Search for: All records

The implementation of intelligent software to identify and classify objects and individuals in visual fields is a technology of growing importance to operatives in many fields, including wildlife conservation and management. To non-experts, the methods can be abstruse and the results mystifying. Here, in the context of applying cutting edge methods to classify wildlife species from camera-trap data, we shed light on the methods themselves and types of features these methods extract to make efficient identifications and reliable classifications. The current state of the art is to employ convolutional neural networks (CNN) encoded within deep-learning algorithms. We outline these methods and present results obtained in training a CNN to classify 20 African wildlife species with an overall accuracy of 87.5% from a dataset containing 111,467 images. We demonstrate the application of a gradient-weighted class-activation-mapping (Grad-CAM) procedure to extract the most salient pixels in the final convolution layer. We show that these pixels highlight features in particular images that in some cases are similar to those used to train humans to identify these species. Further, we used mutual information methods to identify the neurons in the final convolution layer that consistently respond most strongly across a set of images of one particular species. We then interpret the features in the image where the strongest responses occur, and present dataset biases that were revealed by these extracted features. We also used hierarchical clustering of feature vectors (i.e., the state of the final fully-connected layer in the CNN) associated with each image to produce a visual similarity dendrogram of identified species. Finally, we evaluated the relative unfamiliarity of images that were not part of the training set when these images were one of the 20 species “known” to our CNN in contrast to images of the species that were “unknown” to our CNN.

Human activities shape resources available to wild animals, impacting diet and probably altering their microbiota and overall health. We examined drivers shaping microbiota profiles of common cranes (Grus grus) in agricultural habitats by comparing gut microbiota and crane movement patterns (GPS‐tracking) over three periods of their migratory cycle, and by analysing the effect of artificially supplemented food provided as part of a crane‐agriculture management programme. We sampled faecal droppings in Russia (nonsupplemented, premigration) and in Israel in late autumn (nonsupplemented, postmigration) and winter (supplemented and nonsupplemented, wintering). As supplemented food is typically homogenous, we predicted lower microbiota diversity and different composition in birds relying on supplementary feeding. We did not observe changes in microbial diversity with food supplementation, as diversity differed only in samples from nonsupplemented wintering sites. However, both food supplementation and season affected bacterial community composition and led to increased abundance of specific genera (mostly Firmicutes). Cranes from the nonsupplemented groups spent most of their time in agricultural fields, probably feeding on residual grain when available, while food‐supplemented cranes spent most of their time at the feeding station. Thus, nonsupplemented and food‐supplemented diets probably diverge only in winter, when crop rotation and depletion of anthropogenic resources may lead to a more variable diet in nonsupplemented sites. Our results support the role of diet in structuring bacterial communities and show that they undergo both seasonal and human‐induced shifts. Movement analyses provide important clues regarding host diet and behaviour towards understanding how human‐induced changes shape the gut microbiota in wild animals.

Animals generally benefit from their gastrointestinal microbiome, but the factors that influence the composition and dynamics of their microbiota remain poorly understood. Studies of nonmodel host species can illuminate how microbiota and their hosts interact in natural environments. We investigated the role of migratory behaviour in shaping the gut microbiota of free‐ranging barn swallows (Hirundo rustica) by studying co‐occurring migrant and resident subspecies sampled during the autumn migration at a migratory bottleneck. We found that within‐host microbial richness (α‐diversity) was similar between migrant and resident microbial communities. In contrast, we found that microbial communities (β‐diversity) were significantly different between groups regarding both microbes present and their relative abundances. Compositional differences were found for 36 bacterial genera, with 27 exhibiting greater abundance in migrants and nine exhibiting greater abundance in residents. There was heightened abundance ofMycoplasmaspp. andCorynebacteriumspp. in migrants, a pattern shared by other studies of migratory species. Screens for key regional pathogens revealed that neither residents nor migrants carried avian influenza viruses and Newcastle disease virus, suggesting that the status of these diseases did not underlie observed differences in microbiome composition. Furthermore, the prevalence and abundance ofSalmonellaspp., as determined from microbiome data and cultural assays, were both low and similar across the groups. Overall, our results indicate that microbial composition differs between migratory and resident barn swallows, even when they are conspecific and sympatrically occurring. Differences in host origins (breeding sites) may result in microbial community divergence, and varied behaviours throughout the annual cycle (e.g., migration) could further differentiate compositional structure as it relates to functional needs.

The behavioural ecology of host species is likely to affect their microbial communities, because host sex, diet, physiology, and movement behaviour could all potentially influence their microbiota. We studied a wild population of barn owls (Tyto alba) and collected data on their microbiota, movement, diet, size, coloration, and reproduction. The composition of bacterial species differed by the sex of the host and female owls had more diverse bacterial communities than their male counterparts. The abundance of two families of bacteria, Actinomycetaceae and Lactobacillaceae, also varied between the sexes, potentially as a result of sex differences in hormones and immunological function, as has previously been found with Lactobacillaceae in the microbiota of mice. Male and female owls did not differ in the prey they brought to the nest, which suggests that dietary differences are unlikely to underlie the differences in their microbiota. The movement behaviour of the owls was associated with the host microbiota in both males and females because owls that moved further from their nest each day had more diverse bacterial communities than owls that stayed closer to their nests. This novel result suggests that the movement ecology of hosts can impact their microbiota, potentially on the basis of their differential encounters with new bacterial species as the hosts move and forage across the landscape. Overall, we found that many aspects of the microbial community are correlated with the behavioural ecology of the host and that data on the microbiota can aid in generating new hypotheses about host behaviour.

Home range estimation is routine practice in ecological research. While advances in animal tracking technology have increased our capacity to collect data to support home range analysis, these same advances have also resulted in increasingly autocorrelated data. Consequently, the question of which home range estimator to use on modern, highly autocorrelated tracking data remains open. This question is particularly relevant given that most estimators assume independently sampled data. Here, we provide a comprehensive evaluation of the effects of autocorrelation on home range estimation. We base our study on an extensive data set ofGPSlocations from 369 individuals representing 27 species distributed across five continents. We first assemble a broad array of home range estimators, including Kernel Density Estimation (KDE) with four bandwidth optimizers (Gaussian reference function, autocorrelated‐Gaussian reference function [AKDE], Silverman's rule of thumb, and least squares cross‐validation), Minimum Convex Polygon, and Local Convex Hull methods. Notably, all of these estimators exceptAKDEassume independent and identically distributed (IID) data. We then employ half‐sample cross‐validation to objectively quantify estimator performance, and the recently introduced effective sample size for home range area estimation () to quantify the information content of each data set. We found thatAKDE95% area estimates were larger than conventionalIID‐based estimates by a mean factor of 2. The median number of cross‐validated locations included in the hold‐out sets byAKDE95% (or 50%) estimates was 95.3% (or 50.1%), confirming the largerAKDEranges were appropriately selective at the specified quantile. Conversely, conventional estimates exhibited negative bias that increased with decreasing . To contextualize our empirical results, we performed a detailed simulation study to tease apart how sampling frequency, sampling duration, and the focal animal's movement conspire to affect range estimates. Paralleling our empirical results, the simulation study demonstrated thatAKDEwas generally more accurate than conventional methods, particularly for small . While 72% of the 369 empirical data sets had >1,000 total observations, only 4% had an >1,000, where 30% had an <30. In this frequently encountered scenario of small ,AKDEwas the only estimator capable of producing an accurate home range estimate on autocorrelated data.