Predicted Effects of Behavioural Movement and Passive Transport on Individual Growth and Community Size Structure in Marine Ecosystems - Advances in Ecological Research - page 88

grouping consistently changed the steepness of the slopes, a Welch two-sample
t-test assuming unequal variances was used ( RDevelopment Core Team, 2009 ).
An estimate of the slope for the unimodal responses could not be estimated, so
the values of the second-order coefficient were used instead and tested to see if
the quadratic terms were greater for a particular grouping.

III. RESULTS

A. Response Variables Compared

In total, 15 comparisons were made between the different aggregations. Each
response variable will be presented, in turn, following the size structure dimen-
sion sets described in Section II . A summary of the results of all comparisons
done can be found at the end of the results section. The detailed results of each
individual regression conducted are presented in Tables A1 and A2 .

1. Size Structure Dimension Set #1: Trophic Orderings

a. Predator Mass-Prey Mass (Resolution). An LMM was used to estimate
the slopes for each study system, with random effects for predator identity
and predator species identity. The paired t-test ( Figure 4 B) comparing the
aggregations showed that slopes were significantly steeper for the regressions
of the raw data (aggregation A: with a mean slope of 0.694) than for those of
the less resolved species averages (aggregation D: mean slope of 0.334,
t 0.007). The predator body mass-prey body mass

LMMs are shown in Figure 5 A and B.

¼ 3.970, df ¼ 6, p ¼

b. Predator Mass-Prey Mass (Grouping). Slopes were estimated with
LMMs that included a random effect of predator species identity
( Figure 5 C and D). When compared with a paired t-test ( Figure 4 C), the
slopes of the study systems using a taxonomic grouping based on link
averages (aggregation C) were significantly steeper (mean slope of 0.717)
than those of the size-class-based grouping (aggregation F), which had a
mean slope of 0.557 (t ¼ 2.594, df ¼ 6, p ¼ 0.041).

c. Predator Mass-Predator-Prey Mass Ratio (Resolution). Here, an
LMM was used to estimate the slopes for each study system, with both
random effects for predator individuals and predator species ( Figure 6 A
and B). The low-resolution species average aggregation (D) had steeper
slopes (mean of 0.666) than the individual data (aggregation A: mean slope