Introduction
Lack of resources, including availability of nest substrates and food, can limit abundances of small mammals and birds (Aitken & Martin, 2012; Berthier, Leippert, Fumagalli, & Arlettaz, 2012; Hanski, Hansson, & Henttonen, 1991). In turn, predators are frequently limited by prey availability and increases in abundance of small mammals, at least locally, can support increased predator numbers albeit temporally lagged behind abundances of their prey (Hanski et al., 1991). Artificially increasing resources, including structural habitat (e.g. nest boxes), can provide a unique opportunity to examine intrinsic increases in abundances of the targeted population but also potential responses of the broader vertebrate community, including predators (Aitken & Martin, 2012; Cockle & Martin, 2015; Dunn, 1977). Yet, studies of wild small mammal (<1 kg) population response to increases in structural habitat and the potential responses of predators and competitors remains relatively scarce (Newton, 1994).
Predators can reduce abundances of their prey directly through predation (lethal) and indirectly by inducing non-lethal behavioral constraints to foraging, resting, and reproduction (Preisser, Bolnick, & Benard, 2005). Prey can exhibit differential responses to predator foraging mode, including active avoidance of locations where active seeking predators (e.g. weasels, Mustela spp. ; (King & Powell, 2007) have visited or relying on indirect cues, such as overhead cover and light intensity to select areas with high overhead cover to reduce risk while foraging from sedentary ambush predators such as most forest owls (Jaksić & Carothers, 1985; Jędrzejewski & Jędrzejewska, 1990; Jędrzejewski, Rychlik, & Jędrzejewska, 1993; Kotler, Brown, & Hasson, 1991). Prey vulnerability to predators exhibiting different foraging modes may thus vary depending on where encounters occur, whether at prey resting or foraging sites.
The timing and extent of biological phenomena and the statistical models used to interpret can be sensitive to the period in which observations are made (Steenweg, Hebblewhite, Whittington, Lukacs, & McKelvey, 2018). Continuous monitoring, such as remote camera or video devices, enables varying the temporal grain of observation (e.g., time-bin width of 1-hour vs. 1-day), providing insights into statistical sensitivity but also potentially to observed biological phenomenon. The decision to bin continuous data to a coarser temporal grain is often arbitrary or based on properties of statistical models rather than biological phenomena (Sollmann, 2018).
In 2015, Linnell et al. (2018) initiated a study examining the response of arboreal rodents (red tree vole, Arborimus longicaudus ; Humboldt flying squirrel, Glaucomys oregonensis ; Douglas’ squirrel, Tamiasciurius douglasii ) to an increase in nest substrates in young forests (<80 years old), a resource hypothesized to be limiting there as compared to old forests (≥80 years old). They observed a 5.8-fold increase (95% confidence interval(CI) : 2.9, 9.2) in plot-level occupancy of the main target population (red tree voles, henceforth: tree voles), a small arboreal rodent that builds nests and forages exclusively in the live-tree canopy, and that was not likely to be limited by food as their diet primarily consists of conifer needles, which are readily available in conifer forests. Tree voles are important prey for predators that exhibit different foraging modes, including forest owls (Strix occidentalis caurina, Strix varia, Aegolius acadicus ) and weasels, and have low annual survival of 0.15 (95% CI : 0.06, 0.31), due primarily to high predation rates (Forsman, Anthony, & Zabel, 2004; Forsman & Maser, 1970; Swingle, Forsman, & Anthony, 2010; Wiens, Anthony, & Forsman, 2014) .
Herein we describe predation and non-lethal avoidance of the suite of nest predators and competitors of arboreal rodents, in particular tree voles, during monitoring of artificial nest substrates (henceforth, nest platforms) for three years. We describe the indirect and direct short (i.e. ~1 week) and long term effects (12-weeks) of four taxa that are documented predators of tree voles (weasels, owls) or that may exhibit competition but represent low predation risk (flying squirrels, probing or digging birds) on patterns of nest occupancy by tree voles as observed by remote cameras placed directly above nest platforms (Graham & Mires, 2005; Swingle et al., 2010). We predicted that weasels and forest owls would have an immediate lethal effects (mortality) and cause longer term non-lethal avoidance of nest platforms. Interactions between Humboldt flying squirrels (henceforth, flying squirrels) and tree voles, as estimated from nest platform occupancy, are likely to be more subtle involving weak correlations although have the potential to influence nest occupancy if those interactions occur frequently.