Considering group size when estimating wildlife populations

Dall's sheep are social animals and form groups. Group size is influenced by several factors, which, when considered, can improve the accuracy of wildlife population estimates.

NPS/Zack Delisle

Dall’s sheep live in mountainous areas across northwestern North America. They are highly social animals that live in groups often comprised of ewes (adult females) and their lambs or several rams. National Park Service scientists looked at the conditions that influence the size of Dall’s sheep groups in several Alaska parks. The kinds of characteristics they looked at included things like topography, escape routes, and food availability. What they found was:

• Sheep formed larger groups in places where they were more vulnerable to predators.
• They also formed larger groups in places that had large amounts of vegetative food.
• Sheep formed smaller groups in steep terrain where vulnerability to predators was low and in areas with limited amounts of food.

Based on these findings, the scientists next examined how they could use this new knowledge about sheep groups to better estimate population sizes of sheep. In fact, they were able to increase the reliability and predictive accuracy of their population estimates. The improved estimates of sheep population size are useful to park managers for better understanding the health of sheep populations.

Linking behavioral ecology and population monitoring: The importance of group size for spatial population models

Abstract

Modeling the spatial distribution of wildlife abundance is paramount for management. In group-forming species, group size and occurrence may be governed by different ecological processes. Hierarchical models can conveniently address group size and occurrence as separate processes when estimating abundance. Therefore, identifying factors influencing group size may improve models that the management of group-forming species relies upon. We test this premise on Dall's sheep (Ovis dalli; henceforth sheep), a group-forming ungulate for which spatial distance sampling models are used to inform management, but environmental features that affect the group size of sheep have yet to be explored within a spatial distance sampling framework. We first used multi-level Bayesian models to test how spatially explicit indices of predation risk and food availability explained variation in sheep group size. We then demonstrated how including group size covariates within a spatial distance sampling model can improve model inference. Variation in sheep group size was associated with interactions between and among indices of predation risk and food availability. Larger predicted group sizes occurred in areas with higher indexed predation risk and in areas with lower risk but high indexed food availability (decreased competition). Groups tended to be smaller in steep terrain near topographical apexes and in areas with limited forage. Incorporation of these predictors of group size in our application demonstrated how our understanding of spatial patterns in abundance improved when we simultaneously modeled variation in both group occurrence and size. Our findings indicate that sheep are making complex trade-offs between predation risk and food availability when deciding to aggregate with conspecifics. Explicitly modeling these ecological relationships within our spatial distance sampling model improved predictive performance, increased abundance estimates, and mechanistically linked ecological processes with population monitoring and management. Many wildlife species that form groups are of interest to active wildlife management and our grander understanding of wildlife ecology. Therefore, the concepts we developed here are broadly applicable across a wide range of group-forming taxa.

Delisle, Z. J., J. H. Schmidt, E. J. Wald, K. A. Cutting, M. S. Sorum, B. Mangipane, K. Joly, R. B. Fronstin, M. D. Cameron, and B. Borg. 2025. Linking behavioral ecology and population monitoring: The importance of group size for spatial population models. Ecosphere 16(12): e70461.