effects. This raises wider questions about the conventional construction of food
webs, such as whether the perceived prevalence of omnivory, cannibalism and
feeding loops (speciesAeats speciesBeats speciesA) reported in summary webs
might be an artefact of species averaging. Complex feeding loops (AeatsBeatsC
eatsA) among predators, determined by body size, produce dynamics that are
often highly unstable in standard Lotka-Volterra population-based models (Polis,
1991 ) and should therefore be rare in nature. However, aquatic food webs are
typically rich in omnivory (feeding across several trophic levels) and feeding
loops, especially in invertebrate-dominated systems (Woodward & Hildrew,
2002a). Many of these interactions represent ‘life-history’ omnivory, and in
extreme cases this can give rise to feeding loops and reversals in trophic status,
whereby large individuals of small species prey upon small individuals of species
that are ultimately larger (Woodward & Hildrew,2002b). This often occurs when
generations overlap, which can also introduce a seasonal dimension to shifts
within the food web: all of these complexities are lost by averaging across species
and time, and more subtleindividual-based models might provide greater insight
(e.g. Claessen, De Roos & Persson,2000 ).
However, these marked changes in trophic interactions, whilst dramatically
altering the occurrence of different feeding links between species, may still
result in networks structured by size. If the life-history changes are reasonably
synchronous (as is the case in many invertebrates) at any one time species can
still be ordered by size, and the underlying size constraints on feeding still
apply. If life-history changes are very asynchronous, such that many stages of
each species are present at once, then size constraints still operate, but now at
the level of life stages, or individuals. Thus, body size could provide an over-
arching set of rules that might generate relatively stable patterns of structure,
even where links between particular species change markedly. Such structural
constraints are suggested by static food-web models based on the hierarchical
ordering of feeding niches (e.g. Cohenet al., 1986, Warren,1996, Williams &
Martinez,2000, Cattinet al., 2004). The complex changes in feeding interactions
associated with shifts in species’ size distributions are not inconsistent with
these models if size rather than taxonomic identity is the constraint (e.g.
Woodward & Hildrew,2002b).The other side of the equation: community-size distributions
The ‘rules’ by which consumers are size selective might be considered to define
the fundamental niche, whereas the realized niche is determined by the actions
of individual consumers plus the size structure of the underlying community.
The distribution of all possible body-size ratios in a food web (Fig.6.7) is clearly a
function of the size distribution of the different species. Size spectra have often
been used to depict the distribution of body sizes within a community web in
marine systems and, to a lesser but increasing extent, in freshwaters (Schmid,110 G. WOODWARD AND P. WARREN