GridDER: Grid Detection and Evaluation in R

Abstract

Observations and collections of organisms form the basis of our understanding of Earth's biodiversity and are an indispensable resource for global change studies. Geographic information is key, serving as the link between organisms and the environments they reside in. However, the geographic information associated with these records is often inaccurate, thus limiting their efficacy for research. While some tools for identifying erroneous coordinates have been developed, a prominent but less discussed source of inaccuracies arises from the use of gridded survey systems in many regions of the world. Here we present GridDER, a tool for identifying biodiversity records that have been designated locations based on widely used grid systems. Our tool also estimates the degree of environmental heterogeneity associated with grid systems, allowing users to make informed decisions about how to use such occurrence data in their research. We show that a significant proportion (∼13.5%; 261 million) of records on GBIF, the largest aggregator of natural history collection data, have potentially been assigned the centroid coordinates of grid cells of varying size, and demonstrate that our tool can reliably identify such records and quantify the associated uncertainties. GridDER can serve as a tool to not only screen for gridded points, but to quantify the geographic and environmental uncertainties associated with these records, which can be used to inform models and analyses that utilize these data, including those pertaining to global change.