Physical defences wear you down: progressive and irreversible impacts of silica on insect herbivores

Authors


Correspondence author. E-mail: f.p.massey@sussex.ac.uk

Summary

  • 1Silica in the leaves of grasses can act as a defence against both vertebrate and invertebrate herbivores. The mechanisms by which silica affects herbivore performance are not well characterized. Here we expose an insect herbivore Spodoptera exempta to high-silica diets and test two mechanisms by which silica has been proposed to act as a defence. First, that silica reduces the digestibility of leaves and second, that silica causes wear to insect mandibles, both of which could potentially impact on herbivore performance.
  • 2Silica reduced the efficiency with which S. exempta converted ingested food to body mass and the amount of nitrogen absorbed from their food, leading to reduced insect growth rates. The measure of how efficiently herbivores utilize digested food (ECD) was unaffected by silica.
  • 3These effects occurred even with short-term exposure to silica-rich diets, but they also increased markedly with the duration of exposure and affected late instars more than early instar larvae. This appears to be due to the progressive impacts of silica with longer exposure times and suggests that herbivores cannot adapt to silica defences, nor do they develop a tolerance for silica with age.
  • 4Exposure to silica-rich diets caused increased mandible wear in S. exempta. This effect was extremely rapid, occurring within a single instar, further reducing feeding efficiency and growth rates. These effects on insect growth and feeding efficiency are nonreversible, persisting after the herbivore has switched diets. Up to a third of this residual impact can be explained by the degree of mandible wear caused by previous silica-rich diets.
  • 5The impacts of silica on S. exempta larvae were progressive with exposure time and could not be compensated for, even by switching to a different diet. Thus, herbivores cannot easily adapt to physical defences such as silica, suggesting this defence will have major implications for herbivore fitness.

Introduction

Herbivores consume on average 10–20% annual net primary production in terrestrial ecosystems (Cyr & Pace 1993) and plants have developed many strategies to limit biomass loss to herbivores. Mechanisms by which plants defend themselves against herbivores have received far more attention in the context of chemical defences than physical ones, and the mechanisms by which physical defences impact on herbivore performance remain poorly understood (Hochuli 1996). Physical defences in plants, such as toughness and leaf hairs or spines can affect a broad range of herbivores (Lucas et al. 2000; Hanley et al. 2007), and for several plant taxa, physical defences appear to be the primary determinant of herbivore feeding behaviour. For example, physical defences such as silica are considered to be more important than chemical defences in deterring herbivory on grasses (Vicari & Bazely 1993; Massey, Ennos & Hartley 2007).

There is clear evidence to demonstrate that the high levels of silica found in many plants, particularly grasses, have wide-ranging antiherbivore effects on both invertebrates and vertebrates (McNaughton & Tarrants 1983; Goussain, Prado & Moraes 2005; Massey, Ennos & Hartley 2006; Teaford et al. 2006; Kvedaras et al. 2007). Silica increases the abrasiveness of grass leaves which deters folivores (Gali-Muhtasib, Smith & Higgins 1992; Massey et al. 2006; Massey & Hartley 2006). These impacts are so pronounced that silica is able to alter the feeding preference of herbivores for different grass species (Massey et al. 2006, 2007). Silica defences can also affect herbivore performance. Both insect and mammalian folivores have been shown to have reduced growth rates and digestive efficiency when feeding on silica-rich diets (Goussain et al. 2005; Massey & Hartley 2006, Massey et al. 2006, Kvedaras et al. 2007).

There are two mechanisms suggested for the impacts of silica on herbivores, although both currently lack good experimental support. First, increased abrasiveness of leaves due to silica may increase the wear on herbivore mouthparts (Baker, Jones & Wardrop 1959; Vicari & Bazely 1993). Second, silica is known to decrease the efficiency with which herbivores can digest grass leaves (Massey & Hartley 2006). This could be for a number of reasons: silica may act as a physical barrier within leaves preventing access to nitrogen (O’Reagain & Mentis 1989; Vicari & Bazely 1993); there could be a reduction in the mastication of leaf material by the herbivore, not releasing as much nitrogen from plant tissues; or silica may be physically damaging the digestive tracts of herbivores thus reducing digestive efficiency.

Silica may have played an important role in the evolution of both mammalian and insect grass-feeding herbivores (Chapman 1964; Jernvall & Fortelius 2002). The mouthparts of many groups of herbivores show specific adaptations associated with grass feeding. It was long believed that silica abrades teeth, and that hypsodonty in grazing ungulates and hypselodonty in grazing rodents was an adaptation to this abrasive diet (Baker et al. 1959; Janis & Fortelius 1988; Jernvall & Fortelius 2002), although more recent work disputes this (Sanson, Kerr & Gross 2007). The abrasive nature of silica has also been proposed to drive mouthpart adaptations of certain groups of insect herbivores, including enlargement and specialization of mandibles to cut grass leaves (Chapman 1964; Patterson 1984; Hochuli 1996). Again, the impacts of silica remain unclear because insects renew their mandibles with each moult. Several studies have reported increased mandible wear in lepidopteran larvae fed on cultivars of rice with differing silica contents (Djamin & Pathak 1967; Drave & Lauge 1978; Ramachandran & Khan 1991). However, no controlled studies have yet quantified the wear on mandibles in response to the manipulation of silica concentrations, nor have the potential consequences of silica-driven mandible wear for insect performance been assessed experimentally (Hochuli 1993).

To understand the mechanisms by which silica-based defences impact on herbivore performance, it is important to test and disentangle impacts on herbivore mandibles from the potential impacts on food utilization experimentally. It is also important to assess the time-scale over which these mechanisms act. The impacts of chemical defences on insect herbivores are often greatest at early developmental stages (Wiseman, Carpenter & Wheeler 1996; Van Dam, Hermenau & Baldwin 2001). This may not be the case for silica defences, because the proposed mechanisms of mouthpart wear and damage to gut physiology may increase with exposure time. Finally, it is important, especially when considering generalist herbivores, to assess whether these impacts can be alleviated by dietary switching, or whether there are irreversible impacts of silica on the herbivore (Hochuli 1996).

Here we assess the impacts of silica on herbivore growth performance and food utilization efficiency in several ways: first, as immediate effects to naïve herbivores reared on low-silica diets; second, as long-term effects on herbivores reared on experimentally manipulated silica diets; finally, by switching herbivores from diets of manipulated silica levels of both high and low silica levels to a single, standard diet with intermediate silica levels, we assess irreversible or residual impacts of silica defence. In addition, we assess a proposed mechanism for these effects by quantifying wear to mandibles caused by silica and relating it to food utilization efficiency in terms of both the levels of nutrients they extract from ingested food and how efficiently they convert digested food to body mass. We hypothesized that:

  • 1Without prior exposure to high-silica diets, immediate effects of silica on herbivore growth rate would be slight and due to a reduction in pre-digestion food utilization.
  • 2Increased exposure to silica-rich diets would increase the detrimental effects of silica by affecting insect mandible wear as well as reduced pre-digestion food utilization efficiencies.
  • 3The impacts of silica on herbivores will be greater on older instars than younger instars because of the increase in exposure period.
  • 4The effects of silica on insect food utilization efficiency and mandible wear would be irreversible and continue to impact on insect performance after switching to a different diet.

Methods

study species and plant growth conditions

Spodoptera exempta Walker (African armyworm) (Lepidoptera, Noctuidae) is considered to be a grass-feeding generalist herbivore (Lee et al. 2003) and is found extensively throughout Africa as a pest species on cereal crops and grasslands (Parker & Gatehouse 1985). S. exempta has previously been shown to have a limited ability to increase rate of consumption to compensate for a poor-quality diet (Lee et al. 2003; Massey et al. 2006). The S. exempta larvae used in this study came from a culture at the University of Lancaster, originally collected from Tanzania. We selected three grass species (Deschampsia caespitosa L., Festuca ovina L. and Lolium perenne L.) found abundantly in European grasslands, which vary in their natural levels of foliar silica, nutrients and palatability to herbivores (Massey et al. 2007).

All grasses were grown under glasshouse conditions (15–25 °C, 16:8 h L:D) and received water ad libitum. Silica treatment plants (D. caespitosa, F. ovina, L. perenne) were grown in seed trays of inert growth media (perlite) for 12–15 weeks. Half the plants were watered every 3 days with 100 mL Hoagland's solution; the other half received Hoagland's solution containing 150 mg L−1 of soluble silica in the form of NaSiO3·9H2O (Cid et al. 1990). In addition to these silica treatment plants, Poa annua L. (low-silica plants) used for experiment 1 (see below) were grown in seed trays of perlite for 8–10 weeks, watered every 3 days with 100 mL Hoagland's solution. Wheat seedlings (Triticum avenae L.) used in experiment 3 (see below) were grown in seed trays of compost (John Innes No. 2) for 4 weeks and received tap water ad libitum.

experiment 1: short-term impacts of silica

To assess the effects of short-term exposure to high-silica diets on the growth and food utilization efficiency of S. exempta larvae, one batch of larvae (> 72 individuals from a single egg clutch) were reared to second instar and another to fifth instar on leaves of P. annua which is known to have low foliar silica levels (SiO2 = 0·81% dry mass, N = 3·5% dry mass) and on which S. exempta display fast growth rates and high food utilization efficiencies (Massey et al. 2006). Growth performance and food utilization tests were conducted over 24 h using high- and low-silica diets of D. caespitosa, F. ovina and L. perenne leaves (n = 12 larvae and plants) on both the second and fifth instar larvae. Individual larvae were starved for 4 h and weighed, before being placed in a container with a known mass of fresh grass leaf material. Insects were kept at 23–25 °C and allowed to feed for 24 h, after which time they were starved for a further 4 h, to allow all the frass to be passed, before being re-weighed. Faecal samples were collected for nitrogen analysis. Growth and food utilization efficiency measures were calculated according to Slansky (1985), Massey et al. (2006) and Massey & Hartley (2006). Sample sizes were 12 replicates per treatment for all indices except nitrogen absorption where n = 5 for second instar and n = 10 for fifth instar larvae due to the limitation of sufficient faecal mass for nitrogen analysis.

  • • Relative growth rate (RGR), which calculates body mass growth relative to initial body mass, was calculated as: mass gained (g)/ initial mass (g)/time (days).
  • • Efficiency of conversion of ingested food (ECI), which estimates the percentage of food ingested converted to body mass, was calculated as: mass gained (mg change in fresh body mass)/food ingested (mg change in dry mass) × 100
  • • Efficiency of conversion of digested food (ECD), which estimates the percentage of assimilated food that is converted to body mass, was calculated as mass gained (mg change in fresh body mass)/food ingested (mg change in dry mass) – frass mass (mg dry mass) × 100
  • • Approximate digestibility (AD), which estimates the percentage of ingested food that is digested and assimilated, was calculated as food ingested (mg dry mass) – frass (mg dry mass)/food ingested (mg dry mass) × 100
  • • Relative consumption (RC), which estimates the mass of food ingested over 24 h relative to initial body mass, was calculated as: food ingested (mg change in dry mass)/mean body mass over experimental period (mg fresh mass).
  • • Proportion of nitrogen absorbed was calculated as nitrogen in food plant (percentage of dry mass) – nitrogen in faeces (percentage of dry mass)/nitrogen in food plant (percentage of dry mass). Lepidopteran faecal samples contain nitrogen from both undigested plant material as well as the waste nitrogen from metabolic processes. Therefore, the nitrogen absorption values should be regarded as an underestimate of the actual proportion of nitrogen absorbed and any treatment differences an underestimate of the potential differences.

experiment 2: long-term impacts of silica

To assess whether there were progressive effects of silica diets over the duration of larval exposure, S. exempta larvae were maintained throughout development on either a high- or low-silica diet of D. caespitosa, F. ovina or L. perenne. For one batch after 3 days (at second instar) and another batch after 12 days (at fifth instar), the growth performance and food utilization metrics (as in experiment 1) of larvae were measured for each diet treatment (n = 12, except nitrogen absorption where n = 5 for second instar and n = 10 for fifth instar larvae).

Following growth performance tests, fifth instar larvae were weighed individually and killed by immersion in ethanol. The mandibles were dissected out to estimate the degree of wear measured using a method adapted from Raupp (1985). The overall length of the right-hand mandible and the length of the incisor cusps were measured in ventral view at ×100 magnification under a microscope (n = 10). The incisor:mandible length ratio expresses the length of the incisor cusp (i.e. the effective cutting edge) as the proportion of the total mandible length. Therefore, mandible wear can be expressed as the inverse of the incisor:mandible length ratio. The identities of individual larvae were maintained throughout food-utilization tests and mandible wear assessment to compare the degree of mandible wear with levels of nitrogen absorbed from their diet.

experiment 3: irreversible impacts of silica

To assess whether exposure to high-silica diets had any residual effects on S. exempta larval performance, they were reared on silica treatment diets and then switched onto a diet of wheat leaves. Larvae were reared from hatching to fifth instar on either high- or low-silica diets of D. caespitosa, F. ovina, or L. perenne (n = 12). All larvae were reared on source diets for 12 days, in which time all had been in their fifth instar for at least 36 h, and then placed onto a diet of wheat leaves for 24 h to conduct food-utilization tests. Growth performance and food-utilization tests were then conducted on all larvae (n = 12, except nitrogen absorption where n = 10 per source diet treatment), using wheat leaves as a standard plant material of intermediate silica concentration (N = 6·74% dry mass, SiO2 = 2·67% dry mass). After performance tests, larvae were weighed and assessed for mandible wear as above, with individual larval identity maintained to allow comparison with nitrogen absorption.

chemical analyses

Dried faecal samples and food-plant leaf samples (n = 10) from performance tests were ground using a ball mill (‘Pulverisette 23’, Fritsch, Industriestrasse 8, Germany) before analysing nitrogen concentration using flash combustion of dried leaf samples (approximately 2·5 mg) followed by gas chromatographic separation (Elemental Combustion System; Costech Instruments, Milan Italy) calibrated against a standard of composition C26H26N2O2S. Foliar silica concentration was determined by fusing dried leaf samples (approximately 0·2 g) in NaOH followed by analysis using the colorimetric silicomolybdate technique (Allen 1989; Massey et al. 2006). To confirm that our manipulated silica levels were within the range occurring in natural communities, leaf samples of each of the three grass species were collected from 10 replicate plants in 10 sites around East Sussex in August 2005, selected to represent a range of soil types and habitats (Table 1).

Table 1.  Foliar silica and nitrogen content (percentage of dry mass) of grass study species. Values are means ± SE
Silica treatmentSilica content (percentage of dry mass)Nitrogen content (percentage of dry mass)Field silica concentrations (percentage of dry mass)
LowHighLowHighMinimumMeanMaximum
Deschampsia caespitosa1·79 ± 0·236·62 ± 0·443·03 ± 0·202·89 ± 0·161·513·755·62
Festuca ovina0·52 ± 0·042·44 ± 0·163·80 ± 0·053·94 ± 0·050·652·323·54
Lolium perenne0·54 ± 0·104·68 ± 0·345·82 ± 0·135·74 ± 0·140·582·744·52

statistical analyses

For experiments 1 and 2 the RGR, ECI, ECD, AD and RC of S. exempta larvae were assessed using three-way anovas comparing diet species, herbivore age and silica treatments, and using Tukey's post-hoc analysis of significant results. As nitrogen levels differed significantly between species, but not between silica treatments within species (Table 1), the proportion of nitrogen absorbed by S. exempta was compared between herbivore age classes and silica treatments using two-way anovas for each diet species. The degree of mandible wear, represented by the inverse of the incisor:mandible length ratio, was compared between species and diets treatments using a two-way anova. Due to the lack of independence among individuals within treatment groups compared with between treatment groups, the relationship between mandible wear and foliar silica concentration was assessed using a Pearson's correlation on the means of each diet (n = 6). For experiment 3, the RGR, ECI, ECD, AD, RC and proportion of nitrogen absorbed from wheat leaves was assessed using two-way anovas comparing source diets species and silica treatment of source diet. The impact that previous mandible wear had on the proportion of nitrogen absorbed from wheat leaves was assessed using regression analysis after controlling for the influence of previous diet species as a categorical factor in a general linear model (GLM).

Results

foliar chemical analysis

Foliar nitrogen concentrations were unaffected by silica treatments (Table 1, anova: species F2,53 = 28·52, P < 0·001; silica F1,53 = 1·42, P = 0·238; species × silica F2,53 = 0·25, P = 0·777). For all three treatment species, foliar silica concentrations were significantly increased by the silica treatment (Table 1, anova: species F2,53 = 122·05, P < 0·001; silica F1,53 = 652·92, P < 0·001; species × silica F2,53 = 37·20, P < 0·001).

experiment 1: short-term impacts of silica

S. exempta larvae that were reared on low-silica diets and then had exposure to silica treatment diets at second instar displayed significant reductions in RGR on all grass species, and slightly reduced nitrogen absorbed on high-silica L. perenne compared with low silica (Fig. 1, Table 2). There were no significant differences due to silica for other food-utilization measurements for second instars (Fig. 1, Table 2). When S. exempta were reared on a low-silica diet until fifth instar and then switched onto silica treatment diets, foliar silica had significant effects on their growth rates on both F. ovina and L. perenne, with reductions of 36% and 34%, respectively (Fig. 1a, Table 2). Fifth instar larvae absorbed a lower proportion of nitrogen from high-silica diets than low-silica ones from both D. caespitosa and L. perenne (Fig. 1f, Table 3). This was most pronounced for D. caespitosa, with insects suffering a 34% reduction in the proportion of nitrogen they absorbed due to silica. Silica-rich diets of L. perenne significantly reduced the ECI of fifth instar larvae (Fig. 1b, Table 2).

Figure 1.

(a) Relative growth rate (RGR), (b) efficiency of conversion of ingested material (ECI), (c) efficiency of conversion of digested material (ECD), (d) approximate digestibility (AD), (e) relative consumption (RC) and (f) the proportion of nitrogen absorbed from three grass species for second and fifth instar larvae after short-term exposure to silica treatment diets. Values are means ± SE. Where the main factor effects are significant (see Table 2), bars not sharing a common letter differ significantly (Tukey; P < 0·05). Due to interspecific differences in foliar nitrogen content, the proportion of nitrogen absorbed by herbivores was assessed within species (f, see Table 3).

Table 2. anova models assessing the effects of grass species (Deschampsia caespitosa, Festuca ovina, and Lolium perenne), herbivore age state (second and fifth instar) and silica treatment (high and low) on herbivore performance and food utilization over both short- and long-term exposure periods
ResponseFactord.f.Duration of exposure to silica diets
Short term (expt. 1)Long term (expt. 2)
FPFP
  1. Herbivore performance and food utilization indices were as follows: RGR, relative growth rate; ECI, efficiency of conversion of ingested material; ECD, efficiency of conversion of digested material; AD, approximate digestibility; RC, relative consumption.

RGRSpecies (Sp)2 2·120·124 5·800·004
Age (A)110·050·002 0·270·603
Silica (Si)173·76< 0·001105·74< 0·001
Sp × A21·670·193 2·950·056
Sp × Si20·780·460 0·150·863
A × Si13·510·063 0·370·545
Sp × A × Si21·550·216 0·100·908
Error132    
ECISpecies (Sp)28·31< 0·0016·620·002
Age (A)11·28 0·2619·530·002
Silica (Si)115·76< 0·001 91·75< 0·001
Sp × A23·640·0290·690·502
Sp × Si22·330·1011·910·152
A × Si17·610·0076·010·016
Sp × A × Si22·350·0990·590·554
Error132    
ECDSpecies (Sp)20·230·7980·200·816
Age (A)11·010·31713·08< 0·001
Silica (Si)10·310·5764·680·032
Sp × A20·830·4381·160·032
Sp × Si20·110·8943·500·033
A × Si10·600·4410·460·500
Sp × A × Si20·190·8242·040·135
Error132    
ADSpecies (Sp)22·400·0950·040·962
Age (A)120·31< 0·001 92·8< 0·001
Silica (Si)10·050·8319·410·003
Sp × A21·840·1643·260·042
Sp × Si21·000·371 11·12< 0·001
A × Si10·090·7714·310·040
Sp × A × Si21·760·1763·980·021
Error132    
RCSpecies (Sp)212·11< 0·001 14·48< 0·001
Age (A)1122·65< 0·0010·030·865
Silica (Si)10·04 0·8450·680·409
Sp × A212·35< 0·0014·900·009
Sp × Si28·14< 0·0010·440·645
A × Si10·79 0·3742·130·147
Sp × A × Si23·54 0·0320·800·453
Error132    
Table 3. anova models assessing the effects of herbivore age state (second and fifth instar) and silica treatment (high and low) on proportion of nitrogen absorbed by Spodoptera exempta larvae from food plants over both short- and long-term exposure periods
Diet speciesFactord.f.Duration of exposure to silica diets
Short term (expt. 1)Long term (expt. 2)
FPFP
DeschampsiacaespitosaAge (A)151·36< 0·001 4·00 0·056
Silica (Si)117·78< 0·00146·75< 0·001
A × Si1 8·96 0·00712·20 0·002
Error26    
FestucaovinaAge (A)185·57< 0·00122·42< 0·001
Silica (Si)1 2·00 0·169 8·64 0·007
A × Si1 0·81 0·376 1·79 0·192
Error26    
LoliumperenneAge (A)1171·93< 0·001123·00< 0·001
Silica (Si)1 16·30< 0·001 71·00< 0·001
A × Si1 0·27 0·609 14·44 0·001
Error26    

experiment 2: long-term impacts of silica

When reared from hatching and throughout development on silica-rich diets, silica significantly reduced the RGR of S. exempta larvae across all species and at both second and fifth instars, except L. perenne second instar (Fig. 2a, Table 2). Silica also reduced the efficiency with which fifth instar larvae on all three diet species and second instars on L. perenne could convert mass of ingested grass leaves to body mass (Fig. 2b, Table 2). In addition, silica reduced the approximate digestibility of grass leaves to fifth instar larvae on D. caespitosa leaves and both second and fifth instar larvae on F. ovina leaves (Fig. 2d, Table 2). Silica has no consistent impacts on the conversion of absorbed food to body mass (Fig. 2c, Table 2), nor did silica alter the consumption rates of herbivores (Fig. 2e, Table 2). Similar to the results for RGR, long-term exposure to silica-rich diets had large impacts on the proportion of nitrogen larvae could absorb from the leaves. Silica reduced nitrogen absorption during digestion on all diets species at fifth instar and on D. caespitosa and L. perenne at second instar (Fig. 2f, Table 3). Overall, the impacts of silica on RGR, ECI and the proportion of nitrogen absorbed from leaves was greater with continual exposure (experiment 2) compared with short-term exposure (experiment 1), especially at fifth instar. For example, overall in fifth instar larvae, silica caused 60% reductions in RGR with continuous exposure, compared with 27% reductions on single exposure. While there were also reductions in ECI of 55% with continuous exposure versus 48% with single exposure, and of 41% versus 20% in the proportion of nitrogen absorbed.

Figure 2.

(a) Relative growth rate (RGR), (b) efficiency of conversion of ingested material (ECI), (c) efficiency of conversion of digested material (ECD), (d) approximate digestibility (AD), (e) relative consumption (RC) and (f) the proportion of nitrogen absorbed from three grass species for second and fifth instar larvae after long-term exposure to silica treatment diets. Values are means ± SE. Where the main factor effects are significant (see Table 2), bars not sharing a common letter differ significantly (Tukey; P < 0·05). Due to interspecific differences in foliar nitrogen content, the proportion of nitrogen absorbed by herbivores was assessed within species (f, see Table 3).

In addition to an increase in the magnitude of silica effects on larval growth and food utilization efficiency with duration of exposure, we found evidence that S. exempta could adapt to low-silica diets of different species over time, but not to high-silica diets (Figs 1 & 2). Larvae achieved faster RGR on low-silica plants when they had previously been raised on that diet (Fig. 2a), compared with those larvae that had no prior exposure to treatments diets (Fig. 1a). The opposite effect was found for larvae reared on high-silica diets. Larvae raised for 12 days on high-silica diets displayed lower RGR than larvae previously reared on a low abrasive diet (P. annua). The latter result also suggests that there are progressive negative impacts of silica in grasses on herbivores.

Comparisons of insect mandible wear across larvae reared for 12 days on different silica diet treatments revealed significant increases in the degree of wear (1/incisor:mandible length ratio) on high-silica diets of all species (Fig. 3a anova: species F2,54 = 12·02 and P < 0·001, silica F1,54 = 18·40 and P < 0·001, species × silica F2,54 = 0·55 and P = 0·582). In addition, there was a high degree of correlation between the degree of mandible wear and the foliar silica concentrations of each diet treatment (Fig. 3b Pearson's correlation coefficient = 0·810 and P = 0·050).

Figure 3.

(a) Mandible wear (mean ± SE) (1/incisor:mandible length ratios) for fifth instar Spodoptera exempta larvae after 12 days on high- and low-silica diets of Deschampsia caespitosa Festuca ovina and Lolium perenne. Letters above bars denote significant differences between treatments (Tukey; P < 0·05). (b) Relationship between mean foliar silica concentrations (± SE) of treatment plants and mandible wear (mean ± SE) of fifth instar S. exempta.

experiment 3: irreversible impacts of silica

Larvae previously reared on silica diet treatments until fifth instar and then placed on a diet of wheat leaves, displayed residual effects of past diet both on their RGR and food utilization (Fig. 4). Larvae from high-silica source diets had, on average, 45% lower RGR on wheat than did those larvae from low-silica source diets (Fig. 4a, Table 4). The greatest effects were found on larvae previously reared on high-silica D. caespitosa (54% reduction in RGR). There were also consistent reductions, due to previous silica treatment diets, in the efficiency with which larvae could acquire body mass from both the amount of food ingested (Fig. 4b) and the amount of food digested (Fig. 4c, Table 4). Comparisons of foliar and faecal nitrogen levels revealed that larvae previously reared on high-silica diets of D. caespitosa and L. perenne absorbed 42% and 25% less nitrogen from wheat leaves than those reared on low-silica leaves (Fig. 4f, Table 4). These findings may be partly explained by the strong negative relationship found between the degree of mandible wear and the proportion of nitrogen absorbed from wheat leaves, even when controlling for source diet species (Fig. 5).

Figure 4.

(a) Relative growth rate (RGR), (b) efficiency of conversion of ingested material (ECI), (c) efficiency of conversion of digested material (ECD), (d) approximate digestibility (AD), (e) relative consumption (RC) and (f) the proportion of nitrogen absorbed from grass for fifth instar larvae feeding on wheat leaves after prior exposure to high- and low-silica treatment diets of Deschampsia caespitosa Festuca ovina and Lolium perenne. Values are means ± SE. Where main factor effects are significant (see Table 4), bars not sharing a common letter differ significantly (Tukey; P < 0·05).

Table 4. anova models assessing the effects of prior exposure to diets of different grass species (Deschampsia caespitosa, Festuca ovina, and Lolium perenne), and silica treatment (high and low) on herbivore performance and food utilization of fifth instar larvae feeding on wheat leaves (experiment 3)
ResponseFactord.f.FP
  1. Herbivore performance and food utilization indices were as follows: RGR, relative growth rate; ECI, efficiency of conversion of ingested material; ECD, efficiency of conversion of digested material; AD, approximate digestibility; RC, relative consumption; nitrogen absorbed, proportion of nitrogen absorbed from diet.

RGRSpecies (Sp)26·940·002
Silica (Si)131·98< 0·001
Sp × Si22·590·082
Error66  
ECISpecies (Sp)215·93< 0·001
Silica (Si)160·21< 0·001
Sp × Si20·600·550
Error66  
ECDSpecies (Sp)21·560·217
Silica (Si)126·59< 0·001
Sp × Si20·080·920
Error66  
ADSpecies (Sp)210·72< 0·001
Silica (Si)13·020·087
Sp × Si20·680·510
Error66  
RCSpecies (Sp)21·670·195
Silica (Si)10·010·927
Sp × Si24·830·011
Error66  
Nitrogen absorbedSpecies (Sp)2175·41< 0·001
Silica (Si)147·83< 0·001
Sp × Si21·410·252
Error66  
Figure 5.

Relationship between mandible wear (mean ± SE) (1/ incisor:mandible length ratios) and resultant proportion of nitrogen absorbed from wheat leaves (mean ± SE) by fifth instar Spodoptera exempta larvae previously reared on high- and low-silica diets of Deschampsia caespitosa Festuca ovina and Lolium perenne. anova: species F2,56 = 113·58 and P < 0·001, mandible wear F1,56 = 36·43 and P < 0·001, overall r2 = 86·3%, while mandible wear r2 = 28·4%.

Discussion

We have experimentally demonstrated, for the first time, two contrasting mechanisms by which silica in grasses can impact upon insect herbivore performance. First, silica reduced the efficiency of food utilization by S. exempta. Specifically, silica reduced the efficiency with which S. exempta converted ingested food to body mass and the amount of nitrogen absorbed from their food, leading to reduced insect growth rates. This was found to be an immediate response of herbivores to silica-rich diets. The magnitude of this effect increased with duration of exposure to a high-silica diet: the approximate digestibility of food reduced with increased exposure time to silica. The food utilization metrics that were affected by silica content of the food (ECI, AD and nitrogen absorption) are all measures of how much of the nutrients herbivores extract from the food ingested. The measure of how efficiently herbivores utilize digested food (ECD) was unaffected by silica in this study. This demonstrates that silica impacts upon herbivores by their reducing nutrient acquisition from food.

Second, we have demonstrated that silica increases the wear on insect mandibles and that this effect is extremely rapid, occurring within a single instar. We have also shown that mandible wear was correlated with reduced growth rates and reduced absorption of nitrogen. Although it has long been believed to be a key result of a high-silica diet, this is the first study to experimentally quantify silica-mediated mandible wear and assess its impacts on herbivore performance. In addition to these mechanisms, we found that the effects silica has on insect growth and food utilization efficiency are nonreversible, persisting after the herbivore has switched diets. Hence, silica has major long-term and irreversible impacts on insect herbivore performance and potentially their fitness. Although the outcomes of this reduced performance were not measured directly in this study, it is well established that there are strong positive associations between adult mass and lepidopteran fecundity and fitness (Woodrow, Gatehouse & Davies 1987; Lill & Marquis 2001; Iyengar & Eisner 2002).

short-term vs. long-term impacts of silica

Previous studies have shown that silica-rich plants are more abrasive and this may lead to reduced preference for herbivores to feed on them (Gali-Muhtasib et al. 1992; Massey & Hartley 2006; Massey et al. 2006). Our findings suggest that in addition to the effects of silica on preference, there are also very rapid effects on the growth performance of herbivores. The reduction in nitrogen absorbed from the high-silica plants may be particularly important for herbivores such as Spodoptera, with a limited capacity to increase consumption rates in response to diet quality (Lee et al. 2003; Massey et al. 2006). We found no evidence that S. exempta increased rates of consumption to compensate for poor diet quality. For herbivores that are able to increase consumption rates, such as locusts (Raubenheimer & Simpson 2003), the strong effects of silica on mouthpart wear could increase with feeding rate.

In addition to the immediate impacts of silica-based defences, the magnitude of their negative effects on herbivores increased with the duration of exposure to silica-rich diet. This appeared to be due, at least in part, to abrasion of mandibles over time. The degree of mandible wear displayed a strong positive relationship with the silica concentration of herbivore diet, and 28% of the variation in nitrogen absorbed from a standard diet was explained by the wear on herbivore mandibles alone. However, as caterpillars replace their mandibles with each moult, mouthpart wear can only have an impact within each instar. Therefore, mandible wear does not explain fully the increased negative impact of silica with exposure time. We suggest that there may be parallel destructive effects on the mid-gut of caterpillars which may reduce food utilization efficiency. These effects are able to persist between instars because this part of the digestive tract is not shed during a moult (Chapman 1983). This suggestion is supported by our findings that the negative impacts of silica persist after the herbivore has switched onto a different diet, a result which also could not be explained fully by mouthpart wear (Fig. 5).

progressive and irreversible effects of silica

Several previous studies have demonstrated that the impacts of chemical defences can be greater on early instars compared to late instars (Wiseman et al. 1996; Van Dam et al. 2001; Adler 2004). For silica, this does not appear to be the case. For example, long-term exposure to silica-rich diets resulted in an overall 48% reduction in RGR for second instars and a 60% reduction in RGR for fifth instars. Similar results were found for ECI and nitrogen absorption (Fig. 2). For short-term exposure, the impacts on RGR were similar for both second and fifth instars. This appears to be due to the progressive impacts of silica with longer exposure times and suggests that herbivores do not develop a tolerance for high-silica diet with age. Silica is a defence to which herbivores seem unable to adapt, in contrast to many chemical defences (Green, Zangerl & Berenbaum 2001; Ratzka et al. 2002).

For the first time, we have demonstrated that the impacts of silica on herbivore performance and food utilization efficiency persist after the herbivore has switched to a lower-silica diet. Those S. exempta previously reared on high-silica diets displayed lower food utilization on a standard diet in terms of efficiency to convert both ingested and digested food to body mass, and in the levels of nitrogen absorption than S. exempta reared on low-silica diets. These reductions resulted in continued lower growth rates on the new diet. Many herbivores have been shown to perform better when ingesting a mixed diet than a single plant species due to the dilution of particular chemical defences and/or increased nutrient balance (Bernays et al. 1994; Hagele & Rowell-Rahier 1999; Adler 2004; Wiggins et al. 2006). The results of this study could have major implications for both specialist and generalist herbivores that use dietary switching. Herbivores that had experienced diets containing high foliar silica content would not recover full functionality by switching diets, thus demonstrating the strong deterrent effect of silica-based defences. Again, in contrast to many chemical defences, these defences cannot be overcome or compensated for with dietary mixing: the adverse effects remain when the insects are moved to plants with low silica.

implications for herbivores

As foliar nitrogen levels are often the principle limiting nutrient influencing the behaviour and performance of insect herbivores (Slansky & Scriber 1985), the reduction in nitrogen absorbed from grass due to silica-based defences could have large impacts on both herbivore fitness and feeding preferences under field conditions, where inter- and intraspecific variation in grass foliar silica levels is high (O’Reagain & Mentis 1989; Massey et al. 2007). Silica defences result in increased development times which can increase exposure to natural enemies (Haggstrom & Larsson 1995; Benrey & Denno 1997; Kondoh & Williams 2001; Kaplan et al. 2007), as well as reduced adult body mass which can lead to decreased fecundity (Woodrow et al. 1987; Lill & Marquis 2001; Iyengar & Eisner 2002). We have demonstrated that the impacts of silica on S. exempta larvae were progressive with exposure time and could not be compensated for, even by switching to a different diet. Thus, silica defences potentially have major implications for herbivore fitness.

Our study also demonstrates generality in the nature of physical defences across different herbivores and plants. First, the mechanism of a reduction in the nitrogen absorbed from the leaves is consistent with our previous findings from field voles (Massey & Hartley 2006). This consistency of results across insect and mammalian herbivores suggests a universal mechanism for silica defence. Second, similar defence mechanisms to silica have recently been reported for another physical plant defence, calcium oxalate crystals in Medicago truncatula (Korth et al. 2006). The calcium oxalate crystals reduced growth rates and food utilization efficiency of Spodoptera exigua, as well as possible impacts to mouthpart wear. Thus, suggests that different physical defences in a broad range of plants may impact on folivorous insects via similar mechanisms.

Acknowledgements

Jeremy Field, Hefin Jones, Alan Stewart, Adam Vanbergen and three anonymous reviewers made helpful comments on earlier drafts of the manuscript. The research was funded jointly by a Small Ecological Project Grant from the British Ecological Society (awarded to F.M.) and by the School of Life Sciences, University of Sussex.

Ancillary