Invertebrates, critical to life on earth and ecosystem functioning, are declining globally in response to human impacts but there is controversy about the size and scope of declines. Standardized, large-scale, long-term sampling, like that of the National Ecological Observatory Network (NEON), will allow us to quantify current and forecast future impacts on invertebrates. Invertebrate sampling at NEON sites focuses on Carabidae, a diverse family of ground-dwelling beetles. Our aim is to identify environmental variables that the beetle species are sensitive to and to forecast the effects of environmental change over a range of time scales. We use a Bayesian framework to account for multiple sources of uncertainty, including species misidentification. We use environmental and LiDAR-derived microhabitat variables to forecast changes in assemblage composition and abundance at a spatial scale relevant to carabid behavior and biology in subalpine forest at NEON Niwot Ridge.
Results/Conclusions
The misclassification modeling results allow us to take advantage of the entire NEON dataset of Niwot Ridge carabid assemblages with measured certainty in species identifications. All individuals are identified by a non-expert and a subset identified by an expert. Our results reconcile the nine species not matching between parataxonomist and expert taxonomist identifications and the nearly 500 individuals not identified at all by an expert. Using Bayesian inference, we found species-specific differences in abundance and composition of carabid assemblages between evergreen forest and tundra habitats. The fine spatial resolution of airborne remote sensing allows for habitat characterization at a scale appropriate for carabid biology, and simultaneously, its wide spatial extent allows for conservation managers to scale up these data to decision-making levels.