Working with large-scale population trend data in ecology and conservation: methods and applications

Wildlife conservation, at its core, hinges on sustaining and maintaining animal popu- lations. Whether reducing hunting, protecting habitat or creating breeding pro- grammes, so many conservation decisions have population support at their core, and yet methods to accurately work with population data at scale are lacking. Armed with one the world’s largest population trend datasets, 749 species of waterbirds, at 45,475 sites across the world, I develop methods to improve inferences from popula- tion trends, explore methods to combat data dredging and combine my findings to conduct the largest ever robust assessment of protected area effectiveness In Chapter 2 I take the highest quality trends from the waterbird dataset, and artifi- cially degrade them to understand how likely short-term trends are to represent longer term trends. The methods are generalisable to any taxa and allow for rigorous quantification of the reliability of trends derived from different length time series. Trends can be useful for more than just determining the health of a species, they can also be used to assess the impact of interventions on a population. In Chapter 3, I present a pre-analysis plan for a project assessing the impact of protected areas on populations; pre-analysis plans are relatively new in ecology, but present one method to combat against data dredging of large datasets. As I began to carry out this analy- sis, I discovered that current impact evaluation methods in ecology can result in in- accurate conclusions being drawn when applied to population trends; in Chapter 1 I set out a new analysis framework. Finally, in Chapter 4, I combined the methodological developments of Chapters 1 and 2 to conduct, to my knowledge, the largest ever assessment of protected area effec- tiveness. Using the waterbird dataset, I compare the trends of populations in pro- tected areas to the trends of the same populations in the years before protection, as well as to unprotected populations (a Before-After-Control-Impact study design). This approach is far more rigorous than a simple inside-outside comparison ap- proach and, using this design, I show that just 32% of populations are actively bene- fitting from protection, but that large, well-managed sites produce better outcomes.