In a wide-ranging and stimulating TREE review, Reid[1xReid, W.V. Trends Ecol. Evol. 1998; 13: 275–280Abstract | Full Text | Full Text PDF | PubMed | Scopus (305)See all References[1]recently concluded that focusing on a handful of readily surveyed indicator groups could be of limited use in identifying priorities for conserving biodiversity as a whole. His reason was straightforward: a suite of studies[2xPrendergast, J.R. et al. Nature. 1993; 365: 335–337CrossrefSee all References, 3xLombard, A.T. S. Afr. J. Zool. 1995; 30: 145–163See all References, 4xDobson, A.P. et al. Science. 1997; 275: 550–553Crossref | PubMed | Scopus (396)See all References, 5xFlather, C.H. et al. Ecol. Appl. 1997; 7: 531–542CrossrefSee all References]has now shown that, at the relatively fine scales relevant to reserve planning, there is often little correspondence in the richest sites for different groups (see also Refs [6xGaston, K.J. Prog. Phys. Geogr. 1996; 20: 105–112CrossrefSee all References[6]and [7xvan Jaarsveld, A.S. et al. Science. 1998; 279: 2106–2108Crossref | PubMed | Scopus (242)See all References[7]). Hotspots—defined variously as the richest sites for all species, or for endemic, rare or threatened species—seldom coincide across taxa.More importantly (and not explicitly discussed by Reid), other studies[4xDobson, A.P. et al. Science. 1997; 275: 550–553Crossref | PubMed | Scopus (396)See all References, 7xvan Jaarsveld, A.S. et al. Science. 1998; 279: 2106–2108Crossref | PubMed | Scopus (242)See all References]show that sets of priority sites identified by considering between-site complementarity, as well as species richness (which therefore represent diversity far more efficiently than do hotspots[8xPressey, R.L. et al. Trends Ecol. Evol. 1993; 8: 121–128Abstract | Full Text PDF | Scopus (655)See all References, 9xWilliams, P.H. et al. Conserv. Biol. 1996; 10: 155–174CrossrefSee all References]), also show poor cross-taxon overlap. However, although high congruence in the most important sites for conserving different kinds of organisms would be extremely useful for conservation planning (and fascinating from an evolutionary perspective), its absence is not sufficient justification for rejecting the use of indicators for reserve selection. This is because whether the priority sites for one group embrace priority areas for other taxa is fundamentally less important than the extent to which they capture species from those taxa.What really matters in testing the use of indicators for reserve selection is the representation of diversity as a whole in the key sites for just one focal group. Towards the end of his review, Reid mentions unpublished results from Oregon (referred to in Ref. [10xCsuti, B. et al. Biol. Conserv. 1997; 80: 83–97Crossref | Scopus (298)See all References[10]) showing that cross-taxon capture of species by the complementary sites for single taxa is actually rather high. Another recent review cites similar evidence from other studies[5xFlather, C.H. et al. Ecol. Appl. 1997; 7: 531–542CrossrefSee all References[5]. An extensive new analysis across 50 forest reserves in Uganda confirms these findings and provides an explanation for them, which in turn suggests that, in some areas, the use of indicators for reserve selection holds more promise than Reid seems to imply[11xSee all References[11].Working on a dataset covering 2452 species of woody plants, large moths, butterflies, birds and small mammals that took nearly 100 man-years of survey effort to compile, Howard and colleagues[11xSee all References[11]first showed that in Uganda, as elsewhere, patterns of species richness (controlling for forest area) show poor congruence across groups. Nevertheless, complementarity-based sets of priority sites identified using data on single taxa capture species richness across all groups with roughly the same efficiency as sets built up using information on all taxa; data on birds and butterflies perform particularly well. The reason is that Uganda is environmentally quite heterogeneous, and each taxon exhibits similar biogeographical patterns across the country. Consequently, pairs of forests that complement each other well for one taxon do so for other groups too, and a priority set for one taxon embraces all the main habitat types for other groups. In this case, lack of cross-taxon congruence in species richness is not enough to invalidate the use of indicators because of high congruence in complementarity.How general are these results likely to be? Other studies have shown that complementary areas for focal taxa sometimes capture diversity in other groups rather less well than in Uganda, but these deal with temperate situations where the focal taxa themselves are relatively species-poor[12xRyti, R.T. Ecol. Appl. 1992; 2: 404–410CrossrefSee all References[12], or contain far fewer species than there are candidate sites for their conservation[4xDobson, A.P. et al. Science. 1997; 275: 550–553Crossref | PubMed | Scopus (396)See all References[4]. Further work is now needed to assess whether the Uganda results hold for less heterogeneous parts of the tropics, where high species richness coupled with very limited funds for biodiversity surveys suggest that the payoffs of using indicators for reserve selection would be greatest.