Phenotypically Plastic Responses of Larval Tiger Salamanders, Ambystoma tigrinum, to Different Predators

—Studies of prey responses to different predators are needed to investigate costs and benefits of particular antipredator responses and to unravel community-level effects on phenotypic plasticity. We reared laboratory-bred larvae of Arizona Tiger Salamanders, Ambystoma tigrinum nebulosum with either of two common predators, diving beetle larvae (Dytiscus sp.) or dragonfly naiads (Anax junius). Relative to controls, salamander larvae in both predator treatments had shorter snout–vent lengths and deeper tails; these differences may be related to increased swimming ability. In addition, larvae reared with dragonfly naiads had shorter tails than those reared with diving beetle larvae, possibly in response to different predator foraging strategies or differences in strength of selection imposed by each. Salamander larvae from predator treatments weighed less than controls, with salamanders reared with dragonflies weighing the least. This suggests that salamanders respond more strongly to dragonfly naiads than diving beetles and that dragonflies may be a more important predator. Thus, salamander larvae may distinguish between different predators, highlighting the utility of studying effects of multiple predators on phenotypic plasticity of prey. Environmental variation often drives the evolution of phenotypic plasticity, which is common throughout the plant and animal kingdoms (Stearns, 1989). Inducible defenses are a form of phenotypic plasticity driven by selection pressures imposed by predators (Harvell, 1990). Prey exhibit a broad array of inducible defenses, including changes in the chemical composition of plants in response to herbivory (Karban and Baldwin, 1997), spines and other morphological structures in marine and freshwater invertebrates (Harvell, 1984, 1986; Havel, 1987; Spitze, 1992), and changes in tail morphology in anurans (VanBuskirk et al., 1997; VanBuskirk and McCollum, 1999; Relyea, 2001). Amphibians have been the focus of several recent studies of predator-induced defenses (Smith and VanBuskirk, 1995; McCollum and VanBuskirk, 1996; VanBuskirk et al., 1997). Most such studies have been with single-predators, where several species of amphibian larvae show induced changes in shape and color when exposed to predatory dragonfly naiads (McCollum and Leimberger, 1997; VanBuskirk et al., 1997; VanBuskirk and McCollum, 1999; VanBuskirk and Schmidt, 2000). Studying prey responses to single predators, however, may not give an accurate picture of the true costs and benefits of a predator-induced morphology (Relyea, 2003). For example, predator avoidance by Daphnia also increased susceptibility to parasites (Decaestecker et al., 2002). Further, prey may show different responses to different predators, as in the Common Frog (Rana temporaria), whereby tadpoles exhibit a strong behavioral response to predatory dragonfly naiads, but no response to predatory newts (VanBuskirk, 2001). Tadpoles of Wood Frog (Rana sylvatica) produced predator-specific phenotypes when reared with single predators (Relyea, 2003). However, when reared with predator pairs, tadpoles of R. sylvatica developed a phenotype resembling the response when reared with the more dangerous predator alone (Relyea, 2003). Predator-induced changes can also be reflected in larval growth rates, which may be increased or decreased in the presence of predators. Growth rates may be accelerated because larger larvae may remove entirely the threat imposed by gape-limited predators (Alford, 1986; Semlitsch, 1990). In addition, rapid larval growth may allow larvae to increase predator handling time earlier than with slower growth (Formanowicz, 1986). Amphibian larvae with high growth rates can also metamorphose early to minimize their exposure to aquatic predators (Wilbur and Collins, 1973; Werner, 1986). Conversely, growth rate may be inhibited because of decreased activity level, which is a generalized antipredator response in amphibians (Sih, 1992). Decreased activity level commonly leads to reduced feeding, which often translates to decreased growth rate (Sih, 1992; VanBuskirk and Yurewicz, 1998). Herein, we present results of experiments to address phenotypic responses among larvae of the Arizona Tiger Salamander (Ambystoma tigrinum nebulosum) to two common invertebrate predators: dragonfly naiads (Anax junius) and predacious larvae of diving beetles (Dytiscus sp.). We predict differences in tail morphology in Tiger Salamanders in the two predator treatments because dragonflies and diving beetles have different predation strategies. Dragonflies are often ‘‘sit-andwait’’ predators (Corbett, 1980), where initial prey burst speed may be most important for escape. Diving beetles attempt to eat potential prey items upon contact and possibly chase them (AS, pers. obs.); thus, prey endurance may also be important (Nilsson and Svensson, 1994). In addition, we expect that larval growth will be different in both predator treatments relative to a predator-free control, because amphibian larval growth rates have been increased (e.g., to decrease time of vulnerability) or decreased (e.g., because of stress and lowered activity level) in previous studies. 2 Corresponding Author. E-mail: [email protected] MATERIALS AND METHODS Salamander Rearing.—Larvae were full-sibs from a laboratory-bred pair of A. t. nebulosum. Eggs were separated into individual containers within 24 h of laying and reared in dechlorinated water with aeration until hatching. After hatching, 40 salamander larvae were reared individually in each of three treatments (a total of 120 larvae): control container (with cage but no predator present), with caged Anax naiad, or with caged Dytiscid larvae. Both predator treatments thus provided chemical and visual predator cues but precluded physical contact. Salamanders were reared individually in 4 liter white plastic ice cream buckets filled with 3.5 liters of dechlorinated water and fed 0.015 g of brine shrimp daily throughout the experiment at temperatures (16–188C) and light cycle (12:12 L:D) that mimicked natural conditions. Container placement was randomized among shelves in the laboratory. We changed water once per week. Anax naiads and Dytiscid larvae were fed one salamander larva weekly that was genetically unrelated to experimental salamanders. Snout–vent lengths (SVL) and masses of salamanders were measured five weeks posthatching with calipers and a top loading digital balance, respectively. Then, a digital image of each salamander was captured using a Sanyo CCD Video Camera (Model VDC-2524) interfaced with a Wild dissecting microscope and attached to a Pentium computer via an Imagenation digitizing card. We measured tail length (from the cloaca to the tail tip), maximum tail depth (width of tail at its widest point), and maximum width of tail musculature in each salamander with Optimas Version 6.0 software (Media Cybernetics, Inc. Silver Springs, MD, 1998). Salamanders were then euthanized in 0.1% MS-222 and fixed in 10% formalin. Statistical Analyses.—Statistical analyses were performed with SAS Version 8.1 for Windows. We performed a MANOVA on SVL, tail length, tail depth, depth of tail muscle and mass. We then performed individual ANCOVAS, where snout–vent length was used as a covariate for the four remaining variables (tail length, tail depth, depth of tail muscle, and mass). Sequential Bonferroni corrections were performed on individual ANCOVA P-values (because of nonindependence of multiple variables), and for those that were significant, we conducted Fisher’s Least-SignificantDifference multiple comparisons to test for significant differences among the three treatments.