Sampling Simulation in a Virtual Ocean Reveals Strong Sampling Effect in Marine Diversity Patterns

Abstract

AimUndersampling and other sources of sampling bias pose significant issues in marine macroecology, particularly when shaping conservation and management decisions. Yet, determining the extent to which such biases impact our understanding of marine diversity remains elusive. Here, utilising empirical data on sampling efforts, we sampled from virtually established species distributions to evaluate how deep is the influence of sampling bias on estimations of the latitudinal gradient in marine diversity.LocationAtlantic Ocean.Time PeriodPresent.Taxa StudiedOphiuroidea.MethodsWe developed a computer simulation that implements two null models of species distribution (the geometric constraints and the area model) in a two‐dimensional domain, replicates the latitudinal distribution of historical sampling efforts and then quantifies diversity metrics (observed and estimated species richness) and sample completeness for each grid cell and latitudinal band.ResultsWe found consistent patterns of observed species richness across models, noting peaks at midlatitudes regardless of whether the true richness was unimodal or flat. Dips in equatorial diversity persisted even after using different methods of species richness estimation. Additional simulations showed that estimators' accuracy improved with increased sampling efforts, but only when samples were randomly distributed. Spatially aggregated samples inflate completeness without necessarily enhancing estimators' accuracy.Main ConclusionsThis finding emphasises the imperative of bolstering sampling efforts at tropical latitudes and deploying robust statistical techniques to mitigate undersampling effects. Meanwhile, we suggest considering sampling bias as an alternative null hypothesis for recorded marine diversity patterns.