Small mammal abundances are frequently limited by resource availability, but predators can exert strong lethal (mortality) and non-lethal (e.g., nest abandonment) limitations. Artificially increasing resource availability for uncommon small mammals provides a unique opportunity to examine predator-prey interactions. We used remote cameras to monitor 168 nest platforms placed in the live-tree canopy (n = 23 young forest stands), primarily for arboreal red tree voles (tree voles; Arborimus longicaudus), over three years (n = 15,510 monitoring-weeks). Tree voles frequently built nests and were detected 37% of monitoring-weeks, whereas flying squirrels (Glaucomys oregonensis) built nests infrequently but were detected 45% of monitoring-weeks. Most nest predators were detected infrequently (<1% of monitoring-weeks) and were positively correlated with tree vole presence. Weasels (Mustela spp.) were highly effective predators of tree voles (n = 8 mortalities; 10% of detections) compared to owls (n = 1), flying squirrels (n = 2), and Steller’s jays (n = 1). Tree vole activity decreased from 84.1 (95% confidence interval (CI): 56.2, 111.9) detections/week 1-week prior to a weasel-detection to 4.7 detections/week (95% CI: 1.7, 7.8) 1-week post detection and remained low for at least 12 weeks. Interpretations of predator-prey interactions were highly sensitive to how we binned continuously collected data and model results from our finest bin-width were biologically counter-intuitive. Average annual survival of female tree voles was consistent with a previous study (0.14; 95% CI: -0.04 [0.01], 0.32) and high compared to many terrestrial voles. The relative infrequency of weasel detections and inefficiency of other predators did not provide strong support for the hypothesis that predation per se limited populations. Rather, predation pressure, by reducing occupancy of already scarce nest-sites through mortality and nest abandonment, may contribute to long-term local instability of tree vole populations in young forests. Additional monitoring would be needed to assess this hypothesis.

We added 598 nest platforms at a height of 16 ± 4 m to 23 randomly selected young forest sites (17 in 2015, six in 2016) that were located adjacent to old forests that contained sign of tree vole presence (Linnell et al. 2018). At each site, we randomly selected two 100 m² circular plots per hectare and constructed 1 nest platform in the tree at plot center within the live canopy (live limbs vertically above and below). To construct a nest platform, we stretched a length of hexagonal wire mesh (2.54 cm openings) between two or three branches to form an open basket and placed ~8 liters of conifer branch tips and moss within the basket.

We used two sources of data: annual nest platform inspections (n = 1640) that occurred each summer 2016 – 2018, and photographic data from nest platforms monitored with a remote camera (n = 168) June 2015 – October 2018. Each nest platform was inspected annually for diagnostic sign of arboreal rodent nests. Tree vole nests, especially female nests, are frequently large (0.06 m³) and consisted of tunnels and nest chambers formed within discarded resin ducts, fecal pellets, and conifer branch tips whereas flying squirrels primarily built smaller cup-shaped nests from collected moss (Lesmeister and Swingle 2017; Swingle 2005). Tree vole nests were inhabited primarily by one adult except for breeding female nests which often contained juveniles (Swingle 2005).

We deployed remote cameras at a random selection of ~10% (2015) or ~20% (2016) of nest platforms (n = 96). In addition to the randomly selected nest platforms, we placed cameras at 72 nest platforms built in 2015 and containing a tree vole nest identified during the first-year annual inspection in 2016. We combined data collected from randomly (n = 96) and non-randomly (n = 72) placed cameras for all analyses (n = 168 monitored nest platforms).

Remote cameras were mounted 0.6 – 1.0 m above nest platforms and faced down such that the entire nest platform and some adjacent branches were within the field of view of the camera sensor and included in each image. We set each camera to record photos when triggered by motion with a five minute (2015 – 2016) or one-minute (2016 – 2018) quiet period. We tagged each photo with species identity, and for tree voles we also identified age class (juvenile, adult) and, if present, we noted the unique external marking on the tree vole. We tracked photo tagging and estimated a rate of 2616 photos per hour (95% CI: 2406, 2826; n = 175 sessions). Using this rate, we estimated that tagging the 852,000 photos in our data set required 326 hours (95% CI: 301, 354).

Meta-data is uploaded in tab with main data set.