Ringed seal (Pusa hispida) seasonal movements, diving, and haul‐out behavior in the Beaufort, Chukchi, and Bering Seas (2011–2017)

Abstract Continued Arctic warming and sea‐ice loss will have important implications for the conservation of ringed seals, a highly ice‐dependent species. A better understanding of their spatial ecology will help characterize emerging ecological trends and inform management decisions. We deployed satellite transmitters on ringed seals in the summers of 2011, 2014, and 2016 near Utqiaġvik (formerly Barrow), Alaska, to monitor their movements, diving, and haul‐out behavior. We present analyses of tracking and dive data provided by 17 seals that were tracked until at least January of the following year. Seals mostly ranged north of Utqiaġvik in the Beaufort and Chukchi Seas during summer before moving into the southern Chukchi and Bering Seas during winter. In all seasons, ringed seals occupied a diversity of habitats and spatial distributions, from near shore and localized, to far offshore and wide‐ranging in drifting sea ice. Continental shelf waters were occupied for >96% of tracking days, during which repetitive diving (suggestive of foraging) primarily to the seafloor was the most frequent activity. From mid‐summer to early fall, 12 seals made ~1‐week forays off‐shelf to the deep Arctic Basin, most reaching the retreating pack‐ice, where they spent most of their time hauled out. Diel activity patterns suggested greater allocation of foraging efforts to midday hours. Haul‐out patterns were complementary, occurring mostly at night until April‐May when midday hours were preferred. Ringed seals captured in 2011—concurrent with an unusual mortality event that affected all ice‐seal species—differed morphologically and behaviorally from seals captured in other years. Speculations about the physiology of molting and its role in energetics, habitat use, and behavior are discussed; along with possible evidence of purported ringed seal ecotypes.

They are an important component of the Arctic food web as a generalist predator (Crawford, Quakenbush, & Citta, 2015;Dehn et al., 2007;Lowry, Frost, & Burns, 1980), the primary prey of polar bears (Ursus maritimus; Stirling & Archibald, 1977), and a valuable subsistence resource for coastal Inuit people. Ringed seals are considered the most ice dependent of the four "ice associated" seal species in the western Arctic (Smith, Stirling, & Taugbøl 1991), which also include: bearded seals (Erignathus barbatus), spotted seals (Phoca largha), and ribbon seals (Histriophoca fasciata). They are well adapted to wintering within shore fast and pack-ice habitats-using their front claws to maintain breathing holes and to excavate lairs F I G U R E 1 (a) Three ringed seals (Pusa hispida) hauled out on multi-year ice in the southern Beaufort Sea near Utqiaġvik, AK. Note the molting fur on the center seal and the black face of a "rutting" male on the right. Daily CRAWL location estimates (n = 4,083) of the 17 ringed seals instrumented with satellite tracking tags are shown for July-September (b) and December-May (c). Colors distinguish seals tagged in 2011 (red, n = 5) from those tagged in 2014 and 2016 (blue, n = 2 and 10 respectively). The light gray contour at the 65 m isobath corresponds to the vertical line in Figure 7. The colored lines in (b) and (c) indicate the minimum and maximum extent of the sea ice in September and March, respectively, for each of the three tagging periods in snow that has drifted above these holes (Stirling, 1977). Female ringed seals give birth to and nurse their pups within snow lairs, which are important to pup survival because they provide shelter from the elements and concealment from predators (Smith, 1976;Stirling & Archibald, 1977;Smith, 1980;Smith, Stirling, & Taugbøl 1991;Stirling & Smith, 2004). Sea ice also serves as a platform on which ringed seals haul out during their annual pelage molt in spring (Fay, 1974;Smith & Stirling, 1975)-a time when their epidermis and fur are shed and replaced. This close relationship with sea ice suggests that ringed seals may be sensitive to changes in their habitat (Laidre et al., 2008). Given the ecological importance of ringed seals and the ongoing rapid changes to their sea-ice habitats, a characterization of ringed seal movements, diving, and haul-out behavior has practical applications for their management and can contribute to a better understanding of emerging ecological trends in the Arctic.
Previous studies of ringed seal movements in the Beaufort-Chukchi-Bering (BCB) Sea region reported seasonal and demographic movement patterns (Crawford, Frost, Quakenbush, & Whiting, 2012, 2018Harwood, Smith, & Auld, 2012;Harwood, Smith, Auld, Melling, & Yurkowski, 2015;Kelly et al., 2010), though broad-scale variability in these patterns appeared to be associated with capture location. For example, ringed seals tagged during  (Harwood et al., 2015) exhibited more localized movements and remained in the southwestern Archipelago throughout autumn and winter. Though similar in phenology, ringed seals captured in Kotzebue Sound, Alaska (Crawford, Frost, et al., 2012;Crawford et al., 2018) moved into the southern Chukchi and Bering seas during winter. Given the enormity of the BCB, the limited number of tracking studies to date, and the apparent differences in spatial distribution and movements associated with tagging locations, there remains a need to document and characterize the movements and behaviors of ringed seals from other locales within the BCB to more fully document this species' spatial ecology.
Here, we present seasonal movements, habitat use, diving, and haul-out behavior of ringed seals instrumented with satellite transmitters in the vicinity of Utqiaġvik (formerly Barrow), Alaska-a capture region not represented in prior tracking studies. We illustrate and quantify their movements and behaviors with respect to several geographic and demographic covariates. This work contributes to a growing body of literature about ringed seal spatial use, while also informing broader scale Arctic ecosystem monitoring efforts (Moore et al., 2014).

| ME THODS
Ringed seals were captured near Utqiaġvik, AK (71.3° N,156.8° W) during June-July of 2011, 2014, and 2016. All captures were made with nets that were set and continuously monitored near ice floes where seals had been observed. All nets had a lightweight lead-line and a highly visible float-line to ensure that entangled ringed seals could surface to breathe and that observers could readily determine when a capture occurred.
Upon capture, the seals were physically restrained during sampling and instrumentation. Biometric and demographic data were recorded (Table 1; Appendix A), including body mass, standard length, axillary girth, sex, and age class (Geraci & Lownsbury, 2005).
Age was determined by counting the alternating light and dark bands on the front claws. Seals with ≥5 claw bands were classified as adults (McLaren, 1958); otherwise, they were classified as subadults. One seal with no record of claw bands was designated as a subadult based on its small size and weight (Crawford, Frost, et al., 2012).
Satellite transmitters (hereafter "tags") provided location and dive data for each seal using the Argos System (Harris et al., 1990).
Most seals were instrumented with one primary and one secondary tag, and the data from each were combined into a single tracking time series. While secondary tags were expected to reveal haul-out locations for up to 2 years, only two of the secondary (SPOT) tags (n = 28) provided >1 year of data (Appendix A), so we limited this study to the first year of data collection. Because seasonal patterns were of primary interest, we also limited our analyses to seals with tags that provided data beyond December 31 of the deployment year (n 2011 = 5, n 2014 = 2, and n 2016 = 10).
Of the 17 seals in this study, all but two were instrumented with SPLASH tags (Wildlife Computers; 7.6 × 5.6 × 3.2 cm; 125 g in air) as their primary tag, while the remaining two seals were instrumented with CTD tags (Sea Mammal Research Unit; 10.5 × 7 × 4 cm; 545 g in air). All primary tags were attached using 5-min epoxy and/or cyanoacrylate adhesive to either the fur between the shoulder blades or on the head depending on the size of the seal. We anticipated that the primary tags would remain attached to the seals until shed during their annual molt the following spring-a duration of about ten months depending on tagging date. The primary tags provided data on movements, diving, and haul-out behavior. The 2011 primary tags (SPLASH) provided summary statistics for dive duration and maximum dive depth (for dives ≥3.5 m deep) as histograms, summarizing 6-hr time blocks. All primary tags deployed in 2016 recorded the start time, end time, and maximum depth (resolution = 0.5 m, ±1%) of each individual dive, where start and end times were detected by crossing a 1.0 m depth threshold, and they did not collect 6-hr histogram summaries. The primary tags deployed in 2014 were mixed; all collected 6-hr histograms and some recorded individual dives, but the dive end times relied on a saltwater sensor that was prone to incorrectly pool sequential dives when intervening surface events were not detected. Hence, we only analyzed dive-behavior data collected from tags deployed in 2016. All seals were also instrumented with a secondary tag (Wildlife Computers SPOT; 2.0 × 2.0 × 8.3 cm; 50 g in air). All secondary tags were permanently affixed to the rear flipper by screwing into a backing-plate through two holes punched in the interdigital webbing. Note: These seals reported location data beyond December 31 of the tagging year. Duration between the first and last location is presented as elapsed days, while CRAWL days denote the total number of days for which the CRAWL movement model estimated the seal's location with a standard error of <25 km. (3.5% of all location estimates), which was commensurate with the spatial scale of the lowest-resolution environmental data we used in our modeling. CRAWL locations were augmented with habitat metrics describing the distance to mainland (Wessel & Smith, 1996), bathymetry (National Geophysical Data Center, 2006), sea-ice concentration (Cavalieri, Parkinson, Gloersen, & Zwally, 1996;Maslanik & Stroeve, 1999), and distance to the sea-ice edge. The minimum distance to the ice edge was measured from the midday UTC (03:00 local time) location estimate to the nearest periphery of sea ice (≥15% concentration) composed of ≥10 contiguous 25 km pixels.
Because there was a 294-day gap between the higher spatial resolution AMSR-E and AMSR-2 sea-ice data sets, we used the 25 km resolution SSM/I sea-ice data in our analyses. Sea-ice concentration was based on the average value within a 50 km radius circle (excluding land) centered on the midday location.
Diurnal and seasonal haul-out behaviors were quantified using data from the primary tags, which binned daily summaries into 24 one-hour increments. The tag reported the fraction (%) of each hour that its saltwater sensor was wet or dry (sampled at 10 s intervals).
We defined hours that were ≥80% dry as "haul-out" hours. The 80% threshold was robust because the distribution of hourly percent-dry values was strongly bimodal; with 95% of all sampled haul-out hours (n = 62,279) being either ≥80% dry (11.5%) or ≤25% dry (83.5%). We excluded the first week of post-deployment behavior data prior to analysis to reduce potential biases associated with capturing seals close to shore.
Dive-behavior analyses were based on data retrieved from the primary tags deployed in 2016 (n = 10) and included parameters for dive duration, dive depth (maximum), and surface duration. We temporally paired the dive metrics with the nearest 6-hr CRAWL location and associated environmental attributes (e.g., ocean depth and sea ice). Dives were classified as bottom dives when the maximum dive depth was ≥75% of the mapped ocean depth. We did not attempt to classify bottom-dives in shallow water (<10 m) where relationships between dive depths and water depths become increasingly uncertain due to inaccuracies in both the location and bathymetry data.
By comparing successive dive depths and intervening surface intervals (Appendix C), we classified behavior as: (a) resting, when the surface interval between successive dives exceeded 10 min; (b) repetitive diving, when ≥5 sequential dives attain maximum depths within ±15% of either of the two preceding dives -single dives >15% different were allowable within a repetitive-diving episode; and (c) mixed diving, for all remaining dives not classified as repetitive.
We partitioned movement, dive, and haul-out data into two hab- Sea and much of the Chukchi Sea. The factor CapYear was included to assess whether a disease outbreak that began in 2011 may have influenced ringed seal movements and/or behavior. Ultimately designated as an "Unusual Mortality Event" (UME) (NOAA, 2011a(NOAA, , 2011b(NOAA, , 2012(NOAA, , 2014Stimmelmayr et al., 2013), none of the 2011 seals included in our analyses showed obvious symptoms (e.g., alopecia, lethargy, skin inflammation, or unusual molting patterns).
However, several other seals captured in 2011 were symptomatic.
In contrast, during both the 2014 and 2016 field seasons, no seals were observed to be symptomatic.
All models were constructed using R Statistical Software.
Prior to modeling, we removed the first week of data from each seal's time series to reduce any influence of capture and handling. Brooks et al., 2017) with a beta distribution and logit link. We transformed the data following Smithson and Verkuilen (2006) to address zeros and ones. To understand the use of sea ice when it was generally available, we partitioned the data for model set IV by Season, and developed models for the ice-covered period only.
Haul-out Time (model set V) was also analyzed with R-package glm-mTMB, but using a Poisson distribution and log link. In this model set, we were interested in understanding the factors associated with time spent when haul-out occurred (not whether haul-out occurred), and so we filtered our data to include only those days with ≥1 hr spent hauled out. As such, our models did not require adjustments that would otherwise be needed for zero-inflation. Finally, all models included random effects to account for individual variability among seals.
For each model set, we followed a systematic model selection procedure. First, we generated all possible single-and multivariate mixed models. We assessed model performance based on parsimony (using Akaike's information criterion; AIC C ) and then modified the highest performing models (ΔAIC C <2) by adding twoway interaction terms. The full model set was then re-assessed and ranked. Models within each set having the lowest AIC C were considered "best," though other models with ΔAIC C within 2.0 of the highest-ranking model were also deemed comparable (Burnham & Anderson, 2002). Visual inspections of residual plots from all significant models revealed no obvious deviations from homoscedasticity (Zuur, Ieno, Walker, Saveliev, & Smith, 2009). Finally, we used R-package emmeans (Lenth, 2019) to estimate marginal mean values from the best models (Appendix F) and to make multiple comparisons because this method is useful for summarizing the effects of factors when subjects are repeatedly measured and have unequal sample sizes (Lenth, 2016).

| RE SULTS
Of the 39 ringed seals captured and tagged (Appendix A), a total of 17 (Table 1) met the criteria for tag longevity in order to be included in our analyses. These 17 ringed seals included 12 adults (9 ♂, 3 ♀) and 5 subadults (2 ♂, 3 ♀). Mean body length was 102.5 cm (SD = 9.3) for adults and 88.2 cm (SD = 10.0) for subadults. The mean weight of adults was 40.7 kg (SD = 11.2) and subadults was 26.8 kg (SD = 8.0). The mean length (x = 92 cm, SD = 0.8) and weight ( should be noted that the maximum dive durations in each of those bins were consistently 10-12 min-which may represent the physiological dive-duration limit for ringed seals (Lydersen, Ryg, Hammill, & O'Brien, 1992).

Ringed seals instrumented with satellite transmitters near
Utqiaġvik, Alaska provided movement and dive-behavior data that both corroborated and expanded prior knowledge. The ringed seals in our study migrated to the southern Chukchi and Bering seas for winter, like those tagged by Crawford, Frost, et al. (2012) near Kotzebue, Alaska, and those tagged by Harwood et al. (2012) near the entrance of Amundsen Gulf, Canada.  for some seals, extensive for others, and habitats occupied were varied and widely distributed (from 57°N to 70° N latitude). Some seals stayed close to the coast in relatively shallow water, even stopping and maintaining a localized winter residency (Figures 1 and 3), while others went far offshore into the Bering Sea and moved continuously all winter in the dynamic pack-ice. Diversity in behavior and of habitats occupied suggests that, as a species, ringed seals can exploit a breadth of niches. We found some evidence of demographic habitat partitioning (Figure 4; Appendix F). Adults appeared to occupy winter habitats with higher sea-ice concentration, suggesting that different reproductive and life-history states (e.g., mating adults vs. growing subadults) may lead to different habitat requirements Crawford, Frost, et al. (2012). Adult and subadult ringed seals tagged by Crawford, Frost, et al. (2012) near Kotzebue, Alaska, wintered in distinctly different regions, with subadults moving farther south into the Bering Sea, while adults stayed primarily in the southern Chukchi Sea. Though we also noted evidence suggesting demographic differences in habitat use (Figure 4; Appendix F), our results were not statistically significant. Our results indicate, however, that year of tag deployment was important to understanding the movements of ringed seals (Figures 3 and 4), which may be important in light of the UME that began in 2011.
Most of the seals in our study (71%) made brief (~week long) off-shelf forays during summer that appeared to be deliberate and sometimes far-ranging efforts to reach the retreating sea ice ( Figure 5, Appendix G). Broad-scale movements by ringed seals during the open-water season are not unprecedented, such as populations in Svalbard that make long distance movements to seasonally access productive habitats (Freitas, Kovacs, Ims, Fedak, & Lydersen, 2008;Hamilton, Lydersen, Ims, & Kovacs, 2015). Offshelf movements by Utqiaġvik seals were notable because they Note: Sample (%) is the fraction of the month for which we obtained dive-behavior time series data for any given seal month. For each month, at least a 10% sample of the dive-behavior time series data was required for a seal to be included in the respective monthly estimate ("n" is a seal month). Analysis used the 8 SPLASH tags deployed in 2016. Seal months with an average distance from the coast of <5 km were excluded (n = 4 seal months).

TA B L E 2
Monthly estimates of the mean hours per day spent diving, and the proportion (%) of those hours spent engaged in episodes of repetitive diving to similar depths F I G U R E 7 Median dive depth recorded during episodes of repetitive diving to similar depths in relation to the average ocean depth at locations on the day of the diving. Analysis was restricted to days when seals were located in water 10-300 m deep. The solid red line denotes a 1:1 dive depth to ocean depth relationship, and the dotted red line denotes the dive-depth threshold for classification as bottom diving. The gray vertical line denotes the 65 m isobath which is demarcated in Figure 1 with the light gray shading. Note log scales on both axes. See Figure 8 for a summary of dive behavior based on the dive histogram data received from tags deployed in 2011. Dives that implausibly exceeded ocean depth were likely due to errors in the estimated seal locations, errors, or generalizations in the coarse-resolution bathymetry data, or imprecision in assigning locations to dives apparently abandoned more productive continental shelf habitat (Born, Teilmann, Acquarone, & Riget, 2004;Kingsley, Stirling, & Calvert, 1985;Teilmann, Born, & Acquarone, 1999) in favor of deep-water Arctic Basin habitat of generally lower quality (Frey et al., 2016). Given their dive-depth constraints (Lydersen et al., 1992), ringed seals that forage in deep water may have limited access to prey or incur higher foraging costs (Hamilton et al., 2015). Upon reaching the sea ice in the Arctic Basin, ringed seals spent more time hauled out than foraging. That 25% of these seals returned for a second time suggests potential benefits that may result from this behavior. This apparent motivation to haul out may reveal physiological constraints, such as those relating to the energetics of their molt (Crawford, Vagle, & Carmack, 2012;Majewski et al., 2016).
Distinct patterns in the dive data suggest that the ringed seals in our study frequently engaged in focused bouts of repetitive diving, the attributes of which are suggestive of active foraging behavior.
Specifically, tagged seals repeatedly dove to near-constant depths, showed near-constant dive durations and intervening surface times (Appendix C), and exhibited this behavior during substantial portions of the day (Table 2). Focused foraging behaviors can maximize energetic profitability when they result in repeated capture and consumption of aggregated prey-a strategy that makes energetic sense in patchy environments (Schoener, 1971). Repetitive diving also occurred more frequently during midday, when ambient light is brightest (Figure 10), and was spatially allocated in favor of habitats where prey species are known to aggregate-that is, the continental shelf seafloor (Benoit, Simard, Gagné, Geoffroy, & Fortier, 2010). If repetitive-diving bouts are indeed indicative of active foraging efforts, then their prevalence in the data show that ringed seals forage on average >12 hr/day from August through January (Table 2).
The tendency for most repetitive-diving bouts to occur at or near the seafloor (Figure 7) may be related to the ecology of their prey.
Ringed seals prey upon zooplankton (Lowry et al., 1980) and planktivorous fish (Crawford et al., 2018), both of which make synchronous diel vertical migrations (DVM) into deeper waters during the brightest hours of the day (Hays, 2003;Rabindranath et al., 2011;Stich & Lampert, 1981)-but, as potential prey themselves, face trade-offs between their own metabolic needs and predation risk (Pearre, 2003). Among Arctic cod (Boreogadus saida), which are an important forage species for ringed seals (Holst, Stirling, & Hobson, 2001), the larger and more energy-rich adults have greater metabolic stores and lower food limitation that enables them to remain longer at depth-decoupling them from closely following the DVM of zooplankton into shallower water where predation risk is higher (Benoit et al., 2010). Dense aggregations of adult cod that form in the demersal zone can physically displace smaller conspecifics into shallower water (Benoit et al., 2010;David et al., 2016;Farley et al., 2017). Thus, Arctic cod physiology and behavior may set F I G U R E 8 Dive histogram data corroborating daytime bottom-diving behavior among the ringed seals tagged in 2011. SPLASH tags deployed on ringed seals in 2011 (n = 5) provided summarized "histogram" data containing the number of dives to ocean-depth intervals during four 6-hr periods (charted here in local time, UTC-10 hr). We partitioned dive data into days when seals were located where the ocean depths were congruent with the six most commonly visited dive-depth bins and charted the relative proportion of dives in each depth bin, for each 6-hr period. Numbers in parentheses are the number of dives summarized in the respective chart. Results corroborate that most dives attained depths near the ocean bottom (as in Figure 7) and that deeper diving was more common during the midday (10:00-16:00) hours (as in Figure 10 partitioned by prey body-size (i.e., benefits to seals) and prey depth (i.e., cost to seals). Repetitive diving to the bottom may thus reflect optimal foraging (Waddington & Holden, 1979) in which larger and more energy-rich cod are targeted (Bowen, Tully, Boness, Bulheier, & Marshall, 2002). This behavior would be consistent with an energy maximization strategy (Bergman, Fryxell, Gates, & Fortin, 2001;Santini & Chelazzi, 1996) that invests more energy into deeper or longer dives to achieve a higher net energetic intake rate than would be possible by foraging on more accessible but less energetically profitable prey. It may also partially explain the tendency for larger bodied seals to dive less frequently, but for longer durations (Crawford et al., 2018).
Repetitive diving occasionally occurred in the very deep waters of the Arctic Basin (Figure 6b). This behavior has been reported previously (Gjertz, Kovacs, Lydersen, & Wiig, 2000) and may be related to concentrations of primary productivity occurring in the upper water column during the late summer/early fall (Ardyna et al., 2013).

Relatively shallow repetitive-diving bouts over the deep-water Arctic
Basin were occasionally punctuated by single dives to substantially greater depths (Figure 6a). Perhaps exploratory in nature (Simpkins, Kelly, & Wartzok, 2001), these intermittent deep dives are consistent with a strategy of searching alternative foraging patches to minimize lost foraging opportunities (Kohlmann & Risenhoover, 1998;Lima, 1985), which may be more profitable in habitats with lower prey densities, heterogeneously distributed prey, or when a foraging patch is nearing depletion (McNair, 1983). Our observation of ringed seals shifting their repetitive-diving behavior into deeper strata in the water column (Figure 6b) suggests that exploratory dives may have been profitable on occasion.
Temporal patterns of diving, resting at the surface, and hauling out (Figures 10 and 11) suggest that ringed seals modify their daily activities in response to ambient conditions and as an adjustment to the potentially high sensitivity of their prey to light (Berge et al., 2020). Repetitive diving to depths >25 m was more common during midday and became increasingly frequent at midday as day- Resting and haul-out were more prevalent behaviors at night (Figures 10 and 11). We found that during onset of the ice-covered season, seals hauled out more often during the darkest hours of the day (Figure 11), consistent with previously observed patterns of nocturnal haul-out behavior in ringed seals (Crawford et al., 2018;Härkönen et al., 2008;Kelly et al., 2010). Furthermore, diurnal patterns from the binned dive data reported by tags deployed in 2011 were consistent with the aforementioned patterns that ringed seals dove most often to depths near the seafloor and during midday ( Figure 8).
The relative value of habitat and the profitability of behavioral strategies may vary over annual cycles of ringed seal life history. For example, beginning in late spring, ringed seals undergo their annual pelage molt; an important physiological event in which several epidermal layers and the fur are shed and regenerated. This process is facilitated by infusing the epidermis with blood-providing the nutrients, oxygen, and warmth needed for tissue regenerationbut unsustainable levels of heat conduction from molting seals occurs when they are immersed in frigid Arctic waters (Boily, 1995).
The high metabolic demands of the molt (Feltz & Fay, 1966;Ryg, Smith, & Øritsland, 1990) (Young & Ferguson, 2013). Behavioral strategies that lower energetic losses while simultaneously accelerating completion of the molt should be favored (Berta, Sumich, & Kovacs, 2015;McLaren, 1958), as possibly evidenced in our data by long movements to distant sea ice followed by extended haul-out time in lieu of feeding. When considering the long-range movements that ringed seals made to the Arctic Basin in the mid-late summer, it seems plausible that the pursuit of available sea ice for the purpose of hauling out represents a behavioral strategy that weighs the relative quality of habitat against its value toward meeting a seal's physiological requirements.
The inclusion of the factor CapYear, which appeared in four of the five "best" models from our model sets (Table 3; Appendices E and F), was in response to two noteworthy events that occurred in 2011-2012. The first event was the emergence of a disease among ice seals that caused an abnormal molt, skin lesions, lethargy, mortality, and/or the unusual tendency to haul out on land (Herreman, pers. obs.). This disease was ultimately designated as an UME by NOAA.
The second noteworthy event was the unusually early breakup of the sea ice in July of 2011, which was followed in March 2012 by the greatest sea-ice maximum and mean sea-ice concentrations recorded in the Bering Sea since start of the satellite record in 1979 (Fetterer et al., 2017). It is conceivable that annual variations in sea-ice dynamics can drive physiologically mediated seal behavior. Basin where they hauled out more than they foraged. The distinctive morphology, behavior, and spatial distribution of the seals tagged in 2011 do call attention to reports of two purported ringed seal ecotypes: (a) a smaller pelagic "pack-ice seal" and (b) a larger coastal "fast-ice seal" (Fedoseev, 1975;Finley, Miller, Davis, & Koski, 1983;Freuchen, 1935

TA B L E 3
Top models explaining variance in movement rate, distance from the mainland, concentration of sea ice occupied, distance from sea-ice edge, and haul-out duration as a function of sex, age class, season, and capture year for haul-out, prey access, and seal physiology. Under ideal conditions, hauling out on sea ice in high-quality foraging habitat (i.e., continental shelf) could enable molting ringed seals to partially offset energetic costs accrued from reproduction, lactation, and molting (Ryg & Øritsland, 1991), particularly if they can profitably capture prey. However, earlier northward retreat of pack-ice (Comiso, Meier, & Gersten, 2017)  Off-shelf forays in 2014 and 2016 were less frequent and occurred earlier in the summer ( Figure 5). Currently, it is unknown whether a protracted or otherwise complicated molt (e.g., UME) could motivate seals to make late-summer forays to the retreating pack-ice in order to haul out. A more complete understanding of phocid molting physiology with respect to energetics may help clarify the drivers of this behavior, including the relative value of habitat over the course of a seal's annual life-cycle. The quality of a habitat (i.e., its value to an animal's fitness) is a function of local environmental conditions and eco-physiological constraints (Charnov, 1976;Lima, 1983), the interactions of which can shape habitat selection via the profitability of different behaviors. How this occurs may not be straightforward and, given their dynamic environment and the many possible scenarios encountered by ringed seals, is likely the net sum of numerous behavioral adjustments that optimize energy intake given the relative ratios of costs and benefits (Born et al., 2004;Ferguson & Higdon, 2006;Stephens & Krebs, 1986).
Based on their high abundance and wide distribution (Reeves, 1998), ringed seals are a very successful species, likely due to behavioral plasticity that has allowed them to exploit a variety of habitats throughout the circumpolar north. To date, ringed seals in the Bering and Chukchi seas have not exhibited declines in body condition, growth, or reproduction observed in other populations (Crawford et al., 2015). In the face of an accelerating trend toward earlier, more rapid, and/or more extensive summer sea-ice melt (Comiso et al., 2017), as well as recent dramatic losses of winter sea ice in the Bering Sea (Siddon & Zador, 2018), a more comprehensive understanding of the energetic consequences and behavioral tradeoffs (Laidre et al., 2008) faced by ringed seals throughout their life-cycle is needed to help guide their conservation and management.

| CON CLUS IONS
This study adds to a growing body of knowledge about ringed seal movements and behaviors. Seals were captured in a region that had received little prior investigation and were instrumented with satellite transmitters capable of providing location, information about individual dives, and hourly haul-out status. Like other ringed seal tracking studies in the Beaufort and Chukchi seas, most of the seals we tagged near Utqiaġvik moved into the southern Chukchi and Bering seas during winter. They occupied a diversity of habitats and spatial distributions, from close to shore and very localized, to far offshore and wide-ranging in the drifting sea ice. The ringed seals we captured in 2011, concurrent with a UME that affected all ice-seal species, were physically smaller than seals captured in other years and maintained a more pelagic distribution, raising speculation that the UME could have facilitated the tagging of a "pelagic" ringed seal ecotype that would not have otherwise been available for capture nearshore. Many ringed seals, especially those tagged in 2011, made forays into the deep Arctic Basin with an apparent intent to reach the pack-ice to haul out. Focused bouts of repetitive diving occurred over the continental shelf for >12 hr/day, usually to depths at or near the ocean floor. Hauling out tended to be progressively more nocturnal from winter to early spring; but abruptly switched in May to a pronounced daytime haul-out pattern with onset of the molt.
As a "threatened" species (Endangered Species Act [ESA]) (National Marine Fisheries Service, 2012), recovery criteria for ringed seals is drawn from the best available science about their habitat use and behavior; as well as knowledge about the dynamics of pinniped populations overall (Conn et al., 2014). Given the potential for increases in human/wildlife conflicts in the Arctic (Harsem et al., 2015;Smith & Stephenson, 2013), mitigation and recovery strategies for ringed seals will benefit from better information about their movements and behavior. Ongoing conservation efforts for polar bearsanother ESA threatened species-will also benefit from an improved ecological understanding of ringed seals (Durner et al., 2009;Wilson, Horne, Rode, Regehr, & Durner, 2014). And, because the Arctic is a stochastic environment (Walsh, 2008) where rapid climate mediated change is already occurring (Post et al., 2013), continued research that fills gaps in poorly sampled regions will contribute to a more comprehensive understanding of the Arctic as an ecosystem, and therein the eco-physiological processes that are important to the conservation and management of ringed seals-a vulnerable species with high ecological and cultural value (Condon, Collings, & Wenzel, 1995;Huntington, Quakenbush, & Nelson, 2016). Government.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data set analyzed for this study is available from the Dryad Digital Repository https://doi.org/10.5061/dryad.zpc86 6t65 (Von Duyke, Douglas, Herreman, & Crawford, 2020).