Bioversity data availability is growing, but data is affected by temporal, spatial and taxonomic biases. Understanding these and what causes them is important to preventing future bias and filling data gaps.
In this paper, researchers present a novel approach to quantifying sampling effort in Africa by exploring the influence of existing knowledge about a site on attracting researchers, while estimating the time needed to complete sampling.
Initially, the authors mapped GBIF-mediated occurrences of amphibians, mammals and birds to 100x100 km cells to identify the number of estimated sampling events by taxa and country.
By applying a logistic model between sampling completeness, events and year, they discovered that across most of Africa, previous sampling of amphibians and mammals strongly increased the likelihood of resampling. For bird sampling before 1940, however, there was a higher tendency to visit unsampled areas, indicating a preference for discovering new species.
Modelling sampling rate, the authors predicted that it would take between 171 and 273 years to obtain 90 per cent coverage. To record 50 per cent of species within a cell would take on average 11.5 visits for amphibians, 12.7 visits for mammals and 27 visits for birds.