animals, or that the larger interstitial spaces associated with mountain cobble
streams were responsible.
Experimental studies
Experimental studies of body-size patterns can be expected to provide the
clearest demonstration of underlying mechanisms. Taniguchi and Tokeshi
(2004) established artificial substrata whose surface irregularities were quanti-
fied according to their fractal geometry. The substratum plates were cut from
natural serpentine bed material to create five levels of complexity of cavity sizes
by attaching 0.3cm thick stone squares of different dimensions (2020 cm,
10 10 cm, 55 cm, 2.52.5 cm) to a basal plate of constant area (2020 cm).
Replicate plates were left in a flashy Japanese stream for 36 days, repeating
the experiment in summer, winter and spring. Invertebrate body size declined
with habitat complexity in all seasons. In other words, smaller individuals were
commoner on substrata that provided smaller interstices, most likely acting as
refuges against dislodgement in the current (or possibly against predation).
Thus, there seemed to be a match between crevice size and invertebrate size.
Downeset al.(1998) also found a higher proportion of small invertebrates in
substrata with small crevices than on crevice-free substrata.
Taniguchi and Tokeshi’s (2004) well-controlled experiment is illuminating,
but they wisely point to the importance of considering habitat complexity in
conjunction with other environmental factors. Thus, they left some of their
plates to be colonized by algae and to accumulate detritus for one month prior
to the experiment (removing any accumulated invertebrates after the precondi-
tioning period). Across the range of substratum complexity, by the end of the
experiment, mean body size was greater on plates that had previously accumu-
lated food resources. This accords with the prediction that larger invertebrates
may have a competitive advantage in more productive situations, although
greater growth rates on preconditioned plates cannot be ruled out.
Another biotic variable to be treated experimentally is predation. Flecker and
Townsend (1994) established replicate artificial channels containing natural
substratum in a New Zealand stream, allowing algae and invertebrates to colo-
nize before establishing three predation treatments – no fish, native galaxiid
fish or exotic brown trout (Salmo trutta). After ten days the channels with the
voracious trout contained half the density of large invertebrate carnivore spe-
cies compared with those with native or no fish. Moreover, as with Taniguchi
and Tokeshi’s (2004 ) productivity pattern, the effects of fish predation on inverte-
brate size structure can be moderated by structural complexity. Thus, in enclo-
sures in a Californian river, the presence of roach (Hesperoleucas symmetricus) and
steelhead (Oncorhyncus mykiss) in a six-week experiment reduced the density of
large invertebrate predators when the substratum consisted of boulders but not
gravel (Power,1992). This supports the hypothesis that refuges provided in
BODY SIZE IN STREAMS 83