Loss of specificity: native but not invasive populations of Triadica sebifera vary in tolerance to different herbivores
- First Online:
- Cite this article as:
- Carrillo, J., McDermott, D. & Siemann, E. Oecologia (2014) 174: 863. doi:10.1007/s00442-013-2807-4
- 379 Views
During introduction, invasive plants can be released from specialist herbivores, but may retain generalist herbivores and encounter novel enemies. For fast-growing invasive plants, tolerance of herbivory via compensatory regrowth may be an important defense against generalist herbivory, but it is unclear whether tolerance responses are specifically induced by different herbivores and whether specificity differs among native and invasive plant populations. We conducted a greenhouse experiment to examine the variation among native and invasive populations of Chinese tallow tree, Triadica sebifera, in their specificity of tolerance responses to herbivores by exposing plants to herbivory from either one of two generalist caterpillars occurring in the introduced range of Triadica. Simultaneously, we measured the specificity of another defensive trait, extrafloral nectar (EFN) production, to detect potential tradeoffs between resistance and tolerance of herbivores. Invasive populations had higher aboveground biomass tolerance than native populations, and responded non-specifically to either herbivore, while native populations had significantly different and specific aboveground biomass responses to the two herbivores. Both caterpillar species similarly induced EFN in native and invasive populations. Plant tolerance and EFN were positively correlated or had no relationship and biomass in control and herbivore-damaged plants was positively correlated, suggesting little costs of tolerance. Relationships among these vegetative traits depended on herbivore type, suggesting that some defense traits may have positive associations with growth-related processes that are differently induced by herbivores. Importantly, loss of specificity in invasive populations indicates subtle evolutionary changes in defenses in invasive plants that may relate to and enhance their invasive success.
KeywordsExtrafloral nectarExotic speciesGeneralistPlant-herbivore interactionsTradeoffs
Tolerance to herbivory is the ability of a plant to compensate for the negative fitness effects of herbivore damage, which is in contrast to resistance to herbivory, the ability of a plant to avoid the negative effects of herbivore damage (Strauss and Agrawal 1999). Plants employ both tolerance and resistance strategies simultaneously and both strategies can be important to plant fitness (Núñez-Farfán et al. 2007; Carmona and Fornoni 2013), yet the relative induction of each strategy may depend upon the herbivore doing the damage. It is clear that the induced-resistance responses of plants to different herbivores can be specific, e.g., dependent on the feeding mode of the herbivore (reviewed in Ali and Agrawal 2012). Currently, it is unresolved if plant tolerance to herbivory is similarly specific (but see Gavloski and Lamb 2000), rather than a generalized regrowth response independent of damage type or herbivore identity (reviewed in Fornoni 2011). Moreover, we know very little about the evolution of specificity of plant responses to herbivores because there are few studies that examine the specificity of both resistance and tolerance responses with replicated, divergent plant populations (Agrawal 2011; but see Garrido et al. 2012).
Biotic differences among the ranges of native and introduced plant populations are expected to lead to selection favoring divergent plant traits among populations, specifically lowered resistance to herbivores and increased competitive ability, as specialist herbivores are often lacking [e.g., evolution of increased competitive ability hypothesis (Blossey and Nötzold 1995)]. Plant tolerance is expected to depend on the relative strength of selection from different types of herbivores and its relationship to resistance traits and plant growth (ecological and allocation costs of defense), though theory on how tolerance traits will respond during plant invasions is still being developed (e.g., Orians and Ward 2010; Fornoni 2011). In a non-invasive species, Agrawal and Fishbein (2008) observed a macroevolutionary relaxation of resistance strategies against specialists in favor of tolerance strategies, while resistance traits against generalists showed phylogenetic escalation in the milkweed genus, Asclepias (Agrawal et al. 2009). Milkweed specialists commonly sequester secondary metabolites involved in plant resistance for their own use, thus reducing the benefit of some plant-resistance traits and increasing the viability of tolerance traits. Müller-Schärer et al. (2004) and Joshi and Vrieling (2005) have suggested that in the introduced ranges of exotic plants, native generalist enemies may be present that exert selection on introduced species; in the absence of specialists, this scenario has been hypothesized to lead to increased allocation to defense against generalist herbivores and decreased allocation to defense against specialists. If defenses against specialists commonly involve tolerance mechanisms (e.g., Carmona and Fornoni 2013), we might expect introduced populations to evolve higher resistance and lower tolerance than native populations when generalists dominate the herbivore community. Similarly, as resistance and tolerance traits sometimes tradeoff (e.g., Leimu and Koricheva 2006), increased resistance of invasive populations could lead to lower tolerance (see Oduor et al. 2011 for an example of increased resistance and decreased tolerance in invasive populations of Brassica nigra).
Alternately, plant tolerance of herbivory may be expected to increase in invasive populations compared to native populations (Fornoni 2011). For example, if tolerance traits experience correlated selection with growth-related traits, increased selection for competitive ability versus defense against herbivores could increase plant tolerance in invasive populations. Wang et al. (2011) found that faster growing and more competitive invasive populations of Chinese tallow tree had decreased resistance to both a specialist beetle and a specialist caterpillar, but increased tolerance to the specialists compared to native populations in terms of biomass production. Likewise, in a study that examined the combined effects of specialist versus generalist herbivory and competition, Huang et al. (2012) found that the more vigorous, invasive populations of Chinese tallow tree had greater plasticity in growth responses than native populations, displaying greater tolerance to a generalist herbivore than a specialist herbivore, but only when competing with a strong versus a weaker neighbor. In another invasive plant, Stastny et al. (2005) found that larger and more vigorous introduced populations of Senecio jacobea exhibited lower resistance but higher tolerance of damage from a specialist flea beetle, compared to native populations.
Additionally, evidence from a non-invasive, woody hybrid willow indicates that biomass tolerance (compensatory regrowth) and biomass in undamaged controls are significant predictors of reproductive tolerance of saplings (Hochwender et al. 2012). Although it is difficult to determine the relationship between biomass tolerance and lifetime reproductive tolerance in long-lived perennial species, the positive relationship between biomass and reproductive fitness in woody plants in general indicates that the ability to accumulate biomass before and after damage is an important component of reproductive tolerance to herbivory (Stevens et al. 2007). Therefore, and perhaps especially for woody species, selection for increased growth in invasive populations may indirectly select for increased tolerance to herbivory.
We exposed native and invasive populations of Chinese tallow tree (Triadica sebifera) to herbivory by either one of two generalist caterpillars to examine population variation in the specificity of defense responses. We asked:
Is plant tolerance specific (dependent on herbivore identity)?
Do native and invasive populations differ in their tolerance or specificity of tolerance to different generalist herbivore species?
Do native and invasive populations differ in induced resistance [extrafloral nectar (EFN) production] or specificity of resistance to different generalist herbivore species?
What is the relationship between tolerance and induced EFN production and does this relationship depend on plant population origin or herbivore identity?
Materials and methods
Chinese tallow tree [Triadica sebifera (L.) Small (synonyms: Sapium sebiferum (L.) Roxb. (Euphorbiaceae), “Triadica” hereafter] is an aggressive invader throughout the southeastern US where it displace native plants (Bruce et al. 1997; Siemann and Rogers 2003). Triadica is native to China and Japan, and was first introduced to Georgia, US in the late eighteenth century, then to Texas, Florida and Louisiana in the early twentieth century (Bruce et al. 1997). Triadica has low levels of herbivory in the introduced range (Siemann and Rogers 2003), which is thought to have led to the relatively low allocation to resistance but high allocation to growth (and thus tolerance of herbivory) of introduced populations compared to native populations (e.g., Huang et al. 2010; Wang et al. 2011; Carrillo et al. 2012a). Both native and invasive Triadica populations produce EFN (Carrillo et al. 2012a), which has been shown to attract predaceous ants and function as indirect resistance against herbivores (e.g., Heil et al. 2001).
Trichoplusia ni (cabbage looper; Lepidoptera: Noctuidae) is a generalist defoliator native to southern North America (Franklin et al. 2010). It annually produces about four generations in its native range. It has been accidentally introduced to Europe and Asia. It is sympatric with Triadica throughout Triadica’s introduced range and in the southern part of Triadica’s native range (Shorey et al. 1962). Spodoptera exigua (beet armyworm; Lepidoptera: Noctuidae) is an exotic generalist herbivore in the US. It is thought to have been introduced from southeast Asia in 1876 (Mitchel 1979). It has been reported to attack >90 plant species across 18 families in the US and is a serious crop pest (Greenberg et al. 2001). It is sympatric with Triadica throughout all of its native and introduced ranges. Many specialist and generalist herbivores damage Triadica in its native range, including Spodoptera exigua (personal communication, J. Ding). Damage levels in the native range vary, but can reach full defoliation during herbivore outbreak conditions (personal communication, J. Ding). Triadica experiences extremely low levels of herbivory in its introduced range (e.g., Siemann et al. 2006). Therefore both native and invasive populations may have more evolutionary history with Spodoptera exigua than with Trichoplusia ni. As we do not have replication at the level of native versus exotic herbivores, we are not able to answer questions about herbivore novelty and plant response, but differences in the evolutionary history of Triadica with these generalist herbivores may have the potential to drive patterns of defense induction.
We hand collected seeds from 11 populations in the US and eight populations in China in November and December 2009. Seeds were kept in refrigerator storage from the time of collection until they were prepped for the experiment in spring 2010. We prepared seeds by soaking the seeds in a mixture of water and laundry detergent (10 g L−1) to soften the seeds’ waxy coating, which we then removed by scrubbing. The waxy coating is not involved in seed provisioning. We germinated seeds from each population by planting them 1 cm deep into a well-watered peat-based soil mixture (Metromix) in a 8 × 8 × 3-cm plastic tray. We placed the trays in a climate-controlled greenhouse maintained at 31 °C at Rice University, Texas, US. Once seedlings appeared, we individually separated and planted each seedling 3 cm deep into 4 × 4 × 4-cm pots with fresh Metromix soil.
Seedlings grew until October 2010, at which point we chose 147 (67 China, 80 US) seedlings of similar condition (height and leaf number) to use in our experiment. We randomly assigned plants from each population to one of three treatments in a full factorial design: addition of T. ni, addition of S. exigua, or control (no herbivores), with two to four replicates for each population/treatment combination. We randomly arranged the pots on greenhouse tables and rearranged their order four times over the course of the experiment. Three plants that died over the course of the experiment (one China and two US) were excluded from our analysis, and there were missing data for five plants (one China and four US), which were excluded from individual analyses, leaving a total of 139 plants (65 China, 74 US).
We obtained caterpillar eggs of both T. ni and S. exigua from Bioserv Entomology Division, New Jersey, and incubated the eggs at 31 °C. Approximately 6 days after arrival, caterpillars were in their second instar, and we placed 15–20 caterpillars on each caterpillar treatment plant. We caged each pot with a 4 × 4 × 20-cm nylon mesh cage that prevented the herbivores from escaping. The caterpillars were allowed to consume the plants until the plant was fully defoliated (~3–7 days after caterpillar addition), at which point we removed the caterpillars from the cages. We allowed caterpillars to fully consume plants for two reasons: so that damage among experimental treatments was equivalent; and Triadica can be completely defoliated in its native range during herbivore outbreaks, so this is a realistic defoliation level for Triadica seedlings. Caterpillars that we could not find we assumed had died. No moths eclosed in the cages. Seedlings then overwintered in the greenhouse until data collection in the spring.
We collected data on plant height and leaf number 6 months following herbivore addition, allowing for the regeneration of foliage due to herbivore consumption and winter leaf drop. At that point, we recorded EFN production by counting the number of leaves that were exuding fluid from the EFN glands and calculated the total proportion of leaves producing EFN. We then harvested, dried, and weighed the leaves, stems, and roots separately to determine plant mass and root:shoot ratio.
We examined differences in plant growth (above-, belowground and total mass, as well as root:shoot ratio) and in indirect defense investment (number and percentage of leaves producing EFN) among the three herbivore treatments and among native and invasive populations with a generalized linear model ANOVA (proc glm, SAS 9.0) to conservatively estimate differences among origins using variation among populations. The model included origin, caterpillars, and population nested within origin as fixed effects. We were explicitly interested in variation among populations, and population terms were used as the error terms for testing the effects of origin terms. Adjusted means partial difference tests were conducted to examine differences among means for significant terms with more than two levels.
We examined the relationship between tolerance (damage-control population means of aboveground mass) and induced EFN production (damage-control population means of total number of leaves with EFN) first across all populations (using all population means) and then within the US or China populations (using only population means of US or China). Tradeoffs among defense may be more evident when considering populations that differ substantially in traits or evolutionary history (e.g., Strauss and Agrawal 1999), which is the rationale for examining these relationships across all populations. We then restrict these comparisons to either within US populations or within China populations to detect if tradeoffs differ among population groupings with different evolutionary history. We regressed population means of induced biomass production (herbivore-damage treatments-control, either all herbivores together or split by herbivore treatments) versus population means of induced EFN production (herbivore-damage treatments-control, either all herbivores together or split by herbivore treatments). We additionally examined tradeoffs in the above metrics between the two herbivore species.
Specificity of tolerance
Dependence of tolerance and indirect defense variables on Triadica population origin, caterpillar treatment and their interaction
Source of variation
O × C
Aboveground massa (g)
O × C
Belowground massa (g)
O × C
O × C
Number of leaves with EFN
O × C
Percentage of leaves with EFN
O × C
Plants had greater aboveground mass after damage from either herbivore compared to controls [untransformed mean ± SE 0.77 g ± 0.04 (T. ni) and 0.67 g ± 0.04 (S. exigua) versus 0.53 g ± 0.04 (control); Table 1] and had greater belowground biomass after herbivory from T. ni versus after herbivory from S. exigua or no-herbivore controls [untransformed mean ± SE 1.91 g ± 0.10 (T. ni) versus 1.35 g ± 0.11 (S. exigua) and 1.50 g ± 0.10 (control); Table 1]. Root:shoot ratio was smaller for plants damaged by S. exigua versus plants damaged by T. ni and no-herbivore controls (Table 1; Fig. 1b).
Root:shoot ratio and aboveground biomass differed among native and invasive populations with the main effect of plant origin (Table 1; Fig. 1b). Root:shoot ratio was higher for plants from China (native range) [untransformed mean ± SE 2.94 ± 0.14 (China) vs. 2.35 ± 0.14 (US)], and aboveground biomass smaller [untransformed mean ± SE 0.57 g ± 0.03 (China) vs. 0.74 g ± 0.03 (US)] compared to plants from the US (invasive range).
Specificity of EFN production
Relationship between tolerance and EFN production
Induced growth responses of Triadica depended on herbivore species (Fig. 1), indicative of specificity in herbivore tolerance. This is one of the first demonstrations of specificity of plant tolerance using independent manipulations of different herbivores, and shows that even similarly feeding generalist herbivores can induce different defense responses of plants (but see Gavloski and Lamb 2000; Garrido et al. 2012). Previous studies have shown that plants vary in tolerance response to different types of damage. For example, Boalt and Lehtilä (2007) found that Raphanus raphanistrum plants were less tolerant of mechanical apical meristem damage versus mechanical foliar damage, and tolerances of the two types of damage were not correlated across plant families. Tiffin and Rausher (1999) found a significant positive relationship between tolerance to apical meristem damage and folivory in Ipomea purpurea, while Pilson (2000) found that tolerance to natural levels of seed herbivory and folivory were independent in Brassica rapa. These studies did not experimentally manipulate the two herbivores examined, so the specificity of each response is not independent of the other. Gavloski and Lamb (2000) independently examined the compensation responses of Brassica napus and Sinapsis alba to three biting and chewing herbivores (a specialist flea beetle, a specialist caterpillar, and a generalist caterpillar) and found that plants varied in the degree and timing of their tolerance response. The herbivores they chose are similar but have distinct feeding patterns and impacts on the plants (Gavloski and Lamb 2000). Garrido et al. (2012) demonstrated that different populations of Datura stramonium can have specific tolerance responses to sympatric or allopatric herbivore populations, indicating local adaptation. In the current study, we used different species of caterpillars that were both generalists and that fed similarly (leaf chewing) and experimental plants had 100 % of their leaf area removed by the herbivores. Importantly, our results show that tolerance is more than a generalized response to plant damage, as herbivore identity determined plant responses.
The mechanism of such specificity of tolerance responses is unclear, as the plants in our study received equivalent amounts of similar types of damage. However, we found no correlation between tolerances of the two species, neither overall nor when split by population origin. For resistance traits, herbivore-derived molecules have been implicated in induced responses (e.g., Heil and Baldwin 2002); our results indicate that plants also vary in their tolerance response to specific herbivore triggers, (e.g., saliva or frass). Bergman (2002) found support for this hypothesis in Salix caprea, when she demonstrated that moose saliva plus leaf tearing increased branching, a potential tolerance trait, versus leaf tearing alone. Likewise, Zhang et al. (2007) found an increase in relative growth rate of both Artemisia frigidaa and Leymus chinensis plants with the application of sheep saliva to mechanically damaged tissue versus mechanical damage alone. Originally proposed as an indistinct response to many types of abiotic and biotic damage (Belsky et al. 1993), with this study we now have increased evidence that tolerance via regrowth after tissue loss is not only responsive to herbivory but also to different types of herbivory.
The specificity of tolerance responses varied among populations, indicating evolutionary divergence in tolerance of herbivory via induced growth responses. Native, Chinese Triadica populations were more specific in their tolerance responses to herbivore species than invasive, US Triadica populations, as Chinese populations responded more strongly in terms of biomass production to T. ni herbivory than to S. exigua herbivory, while US populations responded similarly to damage from either caterpillar species. This is in contrast to a previous study, where invasive populations of Triadica had higher tolerance of damage from a generalist versus a specialist caterpillar, while native populations had no such specificity of response and had similar tolerances to the two herbivore species (Huang et al. 2010). In both studies, invasive populations were more tolerant in general, which appears to be due to the increased vigor of invasive populations compared to native populations, but it is unclear why populations varied in response to the specific herbivores chosen. The loss of specificity seen in the current study may mean that maintenance of recognition mechanisms likely involved in specific induced responses are costly and selected against in the introduced range, where natural herbivory levels are low (Lankau et al. 2004).
We found no such specificity of response for the indirect resistance trait, EFN production. EFN production (number of leaves with active nectaries) was similar after herbivory by either herbivore and did not depend on population origin (Fig. 2). Previous studies have shown that EFN production depended on feeding mode of the herbivore (chewing vs. sucking) (Carrillo et al. 2012b) and that native populations of Triadica produced greater volume and sweeter nectar than invasive populations (Carrillo et al. 2012a). The caterpillars used in this study are both leaf chewers and it is likely that insect predators attracted to EFN should be equally effective at deterring the two leaf-chewing herbivores. However, at the timescale needed to measure a tolerance regrowth response, any differences in specificity of induced EFN production among origins may have already dissipated. Previous studies have shown that induced EFN responses can occur quickly (e.g., Heil et al. 2000; Mondor and Addicott 2003) and that herbivore specific response can last up to 3 weeks after herbivory (Carrillo et al. 2012b). In our study, the herbivores completely defoliated the plants and many plants had no or little leaf tissue 1–3 weeks post-damage, therefore, we could not measure EFN immediately after our damage treatment. Still, we were able to detect an EFN induction response due to herbivory 6 months post-damage (Fig. 2). We cannot rule out that the timing of measurements may have affected our ability to detect an origin effect on induced EFN production. Despite potential synchronization issues of measuring tolerance and induced EFN in the same study, our analysis of tradeoffs between these two defenses could still reveal a cost of past EFN production on current tolerance response. However, defenses that are induced on different time scales may be less likely to display a tradeoff than defenses that are induced on the same timescale.
In another study that examined direct resistance traits of Triadica induced by herbivory by either one of two generalist caterpillar species or one specialist species, induced flavonoids and tannins did not vary across caterpillar species but varied between population origins (Wang et al. 2012). Manzaneda et al. (2010) examined variation in tolerance and resistance among genotypes of Boechera stricta and found that differences in tolerance (early fecundity, flowering time) existed between real and simulated herbivory for some genotypes but not for others, but that tolerance responses to either a non-native specialist or non-native generalist caterpillar were similar within a genotype. That study found distinct induction responses of a chemical defense trait between the two herbivore species. These results show that different defense responses may be differently induced by herbivores even within the same genotype or population. Indirect defenses such as EFN production, which are often facultative (Heil 2008), may select for less specific responses to similarly feeding herbivores than direct defenses or tolerance responses (but see Wang et al. 2013).
We found that tolerance (biomass) and EFN production (total number of leaves producing EFN) were positively correlated in the T. ni treatment but were independent in the S. exigua treatment (Fig. 3a, b), across native and invasive populations. Within each population grouping, we found that the relationship between the two defenses in the T. ni treatment was positive only within invasive, US populations, not within native, Chinese populations (Fig. 3a, b). Tolerance to either caterpillar species was not related nor was resistance to either caterpillar. As such, there was little evidence for costs of tolerance: relationships among traits were dependent on both population origin and herbivore identity but were mostly independent of each other. Furthermore, we found that biomasses in control and herbivore-damaged plants were positively correlated, suggesting that in general, plant vigor is a better predictor of tolerance responses than herbivore resistance and that resistance and tolerance to different herbivores may evolve independently. It is also worth noting that although we did not consider the responses of the different herbivores to our plants, differences in the defense responses of the native and invasive populations of Triadica may be due in some part to variation in the evolutionary responses of the herbivores to the plants.
Overall our results demonstrate that the specificity of defense responses varies depending on the type of defense considered (tolerance was specific, while EFN was not). Documenting variation in plant defenses among populations that differ in biotic communities, such as that which exists among the native and introduced ranges, is one way to increase our knowledge of the evolution of specificity in plant defense against herbivores. Herbivores have been implicated in the evolution of tolerance for some plants (e.g., Lennartsson et al. 1997), but previous work showing high levels of tolerance in invasive populations of Triadica that experience low levels of natural herbivory suggests that selection for growth can drive tolerance responses (Zou et al. 2008). We found that invasive plants were larger after herbivory from T. ni than controls with no damage. Such overcompensation has been theorized to be a plastic growth response to damage in general (e.g., Belsky et al. 1993), and may be common in fast-growing invasive plants. Future work may consider how specificity of response relates to growth-related processes for a better understanding of the costs of specificity and its evolution.
We thank J. Ding for constructive comments and support from an Alliance for Graduate Education and the Professoriate, Ford Foundation, and American Association of University Women fellowship (J. Carrillo), the US National Science Foundation (DEB 0820560; E. Siemann), and the foreign visiting professorship of the Chinese Academy of Sciences (2009S1-30; E. Siemann).
Conflict of interest
The authors declare they have no conflict of interest.