Fig 1.
Pacific walrus haulout sites surveyed in 2017–2019.
Note: Keniskin Bay (south) is not shown on the figure due to proximity to the Keniskin Bay (main). For the site detailed information and coordinates refer to the metadata. The map generated in R using several open source packages including “ggplot2” [38] using world map data from “rnaturalearth” and “rnaturalearthdata” packages [39].
Table 1.
Survey dates by survey method, year, and haulouts.
Empty cells indicate that a particular type of count was not performed for that location and year.
Fig 2.
Comparing densities calculated using land-based and aerial drone-based counts.
Fig 3.
The distribution of walrus densities (in walruses per m2) across surveyed sites.
The data used for this plot include model polygon counts, complete haulout counts, or a combination of both methods. Each box represents the interquartile range (IQR) of walrus densities at each site, while the horizontal line within each box indicates the median density.
Fig 4.
Predicted herd size plotted against observed count values.
X-axis ‐ area occupied by walruses, y-axis ‐ number of walruses. Predictions for Cape Vankarem are shown separately from the two site types (“mixed beach terrain”) for visualization purposes.
Table 2.
Predicted maximum total abundance of Pacific walrus by haulout site and year.
Predictions with lower and upper CI bound are obtained from non-specific negative binomial GLM counts–complete head count data obtained from orthomosaics; extrapolation–model polygon-based extrapolation results.