A matter of scale: Identifying the best spatial and temporal scale of environmental variables to model the distribution of a small cetacean

The importance of scale when investigating ecological patterns and processes is recognised across many species. In marine ecosystems, the processes that drive species distribution have a hierarchical structure over multiple nested spatial and temporal scales. Hence, multi-scale approaches should be considered when developing accurate distribution models to identify key habitats, particularly for populations of conservation concern. Here, we propose a modelling procedure to identify the best spatial and temporal scale for each modelled and remotely sensed oceanographic variable to model harbour porpoise (Phocoena phocoena) distribution. Harbour porpoise sightings were recorded during dedicated line-transect aerial surveys conducted in the summer of 2016, 2021 and 2022 in the Northeast Atlantic. Binary generalised additive models were used to assess the relationships between porpoise presence and oceanographic variables at different spatial (5, 20 and 40 km) and temporal (daily, monthly a..., , , # A matter of scale: Identifying the best spatial and temporal scale of environmental variables to model the distribution of a small cetacean

[https://doi.org/10.5061/dryad.s7h44j1dv](https://doi.org/10.5061/dryad.s7h44j1dv)

Harbour porpoise sightings were recorded during dedicated line-transect aerial surveys conducted in the summer of 2016, 2021 and 2022. Binary generalised additive models were used to assess the relationships between porpoise presence and oceanographic variables at different spatial (5, 20 and 40 km) and temporal (daily, monthly and across survey period) scales. Selected variables included sea surface temperature, thermal fronts, chlorophyll-a, sea surface height, mixed layer depth and salinity.

## Description of the data and file structure

Each dataset contains presence and absence data of harbour porpoise in a grid cell and associated values of one oceanographic variable at various spatial and temporal scale. The covariates were extracted from the Copernicus Ma...