Results
The size frequency histograms are shown in Fig.5.2. Invertebrates ranged from
1.5 mm to 25 mm, with mean lengths between 2.5 and 4.9 mm and with the
majority of individuals across all sites in this size range. Invertebrate abundan-
ces were significantly influenced by land use (F3,14¼22.33, p<0.01), with native
forest sites having significantly more individuals than other categories, and
pasture and tussock sites having significantly higher abundances than pine
forest sites (Tukey’spost-hoctests). There was no evidence of strong seasonal
variability in body size for the two sites where data were available, and no
apparent pattern involving functional feeding groups.
Contrary to our predictions, body size was not significantly related to dis-
turbance (Fig.5.3a), average substratum particle size (Fig.5.3b) or algal pro-
ductivity (Fig.5.3c). However, body size seemed to increase from low to
intermediate substratum variance but to decrease again in the most hetero-
geneous substrata (Fig.5.3d). Grazers and, in particular, shredders were mainly
responsible for this pattern.
There was a significant effect of land use on body size overall and, separately,
for grazers, shredders and collector gatherers (Fig.5.4). Invertebrates were
larger (both mean and upper quartile lengths) in pasture streams, driven
mainly by the presence of larger collector gatherers. Native forest sites were
typified by the absence of small grazers, resulting in significantly greater
lower quartile lengths for that functional feeding group than in other land
uses. However, interpretation of a land-use effect as independent from a fish
effect was confounded by the presence of trout in all native forest streams (see
below). Tussock streams were typified by small body sizes across all functional
groups.
The fish regime had a significant effect on the body size of predatory inverte-
brates, with sites containing only brown trout having smaller individuals, as
predicted (Fig.5.5).
In the principal components analysis of the proportion of individuals in each
size class, streams scoring highly on Principal Component 1 (PC1) had a high
percentage of mainly very small individuals (1.5–4mm), while those scoring
highly on Principal Component 2 (PC2) had a higher proportion of intermediate
sized individuals (4–10mm). Overlaying the physicochemical data on to the
ordination (Fig.5.6) indicated a trend for high scores on PC1 to be correlated
with high algal productivity, substratum particle size and substratum variability.
A Mantel test including all of the predictors supported this with the best model
(Spearman’s Correlation 0.198) including algal production and average substra-
tum particle size, and all of the ten best models incorporated those factors.
Because PC1 was positively correlated mainly with the smallest size classes,
these results run counter to our predicted positive relationships between size
and both productivity and substratum particle size.88 C. R. TOWNSEND AND R. M. THOMPSON