Big data-big problems? How to circumvent problems in biodiversity mapping and ensure meaningful results

Abstract

Our knowledge of biodiversity hinges on sufficient data, reliable methods, and realistic models. Without an accurate assessment of species distributions, we cannot effectively target and stem biodiversity loss. Species range maps are the foundation of such efforts, but countless studies have failed to account for the most basic assumptions of reliable species mapping practices, undermining the credibility of their results and potentially misleading and hindering conservation and management efforts. Here, we use examples from the recent literature and broader conservation community to highlight the substantial shortfalls in current practices and their consequences for both analyses and conservation management. We detail how different decisions on data filtering impact the outcomes of analysis and provide practical recommendations and steps for more reliable analysis, whilst understanding the limits of what available data will reliably allow and what methods are most appropriate. Whilst “perfect” analyses are not possible for many taxa given limited data, and biases, ensuring we use data within reasonable limits and understanding inherent assumptions is crucial to ensure appropriate use. By embracing and enacting such best practices, we can ensure both the accuracy and improved comparability of biodiversity analyses going forward, ultimately enhancing our ability to use data to facilitate our protection of the natural world.

Optimising Species Distribution Models: Sample size, positional error, and sampling bias matter

“…The quantity and quality of biological observations has improved dramatically over the past decades. However, a certain level of uncertainties is inherently present in such data, resulting in uncertainties of scientific inferences based on it (Hortal et al 2015;Daru and Rodriguez 2023;Hughes et al 2023). Correlative species distribution models (SDMs; aka habitat suitability models or ecological niche models; Sillero 2011) are useful for tackling the gaps in our knowledge of species occurrence (Elith and Leathwick 2009).…”