Recent studies suggest that species with similar functional traits will have similar effects on ecosystems, but evidence for redundancy of species impacts is limited. Here we use a long-term experiment to gain insight into functional relationships within a desert rodent community. Experimental removal of kangaroo rats, Dipodomys spp., coupled with the recent, serendipitous colonization of a single species of large pocket mouse Chaetodipus baileyi yielded treatments that differed in the diversity of large granivorous rodents present. We evaluated functional overlap of C. baileyi and the other resident large granivores (i.e. the kangaroo rats) by comparing total energy use of granivorous rodents and total abundance and species richness of small granivores across treatments before and after the arrival of C. baileyi. We found that C. baileyi almost completely compensated for the changes in these key ecosystem-level properties caused by kangaroo rat removal, but it differentially impacted the population dynamics of individual small granivorous rodent species. Thus, its effects were largely complementary, rather than redundant, to those of the missing kangaroo rats. Although short-term or single-measure analyses may suggest redundancy, our results support the longstanding dictum that niches of coexisting species are often similar but rarely, if ever, identical.