Search

Conserving and managing biodiversity in the face of ongoing global change requires sufficient evidence to assess status and trends of species distributions. Here, we propose novel indicators of biodiversity data coverage and sampling effectiveness and analyze national trajectories in closing spatiotemporal knowledge gaps for terrestrial vertebrates (1950 to 2019). Despite a rapid rise in data coverage, particularly in the last 2 decades, strong geographic and taxonomic biases persist. For some taxa and regions, a tremendous growth in records failed to directly translate into newfound knowledge due to a sharp decline in sampling effectiveness. However, we found that a nation’s coverage was stronger for species for which it holds greater stewardship. As countries under the post-2020 Global Biodiversity Framework renew their commitments to an improved, rigorous biodiversity knowledge base, our findings highlight opportunities for international collaboration to close critical information gaps., Conserving and managing biodiversity in the face of ongoing global change requires sufficient evidence to assess status and trends of species distributions. This study analyzes national trajectories in closing spatiotemporal knowledge gaps for terrestrial vertebrates (1950-2019) based on novel indicators of data coverage and sampling effectiveness.

Conservation science depends on an accurate understanding of what's happening in a given ecosystem. How many species live there? What is the makeup of the population? How is that changing over time? Species Distribution Modeling (SDM) seeks to predict the spatial (and sometimes temporal) patterns of species occurrence, i.e. where a species is likely to be found. The last few years have seen a surge of interest in applying powerful machine learning tools to challenging problems in ecology. Despite its considerable importance, SDM has received relatively little attention from the computer science community. Our goal in this work is to provide computer scientists with the necessary background to read the SDM literature and develop ecologically useful ML-based SDM algorithms. In particular, we introduce key SDM concepts and terminology, review standard models, discuss data availability, and highlight technical challenges and pitfalls., Comment: ACM COMPASS 2021

Conserving and managing biodiversity in the face of ongoing global change requires sufficient evidence to assess status and trends of species distributions. Here we analyze national trajectories in closing spatiotemporal knowledge gaps for terrestrial vertebrates (1950-2019) based on novel indicators of data coverage and sampling effectiveness. Despite a rapid rise in data coverage, particularly in the last two decades, strong geographic and taxonomic biases persist. For some taxa and regions, a tremendous growth in records failed to directly translate into newfound knowledge due to a sharp decline in sampling effectiveness. But nation’s coverage is stronger for species they hold greater stewardship for. As countries under the post-2020 Global Biodiversity Framework renew their commitments to an improved, rigorous biodiversity knowledge base, our findings highlight opportunities for international collaboration to close critical information gaps.

Large networks of weather radars are comprehensive instruments for studying bird migration. For example, the US WSR-88D network covers the entire continental US and has archived data since the 1990s. The data can quantify both broad and fine-scale bird movements to address a range of migration ecology questions. However, the problem of automatically discriminating precipitation from biology has significantly limited the ability to conduct biological analyses with historical radar data. We develop MistNet, a deep convolutional neural network to discriminate precipitation from biology in radar scans. Unlike prior machine learning approaches, MistNet makes fine-scaled predictions and can collect biological information from radar scans that also contain precipitation. MistNet is based on neural networks for images, and includes several architecture components tailored to the unique characteristics of radar data. To avoid a massive human labelling effort, we train MistNet using abundant noisy labels obtained from dual polarization radar data. In historical and contemporary WSR-88D data, MistNet identifies at least 95.9% of all biomass with a false discovery rate of 1.3%. Dual polarization training data and our radar-specific architecture components are effective. By retaining biomass that co-occurs with precipitation in a single radar scan, MistNet retains 15% more biomass than traditional whole-scan approaches to screening. MistNet is fully automated and can be applied to datasets of millions of radar scans to produce fine-grained predictions that enable a range of applications, from continent-scale mapping to local analysis of airspace usage. Radar ornithology is advancing rapidly and leading to significant discoveries about continent-scale patterns of bird movements. General-purpose and empirically validated methods to quantify biological signals in radar data are essential to the future development of this field. MistNet can enable large-scale, long-term, and reproducible measurements of whole migration systems.

Predictive models are used in many fields to characterize relationships between the attributes of an instance and its classification. While these models can provide valuable support to decision-making, they can be challenging to understand and evaluate. While they provide predicted classifications, they do not generally include indications of confidence in those predictions. Typical quality measures for predictive models are the percentage of predictions which are made correctly. These measures can give some insight into how often the model is correct, but provide little help in understanding under what conditions the model performs well (or poorly). We present a framework for improving understanding of predictive models based on the methods of both machine learning and data visualization. We demonstrate this framework on models that use attributes about individuals in a census data set to predict other attributes of those individuals.