Litter deposition and disappearance in Brachiaria pastures in the Atlanticforest region of the South of Bahia, Brazil

Over the last 25 years more than 70 million ha of the native vegetation in Brazil have been replaced by pastures for beef production planted to grasse s of the genus Brachiaria, and to a lesser extent Andropogon gayanus, both of African origin. Some years after implantation, these pastures decline in productivity, probably due to low availability of P, and immobilisation of N in the soil due to the large quantities of senescent leaves (litter) of high C:N ratio deposited on the soil surface. In this paper we report the e ffects of the introduction of a forage legume (Desmodium ovalifolium) and d ifferent animal stocking rates on the deposition and decomposition of plant litter in pastures of Brachiaria humidicola at a site in the coastal Atlan tic forest region of the south of Bahia State (Brazil). Litter existing on the ground, and that deposited in 14-day periods, was monitored at monthly intervals during 3 years of the study. Doubling the stocking rate from 2 to 4 animals ha(-1) caused a highly significant decrease in litter deposition , but the presence of the legume in the sward had little effect. Calculatio ns made directly from the quantities of litter deposited in the 14-day peri ods showed that between 15 and 18 tons of litter dry matter (dm) were depos ited annually, but the relatively small quantities of existing litter (annu al means of 0.8 to 1.5 t dm ha(-1)), showed that decomposition was rapid, s howing values for half life of between 22 and 33 days. This technique was a ssumed to underestimate true litter disappearance rates, as with such rapid decomposition a significant proportion of the litter disappeared within th e 14-day collection periods. An equation was developed to correct for this loss of litter during the collection periods and corrected litter decomposi tion constants of 0.037 to 0.097 g g(-1) day(-1) were recorded resulting in half lives of between 9 and 20 days. Using these data and adding them to e stimates of animal consumption the net aerial primary productivity (NAPP) o f the pastures ranged from 28 to 34 t dry matter ha(-1) yr(-1). Experiments with litter bags, and a "covered litter" system which allowed access of so il fauna to the litter, indicated that soil faunal activity had little impa ct on litter disappearance and such techniques underestimated true litter d ecomposition by at least an order of magnitude. We suggest that this undere stimation is due to the fact, that in contrast to litter bags, in the open field situation fresh litter is being added continuously. As this material consists of both easily degradable ("active") and recalcitrant fractions, t he easily degradable fraction fuels an active microbial biomass which conti nuously degrades the less decomposable material. It is concluded that the a pproach used in this study gives more realistic, and much higher estimates, of net primary aerial production of tropical grasslands and pastures than techniques heretofore utilised.