When can model‐based estimates replace surveys of wildlife populations that span many discrete management units?

Abstract

Abstract Monitoring widely distributed species on a budget presents challenges for the spatio‐temporal allocation of survey effort. When there are multiple discrete units to monitor, survey alternatives such as model‐based estimates can be useful to fill information gaps but may not reliably reflect biological complexity and change. The spatio‐temporal allocation of survey effort that minimizes uncertainty for the greatest number of units within a budget can help to ensure monitoring is optimized. We used aerial survey‐based population estimates of moose (Alces alces) across 30 Wildlife Management Units (WMUs) in Ontario, Canada to parameterize simulated populations and test the performance of different monitoring scenarios in capturing WMU‐specific annual variation and trends. Firstly, we tested scenarios that prioritized conducting a survey for a unit based on one of three management criteria: population state, population uncertainty or number of years between surveys. Also incorporated in the decision framework were WMU‐specific costs and annual budget constraints. Secondly, we tested how using model‐based estimates to fill information gaps improved population and trend estimates. Lastly, we assessed how the utility (based on minimizing population uncertainty) of using a model‐based estimate rather than conducting a survey was impacted by population density, severity of environmental stressors and years since the last survey. Interval‐based monitoring that minimized the number of years between surveys captured accurate trends for the highest number of WMUs, but annual variation was poorly captured regardless of management criteria prioritized. Using model‐based estimates to fill information gaps improved trend estimation. Further, the utility of conducting a survey increased with time since the last survey and was greater for populations with low densities when the severity of environmental stressors was high, while being greater for populations with high densities when environmental severity was low. Overall, the utility of aerial survey monitoring was strongly associated with WMU‐specific monitoring precision and the predictive power of model‐based estimates. If long‐term trends are evident, then there is greater value in using alternatives such as model‐based predictions to replace surveys, but model‐based estimates may be a poor substitute when there is strong annual variation and when using a simple model.

Keywords

• aerial survey
• model‐based estimate
• moose
• population trends
• survey effort
• uncertainty