Quantifying The Relation Between Predator-Induced Behavior And Growth Performance In Larval Anurans - Statistical Data Included

Ecology, Sept, 1999 by Rick A. Relyea, Earl E. Werner

INTRODUCTION

A central goal of community ecology is to predict the consequences of species interactions for community structure. In order to predict the consequences of such interactions, we must understand the underlying mechanisms of the interactions (Schoener 1986, Tilman 1987, Werner 1998). These mechanisms, in turn, are largely functions of individual traits such as behavior, morphology, and life history. Thus, it is crucial to develop the quantitative relations between species' traits and performance in the field if we are going to be able to predict the consequences of species interactions. If particular traits critical to interactions are common to many taxa, then results from such studies may be generalizable to other communities (Werner 1998).

A useful experimental approach to quantifying the relation between species' traits and performance is to take advantage of species whose traits are phenotypically plastic. Traits of individuals can be altered by exposure to different environments and these individuals then subjected to interactions with other species. This approach allows us quantitatively to relate trait changes to consequences for species interactions. One problem with this approach is that important traits such as behavior are often very dynamic and difficult to assess under field conditions. It would be useful, then, to assess if short-term measurements of behavioral responses or other trait changes conducted under controlled laboratory conditions offer any insight into species performance in the field when interacting with other species.

In this study, we examined the relation between short-term behavioral observations in the laboratory and performance under more natural conditions in the field. To accomplish this, we first conducted laboratory experiments examining the responses of two anuran larvae, bullfrogs (Rana catesbeiana) and green frogs (R. clamitans), to three potential predators. These predators, bluegill sunfish (Lepomis macrochirus), mudminnows (Umbra limi), and odonate larvae (Anax junius and A. longipes), are common predators found on the gradient of pond hydroperiod along which the two anuran larvae are distributed (Collins and Wilbur 1979, Werner and McPeek 1994, Wellborn et al. 1996). Because the three predators represent different risks to the anuran larvae (Werner and McPeek 1994; Relyea, unpublished manuscript), we expected that they would generate variation in the magnitude of the behavioral responses of the anuran larvae. We examined activity and spatial responses of the anurans, both of which are ubiquitous responses to predators (Sih 1987, Lima and Dill 1990, Werner 1992). Both of these behaviors also are related to a species' competitive ability and vulnerability to predators (Werner 1991; Werner and Anholt 1993; Relyea, unpublished manuscript).

We then assessed the relation between behavioral responses measured in the laboratory and performance of the anuran larvae when competing under more natural conditions in the field. We hypothesized that the behavioral responses to predators would result in different growth responses and competitive performances. Competition studies conducted in the nonlethal presence of these predators in the field indicated that the behavioral responses in the laboratory were highly correlated with growth responses and relative competitive performance.

METHODS

For both the laboratory and field experiments, we examined larval bullfrog and green frog responses to the absence and presence of different species of caged predators: Anax, Umbra, or Lepomis. Caged predators simulate predation risk without mortality to the experimental populations, since a major part of the predator cue is a water-soluble chemical (Kats et al. 1988, Werner 1991, McCollum and Van Buskirk 1996, Werner and Anholt 1996). Thus, this system provides the opportunity to examine competitive relations between the two anuran prey under a variety of predator combinations that should alter prey behavior in different ways.

Laboratory experiments

We collected multiple clutches of green frog eggs from the University of Michigan's E. S. George Reserve and bullfrog eggs from the Michigan Department of Natural Resources' Saline Fish Hatchery. We hatched the eggs and reared the tadpoles in small, predator-free wading pools for [approximately]3 wk before the experiments were initiated. The experiments were conducted in plastic containers filled with 7 L of aged well water. Each container was equipped with a single predator cage, constructed of two wooden slats suspending a mesh bag in the container (mesh size = 1 x 2 mm). Ten tadpoles (initial mass of bullfrogs, 18.6 [ or -] 1.0 mg; of green frogs, 21.0 [ or -] 1.3 mg) were fed a 3:1 ration of rabbit chow: Tetramin fish flakes three times/wk at a rate of 6% of body mass/d. The laboratory light: dark schedule was 14:10 h.

The first experiment, conducted in 1994, quantified activity and spatial avoidance responses of green frogs and bullfrogs separately in the absence or presence of a single caged Umbra or Anax. We fed the predators 1-2 small tadpoles ([approximately]50 mg) of the appropriate species three times/wk to maintain the predator cue in the water. We quantified the proportion of tadpoles active (moving) and occupying the predator half of the container by approaching a container and scan sampling (Altmann 1974). These observations were repeated 35-40 times over a 5-wk period and included 0-2 observations/d. Examination of the data by weekly intervals provided no evidence of habituation to the predators over time.