Environmental data

Gray whale distribution and abundance were modeled relative to static and dynamic environmental variables selected to describe the nearshore region of the NCC. In particular, variables assessed included features that may regulate nearshore productivity and retention of upwelled nutrients and thereby potentially influence gray whale foraging opportunities (Table ​(Table1,1, Supplementary Materials, Fig. ^S3).

Static and dynamic environmental covariates included in the gray whale density surface models, along with the source of the data and relevant information on how the metrics were computed from the data.

All static environmental layers were computed at a 1 km spatial resolution. Bathymetric depth was obtained from the general bathymetric chart of the oceans (GEBCO, 15 arc-second resolution), and high-resolution coastline information was obtained from the Open Street Map dataset. These two datasets were used to generate continuous layers representing the depth, distance from shore, and shelf width, which was measured as the distance from the coast to the 200 m isobath. Additionally, prominent capes were identified by applying a 200 km continuous smooth to the high-resolution coastline, and locations where the distance between the smoothed and high-resolution coastline was > 5 km were labeled capes. Subsequently, a continuous layer representing distance to the nearest cape was generated for the whole region, with negative values representing locations north of capes and positive values representing locations south of capes. The location and size of river estuaries were obtained from the Pacific Marine and Estuarine Fish Habitat Partnership Estuary Points dataset, which was subset to only include river estuaries > 300 hectares (includes most rivers but excludes small creeks); a continuous layer representing distance to the nearest estuary was generated across the study area. Benthic substrate was classified as either hard or soft based on multibeam sonar mapping conducted, processed, and provided by the Active Tectonics and Seafloor Mapping Lab at Oregon State University for Oregon waters, and the Seafloor Mapping Lab at California State University Monterey Bay for California waters.

Three dynamic environmental variables were examined: sea surface temperature (SST), coastal upwelling, and relaxation. Daily SST data were downloaded from the optimal interpolation SST (OISST, 0.25° resolution) product, which incorporates observations from different platforms (satellites, ships, buoys, and Argo floats) into a regular grid, and spans the full temporal and spatial range of our study. Coastal upwelling and relaxation were obtained from the daily Coastal Upwelling Transport Index (CUTI, 1° latitude bins), which estimates vertical flux off the United States West Coast using regional sea surface height, surface wind stress, and mixed layer depth via regional ocean reanalysis⁵⁸. After a 10-day smoothing filter was applied to the daily CUTI values to reduce the influence of anomalous spikes following previously established methods⁵⁹, a daily cumulative upwelling index was generated for each 1° latitude bin. The cumulative CUTI was used to identify the upwelling season as the period between spring transition index and the end of the upwelling season, following previously established upwelling phenological definitions⁴. Cumulative CUTI was then re-calculated beginning at the spring transition index for subsequent analysis. The mean daily CUTI within the upwelling season was then calculated across the entire study period for each latitude bin, and “relaxation events” were defined as days when the daily CUTI value fell below the mean CUTI during the upwelling season for that latitude bin. Finally, cumulative relaxation was calculated for each day as the cumulative sum of the number of days classified as relaxation events since the spring transition.

Observation conditions and the total number of gray whales were aggregated by survey segment. Environmental data were then extracted at the centroid location of each survey segment for all static layers and SST; cumulative CUTI and cumulative relaxation were assigned to segments by date and latitude bin. While inshore segments were 5 km in length, the offshore segments varied in length due to the sampling scheme; segment length was accounted for in model fitting.

Our study area spanned parts of three established bioregions of the California Current Large Marine Ecosystem⁴⁹ (Supplementary Materials, Fig. ^S1). We fit spatial models separately for each region to account for documented regional oceanographic differences that may influence gray whale habitat use and foraging patterns in distinct ways.