Oikos 123: 886–896, 2014 doi: 10.1111/j.1600-0706.2013.00946.x © 2014 The Authors. Oikos © 2014 Nordic Society Oikos Subject Editor: Susan P. Harrison. Accepted 24 November 2013 Plant damage and herbivore performance change with latitude for two old-field plant species, but rarely as predicted Tania N. Kim T. N. Kim ([email protected]), Dept of Biological Science, Florida State Univ., Tallahassee, FL 32306-4295, USA. Present address: Great Lakes Bioenergy Research Center, Univ. of Wisconsin, Madison, WI 53726-4084, USA. A long standing hypothesis in biogeography is that latitudinal gradients in plant defenses (LGPD) should arise because selection for plant defenses is greater in the tropics compared to temperate areas. Previous studies have focused on plant traits thought to confer resistance, yet many traits may not actually confer resistance (putative resistance) or interact to influence herbivore performance. In this study, I used a multi-trophic approach to examine relationships between latitude, herbivore pressure, and plant resistance (measured as the growth rates of herbivores) of two oldfield plant species (Solanum carolinense and Solidago altissima) using a field survey across a 12 degree gradient in the eastern US combined with laboratory bioassays measuring the performance of generalist and specialist herbivores. I used structural equation modeling to examine the direct and indirect pathways by which latitude influences herbivore pressure and plant resistance. A latitudinal gradient in plant damage was observed in the expected direction for S. caroliense (damage decreased with latitude), but the opposite relationship was observed for S. altissima. Damage to both plant species was mediated by herbivore abundances, which was in turn influenced by predator abundances. Resistance to herbivores also varied with latitude but the form of the relationship was dependent on herbivore and plant species. There were direct, non-linear relationships between latitude and resistance (for Spodoptera exigua and Schistocerca americana feeding on S. altissima; S. exigua and Manduca sexta feeding on S. carolinense). Herbivore growth rates were also mediated by the density of S. carolinense for Leptinotarsa juncta and S. americana feeding on S. carolinense. There was no relationship between plant resistance and herbivore pressure and no indication of feedbacks. Results from this study indicate that latitudinal variation in plant resistance is complex, possibly constrained by resource availability and tradeoffs in plant defenses. Latitudinal gradients in biological traits and processes have fascinated biogeographers for decades (Darwin 1859, Pianka 1966). Latitudinal variation in climatic, historical, and other correlated features of the environment (e.g. area, energy, spatial domain) can be strong drivers of biodiversity, evolution and ecosystem productivity (Hillebrand 2004). Consequently, the intensity of interspecific interactions may vary with latitude as well. For example, species interactions, including herbivory, competition and predation, tend to be stronger at lower latitudes compared to higher latitudes (reviewed by Schemske et al. 2009). Because herbivore pressure on plants (i.e. plant damage and herbivore abundance) is generally greater in the tropics, it had been hypothesized that low latitude plants should evolve more effective plant defenses compared to high latitude plants as a result of increased selection for these defense traits (Coley and Aide 1991, Coley and Barone 1996, Fig. 1). A prediction of this hypothesis is that there should be a latitudinal gradient in plant defenses (LGPD) with low latitude plants being better defended against herbivores than high latitude plants. 886 Recent empirical studies have looked for evidence of latitudinal gradients in herbivory and plant defenses but results are inconsistent (reviewed by Moles et al 2011a). In a global survey of habitats from 14 countries (including rainforests, deserts, tundra, and temperate forests), Moles et al. (2011b) examined latitudinal gradients of 14 plant resistance traits using standardized protocols and phylogenetic contrasts. Only two of the measured plant traits followed the predicted pattern of greater levels at low latitudes compared to high latitudes; six traits had the opposite pattern with greater levels at high latitudes and six traits showed no relationship with latitude. Mixed results have also been shown in continental-scale studies within single habitat types. The predicted patterns of higher herbivory and plant defenses at low latitudes were observed in salt marshes of North America and Europe (Pennings et al. 2001, 2007) but the reverse pattern was observed in hardwood trees in North American forests (Adams and Zhang 2009) and there were no latitudinal gradients in damage or plant defenses for Acacia shrubs in Eucalyptus forests in Australia (Andrew and Hughes 2005). These inconsistent findings, at both global Predator abundance g j f n Damage b Latitude a Herbivore abundance c d e k h m Plant resistance l Plant abundance i Figure 1. Conceptual model of the relationships between latitude, herbivore pressure and plant resistance. Black, solid arrows represent relationships described by the traditional latitudinal gradient in plant defenses hypothesis, which predicts a negative relationship between latitude and herbivore abundance, a positive relationship between herbivore abundance and damage, and a positive relationship between damage and resistance. Gray arrows represent other relationships that might influence damage and plant resistance such as top–down and bottom–up processes. Dashed arrows represent feedbacks. Gray arrows represent relationships are not examined in traditional latitudinal gradient studies. and continental scales, suggest that the predicted pattern of LGPD may only apply to some habitat types and some plant traits. One reason for inconsistent findings could be that the underlying LGPD hypothesis is too simple. According to the LGPD hypothesis (black, solid arrows in Fig. 1), herbivore abundance should decrease with latitude (path a), herbivore abundance positively influences damage (path b), and the resulting damage positively influences selection for or the production of plant defenses (path c). However, the LGPD hypothesis does not account for feedbacks between plant defenses and herbivore abundances (dashed arrows in Fig. 1). A large body of literature has demonstrated that plant resistance can influence subsequent damage (path d), and variation in resistance has the potential to influence herbivore population dynamics (path e, Karban and Baldwin 1997, Underwood 1999). Reduction in herbivore densities could in turn reduce damage, potentially leading to positive latitudinal relationships with herbivores, damage or plant defenses. Latitudinal relationships may therefore be dynamic, making it unlikely that there would be a simple relationship at any one point in time. A simple and static hypothesis may therefore not effectively describe latitudinal variation in plant defenses given the often complex and dynamic relationship among species. Another reason for inconsistent findings could be different (and imperfect) measures of resistance. Although the plant traits typically measured in latitudinal gradient studies are thought to provide resistance against herbivores (e.g. phenolics, toughness, C:N), not all putative defense traits actually result in resistance to herbivores or do so equally amongst different herbivore species (Carmona et al. 2011, Johnson and Rassman 2011). Also, interactions between different chemical compounds in plants are common and these interactions may result in non-additive (or even unpredictable) effects on herbivores (Hay et al. 1994, Steppuhn and Baldwin 2007). Additionally, biochemical changes in insect guts can help avoid toxicity of plant defenses or alter the efficiency of insect digestion (Jongsma and Bolter 1997, Chown and Nicolson 2004) rendering plant defenses ineffective. In order to better understand the relationships among latitude, herbivore pressure, and plant resistance traits, we need to ensure that our measure of resistance is ecologically meaningful. Rather than examining relationships between latitude and individual plant traits, the use of bioassays that measure changes in herbivore growth rates after feeding on plants can be used to assess plant resistance more holistically. Bioassays integrate the effects of multiple plant traits on herbivores and the ability of herbivores to digest plant material, thus revealing the efficiency of those traits in deterring herbivores from future plant damage. Bioassays may also allow us to examine the ecological (and evolutionary) consequences of those traits for herbivore populations. Finally, previous studies examining LGPD make little attempt to incorporate other factors that might influence plant damage and resistance (e.g. predator and plant abundances, gray arrows in Fig. 1). For example, predators can indirectly influence damage by suppressing herbivore abundance (path f, top–down effects, Hairston et al. 1960) or directly influence damage by altering herbivore feeding behavior and physiology (path g, trait-mediated indirect effects, Abrams 1995). These consumptive and/or nonconsumptive effects can influence the production of plant defenses (Griffin and Thaler 2006). Alternatively, the abundance of plants might indirectly influence damage by affecting herbivore abundances (path h, bottom–up effects) or directly influence damage by modifying the feeding behavior or physiology of herbivores (path l ). If competition for resources constrains the production of defenses (Herms and Mattson 1992), then plant density can directly influence the growth rates of herbivores (path i). Finally, predator and plant abundances can also vary with latitude (path j and k, respectively). Because other trophic interactions can interact with herbivore pressure to influence damage and plant resistance, incorporating their effects using a multi-trophic approach may be required to better understand why LGPD have been found in only a few systems (Johnson and Rasmann 2011, Marczak et al. 2011). In this study, I used a multi-trophic approach to examine relationships between latitude and plant resistance to herbivory using the growth rates of herbivores to measure resistance for two common old-field plant species (Solanum carolinense and Solidago altissima) spanning a broad geographic range ( ⬎ 12 degrees). Specifically, I asked 1) are there latitudinal gradients in herbivore abundances and damage (i.e. herbivore pressure) and 2) does plant resistance (measured as herbivore growth rate) vary with latitude? Because of the many ways in which latitude can affect damage and plant resistance (e.g. through variation in herbivore, predator, and/or plant abundances), I used structural equation modeling (SEM) to ask 3) are latitudinal gradients in damage or resistance simply mediated by herbivore abundances or are more complex indirect pathways involving plant and/or predator abundances also important? 887 Methods Study system Solanum carolinense (Carolina horsenettle) and Solidago altissima (tall goldenrod) are perennial herbaceous plants native to the eastern United States but considered invasive in other parts of the US, Europe and Asia (Werner et al. 1980, Wise et al. 2008). Both species co-occur in early successional old-field habitats and support a diversity of generalist and specialist insect herbivores. The primary herbivores of S. carolinense are specialist leaf chewing insects including the false potato beetle Leptinotarsa juncta, tobacco hornworm Manduca sexta and eggplant flea beetle Epitrix fuscula. Common herbivores of S. altissima include generalist sap suckers (e.g. red goldenrod aphids Uroleucon sp., spittle bugs Clastoptera sp.) and specialist feeders (e.g. gall-making midges Eurosta solidaginis and Rhopalomyia solidaginis, leaf miners Microrhopala vittata and leaf chewers Trirhabda virgata). Both plant species also host generalist insects including the beet army worm Spodoptera exigua, cabbage looper Trichoplusia ni and grasshoppers (Melanoplus sp., Schistocerca americana, Aptenopedes sp.). Both plants have chemical and physical traits suspected to deter herbivores (Bosio et al. 1990, Cipollini and Bergelson 2002), and herbivores are known to affect plant performance (Root 1996, Wise and Sacchi 1996), population growth (Carson and Root 1999, Underwood and Halpern 2012), and natural selection (Meyer 1993, Wise and Cummins 2006) of both plant species. All species hereafter will be referred to by genus name. Field survey In 2008, I surveyed 18 old-fields at six locations spanning more than 12 degrees of latitude from Florida to New York, USA (Fig. 2). At each location, I surveyed three early successional fields (⬍ 5 years post disturbance). Field sizes were at least 50 ⫻ 50 m in area, and separated by 2 km. At each field, I estimated the local abundance of Solanum and Solidago using two 50 ⫻ 1 m belt transects, separated by 20 m. I measured leaf damage (percent area removed on all leaves) and recorded herbivores found on 20 randomly selected Solanum and Solidago plants within each belt transect. I characterized the insect herbivore and predator communities at the fieldlevel with sweep net sampling across the entire length of both belt transects. I identified herbivores and predators to the species or morphospecies level. I started field surveys in Florida (late July) and ended in New York (early August), completing all surveys within a two week period. This period corresponded to the middle of the growing season (peak herbivore activity and plant growth) at each location. Herbivore performance To measure plant resistance, I performed no-choice bioassays in the laboratory using plant material collected from the field and measured herbivore growth rates. Note that I did not measure plant quality (the chemical and nutritional components of the leaves) but rather the net effect of these quality traits on herbivore growth rates. Because the effectiveness of plant defenses against herbivores can vary with herbivore species (e.g. generalists may be more affected by variation in defenses than specialists), I examined the performance of several herbivore species known to feed on Solanum and Solidago. I collected five plants per species at each field and dried plant material in a forced-draft oven at 60°C until constant weight. Plant material was then ground to a fine powder using a mortar Figure 2. Locations of surveyed sites. Six locations were surveyed across a latitudinal gradient in the eastern USA including areas near Tallahassee, FL; Rome, GA; Knoxville, TN; Dublin, VA; Boyce, VA; and Ithaca, NY. Three fields per location were surveyed yielding 18 old-fields. 888 and pestle. While low-temperature drying can result in some loss of chemical compounds in the leaves (e.g. phenolics and volatiles, Abascal et al. 2005, Harbourne et al. 2009), low-temperature drying preserves the nutritional content of the leaves and is commonly used in agronomy and ecosystem sciences. Moreover, even with some loss of chemical compounds in leaves, a relative comparison of resistance to herbivores was possible because plant material was prepared identically across all sites and latitudes. For the bioassays, I reconstituted the plant material (0.5 g) in 10 ml of agar solution (1%) to make a standard insect diet following Pennings et al. (2007). I used a control diet consisting of only agar and water to determine how agar influenced herbivore growth. While the use of standard diets eliminates some physical defenses produced by Solanum and Solidago (trichomes were destroyed but Solanum spines remained intact), both these plant species are known to produce a range of phenolics to deter herbivores (Bosio et al. 1990, Cipollini and Bergelson 2002). A large amount of plant material was required for the diet, therefore I combined plant material from all individuals of each plant species at each field. The performance of herbivores on Solanum and Solidago diets was therefore assessed at the field-level rather than the individual plant-level. Generalist herbivores (Schistocerca and Spodoptera) were tested on Solanum and Solidago diets separately. Specialist herbivores were tested only on their respective host plants; Solanum diet was fed to Leptinotarsa and Manduca, and Solidago diet was fed to Trirhabda. Grasshoppers were obtained from a laboratory colony and raised on lettuce and barley. Both caterpillar species were obtained from commercial rearing facilities and fed on artificial diets. Leptinotarsa were obtained from a laboratory colony and raised on Solanum foliage from plants collected in north Florida. Trirhabda beetles were obtained from various fields throughout New York and Virginia. One month prior to the bioassays, both beetle species were fed on a mixture of live plants collected from the different latitudes and grown in the greenhouse to avoid bias based on prior feeding experience. The performance of all herbivores except Trirhabda was assessed by the relative growth rates (RGR) of insect larvae. Seven to ten larvae per site per insect species were used for the bioassays and were fed either Solanum or Solidago diet from each site. Because RGRs can decrease with increasing size (see Rose et al. 2009 for a plant body size example), I used 2nd instar larvae. For Spodoptera bioassays where a large number of individuals were needed (280 individuals), I supplemented with 3rd instar larvae (approximately 40 individuals). Prior to the start of the bioassays, larvae were starved for three hours to remove food content in their digestive tracts and each larva was weighed separately. Standardized pieces of diet (4 cm3) were offered to each larva in 75–150 ml soufflé cups lined with moist filter paper. Herbivores fed on the diet for three days. After three days of feeding, the diet was removed, and the larvae starved for an additional three hours and weighed. The RGRs were calculated as ln (final larva mass) – ln (initial larva mass). To account for agar effects which varied amongst herbivores, the final RGR for each individual was calculated as the RGR (diet) –mean RGR of the herbivore species (agar). For Trirhabda feeding on Solidago diet, adult females were used because larvae were not available. Because the growth rates of adult insects vary little, performance was characterized using the number of eggs laid after feeding on diet for three days. Bioassays were carried out in four temporal blocks because of the large number of bioassays needed to be performed and not all herbivores were available at the same time. Blocks were done between February–June 2009 using Leptinotarsa, Manduca, Spodoptera and Schistocerca. A separate experiment was done using Trirhabda adults in September 2011. Because the supply of insects was limited, a subset of latitudes was used for the Schistocerca bioassays (Florida 30°N, Georgia 34°N, northern Virginia 39°N, New York 42°N) for both Solanum and Solidago. Similarly, only three of the six latitude locations were used for bioassays using Trirhabda (Florida 30°N, southern Virginia 37°N, New York 42°N). The final number of bioassays was 870 (180 using Leptinotarsa, 180 Manduca, 252 Spodoptera, 168 Schistocerca and 90 Trirhabda). Statistics Field was the experimental unit so damage and the relative growth rates of herbivores were averaged across plant individuals within each field for each plant species (n ⫽ 18 for damage; n ⫽ 9 to 18 for herbivore growth rates). The abundance of plants and herbivores were summed across both belt transects for each field (n ⫽ 18). The mean damage per field was log-transformed to meet the assumptions of normality. General linear models To examine the effects of latitude on herbivore abundances and plant damage, I used separate GLMs using Gaussian error in R (ver. 2.12). I also used separate GLMs to examine relationships between latitude and the RGR of each herbivore species. Preliminary results revealed non-significant effect of temporal blocks therefore temporal block was not included as a variable in the model (Supplementary material Appendix 1–2). The RGRs for individuals from all temporal blocks were then averaged per field. After visual examination of the data, some of the relationships with latitude appeared non-linear therefore I included a second order polynomial latitude term in the models. Linear and quadratic latitude terms were centered to reduce multicollinearity between these two variables. Structural equation modeling (SEM) I used SEM to examine the direct and indirect relationships between latitude, herbivore pressure and plant resistance (Fig. 1). According to the LGPD hypothesis (black arrows in Fig. 1), I predicted 1) a negative relationship between latitude and foliar-feeding herbivore abundance (path a), 2) a positive relationship between foliar-feeding herbivore abundance and damage (path b), and 3) a positive relationship between damage and plant resistance (path c). This model is hereafter referred to as the ‘LGPD model’. I tested an ‘alternative model’ that included relationships between latitude (both linear and quadratic terms), plant, and predator abundances (all paths in Fig. 1). Herbivore abundances can be driven by top–down (predator abundances, path f) and 889 bottom–up factors (host-plant abundances, path h), which may themselves vary with latitude (paths j and k, respectively). Damage can also be influenced by predators (path g) through non-consumptive effects on herbivores. Plant density can influence resistance (path i) and damage (path l) if plants are competing for limited resources. While explicitly examining the importance of feedbacks between plant resistance and herbivore abundance using dynamic models is beyond the scope of this paper, SEM allowed me to incorporate reciprocal interactions and a feedback loop in the alternative model to address the potential for feedbacks in this system (paths d, m and n). An assumption of models with feedbacks (non-recursive models) is that they represent a static feedback process at equilibrium (Grace et al. 2007). Non-recursive models have the potential to be under-identified (model degrees of freedom ⬍ 0), therefore two conditions (order and rank conditions) must be met to avoid identification issues (Kline 2011). The order condition states that the number of excluded variables for each endogenous variable must equal or exceed the total (A) number of endogenous variables minus 1. The rank condition states that each variable in the feedback loop must have a unique set of variables directly influencing them from outside of the loop. In order to satisfy the order and rank conditions, I dropped paths m and n from the initial alternative model because I was not explicitly interested in bottom–up effects on predators and top–down effects on host-plants. This modified alternative model (Supplementary material Appendix 3) satisfied both the order and rank conditions therefore the model was not under-identified. Separate SEM analyses were performed for each herbivore and plant species combination, comparing the fits of the LGPD model against the alternative model (14 SEMs in total). Parameter estimation and model fitting was performed using maximum likelihood procedures in AMOS (ver. 22). I assessed model fit using χ2 and associated p-values, comparative fit index (CFI), and root mean square error of approximation (RMSEA). Good model fit is indicated by p-values ⬎ 0.05, CFI values ⬎ 0.95 and RMSEA values ⱕ 0.05. To improve the fit of the alternative model (Fig. 1 all Solidago 8 45 Gallers 40 Hemiptera 35 Orthoptera 30 Lepidoptera 25 Coleoptera 7 Richness (number of species) Abundance (number of individuals) 50 20 15 5 4 3 10 2 5 1 0 FL GA TN S.VA N.VA (B) 0 NY FL GA TN FL GA TN S.VA N.VA NY Solanum 50 8 Hemiptera 40 Lepidoptera 7 Coleoptera 6 Richness (number of species) 45 Abundance (number of individuals) 6 35 30 25 20 15 10 5 4 3 2 1 5 0 0 FL GA TN S.VA N.VA NY Field locations S.VA N.VA NY (increasing latitude) Figure 3. Latitudinal gradients in the total abundance and richness of herbivores on the plants Solidago altissima (A) and Solanum carolinense (B) (n ⫽ 60 plants per field location). Locations include FL ⫽ Florida, GA ⫽ Georgia, TN ⫽ Tennessee, S.VA ⫽ Virginia (south), N.VA ⫽ Virginia (north), NY ⫽ New York, USA. 890 paths), I removed non-significant pathways (p ⬎ 0.05), starting with the greatest p-value. I also evaluated the residual covariances between variables to determine whether paths should be added. Because the ratio of sample size to number of parameter estimates was less than the suggested ratio of 5 (Grace 2006), I confirmed path coefficients estimated via maximum likelihood with parameters estimated using Bayesian methods which are less sensitive to small sample sizes (Lee and Song 2004). Path coefficients where the 95% confidence intervals included zero were considered non-significant and therefore removed from the model. I compared models using the Akaike information criterion (AIC) and deviance information criterion (DIC) where models with lower AIC and DIC (⬎ 2) are considered better fit models (Burnham and Anderson 2002). This step-wise elimination process continued until all paths remaining in the model where significant or when AIC and DIC values increased after a variable was removed. For models where ΔAIC or DIC ⬍ 2 (models of similar fit), I selected the most parsimonious model (Burnham and Anderson 2002). Results Figure 4. Latitudinal gradients in herbivory (% leaf damage) for Solanum carolinense (gray) and Solidago altissima (black). Lines represent significant relationships with latitude and damage (Solanum: F1,15 ⫽ 6.760, plinear ⫽ 0.02; Solidago: F1,15 ⫽ 4.760, pquadratic ⫽ 0.04). Field surveys Solanum received significantly greater leaf damage (15.0%) than Solidago (7.1%, t ⫽ ⫺4.08, DF ⫽ 31, p ⬍ 0.001). Solidago damage increased, non-linearly, with latitude (quadratic F1,15 ⫽ 4.76, p ⫽ 0.04, Fig. 4). On the other hand, damage to Solanum decreased with latitude (linear F1,15 ⫽ 6.76, p ⫽ 0.02), mostly due to Solanum plants in New York that experienced very little damage (Fig. 4). The total abundance of herbivores per field did not vary with latitude (linear F1,15 ⫽ 0.01, p ⫽ 0.91, quadratic F1,15 ⫽ 1.04, p ⫽ 0.32). Sweep net samples revealed that herbivores within the sampled fields were mostly Hemipterans (77% of the abundance of herbivores). Herbivore counts on plants revealed that Hemipterans made up a large proportion of herbivores observed on Solidago (53% of herbivores on Solidago, Fig. 3); however most of herbivores observed on Solanum were Chrysomelid beetles (92.2% of herbivores on Solanum, Fig. 3). When examining foliarfeeding herbivores only, there was a positive linear relationship with latitude (F1,15 ⫽ 9.99, p ⫽ 0.01). Laboratory bioassays Overall, latitude did affect mean herbivore growth rates, but latitude effects depended on diet specialization, herbivore species, and host plant. Generalist herbivores generally lost weight whereas specialists gained weight (Fig. 5, 6), suggesting that specialists were able to meet their nutritional requirements feeding on Solanum or Solidago, whereas generalists did not. For Solanum specialists, Manduca performed best on mid-latitude plants (Fig. 5, quadratic F1,15 ⫽ 7.29, p ⫽ 0.01) but there was no relationship between latitude and the RGR of Leptinotarsa, (linear F1,15 ⫽ 0.04, p ⫽ 0.84, quadratic F1,15 ⫽ 0.03, p ⫽ 0.85). For generalists feeding on Solanum, Spodoptera performed poorly on midlatitude plants (quadratic F1,15 ⫽ 10.77, p ⫽ 0.01). There was no relationship between latitude and the RGR of Schistocerca (linear F1,9 ⫽ 0.45, p ⫽ 0.51, quadratic F1,9 ⫽ 0.03, p ⫽ 0.85). For generalist herbivores feeding on Solidago (Fig. 5), Spodoptera also performed poorly on mid-latitude plants (similar pattern to a Solanum diet, quadratic F1,15 ⫽ 11.33, p ⬍ 0.01). However, unlike Solanum, a positive, slightly non-linear relationship was observed between the RGR of Schistocerca and latitude (linear F1,9 ⫽ 21.20, p ⬍ 0.01, quadratic F1,9 ⫽ 9.11, p ⫽ 0.01), indicating that northern Solidago plants were more nutritious and palatable to Schistocerca. For the specialist leaf feeding beetle, Trirhabda, a negative relationship was observed between the number of eggs and latitude (linear F1,6 ⫽ 6.05, p ⫽ 0.04). Structural equation models The models describing the LGPD hypothesis were very poor fits to both the Solanum and Solidago data (Supplementary material Appendix 4–5). Instead, the alternative models that included relationships between predator and plant abundances improved model fit (ΔAIC ranged from 12.2 to 43.3, Supplementary material Appendix 4–5). Latitude indirectly influenced damage to both Solanum and Solidago through variation in the abundance of foliar-feeding insect herbivores, but not in the direction predicted by the LGPD (Fig. 7, 8). Herbivores were also negatively influenced by predator abundances, indicating top–down control and predators were positively influenced by latitude. For Solidago, predators also had direct effects on damage (without any changes in the abundance of herbivores, Fig. 8b). Solidago damage was also directly influenced by the density of Solidago (positive relationship) but not through variation in herbivore abundances (Fig. 8b–c). 891 Figure 5. Relationships between latitude and the relative growth rates of Solanum specialists (A) Manduca sexta, (B) Leptinotarsa juncta and generalists (C) Spodoptera exigua and (D) Schistocerca americana on Solanum diet. Growth rates for each species are averaged at each site. p-values of significant latitude effects (both linear and quadratic terms) on mean growth rates. NS ⫽ non-significant latitude effects. Solanum and Solidago resistance to herbivory varied with latitude, but this variation was not dependent on herbivore pressure. For the two of the three assayed herbivores on Solidago (Spodoptera and Schistocerca), there were direct relationships between latitude and plant resistance (Fig. 8). For both caterpillar species feeding on Solanum (Spodoptera and Manduca), there were direct, non-linear relationships with latitude. For the other two Solanum herbivores (Leptinotarsa and Schistocerca), herbivore growth rates were mediated by the density of Solanum. For all assayed herbivores on Solanum and Solidago, there was no relationship between damage and plant resistance and no indication of feedback loops and reciprocal interactions between predictor variables (all relationships were unidirectional). Discussion The latitudinal gradient of plant defenses (LGPD) hypothesis predicts that because herbivore pressure (i.e. herbivore abundance and damage) is generally greater at low latitudes, low latitude plants should have evolved better defenses than 892 plants at higher latitudes. In this study, foliar-feeding herbivore abundance and plant damage did vary with latitude though not necessarily in the expected direction (i.e. greater damage and herbivore abundance in lower latitudes). Furthermore, herbivore abundance and damage were influenced by top–down and bottom–up processes, respectively, and damage in the field did not influence resistance. There was no general pattern of herbivore responses across latitude; the exact relationship between herbivore growth rates and latitude depended on plant and herbivore species. Together, these results suggest that the LGPD hypothesis is too simple to explain the complex realities of plant–insect interactions. Relationships between herbivore abundance, damage and resistance are complex The idea that low latitude plants experience greater herbivore pressure than high latitude plants forms the foundation of the LGPD, but this pattern was not observed in this study. Low latitude sites had lower abundances of foliarfeeding insects and this relationship resulted in opposite damage patterns to Solanum and Solidago (Fig. 4) even though both plant species are similar in life-history. For Solidago, the positive relationship between herbivore abundance and damage may be driven by an increase in the abundance of Trirhabda and Microrhopala (two specialists on Solidago) with latitude. The ranges of both beetle species are limited to northern regions of the US and can reach outbreak levels causing high levels of foliar damage (Carson and Root 2000, Uriarte 2000). Trirhabda and Microrhopala were not found in fields at lower latitudes but within their ranges, their abundances generally increased with latitude (sweep net: F1,7 ⫽ 8.83, p ⫽ 0.02, plant surveys: F1,7 ⫽ 18.49, p ⫽ 0.003). For Solanum, on the other hand, the unexpected negative relationship between herbivore abundance and damage can be partially explained by the fact that not all herbivores exert the same amount of damage. Solanum herbivore body sizes varied by an order of magnitude: from Epitrix (2–3 mm in length) to Manduca (10–70 mm). The abundance of large-bodied herbivores (Manduca and Leptinotarsa) decreased with latitude (sweep net: F1,16 ⫽ 5.021, p ⫽ 0.03, plant surveys: F1,16 ⫽ 15.19, p ⫽ 0.001) while the abundance of smallbodied herbivores (Epitrix) increased with latitude (sweep net: F1,13 ⫽ 4.27, p ⫽ 0.06, plant surveys: F1,13 ⫽ 7.39, p ⫽ 0.02). This variation in species composition could explain why there was an overall negative relationship between damage and herbivore abundance. These results suggest that latitudinal variation in herbivore species composition (or identity of herbivores present) may better characterize selection pressures on plants than the total abundance of herbivores alone (Salazar and Marquis 2012). In order to make the connection between herbivore pressure and plant resistance, we need a measure of resistance that is ecologically and evolutionarily meaningful. Previous studies have measured latitudinal variation in plant traits thought to confer resistance and inferred effects on herbivores (reviewed by Moles et al. 2011a). However, some of these traits may not actually confer resistance nor do so equally for all herbivores (Carmona et al. 2011). This was evident in the bioassays where herbivore species responded differently to the same diet. For example, at mid-latitudes, Manduca generally performed well on Solanum but Spodoptera performed poorly. Variation in responses suggests that measuring plant traits alone may not be an adequate measure of overall resistance. Furthermore, some herbivore species had consistent responses across latitude (Leptinotarsa and Schistocerca) while other herbivore species had varying responses across latitude (Manduca, Spodoptera). Because the responses of different herbivore species vary both within sites and across latitude, the ecological and evolutionary consequences of resistance for both plants and herbivores may be difficult to generalize and predict. In this study, factors other than herbivore abundance influenced damage and resistance which could explain why the expected direction of the LGPD was not observed (Fig. 7, 8). For example, Solidago damage increased with predator abundance and the density of Solidago. Variation in predation risk (i.e. predator abundance) and the foraging environment (i.e. density of plants) could influence the feeding behavior and physiology of herbivores, potentially influencing the amount and distribution of damage to Figure 6. Relationships between latitude and (A) the number of eggs per clutch for a Solidago specialist (Trirhabda virgata) (B) the relative growth rates of generalists Spodoptera exigua and (C) Schistocerca americana on Solidago diet. Growth rates for each species and number of eggs are averaged at each site. p-values of significant latitude effects (both linear and quadratic terms). host plants without changing herbivore densities (Belovsky et al. 2011). Predators also had consumptive effects on herbivores. The indirect benefit of predators to plants may not require the production of plant defenses to deter herbivores. Instead, plants may invest in other traits (e.g. floral resources, shelter) to attract insect predators, rather than chemical and physical defenses. Finally, there were direct effects of latitude on damage and resistance. These direct linkages suggest that there were other unmeasured factors, such as climate and 893 Figure 7. Final models examining relationships between latitude, damage, and plant resistance to Solanum specialists (A) Manduca sexta, (B) Leptinotarsa juncta, and generalists (C) Spodoptera exigua, and (D) Schistocerca americana. Field-level characteristics were identical in all four SEMs (predator abundances, foliar-feeding herbivore abundances, Solanum density, and mean Solanum damage (log-transformed)). Only resistance to each herbivore species (measured as the relative growth rates (RGR) of each herbivore species) varied in each SEM. Solid arrows are significant positive relationships and dashed arrows are significant negative relationships. Arrow thickness is scaled to the standardized path coefficients (Supplementary material Appendix 6). χ2- and p-values denote model fit. resource availability, that influenced damage and resistance to both plants species because latitude cannot directly influence damage and resistance. Because damage and resistance can be affected by a range of biotic and abiotic factors, the LGPD hypothesis may be too simple to describe the complexities of plant–herbivore interactions. Constraints to selection The LGPD hypothesis suggests that variation in herbivore pressure should lead to changes in resistance to herbivores, however I did not observe a relationship between herbivore pressure and plant resistance. One reason for this lack of relationship could be that responses to selection pressures may be constrained. For example, resource availability may limit the ability of plants to produce defenses even if herbivore abundances and damage is high. In two of the feeding assays, Solanum density (which varied with latitude) affected herbivore growth rates (Leptinotarsa and Schistocerca) but these relationships were not influenced by herbivore abundance. If plants from high-density fields experienced greater competition for resources, this could limit the production of plant defenses, resulting in lower resistance and greater performance of some herbivore species (Kim 2012, Halpern et al. in press). Furthermore, because multiple herbivores are known to feed on Solanum and Solidago in the field (⬎ 30 species described on Solanum (Wise 2007) 894 and ⬎ 100 species described on Solidago (Root 1996)), diffuse coevolution could constrain the evolution of resistance to all herbivores (Wise and Rausher 2013). Another reason for the lack of latitudinal concordance between herbivore pressure and plant resistance may be that tolerance varies with herbivore pressure instead of resistance. The degree to which plants in a particular location express tolerance versus resistance could influence herbivore abundances (Stinchcombe 2002, Fornoni 2011). Both Solanum and Solidago are known to tolerate damage by herbivores and tolerance has been shown to vary with environmental context such as successional stage (Hakes and Cronin 2012) or intraspecific density (McNutt et al. 2012). To my knowledge, no study has examined latitudinal variation in tolerance, which is likely to be an important missing component in latitudinal gradient studies (Johnson and Rasmann 2011). Summary and future directions In summary, latitudinal relationships between damage and herbivore growth rates were observed, but not in the predicted directions. The LGPD hypothesis suggests that plant defenses should be influenced by latitudinal variation in herbivore pressure. However, plant responses may be influenced by other factors such as predator abundances or constrained by resource availability. To further our multiple points in time and through manipulative experiments may be required. Revisiting the original LGPD hypothesis to incorporate some of the inherently dynamic and complex nature of natural selection and population dynamics through the use of mathematical models would be one fruitful approach. Acknowledgements – I gratefully acknowledge M. Cipollini (Berry College, Georgia, USA), G. Crutsinger and L. Souza (Univ. of Tennessee, Tennessee, USA), M. Wise and D. Carr (Blandy Experimental Farm, Virginia, USA), and S. Campbell (Cornell Univ., New York, USA) for logistical support during latitudinal field surveys. I thank J. Capinera (Univ. of Florida, Florida, USA) for providing grasshoppers for bioassays. I thank J. Stanford for lab and greenhouse assistance. This manuscript was greatly improved by comments from N. Underwood, B. Spiesman, J. Grinath and A. Hakes. The Robert K. Godfrey Endowment Award for the Study of Botany (Florida State Univ., Florida, USA) helped fund this research. References Figure 8. Final models examining relationships between latitude, damage and plant resistance to Solidago specialist (A) Trirhabda virgata, and generalists (B) Spodoptera exigua, and (C) Schistocerca americana. Field-level characteristics were identical in all four SEMs (predator abundances, foliar-feeding herbivore abundances, Solidago density, and mean Solidago damage (logtransformed)). Only resistance to each herbivore species (relative growth rates (RGR) of Spodoptera, and Schistocerca and number of Trirhabda eggs) varied in each SEM. Solid arrows are significant positive relationships and dashed arrows are significant negative relationships. Arrow thickness is scaled to the standardized path coefficients (Supplementary material Appendix 7). χ2- and p-values denote model fit. understanding of why latitudinal gradients in herbivory and plant defenses occur in some systems but not others, future studies should incorporate other factors that affect plant resistance (e.g. resource availability, genetic variation), other forms of defenses (e.g. tolerance, induced defenses, indirect defenses such as predators), and examine how these factors might vary with latitude. Finally, because plant–herbivore interactions are dynamic, where variation in plant defenses can feed back to influence herbivore populations, examining relationships between herbivores and plant resistance at Abascal, K. et al. 2005. The effect of freeze-drying and its implications for botanical medicine: a review. – Phytother. Res. 19: 655–660. Abrams, P. A. 1995. Implications of dynamically variable traits for identifying, classifying, and measuring direct and indirect effects in ecological communities. – Am. Nat. 146: 112–134. Adams, J. M. and Zhang, Y. J. 2009. Is there more insect folivory in warmer temperate climates? A latitudinal comparison of insect folivory in eastern North America. – J. Ecol. 97: 933–940. Andrew, N. R. and Hughes, L. 2005. Herbivore damage along a latitudinal gradient: relative impacts of different feeding guilds. – Oikos 108: 176–182. Belovsky, G. E. et al. 2011. Prey change behaviour with predation threat, but demographic effects vary with prey density: experiments with grasshoppers and birds. – Ecol. Lett. 14: 335–340. Bosio, C. F. et al. 1990. Defense chemistry of Solidago altissima – effects on the generalist herbivore Trichoplusia ni (Lepidoptera, Noctuidae). – Environ. Entomol. 19: 465–468. Burnham, K. P. and Anderson, D. R. 2002. Model selection and multimodel inference: a practical information theoretic approach. – Springer. Carmona, D. et al. 2011. Plant traits that predict resistance to herbivores. – Funct. Ecol. 25: 358–367. Carson, W. P. and Root, R. B. 1999. Top–down effects of insect herbivores during early succession: influence on biomass and plant dominance. – Oecologia 121: 260–272. Carson, W. P. and Root, R. B. 2000. Herbivory and plant species coexistence: community regulation by an outbreaking phytophagous insect. – Ecol. Monogr. 70: 73–99. Chown, S. L. and Nicolson, S. W. 2004. Insect physiological ecology: mechanisms and patterns. – Oxford Univ. Press. Cipollini, D. F. and Bergelson, J. 2002. Interspecific competition affects growth and herbivore damage of Brassica napus in the field. – Plant Ecol. 162: 227–231. Coley, P. D. and Aide, T. M. 1991. Comparison of herbivory and plant defenses in temperate and tropical broadleaved forests. – In: Price, P. W. et al. (eds), Plant–animal interactions: evolutionary ecology in tropical and temperate regions. Wiley, pp. 25–49. Coley, P. D. and Barone, J. A. 1996. Herbivory and plant defenses in tropical forests. – Annu. Rev. Ecol. Syst. 27: 305–335. 895 Darwin, C. 1859. The origin of species by means of natural selection. – John Murray. Fornoni, J. 2011. Ecological and evolutionary implications of plant tolerance to herbivory. – Funct. Ecol. 25: 399–407. Grace, J. B. 2006. Structural equation modeling and natural systems. – Cambridge Univ. Press. Grace, J. B. et al. 2007. Does species diversity limit productivity in natural grassland communities? – Ecol. Lett. 10: 680–689. Griffin, C. A. and Thaler, J. S. 2006. Insect predators affect plant resistance via density- and trait-mediated indirect interactions. – Ecol. Lett. 9: 338–346. Hairston, N. G. et al. 1960. Community structure, population control and competition. – Am. Nat. 94: 421–425. Hakes, A. S. and J. T. Cronin. 2012. Successional changes in plant resistance and tolerance to herbivory in natural landscapes. – Ecology 93: 1059–1070. Halpern, S. L. et al. 2014. Plant-mediated effects of host plant density on a specialist herbivore of Solanum carolinense. – Ecol. Entomol. 39: 217-225. Harbourne, N. et al. 2009. Effect of drying methods on the phenolic constituents of meadowsweet (Filipendula ulmaria) and willow (Salix alba). – LWT-Food Sci. Technol. 42: 1468–1473. Hay, M. E. et al. 1994. Synergisms in plant defenses against herbivores: interactions of chemistry, calcification, and plant quality. – Ecology 75: 1714–1726. Herms, D. A. and Mattson, W. J. 1992. The dilemma of plantsto grow or defend. – Q. Rev. Biol. 67: 283–335. Hillebrand, H. 2004. On the generality of the latitudinal diversity gradient. – Am. Nat. 163: 192–211. Johnson, M. T. J. and Rasmann, S. 2011. The latitudinal herbivory-defence hypothesis takes a detour on the map. – New Phytol. 191: 589–592. Jongsma, M. A. and Bolter, C. 1997. The adaptation of insects to plant protease inhibitors. – J. Insect Physiol. 43: 885–895. Karban, R. and Baldwin, I.T. 1997. Induced responses to herbivory. – Univ. of Chicago Press. Kim, T. N. 2012. Community-level consequences of plant– herbivore interactions. – PhD thesis, Florida State Univ. Kline, R. B. 2011. Principles and practice of structural equation modeling. – The Guilford Press, New York, USA. Lee, S. Y. and Song, X. Y. 2004. Evaluation of the Bayesian and maximum likelihood approaches in analyzing structural equation models with small sample sizes. – Multivar. Behav. Res. 39: 653–686. Marczak, L. B. et al. 2011. Latitudinal variation in top–down and bottom–up control of a salt marsh food web. – Ecology 92: 276–281. McNutt, D. M. et al. 2012. Intraspecific competition facilitates the evolution of tolerance to insect damage in the perennial plant Solanum carolinense. – Oecologia 17: 1033–1044. Meyer, G. A. 1993. A comparison of the impacts of leaffeeding and sap-feeding insects on growth and allocation of goldenrod. – Ecology 74: 1101–1116. Moles, A. T. et al. 2011a. Assessing the evidence for latitudinal gradients in plant defence and herbivory. – Funct. Ecol. 25: 380–388. Supplementary material (available as Appendix oik-00946 at ⬍ www.oikosjournal.org/readers/appendix ⬎). Appendix 1–7. 896 Moles, A. T. et al. 2011b. Putting plant resistance traits on the map: a test of the idea that plants are better defended at lower latitudes. – New Phytol. 191: 777–788. Pennings, S. C. et al. 2001. Latitudinal differences in plant palatability in Atlantic coast salt marshes. – Ecology 82: 1344–1359. Pennings, S. C. et al. 2007. Latitudinal variation in plant– herbivore interactions in European salt marshes. – Oikos 116: 543–549. Pianka, E. R. 1966. Latitudinal gradients in species diversity: a review of concepts. – Am. Nat. 100: 33–46. Root, R. B. 1996. Herbivore pressure on goldenrods (Solidago altissima): its variation and cumulative effects. – Ecology 77: 1074–1087. Rose, K. E. et al. 2009. The costs and benefits of fast living. – Ecol. Lett. 12: 1379–1384. Salazar, D. and B. J. Marquis. 2012. Herbivore pressure increases toward the equator. – Proc. Natl Acad. Sci. USA 109: 12616–12620. Schemske, D. W. et al. 2009. Is there a latitudinal gradient in the importance of biotic interactions? – Annu. Rev. Ecol. Syst. 40: 245–269. Steppuhn, A. and Baldwin, I. T. 2007. Resistance management in a native plant: nicotine prevents herbivores from compensating for plant protease inhibitors. – Ecol. Lett. 10: 499–511. Stinchcombe, J. R. 2002. Can tolerance traits impose selection on herbivores? – Evol. Ecol. 16: 595–602. Underwood, N. 1999. The interaction between induced plant resistance and herbivore population dynamics. – In: Agrawal, A. et al. (eds), Induced plant defenses against pathogens and herbivores: biochemistry, ecology and agriculture. Am. Phytopathol. Soc. Press, pp. 221–229. Underwood, N. and Halpern, S. 2012. Insect herbivores, density dependence and the performance of the perennial herb Solanum carolinense. – Ecology 93: 1026–1035. Uriarte, M. 2000. Interactions between goldenrod (Solidago altissima L.) and its insect herbivore (Trirhabda virgata) over the course of succession. – Oecologia 122: 521–528. Werner, P. A. et al. 1980. The biology of Canadian weeds: Solidago canadensis L. – Can. J. Plant Sci. 60: 1393–1409. Wise, M. J. 2007. The herbivores of Solanum carolinense (Horsenettle) in northern Virginia: natural history and damage assessment. – Southeast Nat. 6: 505–522. Wise, M. J. and Sacchi, C. F. 1996. Impact of two specialized herbivores on reproduction of horse nettle, Solanum carolinense. – Oecologia 108: 328–337. Wise, M. J. and Cummins, J. J. 2006. Strategies of Solanum carolinense for regulating maternal investment in response to foliar and floral herbivory. – J. Ecol. 94: 629–636. Wise, M. J. and Rausher, M. D. 2013. Evolution of resistance to a multiple-herbivore community: genetic correlations, diffuse coevolution, and constraints on the plant’s reponse to selection. – Evolution 67: 1767–1779. Wise, M. J. et al. 2008. Compensation for floral herbivory in Solanum carolinense: identifying mechanisms of tolerance. – Evol. Ecol. 22: 19–37.
© Copyright 2026 Paperzz