To examine size-dependent food web interactions in systems with cannibalism
, we compared the abundances of zooplankton and phytoplankton over 2 years
in four lakes with cannibalistic perch (Perca fluviatilis ) of which two al
so supported the top predator pike (Esox lucius). The abundance of perch 2
years and older was lower in lakes with pike than in lakes with only perch.
In contrast, the abundance of small perch (young-of-the-year and I-year ol
d) was lower in lakes with only perch suggesting that intense cannibalism r
educed these size classes to low levels in lakes lacking pike. Functional r
esponse experiments with differently sized perch and zooplankton showed tha
t the attack rate of small perch susceptible to cannibalism was much higher
than that of large cannibalising perch. The optimal body size of perch wit
h respect to attack rate was also lower for small zooplankton prey than for
large zooplankton. The zooplankton communities in lakes with only perch we
re dominated by the relatively small species Ceriodaphnia quadrangula and B
osmina spp. and total zooplankton biomass was higher in these lakes than in
lakes with both pike and perch. In contrast, the mean size of cladoceran z
ooplankton was largest in lakes with both pike and perch owing to a dominan
ce of the large zooplankton species Holopedium gibberum in these lakes. We
relate these patterns to (1) the low foraging efficiency of large perch on
small zooplankton and (2) the low abundance of small zooplanktivorous perch
(due to cannibalism) in lakes with only perch. The differences in zooplank
ton community structure also resulted in different seasonal dynamics of phy
toplankton between lakes. Cannibalism introduces a vertical heterogeneity t
o food webs that causes consumer-resource dynamics that are not predictable
from linear food chain models.