DISCUSSION
We found that of four carnivore species, only two appeared to actively search for elk and none searched for mule deer neonates. The fact that no carnivores encountered parturient mule deer more often than expected by chance was consistent with our predictions that the young of less abundant species would not be targeted (Figure 1b); mule deer are at least 5 times less abundant than elk in this ecosystem (Oregon Department of Fish and Wildlife, unpublished data). This result suggests that mule deer fawn mortalities in this study area (Jackson et al. 2021) are likely the result of fortuitous encounters (from the perspective of the predator) and that risk to mule deer from incidental predation may thus depend on the amount of overlap between deer and elk parturition habitat for predators that are primarily searching for elk calves. Our finding that two of the four carnivore species actively searched for elk (the more abundant species) partially aligned with our expectations that a predatory response would be greater toward the more abundant prey species. We also expected the more generalist carnivores (bears and coyotes, which are highly omnivorous) would exhibit a more fluid response to an ephemeral prey source. This hypothesis was supported for bears but not coyotes. The lack of response of bobcats and coyotes to both elk and deer may be explained for several reasons. First, elk calves are large prey items for mesopredators such as coyotes and bobcats particularly when subject to maternal defense from elk. While mule deer young are much smaller, their rarity on the landscape may not have warranted search behavior. Nonetheless, we can speculate that had mule deer been the more abundant species in this area, we may have observed some level of search behavior by coyotes and/or bobcats. Second, bobcats are efficient predators of small prey while coyotes in this system gain a substantial amount of protein via scavenging (Ruprecht et al. 2021b), both of which potentially reduce the need to pursue neonates. Finally, our sample size of collared bobcats was small (N = 6) so results may have been subject to Type II error for that species.
Our analyses to evaluate whether carnivores’ use of elk parturition habitat tracked the phenology of the birth pulse revealed further contrasting behaviors among the predators in our study. Use of elk parturition habitat by male bears was best explained by a model with a quadratic effect of Julian week, with its maximum aligning almost perfectly with the peak of the birth pulse (Figure 6b). This result suggests that male bears exhibited a spatial shift in habitat consistent with an effort to maximize encounters with elk neonates immediately following parturition, which is logical given that bears are limited by a short window in which they can efficiently hunt neonates. This idea is further supported in that the number of encounters between GPS-collared bears and parturient elk in the 30 days post-parturition peaked between 7-10 days after an elk gave birth. By contrast, the quadratic effect was not supported for cougars (Figure 6d). Although previous research has shown that juvenile elk constitute a large fraction of the diets of cougars (Clark et al. 2014), several aspects of cougar predatory behavior may explain why cougars did not exhibit a quadratic effect indicative of searching areas used by neonates. First, because elk become solitary to give birth before rejoining the herd several weeks later (Paquet & Brook 2004), it may be inefficient for cougars to target solitary mother-young pairs when they would have access to more individuals by pursuing larger herds of mixed age classes. Such “nursery herds” (Paquet & Brook 2004) would present naïve prey such as yearling elk that may have recently lost maternal guidance, or vulnerable young of the year after the mother-young pair rejoined the herd after parturition. The number of encounters between GPS-collared cougars and parturient elk in the 30 days post birth was greatest around 20 days after an elk gave birth which further supports the idea that cougars pursued calves only after they had matured for several weeks. We were initially concerned that the apparent use of parturition habitat by bears coinciding with the birth pulse could be driven by selection for other dynamic resources such as green vegetation that was correlated with elk parturition habitat, causing a spurious result. However, only male bears exhibited this pattern, and we would expect females to exhibit the same response to bottom-up resources.
Although we are unaware of previous research on search behavior of cougars toward ungulate neonates, our work both aligns and contrasts with patterns described for black bears and coyotes elsewhere. The density of elk neonates in our study area was closer in magnitude to the number of caribou calves in the study by Rayl et al. (2018) that determined bears actively searched for neonates than it was to the study by Bastille-Rousseau et al. (2011), in which bears opportunistically encountered neonates. Our finding that male bears were much more likely to encounter neonates than were females also aligns with Rayl et al. (2015) who found that male bears were more likely to visit caribou calving grounds than were females. Further, several other studies have documented higher predation or meat consumption by male black bears than females (Boertje et al.1988; Jacoby et al. 1999), which should be expected given that previous research has shown that larger, male bears require more animal-borne protein to gain weight than do smaller, female bears (Rode, Robbins & Shipley 2001). Our results did not suggest that coyotes actively searched for elk calves which accords with a cause-specific mortality study of elk calves in this region that found coyote predation to be minor (Johnson et al. 2019). In other ecosystems, however, coyotes have been implicated as nontrivial sources of mortality for elk calves (Barber-Meyer, Mech & White 2008) although there is mounting evidence that coyote predation on mule deer neonates occurs largely only when small mammal populations are low (Hamlin et al. 1984; Hurleyet al. 2011). We unfortunately did not have sufficient data on occurrence and abundance of alternative prey to assess whether this occurred in our study.
It is important to view the distinction between active search and incidental encounter in light of the effects that predator population dynamics may have on prey. If predators employ active search behavior, then a reduction in predator density may not yield increased neonate survival because neonates spared by that individual become targets for the remaining pool of searching predators (unless prey density is such that predation is limited by handling time or satiation). But if neonate predation occurs because of incidental encounters, then a reduction in predator density benefits a focal prey species by reducing encounter rates both because each prey is less likely to be incidentally encountered per unit time and because predators spend time handling other species. Although population growth rates of ungulates are most sensitive to survival of adult females, this demographic rate is consistently high and stable (Gaillard, Festa-Bianchet & Yoccoz 1998) and many ungulate populations are instead limited by insufficient recruitment due to low neonate survival (Raithel, Kauffman & Pletscher 2007). Thus, knowledge of how different species of carnivores search or encounter different species of prey will be needed to determine the extent to which predator control would be an effective strategy for managing ungulate populations.
A necessary assumption in our study was that predator and adult female ungulate locations within a 200-meter proximity at the time of simultaneous GPS recordings constituted an encounter with an ungulate neonate in the weeks after birth. This assumption was required because we did not have GPS transmitters on neonates and thus assumed the location of the adult female was a reasonable proxy for the location of the neonate. This assumption certainly introduced some amount of data contamination that may have obscured a stronger signal than what we observed. Although our dataset was unique in that it included contemporaneous GPS recordings on four species of carnivores and two species of ungulates, with fix intervals up to three hours on predator GPS collars, we are certain to have missed additional encounters. Another limitation in our study was that we were required to use inference methods to predict some of the parturition events for deer and all parturition events for elk which are subject to error in the exact timing of births. Further, some number of neonates likely died in the days following birth which our analysis could not consider. These factors should only have acted to dilute any potential signals in the data and not cause spurious correlations. Nonetheless, we caution that our results are conservative and should be interpreted with the possibility that Type II error may be present. These issues will be minimized as GPS transmitters become sufficiently lightweight to be placed on neonates and as battery life improves such that more frequent GPS readings can be obtained on both predators and prey.
An emerging frontier in animal ecology is understanding if and how behavior influences population dynamics. Consequently, elucidating how predatory tactics (e.g. active search vs. incidental encounters) affect prey populations should feature prominently in future research. Our results suggest there was a behavioral response by two of the four carnivores toward elk, but no response by any of the carnivores toward mule deer. This result combined with previous research suggests that the foraging tactics by predators in response to a pulsed resource differ by both predator and prey species, are likely ecosystem-specific, and can change dynamically through time as the availability of vulnerable neonates fluctuates.