Biology subjects

Getz, W. M.

Publications and source records attributed to Getz, W. M..

7 recordsLinked to original sources

bioRxiv2018

A comparison of visual features used by humans and machines to classify wildlife

In our quest to develop more intelligent machines, knowledge of the visual features used by machines to classify objects shall be helpful. The current state of the art in training machines to classify wildlife species from camera-trap data is to employ convolutional neural networks (CNN) encoded within deep learning algorithms. Here we report on 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 then used 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 are in most, but not all, cases 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, and we then interpret the features in the image where the strongest responses occur. 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 how images that were not part of the training set fell within our dendrogram when these images were one of the 20 species \"known\" to our CNN in contrast to where they fell when these images were \"unknown\" to our CNN.

ecology↗

bioRxiv2018

The global distribution of Bacillus anthracis and associated anthrax risk to humans, livestock, and wildlife

Bacillus anthracis is a spore-forming, Gram-positive bacterium responsible for anthrax, an acute and commonly lethal infection that most significantly affects grazing livestock, wild ungulates and other herbivorous mammals, but also poses a serious threat to human health1, 2. The geographic extent of B. anthracis endemism is still poorly understood, despite multi-decade research on anthrax epizootic and epidemic dynamics around the world3, 4. Several biogeographic studies have focused on modeling environmental suitability for anthrax at local or national scales5-9, but many countries have limited or inadequate surveillance systems, even within known endemic regions. Here we compile an extensive global occurrence dataset for B. anthracis, drawing on confirmed human, livestock, and wildlife anthrax outbreaks. With these records, we use boosted regression trees10, 11 to produce the first map of the global distribution of B. anthracis as a proxy for anthrax risk. Variable contributions to the model support pre-existing hypotheses that environmental suitability for B. anthracis depends most strongly on soil characteristics such as pH that affect spore persistence, and the extent of seasonal fluctuations in vegetation, which plays a key role in transmission for herbivores12, 13. We apply the global model to estimate that 1.83 billion people (95% credible interval: 0.59--4.16 billion) live within regions of anthrax risk, but most of that population faces little occupational exposure to anthrax. More informatively, a global total of 63.8 million rural poor livestock keepers (95% CI: 17.5--168.6 million) and 1.1 billion livestock (95% CI: 0.4--2.3 billion) live within vulnerable regions. Human risk is concentrated in rural areas, and human and livestock vulnerability are both concentrated in rainfed systems throughout arid and temperate land across Eurasia, Africa, and North America. We conclude by mapping where anthrax risk overlaps with vulnerable wild ungulate populations, and therefore could disrupt sensitive conservation efforts for species like bison, pronghorn, and saiga that coincide with anthrax-prone, mixed-agricultural landscapes. Anthrax is a zoonotic disease caused by the Gram-positive bacterium Bacillus anthracis, a generalist soil-transmitted pathogen found on every inhabited continent14, and several islands including Haiti and parts of the Philippines and Indonesia. Worldwide, an estimated 20,000 to 100,000 cases of anthrax occur annually, mostly in poor rural areas15. In clinical presentations of anthrax, case fatality rates are a function of exposure pathway. Respiratory exposure from spore inhalation is important the context of bioterrorism, but is highly uncommon, and accounts for a negligible fraction of the global burden of anthrax cases. Cutaneous exposure to B. anthracis accounts for the majority of human cases worldwide, and typically presents with low mortality; gastrointestinal exposure accounts for the remainder and presents with intermediate to high fatality rates. Cutaneous and gastrointestinal cases of anthrax are most commonly caused by handling and slaughtering infected livestock, or butchering and eating contaminated meat; untreated gastrointestinal cases likely account for most human mortality from anthrax.14-16

epidemiology↗

bioRxiv2018

Using movement data to estimate contact rates in a simulated environmentally-transmitted disease system

Agent-based models have become important tools in ecology, particularly in the study of infectious disease dynamics. Simulations of near-continuous movement paths guided by empirical data offer new avenues of investigation into disease transmission. Here, we simulate the spatiotemporal transmission dynamics of anthrax, the acute disease caused by the bacterium Bacillus anthracis, a pathogen transmitted primarily via environmental reservoirs. We explore how calculations of the probabilities of contact between a host and infectious reservoirs are affected by the scale and method of analysis. At both the landscape and individual scales, empirical movement tracks offer previously unattainable estimates of impacts of movement decisions on contact rate metrics. However, the analytical method selected for the calculation of the probability of contact has notable impacts on the resulting estimates, with convex polygons virtually canceling out variation, and unions of local convex hulls (LoCoH methods) and space-time prisms reflecting reasonable variation, but differing in the magnitude of their estimates. The explicit consideration of behavioral states along movement pathways also impacts evaluations of exposure risk, though its effects differ across methods of analysis. Ultimately, simulations demonstrate that the incorporation of movement data into pathogen transmission analyses helps clarify the role of movement processes underlying the observed dynamics of infectious disease.

ecology↗

bioRxiv2017

Going through the motions: incorporating movement analyses into disease research

Though epidemiology dates back to the 1700s, most mathematical representations of epidemics still use transmission rates averaged at the population scale, especially for wildlife diseases. In simplifying the contact process, we ignore the heterogeneities in host movements that complicate the real world, and overlook their impact on spatiotemporal patterns of disease burden. Movement ecology offers a set of tools that help unpack the transmission process, letting researchers more accurately model how animals within a population interact and spread pathogens. Analytical techniques from this growing field can also help expose the reverse process: how infection impacts movement behaviors, and therefore other ecological processes like feeding, reproduction, and dispersal. Here, we synthesize the contributions of movement ecology in disease research, with a particular focus on studies that have successfully used movement-based methods to quantify individual heterogeneity in exposure and transmission risk. Throughout, we highlight the rapid growth of both disease and movement ecology, and comment on promising but unexplored avenues for research at their overlap. Ultimately, we suggest, including movement empowers ecologists to pose new questions expanding our understanding of host-pathogen dynamics, and improving our predictive capacity for wildlife and even human diseases.

ecology↗

bioRxiv2017

Modeling Epidemics: A Primer and Numerus Software Implementation

Epidemiological models are dominated by SEIR (Susceptible, Exposed, Infected and Removed) dynamical systems formulations and their elaborations. These formulations can be continuous or discrete, deterministic or stochastic, or spatially homogeneous or heterogeneous, the latter often embracing a network formulation. Here we review the continuous and discrete deterministic and discrete stochastic formulations of the SEIR dynamical systems models, and we outline how they can be easily and rapidly constructed using the Numerus Model Builder, a graphically-driven coding platform. We also demonstrate how to extend these models to a metapopulation setting using both the Numerus Model Builder network and geographical mapping tools.

epidemiology↗

bioRxiv2017

The Parasite Extinction Assessment & Red List: an open-source, online biodiversity database for neglected symbionts

Parasite conservation is a rapidly growing field at the intersection of ecology, epidemiology, parasitology, and public health. The overwhelming diversity of parasitic life on earth, and recent work showing that parasites and other symbionts face severe extinction risk, necessitates infrastructure for parasite conservation assessments. Here, we describe the release of the Parasite Extinction Assessment & Red List (PEARL) version 1.0, an open-access database of conservation assessments and distributional data for almost 500 macroparasitic invertebrates. The current approach to vulnerability assessment is based on range shifts and loss from climate change, and will be expanded as additional data (e.g., host-parasite associations and coextinction risk) is consolidated in PEARL. The web architecture is also open-source, scalable, and extensible, making PEARL a template for more eZcient red listing for other high-diversity, data-de1cient groups. Future iterations will also include new functionality, including a user-friendly open data pository and automated assessment and re-listing.

ecology↗

bioRxiv2017

A cross-validation-based approach for delimiting reliable home range estimates

BackgroundWith decreasing costs of GPS telemetry devices, data repositories of animal movement paths are increasing almost exponentially in size. A series of complex statistical tools have been developed in conjunction with this increase in data. Each of these methods offers certain improvements over previously proposed methods, but each has certain assumptions or shortcomings that make its general application difficult. In the case of the recently developed Time Local Convex Hull (T-LoCoH) method, the subjectivity in parameter selection serves as one of the primary impediments to its more widespread use. While there are certain advantages to the flexibility it offers for question-driven research, the lack of an objective approach for parameter selection may prevent some users from exploring the benefits of the method.\n\nResultsHere we present a cross-validation-based approach for selecting parameter values to optimize the T-LoCoH algorithm. We demonstrate the utility of the approach using a case study from the Etosha National Park anthrax system. Utilizing the proposed algorithm, rather than the guidelines in the T-LoCoH documentation, results in significantly different values for derived site fidelity metrics.\n\nConclusionsDue to its basis in principles of cross-validation, the application of this method offers a more objective approach than the relatively subjective guidelines set forth in the T-LoCoH documentation and enables a more accurate basis for the comparison of home ranges among individuals and species, as well as among studies.

ecology↗