with sizes being the average size of each species), as is often done in studies
comparing consumer-resource body-size ratios (e.g. Broseet al., 2005), the true
ratio of predator and prey body mass is underestimated by about one order of
magnitude, when compared with that obtained as the average of the size ratios
of individuals involved in actual feeding events (e.g. Fig.6.5c vs. Figs.6.5d & 6.5e).
This is because species averaging effectively assumes that all sizes of predator
population are equally likely to feed on all sizes of prey; but this is rarely likely to
be the case. One consequence is that the triangularity in the species-averaged
feeding matrix disappears, as the preponderance of smaller than average prey
is revealed in the predators’ diets.
Taking averages at the species level gives a log-normal distribution of body-mass
ratios, with a mean close to one: i.e. predators are, on average, about ten times
larger than their prey. A similar-shaped distribution is obtained when this is
repeated for individual feeding links, but predators are, on average, now about
100 times larger than theirprey (Fig.6.5). Consequently, species averaging severely
underestimated the true size disparity between consumers and resources, such
that the prey that were actually consumed were considerably smaller than those
that potentially could be consumed, highlighting the potential importance of size
refugia from predation at the two extremes of a prey species’ size spectrum.
Ontogenetic size refugia can arise where individuals that survive to become
large become invulnerable to predators (Fig.6.6b prey speciesB) and examples
from aquatic systems are well known (Bechara, Moreau & Hare,1993, Chase,
1999 ; Woodwardet al., 2005c). The effects of these partial overlaps of feeding
niche and prey-size distribution will depend on the contribution that individu-
als in the exploited part of the spectrum make to the reproductive output of the
population. Interactions of this sort, where parts of species’ populations interact
intensely and others little or not at all, can in theory lead to complex dynamics
that differ considerably from the dynamics expected if all individuals are equiv-
alent. This phenomenon often also has a temporal component: for instance,
where the numbers of individuals in the different parts of the size spectrum vary
over time through reproductive pulses (De Roos & Persson,2002; Scheffer &
Carpenter, 2003 ; Bystro ̈m, 2006; Persson & De Roos, this volume).
Clearly, although combining information from all sizes of predators and prey,
and across sampling occasions, is the standard method for constructing most
food webs, such species-averaged webs, whilst depicting a correct representation
of the linkages between species, mask the effects of size partitioning, onto-
genetic and seasonal shifts, and the potential for intraspecific size-refugia from
predation. It is difficult from existing food-web data to get a clear measure of the
magnitude of this effect but, for example, within the Skipwith food web about
36% of links are present only at certain life-stage combinations of consumer and
resource: some of this variation (and certainly much more in terms of frequency
of interaction rather than just presence) can be attributed to individual size
BODY SIZE AND PREDATORY INTERACTIONS IN FRESHWATERS 109