Specificity of extrafloral nectar induction by herbivores differs among native and invasive populations of tallow tree

Abstract

Background and Aims

Invasive plants can be released from specialist herbivores and encounter novel generalists in their introduced ranges, leading to variation in defence among native and invasive populations. However, few studies have examined how constitutive and induced indirect defences change during plant invasion, especially during the juvenile stage.

Methods

Constitutive extrafloral nectar (EFN) production of native and invasive populations of juvenile tallow tree (Triadica sebifera) were compared, and leaf clipping, and damage by a native specialist (Noctuid) and two native generalist caterpillars (Noctuid and Limacodid) were used to examine inducible EFN production.

Key results

Plants from introduced populations had more leaves producing constitutive EFN than did native populations, but the content of soluble solids of EFN did not differ. Herbivores induced EFN production more than simulated herbivory. The specialist (Noctuid) induced more EFN than either generalist for native populations. The content of soluble solids in EFN was higher (2·1 times), with the specialist vs. the generalists causing the stronger response for native populations, but the specialist response was always comparable with the generalist responses for invasive populations.

Conclusions

These results suggest that constitutive and induced indirect defences are retained in juvenile plants of invasive populations even during plant establishment, perhaps due to generalist herbivory in the introduced range. However, responses specific to a specialist herbivore may be reduced in the introduced range where specialists are absent. This decreased defence may benefit specialist insects that are introduced for classical biological control of invasive plants.

INTRODUCTION

Invasive plants are released from potentially co-evolved specialist herbivores (Elton, 1958; Maron and Vila, 2001; Keane and Crawley, 2002) and may encounter novel generalist herbivores in the introduced range (Strauss et al., 2006). Thus, their direct defences (e.g. trichomes, spines and secondary metabolites) and indirect defences [e.g. domatia, volatile organic compounds (VOCs) and extrafloral nectar (EFN)] to herbivory may change from their native to introduced ranges. Selection for increased growth or reproduction in the introduced range (EICA hypothesis; Blossey and Nötzold, 1995) may lead to reductions in defences against natural enemies in populations in the invasive range compared with those in the native range. Although a number of studies have examined variation in direct defences of invasive plants (Müller-Scharer et al., 2004; Orians and Ward, 2010; see review by Bossdorf et al., 2005), little information is available regarding the variation in indirect defences (but see Llusia et al., 2010; Inderjit et al., 2011; Carrillo et al., 2012a; Hanley, 2012).

Indirect defences may be particularly important to invasive plants. As the cost of VOCs and EFN may be lower than the cost of direct defences (O'Dowd, 1979; Katayama and Suzuki, 2011; Villamil et al., 2013), they may be under weaker negative selection when herbivory levels are low. This may lead to a greater relative importance of indirect defences for invasive plants in their introduced ranges compared with their native ranges. However, it has been suggested that indirect defences may be especially effective against toxin-sequestering specialists (Ali and Agrawal, 2012), which might reduce the induction of indirect defences in the introduced range. Furthermore, plant defences could also dramatically change during different ontogeny stages, and induced defence may peak in juvenile plants and decrease with maturity (Boege and Marquis, 2005; Barton and Koricheva, 2009). Thus, study of the induced defence of juvenile plants is important to understand the evolution of defence in invasive plants.

There is evidence that plants have evolved sophisticated mechanisms to detect herbivore-derived molecules that act as signals of herbivore damage and mediate the specificity of plant responses (Heil, 2009; Zas et al., 2011; Meldau et al., 2012). Such a mechanism was demonstrated in a pine weevil, Hylobius abietis (Zas et al., 2011), where the weevil elicited greater induced direct chemical defences in native host plants than in exotic host plants despite comparable induction by methyl jasmonate. The authors suggest that the native species is adapted to the local herbivore but the exotic plant species is not. We assume that such differences would also exist for induced indirect defences (Heil, 2012). Given that specialist herbivores are often lacking in the introduced range and the apparent role of specialist-derived molecules in modifying a general herbivore response, invasive populations of exotic plants may have decreased inducible resistance to specialists, relative to their native populations. To date, however, this has never been tested for any indirect defence trait.

Extrafloral nectar is one of the major forms of indirect defence against herbivory, being described in approx. 1000 plant species ranging over 100 families (Koptur, 1992; Heil, 2008). Extrafloral nectar is generally thought to attract ants (Bentley, 1977), but it may also help sustain other predators (Wooley et al., 2007) and parasitoids (Koptur, 1985). Only three studies to date have examined indirect defences of invasive plants. All of these studies examined EFN production in tallow [Triadica sebifera (L.) Small = Sapium sebiferum (L.) Roxb.] in its invasive range and found higher or comparable constitutive EFN production in native populations and that EFN can be induced by simulated herbivory or a generalist caterpillar, but no differences in induction for invasive vs. native populations were found (Rogers et al., 2003; Carrillo et al., 2012a, b). However, there have been no studies of variation in the induction of indirect defences by specialist and generalist herbivores for any invasive plants from introduced and native populations despite its importance for plant defence theory, particularly during the early ontogenetic stages of plant life history. Especially in the particular case of juvenile plants, few studies have elucidated the general ontogenetic trajectories of this important indirect defence (Barton and Hanley, 2013). Villamil et al. (2013) reported that compared with mature plants, juvenile plants had a significantly different EFN response to herbivores. For Triadica, a previous study found that the plant early stages played an important role in Triadica's invasion success (Bruce et al., 1997). Thus, juvenile invasive plants may represent an excellent opportunity to investigate the evolutionary alternatives of indirect defence during plant development.

In this study, we compared the constitutive EFN production of juvenile plants of native and invasive Triadica populations and used leaf clipping, and damage by a native specialist and two native generalist caterpillars to examine differences in inducible EFN. Specifically, we ask the following questions. (1) Does the type of damage (artificial clipping vs. herbivory; herbivore identity) affect EFN production? (2) How does EFN production of invasive and native populations vary with the type of damage in an early ontogeny stage?

MATERIALS AND METHODS

Study organisms

Tallow [Triadica sebifera (L.) Small = Sapium sebiferum (L.) Roxb. hereafter ‘Triadica’] is a common tree in its native range (China and Japan), growing in cultivation and in the wild (Zhang and Lin, 1994). It was first introduced to Georgia and South Carolina, USA in the late 18th century for agricultural and ornamental purposes, then to Texas, Florida and Louisiana in the early 20th century (Bruce et al., 1997). Triadica aggressively displaces native plants forming monospecific stands (Bruce et al., 1997; Siemann and Rogers, 2003) and has the potential to spread 500 km northward beyond its current invasive range (Pattison and Mack, 2008). Triadica populations from the introduced range are faster growing and less resistant to herbivores than populations in the native range (Siemann and Rogers, 2001; Huang et al., 2010; Wang et al., 2011, 2012a).

Gadirtha inexacta Walker (Lepidoptera: Noctuidae) is a multivoltine moth currently being considered as a potential biological control agent against Triadica in the USA (Wang et al., 2012b). Host range tests in China indicated that it is host specific to Triadica. Cnidocampa flavescens Walker (Lepidoptera: Limacodidae) is a multivoltine generalist moth whose young larvae cause serious damage via window feeding on the lower leaf cuticle before older larvae remove large areas of leaf tissue. In the field, we found that C. flavescens sometimes could consume entire saplings (Wang et al., 2012b). Grammodes geometrica Fabricius (Lepidoptera: Noctuidae) is also a multivoltine generalist moth. It is a generalist defoliator, and caterpillars can cause severe damage to Triadica, especially after the second instar (Wang et al., 2012b).

We collected caterpillars of G. inexacta, C. flavescens and G. geometrica in fields in Wuhan from April to June 2010 and reared them in field cages separately in the Wuhan Botanical Garden, at the Chinese Academy of Sciences, Hubei, China (30 °32'N, 114 °24'E). We used the offspring of these collections for our experiments.

Seeds and plants

We conducted our experiments at Wuhan Botanical Garden in 2010. We used seeds collected from eight populations across south China (hereafter referred to as native populations) and eight populations from the south-east USA (referred to as invasive populations; Supplementary Data Table S1) in late November 2009, collected from 4–10 randomly selected Triadica trees for each population. To evaluate the potential impacts of seed provisioning on seedling performance, 20 seeds from each population were weighed. There was no difference in seed mass of invasive and native populations [nested analysis of variance (ANOVA), F_1,14 =1·652, P = 0·246]. We removed the seed's waxy coats by soaking them in water with laundry detergent (10 g L⁻¹) for 2 d, then we buried the seeds in sand at a depth of 5–10 cm kept in a refrigerator (4 °C) for 40 d.

We planted seeds from the 16 populations on 15 April 2010 and maintained them in a greenhouse. Seeds germinate within 2 weeks, but require a further 6 weeks to pass through the true seedling stage (Hanley et al., 2004) to become established saplings (Jin, 2012). On 20 June 2010, we transplanted saplings individually into pots (16 cm height; 25 cm diameter) containing growing medium (50 % field soil and 50 % sphagnum peat moss) and placed them in an outdoor common garden. We selected similar-sized seedlings for our experiments and randomly assigned them to different treatments. We enclosed the saplings in nylon cages (100 cm height; 27 cm diameter) to exclude herbivores in the common garden. We tested small saplings in our study because a previous study suggested that the early sapling stages play an important role in Triadica's invasion success (Bruce et al., 1997).

Experimental details

On August 10 when the number of average true leaves was 26 and the average sapling height was 50 cm, we artificially damaged plants or subjected them to herbivory by one of three different caterpillar species to compare the responses of plants from invasive vs. native populations to different types of damage. Control plants received no damage treatment. For artificially damaged plants, we used scissors to clip 25 % of the leaf area over 3 d by removing whole leaves towards the base of the stem. For herbivore treatments, we allowed 1–3 caterpillars to damage plants. The number of larvae for each plant varied with caterpillar species because larvae were in different instars at the time of induction treatments and caterpillar species varied in size. When leaf area consumed reached 25 %, we removed caterpillars (after 2 or 3 d). We replicated each treatment eight times, yielding a total of 640 plants (2 continents × 8 populations × 5 treatments × 8 replicates) in the experiment.

Extrafloral nectar production

Eleven days after removing herbivores (24 August), we recorded the number of total leaves, the number of leaves which had EFN on the petiole and the number of leaves with EFN on the underside of the leaf. Previous studies found that Triadica could secrete EFN shortly after simulated damage but the EFN induction could last as long as 3 weeks after the damage (Rogers et al., 2003; Carrillo et al., 2012a, b), thus the timing of measurement was appropriate in this study. We used microcapillary tubes (2 µL) to collect EFN from every leaf and measure the EFN volume. The percentage of soluble solids was estimated with a low volume hand-held refractometer (45-05, Bellingham + Stanley, Kent, UK). The refractive index of a sample is determined by the concentration of dissolved solutes and is expressed as percentage sucrose by mass. The percentage of leaves producing EFN was calculated as (leaves producing petiole EFN + leaves producing underside EFN)/(number of leaves × 100). This index ranges from 0 to 200. The content of soluble solids of EFN was calculated as percentage soluble solids multiplied by EFN volume.

Statistical analyses

We used a series of ANOVAs to examine the effects of damage, artificial vs. herbivore damage, and damage by the three different species of herbivores on the EFN response variables: percentage of leaves with EFN and content of soluble solids of EFN in SAS (v 9·0).

In the analyses focused on the effect of damage, we used the entire data set and parameterized the treatment variable as control or damaged (artificial, specialist – Noctuid, generalist – Noctuid, generalist – Limacodid). The ANOVAs included this two-level treatment variable, geographic origin of populations (invasive or native range) and their interaction. We included population nested in origin as a random variable so we could conduct conservative tests of origin effects that correspond to the null hypothesis that variation between origins is not greater than the variation among populations. In cases in which the origin × treatment term was significant, we conducted adjusted means partial difference tests to examine differences among treatment levels.

In the analyses focused on the effect of artificial vs. herbivore damage, we only included plants from damage treatments (i.e. we excluded control plants) and used a two-level predictor: artificial or herbivore. The ANOVAs included origin, type of damage treatment and their interaction, and population (origin) as a random variable. We do not report origin results because they are equivalent to the tests of differences of means between native-damaged and invasive-damaged in the previous analyses of damage effects. We conducted post-hoc tests for significant origin × treatment results.

In the analyses focused on the effect of damage by different herbivore species, we only included plants from herbivore damage treatments (i.e. excluded control and artificial) and used a three-level predictor. The ANOVAs included origin (not reported – same as means contrasts of invasive-herbivore vs. native-herbivore in control vs. damaged ANOVA), species treatment and their interaction, and population (origin) as a random effect.

RESULTS

In control treatments, invasive populations had a higher proportion of leaves producing EFN than did native populations, but there were no significant differences in content of soluble solids of EFN (Table 1, Fig. 1). Damage (artificial or by one of the three herbivores) significantly increased EFN production (percentage leaves, soluble solids; ‘Induction’ in Table 1; ‘Ind’ in Fig. 1). Increases in the proportion of leaves with EFN were significantly larger for native populations than for invasive populations such that invasive and native populations had similar levels of induced EFN (‘Origin × induction’ in Table 1, Fig. 1A). There were no significant interactions between origin and damage for soluble solids.

Table 1.

The results of the series of ANOVAs to examine the effects of population origin, induction, induction by clipping vs. herbivory (‘Herb’) and induction by the different herbivore species [‘Species’, Generalist – Limacodid (Cnidocampa flavescens), Generalist – Noctuid (Grammodes geometrica), Specialist – Noctuid (Gadirtha inexacta)] on the EFN response variables of juvenile Triadica in terms of percentage of leaves with EFN at petioles or under leaves, and the content of soluble solids of EFN

Factor	Percentage with EFN		Soluble solids
	d.f.	P	d.f.	P
Origin	1,14	0·0240	1,14	0·7820
Induction	1,632	0·0001	1,364	0·0001
Origin × induction	1,14	0·0309	1,14	0·8207
Herb	1,504	0·0421	1,309	0·0028
Origin × herb	1,14	0·9835	1,14	0·8865
Species	2,374	0·0001	2,238	0·0001
Origin × species	2,14	0·0217	2,14	0·0016

Significant values are highlighted in bold (P < 0·05).

Fig. 1.

Effects of induction (Con = constitutive, Ind = average effect of induction) by artificial clipping (Clip), by one of two generalist herbivores [G-Noc, generalist – Noctuid (Grammodes geometrica); G-Lim, generalist – Limacodid (Cnidocampa flavescens)] or by a specialist herbivore [S-Noc, specialist – Noctuid (Gadirtha inexacta)] on the EFN response variables of juvenile Triadica. (A) Percentage of leaves with EFN, and (B) content of soluble solids of EFN. Values are adjusted means + s.e. Different lower-case letters indicate origin × induction means that were significantly different in post-hoc tests (P < 0·05). Different upper-case letters indicate origin × species means that were significantly different in post-hoc tests (P < 0·05).

Although increases were variable (Fig. 1), analyses indicated that in general herbivory increased EFN production (percentage leaves, soluble solids) significantly more than did artificial damage (‘Herb’ in Table 1, Fig. 1). However, this effect was not influenced by origin.

There was a significant effect of herbivore species on EFN production (percentage leaves, soluble solids) that varied with origin (‘Species’ in Table 1). Native populations had significantly greater EFN soluble solids after damage by the specialist compared with either generalist (Table 1, Fig. 1). Invasive populations had more similar EFN production (percentage leaves, soluble solids) after different types of herbivore damage. For both EFN metrics, induction by generalist Limacodid herbivory was on average lower than for the other two species (Table 1, Fig. 1).

DISCUSSION

Extrafloral nectar production of native populations of Triadica was more strongly induced by a specialist herbivore than by either generalist herbivore (soluble solids: 2·1 times higher; Fig. 1B). However, this specific strong induction response to a specialist herbivore was limited in invasive Triadica populations that have had no association with this or any other specialist herbivores for 100–200 years. Our study presents the first evidence that native populations of invasive plants could have distinct induced defence responses toward generalist vs. specialist herbivores that invasive populations lack.

Previous studies indicate that different herbivores can induce distinct plant defence responses (e.g. Ali and Agrawal, 2012; Heil, 2012; Meldau et al., 2012). Those studies identified several herbivore-derived fatty acid–amino acid conjugates or other herbivore-derived molecules that appear to be signals of the type of herbivore damage to plants (Heil, 2009; Hilker and Meiners, 2010; Ali and Agrawal, 2012). Assuming that there is a metabolic cost to the plant to maintain these herbivore defence elicitors, in the absence of specialist herbivores in invasive populations there may be selection against specialist-specific responses. However, in a similar study that looked at the direct defences of Triadica (e.g. foliar tannins and flavonoids), these types of defences did not show such specificity of induction by specialists and generalists; rather, comparable responses to damage were found that did not vary between native and invasive populations (Wang et al., 2012a). This indicates that loss of herbivores in the invasive range does not explain the patterns for all defences; rather there are differences in herbivore-specific induction of direct vs. indirect defences. Simple trade-offs between induced direct and indirect defences do not explain this result (Heil, 2012). Ali and Agrawal (2012) predicted that induced indirect defences may be a particularly effective defence against toxin-sequestering specialists for which direct defences may be ineffective (such as caterpillars). They note in that review that the majority of generalist vs. specialist herbivore studies only included a single specialist and single generalist species. In this study, we included one specialist and two generalist species; greater replication of herbivore species is needed to extend these results to predict general patterns for generalist herbivores vs. specialist herbivores.

Our study shows that invasive Triadica populations had a higher proportion of leaves producing constitutive EFN than native populations (Fig. 1A), suggesting that although the herbivore loads are low in the invaded range (no specialists and few generalists), Triadica has maintained this indirect resistance. The higher constitutive EFN production of invasive populations is in contrast to other studies with Triadica which found that invasive populations had lower (Carrillo et al., 2012a) or similar (Rogers et al., 2003; Carrillo et al., 2012b) constitutive EFN production. There are differences in methodology between this study and the other Triadica studies in terms of soil type (e.g. potting mix vs. field soil), plant age, duration of nectar accumulation and experimental setting (caged in field vs. greenhouse). EFN has previously been shown to depend on environmental conditions such as temperature, light and water availability (Heil, 2009). Our results suggest that there are also important genetic by environment interactions (which may include biotic factors as well as abiotic factors) that can have strong effects on EFN production in the absence of above-ground herbivore damage.

For all forms of induction other than specialist herbivory, we found that EFN responses depended on type of damage but that native and invasive populations had comparable responses to those different types of induction. We found EFN production was induced in Triadica by both artificial damage and herbivory (Table 1, Fig. 1) but that herbivory induced more EFN production than artificial damage, which supports previous results indicating that clipping does not always mimic damage from herbivory (Baldwin, 1990; Pulice and Packer, 2008). We found significant variation in EFN induced by the two generalist herbivores (Table 1, Fig. 1) that was comparable for invasive and native populations. This was unexpected given the difference in induction by specialists vs. generalists for these populations. Together these results suggest that control of the induction of indirect defences may be complex; studying the expression of defence traits in novel conditions may help to understand the ecology and evolution of specificity of herbivore defence. The reduced EFN responses specific to a specialist herbivore in the introduced range may facilitate classical biological control of invasive plants, because decreased EFN induction by specialist insects that are introduced for biocontrol will lower the risk of predation and potentially benefit specialist populations. Compared with the native flora, many non-native plants appear to exhibit an enhanced ability to establish during the juvenile stage, which may play an important role in their ability to invade (Bruce et al., 1997). Consequently, understanding EFN production in juvenile Triadica plants is a vital step not only in elucidating defensive allocation during early ontogeny, but also in understanding how and why this highly invasive species has an establishment advantage over native species.

SUPPLEMENTARY DATA

Supplementary data are available online at www.aob.oxfordjournals.org and consist of Table S1: native and invasive populations of Triadica sebifera that were used in this study.