Pither & Aarssen (2005) propose a null model approach to assess the proportion of niche specialist taxa along ecological gradients. They apply this methodology to a large data set of lacustrine diatom assemblages and conclude that a majority of the taxa are generalists on a pH gradient. This conflicts with previous work, which shows that many diatom taxa have a statistically significant relationship with pH. We demonstrate the methods used by Pither & Aarssen (2005) have a high Type II error for rare taxa, and that this problem is compounded by the non-uniform sampling of the pH gradient which effectively precludes acid-lake specialist diatoms from being recognized as such. We re-analyse the data used by Pither & Aarssen (2005) and show that most of the diatoms have a statistically significant relationship with pH, and we thus refute their conclusions that few diatom species are specialists.