HOW RICH IS THE FLORAOF BRAZILIAN CERRADOS?'A. A. J. F. Castro,2 F. R. Martins,' J. Y.Tamashiro,: and G. J. Shepherd3ABSTRACT An attempt is made to summarize what is known about the richness of the total terrestrial angiosperm flora of the"cerrados" (as a complex of formations) in Brazil, based on published surveys and species lists. A "refined" lisi ofarboreal and shrubby species was compiled from a total of 145 individual lists from 78 localities, taking into accountsynonymy and recent taxonomic changes. The refined list had 1709 references to taxonomic entities at the species level(973 identified with confidence and 31 with aff. or cf:), 572 references to generic entities (363 genera identified withconfidence), and 210 references to the family level (88 families identified with confidence). There are many unidentifiedarboreal and shrubby taxa at the specific, generic, and family levels, indicating that a considerable amount of taxonomicresearch remains to be done on the cerrado flora, and that this flora may be much richer than is generally assumed.Depending on the assumptions made, these data suggest a total of around 1000 to 20(X)00 arboreal and shrubby speciesand 20(X) to 50(X) herbaceous ones, yielding estimates for the total cerrado flora (terrestrial angiosperms) ranging from30(X) to 70(X) species. These limits, especially the upper one, are dubious, but give an idea of the magnitude of theangiosperm flora in the Brazilian cerrados. Surveys of cerrados are very unevenly distributed, and studies of relativelyunknown sites may reveal much more diversity than that presently known. On reading accounts of floristic studies on cer-rados in Brazil, one rapidly cornes to realize thatthe majority of authors, either implicitly or explic-itly, consider the cerrado flora to be well known andto have low richness. For example, Rizzini (1963,1971) estimated around 600 species and a littleover 200 genera for the whole cerrado arboreal andshrubby flora, but Heringer et al. (1977) cited 193arboreal and shrubby species and confirmed lessthan 150 genera. Even recent studies (e.g., Leit�oFilho, 1992; Ratter et al., 1997) have estimated thenumber of arboreal-shrubby species for the cerra-dos as being around 800. Eiten (1990) has beenone of the few authors to suggest that the thick-stemmed arboreal-shrubby flora contains more than1000 species and that the denser physiognomiesmay reach more than 150 arboreal and shrubbyspecies per hectare. Castro (1994; see Ratter et al.,1997) made an extensive survey of the literature inorder to gather support for the idea that the arborealand shrubby flora of the Brazilian cerrados is muchricher than previously assumed. It could be argued that as the cerrados consistof physiognomies that are predominantly grass-lands, the greatest floristic richness should be en-countered in the non-woody (herbaceous-subshrub-by) component of the vegetation. Surveys of thiscomponent have been rare in Brazil (Mantovani &Martins, 1993). Comparing the non-woody compo-nent in different localities in Brazil, Mantovani(1983) found a local richness that varied between165 species in the Serra Dourada (state of Goi�s)and 640 in the municipality of Lagoa Santa (stateof Minas Gerais). In an area of 343.42 ha of a cer-rado in the Reserva Biol6gica de Moji Gua�u (stateof S�o Paulo), Mantovani and Martins (1993) found403 species of non-woody angiosperms. The her-baceous-subshrubby angiosperm flora of the cer-rados therefore appears to be richer than the ar-boreal-shrubby flora, but its richness varies withphysiognomy (Mantovani, 1987). It can also be argued that the maximum physi-ognomic and floristic expression, together withmaximum spatial continuity, should occur in the"nuclear" (Labouriau, 1966), "central" (Rizzini,1963), or "core" (Eiten, 1972; Ferri, 1977a) areas.An implication of this reasoning is that marginaland disjunct areas (Ratter et al., 1988a) shouldhave a relatively impoverished flora in comparisonto the nuclear area, although they may be supple-mented by floristic elements from the sdhaoundingvegetation formations (Eiten, 1972; Fernandes &Bezerra, 1990; Rizzini, 1963). These elements,which occur preferentially in other formations and ' Research was developed in the Curso de P6s-Gradua��o em Biologia Vegetal, Instituto de Biologia. UniversidadeEstadual de Campinas, Brazil. We thank Esmeralda Zancheta Borges for preparation of the map in Figure 1. 2 Departamento de Biologia, Centro de Ci�ncias da Natureza, Universidade Federal do Piauf, Campus da Ininga,Teresina 64049-550, PI, Brazil. SD)epartamento de Bot�nica, Instituto de Biologia, Universidade Estadual de Campinas, Caixa Postal 6109, Campinas13083-970, SIP Brazil. * To whom correspondence should be sent.ANN. MIssouRI BOT. GARD. 86: 192-224. 1999.