Population cycles in small rodents seen through the lens of a wildlife camera

Population cycles in small rodents have attracted attention from ecologists for more than a century. This spectacular phenomenon is crucial for the functioning of many northern food-webs and has intrigued ecologist because of its lessons for general ecology. Knowledge about the rodent cycle has, however, been hampered by the lack of reliable monitoring methods both for rodents and some of their assumed interactants (e.g. the small mustelids). In resent decades, camera traps have become widely used in ecology as they provide a cost-efficient and non-invasive method for wildlife monitoring. In this thesis, consisting of four studies, I will investigate how camera trap tunnels tailored for small mammals can enhance rodent monitoring. First, in study I, I together with colleagues conducted the first large scale assessment of the applicability of tunnel-based camera traps to estimate population parameters in a small mammal community, including during a long Arctic winter. We showed that the camera trap provide estimates of rodent occupancy also under the snow during winter. Further we give recommendation on micro-scale placement of the traps to maximize technical functionality in order to avoid loss of data. Then, in study II, we expand on dynamic occupancy models for interacting species by including two nested spatial scales. This allows for camera trap-based investigation of the rodent-mustelid interaction on both a local and a landscape scale. Features of this interaction are assumed to be a key to understand the cause(s) of the rodent population cycles. In study III, we apply the statistical framework developed in study II to a dataset derived from the long-term monitoring program Climate-ecological Observatory for Arctic Tundra (COAT). Our results show that presence of mustelids increased the extinction probability of rodents on both a local and a landscape scale. Furthermore, we demonstrate a clear habitat dependence and indications of a season-dependency in the rodent-mustelid interaction strength. Finally, in study IV, we assess whether camera trap-based abundance indices can be used to study population dynamics of two rodent species (gray-sided vole (Myodes rufocanus) and tundra vole (Microtus oeconomus)). This was done by comparing camera trap-based abundance indices to abundance estimated from capture-mark-recapture (CMR). For gray-sided voles a single camera trap provided reliable abundance indices with camera trap counts aggregated over 5-days. For tundra voles counts from four spatially replicated camera traps from a single day within the same local population needed to be aggregated to obtain a good correspondence to the abundance estimated from CMR. Such species-differences imply that the design of camera trap studies should be adapted to the species in focus. This study further highlight that camera traps yield much more temporally resolved abundance metrics than alternative methods. To conclude, the work presented in this thesis demonstrates how camera trap-based rodent monitoring provides multiple improvements compared to previous methods. Camera traps are non-invasive avoiding the ethical issues kill-trapping are fraught with. Further, camera traps provide data year round - including from under the snow- on a fine temporal scale. In addition camera traps are not species specific and provide data on the whole small mammals community including small rodents, shrews and small mustelids. In addition, camera traps provide a reliable abundance index at least for two of the most ecologically important rodents species in northern Fennoscandia. Furthermore, this thesis present a statistical framework for investigating mustelid-rodent interactions based camera trap data and exemplify how this framework can improve on the knowledge on one of the longest standing mysteries in ecology