Variable effects of temperature on insect herbivory

Introduction

Environmental temperature drives a number of important ecological interactions, including competition, predation, and herbivory, by determining the metabolic rates of ectothermic organisms (Vassuer & McCann, 2005; O’Connor et al., 2009; Vucic-Pestic et al., 2011). As metabolic demands increase exponentially with rising temperatures, consumers generally either increase food intake or switch to higher quality diets to offset the rising costs of metabolism (O’Connor, 2009; Lemoine et al., 2013). As a result, predation and herbivory rates tend to increase exponentially with increased temperature (Hillebrand et al., 2009; Vucic-Pestic et al., 2011). However, both consumption rates and fitness can decline precipitously once a species encounters temperatures beyond its thermal optimum (Lemoine & Burkepile, 2012). A suite of adaptive and evolutionary factors determine these thermal optima, such that a given consumer community may contain species with considerably different thermal response curves (Angilletta, Steury & Sears, 2004; Angilletta, 2009). To date, however, few (if any) studies have examined variation in thermal response curves for a community of co-occurring herbivores (Buckley, Nufio & Kingsolver, 2014).

Insect herbivores can be an important top-down force in terrestrial ecosystems, controlling plant biomass (Carson & Root, 2000), maintaining species diversity (Bagchi et al., 2014), reorganizing competitive hierarchies (Kim, Underwood & Inouye, 2013), and enhancing nutrient cycling (Metcalf et al., 2014). Because insects are ectothermic, their physiological rates, including consumption and growth rates, are directly tied to environmental temperature. By extension, top-down control of plant biomass is also likely to be temperature dependent. Theoretical and experimental studies suggest that herbivory rates should increase exponentially with rising temperatures (O’Connor, Gilbert & Brown, 2011). One potential shortcoming of theoretical examinations of temperature-driven herbivore-plant interactions is that they do not often incorporate variation in thermal response curves within an assemblage of herbivore species. In part, this is because thermal response curves for multiple herbivore species on a single host plant remain mostly uncharacterized. Species are often examined singly, and ecologists have focused on a few readily available model species like Spodoptera spp. (Stamp & Yang, 1996) and Manduca sexta (Kingsolver & Woods, 1998). In contrast, the herbivore guild within a community, or even on a single host plant, can vary from 1–100s of species, each with different life histories, climatic niches, evolutionary histories, and dietary needs that may drive vastly different thermal response curves (Buckley, Nufio & Kingsolver, 2014).

Multiple factors aside from evolutionary history and local adaptation can determine the shape of a species’ thermal response curve. For example, plant chemical defenses can become more or less effective at high temperatures depending on the identity of the herbivore, plant, and chemical compounds in question (Stamp & Osier, 1998; Stamp, Yang & Osier, 1997). Similarly, different insect species can become more or less nutrient-limited at higher temperatures, which is also contingent on host plant quality (Kingsolver et al., 2006; Kingsolver & Woods, 1998; Lemoine et al., 2013). Thermal response curves therefore likely differ among herbivore species and within a single herbivore species utilizing different hosts. Given the interest in predicting the effects of climate change on trophic interactions and community structure (Singer, Travis & Johst, 2013; Urban, Tewksbury & Sheldon, 2012), we sought to determine whether temperature influences herbivory in a predictable manner based on a few easily measured variables of plant nutritional quality.

Here, we report thermal response curves of consumption rates for 21 herbivore-plant pairs, encompassing 14 herbivores and 12 plant species (Table 1). We asked two specific questions: (1) what is the extent of within- and among-species variation in thermal response curves of consumption for insect herbivores? and (2) can plant nutritional quality explain variation in thermal response curves? By working with multiple species of both herbivores and plants, we demonstrate that thermal response curves vary substantially both among and within herbivore species. However, we were unable to detect any influence of plant nutritional quality on the overall shape of the thermal response curve across taxa, suggesting that thermal response curves are idiosyncratic and highly variable among plant-herbivore pairs.

Methods

All experiments were conducted at the Smithsonian Environmental Research Center (SERC), in Edgewater, MD, USA from June–August 2012. Using laboratory feeding assays, we evaluated the feeding performance of 14 herbivore species from three Orders (Lepidoptera, Coleoptera, Hymenoptera) on 12 plant species (Fig. 1, Table 1). Herbivores were collected by hand from host plants in the forests and fields on the SERC premises throughout the summer. All herbivores were kept in a cage and fed leaves from the plant species on which they were collected. Individuals were used in feeding assays within 24 h of collection. No individual was used more than once. As herbivores were opportunistically collected, the number of replicates per host plant/temperature combination varied depending on the number of herbivores found (Table S1). Gregarious species (e.g., Hyphantria cuneata) have higher replicate numbers than do rare, non-gregarious species (e.g., Danaus plexippus). A single lepidopteran species could not be identified beyond the Tortricidae family. We focused on folivorous individuals, using larvae from lepidopteran and hymenopteran species and adults only from a single coleopteran species (Chrysochus auratus).

Results

The exponential parameter (b) of overall consumption rates was significantly greater than zero, indicating that overall consumption rates did increase exponentially with temperature (Figs. 2 and 3). However, the increase was restricted to temperatures between 20 and 30 °C (Fig. 1). The 95% CI of the Gaussian parameter (c) narrowly included zero, but the bulk of the posterior distribution for this parameter lay below zero, indicating that overall consumption rates began to level off at temperatures above 30 °C (Pr(<0) = 0.96, Fig. 3). Accordingly, our model predicts relatively little change in overall consumption rates between 30 and 35 °C (Fig. 2). Variance in consumption rates among herbivore-plant pairs also increased substantially with rising temperature. At 20°, variance among mean herbivore-plant consumption rates was 0.45, while at 35° this variance increased to 1.43. Thus, variability in consumption rates among herbivore-plant pairs increased by over 300%. As a result, at 20 °C the predicted mean community-level consumption rates lie between 0.44 and 1.13 g per day (95% CI). Estimates of mean community-level consumption were more uncertain at higher temperatures, lying between 0.93 and 2.13 g per day (95% CI).

Uncertainty regarding overall consumption rates at higher temperatures stems from idiosyncratic thermal response curves among herbivore-plant pairs (Figs. 4 and 5). Six herbivore-plant pairs (Arge scapularis–Ulmus rubra, Chrysochus auratus–Apocynum cannibinum, Hyphantria cunea–Acer negundo, H. cunea–Liquidambar styraciflua, Melanophia canadaria–Lindera benzoin, Papilio troilus–S. albidum) had Gaussian parameters (c) that were moderately or significantly different from zero, indicating decreasing consumption rates at higher temperatures (Figs. 4 and 5). An additional eleven herbivore-plant pairs (Atteva aurea–Ailanthus altissima, Danaus plexippus–Asclepias syriaca, Epimecis hortaria–L. benzoin, E. hortaria–S. albidum, Euchaetes egle–Asclepias syriaca, Malacosoma americanum–Prunus serotina, Melanophia canadaria–Acer negundo, Nematus tibialis–Robinia pseudoacacia, P. troilus–L. benzoin, Saucrobotys futilalis–A. cannibinum, Unidentified Tortricid – L. benzoin) increased consumption with warming throughout the entire temperature range, where the exponential parameter (b) was significantly or moderately different from zero. In some cases the parameter value was small enough that the fit was approximately linear (e.g., Epimecis hortaria–Sassafras albidum, Figs. 4 and 5). An additional four herbivore-plant combinations (Danaus plexippus–A. syriaca, E. hortaria–Liriodenron tulipifera, Euchaetes egle–Apocynum cannibinum, M. canadria–S. albidum, Papilio polyxense–Foeniculum vulgare) showed no detectable change in consumption rate with increasing temperature.

Even within herbivore species, thermal response curves varied considerably. For example, E. hortaria consumption of L. benzoin, and to a lesser extent S. albidum, increased exponentially with temperature (Fig. 4). However, E. hortaria consumption of L. tulipifera did not vary significantly over the observed temperatures (Figs. 4 and 5). Likewise, P. troilus increased consumption of L. benzoin across temperatures, but consumption of S. albidum began to decline at 35° (Fig. 4). We were not able to detect any effect of plant nutritional content on the shape of thermal response curves among herbivore-plant pairs (Fig. 6).

High intra- and interspecific variability among thermal response curves for each herbivore-plant pair led to variable effects of increasing temperatures on potential top-down control of plant biomass (Fig. 7). For example, a 3 °C increase in temperatures resulted in less than a 20% increase in cumulative consumption for twelve herbivore-plant pairs compared to what is predicted for current temperatures, while leading to a >30% increase for four herbivore-plant pairs (Fig. 7). A 5 °C increase in temperature exacerbated this variability, as five herbivore-plant pairs exhibited <20% increase in cumulative consumption and four herbivore-plant pairs exhibited a >50% increase (Fig. 7). Further, increased warming from 3 to 5 °C had highly variable impacts on cumulative consumption rates among herbivore-plant pairs. For example, Atteva aurea nearly doubled its estimated consumption of Ailanthus altissima as warming increased from 3 to 5 °C, whereas estimated consumption by Melanophia canadaria was unaffected by temperature increases beyond 3 °C (Fig. 7). Thus, high intra- and interspecific variability in the herbivore-plant thermal response curves led to high variability in potential climate change effects on top-down control of plant biomass over an entire growing season.

Discussion

Temperature influences herbivory rates via direct effects on insect herbivore physiology. However, it is currently unclear how temperature affects top-down control of plant biomass at the community or species level. Our data suggest that the influence of rising temperature on potential top-down control of plant biomass via herbivory depends upon the identity of the herbivore-plant pair under consideration. Such variability in the relationship between consumption rate and temperature will make it difficult to predict the effects of temperature changes, i.e., climate change, on top-down control of plant biomass.

Theory predicts that herbivory rates should increase exponentially with rising temperature more quickly than primary production, reducing standing plant biomass at higher temperatures (Gillooly et al., 2001; O’Connor et al., 2009; O’Connor, Gilbert & Brown, 2011) However, meta-analyses of thermal response curves report substantial variability among species. Indeed, approximately 40% of the thermal response curves examined by Dell, Pawar & Savage (2011) exhibited curvature, wherein the thermal response curve began to decrease at high temperatures. In our study, 33% of the herbivore-plant pairs exhibited substantial curvature, reducing consumption of plant biomass at high temperatures, thereby contradicting theoretical predictions of exponential increases in top-down control of plant biomass at high temperatures. Often, reduced consumption rates at high temperatures result from metabolic demand exceeding energetic supply, such that energy available for tasks beyond cellular maintenance, such as movement, feeding, or digestion, decreases sharply at high temperature (Somero, 2011). This results in rapid decreases in consumer fitness at temperatures beyond an organism’s thermal optimum (Lemoine & Burkepile, 2012). We show that community-level herbivory rates display the same, albeit much less pronounced, curvature as do some individual species. The slow decline at higher temperatures, rather than a rapid drop-off beyond some threshold value, is a result of species-specific variation in thermal response curves. Almost half (43%) of plant-herbivore pairings did not show signs of decreased consumption at higher temperature, while one showed evidence of declining consumption beyond 30 °C, leading to no net change in overall consumption rates at higher temperatures.

Such variation in thermal response curves makes predicting the effects of temperature changes (i.e., microhabitat variation, seasonal effects, climate change) on herbivore-plant interactions challenging in the absence of species-specific information. Indeed, a generalist herbivore may have as many thermal response curves as host plants (Lemoine et al., 2013) We report similar patterns here. For example, Epimecis hortaria, the tulip tree beauty moth, rapidly increased consumption of both Lindera benzoin and Sassafras albidum with warming, but the increased consumption of S. albidum was much slower. In contrast, E. hortaria showed no relationship between consumption of Liriodendron tulipifera and temperature. Similarly, Papilio troilus increased consumption of both L. benzoin and S. albidum with increasing temperature, but consumption of S. albidum began to decrease at 35 °C, and consumption of L. benzoin showed no curvature.

Given the high variation in thermal response curves among herbivore-plant combinations, predicting the effects of climate change on the top-down control of plant biomass remains challenging. Some studies have ascribed a single thermal response curve to herbivore species, demonstrating that plant biomass will decrease in a warming world as herbivory rates outpace primary production (O’Connor, Gilbert & Brown, 2011) Our results suggest that using a single consumption-temperature relationship for all herbivores can substantially overestimate the impact of climate change on plant biomass. For example, between 20 and 30 °C, both Chrysochus auratus and Saucrobotys futilalis increased consumption of Apocynum cannibinum. However, at 35° C. auratus decreased consumption while S. futilalis continued to increase consumption, resulting in little change in overall consumption rates on A. cannibinum beyond 30 °C.

When we integrated these changes in consumption over a full growing season, we showed that top-down control on plant biomass is likely to increase with increasing temperature but the magnitude of the increase depends on the herbivore-plant combination. Over the course of a summer, simulated warming resulted in >20% increase in cumulative consumption for 10 herbivore-plant pairs, just under half of the pairings examined. Conversely, simulated climate change resulted in >40% increase in cumulative consumption for five herbivore-plant pairs. Overall, the change in consumption ranged from no change to an increase of over 60%. This variability in consumption may explain why studies documenting significant effects of warming on top-down control of plant biomass typically examine one herbivore species (Chase, 1996; Barton, Beckerman & Schmitz, 2009) while studies focusing on entire herbivore communities report weak or negligible effects of warming (Richardson et al., 2002).

Surprisingly, we were unable to detect any influence of plant nutritional quality on the shape of thermal response curves. Based on previous work (Lemoine et al., 2013) we expected consumption rate to increase more rapidly on plants of higher nutritional quality. Conversely, compensatory feeding predicts that consumption rates should increase more rapidly with temperature for plants of low nutritional quality as herbivores attempt to fuel rising metabolic demands (e.g., Williams, Lincoln & Thomas, 1994). Our data suggest that plant nutritional content had little effect on thermal response curves among herbivore species. However, prior work has found that the relationship between temperature and consumption rates within a given species can vary with dietary quality. For example, the Japanese beetle Popillia japonica increased growth and consumption rates at high temperatures only on host plants with high nitrogen and carbon concentrations (Lemoine et al., 2013). This may also be the case in our data. Within an herbivore species, we can distinguish some patterns related to plant quality. For example, the generalist herbivore Epimecis hortaria increased consumption rapidly with warming only on higher nitrogen plant species. Within a plant species, however, patterns were less clear, as particular herbivore species were equally likely to have unimodal or exponential curves when feeding on the same plant. Thus, across all 21 plant-herbivore pairings, we were unable to detect an overall pattern relating plant quality to multiple thermal reaction norms. Thus, dietary quality may be more important for determining thermal response curves within herbivore species but cannot predict the shape of consumption thermal response curves among herbivore and plant species.

We focused on examining herbivore response to increasing temperature while holding plant phytochemistry constant, but rising temperatures might also affect plant phytochemistry. Plant growth rates increase with rising temperatures (Veteli et al., 2002), which could alter nutritional content or concentrations of defensive compounds as plants shuttle more resources into growth (Coley, Bryant & Chapin, 1985). However, studies have shown that the effects of temperature on plant secondary chemistry are highly idiosyncratic among species (Veteli et al., 2002; Zvereva & Kozlov, 2006). Furthermore, although variable temperature can alter nutritional quality within a species, variation caused by temperature is substantially lower than inherent variation among plant species (Aerts et al., 2009). However, the effects of rising temperature on plant chemistry must be considered more completely before applying results such as ours in a climate change context.

One potential caveat of our study is small sample size at many herbivore-plant-temperature combinations. Given that we used field-collected organisms, sample size varied considerably depending on the rarity of the species. Common and/or gregarious species, like E. hortaria, M. canadaria, and H. cunea, have much higher sample sizes than rare or cryptic species, like D. plexippus and A. scapularis. Thus, some herbivore-plant pairs show considerable variability in the estimated thermal response curve and, in some cases, the prediction was heavily influenced by the overall response. However, most work regarding the influence of temperature on herbivory and its interaction with diet quality focus on a few readily available lepidopteran herbivores (Kingsolver & Woods, 1998; Kingsolver et al., 2006). The influence of temperature on herbivory by the majority of the species reported here was heretofore unknown, and our research adds considerably to the body of work documenting the importance of temperature on rates of herbivory.

In summary, we show that herbivory, and therefore potential top-down control of plant biomass, is highly contingent upon environmental temperature. While theoretical predictions suggest that climate change might increase top-down control of plant biomass, our results indicate that the effects of temperature on herbivory rates are highly variable. A single plant species might experience more or less herbivory at higher temperatures, depending on the identity of the herbivores present. Insects often control plant community structure (Carson & Root, 2000) and dominance hierarchies among plant species (Kim, Underwood & Inouye, 2013) Thus, studies documenting the species-specific effects of temperature on insect herbivory levels will be crucial to understanding how climate change might affect community composition in the plant-herbivore assemblages of the future.

Supplemental Information