Although omnivory (the consumption of resources from more than one trophic
level) is widespread, this fundamental limitation to the applicability of f
ood chain theory to real communities has received only limited treatment. W
e investigated effects of enrichment (increasing carrying capacity, K, of t
he resource) on a system consisting of a resource (R), an intermediate cons
umer (N), and an omnivore (P) using a general mathematical model and tested
the relevance of some of its predictions to a laboratory system of mixed b
acteria (=R) and the ciliates Tetrahymena (=N) and Blepharisma (=P). The mo
del produced six major predictions. First, N may facilitate or inhibit P. E
nrichment may revert the net effect of N on P from facilitation to inhibiti
on. Second, along a gradient of K, up to four regions of invasibility and s
table coexistence of N and P may exist. At the lowest K, only R is present.
At somewhat higher K, N can coexist with R. At intermediate it, either N a
nd P coexist, or either consumer excludes the other clef ending on initial
conditions. At the highest K, N may be excluded through apparent competitio
n and only R and P can coexist. The pattern of persistence of Tetrahymena a
nd Blepharisma along an enrichment gradient conformed fairly well to the sc
enario allowing coexistence at intermediate K. Third, fur stable equilibria
of the omnivory system, R always increases and N always decreases with R.
The abundances of bacteria and Tetrahymena were suggestive of such a patter
n but did not allow a strict test because coexistence occurred at only one
level of enrichment. Fourth, an omnivore can invade an R-N system at a lowe
r K than an otherwise identical specialist predator of N. Fifth, an omnivor
e can always invade a food chain with such a specialist predator. Sixth, ov
er ranges of K where both omnivory systems and otherwise identical three-le
vel food chains are feasible, N is always less abundant in the omnivory sys
tem, whereas the relative abundances of R and P in omnivory systems compare
d to food chains may change with K. It is thus possible that total communit
y biomass at a given It is lower in an omnivory system than in a food chain
. Both the model and the experimental results caution that patterns of trop
hic-level abundances in response to enrichment predicted by food chain theo
ry are not to be expected in systems with significant omnivory.