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

Aim:
Undersampling 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.
Location:
Atlantic Ocean.
Time Period:
Present.
Taxa Studied: Ophiuroidea.
Methods:
We 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.
Results:
We 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