Understanding biodiversity patterns over space and time requires using data collected across large scales. One way to collect these data is to compile databases of multiple independent studies. However, compiled databases often magnify problems with sampling bias, where some taxa, geographic regions, and ecological questions have been studied far more than others. In our study, we focus on parasites, a group that is very biodiverse but where sampling is very limited relative to true diversity and sampling bias can be extreme. We use simulations to compare the precision and bias of different methods to control for sampling effort in these databases; our results show that, without substantial increases in sampling effort, it is difficult to accurately estimate parasite diversity. Using these biodiversity estimators is more robust in comparative analyses than for creating raw estimates of parasite diversity, but they tend to underestimate effect sizes, which is important for interpreting the results of previous studies that compare parasite diversity across host taxa. Our results are also relevant beyond parasite diversity because any database compiled from multiple published studies inevitably contains substantial sampling bias.
Teitelbaum, C.S., Amoroso, C.R., Huang, S., Davies, T.J., Rushmore, J., Drake, J.M., Stephens, P.R., Byers, J.E., Majewska, A.A. and Nunn, C.L. (2020), A comparison of diversity estimators applied to a database of host–parasite associations. Ecography. doi:10.1111/ecog.05143