Analytical approaches to investigating seabird-environment interactions: a review

A goal of seabird ecology is to relate the physiology, population dynamics, distribution and behaviour of seabirds with their biotic and abiotic environments. One of the most challenging aspects is to understand how seabirds interact with their environment when direct observations are not always possible or practical. In the present paper, we reviewed 218 published studies that exam- ined associations between seabird distribution, behaviour and their environment, in order to assess current trends, weaknesses and the future directions of research. Based on the number of publica- tions, it is evident that the field is growing rapidly and that methods for evaluating seabird distribu- tion are becoming increasingly more sophisticated and are changing from Eulerian (grid-like) to Lagrangian (particle-like) data types. This has been accompanied by a reduction in the spatial and temporal scale of observation, where, in most cases, no behavioural information is inferred from Lagrangian data; instead they are often used as if they were Eulerian data. In parallel, environmen- tal remote sensing is becoming more common; however, we did not record significant changes in the statistical approaches used to describe seabird distributions and used to link them with oceano- graphic variables. In particular, despite the spatially explicit nature of the data, spatial statistics have rarely been used. The vast majority of studies used environmental variables that described water masses (descriptive approach), whereas a few studies determined oceanographic features that enhance prey availability to seabirds (process-based approach). Future studies could enhance their ecological interpretation of seabird-environment interactions by making greater use of ad hoc statis- tical approaches that facilitate appropriate pattern detection (e.g. area-restricted searching pattern for birds, mesoscale patterns for environment). Furthermore, appropriate hypothesis testing and modelling that accounts for the spatially explicit, multiscale and multivariate nature of the interaction between seabirds and their habitats is recommended. Although quantitative methods currently exist (but are rarely used), further application could greatly improve our understanding of the processes linking seabird distribution to their environment.