Geoscience Reference
In-Depth Information
(i.e. sites 03, 15, and 20), for which the AIC obtained negative infinite values.
For the species-averaged PPMR, three of the models (with the exception of
the null model) exhibited the same performance at sites 01, 09, 12, 14, and 19.
This may be attributed to the fact that these sites include just two predator
species. Interestingly, at other sites (i.e. sites 02, 16, and 18), an allometric
model was commonly selected for species-averaged PPMR, although this
was not always the case for link-averaged PPMR. If more predator species
are sampled, a pattern showing that body mass is crucial for species-averaged
and link-averaged PPMR may emerge.
Individual-predator and individual-link PPMRs were generally best
explained by the combined model that included both species identity and
body mass (7/11 sites for individual-predator PPMR and 9/11 sites for
individual-link PPMR;
Table 2
). A null model was not selected in any of
the sites. This result has two implications. First, interactions between preda-
tor and prey individuals are critically affected by both species identity and
body mass. This develops the previous argument by
Barnes et al. (2010)
, who
only emphasised the effect of body mass on PPMR. Second, the determinants
of PPMR may become clearer with increasing resolution of data analysis, as
indicated by the result that a particular model was selected for individual-
predator and individual-link PPMRs, while the best model was unclear for
species-averaged and link-averaged PPMRs.
We evaluated in more detail how PPMR is determined by species identity
and body mass by highlighting two of the models that explain individual-
predator PPMR of each species. These comprised the model with species
identity alone and the combined model with both species identity and body
mass (see
Barnes et al., 2010
for the effect of body mass alone). The model
with species identity alone is based on the assumption that PPMR is common
within species. The analysis showed that there were significant interspecific
differences in PPMR (i.e. the 95% confident interval of at least one of the
species did not overlap with that of any other species) in 8 of 11 sites (i.e. sites
01, 02, 09, 12, 15, 16, 18, and 20;
Figure 4
). The most distinct differences were
found at site 20, which contained almost one-third of all interaction records
for the 21 sites (n
¼
10,994;
Table 1
). At site 20 (NE US Continental Shelf),
Merluccius bilihearis (commonly named silver hake) and Mustelus canis
(smooth dogfish) had the lowest and highest PPMRs of 10
1.25
0.05
and
10
2.94
0.09
, respectively, indicating a difference of about a 50-fold. No signif-
icant interspecific differences in PPMR were observed for sites 03, 14, and 19,
probably due to small sample sizes (n
115, 163, and 414, respectively).
We evaluated the interaction effect of species identity and body mass by
comparing the regression slope of the relationship between log
10
(PPMR)
and log
10
(individual body mass) of each predator species, as in the meth-
odology of
Barnes et al. (2010)
at the community level. If the slope is
positive (or negative),
¼
then it
indicates that
the relationship between
