Deep learning with citizen science data enables estimation of species diversity and composition at continental extents

Abstract

Effective solutions to conserve biodiversity require accurate community and species‐level information at relevant, actionable scales and across entire species' distributions. However, data and methodological constraints have limited our ability to provide such information in robust ways. Herein we employ DMVP‐DRNets, an end‐to‐end deep neural network framework, to exploit large observational and environmental datasets together and estimate landscape‐scale species diversity and composition at continental extents. We present results from a novel year‐round analysis of North American avifauna using data from 9 M eBird checklists and 72 environmental covariates. We highlight the utility of our information by identifying critical areas of high species diversity for a single group of conservation concern, the North American wood warblers, while capturing spatiotemporal variation in species' environmental associations and interspecific interactions. In so doing, we demonstrate the type of accurate, high‐resolution information on biodiversity that deep learning approaches such as DMVP‐DRNets can provide and that is needed to inform ecological research and conservation decision‐making at multiple scales.

Combining environmental DNA and remote sensing for efficient, fine-scale mapping of arthropod biodiversity

“…It is natural to think about combining novel community data with copious environmental-covariate information in the form of continuous-space remote-imagery layers (and/or with continuous-time acoustic series) to produce continuous spatio(-temporal) biodiversity data products [9,30,[33][34][35][36][37][38][39][40]. Here, we do just this, combining a point-sample dataset of Malaise-trapped arthropods with continuous-space Landsat and LiDAR imagery within a joint species distribution model (JSDM [40][41][42][43]). We were able to produce distribution maps for 76 arthropod species across a forested landscape.…”

Combining environmental DNA and remote sensing for efficient, fine-scale mapping of arthropod biodiversity

“…It is natural to think about combining novel community data with copious environmental-covariate information in the form of continuous-space remote-imagery layers (and/or with continuous-time acoustic series) to produce continuous spatio(-temporal) biodiversity data products (Bush et al, 2017; He et al, 2015; Kwok, 2018; Leitão and Santos, 2019; Lin et al, 2021; Pettorelli et al, 2018; Cavender-Bares et al, 2022; Hartig et al, 2023; Müller et al, 2023; Davis et al, 2023). Here we do just this, combining a point-sample dataset of Malaise-trapped arthropods with continuous-space Landsat and lidar imagery within a joint species distribution model (JSDM Ovaskainen and Abrego, 2020; Pichler and Hartig, 2021; Warton et al, 2015; Davis et al, 2023). We were able to produce distribution maps for 76 arthropod species across a forested landscape.…”