1. Two contrasting explanations of niche separation in ruminants empha
size the importance of body mass-related trends in energy requirements
and food-processing ability (the diet-quality assumption) and of phys
iological adaptations to diet type (grass or browse; the diet-type ass
umption). 2. The quality of the diet consumed in the dry season by 21
species of African ruminants was estimated from observed fermentation
rates in the rumens of culled animals using a simulation model of dige
sta kinetics. The model predicts the effect of food composition (inclu
ding tannins) on a number of variables on which previous arguments abo
ut physiological adaptations have been based: cellulolytic activity, d
igestion kinetics and volatile fatty acid proportions. 3. The predicte
d diet quality (potential digestibility) varied from 0.90 in small ani
mals to 0.75 in large ones, and was negatively related to body mass wi
th a shallow allometric exponent(- 0.035 +/- 0.0109). 4. The allometri
c exponent scaling body mass to predicted energy assimilation rate was
0.852+/-0.0474. The abundance and quality of food available were appa
rently adequate to allow greater net energy intake, relative to requir
ements, by large animals than small ones. 5. No difference existed in
the predicted potential digestibility or net energy yield from the die
t of browsing and grazing species after controlling for body mass. The
refore, although the diets differed in their botanical and chemical co
mposition, the end result had little effect on the nutritional ecology
of the animals with different diet types. 6. The model predicted that
the level of cellulolytic activity of microbes in the rumen can affec
t the energy obtained from the diet. Microbial adaptation would cause
browsers to suffer a penalty on switching from browse to grass, thereb
y reinforcing diet choice. 7. The presence of tannins in the diets of
browsers was predicted to decrease rumen fermentation rate and increas
e the digestibility of the diet required to match observed fermentatio
n rates.