By studying animal movements, researchers can gain insight into many of the ecological characteristics and processes important for understanding population-level dynamics. We developed a Brownian bridge movement model (BBMM) for estimating the expected movement path of an animal, using discrete location data obtained at relatively short time intervals. The BBMM is based on the properties of a conditional random walk between successive pairs of locations, dependent on the time between locations, the distance between locations, and the Brownian motion variance that is related to the animal's mobility. We describe two critical developments that enable widespread use of the BBMM, including a derivation of the model when location data are measured with error and a maximum likelihood approach for estimating the Brownian motion variance. After the BBMM is fitted to location data, an estimate of the animal's probability of occurrence can be generated for an area during the time of observation. To illustrate potential applications, we provide three examples: estimating animal home ranges, estimating animal migration routes, and evaluating the influence of fine-scale resource selection on animal movement patterns.

1. Global positioning system (GPS) technology enables researchers to evaluate wildlife movements, space use and resource selection in detail for extended periods of time. Two types of errors, missed location fixes and location error, are inherent to GPS telemetry and can bias location data sets. Habitat characteristics can influence both types of errors, but no studies have reported how continuous ranges of canopy cover and terrain simultaneously affect location error at different positional dilution of precision (PDOP) and signal quality levels. This information can assist in developing a protocol for removing large location errors from GPS data sets. 2. The objectives of this study were to quantify how canopy cover and terrain affected GPS collar performance within a mountainous region of northern Idaho, USA, and evaluate different data-screening options for GPS location data sets from stationary test collars and free-ranging black bears Ursus americanus . 3. The fix rate for test collars was very high in all habitats (mean 99·5%, SE 0·14, range 97·9-100%) and was not related to canopy cover or terrain obstruction. However, habitat variables strongly influenced location error, PDOP values and proportion of three-dimensional (3-D) fixes. The 95% circular error probable (CEP) equalled 106·8 m for locations at all test sites, and varied substantially with canopy cover, terrain obstruction and signal quality categories, ranging from 14·3 m to 557·0 m. Location errors for two-dimensional (2-D) fixes were more variable at higher PDOP values and were significantly larger compared with 3-D fixes. 4. Data screening increased the accuracy of test collar location data sets by removing large location errors that were associated with high PDOP values. Data-screening options that focused on screening 2-D locations were most effective in reducing location error and retaining the greatest number of locations. For black bear data sets, the four data screening options resulted in data reduction ranging from 8% to 35%. Synthesis and applications.We have demonstrated how location data can be analysed and screened based on 2-D and 3-D fixes in relation to PDOP values to eliminate locations with potentially large location errors. This information can be applied to GPS location data for individual animals to increase data accuracy for analyses.

Motion‐activated cameras (“camera traps”) are increasingly used in ecological and management studies for remotely observing wildlife and are amongst the most powerful tools for wildlife research. However, studies involving camera traps result in millions of images that need to be analysed, typically by visually observing each image, in order to extract data that can be used in ecological analyses. We trained machine learning models using convolutional neural networks with the ResNet‐18 architecture and 3,367,383 images to automatically classify wildlife species from camera trap images obtained from five states across the United States. We tested our model on an independent subset of images not seen during training from the United States and on an out‐of‐sample (or “out‐of‐distribution” in the machine learning literature) dataset of ungulate images from Canada. We also tested the ability of our model to distinguish empty images from those with animals in another out‐of‐sample dataset from Tanzania, containing a faunal community that was novel to the model. The trained model classified approximately 2,000 images per minute on a laptop computer with 16 gigabytes of RAM. The trained model achieved 98% accuracy at identifying species in the United States, the highest accuracy of such a model to date. Out‐of‐sample validation from Canada achieved 82% accuracy and correctly identified 94% of images containing an animal in the dataset from Tanzania. We provide an r package (Machine Learning for Wildlife Image Classification) that allows the users to (a) use the trained model presented here and (b) train their own model using classified images of wildlife from their studies. The use of machine learning to rapidly and accurately classify wildlife in camera trap images can facilitate non‐invasive sampling designs in ecological studies by reducing the burden of manually analysing images. Our r package makes these methods accessible to ecologists.

Biotic and abiotic factors are increasingly acknowledged to synergistically shape broad-scale species distributions. However, the relative importance of biotic and abiotic factors in predicting species distributions is unclear. In particular, biotic factors, such as predation and vegetation, including those resulting from anthropogenic land-use change, are underrepresented in species distribution modeling, but could improve model predictions. Using generalized linear models and model selection techniques, we used 129 estimates of population density of wild pigs (Sus scrofa) from 5 continents to evaluate the relative importance, magnitude, and direction of biotic and abiotic factors in predicting population density of an invasive large mammal with a global distribution. Incorporating diverse biotic factors, including agriculture, vegetation cover, and large carnivore richness, into species distribution modeling substantially improved model fit and predictions. Abiotic factors, including precipitation and potential evapotranspiration, were also important predictors. The predictive map of population density revealed wide-ranging potential for an invasive large mammal to expand its distribution globally. This information can be used to proactively create conservation/management plans to control future invasions. Our study demonstrates that the ongoing paradigm shift, which recognizes that both biotic and abiotic factors shape species distributions across broad scales, can be advanced by incorporating diverse biotic factors.

No abstract