Search for: All records

Abstract Ecological forecasting models play an increasingly important role for managing natural resources and assessing our fundamental knowledge of processes driving ecological dynamics. As global environmental change pushes ecosystems beyond their historical conditions, the utility of these models may depend on their transferability to novel conditions. Because species interactions can alter resource use, timing of reproduction, and other aspects of a species' realized niche, changes in biotic conditions, which can arise from community reorganization events in response to environmental change, have the potential to impact model transferability. Using a long‐term experiment on desert rodents, we assessed model transferability under novel biotic conditions to better understand the limitations of ecological forecasting. We show that ecological forecasts can be less accurate when the models generating them are transferred to novel biotic conditions and that the extent of model transferability can depend on the species being forecast. We also demonstrate the importance of incorporating uncertainty into forecast evaluation with transferred models generating less accurate and more uncertain forecasts. These results suggest that how a species perceives its competitive landscape can influence model transferability and that when uncertainties are properly accounted for, transferred models may still be appropriate for decision making. Assessing the extent of the transferability of forecasting models is a crucial step to increase our understanding of the limitations of ecological forecasts. 

Wildlife population monitoring over large geographic areas is increasingly feasible due to developments in aerial survey methods coupled with the use of computer vision models for identifying and classifying individual organisms. However, aerial surveys still occur infrequently, and there are often long delays between the acquisition of airborne imagery and its conversion into population monitoring data. Near real‐time monitoring is increasingly important for active management decisions and ecological forecasting. Accomplishing this over large scales requires a combination of airborne imagery, computer vision models to process imagery into information on individual organisms, and automated workflows to ensure that imagery is quickly processed into data following acquisition. Here we present our end‐to‐end workflow for conducting near real‐time monitoring of wading birds in the Everglades, Florida, USA. Imagery is acquired as frequently as weekly using uncrewed aircraft systems (aka drones), processed into orthomosaics (using Agisoft metashape), converted into individual‐level species data using a Retinanet‐50 object detector, post‐processed, archived, and presented on a web‐based visualization platform (using Shiny). The main components of the workflow are automated using Snakemake. The underlying computer vision model provides accurate object detection, species classification, and both total and species‐level counts for five out of six target species (White Ibis, Great Egret, Great Blue Heron, Wood Stork, and Roseate Spoonbill). The model performed poorly for Snowy Egrets due to the small number of labels and difficulty distinguishing them from White Ibis (the most abundant species). By automating the post‐survey processing, data on the populations of these species is available in near real‐time (<1 week from the date of the survey) providing information at the time scales needed for ecological forecasting and active management. 

Abstract Regional long‐term monitoring can enhance the detection of biodiversity declines associated with climate change, improving future projections by reducing reliance on space‐for‐time substitution and increasing scalability. Rodents are diverse and important consumers in drylands, regions defined by the scarcity of water that cover 45% of Earth's land surface and face increasingly drier and more variable climates. We analyzed abundance data for 22 rodent species across grassland, shrubland, ecotone, and woodland ecosystems in the southwestern USA. Two time series (1995–2006 and 2004–2013) coincided with phases of the Pacific Decadal Oscillation (PDO), which influences drought in southwestern North America. Regionally, rodent species diversity declined 20%–35%, with greater losses during the later time period. Abundance also declined regionally, but only during 2004–2013, with losses of 5% of animals captured. During the first time series (wetter climate), plant productivity outranked climate variables as the best regional predictor of rodent abundance for 70% of taxa, whereas during the second period (drier climate), climate best explained variation in abundance for 60% of taxa. Temporal dynamics in diversity and abundance differed spatially among ecosystems, with the largest declines in woodlands and shrublands of central New Mexico and Colorado. Which species were winners or losers under increasing drought and amplified interannual variability in drought depended on ecosystem type and the phase of the PDO. Fewer taxa were significant winners (18%) than losers (30%) under drought, but the identities of winners and losers differed among ecosystems for 70% of taxa. Our results suggest that the sensitivities of rodent species to climate contributed to regional declines in diversity and abundance during 1995–2013. Whether these changes portend future declines in drought‐sensitive consumers in the southwestern USA will depend on the climate during the next major PDO cycle.