Foraging Resource Partitioning in the California Sea Lion (Zalophus californianus) from the Southwestern Gulf of California

Abstract

California sea lion (CSL, Zalophus californianus) abundance has declined in different localities across this species’ Mexican distribution. However, Los Islotes rookery in the southwestern Gulf of California (GoC) deviates from this pattern. It is vital to gather ecological knowledge of this CSL settlement and its surroundings to better understand its population in the GoC. This study aimed to determine the foraging habits of different CSL sex and age classes. Sixty-five CSL samples were collected in Los Islotes and its surroundings for stable isotope analysis (δ¹³C and δ¹⁵N). The data were analyzed using a hierarchical Bayesian model, and isotopic areas were estimated using the SIBER package in R. Our findings evidenced resource partitioning. Adult females had lower δ¹⁵N values than most classes, reflecting the regional ¹⁵N-enrichment of the GoC. Conversely, subadult males showed low δ¹⁵N values, carrying foraging information from the ¹⁵N-depleted Pacific Ocean into the GoC. Adult males presented the highest δ¹⁵N values (after pups), suggesting a higher trophic position than adult females and values corresponding to the GoC. Moreover, juveniles had the most negative δ¹³C values and the largest isotopic areas, indicating offshore foraging habits and a mixed consumption of maternal milk and their first prey. Pups showed the highest mean δ¹⁵N value due to maternal milk consumption, reflecting the mother’s δ¹⁵N value and their enrichment. Our findings suggest that segregation is explained by unique life history traits and a possible strategy to avoid potential competition

1. Introduction

The California sea lion (CSL, Zalophus californianus) population in the Gulf of California declined by 65.1% from 1991 to 2019 due to environmental changes resulting from a sea surface temperature increase during the same period, consequently modifying CSL trophic dynamics [1]. Other CSL rookeries in the Mexican Pacific Ocean have declined in recent years or decades because of warm sea surface temperatures. For example, the CSL colony in Santa Margarita Island declined by 75% in the previous 36 years [2], while the San Benito Archipelago rookery declined by 50–60% following the 2015–2016 El Niño event [3].

Los Islotes, the southernmost reproductive colony in the Gulf of California, is exempt from this declining pattern across this species’ Mexican distribution. In contrast with other rookeries, Los Islotes has increased its abundance in recent years [1], reaching approximately 700 individuals (Elorriaga-Verplancken, unpublished data). One of the critical factors related to the success of this colony could be the oceanographic conditions that prevail around La Paz Bay and the southwestern Gulf of California, providing enough prey for these CSLs. Thus, they could adopt strategies to consume these resources, including probable intra-specific niche partitioning among different sex and age classes to avoid potential competition [4]. This foraging diversification at an otariid population level can occur based on phenotype, reflected by age or sex, e.g., [5,6].

Regarding available resources for CSLs, there is a high biological production year-round in La Paz Bay and its surroundings because of a local mesoscale gyre that appears in the summer [7,8]. As a result of these environmental conditions, the CSLs from Los Islotes prey on several species, mainly on the deepwater serrano (Serranus aequidens), threadfin bass (Pronotogrammus multifasciatus), flagfins (Aulopus sp.), bigeye bass (Pronotogrammus eos), California anchovy (Engraulis mordax), and the enope squid (Abraliopsis affinis), among others [9,10]. Each CSL rookery in the Gulf of California seems to have a distinct prey spectrum, as the remaining colonies in the Gulf show their trophic attributes. This results in significant foraging segregation among the Gulf of California rookeries [9,11] that has been evidenced through scats that reflect the diet of primarily adult females, which is the most abundant class within CSL rookeries [1], and stable isotopes in pup fur, which is a maternal foraging indicator [11]. The latter is based on δ¹⁵N values used to assess consumers’ trophic breadth and position, while δ¹³C values provide habitat use information [12,13,14]. Both isotope ratios are also useful to investigate the migratory attributes of consumers [6,15,16].

Our study fills a significant gap based on δ¹⁵N and δ¹³C analysis to determine the foraging habits of all CSL sex and age classes from Los Islotes rookery and its surroundings, expecting a segregation based on life history traits of each sex and age class. This knowledge is vital for a better understanding of the CSL population from the Gulf of California based on an ecologically highly relevant rookery and its surroundings. As previously mentioned, Los Islotes rookery does not follow the declining pattern of most CSL rookeries in the Gulf.

2. Materials and Methods

The rookery of Los Islotes is at the northeastern limit of La Paz Bay in the southwestern Gulf of California (24°35′ N and 110°23′ W), Mexico (Figure 1).

This is formed by two rocky islets that cover an area of around 0.046 km². Los Islotes is part of the Espíritu Santo Archipelago National, which is managed by the National Commission for Protected Natural Areas (CONANP). This area and its surroundings are also an IMMA (Important Marine Mammal Area) denominated as “La Paz Bay and Surrounding Islands” (area of 4929 km²). An IMMA is defined as a discrete portion of habitat that is important to the marine mammal species that inhabit it and that has the potential to be delimited and managed for conservation (https://www.marinemammalhabitat.org/imma-eatlas/, accessed on 27 December 2024). The southwestern region in the Gulf of California is influenced by the convergence of three water masses with different oceanographic characteristics that come from (1) the California Current, (2) the Tropical Pacific Ocean, and (3) water from the Gulf of California [17].

Sample Collection

All CSL fur samples were taken in the southwestern region of the Gulf of California between 2013 and 2023. Pups (30), adult females (6), and adult males (7) were sampled on Los Islotes rookery. These individuals were immobilized using nets to perform the sampling, except for adult males, which were sampled using a 3 m-long pole with adhesive tape at the tip, based on [6]. Most of the samples from juveniles (9; due to the near-absence of females within this class, they were all classified as “juveniles”) and all subadult males (13) were obtained from strandings in this region. The furthest CSL stranding took place at no more than 40 km from Los Islotes. All CSL classes were identified based on their size, proportions, pelage color, and presence/absence of sagittal crest [18]. All fur samples were taken from the dorsal region (3 × 3 cm area) using scissors. All of these were cleaned with a chloroform/methanol solution (1:1 ratio), cut into small segments, and homogenized with an agate mortar. Using an analytical microbalance (precision of 0.0001 mg), 0.8–1.2 mg subsamples were taken and stored in tin capsules (8 × 5 mm) for their analysis in an isotope ratio mass spectrometer. Stable isotope proportions were represented as delta (δ) with units expressed as parts per thousand ‰, [19]: δ¹⁵N or δ¹³C = 1000 [(Rsample/Rstandard) − 1], where: Rsample and Rstandard are the ratios of ¹⁵N/¹⁴N or ¹³C/¹²C for each sample and the standard, respectively. The recognized international standards for these elements are Vienna Pee Dee Belemnite (V-PDB) for carbon and atmospheric nitrogen (N₂) for nitrogen [19].

We analyzed the data using a hierarchical Bayesian model. Bayesian Inference (BI) reallocates the credibility of parameter values among candidate possibilities, using Bayes’ theorem for evaluation, given the data, the model, and prior knowledge about the parameters [20,21]. Moreover, hierarchical Bayesian models partially pool the data and “shrink” the group-level estimates towards the shared estimates [22]. The model was specified as follows, where $i$ represents the isotopic ratio and $j$ the CSL class (pup, juvenile, subadult male, adult female, adult male):

• Shared priors
  • ${\mu }_i\sim \mathrm{Normal}\left(\mu =\overline{{\delta }_i},\sigma =SD\left({\delta }_i\right)\ast 10\right)$: A regularizing normal distribution centered on each isotope’s mean and standard deviation (times 10).
  • ${\sigma }_i\sim \mathrm{Exponential}\left(\lambda =1\right)$: A weakly informative exponential distribution.
• Group-level priors
  • ${\mu }_{i,j}\sim \mathrm{Normal}\left(\mu ={\mu }_i,\sigma ={\sigma }_i\right)$: A normal distribution centered on each isotope’s global mean with the global standard deviation.
  • ${\sigma }_{i,j}\sim \mathrm{HalfCauchy}\left(\beta =5\right)$: Regularizing Half Cauchy distribution for the standard deviations of each isotope and category.
  • ${\nu }_{i,j}\sim \mathrm{Exponential}\left(\lambda =1/30\right)+1$: A shifted exponential distribution centered around 30, spreading the credibility across heavy and light-tailed (normal-like) distributions.
• Likelihood:
  • ${\delta }_{i,j}\sim \mathrm{Student}\mathrm{T}(\mu ={\mu }_{i,j},\sigma ={\sigma }_{i,j},\nu ={\nu }_{i,j})$: A Student t likelihood that assigns a higher probability to extreme values, thus making it useful for robust estimations of parameters [20].

The model was created using the PyMC module (v.5.8.1) [23,24] for Python (v.3.11.5) [25]. The posterior distributions were sampled using four Markov Chain Monte Carlo (MCMC) algorithms using a No-U-Turn Sampler (NUTS) [26].

The differences in mean isotopic values were then calculated by subtracting each group’s posterior means from every other class, and expressed in terms of the mean posterior difference and the probability of the difference being greater or lower than zero (no differences).

Additionally, we estimated each category’s isotopic niche amplitude using the SIBER package (v. 2.1.9) [27] for R (v. 4.3.2) [28]. The widths provided are (a) the total area (TA) covered by the data and (b) Bayesian Standard Ellipse Areas $(SE{A}_B$), which are unbiased to sample size and represent the population’s isotopic niche amplitude, calculated by an eigendecomposition of the covariance matrix’s posterior distribution.

The MCMC were run until convergence, i.e., zero divergent posterior samples, Bayesian Fractions of Missing Information close to 1 [29], and potential scale-reduction factors [22] <1.01 for every parameter. The effective sample size for every parameter was greater than 3000.

Every posterior distribution (either from the parameters or the differences in means) was summarized with its mean and 95% Highest Density Interval ($HD{I}_{95\mathrm{\%}}$). Lastly, although the hierarchical model was created in Python, the posterior samples were imported into R and plotted using the ggplot2 package (v. 3.4.4) [30].

3. Results

The global posterior means for δ¹³C and δ¹⁵N were $-15.5\pm 2.2‰$ and $21.0\pm 2.2‰$, respectively. The posterior means per isotopic ratio and class are presented in

Table 1. Sample sizes and summaries of the posterior distributions (mean, [95% Highest Density Intervals]) for the mean isotopic values and Bayesian Standard Ellipse Areas ($SE{A}_B$), as well as the total areas (TA).

Class	n	δ13C (‰)	δ15N (‰)	TA (‰2)	$\mathit{S}\mathit{E}{\mathit{A}}_{\mathit{B}}$ (‰2)
Pups	30	−15.7; [−15.9, −15.5]	22.7; [22.5, 22.8]	2.9	0.8; [0.5, 1.1]
Juveniles	9	−16.3; [−17.0, −15.5]	21.1; [19.8, 22.3]	5.9	4.5; [1.9, 7.8]
Subadult males	13	−15.2; [−16.1, −14.4]	19.4; [18.9, 20.0]	9.1	4.0; [2.0, 6.4]
Adult females	6	−15.1; [−16.1, −14.1]	20.3; [19.6, 21.0]	1.9	1.9; [0.6, 3.7]
Adult males	7	−14.3; [−15.9, −14.8]	21.4; [20.8, 22.0]	1.2	1.1; [0.4, 2.0]

and in the margins of Figure 2 with the isotopic niches. The isotopic niche amplitudes are also included in Table 1 and the main plot of Figure 2.

The posterior mean differences are shown in Figure 3. Overall, most comparisons had highly probable differences (>75%). The highest probabilities of differences in δ¹³C were found in pups and juveniles relative to the other classes ($P>75\mathrm{\%}$ in all cases). The largest difference was found between adult females and juveniles (${\mu }_{\mathrm{diff}}=1.2‰;P\left(AF>J\right)=97.1\mathrm{\%}$), where juveniles had the more negative values.

Regarding δ¹⁵N, only the differences between adult males and juveniles were not highly probable (${\mu }_{\mathrm{diff}}=0.3‰;P\left(AM>J\right)=68.3\mathrm{\%}$). The largest positive difference for δ¹⁵N was found between adult and subadult males (${\mu }_{\mathrm{diff}}=2‰;P\left(AM>SM\right)=100\mathrm{\%}$), while the largest negative was found between juveniles and pups (${\mu }_{\mathrm{diff}}=-1.6‰;P\left(P>J\right)\approx 100\mathrm{\%}$).

4. Discussion

Our findings evidenced resource partitioning that resulted from foraging segregation among CSL sex and age classes in the southwestern Gulf of California. These differences relate to each class’s energetic requirements and life history traits. Individuals who passed the juvenile stage foraged on resources with similar δ¹³C and different δ¹⁵N values. Our inferences are only valid within the fur timeframe, reflecting the isotopic diet composition during tissue synthesis [31], that is, from 3 to 4 months in the case of pups to several months per year in the remaining classes.

CSL adult females presented lower δ¹⁵N values than most classes, showing highly probable differences compared to most. Adult females are regional indicators since they do not migrate and must lactate for about one year [32]. Therefore, they perform foraging trips around their rookery within a 30–50 km radius [33], taking advantage of available regional resources [9]. This local attribute reflects the ¹⁵N-enriched characteristics of the Gulf of California due to an intense denitrification process compared to the Pacific Ocean at latitudes of the Baja California Peninsula [34]. These δ¹⁵N baseline differences between the Gulf and the Pacific Ocean are used to distinguish several marine mammal species from both provenances [35]. Therefore, we suggest that subadult males, which showed lower δ¹⁵N values than adult females, carry foraging information from the Pacific Ocean (¹⁵N-depleted) into the southwestern Gulf of California, probably from the west coast of the Baja California Peninsula during summer, when they are mostly absent from La Paz Bay and its surroundings [36]. When baseline isotopic differences are not a relevant decisive factor, dimorphic otariid males evidence higher δ¹⁵N values than females, apparently related to their higher trophic position due to their larger body mass and potential consumption of larger, higher trophic-positioned prey. This variation (males > females) occurs in CSLs from the Mexican Pacific Ocean [37] and in other otariids from the northeastern Pacific Ocean/California Current Ecosystem, such as the Guadalupe fur seal, Artcocephalus townsendi [6], or from other regions, in the case of the Antarctic fur seal, Arctocephalus gazella [5]. Subadult males also showed the largest isotopic niche (TA and SEA_B) among all CSL classes, given by a high δ¹⁵N and δ¹³C variability, which is a consequence of their migratory capacity, not only latitudinally but also longitudinally along a shore–offshore gradient.

We suggest that adult males foraged in the Gulf of California based on the arguments above and their high probability of being different from adult females due to their high δ¹⁵N values (highest after pups). They also reflected one of the less variable isotopic niches among all classes. Moreover, their δ¹⁵N values indicated a higher trophic position than adult females. These adult males presented high isotope ratio values, similar to other marine mammal top predators from the ¹⁵N-enriched Gulf of California that showed a trophic position between 4 and 5 [35]. The difference in δ¹³C values between adult males and females was not highly probable, meaning there would not be a marked variation regarding their inshore–offshore habits gradient, with a relatively higher inshore influence in males. In this regard, ¹³C-enriched benthic resources could additionally influence (not significant) adult males due to their high diving capacity compared to adult females [38], as also suggested for the same species in the Mexican Pacific Ocean [37]. These authors also suggested that, in that region, adult males have a higher trophic position than adult females, based on δ¹⁵N values. CSL adult males had a 19‰ δ¹⁵N mean value in that study [37], indicating a significant δ¹⁵N difference (more than 3‰) between individuals from the Pacific Ocean and adult males from the Gulf of California (this study) due to regional baseline δ¹⁵N variations [34]. These findings enhance our knowledge of this species in the Gulf of California since they provide critical knowledge regarding the unknown foraging grounds of adult males from this region (individuals did not leave the Gulf even though they were close to the mouth).

Although our sample sizes for adult individuals were small (six females and seven males), the use of a hierarchical Bayesian model mitigates the differences in sample sizes with other classes, while also providing robust estimates due to the Student T likelihood. Nonetheless, further studies should increase the sample size for these classes to reduce potential biases or complement information by using other approaches, such as telemetry, to compare the movements of both sexes during adulthood.

Juveniles showed the most negative δ¹³C values among all CSL classes, suggesting offshore foraging habits related to potentially distant foraging trips. This inference is supported by some of the largest isotopic niches (TA and SEA_B) shown by this age class relative to the rest. A greater horizontal dispersion by juveniles could result from a reduced oxygen storage capacity compared to adults, limiting their vertical diving [39]. In contrast to other classes, including adult females, juvenile Guadalupe fur seals from the Northeastern Pacific Ocean were also suggested to have large isotopic niches due to their dispersal ability [6]. Another explanation for this high variability in juveniles is the mixed consumption of maternal milk and their first prey. The lactation period in this species takes place during the first year of life [32], ending when the juvenile stage begins. However, there can be delayed weaning in a non-consistent manner [37,40], where some juveniles are still in a transitional period between milk dependence and the establishment of their foraging skills. Milk in juveniles would provide high δ¹⁵N values (see explanation below for pups, which showed the highest δ¹⁵N values). Concurrently, their first prey is likely small and easy to hunt due to a lack of foraging experience, resulting in lower trophic levels that provide low δ¹⁵N values to consumers. In this regard, low trophic organisms such as the pelagic red crab (Pleuroncodes planipes) have been mentioned as a likely important prey with low δ¹⁵N values for juvenile CSLs in the Mexican Pacific Ocean [37,41]. This “prey factor” for young individuals has also explained isotopic differences between age classes in other otariids, such as the northern fur seal, Callorhinus ursinus [42]. Therefore, the overall consequence of this resource combination (milk and first prey) could have created a large isotopic niche for juveniles. Most of the of the samples of this age class, as well as all those of subadult males, were obtained from stranded individuals in this region. Therefore, they could have been nutritionally stressed, sick, or under other conditions that could alter stable isotope values. This scenario cannot be totally discarded; however, since fur reflects the isotopic diet composition during this tissue synthesis [31], which in the case of these individuals is of several months, we suggest that this stranding effect should not be significant.

The pups sampled (around one month old) were not independent prey consumers. They only consumed maternal milk, resulting in the highest δ¹⁵N values among all CSL classes. Pups assimilate the isotopic signal of their mothers through the maternal tissue catabolism used to produce milk, plus an additional enrichment that emulates the trophic relationship between a predator and its prey, showing an approximate 1‰–2‰ δ¹⁵N increase relative to adult female otariids [6,11,43]. Conversely, pups showed lower δ¹³C values than adult females, subadult males, and adult males. Since pups are not foraging in the ocean, these negative δ¹³C values are related to ¹³C-depleted lipids that characterize the fatty milk consumed by pinniped pups, compared to the diet (prey) of the other age classes [6,44].

Based on the findings of this study, we acknowledge that all classes should be analyzed under the same environmental conditions; however, it is difficult to access large sample sizes of each class within a narrow timeframe. Thus, our sampling period extended between 2013 and 2023. Future studies should assess the interannual environmental effect on the foraging habits of the different CSL classes, especially due to the population decline of this species in the Gulf of California in recent decades (not in the southwestern region, however). Nevertheless, our study evidenced a clear CSL foraging segregation in the southwestern Gulf of California, mainly explained by known life history traits of each sex and age class. Additionally, our foraging segregation results agree with previous studies that have reported this same pattern in other otariid species, such as the Galapagos fur seal (Arctocephalus galapagoensis) [45], the New Zealand fur seal (A. fosteri) [46], the northern fur seal (Callorhinus ursinus) [47], the Antarctic fur seal (A. gazella) [5], and the Guadalupe fur seal (A. townsendi) [6], among others.

This apparent foraging flexibility in CSLs from the southwestern Gulf of California among different sex and age classes as a strategy to avoid potential competition for resources could partially explain why the abundance in this rookery is not declining, as others from the Gulf of California are due to environmental changes that could have negatively impacted CSL trophic dynamics [1991–2019; 1]. Similar future studies should be undertaken in other rookeries from the Gulf to assess isotopic niche partitioning patterns among classes in other localities. This would allow a better ecological understanding of the CSL population in this region, providing tools for improved conservation programs across this region’s different natural protected areas.