Aim: Anthropogenic food subsidies are increasingly present in ecosystems, but their impacts remain poorly understood. Big game hunting is a growing activity that annually subsidizes ecosystems with tonnes of carrion world‐wide. By feeding on carrion, scavengers support key ecosystem functions and services, becoming key vectors to transfer the impacts of human‐mediated food subsidies across ecosystems. We characterize and compare the structure of vertebrate communities feeding on these subsidies, namely big game hunting remains, at a global scale.Location: Global.Methods: We collected data from a countrywide field study in Spain and broadened it up to nine regions in four continents by reviewing scientific literature. We analysed the structure of the scavenger communities considering species composition, richness and scavenging frequency.Results: Seventy‐nine vertebrate species, 19% globally threatened, scavenged food subsidies from big game hunting world‐wide. Scavenger richness (2.0–11.0% of vertebrates/region) positively correlated with total vertebrate richness. Although scavenger communities at hunting remains varied among regions, we describe a general structural pattern. Birds and mammals dominate consumption, with birds scavenging twice more frequently than mammals – but more mammal species scavenge compared to birds. Generalists dominate scavenging globally, especially where the presence of obligate scavengers (vultures) and apex predators (e.g. wolves, hyenas, eagles) is low.Main conclusions: Anthropogenic food from hunting subsidizes many vertebrate species from different trophic levels and conservation status and thus is expected to affect from populations to ecosystems. Obligate scavengers and apex predators seem to play a key role structuring the scavenger community through top‐down mechanisms. The general structure of scavenger communities we describe here provides a benchmark for comparisons of subsidized and non‐subsidized communities. More data on the spatio‐temporal availability of anthropogenic food subsidies and their consumption by scavengers world‐wide are needed to efficiently preserve biodiversity, and the associated ecological functions and services, in increasingly subsidized ecosystems.

Summary: 1. Illegal human behaviour such as those affecting natural resource use or resulting from human–wildlife conflicts threaten the sustainable management of ecosystems and the conservation of biodiversity worldwide. However, the frequently scarce and incomplete data owing to the sensitive nature of illegal activities pose a challenge to developing tools to properly understand and prevent those activities.2. We used species distribution models to identify factors related to a prominent illegal activity, wildlife poisoning, and to produce detailed, spatially explicit maps of the risk of occurrence in NW Spain. We alleviated the constraints of imperfect information and occurrence of absences by using presence‐only methods, that is, maximum entropy modelling (MaxEnt). To our knowledge, this is the first time that this method has been used in the context of illegal activities affecting wildlife.3. A total of 112 poisoning events involving 228 individuals of 25 different species were reported in the study area from 2000 to 2010. Most of the reported deaths (90·8%) were birds of prey (52·6%) and mammalian carnivores (38·2%), of which 95·2% were scavengers. Illegal poisoning affected eleven species classified as endangered at national and/or global level.4. Our models highlighted the perceived risk of livestock predation by wolves Canis lupus, although not by bears Ursus arctos, as a major motivation for poisoning. The existence of protected areas was positively correlated to this illegal practice, while socioeconomic factors had less influence on predicting its occurrence. Over 56% of the study area was predicted to be under risk of illegal poisoning.5. Synthesis and applications. We demonstrate a new use for presence‐only models, illustrated using MaxEnt, to assist conservation managers dealing with illegal activities. This approach allows the main causes of an illegal practice to be identified and generates spatially explicit risk maps. Managers can take advantage of this modelling approach to allocate the scarce resources available in conservation to key sectors and locations. In our study system, actions against illegal poisoning should aim to resolve the potential conflict existing between cattle‐farming and wolves, especially in protected areas.

Hierarchical partitioning (HP) is an analytical method of multiple regression that identifies the most likely causal factors while alleviating multicollinearity problems. Its use is increasing in ecology and conservation by its usefulness for complementing multiple regression analysis. A public-domain software “hier.part package” has been developed for running HP in R software. Its authors highlight a “minor rounding error” for hierarchies constructed from >9 variables, however potential bias by using this module has not yet been examined. Knowing this bias is pivotal because, for example, the ranking obtained in HP is being used as a criterion for establishing priorities of conservation. Using numerical simulations and two real examples, we assessed the robustness of this HP module in relation to the order the variables have in the analysis. Results indicated a considerable effect of the variable order on the amount of independent variance explained by predictors for models with >9 explanatory variables. For these models the nominal ranking of importance of the predictors changed with variable order, i.e. predictors declared important by its contribution in explaining the response variable frequently changed to be either most or less important with other variable orders. The probability of changing position of a variable was best explained by the difference in independent explanatory power between that variable and the previous one in the nominal ranking of importance. The lesser is this difference, the more likely is the change of position. HP should be applied with caution when more than 9 explanatory variables are used to know ranking of covariate importance. The explained variance is not a useful parameter to use in models with more than 9 independent variables. The inconsistency in the results obtained by HP should be considered in future studies as well as in those already published. Some recommendations to improve the analysis with this HP module are given.