Rapid restructuring of the odontocete community in an ocean warming hotspot

Abstract

Cetaceans are important consumers in marine ecosystems, but few studies have quantified their climate responses. The rapid, directional warming occurring in the Northeast United States (NEUS) provides a unique opportunity to assess climate impacts on cetaceans. We used stranding data to examine changes to the distribution and relative abundance of odontocetes from 1996 to 2020 in both the NEUS and the Southeast United States (SEUS), which is not warming. We conducted simulations to determine the number of stranding events needed to detect a distributional shift for each species given the speed of the shift and the spatial variability in strandings. We compared observed shifts to climate velocity. Smaller sample sizes were needed to detect more rapid poleward shifts, particularly for species with low spatial variability. Poleward shifts were observed in all species with sufficient sample sizes, and shifts were faster than predicted by climate velocity. For species whose trailing edge of distribution occurred in the NEUS, the center of distribution approached the northern limit of the NEUS and relative abundance declined through time, suggesting shifts north out of US waters. The relative abundance of warm water species in the stranding record increased significantly in the NEUS while that of cool water species declined significantly as their distributions shifted north out of the NEUS. Changes in the odontocete community were less apparent in the SEUS, highlighting the importance of regional warming. Observed poleward shifts and changes in species composition suggest a reorganization of the odontocete community in the NEUS in response to rapid warming. We suggest that strandings provide a key dataset for understanding climate impacts on cetaceans given limitations of survey effort and modeling approaches for predicting distributions under rapidly changing conditions. Our findings portend marked changes to the distribution of highly mobile consumer species across international boundaries under continued warming.

Keywords: cetacean; climate response; distributional shift; rapid warming; regional change; trailing edge.

FIGURE 1

(a) Temporal gradient (°C year⁻¹) in sea surface temperature (SST) in the Northwest Atlantic from 1996 to 2020. (b) Annual trends in mean SST anomaly in the northeast and Southeast United States, respectively (NEUS and SEUS) from 1996 to 2020. (c) Climate velocity in NEUS and SEUS from 1996 to 2020.

FIGURE 2

Changes in the relative abundance of odontocete strandings by climatic grouping in the Northeast United States (NEUS) and the Southeast United States (SEUS) through time (a, c) and relative to annual sea surface temperature anomaly (b, d) from 1996 to 2020. Arctic and cosmopolitan odontocete species made up less than 2% of strandings in each year in the NEUS, as did cosmopolitan species in the SEUS, and arctic species were not observed in the SEUS during the study period.

FIGURE 3

Shifts in poleward distance, measured as the along‐shelf distance from the southern tip of Florida (km), in strandings from 1996 to 2020 for six species of odontocete with sufficient data to be assessed using linear models following results of simulations. Changes in the center of distribution are shown in blue for cool water species and red for warm water species, while changes in the trailing edge of the distribution are shown in orange for the three trailing edge species. The black dotted line represents the location of Cape Hatteras, separating the Southeast and Northeast United States.

FIGURE 4

Changes in the spatial distribution of the stranding events of three cool water odontocete species, harbor porpoise, Atlantic white‐sided dolphin and long‐finned pilot whale, in 5‐year periods along the eastern seaboard of the United States. Histograms reflect the proportion of stranding events for that species in each time period binned based on latitude.

FIGURE 5

Proportion of 1000 model runs with a significant trend (p‐value <.05) and a slope estimate with the correct sign compared to the corresponding true slope for simulated trends in changes in the poleward distance through time, shown here for common dolphins. True slope values are indicated in gray at the top of each panel. Simulated data shown are based on observed variance in common dolphin stranding location and annual stranding event sample sizes (n = 2, 5, 6, 7, 8, 10, 24, 25, 50, 66, 86, 100, 250, 331), with slope values ranging from −2 to 40 km year⁻¹. The green line and number indicates the minimum sample size necessary to identify a significant slope estimate of the correct sign 95% of the time, whereas the orange dotted line indicates a value of 0.95.