Stenglein, J. | Search bioRxiv

bioRxiv2020

Integrating remote sensing and jurisdictional observation networks to improve the resolution of ecological management

The emergence of citizen science, passive sensors (e.g., trail cameras and acoustic monitoring), and satellite remote sensing have enabled biological data to be collected at unprecedented spatial and temporal scales. There is growing interest in networking these datastreams to expedite the collection and synthesis of environmental and biological data to improve broad-scale ecological monitoring, but there are no examples of such networks being developed to directly inform decision-making by managing agencies. Here, we present the implementation of one such jurisdictional observation network (JON), Snapshot Wisconsin (SW), that links satellite remote sensing (RS) with a volunteer-based trail camera network to generate new insights into wildlife distributions and improve their management by the state agency. SW relies on citizen scientists to deploy trail cameras across the state and classify images of wildlife. As of early 2020 SW comprises nearly 1800 volunteers hosting >2100 active cameras recording >37 million images across a sampling effort of >2000 combined trap-years at >3300 distinct camera locations. We use a set of case studies to demonstrate the potential power of a JON to monitor wildlife with unprecedented combinations of spatial, temporal, and biological resolution and extent. Specifically, we demonstrate that SW markedly improves the spatial and temporal resolution with which black bear distributions can be monitored or forecast, in turn improving the resolution of decision-making. Enhancing the biological resolution of monitoring (e.g., monitoring the distribution of species traits or behaviors) may provide new insights into population drivers, such as the connection between vegetation productivity and white-tailed deer foraging behaviors. Enhanced taxonomic extent provided by trail cameras and other passive sensor networks provide managers new information for a wide range of species and communities that are not otherwise monitored. Our cases further show that JONs synergize existing monitoring practices by serving as a complementary and independent line of evidence or as a tool to enhance the extent and precision of existing models through integrated modeling approaches. SW and other JONS are a powerful new tool for agencies to better achieve their missions and reshape the nature of environmental decision-making.

ecology↗

bioRxiv2020

Improving the accessibility and transferability of machine learning algorithms for identification of animals in camera trap images: MLWIC2

O_LIMotion-activated wildlife cameras (or "camera traps") are frequently used to remotely and non-invasively observe animals. The vast number of images collected from camera trap projects have 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. C_LIO_LIWe 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." C_LIO_LIOur 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 empty-animal model achieved an accuracy of 91-94% on out-of-sample datasets from different continents. C_LIO_LIOur 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 North America. We also found that our empty-animal model can facilitate removal of images without animals globally. We provide the trained models in an R package (MLWIC2: Machine Learning for Wildlife Image Classification in R), which contains Shiny Applications that allow scientists with minimal programming experience to use trained models and train new models in six neural network architectures with varying depths. C_LI

ecology↗