INVERTEBRATE PREDATION ON PATCHY AND MOBILE PREY IN STREAMS

(1) A series of field experiments with predatory net-spinning larvae of the caddisfly Plectrocnemia conspersa (Curtis) were carried out in a small fishless stream to determine any impact of this predator on the abundance of its prey. Densities of fifth instar caddisfly larvae were manipulated in enclosures permeable to most other invertebrates in the stream. (2) Three experiments were carried out (spring, summer and winter), each with a 2-week preliminary colonization period to establish prey assemblages inside enclosures. Then zero, one or four predators were introduced per cage (low, ambient or high predator densities). Significant predator impacts were observed in winter only, and results varied with prey taxon. These observations could be attributed either to largely trivial effects of predation, or to technical difficulties in its detection. (3) Two subsequent experiments (A and B) were designed to assess how the detection of predator impact in enclosures is influenced by the patchy microdistribution of lotic invertebrates, and by the continuous exchange of mobile prey with the benthos. (4) In experiment A, prey exchange rates were manipulated by wrapping some cages with fine mesh after the initial colonization period. The evidence that exchange rate influenced the detection of impact by predators was equivocal and may have been obscured by the natural, spatial variability of lotic invertebrates. (5) In experiment B, exchange rates were again altered by manipulating the mesh size of enclosures but, in addition, initial variability in the number of prey per cage was reduced by stocking rather than by allowing colonization. The abundance of total prey and of individual prey groups was reduced significantly in predator treatments compared with controls. Predator impacts were most pronounced at low exchange rates, and were less distinct at high exchange. The effect of mesh size on prey exchange varied among taxa in relation to species-specific morphology and dispersal behaviour. The experimental detection of predator impacts therefore appears to be highly dependent on both prey mobility and spatial heterogeneity.