Where people hunt and what mitigates unsustainable hunting.

Left panel: probability of high threat to wildlife by illegal hunting (killing/hunting/poisoning) inside PAs related to (a) community characteristics and the separate components of the index community characteristics, (b) attitudes of local communities towards conservation, (c) mutual trust levels between park management and organisations, and (d) conservation-friendly local culture (the data underlying this figure can be found in S8 Data). Right panel: threat levels outside park boundaries across (e) community attributes, (f) attitudes, (g) mutual trust levels, and (h) local culture (S9 Data). Threat levels decreased equally outside and inside parks, while the effects were more pronounced outside parks. The models, including mutual trust, best explained the results. (i) The scale of inclusion of local communities in decision-making showed a positive relationship with the index community characteristics (S10 Data). The “scale of inclusion” was alongside the “provision of benefits to communities”, “implementation of livelihood projects”, and “awareness creation”, the only predictor that showed an effect. The dashed line depicts the expected mean of the predicted posterior distribution. The coloured areas are the 33% and 66% credibility intervals. Transparent points are binned probabilities. The size of the bubbles corresponds to the respective number of PAs. Filled points are original data.