Scale of inference: the sensitivity of habitat models for wide-ranging marine predators to the spatial and temporal resolution of environmental data
Kylie L. Scales (1), Elliott L. Hazen (2), Michael G. Jacox (3), Christopher A. Edwards (4), Andre M. Boustany (5), Matthew J. Oliver (6) & Steven J. Bograd (7)
1 Cooperative Institute for Marine Ecosystems and Climate, Institute of Marine Sciences, University of California, Santa Cruz, USA / NOAA SWFSC Environmental Research Division, Monterey, CA 93940, USA. kylie.scales@noaa.gov
2 NOAA SWFSC Environmental Research Division, Monterey, CA 93940, USA. elliott.hazen@noaa.gov
3 Cooperative Institute for Marine Ecosystems and Climate, Institute of Marine Sciences, University of California, Santa Cruz, USA / NOAA SWFSC Environmental Research Division, Monterey, CA 93940, USA. michael.jacox@noaa.gov
4 Institute of Marine Sciences, University of California, Santa Cruz, USA. cedwards@ucsc.edu
5 Nicholas School of the Environment, Duke University, Durham NC 27708, USA. andre.boustany@duke.edu
6 School of Marine Science and Policy, University of Delaware, Lewes, DE 19958, USA. moliver@udel.edu
7 NOAA SWFSC Environmental Research Division, Monterey, CA 93940, USA. steven.bograd@noaa.gov
Understanding and predicting the responses of wide-ranging marine predators such as seabirds, cetaceans, sharks, turtles, pinnipeds and large migratory fish to oceanographic conditions requires habitat-based models that can sufficiently capture their environmental preferences. Marine ecosystems are inherently dynamic, and animal-environment interactions are known to occur over multiple, nested spatial and temporal scales. The spatial resolution and temporal averaging of environmental data layers are therefore key considerations in modelling the environmental determinants of habitat selection. The utility of surface data contemporaneous to animal presence or movement (e.g. daily, weekly), versus synoptic products (monthly, seasonal, climatological) in habitat-based models is currently debated, as are the trade-offs between near real-time, high resolution and composite (i.e. synoptic, cloud-free) environmental data fields. Using movement simulations with built-in environmental preferences (correlated random walks, hidden Markov models) combined with modelled and remotely-sensed (ROMS, MODIS-Aqua) sea surface temperature fields, we explore the effects of spatial and temporal resolution (3km – 1 degree, daily – climatological) on model accuracy. Results indicate that models using seasonal or climatological data fields can overfit environmental preferences in presence-availability designs that use animal movement datasets, particularly in highly dynamic ocean domains. These effects were pronounced where models were constructed using seasonal or climatological fields of coarse (>0.25 degree) spatial resolution. We also observed a divergence between models selected using common performance metrics (AICc, AUC) and those that accurately reproduced known environmental preferences. These findings have important implications for identifying key habitats for marine predator populations of conservation concern, and in forecasting future climate- mediated ecosystem changes.