Determinants of Density–Body Size Scaling Within Food Webs and Tools for Their Detection - Advances in Ecological Research - page 26

self-thinning or CCSR was analysed with a linear regression between the
mean body size of all the organisms in the local community and the density of
individuals in that community.
The simultaneous consideration of the available approaches for the analy-
sis of the DMR in the ponds' metacommunity clearly highlights its main role
in influencing the inferred pattern ( Figure 6 ). Analyses based on recorded
abundance at the metacommunity ( Figure 6 A) or local community levels
( Figure 6 B) suggest a poor association between abundance and body size.
The analysis at the metacommunity level attempts to resemble the geograph-
ic analysis of DMR drawing abundance estimates from a large scale, but the
lack of a clear relationship and its similarity with the pattern observed at the
community level requires caution on this interpretation. In these two cases,
the explained variance is notably low (3.8% and 6.5%, respectively) and the
tendency is in fact positive. The fit of linear and non-linear quantile regres-
sions did not indicate significant trends in maximum densities, with zero
slopes within the confidence interval. On the other hand, the analysis of the
frequency distribution of body sizes at the individual ( Figure 6 C) and species
level ( Figure 6 D) identified regimes with a clear power-law scaling. In order to
evaluate whether a linear, non-linear, or segmented relationship better captures
the observed pattern, the AIC of alternative models were estimated (see
Figure 6 C and D). In this sense, the ISD and SMSD presented three different
regimes of scaling. The scaling exponents estimated for the distribution of
individual and species sizes were

1.77, respectively. These scaling
are steeper than those predicted from the energetic equivalence rule, with an
expected slope of

2.54 and

0.75, even when changes in trophic position with body size
are considered, for which a slope of 1isexpected( Brown and Gillooly, 2003;

Reuman et al.,2009 ). Finally, the self-thinning pattern among communities
indicated that half of the variation in individual densities among communities
was associated with the mean size of the individuals that composed those
communities ( Figure 6 E). The estimated slope for this relationship is congruent
with an energetic limitation of total abundance within communities ( Brown
et al., 2004; Damuth, 1981 ), causing a reduction in the number of individuals as
a consequence of the increase in resources demand with individual size. These
five relationships represent the main different approximations to the DMR so
far considered in the literature ( Reuman et al., 2008; White et al.,2007 ).
However, this is the first time that they have been presented simultaneously
for the same metacommunity. The first message from this result is that an
important fraction of the variation in reported DMRs probably represents
differences in the methodological approach, rather than real variation among
ecosystems (although the two are not necessarily mutually exclusive).
The scaling estimated from the two size distributions are amongst the largest
ever reported and merit further consideration. It is possible that the nature of
the study system (i.e. temporary ponds) accounts for the large scaling. In this