2.1 Visual foraging observations
We observed sea otter foraging behavior from May to August 2018 on the western side and neighboring islands of POW. Sampling was stratified by time since recolonization, based on US Fish and Wildlife Service (USFWS) aerial surveys. Three periods were denoted from the surveys: zone 1 (> 30 years present), zone 2 (< 30 years and > 15 years present), and zone 3 (< 15 years and > 7 years present) (Fig. 1). In each zone, a minimum of 300 foraging dives were recorded. Because zone 2 makes up a majority of POW, most foraging dives occurred in this zone.
Foraging observations were made from shore to assess sea otter diet composition. Questar telescopes (50X) were used to follow individual sea otters for one foraging bout (up to 20 dives per sea otter). The observer recorded the following foraging metrics: prey item (to species level when possible), prey size (based on an estimated sea otter paw width of 5 cm and categorized into < ⅓ of the paw, > ⅓ and < ⅔ of the paw, or the whole paw), the proportion of the prey item consumed, GPS location (approximated based on GPS location of the telescope and distance/bearing to the sea otter), prey handling time (defined as the amount of time the sea otter spent manipulating and eating the prey), time spent diving, and total time spent at the surface. The following sea otter metrics were also recorded for each foraging bout: sex, reproductive status, and age class. Males were identified by the presence of a penile bulge, whereas females were identified if there was a clear lack of penile bulge, or if they had a pup. If sex was not confirmed, nor pup was observed, the sex was categorized as “unknown.” When possible, age class was determined as adult or juvenile by visual assessment of size and amount of grizzled fur .
We calculated the caloric intake for sea otters based on visual foraging observations using the Sea Otter Foraging Analysis (SOFA) program, which is based in Matlab (MathWorks) and maintained by the USGS Alaska Science Center in Anchorage, AK. SOFA uses a Monte Carlo-based simulation to account for unknown prey items and potential sampling bias. SOFA is a Bayesian model that provides the estimated biomass for individual prey types across time since recolonization, reproductive status, and sex. All SOFA outputs are reported as means with standard deviation. The consumption rates for each prey species were assigned for each foraging bout using the estimated prey size relative to a sea otter paw width. Prey diversity for each region was calculated using the Shannon-Wiener Index . Success rate, which is defined as the percentage of dives in a bout where the sea otter came up with food, was calculated for each sea otter metric.