Refining Aging Criteria for Northern Sea Otters in Washington State

Reliable age estimation in wildlife species is invaluable for determining age-specific demographic data (e.g., mortality and reproduction) used to inform policy and management priorities. On the Pacific coast of North America, sea otter recovery has been a priority for the two subspecies: the northern sea otter, E. l. kenyoni, which ranges from Alaska south to the Oregon coast; and the southern sea otter, E. l. nereis, found along the California coast. Recovery strategies greatly differ for southern sea otters, whose decline in the late 1990s has been attributed to mortality among prime-age breeding females (Tinker 2004) as compared with collapse across all ages and genders of northern sea otters in the Aleutian Islands (Estes et al. 2005). Demographic data are particularly limited for northern sea otters in Washington State as a result of geographic isolation (Jameson and Jefferies 2010); they reside along the predominantly undeveloped coastline of the Olympic Peninsula, which makes long-term observational studies difficult. In fact, population models have relied heavily on data such as habitat use from sea otters in other geographic locations (Laidre et al. 2002; Gerber et al. 2004). Age determination of beach-cast carcasses combined with survey data has also been used to develop demographic models for sea otters in Alaska and California (Monson et al. 2000; Tinker et al. 2006), and this approach may be particularly suitable for Washington sea otters, which are currently protected under Washington law with State Endangered status (Washington Administrative Code 232-12-014) and the Marine Mammal Protection Act of 1972 (16 U.S.C. 1362-1407). However, a critical component of these models is the ability to reliably estimate age in beach-cast carcasses.

Previous attempts to develop reliable criteria for estimating age in sea otters have included methods such as cementum annuli analysis, tooth eruption and wear patterns, skull measurements and suture closure, body weight and length measurements, numbers of ovarian corpora albicans, baculum measurements, and pelage coloration (Lensink 1962; Schneider 1973; Morejohn et al. 1975; Garshelis 1984; Bodkin et al. 1997). Cementum annuli analysis became the primary standard of aging after Bodkin et al. (1997) established an accuracy estimate for sea otters in Alaska that was comparable to that of other long-lived carnivore species, with an upper accuracy limit (proportion of age estimates without error) of 0.85. Despite the potential for errors (>1 y) in aging estimates based on cementum annuli (Bodkin et al. 1997), agreement between field aging estimates and cementum annuli appears relatively good for sea otters in Alaska (Monson et al. 2011). However, discrepancies between cementum annuli and pathologist-assigned age at necropsy (based on total length, tail length, presence of thymus, reproductive status, and pelage coloration/grizzle) were noted for 17 of 65 sea otters analyzed from Washington (accuracy estimate of 0.73; N.J. Thomas, personal communication), and discrepancies also appear to be more pronounced for California sea otters (J.A. Ames, California Dept. of Fish and Wildlife, personal communication).

More marked seasonal changes in food availability and daylight in Alaska could result in more prominent cementum layering compared with those in California and Washington, or may be the result of differences in diet, nutritional status, reproduction, dental disease, or trauma (Grue and Jensen 1979; Bodkin et al. 1997; Costello et al. 2004; Von Biela et al. 2007; Medill et al. 2009). Differences in Washington sea otters may also be due to phenotypic divergence resulting from founder effect because they descended from relatively few individuals (≤43 sea otters survived post translocation; Lance et al. 2004), and the genetic diversity within Washington is reduced compared with both the source population at Amchitka Island and the pre–fur-trade sea otter population (Jameson et al. 1982; Bodkin et al. 1999; Larson et al. 2002, 2012). The objective of this study was to develop methods to improve accuracy of demographic data by identifying a set of characteristics that could reliably categorize individual sea otters from Washington into established age-class categories: pups (dependent), juvenile (weaning-age), subadults (near or at full growth, but not reproductively active), adults (reproductively active), and aged adults (>10 y old; Bodkin et al. 2000; Tinker 2004).

It was not possible to obtain known-age sea otter specimens from Washington. Thus, we compared two established aging methods (i.e., cementum annuli counts and skull ossification patterns) to determine if there was sufficient agreement between results to support use of one or both methods (Lensink 1962; Marks and Erickson 1966; Hoffmeister and Zimmerman 1967; Zimmerman 1972; Morejohn et al. 1975; Junge and Hoffmeister 1980). We then assessed a suite of morphologic, physiologic, and developmental characteristics to determine variables that provide aging results similar to those estimated by the established aging methods.

Between February 1989 and July 2011, 82 beach-cast sea otter carcasses from Washington State (males: n = 52; females: n = 30) were shipped to the U.S. Geological Survey – National Wildlife Health Center (Madison, Wisconsin) for a complete diagnostic necropsy to determine cause of death. A maxillary premolar (n = 65) was removed and sent to Matson's Laboratory (Milltown, Montana) for cementum annuli analysis as described by Bodkin et al. (1997). A subset of skulls (males: n = 30; females: n = 20; see Table S1, Supplemental Material) were cleaned at the University of Wisconsin Zoological Museum (Madison, WI) by dermestid beetle larvae in stainless steel tanks. The skulls were then soaked in 50% ammonium hydroxide solution for 10–30 min, rinsed, and dried. Ammonium hydroxide immersion was not used for skulls from neonates to avoid disarticulation of skull sutures.

We categorized the skulls into three age classes based on methods established in previous sea otter aging investigations (Lensink 1962; Morejohn et al. 1975; Roest 1985; Hattori et al. 2003: pup–juveniles, subadults, and adults–aged adults. We differentiated pup–juveniles from subadults by obliteration of the exoccipital–basioccipital suture (Lensink 1962) and rostral growth of the caudal lip of the glenoid fossa in subadults (Morejohn et al. 1975). We categorized a specimen as an adult–aged adult if the basioccipital–basisphenoid suture was obliterated and the mandibular condyles were locked (Morejohn et al. 1975). We refer to this method herein as the “established skull-aging method.”

However, obliteration of the basioccipital–basisphenoid suture occurs around 3 y of age (Lensink 1962; Hattori et al. 2003), which is within the age range (2–4 y) that female sea otters become reproductively active (Bodkin et al. 1993); however, males may not become reproductively active until 5–6 y of age (Schneider 1978). Based on the premise that cranial sutures become obliterated as animals age, we expanded on the established skull-aging method to determine additional skull ossification or development patterns that could be used to differentiate male subadults (1–5 y) from adults (6–10 y), as well as separate adults from aged adults in both genders. We refer to this expanded method herein as the “alternative skull-aging method” (Figure 1).

Using a Fowler/Sylvac digital caliper (150 mm, model Ultra-Cal Mark III, resolution 0.01 mm; Newton, MA), we measured seven morphometric characters (Figure 2) based on previous investigations of sea otter skulls (Scheffer 1951; Lensink 1962; Roest 1973, 1985; Morejohn et al. 1975; Hattori et al. 2003) to investigate differences between adults and aged adults. Due to perimortem traumatic skull damage and brain removal at necropsy, sample sizes varied (range of n = 42–48) for measurable skull characters. We conducted all measurements as distances between landmarks.

To distinguish between pups and juveniles, we detailed dental eruption patterns in a closely related subspecies (E. l. nereis) using known-age, live pups. We compiled longitudinal tooth-eruption records for live-stranded southern sea otter pups (males: n = 13; females: n = 8) undergoing rehabilitation at the Monterey Bay Aquarium between 2004 and 2011. To reduce potential error due to inaccuracy, we limited analysis to include only pups estimated to be <28 d of age at stranding (x̄ = 14; SD = 8.2). We estimated age at stranding for each individual by comparing total length, body weight, dental eruption pattern, presence or absence of remnant umbilicus, evidence of meconium in feces, and initial observations of behavior (Kenyon 1969; Payne and Jameson 1984).

We performed intraoral examinations on each pup at irregular intervals during rehabilitation until release at 32–40 wk of age. We noted eruption status (first erupting, partially erupted, or fully erupted) of each deciduous and permanent tooth at each examination (Table 1), which we performed on conscious individuals. We restrained animals in dorsal recumbency, and opened the oral cavity using gloved fingers (for pups <12 wk of age), or a combination of gloved fingers and flexible plastic bite-bar (for pups >12 wk of age). For each pup, we performed at least one oral examination while the pup was anesthetized for transmitter implant surgery (in preparation for release) between 24 and 28 wk of age. Frequency of intraoral examinations decreased with age, as pups became significantly more difficult to restrain. As a result, sample sizes for estimating timing of later erupting teeth are smaller than for deciduous and earlier erupting permanent teeth (Table 1). We developed a standardized dental chart with occlusal and buccal views using the modified Triadan system for tooth identification (Floyd 1991) and tooth condition scores adapted from Pattison et al. (1997) for scoring tooth-wear pattern and severity in the skull specimens (Figure S1, Supplemental Material).

We used the necropsy report for each specimen to determine standard total length (cm; n = 78), evidence of reproductive maturity in females (presence of mammary gland development, n = 29; presence of placental bands or scars, a fetus, and/or uterine development, n = 25), thymus presence (present or absent; n = 66) and size (large, moderate, small, or remnant; n = 57), and grizzle rating (1—none to slight amount at the tip of the nose, 2—extending to the eyes, 3—extending to the lamboidal crest, 4—extending to the chest, and 5—extending to the tail; n = 51). Sample size varied for each characteristic (range of n = 25–78) because all characters were not measured on each individual. We did not assign thymus characteristics to individuals that died from morbillivirus infection (n = 6) because of the potential for viral proliferation in lymphoid tissue and thymus atrophy (McCullough et al. 1974; Krakowka et al. 1980; von Messling et al. 2004; Di Guardo et al. 2005). We did not use body weight because of dependence on antemortem (e.g., disease, nutritional plane, and reproductive status) and postmortem factors (e.g., scavenging and decomposition).

We fit a linear regression model to the data to determine whether measurements associated with both established and alternative skull methods were significantly associated with cementum annuli, while controlling for sex and county of origin. We defined statistical significance of the correlation between cementum annuli and age based on skull ossification and development using both established and alternative skull methods according to an added-last (Type III) variable-selection test at α-level = 0.05. We then categorized individuals as subadult (males: ≥1 y, <6 y; females: ≥1 y, <3 y; n = 28), adult (males: ≥6 y, ≤10 y; females: ≥3 y, ≤10 y; n = 18), and aged adult (>10 y; n = 11) categories based on cementum annuli in preparation for analyzing the remaining data (skull measurements, tooth condition, and physiologic and morphologic characters). There were a small number of aged adults, so we ran analyses with aged adults combined with adults (adult–aged adult), and then separated from adults. However, a cementum annuli result of zero was associated with both the pup and juvenile categories; therefore, we included neither of these categories in the following analyses. We used logistic regression (response variables: subadult and adult–aged adult) and ordered probit regression (response variables: subadult, adult, and aged adult) to determine if individual skull, morphologic, and physiologic characters were significant predictors of age class

We ran the regressions with gender and county as covariates. We defined models used in the classification-tree-fitting processes according to a two-stage regression approach. The first stage of regressions consisted of simple regressions that included only one predictor variable and covariates for sex and county of origin. This first stage involved 38 regressions associated with skull, tooth condition, and physiologic morphometric variables. The second stage of regressions consisted of conducting multivariable regressions, which included all significant variables from stage 1 for a given morphometric category, along with sex and county as covariates. We considered morphometric variables to be significantly associated with age class according to Wald tests corresponding to slope parameter estimates. This is akin to Type III analysis of variance tests, or added-last tests. We adjusted P-values from these tests for multiple comparisons using the false discovery rate approach.

We then applied recursive partitioning only to the significant traits defined by the above regressions. The procedure associated with constructing classification trees defines nodes (and cut-points) according to values that minimize the impurity (or maximize the homogeneity) of groups at each split. We defined impurity for the classification trees fit to this data according to proportions of observations in each of the categories defined by the splits. We constructed full trees (with a maximum number of splits), and then pruned, or simplified them, using a cross-validation procedure to determine the most parsimonious tree that minimized deviance. The cross-validation procedure involved randomly selecting 10% of observations to remove from the sample, and then calculating deviance from this subsample. We repeated this process 10 times. We selected the smallest tree whose deviance was close to the minimum as the most parsimonious (i.e., best fit) classification tree (Venables and Ripley 2002). We calculated misclassification rates as the overall proportion of observations used in the tree-fitting process that were misclassified by the final classification tree. We constructed 95% confidence intervals for misclassification rates using the binomial exact method.

We calculated age at eruption (in days) for each tooth as the age of the pup on the date in which a tooth was first visibly observed minus one-half of the time interval between that date and the previous examination date. This correction factor accounted for the unknown period of time that the tooth was emerged since the previous examination (Kuykendall et al. 1992). We calculated age at full eruption for each tooth in a similar manner because the correction factor equaled one-half of the time interval between the examination in which a tooth first appeared fully erupted and the previous examination. We calculated age at eruption and age at full eruption only when a tooth was initially observed as either first erupting or partially erupted. We developed this criterion to eliminate cases in which duration of time between examinations was too long for reliable estimation of the variable. Neither median age at eruption nor age at full eruption of deciduous and permanent teeth differed between male and female pups; therefore, we pooled the data.

Using the alternative skull-aging method, the squamosal–jugal suture at the ventral glenoid fossa closed after the basioccipital–basisphenoid suture. Closure of the squamosal–jugal suture has been used to distinguish between juvenile and adult mink Mustela vison and identify raccoons Procyon lotor older than 4.5 y of age (Greer 1957; Fiero and Verts 1986). Therefore, the male subadult category was expanded to include animals that had this suture open or obscured (Table 2). None of the additional skull ossification or development characteristics investigated in the alternate skull-aging method were useful for further differentiating adults from aged adults for males or females.

Age class based on skull ossification and development was significantly correlated with cementum annuli using the alternative skull-aging method (F_{1, 36} = 21.12, P < 0.001), but not the established skull-aging method (F_{1, 36} = 1.62, P = 0.212). When aged adults were combined or separated from adults, tooth condition, standard total length, thymus presence and size, grizzle rating, canine width at the alveolus, zygomatic width (ZW), ZW as a proportion of condylobasal length (CBL), basal length, basal length as a proportion of CBL, interorbital width, postorbital width (PW), and PW as a proportion of CBL were identified by regression analyses as significant for predicting age class. Standard total length was the only character that varied significantly with gender.

When aged adults and adults were combined, results from the recursive partitioning process suggested that postorbital width (PW), proportional basal length (BL/CBL), and proportional zygomatic width (ZW/CBL) were the most accurate way to differentiate between subadults and adults–aged adults (misclassification rate = 0.057, 95% CI = 0.007–0.192; pseudo-R² = 0.835). Cut-offs for these measures were 43.77 mm, 0.90 mm, and 0.78 mm for PW, BL/CBL, and ZW/CBL, respectively, where values lower than the cut-offs corresponded to classification as subadults, and values above or equal to the cut-offs corresponded to classification as adults (Figure 3A). Although classification via PW, BL/CBL, and ZW/CBL was the most accurate, classification via tooth condition score was the simplest way to differentiate between subadults and adults–aged adults (misclassification rate = 0.098, 95% CI = 0.027–0.231; pseudo-R² = 0.535). Standard total length was also a moderately useful predictor of age class (misclassification rate = 0.167, 95% CI = 0.021–0.481; pseudo-R² = 0.341; Figure 3B). Specifically, individuals with a tooth condition score <2.38 should be classified as a subadult, while those with a score ≥2.38 should be classified as an adult–aged adult (Figure 3C; Data S1, Supplemental Material). Small sample sizes for reproductive activity data precluded statistical analysis; however, females with evidence of reproductive activity (e.g., mammary gland development, placental bands or scars, a fetus, and/or uterine development) would be classified as adults by definition.

Misclassification rates of classification schemes derived from recursive partitioning increased when aged adults were separated from adults. Results from these analyses suggest that ZW/CBL, standard total length, and tooth condition were the most useful classification variables for age class. Cut-off values and misclassification rates associated with each of these variables are provided in Figure 4. Across these variables, however, adults and aged adults could not be differentiated because of the high degree of overlap in these measures between the two age classes. All data are available through the U.S. Geological Survey's ScienceBase catalog (doi:10.5066/F72J69B7; Data S2, Text S1, Supplemental Material).

Tooth eruption data from southern sea otters were summarized with median (x̃) and range values (Table 1). The teeth present at day zero, also called milk teeth, included i³, c, and p1. The i¹, i₁, and i², but not i₂, were occasionally present above the gingiva (as also observed in 4 of 7 northern sea otter pup skull specimens), but in all cases, these teeth were extremely narrow, small, and loosely attached. The remaining deciduous teeth (p2 and p3) emerged sequentially starting with p₂ at 3–8 (x̃= 6) d, then p² at 14–33 (x̃ = 17) d. The third premolars followed a similar pattern with p₃ emerging first at 14–28 (x̃ = 23) d and p³ at 26–50 (x̃ = 35) d. The p³ emerged at approximately the same time as the first permanent tooth (I¹; x̃ = 34 d). The deciduous teeth were all fully erupted at 42–80 (x̃ = 65) d, which was similar to when I¹, I², and I₁ were fully erupted (x̃ = 61 d). The remaining permanent incisors (I³ and I₂) emerged (x̃ = 55 and 56 d, respectively) and were fully erupted (x̃ = 70 d) during a similar time frame. The first premolars and canines displayed a similar pattern. P¹ and P₁ emerged at x̃ = 53 and x̃ = 56 d, respectively, and became fully erupted at x̃ = 76 and x̃ = 77 d, respectively. Maxillary and mandibular canine teeth emerged at x̃ = 105 and x̃ = 108 d, respectively, and became fully erupted at x̃ = 125 and x̃ = 127 d, respectively. The remaining permanent teeth emerged and became fully erupted in the following pattern: M₁ → M¹ → M₂ → P₂ → P² → P³ → P₃. A full set of permanent teeth were fully erupted at 210–242 (x̃ = 224) d.

Based on tooth eruption pattern, tooth condition score, and evidence of reproductive activity (females only), individuals were placed into the pup, juvenile, subadult, or adult–aged adult categories (Figure 5). Individuals with deciduous teeth present were classified as either a pup or juvenile. If deciduous canine teeth were present or permanent canine teeth were first erupting, the individual was classified as a pup. The animal was classified as a juvenile when the permanent canine teeth were partially to fully erupted, which occurred at 4–5 mo of age, coinciding with the earliest possible weaning age for sea otters (Siniff and Ralls 1991). Tooth condition score and evidence of reproductive activity were then used to classify individuals with a full set of permanent teeth as either subadult (<2.38 and no evidence of reproductive activity if female) or adult–aged adult (≥2.38 and/or evidence of reproductive activity if female). Using these criteria, the accuracy estimate between cementum annuli and observer-assigned age class increased from 0.73 to 0.91, with discrepancies in only five individuals. These discrepancies included four individuals at the age cut-off between age classes (e.g., 3-y-old female or 6-y-old male) and a 4-y-old male classified as an adult due to a tooth condition score of 3.00.

The primary finding of this investigation was a significant correlation between cementum annuli and skull ossification pattern, suggesting that both of these methods are valid for aging sea otters from Washington State. We also identified ossification of the squamosal–jugal suture at the ventral glenoid fossa as useful to differentiate male subadults from adults in comparison with the basioccipital–basisphenoid suture, which is more accurate to distinguish female subadults from adults because they reach sexual maturity at a younger age. Based on the time and resources required to clean skull specimens, cementum annuli remains the most readily available and cost-effective method for aging sea otters from Washington. The use of cementum annuli to predict age does have limitations, however, due to environmental influences such as diet, seasonality, nutritional status, disease, and reproduction (Grue and Jensen 1979; Bodkin et al. 1997; Costello et al. 2004; Von Biela et al. 2007; Medill et al. 2009). Furthermore, these influences are not static, but may change as an individual matures, population size or resource availability changes, or stochastic events occur. Thus, other characters, such as skull ossification, that predict age may be useful for assessing the accuracy of the cementum annuli aging method in sea otters from Washington in the future.

Another major finding of this investigation was an additional prediction method that could be used in the field to age sea otters from Washington. Specifically, tooth condition score, evidence of reproductive activity in females, and tooth eruption pattern were identified as criteria for placing sea otters from Washington into four age classes: pups, juveniles, subadults, and adults–aged adults. Using these three criteria improved the accuracy estimate between cementum annuli and observer-assigned age class and can be used to assign age class when cementum annuli results are not available. Additionally, using these three criteria improved the upper accuracy estimate proposed by Bodkin et al. (1997) when using cementum annuli results alone (0.91 vs. 0.85). Though a small number of individuals were misclassified using these criteria, the majority of misclassifications included individuals at the age threshold between age classes. Thus, we have identified and validated three aging methods for sea otters from Washington: cementum annuli, skull ossification, and a decision tree based on teeth and reproduction.

Although significant tooth wear has been suggested as an indicator of old age in sea otters (Riedman and Estes 1990), this is the first investigation in sea otters to rigorously test and identify tooth wear as a predictor of age. Tooth wear has been investigated as a predictor of age in many other species, but with mixed success (Smuts et al. 1978; Bowen 1982; Fiero and Verts 1986; Cowan and White 1989; Harris et al. 1992; Stander 1997; Gipson et al. 2000). A limitation is that tooth wear pattern is dependent on prey selection, which is associated with population size, dispersal, and resource availability (Estes et al. 1982; Laidre and Jameson 2006; Tinker et al. 2008), as well as dietary preference (Estes et al. 2005) and intraspecific aggression among sexually mature males (M. Murray, Monterey Bay Aquarium, personal communication). For example, subadult males tend to form groups at the periphery of the population range and are more likely to disperse into unoccupied habitats (Lance et al. 2004), which are dominated by different prey species than are found in the center of the range or densely occupied habitats. Crabs, urchins, and fish may be “softer” prey items that cause less tooth wear than clams, which are often opened using the teeth (Kenyon 1969). In fact, male sea otters in Washington consume more crabs and less clams compared with females (Walker et al. 2008); and in this study, we found that males are older when they develop marked tooth wear compared with females. Thus, tooth condition scoring may not only be different among sea otters from varying geographic locations, but also change over time within a given geographic location, illustrating the importance of having multiple aging methods available for each geographic location.

In the original tooth-condition system for sea otters (Pattison et al. 1997), each score was based on a subjective description of tooth wear and all of the teeth were evaluated as a whole. Though Pattison's method provided an excellent foundation for evaluating tooth wear, many lesions and lesion severity were absent from the descriptions, and thus the system was biased to poorer tooth condition scores. Thus, for animals with an unusual tooth-wear pattern, such as those from San Nicolas Island, in which the incisors were severely damaged, but the remaining teeth were in relatively good condition (Bentall 2005), subadults would likely be assigned a tooth condition score of 4.00, even though the premolars and molars may be in excellent condition. The tooth-condition scoring system developed in this investigation expands the description for each score with additional lesions and lesion severity. Furthermore, overall tooth-condition score was determined by evaluating each tooth type separately (e.g., incisors separately from canines). Thus, a subadult animal with cracked and broken incisors but no to few lesions in the remaining teeth would be assigned a score that is consistent with a subadult.

The standardized dental chart that was developed for evaluating tooth wear can also be used for charting tooth eruption (Figure S1, Supplemental Material) and further adapted for oral health assessments (e.g., in field-captured or captively housed sea otters) that could include evaluation of the gingiva, periodontal health, and regional lymph nodes. Tooth eruption pattern was described by Kenyon (1969), Lensink (1962), and Morejohn et al. (1975), but the previous investigations did not use known-age animals and were based on an unknown sample size. The longitudinal tooth-eruption data in this investigation allowed us to distinguish pups from juveniles, as well as juveniles from older age classes. Using tooth eruption as a guide to aging younger sea otters is necessary because both age classes can have a cementum annuli result of zero (i.e., low specificity). In addition to tooth eruption, dependence on the adult female, as evidenced by concurrent capture or exchanged vocalizations with an adult female, could also be assessed in the field. Differentiating between these younger age classes is important for understanding demography because pups are dependent on the adult female, whereas juveniles are independent but often exhibit a highly variable, and frequently high, mortality rate compared with older age classes (Kenyon 1969; Monson et al. 2000, 2011; Ballachey et al. 2003; U.S. Fish and Wildlife Service 2013).

Juveniles were differentiated from pups by presence of partially to fully erupted permanent canine teeth, which occurred at 4–5 mo of age, coinciding with the earliest possible weaning age for sea otters (Siniff and Ralls 1991). These findings are similar to those of Lensink (1962), indicating that the deciduous canines were lost and replaced around 4–5 mo of age. In this investigation, a complete set of permanent teeth were fully erupted at 8.5–9 mo of age. Lensink (1962) found that the permanent teeth were erupted around 16 mo of age, whereas Kenyon (1969) stated that this occurred around 12 mo of age. We considered that the age range between 8.5 and 16 mo of age was a reasonable approximation of the timing of juvenile transition to the subadult stage (at 1 y old).

With the small sample size available for aged adults (males: n = 5; females: n = 6) as determined by cementum annuli, we did not identify criteria that reliably predicted this age category. Baculum measurements, such as length, weight, and volume, have been shown to correlate with age in southern sea otters and mink (Elder 1951; Morejohn et al. 1975). Baculum measurements have only recently become part of the standard sea otter necropsy protocol at the National Wildlife Health Center, and thus small sample size precluded analysis. As more sea otters from Washington become available for necropsy, skull morphometric measurements, baculum measurements, and tooth-condition score data can be routinely collected to increase the sample size and determine criteria that predict the aged adult category.

We made the assumption that carcass collection was not biased. However, up to 90% of the sea otter carcasses were from areas with easier access and higher visitation rates, and <25% of reported strandings were provided for necropsy on account of decomposition (D. Lynch, U.S. Fish and Wildlife Service, personal communication). It is therefore possible that the specimens examined here represent a limited cross-section of sea otters from Washington, but necropsy findings from these same specimens have been used to inform disease occurrence, causes of mortality, and toxicology (Lance et al. 2004; Kannan et al. 2008), and limited access to specimens is a common shortcoming when studying a protected, nongame species.

In summary, enumeration of cementum annuli remains the preferred method for aging sea otters from Washington and also represents the most precise method because it provides an age in years vs. a broader age class. For animals <1 y of age, however, tooth eruption pattern is the most precise method of age determination. Finally, tooth wear and suture closure patterns can be used to differentiate between subadults and older animals. These latter techniques could be extended for use in sea otters from other geographic locations, not only to verify the accuracy of cementum annuli, but also in the field or when cementum annuli results are not available. These techniques offer a replicable, cost-conscious methodology to gather essential biological information from sea otters, which should be especially useful for citizen-science programs that are often utilized for beach monitoring and stranding networks.

Please note: The Journal of Fish and Wildlife Management is not responsible for the content or functionality of any supplemental material. Queries should be directed to the corresponding author for the article.

Data S1. Data for Figures 3 and 4 from an investigation to refine aging criteria and methods for northern sea otters collected between 1989 and 2011 in Washington, Enhydra lutris kenyoni. Age classification as assigned by pathologist at necropsy, total length from nose to tip of tail (cm), tooth condition scoring based on the new classification criteria given in this study, and zygomatic width to the condylobasal length of the skull. Tooth condition score 1 (excellent) was assigned to teeth with no signs of wear; canines were entire and pointed. Tooth condition score 2 (good) was assigned to teeth with light wear; canines may be slightly rounded and the postcanines (premolars and molars) may have mild flattening of the cusps. Tooth condition score 3 (fair) was assigned to teeth with moderate wear; canines are rounded or blunted, obvious cusp flattening is present in the postcanines, pinpoint pitting may be present in the postcanines, and some teeth (<2) may be broken or missing. Tooth condition score 4 (poor) was assigned to teeth with heavy wear; teeth are worn nearly to the gingiva, pitting and/or caries are larger than pinpoints, and broken teeth are common. Tooth condition scores 5 or 6 were assigned to specimens with all or some milk teeth present, respectively.

Found at DOI: http://dx.doi.org/10.3996//052017-JFWM-040.S1 (13 KB XLSX); also available at the U.S. Geological Survey's ScienceBase catalog (doi: 10.5066/F72J69B7).

Data S2. Complete data set from an investigation to refine aging criteria and methods for sea otters collected between 1989 and 2011 in Washington, Enhydra lutris kenyoni, provided for review. Final data are archived in U.S. Geological Survey's ScienceBase catalog (doi: 10.5066/F72J69B7). See Text S1. for field explanations.

Found at DOI: http://dx.doi.org/10.3996//052017-JFWM-040.S2 (64 KB XLSX).

Text S1. README.TXT. Field headings and explanations for “An investigation to refine aging criteria for northern sea otters collected between 1989 and 2011 in Washington State.”

Found at DOI: http://dx.doi.org/10.3996//052017-JFWM-040.S3 (15 KB DOCX).

Table S1. List of measured skull specimens, Enhydra lutris kenyoni, (n = 50) collected between 1989 and 2011 by reference or case number used in an investigation to refine aging criteria and methods for sea otters in Washington. At the time of manuscript submission, the specimens were housed at the University of Wisconsin Zoological Museum (Madison, Wisconsin), but they will be permanently vouchered at the Burke Museum of Natural History and Culture (Seattle, Washington). FWS: U.S. Fish and Wildlife Service; National Wildlife Health Center: U.S. Geological Survey - National Wildlife Health Center. ♂ = male, ♀ = female.

Found at DOI: http://dx.doi.org/10.3996//052017-JFWM-040.S4 (12 KB DOCX).

Figure S1. Sea otter, Enhydra lutris, standardized dental chart, developed as part of an investigation to refine aging criteria and methods for sea otters in Washington state, with occlusal and buccal views for scoring tooth-wear pattern and severity in skull specimens collected between 1989 and 2011. Tooth condition score 1 (excellent) is assigned to teeth with no signs of wear; canines are entire and pointed. Tooth condition score 2 (good) is assigned to teeth with light wear; canines may be slightly rounded and the postcanines (premolars and molars) may have mild flattening of the cusps. Tooth condition score 3 (fair) is assigned to teeth with moderate wear; canines are rounded or blunted, obvious cusp flattening is present in the postcanines, pinpoint pitting may be present in the postcanines, and some teeth (<2) may be broken or missing. Tooth condition score 4 (poor) is assigned to teeth with heavy wear; teeth are worn nearly to the gingiva, pitting and/or caries are larger than pinpoints, and broken teeth are common. Tooth condition scores 5 or 6 are assigned to specimens with all or some milk teeth present, respectively. The incisors (# 101–103, 201–203, 301–302, and 401–402), canines (# 104, 204, 303, and 403), premolars (# 105–107, 205–207, 304–306, and 404–406), and molars (# 108, 208, 307–308, and 407–408) are scored as separate groups. These four scores per animal are then averaged to obtain an overall tooth condition score ranging between 1 and 4.

Found at DOI: http://dx.doi.org/10.3996//052017-JFWM-040.S5 (1563 KB PDF).

Reference S1. Riedman ML, Estes JA. 1990. The sea otter (Enhydra lutris): behavior, ecology, and natural history. U.S. Department of the Interior, Fish and Wildlife Service, Biological Report 90.

Found at DOI: http://dx.doi.org/10.3996//052017-JFWM-040.S6; also available at https://www.fort.usgs.gov/sites/default/files/products/publications/2183/2183.pdf (16583 KB PDF).

Reference S2. U.S. Fish and Wildlife Service. 2013. Southwest Alaska distinct population segment of the northern sea otter (Enhydra lutris kenyoni) – Recovery plan. U.S. Fish and Wildlife Service, Region 7, Alaska.

Found at DOI: http://dx.doi.org/10.3996//052017-JFWM-040.S7; also available at https://www.fws.gov/alaska/fisheries/mmm/seaotters/pdf/Recovery%20Plan%20SW%20AK%20DPS%20Sea%20Otter%20Aug13.pdf (2519 KB PDF).

We thank D. Lynch, U.S. Fish and Wildlife Service, and Washington Department of Fish and Game, and the volunteers who assisted with collection and submission of sea otters to the National Wildlife Health Center. The University of Wisconsin-Madison provided postdoctoral funding to BBB. We also thank L. Jones, G.R. VanBlaricom, B.E. Ballachey, D.H. Monson, and J.L. Bodkin for substantial improvements to this manuscript, as well as the Associate Editor of the journal for thorough review and feedback.

Use of trade, produce, website, or firm names is for descriptive purposes only and does not imply endorsement by the U.S. Government.