Models of optimal territory size are usually tested only by demonstrating t
hat territory size is inversely related to food abundance or intruder numbe
r. The most fundamental predictions of the models, however, have rarely bee
n tested: i.e. the fitness of the defender is a function of territory size
and the optimal territory is one of intermediate size. We tested these pred
ictions by measuring the growth rare of large convict cichlids (Archocentru
s nigrofasciatus, formerly Cichlasoma nigrofasciatum) while defending food
patches against smaller intruders over a 10-day period. Food patches differ
ed in area by more than two orders of magnitude. We manipulated food abunda
nce so that it increased with patch size in a decelerating way. As assumed.
the realized benefits of defence (weight of food eaten by the defender) an
d the costs of defence (chase rate and chase radius) both increased in a de
celerating way with increasing parch area. As predicted, the growth rate of
the defender first increased and then decreased with increasing patch size
. The initial increase in defender growth rate with increasing patch size w
as related to an increase in food eaten, but the decrease in growth rate fo
r fish defending the largest patches was related to the costs of defence. F
ish defending large patches had a low growth efficiency, apparently because
of the social stress caused by intruders in their territories. Taken toget
her, these results support the assumptions and predictions of optimal terri
tory size models.