BRANDON T. BARTON 1. School of Forestry and Environmental Studies, Yale University, 370 Prospect Street, New Haven, Connecticut USA

Reports Ecology, 91(10), 2010, pp. 2811 2818 Ó 2010 by the Ecological Society of America Climate warming and predation risk during herbivore ontogeny BRANDON T. BARTON 1 School of Forestry and Environmental Studies, Yale University, 370 Prospect Street, New Haven, Connecticut 06511 USA Abstract. Phenological effects of climate change are expected to differ among species, altering interactions within ecological communities. However, the nature and strength of these effects can vary during ontogeny, so the net community-level effects will be the result of integration over an individual s lifetime. I resolved the mechanism driving the effects of warming and spider predation risk on a generalist grasshopper herbivore at each ontogenetic stage and quantified the treatment effects on a measure of reproductive fitness. Spiders caused nymphal grasshoppers to increase the proportion of herbs in their diet, thus having a positive indirect effect on grasses and a negative indirect effect on herbs. Warming strengthened the top-down effect by affecting spiders and grasshoppers differently. In cooler, ambient conditions, grasshoppers and spiders had a high degree of spatial overlap within the plant canopy. Grasshopper position was unaffected by temperature, but spiders moved lower in the canopy in response to warming. This decreased the spatial overlap between predator and prey, allowing nymphal grasshoppers to increase daily feeding time. While spiders decreased grasshopper growth and reproductive fitness in ambient conditions, spiders had no effect on grasshopper fitness in warmed treatments. The study demonstrates the importance of considering the ontogeny of behavior when examining the effects of climate change on trophic interactions. Key words: climate warming; fitness effects; grasshopper behavior; indirect effects; Melanoplus femurrubrum; ontogeny; phenology; Pisaurina mira; predation risk; spider behavior. INTRODUCTION Phenological responses to climate change are expected to differ among species, thereby altering interactions within ecological communities (Walther et al. 2002, Parmesan 2006). Most phenological research has focused on the immediate effects of altering the timing of a single life-history event on specific ecological interactions (e.g., timing of migration and resource availabilty; Visser and Both 2005). However, many interactions extend throughout an individual s life cycle, and the nature and strength of such interactions can vary during ontogeny (Logan et al. 2006, Yang and Rudolf 2009). Because climate change may alter the development rates of individuals, the relative effect of stage-dependent interactions on long-term ecological processes may also change (Logan et al. 2006). Consequently, understanding the net effect of climate Manuscript received 6 December 2009; revised 10 May 2010; accepted 2 June 2010. Corresponding Editor: D. A. Spiller. 1 Present address: Department of Zoology, University of Wisconsin, 430 Lincoln Drive, Madison, Wisconsin 53706 USA. E-mail: btbarton@wisc.edu 2811 change on ecological communities will require explicit consideration of stage-specific responses of species to climate effects (Yang and Rudolf 2009). I conducted a series of experiments to elucidate how climate warming alters the effects of predation risk on herbivore behavior, growth rate, and relative fitness. I examined ontogenetic stage-specific behavioral responses to temperature and predation risk to determine how climate warming may alter top-down effects in a grassland food web comprised of spider predators, grasshopper herbivores, and herbaceous plants. Field studies in this system have revealed strong top-down control in which a spider (Pisaurina mira) acts as an indirect keystone predator whose effects are mediated by the adaptive foraging of a grasshopper herbivore (Melanoplus femurrubrum; Schmitz 2004). P. mira predation is compensatory to natural grasshopper mortality (Ovadia and Schmitz 2002), and instead affect the system by altering grasshopper diet composition (Schmitz 2004). When exposed to predation risk by P. mira, grasshoppers seek refuge in the competitively dominant herb Solidago rugosa, which is reduced in

2812 BRANDON T. BARTON Ecology, Vol. 91, No. 10 abundance due to increased grasshopper herbivory (Beckerman et al. 1997, Rothley et al. 1997). Grasshoppers respond to predation risk differently during their development (Ovadia and Schmitz 2002). While predator effects may be significant early in ontogeny, these effects attenuate as the end of the growing season approaches and individuals risk not maturing in time to reproduce (Ludwig and Rowe 1990). Warming can accelerate grasshopper development, reducing the duration of nymphal stages and allowing grasshoppers to reach maturity more quickly (Logan et al. 2006). However, quickly developing nymphs may have higher nutritional demands, requiring them to adapt their feeding behaviors in ways that may increase their exposure to predators (Ovadia and Schmitz 2002). Spiders are more sensitive to warming temperatures than grasshoppers (Li and Jackson 1996, Joern et al. 2006), and respond to warming by seeking thermal refuge lower in the plant canopy where temperatures are cooler (Barton and Schmitz 2009). Grasshoppers capitalize on this difference in thermal tolerances by concentrating their foraging efforts to the hottest parts of the day when spiders are less active (Schmitz et al. 1997). These physiological differences between spider and grasshoppers suggest that warming may decrease topdown control and weaken the strength of the indirect effect of predators on plants. However, a recent synthesis of data collected within this system demonstrated the opposite: Warmer summers increased the strength of the indirect effect of predators on grasses and herbs (Barton et al. 2009). Thus, physiological differences alone are insufficient to explain how climate affects interactions within this community. To resolve this contradiction between expected and observed effects of temperature on top-down effects, I report on a series of experiments conducted to gain mechanistic insight as to how climate warming increases the strength of topdown control in this system. METHODS This study was conducted at the Yale-Myers Research Forest in northeastern Connecticut, USA, during the summers (June August) of 2007 and 2008. The summer climate of this region is mild, with daily maximum temperature (T max ) averaging 228C, and monthly precipitation averaging 4 cm (Barton et al. 2009). Climate models predict that temperatures in this region will warm 2 48C and receive as much as 25% more precipitation by 2100 (U.S. Global Change Research Program 2003). Behavioral observations In the summer of 2007 I experimentally manipulated temperature and predation risk in terraria and observed the effect on grasshopper foraging behavior. I constructed eight terraria from 45 3 30 cm rectangular plywood bases that contained sod cut from the same field in which I conducted the field experiment. The sod and vegetation were enclosed by a 75 cm tall piece of aluminum insect screen that was securely attached to each wooden base at the bottom and a 45 3 30 cm insect screen lid at the top to prevent arthropods from escaping during the experiments. I drew a grid of 15 3 15 cm quadrats on each side and top of each terrarium to quantify movement in three-dimensional space. The terraria were placed on bench tops outdoors in a grass field at the Yale-Myers Research Forest camp facility, where electricity was available to power infrared heaters. Terraria remained outside and exposed to ambient light, temperature, and moisture during the duration of the experiment. I randomly assigned treatments to terraria in a fully replicated cross of temperature (ambient or warmed) and predation risk (predators absent or present). The temperature within warmed terraria was raised to an average of 38C above ambient, which is within the range of predicted temperature increase for this region by 2100 (U.S. Global Change Research Program 2003). Warming was achieved by mounting a single infrared heat lamp that emits no visible light (250 W, Exo-Terra Heat Wave Lamp, Mansfield, Massachusetts, USA) level with the top of each terrarium and ;15 cm from the side. By angling the lamp at 458 toward the terraria, I created a vertical temperature gradient that matched the temperature profile within the field (Barton and Schmitz 2009). Finally, each terrarium was fitted with a thermometer to determine daily T max. I collected grasshoppers and spiders from local fields with sweep nets and marked their thorax (grasshoppers) or abdomen (spiders) with different colors of enamel paint (Testors, Rockford, Illinois, USA). Four grasshoppers (all treatments) and two spiders (predation treatment) were stocked into terraria. These densities are within the high end of the natural densities observed in these fields (B. T. Barton, personal observation) and are moderate in comparison to densities used in other experimental designs with this system (see Beckerman et al. 1997, Schmitz et al. 1997). Arthropods were allowed to acclimate to terraria and experimental treatments for one day before beginning observations. At 30-min intervals between 06:00 and 21:00 I recorded each grasshopper s three-dimensional grid location, its behavior (feeding, resting, moving, etc.), and the type of plant (herb or grass) being eaten. I also recorded the grid location of each spider. Individuals were removed and released after each daily observation period. I simultaneously conducted two independent replicates during each day, and replicated the design three times for each developmental stage (except second instar, which was repeated on four days). Before each observation period, I randomized terraria treatment assignments and used a new compliment of individuals to maintain independence among each set of observations. I estimated the treatment effect on each species location with a log-linear model (Barton and Schmitz 2009). For each day I counted the total number of times

October 2010 CLIMATE EFFECTS DURING ONTOGENY 2813 an individual was observed in each 15 3 15 3 15 cm quadrat. I calculated the three-day (four-day for second instar) average occupancy for each three-dimensional quadrat and tested for treatment effects using JMP version 8 (SAS Institute 2009). I used a maximum likelihood approach to determine if temperature altered the frequency of quadrat occupancy of grasshoppers and spiders in three dimensions during the behavioral observations. I also examined the effect of T max on grasshopper and spider average vertical height with linear regressions. I estimated total daily feeding time in minutes for each individual grasshopper by making the assumption that each time I observed a grasshopper feeding, that individual had been feeding during the entire 30-min observational interval (Belovsky and Slade 1986). I multiplied the number of observed feeding bouts by 30 minutes to determine total daily feeding time. I determined diet composition by calculating the proportion of feeding events on herbs and on grasses. I pooled daily values for all grasshoppers in each terrarium to obtain an independent estimate for statistical analysis. Diet composition data were log-transformed, and all data was confirmed to meet the assumptions of appropriate statistical tests before analysis. I examined the effects of predator presence and warming on daily feeding time and diet composition for all grasshoppers and each ontogenetic stage separately using two-way analyses of variance (ANOVA) followed by a Tukey test whenever a significant difference was detected. I pooled data from all grasshopper developmental stages and used linear regression to examine the effect of T max on grasshopper daily feeding time when predators were absent or present, and the effect of mean spider height in the canopy on grasshopper daily feeding time. I calculated the direct effect magnitude of spiders on grasshopper feeding time using the log ratio [ln (F Pþ /F P )] where F Pþ and F P are, respectively, the minutes of feeding in the presence and absence of predators (Schmitz et al. 2000). I tested whether the effect of spiders on grasshopper feeding time (total feeding time and grasses and herbs independently) varied across instars using two-way ANOVA followed by Tukey tests whenever a significant difference was detected. Finally, I pooled values for all instars and tested whether warming altered the average spider effect magnitude on grasshopper feeding time with a paired t test. Field experiment During summer 2008 I conducted a standard field enclosure experiment (Schmitz 2004) to assess the effects of temperature and predators on grasshopper density, body growth, and fitness components. I constructed 20 1 m 2 3 0.8 m cylindrical enclosures using aluminum insect screen secured to vinyl-coated garden fencing for sides and fiberglass insect screen to enclose the top. For the warmed treatments, I passively raised the temperature within the enclosure by securing plastic sheeting (4-mm Film Gard, Covalence Plastics, Minneapolis, Minnesota, USA) to the outer structure of each cage, while leaving the top covered only by screen. The resulting greenhouse effect elevated enclosure temperature by 2.58C to 3.98C, while maintaining a vertical temperature gradient and rainfall patterns similar to those observed in natural field conditions (Barton and Schmitz 2009, Barton et al. 2009). I placed enclosures over naturally growing vegetation, but ensured enclosures contained a similar composition of grasses and herbs. Each enclosure was surveyed for existing M. femurrubrum and P. mira, which were removed when present. Enclosures were spaced ;2 m apart to maintain independence. The enclosures were arrayed in a randomized-block experimental design (five blocks) crossing temperature treatment (ambient or warmed) and predator treatment (with or without predators). In mid-june, I stocked second-instar grasshoppers at a density of 10 individuals per enclosure and adult P. mira spiders at a density of two individuals per enclosure. These densities approximately matched field densities at the time of stocking (B. T. Barton, personal observation). The experiment proceeded for 75 days, at which time grasshoppers had fully developed into adults. I terminated the experiment by removing, counting, and weighing all surviving grasshoppers. I examined the effects of warming and predator presence on grasshopper density and body mass using a two-way ANOVA followed by a Tukey test whenever a significant difference was detected. I brought adult female grasshoppers to the laboratory to estimate components of each individual s fitness. In grasshoppers, maternal effects determine the total number of ovarioles that each female has available to produce oocytes (Bellinger et al. 1987). But, during ovulation, environmental factors and physiological stress can influence whether an oocyte is resorbed or matured into an egg (Sundberg et al. 2001). Ovarioles that produce mature oocytes retain evidence of successful ovulation in the form of a cream-colored follicle resorption body (FRB). However, if an oocyte is resorbed, an orange-colored oocyte resorption body (ORB) is evident in the ovariole, indicating failed egg production (Sundberg et al. 2001). Comparing the relative frequencies of FRBs to ORBs is an appropriate method to determine the relative effect of different experimental treatments on grasshopper fitness, as measured by the proportion of successfully ovulated oocytes (Sundberg et al. 2001). I dissected adult female grasshoppers under a dissecting microscope and counted the number of FRBs and ORBs in a subset of ovarioles haphazardly selected along the length of the reproductive tract of each grasshopper. I sampled 15 ovarioles from each individual, which is approximately onequarter to one-half of the total ovarioles in an adult M. femurrubrum female (Bellinger et al. 1987). I averaged the ratio of FRB:ORB of all females in each enclosure to determine the enclosure average. Ratio data

2814 BRANDON T. BARTON Ecology, Vol. 91, No. 10 FIG. 1. (a) Maximum daily temperature (T max ) did not affect grasshopper (Melanoplus femurrubrum) daily feeding time when predators (spiders, Pisaurina mira) were absent (P. 0.05), but (b) increased temperature increased daily feeding time when predators were present (P, 0.001, r 2 ¼ 0.33). (c) Grasshopper vertical height was unaffected by T max (closed circles, P. 0.05), but spiders moved lower in the plant canopy as T max increased, as indicted by the regression line (open circles, P, 0.001, r 2 ¼ 0.53). (d) By affecting the spiders and grasshoppers differently, warming decreased the spatial overlap between the two species, and grasshoppers responded by increasing daily feeding time (P, 0.001, r 2 ¼ 0.33). Data points represent daily averages by enclosure, and n ¼ 64 for all regressions. were log-transformed and confirmed to meet the assumptions of a two-way ANOVA, which was used to test the effect of predator presence and warming on FRB:ORB. RESULTS Behavioral observations Grasshoppers in all developmental stages did not change their location in three-dimensional space in response to predator presence or warming (log-linear model, P. 0.37). Warming had no effect on mean horizontal location of spiders (log-linear model, P ¼ 0.61), but decreased spider height in the plant canopy (log-linear model, G 2 ¼ 15.82, P, 0.001). Linear regression on pooled data revealed that increasing T max decreased spider height in the canopy (P, 0.001, r 2 ¼ 0.53, n ¼ 64; Fig. 1), and grasshopper daily feeding time increased as spider height decreased (P, 0.001, r 2 ¼ 0.28, n ¼ 64; Fig. 1). Consequently, T max indirectly increased grasshopper daily feeding time when predators were present (P, 0.001, r 2 ¼ 0.33, n ¼ 64; Fig. 1), but had no effect when predators were absent (P ¼ 0.16; Fig. 1). Predation risk and temperature had varying effects on average daily feeding time across grasshopper instars (Fig. 2; see also Appendix A for complete two-way ANOVA table). Two-way ANOVA revealed a significant predator effect on grasshopper feeding time for second (F 1,28 ¼ 3.20, P ¼ 0.004) and third instars (F 1,20 ¼ 3.20, P ¼ 0.004), but not fourth instar, fifth instar, or adults (P 0.072). The warming effect on feeding time was significant only for the fourth instar (F 1,20 ¼ 2.64, P ¼ 0.016; P 0.079 for all other stages). A significant predator 3 warming interaction was revealed in second, third, and fourth instars (P 0.025). Tukey tests revealed that predators decreased the daily feeding time of second, third, and fourth instars under ambient temperatures (P, 0.05) but not warmed treatments (Fig. 2). Two-way ANOVA revealed significant treatment effects on grasshopper diet composition for all nymphal stages (see Appendix A for complete two-way ANOVA table). Predators increased the proportion of herbs in

October 2010 CLIMATE EFFECTS DURING ONTOGENY 2815 FIG. 2. Predation risk and warming effects on grasshopper daily feeding time differed among ontogenetic stages. In ambient conditions (black), spiders decreased daily feeding time for early nymphal stages (second through fourth instars), but not fifth instars and adults. However, in warmed treatments, spiders had no effect on daily feeding time for any developmental stage. The average effect of warming across all developmental stages was to increase daily feeding time by 40% when predators were present. Values are means þ SE. Letters above the bars identify significantly different (a ¼ 0.05) treatment means based on two-way ANOVA followed by a Tukey test. nymphal grasshopper diet (second, F 3,28 ¼ 4.22, P ¼ 0.004; third, F 3,20 ¼ 4.22, P ¼ 0.004; fourth, F 3,20 ¼ 2.99, P ¼ 0.007; fifth, F 3,20 ¼ 3.40, P ¼ 0.003; Fig. 3), but not adults (P ¼ 0.283). There was no warming effect (all P 0.263) or interaction effect (all P 0.810) on grasshopper diet composition at any developmental stage. The effect magnitude of spiders on grasshopper feeding time increased with temperature. Warming weakened the effect magnitude on rates of feeding by grasshoppers on grass (P ¼ 0.037, r 2 ¼ 0.07, n ¼ 64, slope ¼ 0.09) and intensified it on herbs (P ¼ 0.001, r 2 ¼ 0.16, n ¼ 64, slope ¼ 0.50). The effect of spiders on grasshopper feeding time on all plants and grasses and herbs independently differed among instars (all, F 4,54 ¼ 6.54, P ¼ 0.001; grasses, F 4,54 ¼ 10.98, P, 0.001; herbs, F 4,54 ¼ 20.39, P, 0.001), such that the effect of spiders was strongest at early instars and tended to be weaker for later instars and adults (Fig. 4). Warming increased the average effect magnitude of spiders on grasshopper feeding time for all plants (t ¼ 3.37, P ¼ 0.002) and herbs (t ¼ 2.23, P ¼ 0.034), but not grasses (P ¼ 0.63) (Fig. 4). Field experiment End-of-season grasshopper density was unaffected by treatments (two-way ANOVA, P ¼ 0.21; see Appendix B for complete output table). Two-way ANOVA revealed that both warming (F 1,16 ¼ 5.67, P, 0.001) and predators (F 1,16 ¼ 3.89, P ¼ 0.001) affected final grasshopper mass, but there was not a significant interaction effect (P ¼ 0.31). Tukey tests revealed that predators decreased mass in ambient treatments (Tukey test, P ¼ 0.018; Fig. 5), but not warmed treatments (P ¼ 0.17). Two-way ANOVA revealed a significant predator effect on grasshopper relative fitness, as measured by the ratio of FRBs to ORBs (F 1,16 ¼ 2.95, P ¼ 0.009; Fig. 5). Predators reduced the ratio of FRBs to ORBs (i.e., increased proportion of failed egg production attempts) in ambient temperature treatments (Tukey test P ¼ 0.016), but not warmed treatments (P ¼ 0.75). DISCUSSION The grassland communities where this study was conducted are characterized by strong top-down control (Schmitz 2004). Spider predators cause grasshopper herbivores to reduce the proportion of feeding time on grasses and consume more herbs (Beckerman et al. 1997, Schmitz et al. 1997). Thus, spiders indirectly affect the plant community by releasing grasses from herbivory and increasing herbivore damage on herbs (Schmitz et al. 1997). The magnitude of the spider effect is related to summer temperatures (Barton et al. 2009); however, the

2816 BRANDON T. BARTON Ecology, Vol. 91, No. 10 FIG. 3. All nymphal grasshoppers (second through fifth instars) responded to the presence of P. mira spider predators by decreasing the proportion of time spent feeding on grasses and increasing the proportion of time spent feeding on herbs. Warming (ambient vs. warmed temperature) had no effect on grasshopper diet composition. specific mechanism by which warming increases topdown control in this system was previously unknown. Warming strengthened top-down control in this community through direct effects on spider behavior that altered grasshopper feeding time. Grasshopper daily feeding time was unaffected by temperature when predators were absent, but increased in response to warming when predators were present (Fig. 1). This indirect effect arose because warming affected grasshoppers and spiders differently. At cooler, ambient temperatures, grasshoppers and spiders were active within the upper parts of the plant canopy and had a high degree of spatial overlap (Fig. 1). Consequently, spiders exerted a strong negative effect on grasshopper feeding time (Fig. 2, ambient). Warming decreased the spatial overlap between grasshoppers and spiders by affecting these species differently. Warming had no effect on grasshopper position within the plant canopy, but caused spiders to move down into cooler microclimates (Fig. 1; see also Barton and Schmitz 2009 regarding temperature gradient within the plant canopy). Thus, the degree of spatial overlap between grasshoppers and spiders decreased with warming, and grasshoppers responded by increasing daily feeding time (Fig. 1). Although warming reduced the predator effect on grasshopper daily feeding time, warming had no effect on diet and grasshoppers continued to feed predominately on herbs (Fig. 3). Thus, warming increased the indirect effect of spiders on herbs (Fig. 4) because grasshoppers fed longer each day and consumed the diet expected under predation risk. Predation risk effects attenuated during grasshopper ontogeny (Fig. 4). At early ontogenetic stages, grasshoppers were highly vulnerable to predation and so adapted highly risk-adverse strategies to decrease exposure to predators (Fig. 2). However, as they developed into large stages, individuals became less vulnerable to predation such that grasshopper behavior represented a weighting between the need to feed to reach a reproductive size by season s end against the risks of predation. This happened for two reasons. First, predator effects are size dependent and decrease as prey

October 2010 CLIMATE EFFECTS DURING ONTOGENY 2817 proportion of ovulated oocytes that successfully matured into eggs and suggests predators had a negative effect on grasshopper fitness. At the community level, spiders increased grass biomass by 65% and decreased herb biomass by 30% (B. T. Barton, personal observation). Warming, however, altered the way in which grasshoppers balanced the conflicting effects of predation risk and nutritional demands. When spiders were present, grasshoppers engaged in predator avoidance behaviors that reduced feeding time. However, warming decreased grasshopper exposure to spiders by separating the two species spatially (Fig. 1). As a result, nymphal grasshoppers maintained high rates of feeding (Fig. 2) and, consequently, spiders had no effect on grasshopper end of season mass or FRB:ORB (Fig. 5). In warming treatments, spiders had a stronger effect on plants, increasing grass biomass by 110% and decreasing herb biomass by 45% (B. T. Barton, personal observation). Indeed, warming significantly altered the way in which spiders and grasshoppers interact to influence the plant community. I explored the effects of summer temperature on food web interactions because previous synthesis (Barton et al. 2009) indicated temperature had a significant FIG. 4. The effect magnitude of spiders on grasshopper daily feeding time (see Methods: Behavioral observations for details) differed among instars and warming treatments for different types of grasshopper food plants. The x-axis shows grasshopper developmental stage (second through fifth instar and adult) and the pooled average (All). Values are means 6 SE. Letters above the bars identify significantly different (a ¼ 0.05) treatment means based on a two-way ANOVA followed by a Tukey test (individual developmental stages) or paired t test (All). grow and approach a predation size refuge (Ovadia and Schmitz 2002, Logan et al. 2006). Second, grasshoppers are an annual species and must complete reproduction before the end of the summer growing season. The net effect of predation risk throughout grasshopper ontogeny was to reduce grasshopper end-of-season mass and the ratio of FRBs to ORBs (Fig. 5). Although FRB:ORB does not quantify absolute egg production, this approach does indicate that predation risk reduced the FIG. 5. Predators reduced grasshopper relative fitness (top) and end-of-season body mass (bottom) in ambient treatments but had no effect in warmed treatments. Values are means þ SE. Letters above the bars identify significantly different (a ¼ 0.05) treatment means based on two-way ANOVA followed by a Tukey test.

2818 BRANDON T. BARTON Ecology, Vol. 91, No. 10 influence in this community. However, multiple interacting abiotic components of climate, including temperature, precipitation, and others, may change during this century. While this study did not address such complex interactions, including such complexity into future experimental designs is critical. Further, this study was conducted within single grasshopper generations, and therefore could not assess long-term effects on population dynamics or adaptive responses (e.g., evolution or phenotypic plasticity). Resolving such long-term effects is paramount to our ability to accurately predict the effects of global change. This study demonstrated that warming affected this community by altering the stage-dependent life-history trade-offs exhibited by grasshoppers. This translated into reduced direct effect of predators on prey feeding time, but not diet selection, which in turn explains the strengthened indirect effect on plants observed previously (Barton and Schmitz 2009, Barton et al. 2009). These results show that consideration of species phenological shifts alone is insufficient to understand climate effects on trophic interactions in communities. Instead, it will require insight into how warming shapes the ontogeny of behavior that affects the nature and strength of trophic interactions. ACKNOWLEDGMENTS I thank O. Schmitz for motivating this research and for comments and discussion. N. David, F. Douglas, D. Hawlena, and K. Hughes provided invaluable field assistance. This manuscript benefited from comments by two anonymous reviewers. This research was supported by NSF DEB 0910047. LITERATURE CITED Barton, B. T., A. P. Beckerman, and O. J. Schmitz. 2009. Climate warming strengthens indirect interactions in an oldfield food web. Ecology 90:2346 2351. Barton, B. T., and O. J. Schmitz. 2009. Experimental warming transforms multiple predator effects in a grassland food web. Ecology Letters 12:1317 1325. Beckerman, A. P., M. Uriarte, and O. J. Schmitz. 1997. Experimental evidence for a behavior-mediated trophic cascade in a terrestrial food chain. Proceedings of the National Academy of Sciences USA 94:10735 10738. Bellinger, R. G., F. W. Ravlin, and R. L. Pienkowski. 1987. Maternal environment and variation in ovariole number among populations of M. femurrubrum and M. scudderi scudderi. Entomologia Experimentalis et Applicata 44:75 80. Belovsky, G. E., and J. B. Slade. 1986. Time budgets of grassland herbivores: body size similarities. Oecologia 70:53 62. Joern, A., B. J. Danner, J. D. Logan, and W. Wolesensky. 2006. Natural history of mass-action in predator-prey models: a case study from wolf spiders and grasshoppers. American Midland Naturalist 156:52 64. Li, D. Q., and R. R. Jackson. 1996. How temperature affects development and reproduction in spiders: a review. Journal of Thermal Biology 21:245 274. Logan, J. D., W. Wolesensky, and A. Joern. 2006. Temperature-dependent phenology and predation in arthropod systems. Ecological Modelling 196:471 482. Ludwig, D., and L. Rowe. 1990. Life-history strategies for energy gain and predator avoidance under time constraints. The American Naturalist 135:686 707. Ovadia, O., and O. J. Schmitz. 2002. Linking individuals with ecosystems: experimentally identifying the relevant organizational scale for predicting trophic abundances. Proceedings of the National Academy of Sciences USA 99:12927 12931. Parmesan, C. 2006. Ecological and evolutionary responses to recent climate change. Annual Review of Ecology Evolution and Systematics 37:637 669. Rothley, K. D., O. J. Schmitz, and J. L. Cohon. 1997. Foraging to balance conflicting demands: novel insights from grasshoppers under predation risk. Behavioral Ecology 8:551 559. SAS Institute. 2009. JMP. Version 8. Cary, North Carolina, USA. Schmitz, O. J. 2004. From mesocosms to the field: the role and value of cage experiments in understanding top-down effects in ecosystems. Pages 277 302 in W. W. Weisser and E. Siemann, editors. Insects and ecosystem function. Springer Series in Ecological Studies. Springer, New York, New York, USA. Schmitz, O. J., A. P. Beckerman, and K. M. Obrien. 1997. Behaviorally mediated trophic cascades: Effects of predation risk on food web interactions. Ecology 78:1388 1399. Schmitz, O. J., P. Hamback, and A. P. Beckerman. 2000. Trophic cascades in terrestrial systems: a review of the effect of top predator removals on plants. American Naturalist 155: 141 153. Sundberg, S. V., M. H. Luong-Skovmand, and D. W. Whitman. 2001. Morphology and development of oocyte and follicle resporption bodies in the Lubber grasshopper, Romalea microptera (Beauvois). Journal of Orthoptera Research 10:39 51. U.S. Global Change Research Program. 2003. New England regional assessment. U.S. Global Change Research Program, Washington, D.C., USA. hhttp://www.necci.sr.unh.edu/ 2001-NERA-report.htmlii Visser, M. E., and C. Both. 2005. Shifts in phenology due to global climate change: the need for a yardstick. Proceedings of the Royal Society B 272:2561 2569. Walther, G. R., E. Post, P. Convey, A. Menzel, C. Parmesan, T. J. C. Beebee, J. M. Fromentin, O. Hoegh-Guldberg, and F. Bairlein. 2002. Ecological responses to recent climate change. Nature 416:389 395. Yang, L. H., and V. H. W. Rudolf. 2009. Phenology, ontogeny and the effects of climate change on the timing of species interactions. Ecology Letters. 13:1 10. APPENDIX A Two-way ANOVA output for the behavioral experiment (Ecological Archives E091-200-A1). APPENDIX B Two-way ANOVA output for the field experiment (Ecological Archives E091-200-A2).