Motion-activated wildlife cameras (or "camera traps") are frequently used to remotely and noninvasively observe animals. The vast number of images collected from camera trap projects has prompted some biologists to employ machine learning algorithms to automatically recognize species in these images, or at least filter-out images that do not contain animals. These approaches are often limited by model transferability, as a model trained to recognize species from one location might not work as well for the same species in different locations. Furthermore, these methods often require advanced computational skills, making them inaccessible to many biologists. We used 3 million camera trap images from 18 studies in 10 states across the United States of America to train two deep neural networks, one that recognizes 58 species, the "species model," and one that determines if an image is empty or if it contains an animal, the "empty-animal model." Our species model and empty-animal model had accuracies of 96.8% and 97.3%, respectively. Furthermore, the models performed well on some out-of-sample datasets, as the species model had 91% accuracy on species from Canada (accuracy range 36%-91% across all out-of-sample datasets) and the emptyanimal model achieved an accuracy of 91%-94% on out-of-sample datasets from different continents. Our software addresses some of the limitations of using machine learning to classify images from camera traps. By including many species from several locations, our species model is potentially applicable to many camera trap studies in | 10375 TABAK eT Al.

We investigated the spatial attributes of mink (Mustela vison) and muskrat (Ondatra zibethicus) interactions in Canada using 160 geographically paired historic time series of mink (n=80) and muskrat (n=80) harvest data obtained from Hudson's Bay Co. Archives. All series were 25 years in length (1925–1949) and were distributed primarily throughout five ecozones. We used autoregressive models and cross‐correlation analysis to characterize the interactions between mink and muskrat. Model selection results did not differ among ecozones, and indicated that a predator‐prey autoregressive model incorporating a delayed density‐dependent term best described both the mink and muskrat harvest time series. Subsequent analysis of autoregressive coefficients and estimated lags indicated that mink and muskrat interactions vary throughout Canada. In western Canada, the trophic interactions appear to be strong, and mink population cycles lag behind muskrats 2–3 years. In central Canada, mink harvests lagged behind muskrats 1 year, and mink and muskrat interactions in central Canada, with the exception of the Hudson Plains ecozone, were intermediate. In eastern Canada, the trophic interactions appeared weakest, and there were no distinct time lags between mink and muskrat. Stronger interactions in western Canada may be a result of decreased prey diversity, forcing mink to specialize more on muskrats, whereas comparatively stronger perturbations stemming from other trophic interactions may alter the estimated interaction between mink and muskrat in eastern Canada.

Information is sparse about aspects of female wolf (Canis lupus) breeding in the wild, including age of first reproduction, mean age of primiparity, generation time, and proportion of each age that breeds in any given year. We studied these subjects in 86 wolves (113 captures) in the Superior National Forest (SNF), Minnesota (MN), during 1972–2013 where wolves were legally protected for most of the period, and in 159 harvested wolves from throughout MN wolf range during 2012–2014. Breeding status of SNF wolves were assessed via nipple measurements, and wolves from throughout MN wolf range, by placental scars. In the SNF, proportions of currently breeding females (those breeding in the year sampled) ranged from 19% at age 2 to 80% at age 5, and from throughout wolf range, from 33% at age 2 to 100% at age 7. Excluding pups and yearlings, only 33% to 36% of SNF females and 58% of females from throughout MN wolf range bred in any given year. Generation time for SNF wolves was 4.3 years and for MN wolf range, 4.7 years. These findings will be useful in modeling wolf population dynamics and in wolf genetic and dog-domestication studies.

We investigated the dynamic properties of population cycles in Canadian muskrats (Ondatra zibethicus). Ninety-one historic time series of muskrat-harvest data obtained from the Hudson's Bay Company Archives were analyzed. Most series were 25 years in length (1925–1949) and were distributed primarily throughout five ecozones. For each series, we estimated period length and coefficients for a second-order autoregressive model. Estimated period length varied between 3 and 13 years, with 3- to 5-year periods located in Subarctic-Arctic ecozones. We hypothesize that the 4-year cycles are largely a result of predation by red fox (Vulpes vulpes), which exhibit 4-year cycles in Arctic regions. The remaining ecozones generally averaged 8–9 years in period length. However, the relative contributions of direct and delayed density dependence varied along a latitudinal gradient. We hypothesize that both social and trophic interactions are necessary to produce the observed dynamics, but that shifts in the nature of mink predation were responsible for the changes in the relative contribution of direct and delayed density dependence. Essentially, there is a tension between population-intrinsic and trophic interactions that may bound the length of the cycle.

We offer an evaluation of the Caughley and Krebs hypothesis that small mammals are more likely than large mammals to possess intrinsic population regulating mechanisms. Based on the assumption that intrinsic regulation will be manifest via direct density‐dependent feedbacks, and extrinsic regulation via delayed density‐dependent feedbacks, we fit autoregressive models to 30 time series of abundance for large and small mammals to characterize their dynamics. Delayed feedbacks characterizing extrinsic mechanisms, such as trophic‐level interactions, were detected in most time series, including both small and large mammals. Spectral analyses indicated that the effect of such delayed feedbacks on the variability in population growth rates differed with body size, with large mammals exhibiting predominantly reddened and whitened spectra in contrast with predominantly blue spectra for small mammals. Large mammals showed less variance and more stable dynamics than small mammals, consistent with, among other factors, differences in their potential population growth rates. Patterns of population dynamics in small versus large mammals contradicted those predicted by the Caughley and Krebs hypothesis.