Marine protected areas (MPAs) can be an effective tool for marine biodiversity conservation, yet decision-makers usually have limited and biased datasets with which to make decisions about where to locate MPAs. Using commonly available abiotic and biotic datasets, I asked how many datasets are necessary to achieve robust patterns of conservation importance. I applied a decision support tool for marine protected area design in two regions of British Columbia, Canada, and sequentially excluded the datasets with the most limited geographic distribution. I found that the reserve selection method was robust to some missing datasets. The removal of up to 15 of the most geographically limited datasets did not significantly change the geographic patterns of the importance of areas for conservation. Indeed, including abiotic datasets plus at least 12 biotic datasets resulted in a spatial pattern similar to including all available biotic datasets. It was best to combine abiotic and biotic datasets in order to ensure habitats and species were represented. Patterns of clustering differed according to whether I used one set alone or both combined. Biotic datasets served as better surrogates for abiotic datasets than vice versa, and both represented more biodiversity features than randomly selected reserves. These results should provide encouragement to decision-makers engaged in MPA planning with limited spatial data.