The competition–relatedness hypothesis of Darwin states that competition is greater among species that are phylogenetically closely‐related, and such species will tend to appear in separate communities (i.e. the species within communities will be phylogenetically overdispersed). Many studies have tested (and mainly refuted) this hypothesis for plant and bacterial communities. Results for the few studies with avian species are not conclusive. We tested Darwin's hypothesis for waterbirds using a set of open, artificial fish ponds in Doñana, south‐western Spain, that provide relatively homogeneous habitat where competition is likely to be intense. Monthly counts of 38 ponds (for 11 months, i.e. 418 censuses) recorded 76 bird species. Darwin's hypothesis predicted that species appearing in the same pond would be less related phylogenetically than expected if species occurred randomly across ponds and months according to the structure of the overall community across the entire pond complex. However, the waterbird community did not show a predominantly overdispersed pattern, suggesting that interspecific competition among phylogenetically related species was not the main force structuring communities. In contrast, the proportion of clustered communities was higher than expected throughout the annual cycle, indicating that related waterbirds tend to co‐occur on the same site, probably because they have similar microhabitat preferences. Clustering patterns were mainly driven by abundant and closely related duck species, and also by shorebirds. However, few individual pond communities remained significantly different from random after correction for multiple testing. Furthermore, the probability of co‐occurrence of a given species pair was negatively related to the phylogenetic distance between them. In conclusion, our study shows waterbird communities are mainly phylogenetically clustered or random, and do not support the competition‐relatedness hypothesis.