Discussion
Our study identified strong direct and indirect effects of active, seeking predators (weasels) on arboreal rodent presence, activity, and survival at nest platforms. Arboreal rodents, primarily tree voles, were killed most frequently by weasels and their activity several weeks post weasel-detection remained depressed. Secondary nest predators, thrushes and jays, probed and dug out nests in apparent pursuit of invertebrates in discarded and decaying nest materials of tree voles, and were positively correlated with flying squirrel presence, perhaps indicating that flying squirrels were attracted to nest disturbances (Forsman & Swingle, 2007).
Predator foraging mode can strongly influence the timing and location where prey are killed. In our study, weasels (i.e., active, seeking predator) were the strongest nest predators whereas owls and flying squirrels exhibited comparatively weak effects at prey nest sites. Owls were detected at similar rates as weasels but may simply be ineffective nest predators, unable to reliably enter or drive tree voles from their nests. Our inferences were limited to nest platforms in young forests and did not extend to foraging tree voles or to the old forests in which strong relationships between tree voles and one of their main predators, the northern spotted owl (Strix occidentalis caurina ), have been established (Forsman, Anthony, Meslow, et al., 2004). Nonetheless, our results provide evidence that tree voles face strong pressure from weasels at nests and owls while tree voles are foraging (Forsman, 1974; Swingle et al., 2010).
Although foraging mode may have differed, diel activity patterns of predators broadly overlapped those of tree voles except for diurnally active digging birds. Weasels can be active throughout the day (Linnell, Epps, Forsman, & Zielinski, 2017) but appeared to be more effective at capturing tree voles in the morning hours (Fig. 3b), a pattern similar to least weasels (M. nivalis nivalis ) which were more active at sunset but captured most Microtus field voles in the morning (Sundell, Norrdahl, Korpimäki, & Hanski, 2000). Determining if this reflects a temporal vulnerability of prey will require additional studies, including where and how weasels use cues to locate arboreal prey.
Weasels can cue into rodent scent when hunting, and scent can accumulate at or near small rodent nests and is hypothesized to increase predation risk (Sharpe & Millar, 1990; Ylönen, Sundell, Tiilikainen, Eccard, & Horne, 2003). Female tree voles may be especially at risk of nest-site mortality because they have high fidelity (typically only one nest used per month) to nests that they invest substantial effort into building, resulting in extensive structures (0.06 m3) (Swingle 2005; Sharpe and Millar 1990). Large, old tree vole nests typically contain multiple tunnels and chambers within layers of soil, providing shelter and escape routes from predators (Maser, 1966). The relative tradeoffs of long-term habitation with predation risk from predators cuing into accumulated scent remains unknown but the higher visitation rates by weasels after year three in our study indicates a response of these predators to prolonged tree vole occupancy of nests. Whether this was due to a functional response to environmental cues such as scent or to numerical response of the predators is unknown, but indicates predation risk can limit long-term population growth of tree voles regardless of an increase in resource availability (i.e., new nests).
Homogeneity of nest platform placement could have provided a visual cue to avian nest predators (Santisteban, Sieving, & Avery, 2002) although vertical placement within the live tree canopy was broadly representative of natural nest substrates found in young forests (Linnell et al., 2018; Swingle, 2005). In contrast, older forests have much higher heterogeneity in tree height and natural substrates for tree vole nests vary in type (Lesmeister & Swingle, 2017), location within the canopy, and may be more numerous (Swingle, 2005). Moreover, higher nests in a more heterogenous environment may disperse scent, decoupling cues from terrestrial predators. These characteristics of tree vole nests in old forest—more numerous nest substrates, varied height, size, and substrate—may make those nests more difficult to locate for predators using visual or olfactory cues.
Predators can effectively reduce resource availability for prey through non-lethal avoidance and non-lethal effects can be of higher magnitude than direct consumption, an effect closely tied to prey population densities (Preisser et al., 2005). In the case of weasels in our study, we observed potentially lethal effects followed by reduction of activity (a non-lethal effect) by tree voles for up to several weeks post weasel detection. Most weasel detections occurred towards the end of our study, limiting our inferences with regards to longer-term non-lethal effects of weasel presence on tree vole nest platform occupancy. But given the limited nest substrate availability in young forests, removal of even several productive nest substrates through predation followed by non-lethal avoidance could cause a substantial limit on tree vole populations. Continued monitoring of nest platforms to assess long-term lethal and non-lethal effects could provide a measure of the effect nest predators have on limiting tree vole populations in young forests.
The relative degree of specialization of predators can determine their functional response to numerical increases in their prey (Sundell et al., 2000). Least weasels, similar-sized to short-tailed weasels in our study, show a type II functional response to higher densities of field voles (rapid initial increased predation rate with prey density) in boreal ecosystems with a lower diversity of prey available (Sundell et al., 2000). In temperate forests weasels have access to a higher diversity of prey but specialization depends on sex and age of weasels with adult females typically the most specialized (King & Powell, 2007). Weasels seem unlikely to exhibit a type II functional response to tree voles, particularly because tree voles do not reach sufficiently high densities to elicit such a response and functional response of weasels may be more similar to a generalist predator (type III) than a specialist (type II) with regards to tree voles (Sundell et al. 2000).
We observed a marked difference in model strength and direction at our finest temporal grain size (1-hour bin-width) compared to coarser bin-widths (1-day and 1-week), which could dramatically affect interpretation of the observed species interactions. Weasels clearly exhibited the strongest indirect (reduction in tree vole activity) and direct (predation) effects on tree voles but in models using 1-hour bin-widths these effects were indistinguishable from other species and the direction of the relationship was negative, opposite that of models with coarser bin-widths. Using multiple bin-widths provided additional insights into predator-prey interactions. For example, the lack of positive correlations of weasels and tree voles at 1-hour bin-widths indicates that weasels were unlikely to have visually cued into tree vole activity as they did not overlap directly the exact hours in which tree voles were detected even though our sampling schedule (1 photo per minute) should have been more than adequate to detect such overlap. For the predator-prey interactions that we observed in our study, temporal grain sizes appeared to have strong effects in interpretation of model output particularly at the finest temporal grain.
Our models were likely inadequate at detecting weak non-lethal interactions occurring over longer time periods, such as potential interference competition between flying squirrels and tree voles. Flying squirrels were ubiquitous but weak lethal predators of tree voles at nests killing them at a rate of 0.03% (n = 2 observed mortalities) compared to the 10% (n = 8 observed mortalities) predation rate of weasels (Table 1). Nonetheless, the observed weak decrease in tree vole activity up to several weeks after detection of a flying squirrel (Fig. 3b) providing circumstantial evidence that flying squirrels exhibit competition with tree voles, albeit weak and potentially lagged and further evidence would be needed to corroborate this observation.
Our technique of artificial marks and visual recaptures using multiple remote cameras at each site appeared to function well for females, but this technique may have been less effective for males which have lower nest fidelity (and conversely, higher local emigration rates; Swingle 2005). Furthermore, our hair clipping technique was effective, but limited the temporal extent of visual recaptures as fur grew back, obscuring our marks. This technique, narrowly applied to monitoring nests of high nest fidelity animals such as female tree voles, may provide accurate estimates of apparent survival as emigration is expected to be relatively low and observations of nest predators can corroborate the fate of marked individuals.
Tree voles responded with strong population growth in 2016 1-year after addition of nest platforms (Linnell et al., 2018). Herein we demonstrated that predator activity increased over a 3-year monitoring period and had strong direct effects on tree voles at nest platforms. The success or failure of nest-box and nest platform to increase prey population size may ultimately depend on longer term predator-prey dynamics inherent to the cover types in which artificial nest substrates are placed (Sonerud, 1985). Our evidence indicates that predators, particularly weasels, can exert strong acute effects on tree vole activity at nests in young forests. Determining whether predators in young forests represent a substantial limit to tree vole populations as compared to old forests per se, will require additional monitoring, ideally of naturally occurring nest substrates or nest platforms (or both) in old forests.