Change-point models for identifying behavioral transitions in wild animals.

Animal behavior can be difficult, time-consuming, and costly to observe in the field directly. Innovative modeling methods, such as hidden Markov models (HMMs), allow researchers to infer unobserved animal behaviors from movement data, and implementations often assume that transitions between states occur multiple times. However, some behavioral shifts of interest, such as parturition, migration initiation, and juvenile dispersal, may only occur once during an observation period, and HMMs may not be the best approach to identify these changes. We present two change-point models for identifying single transitions in movement behavior: a location-based change-point model and a movement metric-based change-point model. We first conducted a simulation study to determine the ability of these models to detect a behavioral transition given different amounts of data and the degree of behavioral shifts. We then applied our models to two ungulate species in central Pennsylvania that were fitted with global positioning system collars and vaginal implant transmitters to test hypotheses related to parturition behavior. We fit these models in a Bayesian framework and directly compared the ability of each model to describe the parturition behavior across species. Our simulation study demonstrated that successful change point estimation using either model was possible given at least 12 h of post-change observations and 15 min fix interval. However, our models received mixed support among deer and elk in Pennsylvania due to behavioral variation between species and among individuals. Our results demonstrate that when the behavior follows the dynamics proposed by the two models, researchers can identify the timing of a behavioral change. Although we refer to detecting parturition events, our results can be applied to any behavior that results in a single change in time.

Surveillance and movements of Virginia opossum (Didelphis virginiana) in the bovine tuberculosis region of Michigan.

Wildlife reservoir hosts of bovine tuberculosis (bTB) include Eurasian badgers (Meles meles) and brushtail possum (Trichosurus vulpecula) in the UK and New Zealand, respectively. Similar species warrant further investigation in the northern lower peninsula of Michigan, USA due to the continued presence of bTB on cattle farms. Most research in Michigan, USA has focused on interactions between white-tailed deer (Odocoileus virginianus) and cattle (Bos taurus) for the transmission of the infectious agent of bTB, Mycobacterium bovis, due to high deer densities and feeding practices. However, limited data are available on medium-sized mammals such as Virginia opossum (Didelphis virginiana; hereafter referred to as opossum) and their movements and home range in Michigan near cattle farms. We conducted surveillance of medium-sized mammals on previously depopulated cattle farms for presence of M. bovis infections and equipped opossum with Global Positioning System (GPS) technology to assess potential differences in home range between farms inside and outside the bTB core area that has had cattle test positive for M. bovis. On farms inside the bTB core area, prevalence in opossum was comparable [6%, 95% confidence interval (CI) 2.0-11.0] to prevalence in raccoon (Procyon lotor; 4%, 95% CI 1.0-9.0, P=0.439) whereas only a single opossum tested positive for M. bovis on farms outside the bTB core area. The prevalence in opossum occupying farms that had cattle test positive for M. bovis was higher (6.4%) than for opossum occupying farms that never had cattle test positive for M. bovis (0.9%, P=0.01). Mean size of home range for 50% and 95% estimates were similar by sex (P=0.791) both inside or outside the bTB core area (P=0.218). Although surveillance efforts and home range were not assessed on the same farms, opossum use of farms near structures was apparent as was selection for farms over surrounding forested habitats. The use of farms, stored feed, and structures by opossum, their ability to serve as vectors of M. bovis, and their propensity to ingest contaminated sources of M. bovis requires additional research in Michigan, USA.

Stable isotope ratio analysis to differentiate temporal diets of a free-ranging herbivore.

Stable isotope ratio analysis (SIRA) of carbon (delta13C) and nitrogen (delta15N) in tissue samples of herbivores can identify photosynthetic pathways (C3 vs. C4) of plants consumed. We present results from free-ranging Rocky Mountain elk (Cervus elaphus) that highlight the ability to differentiate diets using tissue delta13C and delta15N. The signatures of delta13C and delta15N differed in tissues of varying metabolic activity: muscle, a short-term dietary indicator (i.e., 1-2 months) and hoof, a long-term dietary indicator (i.e., 3-12 months). We also documented that delta13C and delta15N values along elk hooves (proximal, middle, distal sections) elucidated temporal shifts in dietary selection. The carbon isotopes of the composite hoof were similar to those of the middle section, but the composite hoof differed in delta(13)C from the distal and proximal sections. The delta13C and delta15N signatures also differed among elk populations, indicating temporal dietary shifts of individuals occupying disparate native range and human-derived agricultural landscapes. Analyses of stable isotopes in various tissues highlighted carbon and nitrogen assimilation through time and differences in the foraging ecology of a rangeland herbivore.