612 Reports Ecology, 83(3), 2002, pp. 612–618  2002 by the Ecological Society of America PREDATOR CHEMICALS INDUCE CHANGES IN MAYFLY LIFE HISTORY TRAITS: A WHOLE-STREAM MANIPULATION BARBARA L. PECKARSKY, 1,2,6 ANGUS R. MCINTOSH, 2,3 BRAD W. TAYLOR, 2,4 AND JONAS DAHL 2,5 1 Department of Entomology, Cornell University, Ithaca, New York 14853 USA 2 Rocky Mountain Biological Laboratory, P.O. Box 519, Crested Butte, Colorado 81224 USA 3 Department of Zoology, University of Canterbury, Christchurch, New Zealand 4 Department of Zoology and Physiology, University of Wyoming, Laramie, Wyoming 82071 USA 5 Department of Ecology, Lund University, Lund, Sweden Abstract. In high-elevation streams of western Colorado, mayflies (Baetis bicaudatus) develop faster, but mature at a smaller size where trout are present compared to streams where fish are absent. These life history traits reduce the time of larval exposure to trout predation, but cost reduced fecundity. We designed a field experiment involving manipu- lation of whole streams to determine whether these changes were caused by the presence of brook trout, and specifically, whether they could be triggered by trout chemical cues. In 1999 and 2000, we introduced water from containers with brook trout (Salvelinus fon- tinalis) into five naturally fishless streams, and fishless stream water into five adjacent control streams, to determine whether these cues alone could induce the mayfly life history traits we have observed in natural trout streams. As in previous small-scale experiments, the size at which mayflies matured declined significantly in streams with added trout chem- icals but did not change in streams with fishless water only. Thus, life history traits similar to those observed in the field were induced within the natural variability inherent in streams. These results demonstrate the strength of this predator–prey interaction and indicate that brook trout are an important agent of natural selection on mayfly life history traits. Key words: chemical induction; field experiment; life history traits; mayflies; nonconsumptive effects; predation; scale; size at emergence; streams; trout. INTRODUCTION Interesting patterns observed in nature are difficult to attribute definitively to specific mechanisms without the use of experiments. However, experiments are usu- ally conducted in simplified environments at small scales, necessitated by the requirement of achieving control of all but the variables of interest (Carpenter 1996). Furthermore, observations made at small spatial and temporal scales in simple environments often can- not be extrapolated to scales relevant to the organisms or processes under investigation (Peckarsky et al. 1997, Petersen et al. 1999, Bro ¨nmark and Hansson 2000). The larger the scale of observation, the more difficult it is to manipulate the variables of interest. Therefore, ecologists often rely on making inferences from ob- servations in nature to explain large-scale patterns (Power et al. 1998). However, the strength of these inferences can be substantially improved by innovative experiments carried out at relevant scales (e.g., Reznick et al. 1990, Carpenter et al. 1995, Wallace et al. 1999, Thrush et al. 2000). We have observed a striking natural pattern in high- elevation streams of Western Colorado. Mayflies (Bae- tis bicaudatus) common in the diet of brook trout (Sal- velinus fontinalis) (Allan 1981) matured at significantly Manuscript received 21 June 2001; accepted 12 July 2001; final version received 27 August 2001. 6 E-mail: BLP1@cornell.edu smaller sizes in streams with trout compared to streams where trout were excluded by natural barriers to dis- persal (Peckarsky et al. 2001). Also, the size at which Baetis metamorphosed declined over the emergence pe- riod, but only in streams that contained trout. These field patterns are consistent with the hypothesis that mayflies adaptively shifted their larval development in response to variable risk of predation (Peckarsky et al. 2001). Mayflies have relatively mobile larvae and ter- restrial adults that may disperse from their natal streams to oviposit (Sweeney et al. 1986, Schmidt et al. 1995, Bunn and Hughes 1997, Peckarsky et al. 2000). Thus, they typically encounter a variety of en- vironmental conditions within one generation, and should benefit by evolving flexible life history traits (Wilbur and Collins 1973, Newman 1989, Rowe and Ludwig 1991, Palmer and Poff 1997). Using field observations of larval growth rates and development times (from egg to maturity) we deter- mined that mayflies maturing in trout streams accel- erated their larval development, which reduced their exposure time to predatory trout, thereby increasing their probability of survival in trout streams (Peckarsky et al. 2001). However, the cost of accelerating devel- opment was smaller body size at emergence, which reduced female fitness (lower fecundity; McPeek and Peckarsky 1998), but not male fitness. Previous work suggests that large males have no mating advantage over smaller males, and there is no size-assortative mat-