Geoscience Reference
In-Depth Information
following
Woodward et al. (2010)
, as many size classes were used as there are
species for each web. For each study system, the size ranges were equal on a
logarithmic scale, with the total range set by the minimum and maximum
sized individual in that system.
Hereafter, comparisons done between any two different taxonomic group-
ings (including the raw data,
Figures 1
A-D and
2
A-D) are referred to as
resolution comparisons and comparisons done between one taxonomic group-
ing (
Figures 1
B-D and
2
B-D) and one size-class grouping (
Figures 1
E-F and
2
E-F) as grouping comparisons.
3. Food Web Aggregations
In order to calculate some of the response variables (see Section
II.C
), such as
TH, aggregated data (as described above) were used. It was possible to do this
for groupings for the taxonomic approach using aggregation level D and using
aggregation level F for the size-class approach (
Figure 1
). In the taxonomic
food web (hereafter denoted level D*), a species was defined as predating on
another if at least one prey species individual was found in the gut of a predator
species individual, a criterion that is commonly used when constructing ''tradi-
tional'' species-based food webs (e.g.
Woodward andHildrew, 2001
). Likewise,
in the size-class food web (hereafter denoted level F*), a feeding link was
assigned if at least one prey item within a size class was found in the gut of a
predator of another size class, irrespective of their taxonomy. The body masses
assigned to each node were from aggregations D and F. Thus, levels D* and F*
are two sides of the same coin: both are derived ultimately from the same
individual data, but they take species- and size-based perspectives, respectively.
Figure 3
illustrates how the links were assigned between nodes in these two
approaches, using the example from
Figure 1
.
C. Response Variables Analysed
Throughout this study, we examined the allometries of nine variables that
relate to trophic interactions. Some of these variables could only be measured
at particular levels of aggregation (for instance, TH could only be calculated
for species or size classes and not individuals; see below for more details),
while others could be measured at any level. For clarity in the comparisons
made, only the relationships of the least and most aggregated data possible
were examined (see
Table 2
for a guide to the comparisons carried out).
Hence, the species-averaged aggregation (D) was compared with the aggre-
gation closest to the raw data describing the individual predation events. In
the comparisons between the size class and taxonomic groupings, the lowest
