Population and Distribution Surveys

From 2017 through 2023, winter surveys of sea otters were done in early to mid-February, spring surveys in mid to late April, summer surveys in July, and fall surveys in late October to early December (as weather conditions and island access permitted). Seasonal surveys were suspended from spring 2020 through spring 2021 due to limited access of USGS personnel to the island during the coronavirus disease 2019 (COVID-19); only one fall survey was completed in December 2020, and it was done with a reduced survey team. Seasonal surveys were resumed in summer 2021 and have continued uninterrupted to 2023, except for the winter 2023 survey, which was cancelled after strong storms caused extended road closures and poor overall survey conditions.

Surveys were completed following similar protocols as the annual range-wide southern sea otter census in California (Hatfield and others, 2019), with the exception that an experienced observer surveyed alone at SNI (whereas a principal observer usually surveys with an assistant during the mainland range-wide census), and there were usually two or three attempts to count the entire island on different days of the survey trip. For each survey day, each observer was assigned to a section of the island spanning one or more of the nine survey areas (fig. 1). Each observer was equipped with a high power 50–80X spotting scope (Questar Inc., New Hope, Pennsylvania) and binoculars (10X). The observers started at one end of their assigned section and selected an observation point that provided good vantage of a viewable segment of habitat. In most cases, observers used the same counting locations year after year, for consistency. The observers scanned the segment with unaided eye and with binoculars for sea otters or objects that were suspected to be sea otters. Large groups, distant areas, and suspicious objects were scanned with the spotting scope. After sufficient time elapsed (15–30 minutes, depending on viewing conditions and extent) to make a thorough count of all sea otters in the first surveyed segment of habitat, the observer moved to another location that provided good viewing of the next segment of habitat, contiguous with the first segment. This process was repeated until the entire section was counted.

All sea otter observations were marked as “points” directly onto paper field maps, using the area’s primary stationary features, such as the shoreline and offshore rocks, for spatial reference. Kelp canopies from an earlier year also were displayed on the maps for use as secondary reference features due to their tendency to persist; however, kelp contracts, expands, and shifts across months and years, making the kelp distributions somewhat less reliable than stationary features for marking the precise locations of sea otters for comparisons across surveys. The kelp features, however, allowed sea otter locations to be mapped relative to all available visual cues within the same survey, which reduced the chances of double-counting or undercounting when the observer moved to the next observation point.

To minimize double-counting or detection biases because of sea otter movements, each total island count included surveys of all nine areas of the island on one survey day. In addition to the sea otter locations, other data recorded directly onto the maps included the number of independent sea otters, number and relative age of pups (pups are classed as either small or large depending on the presence or absence of natal pelage, relative size, and behavior), behavior (resting, foraging, or other), group size, and micro-habitat type (open water or kelp). Time and general survey conditions also were recorded. Viewing conditions were rated from excellent to poor (coded 4=excellent, 3=very good, 2=good, 1=fair, 0=poor), and survey trips were done when fair or better conditions could be anticipated. Shore-based surveys are estimated to detect about 90–95 percent of the sea otters located in the nearshore area (Estes and Jameson, 1988), but distances exceeding 1-kilometer (km) offshore and factors such as viewing conditions, sea otter grouping, and behaviors can influence detectability. Surveys were repeated on multiple days during each seasonal survey trip to SNI, and the maximum single-day total was taken as the island-wide count to minimize undercounting due to detection biases.

Ecosystem Monitoring

Underwater divers performed subtidal monitoring of kelp forest species community composition at four sets of permanently established transects at SNI (areas 1, 5, 7, and 8) semi-annually in April and October in most years since October 1980. Information on these surveys through fall 2019 are reported in detail in Kenner and Tomoleoni (2021). Subtidal kelp forest monitoring was suspended in 2020 and 2021 due to the COVID-19 pandemic, was partially resumed in 2022, and was fully resumed by spring 2023. Because sea urchins are a primary prey item of sea otters, we examined the subtidal counts of red sea urchins (Strongylocentrotus franciscanus) and purple sea urchins (S. purpuratus) and compared their patterns during the same periods as the sea otter surveys.

In addition to subtidal kelp forest monitoring, we obtained multi-spectral digital island-wide imagery of the survey areas around SNI annually (during fall 2014 through 2019 and 2022). This work was contracted by the U.S. Navy and collected from aircraft by the Ocean Imaging Corporation (http://www.oceani.com). The digital imagery was recorded with 0.2 meter and 1.0 meter resolution as red, green, blue, and near-infrared bands, and pixels were subsequently classified into exposed or submerged giant kelp (Macrocystis pyrifera) canopies based on supervised learning algorithms. We obtained the georeferenced aerial imagery and kelp classifications in raster and shapefile format and used them with geographic information systems (GIS) to develop maps of sea otter distributions relative to areas with kelp canopy.

Foraging Surveys

Sea otter foraging data were collected during the same seasonal trips as the sea otter population surveys, opportunistically during the times that were not already dedicated to those population surveys. Observation methods followed well-established protocols (Estes and others, 1981; Ralls and others, 1995; Tinker and others, 2008). Field observations were done during daylight hours by teams of 1–2 observers that searched for foraging sea otters near enough to shore that they could be visually distinguished from other sea otter individuals. The sea otters were monitored using a 50–80X spotting scope. Foraging bouts, which are defined as contiguous sequences of feeding dives made by a focal sea otter, were observed until data from at least 20 dives in the bout were recorded or until visual contact with the focal sea otter was lost. The information recorded during these bouts included location, date and time, duration of the subsurface dive interval (dive time) and the post-dive surface interval (surface time) for each feeding dive (in seconds), outcome of each dive (in other words, whether or not prey was captured), species of prey captured, number and size of prey items, per-item handling time (number of seconds required to handle and consume each item), whether or not tools were used to handle the prey, and ambient conditions (including sea-state, wind, and so forth). Prey size was recorded as the estimated diameter of the shell or maximum body dimension (excluding appendages) and categorized into 5-centimeter incremented size-classes. Each 5-centimeter increment represents one typical sea otter paw-width and was subdivided further into thirds within that size-class for increased precision. For observations where prey could not be reliably identified to species, the items in question were assigned to the lowest possible taxonomic unit. Any items that could not be reliably categorized to any taxonomic level were listed as unknown prey, although size class and number of items were still recorded. Additional information recorded by observers included numbers of prey items that were carried over from a previous dive or stolen by or from the focal animal and, in the case of females with dependent pups, the amounts of these prey that were shared with the pups.

Analysis

Abundance and Trends

Shortly after the translocation of 140 sea otters to SNI from 1987 to 1989, most of those sea otters disappeared from SNI, leaving a population of fewer than 20 sea otters during most of the first decade (table 1). However, this population had gradually grown to exceed 100 individuals by the time of the spring 2016 survey and continued to exceed 100 individuals in all 14 surveys from fall 2018 through spring 2023 (fig. 2). Current survey estimates average about 150 total individuals, which is a sixfold increase in the population since 2000 when sea otters numbered fewer than 25 (table 2; figs. 2, 3).

The sea otter count, as of the last spring survey in April 2023, was 146, which included 134 independents and 12 pups, and the estimated rate of population growth during the 2020–2023 period was ${\lambda }_{2023}=1.10$ (95-percent CI =1.00–1.20), corresponding to a 10-percent annual growth rate during this period (95-percent CI = 0–20 percent). Annual population growth rates varied during the past three decades, with about half of the estimates and 95-percent CI exceeding 1, representing statistically positive annual growth, with most growth occurring after 2010 (figs. 2, 4).

The pup-to-independent ratio for surveys in the 2020–2023 period was 8.8 pups to 100 independent sea otters, which was lower than the pup-to-independent ratio for surveys in the previous period (2017–2019; 12 pups per 100 independent sea otters). Lower pup ratios are associated with periods following larger rates of growth as more juveniles and subadults recruit into the pre-breeding population.

Densities and Distributions

We compared the densities and distributions of sea otters during surveys from the 2020–2023 period with those of the 2017–2019 period. Sea otter locations were aggregated across all surveys during each period to develop kernel density surfaces representing the relative density of cumulative counts (fig. 5). In these maps, sea otters appear to have shifted from concentrations distributed around the island, primarily at the west end (areas 1–3) and north side (areas 4 and 5) during 2017–2019, to a large concentration at the east end (area 7) during 2020–2023, leaving less dense concentrations in the west and northwest.

The maps on figure 5 also display sea otter densities overlaid with kelp canopy distributions derived from aerial imagery collected during the last fall within each respective sea otter survey period. Surface kelp changes significantly over the seasons, with the maximum extent typically occurring in the fall. Kelp surveys were done once annually and were scheduled in the fall with the intent of capturing images of the maximum extent of kelp canopy for that year. After winter storms, the area of surface kelp is largely reduced, and the distribution changes seasonally. Thus, these kelp distributions are not strictly representative of where kelp beds were across the other seasons when sea otter surveys were done, but they are an indicator of where kelp beds had the most potential to be present.

Radial plots (also known as radar or spider plots) provide a graphical summary of the distribution of sea otters counted around the perimeter of the island (Bion, 2020; Nakazawa, 2023). The nine areas defined on figure 1 are arranged on circular plots in the general cardinal direction as they are oriented around the island (figs. 6, 7). Most of the population distribution, which was formerly concentrated on the island’s west end (areas 1 and 3), expanded and increased on the north and south side during 2017–2020 but has since concentrated more toward the southeast end (areas 7 and 8) during the past several years (fig. 7).

Radial plots display the distribution of sea otter use around the island as it varied seasonally and by year (fig. 7). Areas 1 and 3, which include The Boilers, on the west end have consistently had among the largest sea otter counts in the summer and fall from 2017 through 2022. From summer 2019 through winter 2020, areas 4 and 5 combined had a visible spike in sea otter use, starting in area 5 and then shifting more to area 4. The south side of the island did not have much use except sometimes at area 8 near the east end (area 7) where sea otter use has increased from 2020 to 2023, mostly in the spring but also in the summer and fall. Area 6, which is on the eastern end of the north side, where there is almost no kelp, has had very few sea otters in all seasons and years.

Spatial density maps for each of the four seasons reflect these seasonal spatial distribution patterns in greater detail for the recent years (from 2020 through 2023; fig. 8). Sea otters continued to frequent areas at the island’s west end in all seasons and years. However, during 2020–2023, sea otters were observed to aggregate in higher concentrations near Daytona (in area 7 at the east end), notably in spring, summer, and fall, and they were observed near Dutch Harbor in area 8, on the south side in spring.

To estimate and test seasonal differences in the spatial distribution of sea otters generally, we aggregated sea otter counts within the west end, north side, east end, and south side sections (fig. 1). We detected significant differences in distribution among sections between the 2017–2019 and 2020–2023 periods (${\chi }_3^2=40.3$, $p<0.0001$) and season (${\chi }_9^2=54.6$, $p<0.0001$), but there was no significant evidence of trends among years within either of the two periods (${\chi }_3^2<5.3$, $p>$0.15) or interaction effects between period and season (${\chi }_9^2=9.1$, $p=0.42$). Concentrations of sea otters were on the west end each season, in areas associated with historically persistent kelp, but concentrations tended to be lowest in winter or spring (figs. 8, 9). Despite indications that sea otter distributions expanded to both the north and south sides of the island during the 2017–2019 period, when the south side was formerly defined to include area 7 (Yee and others, 2020), there was little indication that this pattern of growth on the north side continued into the 2020–2023 period (figs. 5, 6, 7, 8). However, large concentrations of sea otters continued to persist and grow in area 7, herein referred to as “the east end,” and in area 8 of the south side during the 2020–2023 period (area 7; figs. 5, 6, 7, 8, 9).

Foraging

Sea otter foraging data were collected opportunistically as time and viewing conditions allowed. Consequently, the numbers of foraging observations varied by survey and location. A total of 60 sea otter foraging bouts were observed from winter 2020 through spring 2023, for a total of 773 diving observations (tables 3, 4). The amount of forage data collected from 2020 through spring 2023 was substantially less than that collected in the previous period 2017–2019 (2,581 dives) due to fewer surveys completed during the COVID-19 pandemic and sometimes poorer survey conditions. Most foraging bouts were recorded on the east end (70 percent) compared to the north side (18 percent), south side (7 percent), and west end (5 percent). Among all dives, 368 were observed on 21 female foraging bouts, 179 were observed on 15 male foraging bouts, and the remaining 226 were observed on 24 foraging bouts of unknown sex.

We classified identifiable prey species into taxonomic groups based on highest prevalence in the diet (urchins, bivalves, snails, crabs) or energetic content of less prevalent prey species (abalone, lobsters, octopus) and lumped all other species (other). Of the 587 successful dives observed during 2020–2023, 32 percent resulted in unidentified prey items (table 5). We used Monte-Carlo simulations to estimate the likely caloric values of unidentified prey based on partial data about the prey, such as their size and the lowest taxonomic level to which they could be identified. These simulations use empirical information about the likelihood of different prey species and the associated caloric values for their size class, based on the frequency with which specific species and sizes were identified in the data. This process enabled us to derive 95-percent confidence intervals to account for prey uncertainty. These simulations require robust sample sizes to provide sufficient information on prey. Because our foraging observations in the 2020–2023 period were fewer compared to those of previous years, we aggregated those data across years to calculate an average energy intake rate for the 2020–2023 period.

Sea urchins, including the red sea urchin (Strongylocentrotus franciscanus) and purple sea urchin (S. purpuratus), composed the majority (66 percent) of identified prey types among all successful diving outcomes, with the remaining successful diving outcomes comprised mostly of bivalves (15 percent), snails (12 percent), and crabs (5.2 percent; fig. 10). Three abalone, two lobsters, and one mollusk also were observed to be consumed (table 5). Bivalves comprised a larger proportion of prey in the 2020–2023 period than previously observed and were recorded for 61 dives from 11 bouts, including 1 bout where 21 rock jingles (Pododesmus macroschisma) and 3 clams were observed to be consumed and another 8 bouts where a total of 23 giant rock scallops (Crassodoma gigantea) were observed to be consumed.

We examined the composition of sea urchin species, which is the most identified prey among successful sea otter foraging dives. Red sea urchins generally are larger (median diameter =53 millimeters, interquartile range [IQR; 25–75-percentile] = 41–65 mm) at SNI and thus have higher caloric value than purple sea urchins (median diameter =28 mm, IQR =21–36 mm; Kenner and Tomoleoni, 2020). Because sea otters are size-selective predators, they tend to prefer red sea urchins when larger size classes of red sea urchins are more available. As such, changes in the ratio of red to purple sea urchins consumed may correlate with prey availability. We inspected this ratio during the 2020–2023 period and compared it to ratios calculated annually from 2003 to 2006 and from 2017 to 2019 when similar sea otter foraging surveys were done at SNI. Results indicated a decreasing trend in the ratio of red to purple sea urchins consumed, detectable in 2017–2019, and pronounced by 2020–2023 (fig. 11).

Simulated estimates of energy intake rates averaged 14.0 kilocalories per minute (kcal/min; 95-percent CI =10.8–17.2) during 2020–2023. This estimated rate, slightly lower than the average rate that was previously estimated for 2019 (15.5 kcal/min; 95-percent CI =12.9–18.0), continues a declining trend of energy intake rate estimates beginning with 2017 (20.9 kcal/min; 95-percent CI =16.4–25.3; fig. 12; Yee and others, 2020); however, the confidence intervals of these estimates overlapped and did not conclusively indicate a decline. Dietary diversity trended positively from 2017 to 2018 and 2019 (Shannon-Weiner index [SWI] =0.67, 0.86 and 0.95, respectively), but this pattern of increase did not continue into 2020–2023 (SWI =0.81).

Subtidal monitoring at four sites surrounding SNI following the methods of Kenner and Tomoleoni (2021) showed variations in abundance and trends of sea urchins, by site and by species (fig. 13). Counts of red sea urchins increased by 1–2 orders of magnitude at the west end site but decreased by the same degree at the Daytona site on the east end. Because of these site differences, we did not detect an overall trend in the abundance of red sea urchins (${\chi }_1^2=0.65$, $p=0.42$). However, purple sea urchins experienced larger and more substantial increases over time (${\chi }_1^2=31.1$, $p<0.0001$) compared to red sea urchins, especially at the Daytona site in area 7 on the east end. Counts of purple sea urchins were larger than or close to those of red sea urchins across sites during the 2003–2006 period, when foraging survey results indicated that sea otters had consumed disproportionately high numbers of red sea urchins compared to purple sea urchins. However, by 2020–2023, purple sea urchins exceeded red sea urchins by 1–3 orders of magnitude across all sites, coinciding with the distributional shift of sea otters to the east end and the dietary shift toward higher relative consumption of purple sea urchins.