The semi-humid or humid tropics are ideal for the production of large quant
ities of biomass from fast-growing C-4 grasses, but high yields normally re
quire large quantities of fertiliser, especially N, which has a very high i
nput from fossil fuels (natural gas). A program has been started recently t
o use elephant grass (Pennisetum purpureum Schum.) to substitute firewood a
s a fuel and also to make charcoal for iron production. In this case, any l
arge N fertiliser additions would mean that the yield of bio fuel per unit
of fossil fuel invested would be detrimentally affected. In this study, we
report on the potential for the selection of genotypes of fast-growing C-4
tropical grasses of the genera Pennisetum and Brachiaria for their capacity
to obtain N inputs from plant-associated biological nitrogen fixation (BNF
). Fourteen genotypes each of Brachiaria and Pennisetum were screened for B
NF contributions by growing them in N-15-labelled soil. In the case of the
Pennisetum, after a suitable cutting height for the crop had been selected,
there were large differences in dry matter production, N accumulation and
N-15 enrichment. The differences in N-15 enrichment between genotypes were
statistically significant and BNF inputs were estimated as high as 41% of a
ccumulated N. In the study on Brachiaria genotypes, potential inputs of BNF
seemed lower. Only one or two genotypes of B. brizantha and B. ruziziensis
obtained more then 20% of their N from BNF. The N-2-fixing bacteria that w
ere most commonly associated with shoots and roots the Pennisetum genotypes
were of the genus Herbaspirillum, but predominantly of a recently describe
d new species. The Brachiaria spp. from three different sites (Rio de Janei
ro, Goania, Bahia) were predominately colonised by Azospirillum spp., most
of the isolates being of the species Azospirillum amazonense. Very few Herb
aspirilla were isolated from these plants.