Geoscience Reference
In-Depth Information
that patterns found when using species-aggregated data deviate from those
when individual data are used, for a wide range of parameters and across
multiple study systems. Specifically, for all seven systems, we found that the
slope of prey mass as a function of predator mass was consistently under-
estimated and the slope of PPMR as a function of predator mass was
overestimated, when species averages were used instead of the individual-
level data (
Figure 4
B and D). It is also worth noting that none of the three
Chilean rivers had a significant slope of prey mass as a function of predator
mass when species averages were used but did when individual-level
data were used (
Figure 4
B and
Table A1
). The other response variable
sets (diet and predator variation) were not affected by the degree of
resolution (
Figure 8
B, D and
11
B, D).
The prey mass and PPMR response variables are directly related—the
slope of the PPMR-predator mass relationship equals 1 minus the slope of
the prey mass-predator mass relationship, and the intercepts have the same
magnitude but opposite signs (for an analytical proof, see
Box 1
). The high-
and low-resolution prey mass-predator mass relationships had slopes be-
tween 0 and 1, except for Trancura River (slope
1 in resolution A, D and C)
>
and Coilaco (slope
0 in resolution D). The slopes of the prey mass-predator
mass and PPMR-predator mass relationships give us valuable information
on the size structure of a community. However, to be able to compare the
PPMR between resolutions within a system, we also need to consider the
intercepts of the scaling relationships. The regression lines in
Figures 14 and
15
illustrate prey mass and PPMR as functions of predator mass for the
different resolutions (individual-level data (A) and species averages (D)) for
each of the seven systems. For all systems, except Trancura River, the slopes
of the PPMR-predator mass relationships derived from species averages are
steeper than those derived from individual-level data. Hence, the strength of
the PPMR scaling with predator mass based on species averaging would
nearly always be exaggerated. Moreover, for all systems except Tadnoll
Brook and Trancura River, the high- (individual-level data) and low-(species
averages) resolution regression lines cross somewhere within the observed
size range of predator individuals. Thus, using species averages would result
in an underestimate of PPMR for predators in the lower end of the size
spectrum (to the left of the point of intersection) and an overestimate for
predators in the higher end (to the right of the point of intersection).
These findings support and bring additional clarity to the results of
Woodward et al. (2005a) and Woodward and Warren (2007)
, who, when
examining average body masses of species only, found that some predator
species appeared to feed on prey nearly 100 times larger than themselves
(
Woodward et al., 2005a
). When revisiting the same data from an individual-
based perspective, however,
Woodward and Warren (2007)
showed that at
the level of individual interactions, no predator consumed prey larger than
<
