Search for: All records

ABSTRACT AimHalting widespread biodiversity loss will require detailed information on species' trends and the habitat conditions correlated with population declines. However, constraints on conventional monitoring programs and commonplace approaches for trend estimation can make it difficult to obtain such information across species' ranges. Here, we demonstrate how recent developments in machine learning and model interpretation, combined with data sources derived from participatory science, enable landscape‐scale inferences on the habitat correlates of population trends across broad spatial extents. LocationWorldwide, with a case study in the western United States. MethodsWe used interpretable machine learning to understand the relationships between land cover and spatially explicit bird population trends. Using a case study with three passerine birds in the western U.S. and spatially explicit trends derived from eBird data, we explore the potential impacts of simulated land cover modification while evaluating potential co‐benefits among species. ResultsOur analysis revealed complex, non‐linear relationships between land cover variables and species' population trends as well as substantial interspecific variation in those relationships. Areas with the most positive impacts from a simulated land cover modification overlapped for two species, but these changes had little effect on the third species. Main ConclusionsThis framework can help conservation practitioners identify important relationships between species trends and habitat while also highlighting areas where potential modifications to the landscape could bring the biggest benefits. The analysis is transferable to hundreds of species worldwide with spatially explicit trend estimates, allowing inference across multiple species at scales that are tractable for management to combat species declines. 

Abstract While the quantity, quality, and variety of movement data has increased, methods that jointly allow for population- and species-level movement parameters to be estimated are still needed. We present a formal data integration approach to combine individual-level movement and population-level distribution data. We show how formal data integration can be used to improve precision of individual and population level movement parameters and allow additional population level metrics (e.g., connectivity) to be formally quantified.We describe three components needed for an Integrated Movement Model (IMM): a model for individual movement, a model for among-individual heterogeneity, and a model to quantify changes in species distribution. We outline a general IMM framework and develop and apply a specific stochastic differential equation model to a case study of telemetry and species distribution data for golden eagles in western North American during spring migration.We estimate eagle movements during spring migration from data collected between 2011 and 2019. Individual heterogeneity in migration behavior was modeled for two sub-populations, individuals that make significant northward migrations and those that remained in the southern Rocky Mountain region through the summer. As is the case with most tracking studies, the sample population of individual telemetered birds is not representative of the population, and underrepresents the proportion of long-distance migrants in. The IMM was able to provide a more biological accurate subpopulation structure by jointly estimating the structure using the species distribution data. In addition, the integrated approach a) improves accuracy of other estimated movement parameters, b) allows us to estimate the proportion of migratory and non-migratory birds in a given location and time, and c) estimate future spatio-temporal distributions of birds given a wintering location, which provide estimates of seasonal connectivity and migratory routes.We demonstrate how IMMs can be successfully used to address the challenge of estimating accurate population level movement parameters. Our approach can be generalized to a broad range of available movement models and data types, allowing us to significantly improve our knowledge of migration ecology across taxonomic groups, and address population and continental level information needs for conservation and management. 

Abstract An occupancy model makes use of data that are structured as sets of repeated visits to each of many sites, in order to estimate the actual probability of occupancy (i.e. proportion of occupied sites) after correcting for imperfect detection using the information contained in the sets of repeated observations. We explore the conditions under which preexisting, volunteer-collected data from the citizen science project eBird can be used for fitting occupancy models. Because the majority of eBird’s data are not collected in the form of repeated observations at individual locations, we explore 2 ways in which the single-visit records could be used in occupancy models. First, we assess the potential for space-for-time substitution: aggregating single-visit records from different locations within a region into pseudo-repeat visits. On average, eBird’s observers did not make their observations at locations that were representative of the habitat in the surrounding area, which would lead to biased estimates of occupancy probabilities when using space-for-time substitution. Thus, the use of space-for-time substitution is not always appropriate. Second, we explored the utility of including data from single-visit records to supplement sets of repeated-visit data. In a simulation study we found that inclusion of single-visit records increased the precision of occupancy estimates, but only when detection probabilities are high. When detection probability was low, the addition of single-visit records exacerbated biases in estimates of occupancy probability. We conclude that subsets of data from eBird, and likely from similar projects, can be used for occupancy modeling either using space-for-time substitution or supplementing repeated-visit data with data from single-visit records. The appropriateness of either alternative will depend on the goals of a study and on the probabilities of detection and occupancy of the species of interest. 

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 a Deep‐Reasoning Network implementation of the Deep Multivariate Probit Model (DMVP‐DRNets), an end‐to‐end deep neural network framework, to exploit large observational and environmental data sets 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 over nine million 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. 

Abstract Population size is a key metric for management and policy decisions, yet wildlife monitoring programmes are often limited by the spatial and temporal scope of surveys. In these cases, citizen science data may provide complementary information at higher resolution and greater extent.We present a case study demonstrating how data from the eBird citizen science programme can be combined with regional monitoring efforts by the US Fish and Wildlife Service to produce high‐resolution estimates of golden eagle abundance. We developed a model that uses aerial survey data from the western United States to calibrate high‐resolution annual estimates of relative abundance from eBird. Using this model, we compared regional population size estimates based on the calibrated eBird information with those based on aerial survey data alone.Population size estimates based on the calibrated eBird information had strong correspondence to estimates from aerial survey data in two out of four regions, and population trajectories based on the two approaches showed high correlations.We demonstrate how the combination of citizen science data and targeted surveys can be used to (a) increase the spatial resolution of population size estimates, (b) extend the spatial extent of inference and (c) predict population size beyond the temporal period of surveys. Findings based on this case study can be used to refine policy metrics used by the US Fish and Wildlife Service and inform permitting regulations (e.g. mortality/harm associated with wind energy development).Policy implications: Our results demonstrate the ability of citizen science data to complement targeted monitoring programmes and improve the efficacy of decision frameworks that require information on population size or trajectory. After validating citizen science data against survey‐based benchmarks, agencies can harness strengths of citizen science data to supplement information needs and increase the resolution and extent of population size predictions. 

Abstract The research and conservation community has successfully harnessed the wealth of ecological knowledge found in unprecedented volumes of citizen science (CS) data world‐wide. However, few examples exist of the use of CS data to directly inform policy.Current examples of applications of CS data mainly stem from programs that are restricted in scope (e.g. defined protocols, restricted sampling time frame), and the potential use of unrestricted CS data to inform policy remains largely untapped.Here, we make a call for moving beyond questioning the reliability of CS data and present a case study of how the US Fish and Wildlife Service (USFWS) used information from an unrestricted CS program (eBird) to inform levels of exposure to collision risk for wind energy development.Policy implications. The USFWS made the technical recommendation to use eBird abundance estimates for the bald eagle as the only source of information to define low‐risk collision areas as part of the agency's wind energy permitting process. Our study contributes a clear pathway of how to realize the potential of unrestricted CS programs for generating the evidence base needed to inform policy decisions.