spaces or refuges). But other things are not equal and, to be honest, we would
have been surprised to find a clear relationship between size and algal produc-
tivity in the present study: stream invertebrates depend on both detrital and
algal resources and the relative importance of these differs dramatically in
forested and grassland streams. In contrast, we found in the same set of streams
a positive relationship between food-chain length and algal productivity
(Townsendet al., 1998; Thompson & Townsend,2005).
Patterns with individual factors must be particularly strong if they are to
be detectable in our multivariate investigation. This may be the case for
substratum particle size heterogeneity, where body size increases from low
to intermediate heterogeneity before declining again, perhaps reflecting the
(unmeasured) details of the fractal nature of our streambeds (Schmid & Schmid-
Araya, this volume). The availability of interstitial spaces and refugia in
streambeds is likely to be a complex interaction between the size of substratum
particles, degree of infilling by fine sediments and the shapes of bed particles.
Streams in this study with high heterogeneity tended to include very small
particles, which filled interstitial spaces. Low heterogeneity streams were
typified by small particles, which may not develop interstitial spaces. In
contrast, the sites with intermediate heterogeneity lacked very small particles
but included large particles, providing a range of interstitial habitats. We
also obtained results supporting the prediction that brown trout, through size-
selective predation, would limit the size of predatory invertebrates in the
streams.
Turning to patterns in the complex land-use gradient, the larger body sizes in
pasture than tussock streams is consistent with higher algal productivity in
pasture (because of higher nutrient concentrations), although this is not sup-
ported by our short-term productivity measures. Alternatively, the pattern may
relate to an unmeasured but potentially important effect of greater concentra-
tions of contaminants in pasture settings. We have seen that contaminants can
have stronger adverse effects on smaller invertebrates because of their larger
surface-area-to-volume ratios; this is an area that merits further investigation.
In conclusion, macroinvertebrate body-size patterns are sometimes apparent,
but they rarely account for a large proportion of variation along the dimensions
of interest. This is no doubt partly because several factors operate simultane-
ously (even in partially controlled experiments) and, if these engender different
size-related outcomes, the patterns will be muted or obscured. An additional
problem is that different macroinvertebrates may have different solutions to
environmental stressors. For example, species characteristic of disturbed set-
tings might be ‘resistant’ large animals that are clingers or streamlined, or
‘resilient’ small animals with rapid recolonization and high population-growth
rates. Moreover, there may be large animals whose resilience depends on their
ability to move away before the disturbance reaches full intensity, to return
BODY SIZE IN STREAMS 93