Managing wildlife for ecological, socioeconomic, and evolutionary sustainability.

Predicting the Consequences of Selective Harvesting Selective harvesting of animals is widespread throughout the marine, freshwater, and terrestrial environments and affects a diverse list of species, including fish, mammals, birds, and reptiles (1). Such harvesting can cause changes in the distribution of phenotypic traits within target populations, often with undesirable biological and economic consequences. For example, selective harvesting has been linked to declines in the size of trophy horns in two antelope species in Zimbabwe (2) and of antlers in red deer (Cervus elaphus) in Europe (3, 4), as well as to earlier maturation in some fish species (5). However, the extent to which these changes are the result of ecological or evolutionary mechanisms has been much debated (1). In PNAS, Traill et al. (6) approach this question from a novel angle by developing stochastic two-sex integral projection models (IPMs) capable of differentiating between the ecological and evolutionary effects of selective harvest. Their finding that evolutionary mechanisms contribute relatively little to observed changes in the body mass of bighorn sheep (Ovis canadensis) is an intriguing contribution to the debate over the evolutionary consequences of selective offtake, contradicting earlier studies (7). In addition, Traill et al. (6) suggest that their method could be adopted more widely to allow wildlife managers and conservation practitioners to incorporate the potential evolutionary effects of selective harvesting into their management planning. Here, we explore this suggestion by discussing key challenges that would need to be addressed to translate the approach by Traill et al. (6) from a purely biological model to an effective management model, focusing particularly on issues of data availability and the incorporation of different forms of uncertainty. Long-Term Individual-Based Data The first challenge, if IPMs are to achieve widespread use in the management of harvested species, is their dependence on longterm individual-based data. The model by Traill et al. (6) is parameterized for a species, the bighorn sheep, which has been the subject of extensive study (7). However, the combination of long-term, individual traitbased data and detailed records of harvesting offtake is likely to be rare for (i) the species of most conservation concern and (ii) species of social, cultural, and economic importance (e.g., those targeted by fisheries, recreational, and subsistence hunting). For example, one of the longest published datasets for trophy hunted species of conservation concern suggests that declines in lion (Panthera leo) and leopard (Panthera pardus) populations are linked to trophy hunting (8), yet even here it is not clear whether the individual-level trait data needed to construct an IPM are also available. In the absence of such data, Traill et al. (6) suggest that allometric relationships could be used to parameterize IPMs, but acknowledge that further work would be needed to determine how reliable this approach would be. In principle, technologies such as global positioning system (GPS) collars and satellite imagery might allow long-term data to be collected for other species in the future (9). However, the tradeoffs arising from any large-scale investment in long-term monitoring should always be considered (10). In particular, managers should seek to determine whether the benefits gained from understanding long-term evolutionary effects outweigh those that could be achieved if resources were invested to reduce uncertainties in other components of the harvesting system (e.g., the behavior of resource users, see below). The Importance of Uncertainties in the Management Process In their model, Traill et al. (6) assume a simple proportional harvesting strategy and test their method for harvest pressures ranging from 1% to 85% of males in the bighorn sheep population. Similar assumptions are common in harvesting models, but fail to capture important sources of uncertainty present in real-world systems. The outcomes of harvesting arise from the interactions between management authorities, legal and illegal resource users, the exploited resource and the environment, and the effectiveness of management strategies can be strongly influenced by uncertainty arising from any one of these components (11, 12). An illustrative example concerns the effects Management actions