INVESTIGATING ASSOCIATIONS AMONG RELATEDNESS, GENETIC DIVERSITY, AND CAUSES OF MORTALITY IN SOUTHERN SEA OTTERS (ENHYDRA LUTRIS NEREIS)

In some species, low genetic diversity and high relatedness of individuals are linked with decreased fitness and increased disease susceptibility (Allendorf et al. 2013; Tompkins et al. 2015). In addition, certain populations lacking genetic variation due to inbreeding or genetic bottlenecks (homogenous populations) suffer greater consequences from pathogen exposure when compared with more genetically diverse populations (Sherman et al. 1988; Altizer et al. 2003; King and Lively 2012). Causes of morbidity and mortality often differ in relationships with host genetic diversity. For example, stranded California sea lions (Zalophus californianus) with carcinoma or helminth infections had higher internal relatedness (lower individual genetic diversity) than those impacted by algal toxins or trauma (Acevedo-Whitehouse et al. 2003), although later work involving more samples suggested this may have been affected by confounding factors (Gulland et al. 2020). Our study assessed the associations between common causes of death and host genetic (familial) relatedness and genetic diversity in southern sea otters (Enhydra lutris nereis), a threatened marine mammal with low genetic diversity.

Sea otters are an important predator and keystone species inhabiting nearshore marine waters of the Pacific Rim (Estes and Palmisano 1974; Bodkin 2015). Sea otters were hunted to near extirpation throughout their range during the fur trade in the mid to late 1800s, leaving 13 small (10–100 individuals), widely scattered remnant populations. This bottleneck and isolation among surviving remnant populations contributed to reduction of genetic diversity (Kenyon 1969; Lidicker and McCollum 1997; Bodkin et al. 1999; Larson et al. 2002; Gagne et al. 2018). The southern sea otter, one of three subspecies, was reduced to an estimated 50 individuals during that time; it currently persists along the central coast of California (Fig. 1; Leatherwood et al. 1978; Wilson et al. 1991; Jessup et al. 2007; Hatfield et al. 2019). Several studies have posited that reductions in genetic diversity prior to the fur trade may have also impacted the southern sea otter population (Aguilar et al. 2008; Larson et al. 2012; Beichman et al. 2019). Sea otter protection and management began with the North Pacific Fur Seal Convention of 1911 (Protocols of the International Fur Seal Conference 1911), and while sea otter populations have since increased (Hatfield et al. 2019), genetic diversity in southern sea otters remains lower than the estimated prefur trade levels (Larson et al. 2012; Gagne et al. 2018). This low genetic variation could leave them vulnerable to stochastic events and increased disease susceptibility (Wildt et al. 1987; Crnokrak and Roff 1999; O'Grady et al. 2006; Hedrick and Garcia-Dorado 2016). Despite federal protection, southern sea otters suffer from many causes of morbidity and mortality that may threaten population recovery, including cardiomyopathy, acanthocephalan (Profilicollis spp.) peritonitis, systemic protozoal infections with Toxoplasma gondii and Sarcocystis neurona, domoic acid intoxication, end-lactation syndrome, and white shark (Carcharodon carcharias) bite (Thomas and Cole 1996; Kreuder et al. 2003; Chinn et al. 2016; Tinker et al. 2016; Miller et al. 2020).

Cardiomyopathy, a highly prevalent cause of death for southern sea otters (Kreuder et al. 2003; Miller et al. 2020), is of particular interest because genetic variants have been linked to some forms of cardiac disease in other species (e.g., Hershberger et al. 2013; Kittleson et al. 2015). Factors such as age, location of stranding, exposure to domoic acid, and infection with T. gondii or S. neurona are known to influence risk of sea otter death from cardiomyopathy (Miller et al. 2020; Moriarty et al. 2021); we investigated if genetic diversity or familial relatedness may also contribute to this risk. In humans, point mutations in over 30 genes have been found to contribute to dilated cardiomyopathy (Hershberger et al. 2013). Cardiomyopathy also has a heritable basis in several breeds of domestic dogs (Canis lupus familiaris) and cats (Felis catus), including Great Danes, Doberman pinschers, Maine coon cats, and Norwegian Forest cats (Kittleson et al. 1999; Meurs et al. 2001, 2007; März et al. 2015). Additionally, the high energy demand of raising a pup (Thometz et al. 2014) can lead to extreme emaciation and death in female southern sea otters during the late stages of pup care or after the pup is weaned, a pattern described as “end-lactation syndrome” (Chinn et al. 2016). Nutritionally stressed females, therefore, experience a trade-off between ensuring their own survival probability and investment in pup survival (Monson et al. 2000; Chinn et al. 2016). End-lactation syndrome has been linked to population density and resource availability (Chinn et al. 2016; Miller et al. 2020); we investigated if it might also be associated with genetic diversity and familial relatedness.

Additionally, diet specialization has been shown to put sea otters at risk for parasitic infections (Johnson et al. 2009). Sea otters become infected with acanthocephalans (Profilicollis spp.) through consumption of intermediate hosts: sand crabs (Emerita analoga) and spiny mole crabs (Blepharipoda occidentalis; Mayer et al. 2003). Consumption of marine snails has also been shown to pose a higher risk for T. gondii infection (Johnson et al. 2009), while a diet with a high proportion of clams can put southern sea otters at a higher risk for S. neurona infection (Burgess et al. 2020). The diets of individual southern sea otters, specifically those high in crabs and clams, has been potentially linked to genetic relatedness (Ralls et al. 2017). Our study investigated a direct link between parasitic causes of death and familial relatedness.

Shark bite and domoic acid intoxication are causes of death with demonstrated associations with geography. For example, domoic acid poisoning, resulting in an unusual mortality event of sea otters in 2003, was associated with a bloom of toxic marine diatoms due to land-sea runoff (Jessup et al. 2007). Shark bite is also linked to geography; the frequency of southern sea otters impacted by shark bite has greatly increased in the southern part of the range from 1990 to 2013 (Tinker et al. 2016; Miller et al. 2020). Although strong genetic structure has not been reported in southern sea otters, there is evidence for genetic isolation by distance in this population (Gagne et al. 2018), and individual sea otters have relatively small lifetime home ranges (Tarjan and Tinker 2016; Breed et al 2017). Therefore, we investigated these geographically related causes of death for potential associations with genetic diversity and familial relatedness.

Genetic diversity (e.g., heterozygosity) is an important population-level indicator and was used to investigate associations with disease outcome, because greater diversity has been shown to be correlated with higher population fitness (Reed and Frankham 2003). Microsatellite markers can also be used to assess individual levels of genetic diversity (Aparicio et al. 2006). In this study, we assessed both averaged and individual levels of genetic diversity in sea otters by using microsatellite DNA markers. In addition, measuring genetic relatedness has diverse applications (Weir et al. 2006); here, we used pairwise genetic relatedness to investigate whether common causes of southern sea otter death had a familial basis (i.e., individuals that share the same cause of mortality are more related than expected by chance).

Because a familial basis has been demonstrated for cardiomyopathy in other animals, we hypothesized that southern sea otters with fatal cardiomyopathy would be more genetically related than expected by chance. We predicted that death due to acanthocephalan (Profilicollis spp.) and S. neurona infection might also be associated with familial relatedness because specialization on the prey types associated with these parasites may be associated with higher relatedness in southern sea otters (Ralls et al. 2017). Based on results from previous studies, we did not expect an association with genetic diversity or familial relatedness for toxoplasmosis, domoic acid intoxication, end-lactation syndrome, or shark bite, but we investigated these causes of death because they are common in southern sea otters.

Veterinary pathologists, biologists, and diagnosticians compiled demographic and cause of death data for 560 southern sea otters, hereafter referred to as “otters,” necropsied from 1998 to 2012 through a multiagency collaboration among the California Department of Fish and Wildlife, University of California Davis, US Geological Survey, Monterey Bay Aquarium, and the Marine Mammal Center (Miller et al. 2020). Based on findings from gross necropsy, histopathology, and diagnostic testing, each otter was assigned a primary cause of death and up to three contributing causes of death. The primary cause of death was the etiology that was most severe and probably directly caused the animal to die. Contributing causes of death were those causing moderate to severe damage and were determined to independently contribute to mortality or would have caused mortality if the primary cause had not intervened (e.g., severe acanthocephalan infection in an otter that died from shark bite; Miller et al 2020). Demographic information collected included sex, age, and the location of stranding. Otters were divided into three age classes: subadult (1–3 yr old), adult (4–10 yr), and aged adult (>10 yr) based on pelage characteristics, tooth wear, and characteristics of the gonads and reproductive tract. Adult and aged adult otters were pooled and are referred to as “adults” to distinguish them from subadults that were excluded from some analyses due to age being a risk factor. Each otter's stranding location was recorded as the nearest “as the otter swims” point, which are established geographic locations spaced every 500 m along the California coast (Kreuder et al. 2003). We focused on seven common causes of southern sea otter death: cardiomyopathy, acanthocephalan (Profilicollis spp.) peritonitis, toxoplasmosis, sarcocystosis, probable domoic acid intoxication, end-lactation syndrome, and shark bite (the Supplementary Material and Miller et al. 2020 describe how each cause of death was determined).

More than 1,000 (n=1,006) southern sea otters have been genotyped by using a 38 microsatellite panel (Gagne et al. 2018); 356 otters with both genotype and necropsy data were included in this study. Retained loci had a minimum of two alleles; one monomorphic microsatellite locus (Mvis 072) was dropped from analyses (see Supplementary Material Table S1 for retained loci information; see Supplementary Materials for genotyping methods). Figure 1 displays the locations of the otters analyzed in this study.

Necropsied otters were grouped by cause of death. A necropsied otter was considered a case for a certain condition if that condition was listed as one of the primary or contributing causes of death. A necropsied otter was considered a control for a given condition if this condition was not noted as one of the primary or contributing causes of death. Table 1 provides a complete description of how otters were characterized. Each cause of death was analyzed separately. To determine if pairwise relatedness was significantly different between case and control otters, we used the R package related (Pew et al. 2015), which calculates the average observed pairwise relatedness within each group. We used Queller and Goodnight's (1989) estimator of relatedness because it was found to best estimate relatedness among confirmed sea otter mother-pup pairs in a previous study by using the same microsatellite markers (Ralls et al. 2017). To create a null expectation, case and control individuals were randomly shuffled between the two groups 100,000 times, and pairwise relatedness was calculated within each randomized group for every iteration. The P value represents the proportion of iterations in which the null expectation was equal to or greater than the observed relatedness value (Pew et al. 2015). A P value of less than or equal to 0.05 was considered significant. All related analyses were run by using University of Wyoming's Teton Computing Environment, Intel x86_64 cluster (Advanced Research Computing Center 2018). For cardiomyopathy and end-lactation syndrome, we only analyzed adult and aged adult animals, because most animals died from these conditions as adults (Kreuder et al. 2005; Chinn et al. 2016; Miller et al. 2020). Both sexes were analyzed separately and together, except end-lactation syndrome, which only affects sexually mature females (Chinn et al. 2016). This was to account for males and females having different dispersal patterns that could potentially confound our analysis: male sea otters generally inhabit areas that are separate from females and their pups (Kenyon 1969; Jameson 1989), and males tend to travel further distances over time compared with females (Ralls et al. 1996).

Genetic diversity was assessed to determine if significant differences existed between case and control otters for each common cause of death (Table 1). GenAlEx (Peakall and Smouse 2006, 2012) was used to calculate expected heterozygosity (Nei 1987) in each group. Allelic richness corrected for sample size (El Mousadik and Petit 1996) was calculated by using the PopGenReport package (Adamack and Gruber 2014) in Program R (R Core Team 2019). We used a t-test to determine if these metrics were significantly different between groups with the stats package in Program R (R Core Team 2019).

To determine if individual genetic diversity was a significant predictor of each cause of death, we completed multivariate logistic regression analysis for each cause of mortality (1, disease of interest; 0, other cause), while controlling for other variables reported as important risk factors in previous studies: sex, age class, geographic region of stranding, stranding quarter, and density index (K index; Mayer et al. 2003; Kreuder et al. 2005; Johnson et al. 2009; Chinn et al. 2016; Tinker et al. 2016; Miller et al. 2020). Sex (male, female) and age class (subadult, adult, aged adult) were categoric variables. The geographic region of stranding was divided into three categories (north, central, south), as described in Tinker et al. (2016) and Miller et al. (2020). Season was defined as the following: spring, February–April; summer, May–July; fall, August–October; and winter, November–January, as in Miller et al. (2020). In the model for end-lactation syndrome, season was defined differently: spring-summer, April–July; and fall-winter, August–March, to account for reproduction timing and peaks in end-lactation syndrome (Miller et al. 2020). The K index was defined as how close the otter population is to carrying capacity at each enrolled otter's stranding location, with values from 0 (not close to carrying capacity) to 1 (at carrying capacity; Miller et al. 2020).

We calculated individual metrics of genetic diversity for every otter in the study by using GENHET (Coulon 2010). Because all metrics were highly correlated with each other (abs(r)>0.96), we used homozygosity by loci as our estimate of individual genetic diversity, because it weighs loci by allelic diversity and can outperform internal relatedness (Aparicio et al. 2006). Homozygosity by loci ranges from 0 to 1, where lower values indicate the individual is more genetically diverse and higher values indicate less genetic diversity. Logistic regression models were run by using the glm function in the R stats package (R Core Team 2019) with and without homozygosity by loci to see if this variable significantly improved model fit (i.e., lowered the Akaike information criterion (AIC) score by two or more points; Akaike 1974). The area under the curve (AUC) was calculated by using the pROC package in R (Robin et al. 2011). We calculated 95% confidence intervals (CI) by using the confint function in the stats package in R by the Wald test method (R Core Team 2019).

Significant associations with familial relatedness or individual genetic diversity (P<0.05) were observed for cardiomyopathy, acanthocephalan peritonitis, and sarcocystosis. We found these associations to vary by sex.

Adult male cardiomyopathy cases (n=86) were more related than the null expectation (P<0.049). Conversely, adult female cardiomyopathy controls (n=23) trended toward being more related than the null expectation (P<0.056; Fig. 2).

Relatedness for female acanthocephalan peritonitis cases (n=37) fell within the null distribution, but female acanthocephalan peritonitis controls (n=110) were significantly more related than the null expectation (P<0.004; Fig. 3). In the logistic regression model, including homozygosity by loci significantly improved model fit (AIC: 353.8, AUC: 0.744) compared with models without this variable (AIC: 356.6, AUC: 0.737). In the best fit model, increased homozygosity by loci (higher homozygosity, lower individual genetic diversity) was associated with lower odds of dying from acanthocephalan peritonitis (odds ratio [OR], 0.03; 95% CI, 0.001–0.70; see Supplementary Material Table S2).

Male sarcocystosis controls (n=158) were significantly more related than the null expectation (P<0.011), while male sarcocystosis cases (n=26) were not significantly more related than the null expectation (Fig. 4). The logistic regression model that included homozygosity by loci as a risk factor for fatal sarcocystosis (see Supplementary Material Table S2) was a slightly better fit to the data (AIC: 194.2, AUC: 0.829) than the model that did not include this metric as a risk factor (AIC: 195.3, AUC: 0.822). Including homozygosity by loci in our model showed a tendency (P=0.077) that southern sea otters with lower individual genetic diversity were at higher risk for sarcocystosis-mediated mortality (OR, 70.16; 95% CI, 0.66–8720.36; see Supplementary Material Table S2).

For all other causes of death and combinations of sexes (females only, males only, females and males together) pairwise relatedness fell within null expectations (see Table 2 and Supplementary Material Figs. S1–S4).

With the possible exception of sarcocystosis, homozygosity by loci was not a significant predictor in multivariate logistic regressions, except for acanthocephalan peritonitis. Genetic diversity was not significantly different between averages for case and control groups for any of the seven causes of death (see Supplementary Material Table S3).

We hypothesized, based on previous studies, that we would find an association with familial relatedness or genetic diversity for cardiomyopathy, acanthocephalan peritonitis, and sarcocystosis-mediated mortality. These were the only three causes of mortality for which we observed significant associations in our study, yet some of the directions of associations were opposite to the direction expected.

One previous study of southern sea otters reported that females were more likely to have cardiomyopathy as a primary cause of death (Kreuder et al. 2003), although this bivariate association did not account for the effects of additional risk factors, and the small sample may not be representative. Later multivariate models with larger sample populations concluded that sex was not a significant risk factor for fatal cardiomyopathy in southern sea otters (Kreuder et al. 2005; Miller et al. 2020). Our multivariate logistic regression also did not find sex to be a significant predictor for fatal cardiomyopathy; however, our pairwise relatedness results suggest that southern sea otters may be experiencing some degree of sex-specific genetic variation associated with fatal cardiomyopathy. Related female southern sea otters may carry a genetic factor that protects them from fatal cardiomyopathy, while related males may be genetically predisposed to cardiomyopathy as a primary or contributing cause of death. Possibly, specific gene variants associated with cardiac function found only in certain otter family lines are responsible for these sex-associated risk patterns.

Studies of cardiomyopathy in humans and nonhuman primates have documented sex-specific variations in gene expression (Yan et al. 2004; Haddad et al. 2008; Fairweather et al. 2013). Genes involved in metabolism and energy production are upregulated in cardiomyocytes of female humans with idiopathic dilated cardiomyopathy, compared with males (Haddad et al. 2008). These differences may contribute to increased survival of females with idiopathic dilated cardiomyopathy (Haddad et al. 2008). An X-linked mechanism for the development of dilated cardiomyopathy has also been reported in some human families (Berko and Smith 1987; Muntoni et al. 1993). Future research directions using whole-genome methods could examine whether specific genes play important roles to influence the risk of fatal cardiomyopathy in otters.

Our finding that female southern sea otter acanthocephalan peritonitis controls were more genetically related than the null expectation suggests the possibility of a genetic factor that helps protect females from mortality following enteric infection by Profilicollis spp. parasites. However, the distribution of Profilicollis spp. acanthocephalan parasites along the California coast is patchy, corresponding with areas of sandy habitat favored by the crustacean intermediate hosts (Mayer et al. 2003; Miller et al. 2020). Sampling was not uniform across the southern sea otter range, and this population may exhibit slight genetic isolation by distance (Gagne et al. 2018); therefore, alternatively, it is possible that related female southern sea otter controls lived in similar areas and were not exposed to geographic and diet-related risk factors for acanthocephalan peritonitis (Mayer et al. 2003; Miller et al. 2020). When considering other risk factors, our multivariate logistic regression revealed that higher individual genetic diversity increased risk of death from acanthocephalan peritonitis, which contradicted our initial hypothesis. The Simpson paradox, whereby unmeasured confounding variables lead to different results than if those variables were considered (Blyth 1972), could have contributed to these unexpected results. Perhaps, as one example, acanthocephalan peritonitis controls were animals that did not have acanthocephalan peritonitis as a primary or contributing cause of mortality, but still had Profilicollis spp. enteric parasitism, which could have blurred the distinction between our case and control groups. Given the complicated nature of the data, finer scale genomic markers may be needed to determine whether confounding influences better explain our results for acanthocephalan peritonitis.

Although female controls were more related for cardiomyopathy and acanthocephalan peritonitis, we found male controls were more related when investigating for S. neurona–mediated mortality in southern sea otters. Additionally, our logistic regression analysis showed that males were at higher odds of dying from sarcocystosis compared with females, although this was not significant (P=0.09). Miller et al. (2020), investigating different predictors, found males to be at significantly higher odds of mortality from sarcocystosis compared with females; that study had a larger sample size of 560 necropsied southern sea otters (including many of the same animals enrolled in the current study), which may have enabled significance to be reached. Studies of other protozoan parasites (e.g., T. gondii and Leishmania spp.) in other species have reported differences in susceptibility by sex (Kittas and Henry 1980; Alexander 1988; Mock and Nacy 1988; Roberts et al. 1995; Snider et al. 2009). Sex-associated differences in host immunity to S. neurona have not been well studied, but our data suggest that there may be a sex-driven difference in familial relatedness as associated with southern sea otter mortality from sarcocystosis that merits further investigation, especially because sex is a risk factor for this cause of mortality in this species (Miller et al. 2020). Additionally, given the high OR and low, though nonsignificant, P value for individual genetic diversity in our model for sarcocystosis, we believe genetic diversity may also play a role in influencing risk for this cause of death.

Our findings suggest that mortality from cardiomyopathy, acanthocephalan (Profilicollis spp.) peritonitis, and sarcocystosis may be influenced by host genetic diversity and familial relatedness. Our analyses have limitations due to the complex and nonmutually exclusive nature of the data set: each individual often had several contributing causes of death and were, therefore, considered as cases for more than one condition. In addition, many biologic, environmental, and demographic factors such as age, sex, diet, location and reproductive status have been shown to significantly influence the risk of southern sea otter death from common diseases (Mayer et al. 2003; Kreuder et al. 2005; Johnson et al. 2009; Chinn et al. 2016; Tinker et al. 2016; Miller et al. 2020; Moriarty et al. 2021). We incorporated as many of these factors as possible into our logistic regression modeling. However, we note there are likely other confounding factors that were unmeasured due to opportunistic stranded otter sample collection (Miller et al. 2020) and the dynamic nature of sea otter ecosystems (Estes and Palmisano 1974). We therefore present these initial research findings to inform future genomic studies.

Gaining an in-depth understanding of southern sea otter genetic assets and liabilities can provide essential information for wildlife managers. Future reintroductions have been suggested as a strategy to expand the southern sea otter range (Davis et al. 2019). Range expansion may eventually lead to mixing of the geographically separated subspecies, which may lead to changes in genetic diversity. Understanding the role that genetic diversity and familial relatedness play in disease outcome for southern sea otters may help managers to develop optimal conservation strategies.

Thank you to the many individuals involved in collecting necropsy samples, genetic samples, and data for this work, including F. Batac, L. Carswell, L. Dalbeck, E. Dodd, T. Drazenovich, J. Estes, T. Gilliland, K. Greenwald, M. Harris, B. Hatfield, L. Henkel, A. Reed, M. Tarjan, J. Tomoleoni, and C. Young. F. Gulland and the staff and volunteers at the Marine Mammal Center archived samples and collected and cared for animals. Thank you to the University of Wyoming Advanced Research Computing Cluster for use of the Teton Computing Environment for analyses. A. Buerkle, B. Shumaker, B. Hoar, J. Merkle, and K. Gustafson provided valuable help with statistics. M. Murphy, M. LaCava, L. Johnson, M. Dudenhoeffer, and anonymous reviewers provided insightful comments on this manuscript. This project was supported by the US Geological Survey, US Fish and Wildlife Service, California Department of Fish and Wildlife, Monterey Bay Aquarium, University of California Santa Cruz, University of California Davis, California Coastal Conservancy, and the University of Wyoming.

Supplementary material for this article is online at http://dx.doi.org/10.7589/JWD-D-21-00019.