- and intraspecific variation in sensitivity to wildflower harvesting for the 26 study species. This held for both 'nonsprouters' and 'resprouters' (species with low and high fire persistence ability, respectively). Intraspecific variation in sensitivity to harvesting showed varying geographical patterns and associated with environmental variation. Notably, sensitivity was high towards range edges and at the climatic extremes of species ranges respectively. Synthesis and applications. We show the importance of combining spatial demographic data, density-dependent population dynamics and environmental variation when assessing sensitivity to harvesting across species' geographical ranges. Our findings caution against the application of general harvesting guidelines irrespective of species, geographical location or local environmental conditions. Our range-wide population viability analyses provide insights for developing species-specific, spatially nuanced guidelines for conservation management. Our approach also identifies species and areas to prioritise for monitoring to prevent the overexploitation of harvested species.