The recent past and promising future for data integration methods to estimate species’ distributions

Abstract With the advance of methods for estimating species distribution models has come an interest in how to best combine datasets to improve estimates of species distributions. This has spurred the development of data integration methods that simultaneously harness information from multiple datasets while dealing with the specific strengths and weaknesses of each dataset. We outline the general principles that have guided data integration methods and review recent developments in the field. We then outline key areas that allow for a more general framework for integrating data and provide suggestions for improving sampling design and validation for integrated models. Key to recent advances has been using point‐process thinking to combine estimators developed for different data types. Extending this framework to new data types will further improve our inferences, as well as relaxing assumptions about how parameters are jointly estimated. These along with the better use of information regarding sampling effort and spatial autocorrelation will further improve our inferences. Recent developments form a strong foundation for implementation of data integration models. Wider adoption can improve our inferences about species distributions and the dynamic processes that lead to distributional shifts.