Tropical livestock is often maintained on roughage-based diets deficient in N, and therefore requires supplementation with protein-rich substrates to achieve reasonable production levels. The optimum inclusion rate of a potential supplement is usually determined by in vivo feeding trials or by in vitro incubation of the diet components to estimate the feed value of the complete diet. The present work simulates a supplementation experiment in vitro, by incubating a pure roughage (barley straw), a pure supplement (Sesbania pachycarpa leaves) and mixtures of the two, with increasing inclusion levels of the supplement, in a short-term batch incubation system. Fermentation kinetics were followed by the release of fermentation endproducts (gas and short-chain fatty acids). Microbial biomass was estimated using ribosomal (r) RNA as internal marker for bacteria and eukaryotes separately. Cell-wall-degrading subpopulations were quantified by hybridisation with taxon-specific oligonucleotide probes targeting Chytridiomycetes, Fibrobacter spp., Ruminococcus albus and R. flavefaciens. Carboxymethylcellulase (CMCase) was assayed as an indicator for cell-wall-degrading activity. The addition of S. pachycarpa leaves stimulated fermentation in all cases. Gas production, and especially rRNA concentration, showed clear maxima at 40 % S. pachycarpa inclusion, rates that significantly exceeded the values interpolated from the incubations of the pure substrates. Short-chain fatty acid yield changed only slightly, but in the same way. The analysis of the microbial population structure showed that the positive effects were mainly mediated through enhanced growth of Ruminococcus spp. Increasing proportions of S. pachycarpa leaves in the diet led to a drastic decline in the total eukaryotic population. This points to a defaunation, which may also have added to the positive effects. The eukaryotic subpopulation of the rumen fungi were affected to a lesser degree. Although the cell-wall-degrading organisms showed positive responses to the supplementation, the CMCase activity was not affected significantly by the supplementation. The present work shows that it is possible to predict optimum inclusion levels for a new feed supplement in vitro and thus reduce in vivo experiments. It was also demonstrated that true supplementation effects occur particularly for the microbial biomass production, which is the primary source of amino acids for the ruminant animal. The analysis of microbial population structure in context with conventional metabolic measurements adds valuable information to interpret the observed effects on production-related variables.
