Several species of African grasses brought to Brazil as cattle forage have spread widely, outcompeting native herb species. The open forms of Brazilian savanna ("campo cerrado" and "campo sujo") are the most affected by such invasions, because their structure is open, permitting enough sunlight into the lower strata. The invasion of alien forage grasses occurs in almost every cerrado nature reserve. This study was carried out in the "Cerrado de Emas Biological Reserve", Pirassununga, Sao Paulo State, Brazil, with the following objectives: (a) to compare the abundance of native and alien grass species; (b) to verify the importance of such alien grasses in the community; (c) to identify distribution patterns for the alien grass species in a gradient from the edge (highly disturbed) to the center (less disturbed) of the reserve; and (d) to explore the distribution of native and alien grasses in the search for possible competitive interactions. Using the "point method," a total of 260 points was sampled and 52 species were recorded. The four most frequent species (FA = absolute frequency) were two native (Echinolaena inflexa [Poir.] Chase [FA = 38.85%] and Diandrostachya hrysotrix [Nees] Jacues Felix [FA = 15.38%]) and two alien African species (Melinis minutiflora Beauv. [FA = 33.08%] and Brachiaria decumbens Stapf [FA = 13.85%]). M. minutiflora and E. inflexa had higher values of absolute vigor (67.69 and 59.62%, respectively), relative vigor(28.16 and 24.80%, respectively), and cover (100.77 and 98.47, respectively), indicating higher biomasses and densities and their dominance in the community. B. decumbens presented the highest number of contacts per point, showing the highest stratification. To detect possible edge-center distribution gradients, correspondence analysis was done, initially using all the recorded species and subsequently only the four more frequent grasses, with similar results: (a) the alien grasses, especially M. minutiflora, did not show a distinct distribution gradient from edge to center, but occurred over the whole reserve; (b) no distinct ecotonal band around the reserve (edge-belt) was detected, the whole reserve seeming to be "ecotonal"; and (c) E. inflexa and M. minutiflora showed similar phytosociological patterns, and spatial distribution; association between these two species was statistically significant.
