4.3 Correlation between relative specialization and prey species composition
Our data suggest that the higher proportion of benthic species consumed, the relatively more specialized the diet of the predator. This pattern was observed in the full dataset, as well as when female and male data were considered separately (Tables 2-5). This information ties to our knowledge of the foraging patterns of male versus female harbor seals in the region. Females more often perform deeper foraging dives (Wilson et al., 2014) and eat more benthic species than males, who eat more pelagic species (Schwarz et al., 2018). In Scotland, harbor seal scat samples represented either a largely pelagic foraging strategy or largely benthic foraging strategy (Tollit, Greenstreet, & Thompson, 1997), and males had larger range and duration in foraging trips (Thompson, Mackay, Tollit, Enderby, & Hammond, 1998), suggesting that the separation between the two foraging strategies is not just a regional phenomenon.
The consistency across sexes indicates this pattern is reflective of foraging strategies specific to the ecology of prey species, and not just indicative of different diet preferences between males and females. We hypothesize that this pattern was caused by higher variability in benthic environments (Lalli & Parsons, 1997). If prey have more variable life strategies, a single foraging strategy will not suffice to catch them all. Because an organism is likely limited in the number of foraging strategies at which it can be effective, an individual could be limited in the number of prey species it can exploit.
There is the possibility that the sex of the individual determines the level of specialization regardless of the prey consumed. However, the consistency of benthic prey being associated with a relatively specialist diet, and pelagic prey being associated with a relatively generalist diet, in the complete, only female, and only male datasets suggests that the ecology of the prey species and the sex of the seal was driving the observed pattern. This idea is supported in other literature as well. Individual male harbor seals in Nova Scotia use different behaviors when foraging for benthic versus pelagic prey (Bowen, Tully, Boness, Bulheier, & Marshall, 2002) and larger seals are more likely to forage in pelagic environments regardless of sex (Bjorkland et al., 2015).
If prey species ecology is driving specialization levels, it is especially interesting to consider harbor seal consumption of juvenile Salmoniformes. For example, juvenile sockeye (Oncorhynchus nerka , Walbaum 1792) correlated with a generalist diet (rho = 0.22, p = 0.004). This could indicate that seals were not seeking out juvenile Salmoniformes specifically but rather eating them as a byproduct of focusing on fish that match the image of forage fish (e.g. small and silver) while conducting pelagic foraging strategies. This is just one example of how understanding the level of specialization could deepen our scope of knowledge regarding harbor seal impacts on prey species of concern.
4.4 Study limitations
There are a few notable limitations to this study. First, there was the potential for variation in sample size to introduce bias. However, there were no discernable patterns between sample size and average relative specialization by season (Figures 2, 3). We also included sample size as a random factor in the model to account for any bias introduced there. Hence, any bias introduced by sample size was likely minimal. Second, because scat were collected from the same haul-out multiple times there is a chance that some scat collected came from the same individual. However, this chance is low as Rothstein et al. (2017) estimated the sampling scheme to track five individuals at Cowichan Bay (i.e. a single haul-out) as 440 samples over 22 sampling bouts. Compared to the 1,083 samples used in this analysis from five different haul-outs, it seems unlikely there was a high rate of resampling the same individuals. Third, there are biases in the metabarcoding PCR process for determining diet (Thomas, Jarman, Haman, Trites, & Deagle, 2014). The prey proportions recorded for each sample are not directly proportional to the amount of prey that was ingested (Bowen & Iverson, 2012; Thomas et al., 2014). However, this approach is accepted to be semi-quantitative, biases are assumed to be consistent between samples (Thomas et al., 2014), and the approach has been successfully used in other studies (Deagle et al., 2010; Pompanon et al., 2012; Schwarz et al., 2018; Thomas et al., 2014; Thomas et al., 2017). Furthermore, this approach is superior to the alternative occurrence-based methods for generating diet proportions (Deagle et al., 2019). On a related note, these molecular methods do not provide data that directly equate to counts of prey consumed. But, if individuals within a local group encounter the same size distribution of a given prey species, then diet proportions represent the same relative relationship of prey capture decisions. Further investigation into potential biases introduced by using proportion type data would be useful as this methodology has many benefits and is a valuable molecular technology that should be applied in the future.