Disentangling data discrepancies with integrated population models

Abstract

A common challenge for studying wildlife populations occurs when different survey methods provide inconsistent or incomplete inference on the trend, dynamics, or viability of a population. A potential solution to the challenge of conflicting or piecemeal data relies on the integration of multiple data types into a unified modeling framework, such as integrated population models (IPMs). IPMs are a powerful approach for species that inhabit spatially and seasonally complex environments. We provide guidance on exploiting the capabilities of IPMs to address inferential discrepancies that stem from spatiotemporal data mismatches. We illustrate this issue with analysis of a migratory species, the American Woodcock (Scolopax minor), in which individual monitoring programs suggest differing population trends. To address this discrepancy, we synthesized several long-term data sets (1963–2015) within an IPM to estimate continental-scale population trends, and link dynamic drivers across the full annual cycle and complete extent of the woodcock's geographic range in eastern North America. Our analysis reveals the limiting portions of the life cycle by identifying time periods and regions where vital rates are lowest and most variable, as well as which demographic parameters constitute the main drivers of population change. We conclude by providing recommendations for resolving conflicting population estimates within an integrated modeling approach, and discuss how strategies (e.g., data thinning, expert opinion elicitation) from other disciplines could be incorporated into ecological analyses when attempting to combine multiple, incongruent data types.