Herbivore-induced plant volatiles provide associational resistance against an ovipositing herbivore


  • Ali Zakir,

    Corresponding author
    • Swedish University of Agricultural Sciences, Division of Chemical Ecology, Department of Plant Protection Biology, Alnarp, Sweden
    Search for more papers by this author
  • Medhat M. Sadek,

    1. Swedish University of Agricultural Sciences, Division of Chemical Ecology, Department of Plant Protection Biology, Alnarp, Sweden
    2. Department of Zoology, Faculty of Science, Assiut University, Assiut, Egypt
    Search for more papers by this author
  • Marie Bengtsson,

    1. Swedish University of Agricultural Sciences, Division of Chemical Ecology, Department of Plant Protection Biology, Alnarp, Sweden
    Search for more papers by this author
  • Bill S. Hansson,

    1. Swedish University of Agricultural Sciences, Division of Chemical Ecology, Department of Plant Protection Biology, Alnarp, Sweden
    2. Max Planck Institute for Chemical Ecology, Jena, Germany
    Search for more papers by this author
  • Peter Witzgall,

    1. Swedish University of Agricultural Sciences, Division of Chemical Ecology, Department of Plant Protection Biology, Alnarp, Sweden
    Search for more papers by this author
  • Peter Anderson

    1. Swedish University of Agricultural Sciences, Division of Chemical Ecology, Department of Plant Protection Biology, Alnarp, Sweden
    Search for more papers by this author

Correspondence author. E-mail: ali.zakir@slu.se


  1. Plants produce herbivore-induced plant volatiles (HIPVs) in response to damage by herbivores. Although HIPVs are known to enhance plant resistance by affecting herbivore host plant preferences and by attracting natural enemies, little is known about the role of HIPVs on the resistance of neighbouring plants and the mechanism behind this associational resistance.
  2. This study examined the effect of HIPVs from herbivore-damaged host plants (alfalfa (Medicago sativa), clover (Trifolium alexandrinum) and cotton (Gossypium hirsutum)) on oviposition by Egyptian cotton leafworm (Spodoptera littoralis) (Lepidoptera: Noctuidae) on neighbouring, undamaged host plants.
  3. There was a significant reduction in oviposition by S. littoralis on undamaged plants adjacent to herbivore-damaged cotton plants under both field and laboratory conditions. The results showed that the associational resistance by HIPVs depends on direct effects on oviposition behaviour in S. littoralis. There were also indications that other mechanisms may be involved.
  4. Associational resistance via HIPVs was not observed for all plant species tested. Emission of HIPVs from damaged cotton increased the resistance of undamaged cotton and alfalfa plants to oviposition by S. littoralis, but HIPVs from damaged alfalfa and clover neighbours did not provide resistance to undamaged cotton plants.
  5. Synthesis. Our results suggest that the presence of HIPV-emitting plant neighbours can reduce herbivory on undamaged plants and enhance plant resistance by affecting oviposition behaviour in insect herbivores.


The distribution of plants within a habitat can show large spatial heterogeneity. This may affect the fitness of individual plants and influence interactions among plants as well as interactions between plants and herbivores (Agrawal, Lau & Hambäck 2006; Karban 2008). The attractiveness of an individual plant to vertebrate and invertebrate herbivores shows large variations depending on its neighbouring plants. This associational effect has been found both on land and in marine environments. In diverse vegetation, neighbouring plants may influence the resistance of the host plants in several ways: by attracting natural enemies of the herbivores, by attracting and sustaining herbivores or by repelling herbivores (Atsatt & O′Dowd 1976; Hambäck, Ågren & Ericson 2000; Barbosa et al. 2009).

Emission of volatiles by non-host and damaged neighbours can reduce herbivore attack on host plants (Arimura, Shiojiri & Karban 2010; Dicke & Baldwin 2010; Heil & Karban 2010; Hare 2011; Jactel et al. 2011). Several mechanisms have been identified for such associational resistance, based on studies under laboratory and field conditions. Plant neighbours can provide resistance to host plants by producing repellent volatile compounds that make herbivores either avoid a host plant patch or fail to detect a host plant within the patch (Atsatt & O′Dowd 1976; Agrawal, Lau & Hambäck 2006). Another mechanism is that volatiles from neighbouring plants induce resistance in the intact plants and reduce their attractiveness and suitability against herbivore attack (Karban et al. 2006; Heil & Silva Bueno 2007; Frost et al. 2008; Arimura, Matsui & Takabayashi 2009). Lastly, repellent volatiles from a neighbouring plant can be adsorbed on the host plant surface and increase its resistance against herbivore attack upon re-emission (Himanen et al. 2010; Karban 2010).

Host plant choice is fundamental for the fitness of insect herbivores, and female choice of a suitable oviposition site is in many cases vital for her progeny (Schoonhoven, vanLoon & Dicke 2005). Acceptance of a host plant for mating and oviposition is based on the balance between positive stimuli (attractants and stimulants) and negative stimuli (repellents and deterrents) (Renwick & Chew 1994). Volatile cues from plants are important for many herbivores during orientation towards a plant and final assessment of that plant (Bruce & Pickett 2011). The suitability of a plant for growth and offspring development depends on plant nutritional quality, but also on occupation by other insects or attractiveness to natural enemies (Bernays 2001). Many plants produce herbivore-induced plant volatiles (HIPVs) in response to herbivore feeding that affects the defence of the producing plant and also the resistance and fitness of the intact plants in the neighbourhood (Barbosa et al. 2009; Hare 2011). These HIPVs have been shown to reduce herbivore attack by affecting host selection behaviour in herbivores or by attracting the natural enemies of the herbivores (Karban 2010).

The Egyptian cotton leafworm, Spodoptera littoralis (Lepidoptera: Noctuidae), is a generalist herbivore that feeds on a wide range of wild and cultivated plants including cotton (Gossypium hirsutum), alfalfa (Medicago sativa) and clover (Trifolium alexandrinum) in the agro-ecosystems of North Africa and the Middle East (Sadek, Hansson & Anderson 2010). Cotton plants respond to herbivory by producing HIPVs both locally, at the site of damage, and systemically, distal to the damaged parts (McCall et al. 1994; Röse & Tumlinson 2005). These HIPVs have been shown to attract predators and parasitoids (Paré & Tumlinson 1997). Furthermore, female S. littoralis moths avoid oviposition on cotton plants damaged by conspecific larvae (Anderson & Alborn 1999) and by a root herbivore (Anderson, Sadek & Wäckers 2011). Electrophysiological studies on female S. littoralis have shown that cotton odours, including many of the HIPVs, are detected by olfactory receptor neurons housed in sensilla on the antennae (Jönsson & Anderson 1999) and processed in the central nervous system (Sadek et al. 2002; Saveer et al. 2012). The starting hypothesis in this study was that HIPV emissions from herbivore-damaged host plant neighbours provide resistance to undamaged host plants by reducing herbivore attack. Bioassays were performed in the laboratory and under field conditions to test the function and possible mechanism of HIPV interactions with ovipositing S. littoralis moths. The objectives were to determine: (1) whether emission of HIPVs from damaged cotton plants provides associational resistance against oviposition behaviour in S. littoralis on undamaged host plants; (2) whether HIPVs protect both conspecific and heterospecific host plants and whether the effect is reciprocal; (3) at what distance undamaged plants are protected by emission of HIPVs and (4) what underlying mechanism confers such resistance, that is, whether HIPVs affect oviposition behaviour directly or indirectly through plant–plant communication.

Materials and methods

Laboratory and Field Insects and Plants

For the laboratory experiments, S. littoralis were obtained from a culture established in 2007 from insects collected in the Alexandria region of Egypt. Wild insect material was added to the laboratory culture once or twice annually. The insects were reared on a semi-synthetic diet (Hinks & Byers 1976) and kept at 25 ± 2 °C, 65 ± 5% RH and L16/D8 photoperiod. Insects were sexed and kept in separate rearing chambers until used in the experiments. For the field experiments, the moths were taken from a culture maintained at the Department of Zoology, Assiut University, Egypt. The insects were collected from the field during spring 2011 to establish the laboratory culture. The insects were reared on artificial diet based on wheat germ and casein and kept at 25 ± 1 °C, ≥70% RH and L16/D8 photoperiod. For the laboratory experiments, individual cotton seeds (Gossypium hirsutum L., var. Delta pineland 90) or 15–20 seeds of alfalfa (Medicago sativa L.) or clover (Trifolium alexandrinum L., var. Winner) were planted in pots (D 14 cm) with soil and placed in an acclimatized glasshouse at 25 ± 2 °C, 65 ± 5% RH and L16/D8 photoperiod situated at Alnarp, Sweden (55°39′30″ N, 13°5′0″ E). Artificial light (SON-T, 400 W, Philips, The Netherlands) was provided in addition to natural light, 1–2 m above the plants. Plants of the same size (5–6 true leaves per plant for cotton, 12–14 lush green branches per pot for alfalfa and clover) were used in the experiments. The plants had not started to form flower buds at the time of the experiments.

In the field experiments, cotton plants were grown in an experimental field near the city of Assiut, Egypt (27°10′58″ N, 31°10′58″ E). Potted cotton plants were grown individually from seeds in a glasshouse at the Department of Botany, Assiut University. The pots initially contained about 2 L of soil, and the temperature in the glasshouse oscillated between about 20 °C during night and 35 °C during daytime. The plants had 8–10 true leaves when they were used in the experiments.

Oviposition Experiments

In the laboratory, two pairs of undamaged plants, serving as oviposition plants, were placed on opposite sides (80 cm apart) of a cage (120 × 80 × 60 cm). Another two pairs of plants were placed just outside the far ends of this cage, serving as neighbouring plants (Fig. 1). Two plant ‘patches’ were thus created, one where a pair of undamaged plants inside the cage was adjacent to undamaged plants outside the cage (undamaged emitters) and one where the other pair of undamaged plants inside the cage was adjacent to plants with ongoing herbivory by S. littoralis larvae (damaged emitters) outside the cage. The experiments were performed in conditioned glasshouse chambers with an ambient airflow and 65 ± 5% RH and 25 ± 2 °C. In addition to normal daylight, the chamber was illuminated from 6.00 till 18.00 by a mercury vapour lamp (SON-T, 400 W, Philips, The Netherlands). To produce damaged emitters, 10–12 second to third instar S. littoralis larvae were released on plants 48 h before the oviposition experiment started. The larvae were allowed to feed continuously during the oviposition experiment, and larvae that were lost or stopped feeding were replaced with new ones daily. Eight female and 10 male pupae of S. littoralis close to eclosion were placed in the centre of the oviposition cage and were allowed to emerge and mate inside the cage. Eggs deposited during the 4 days after the first egg batch appearance in an oviposition cage were collected and weighed.

Figure 1.

Laboratory set-up. Four undamaged plants were placed inside the oviposition cage with 2 plants on each side, 80 cm away from each other. A pair of undamaged plants was placed on one side outside the cage, and another pair of plants damaged by the larvae of Spodoptera littoralis was placed on the other side of the cage, adjacent to the oviposition plants.

In the field, oviposition experiments were performed using glasshouse-potted cotton plants as a source of HIPVs. Oviposition cages (140 × 110 × 90 cm) were placed in the field to surround groups of the field-cultivated cotton plants. To avoid directional effects, the cages were placed in pairs with the 110 × 90 cm sides facing each other and separated by a distance of about 50 cm. However, the minimum distance between different pairs was 12 m over the field area of about 1400 m2. Three pairs of cages were used in a first trial, which started on 26 June 2011, and another replicate of three pairs was set up on 12 July 2011, making a total sample size of 12 cages. Potted plants were taken to the field at the time expected for the beginning of oviposition. For oviposition, 8 female and 8 male pupae of S. littoralis close to eclosion were placed in the centre of the oviposition cage, buried under a few mm of moist soil and were allowed to emerge and mate inside the cage. After placing the cages in the field, four potted plants that had been damaged by S. littoralis larvae for 7 days were placed between the two cages of each pair in the first trial and four undamaged plants were placed outside the far side of each cage in the pair (Fig. 2). In the second trial, four damaged plants were placed outside the far sides of the cages while four undamaged plants were placed between the two cages of a pair. This was done to control for directional or wind-drift effects, although wind in this region of Egypt, if present at all, is very mild (Sadek, Hansson & Anderson 2010). The area inside each cage was divided into two roughly equal areas, one for the side of the damaged plants and the other for the side of the undamaged plants. All the plants in each area were checked for egg batches for at least 10 days after the appearance of the first egg batch. In both the laboratory and field experiments, the number of eggs was calculated using the equation [Number of eggs = weight of eggs (mg) × 20], and for each cage, the total numbers of eggs per treatment were pooled.

Figure 2.

Field set-up. Experiments were performed in cages with field-grown cotton plants for oviposition (1), while glasshouse-grown potted cotton plants with (2) or without (3) larval damage were placed outside the cage adjacent to the oviposition plants. In a second series of experiments, the plants at positions (2) and (3) were exchanged.

In the laboratory, three experiments were conducted to investigate the effects of associational resistance via HIPVs within heterospecific plant patches. In each experiment, plants from two species were tested at a time by placing them either at the oviposition (receiver) or at the neighbouring (emitter) plant positions. In the first and second oviposition experiment, undamaged cotton plants were used as receivers, and damaged and undamaged alfalfa (= 12) or clover (= 8) plants were used as emitters. In the third oviposition experiment, undamaged alfalfa plants were used as receivers, while damaged and undamaged cotton plants were used as emitters (= 6).

To investigate the active range of HIPVs on oviposition behaviour, the undamaged cotton plants were distributed inside the oviposition cage at 3 different distances, 30, 60 and 90 cm away from the damaged plants that were placed outside the cage. No undamaged plants were placed outside the opposite end of the cage. Two-day-old mated S. littoralis moths were released in the cages to select among the most suitable plants for oviposition. Eggs were collected and weighed as above.

Plant Exposure to HIPVs

To test whether HIPVs from damaged plants directly affected the resistance in neighbouring plants, a pair of undamaged cotton plants were placed inside a cage at a distance of 30 cm from damaged cotton plants outside the cage infested with S. littoralis larvae (Fig. 3a). The plants inside the cage were pre-exposed to the damaged odour in the same way as in the oviposition experiments above. The damaged plants were subjected to continuous feeding by the larvae over a period of 96 h. When older larvae stopped feeding, they were replaced with young ones. Undamaged plants pre-exposed to HIPVs were carefully observed to make sure that they remained free of direct larval feeding. Control plants were kept adjacent to undamaged plants.

Figure 3.

Set-up of experiment on pre-exposure to HIPVs. (a) A pair of undamaged cotton plants on one side of the cage was exposed to volatiles from damaged cotton and to volatiles from undamaged cotton on the other side of the cage for a period of 96 h. (b) Pre-exposed plants were exposed to oviposition by Spodoptera littoralis moths in the absence of damaged and undamaged emitters.

The effect of exposure to HIPVs on neighbouring undamaged plants was tested in both oviposition and larval feeding experiments. For the oviposition experiments, a pair of plants pre-exposed to HIPVs was placed in one end of the oviposition cage, while a pair of undamaged plants unexposed to HIPVs was placed in the opposite end of the cage. No plants were placed outside the cage (Fig. 3b). The experiments then followed the same protocol as the first set of oviposition experiments.

Larval Feeding Experiment

A larval feeding assay was performed with detached leaves in plastic containers (24 × 18 × 7 cm). One detached leaf from a HIPV-exposed plant (as above) and one from an undamaged plant were placed in opposite sides of an experimental box. The cut petioles were put into glass vials containing water and sealed with Parafilm. A larva of S. littoralis was released in the centre of the boxes, between the two leaves. The selected leaves were photocopied before the onset and again after the termination of the experiment. Leaf area consumption was calculated using a 5 × 5 mm grid.

Larval feeding bioassays were performed by comparing cotton plants pre-exposed to HIPVs with unexposed plants. The difference between the plants was tested by comparing the area consumed by the larvae of S. littoralis during 24 h on leaves from four different positions, namely 2nd oldest true leaf, 5th oldest true leaf, top side-shoot leaf and youngest leaf.

Data Analysis

The effect of damaged neighbours on the number of eggs laid on undamaged plants within cotton, alfalfa or clover plant patches, in both laboratory and field experiments, was analysed using paired-sample t-test, and the level of significance was selected as α = 0.05. Similarly, larval feeding preference in terms of the area consumed on leaves from cotton plants either pre-exposed or unexposed to HIPVs was analysed using the paired-sample t-test. The distance effect of HIPVs on number of eggs laid on the plants was analysed with GLM-anova with cages as blocks, and Tukey's post hoc test was used for pairwise comparisons. Normality and homoscedasticity of the residuals were checked graphically. Post hoc power analysis was used to analyse the likelihood of a significant effect of plant–plant communications. All statistical analyses were performed using minitab 16 software (v; Minitab; State college, PA, USA). Microsoft® office excel 2008-Software and Adobe® illustrator cs4 2008 were used for calculations and graphical representation of the data.


Intraspecific Effect

In the laboratory oviposition experiments, female moths of S. littoralis laid more eggs on cotton receivers that were adjacent to undamaged cotton emitters (72% ± 4.16 SE), than on receivers that were adjacent to cotton emitters damaged by S. littoralis larvae (28% ± 1.62 SE) (paired-sample t-test; t12 = 3.74, = 0.007) (Fig. 4a). Similar results were obtained under field conditions, where females laid a significantly higher proportion of egg batches on receivers with undamaged cotton emitters (69% ± 2.66 SE) than on receivers with damaged cotton emitters (31% ± 1.26 SE) (paired-sample t-test; t12 = 4.55, = 0.001) (Fig. 4b).

Figure 4.

Oviposition choice of Spodoptera littoralis for cotton receivers neighbouring either damaged (black bars) or undamaged cotton (white bars) emitters. A paired-sample t-test was used to test the statistical significance of differences between the damaged and undamaged sides under (a) laboratory (= 12) and (b) field (= 12) conditions. Bars represent mean (±SE) percentage of eggs, with different letters on bars indicating significant differences between damaged and undamaged plant patches at α = 0.05.

Interspecific Effect

In heterospecific plant combinations, the presence of damaged or undamaged alfalfa or clover emitters did not affect the egg laying of S. littoralis on cotton receivers. Females laid 46% (±4.51 SE) of the eggs on cotton receivers adjacent to undamaged alfalfa emitters, while they laid 54% (±4.97 SE) of the eggs on those close to damaged alfalfa emitters (paired-sample t-test; t12 = 0.77, = 0.456) (Fig. 5a). On clover too, eggs were equally distributed between cotton receivers close to undamaged (58% ± 6.05 SE) and damaged clover emitters (42% ± 8.34 SE) (paired-sample t-test; t8 = 0.96, = 0.370) (Fig. 5b). However, S. littoralis females laid fewer eggs on alfalfa receivers close to damaged cotton emitters (17% ± 1.13 SE) compared with undamaged cotton emitters (83% ± 3.65 SE) (paired-sample t-test; t6 = 2.88, = 0.035) (Fig. 5c).

Figure 5.

Oviposition preference of Spodoptera littoralis where female moths were allowed to oviposit on cotton receivers in the presence of either (a) damaged (black bar) or undamaged alfalfa (white bar) emitters (= 12) and (b) damaged (black bar) or undamaged clover (white bar) emitters (= 8). (c) Oviposition preference for alfalfa receivers in the presence of either damaged (black bar) or undamaged cotton (white bar) emitters (= 6). A paired-sample t-test was used to test for statistical significance. Bars represent mean (±SE) percentage of eggs oviposited under damaged and undamaged alfalfa, clover and cotton plant emitters. Different letters indicate significant differences between damaged and undamaged plant patches at α = 0.05.

Distance Effect

When female moths were given a choice of ovipositing on undamaged plants at different distances from damaged plants, a significant increase in the proportion of eggs laid was observed on moving from plants at 30 cm (10% ± 0.53 SE) to 60 cm (34% ± 0.93 SE) and 90 cm (56% ± 1.32 SE) from the damaged plants (anova: F2, 8 = 16.03, < 0.001, = 10) (Fig. 6).

Figure 6.

Oviposition preference of Spodoptera littoralis among three undamaged cotton plants placed at different distances from damaged cotton plants (= 12). GLM-anova with cages as blocks and Tukey's post hoc test were used for statistical analysis. Bars represent mean (±SE) percentage of eggs on all plants within the cages, and different letters on bars indicate significant differences in selection among the undamaged plant over distance for oviposition (α = 0.05).

Pre-Exposure to HIPVs

No significant difference was found in the oviposition experiments between cotton plants unexposed (58% ± 6.71 SE) and pre-exposed to HIPVs (42% ± 7.54 SE) (paired-sample t-test; t12 = 1.93, = 0.079). However, a power analysis showed that with the given effect size 0.56 (mean = 881, SD = 1581) and = 12, the power was 42%. For a power of 90% and the given effect size, = 36 would have been required. No difference in larval feeding was found between leaves from cotton plants pre-exposed to HIPVs and unexposed cotton plants (Table 1). The overall area consumed was 49% for leaves of cotton plants pre-exposed to HIPVs and 51% for unexposed cotton plants.

Table 1. Leaf area consumption from youngest leaf (YL1), top side-shoot leaf (TSSL), 5th oldest true leaf (OTL5) and 2nd oldest true leaf (OTL2) from HIPV pre-exposed vs. unexposed cotton G. hirsutum plants by third instar larvae of S. littoralis during a period of 12 hours
TreatmentsLarval feeding duration (h)Leaf area consumption (mm2)
  1. a

    Paired-sample West was used to compare the leaf area consumption between cotton plants pre-exposed and unexposed to HIPVs in all replications (N = 15–18).

Pre-exposed12120 ± 15100 ± 20200 ± 25140 ± 20
Unexposed 130 ± 590 ± 15240 ± 30150 ± 12
N 17181517
P a 0.8630.7000.6280.831


Our results showed a reduction in oviposition by S. littoralis on undamaged plants in the presence of ongoing larval damage to cotton plant neighbours. Emission of HIPVs from herbivore-damaged cotton plants changed the attractiveness of plants close to a damaged plant and influenced host plant patch selection in Slittoralis. We also demonstrated that the associational resistance via HIPVs was not a general phenomenon among the host plants of S. littoralis and that cotton as an emitter affected the resistance of both conspecific and heterospecific neighbouring plants. On the other hand, intact cotton receiving HIPVs from other host plants of S. littoralis, clover and alfalfa proved not to be more resistant to oviposition, showing that the associational resistance was directional and limited to plants exposed to HIPVs from damaged cotton plants. Species-specific limitations have also been observed in other plant species. For example, wild tobacco (Nicotiana attenuata) plants growing near experimentally clipped sagebrush displayed significantly less leaf loss than tobacco plants growing near unclipped sagebrush plants, whereas reciprocal effects were not observed (Karban & Maron 2002). Furthermore, clipped sagebrush plants did not protect other neighbouring herbaceous plants from herbivore damage in the field (Karban et al. 2003).

For ovipositing females, HIPVs can indicate lowered food quality, increased pressure from natural enemies and risk of competition on the damaged plant (Rasmann et al. 2005; Dicke & Baldwin 2010; Heil & Karban 2010). In addition, our results indicate that it can be beneficial for ovipositing females to use HIPVs to avoid undamaged host plants in patches of damaged plants. Thus, the plants already under herbivore attack can serve as a source of larvae migration onto neighbouring undamaged plants and affect the suitability of these plants as food (van Dam, Hadwich & Baldwin 2000). In a study on S. frugiperda larvae, growth rates were found to be reduced at higher larval densities on tomato plants and it would have been advantageous for the larvae to leave infested plants if neighbouring plants offered food of better quality (Underwood 2010). Feeding by migrating larvae would reduce food quality, but would also lead to food resource competition (Underwood, Anderson & Inouye 2005). Migration of younger larvae from the oviposition plant has been observed in S. littoralis in both field and laboratory experiments, while less movement of larvae towards previously damaged plants has been observed (Anderson, Sadek & Wäckers 2011; Sadek 2011). In those studies, the oviposition plants received less feeding from late instar larvae and experienced overall lower defoliation levels than initially undamaged plants. The response to HIPVs, linked to changes in plant quality, can thus influence the spatial distribution of S. littoralis between patches of plants and thereby the population dynamics of both the herbivore and the plants (Underwood, Anderson & Inouye 2005).

The effects of the associational resistance may extend beyond the species studied in these experiments. We found that emission of HIPVs from cotton influenced the distribution of eggs and the dispersal of the generalist S. littoralis, so it is possible that other generalist herbivores would be affected in the same way. On the other hand, specialist herbivores on different plant species may be less affected by HIPVs from a neighbouring plant (Ali & Agrawal 2012). Different responses of specialist and generalist herbivores to plant defence responses could lead to changes in the temporal and spatial dynamics within the habitat, depending on the other plants with which the host plant is associated (Whitham et al. 2012). In addition, HIPVs affect the distribution and dispersal of natural enemies between and within plant patches (Turlings & Wäckers 2004; Tentelier & Fauvergue 2007), which can have direct effects on the reproduction and distribution of both generalist and specialist herbivores. Thus, neighbouring plants can have direct effects on plant fitness and this can lead to either increased susceptibility or resistance of the host plant (Barbosa et al. 2009), but can also have wider effects on the ecological interactions within a habitat (Bascompte 2009; Whitham et al. 2012).

We found that HIPVs alone were sufficient to affect oviposition behaviour in female S. littoralis, as the females had no physical access to the damaged plants. Our results suggest a mechanism of associational resistance through HIPVs that is based on repellent plants affecting the oviposition decisions of the herbivore during host plant choice (Atsatt & O′Dowd 1976; Agrawal, Lau & Hambäck 2006). The volatiles either repelled the females during orientation towards the plant or affected female oviposition behaviour during assessment after landing. Repellent effects of HIPVs have been reported in other Lepidopteran species (De Moraes, Mescher & Tumlinson 2001; Kessler & Baldwin 2001). Although we found no significant effect on female oviposition by intact cotton plants exposed to HIPVs, we cannot exclude the possibility that other mechanisms may be involved in the repellency. It is possible that in addition to the effect on female oviposition behaviour, the volatiles from the damaged plant also induced a resistance response in the neighbouring plant that affected herbivore behaviour. On the other hand, no effect was found in the larval feeding experiment, whereas previous experiments have shown that herbivore-induced changes in cotton plants affect larval feeding behaviour in S. littoralis (Anderson, Jönsson & Mörte 2001; Anderson & Agrell 2005). However, it is possible that larval feeding and female oviposition choice in S. littoralis are guided by different induced cues. Recent studies on plant–plant communication have shown that exposure to HIPVs can induce resistance in neighbouring plants through eavesdropping (Heil & Karban 2010). For instance, in Lima bean (Phaseolus lunatus), exposure to HIPVs from conspecifics induced increased production of HIPVs as well as extrafloral nectar when the plant was subjected to herbivory (Heil & Kost 2006; Kost & Heil 2006). Similar responses have been observed between neighbouring plants of different species (Kessler et al. 2006). Another possible mechanism is that volatiles emitted from the damaged plants were adsorbed on the neighbouring undamaged plants and then re-emitted during the oviposition experiment (Himanen et al. 2010; Karban 2010). However, considering that the oviposition experiments were performed up to 4 days after volatile exposure and that we expected the volatiles to be re-emitted soon after exposure, this explanation is less likely.

The associational resistance extended to at least 60 cm from the nearest damaged cotton plant. This corresponds well with the distance of HIPV-induced resistance observed in sagebrush (Karban et al. 2006) and lima bean (Heil & Adame-Alvarez 2010). However, HIPVs from cotton may act over even longer distances, as female choice in our experiments was limited by the experimental cage size. Volatiles from damaged alder trees have been reported to reduce resistance at least 1 m from the damaged tree (Dolch & Tscharntke 2000). It is possible that the presence of damaged plants can affect the attraction of herbivores to larger patches of plants and not only to plants directly neighbouring damaged plants (Bukovinszky et al. 2010).

Cotton plants respond with both qualitative and quantitative changes in volatile emissions after larval feeding (McCall et al. 1994). The induced compounds are emitted locally from damaged leaves as well as systemically from undamaged leaves distal to the damaged site (Paré & Tumlinson 1998; Röse & Tumlinson 2005). Alfalfa and clover also have been shown to display damage-induced changes in emission of volatiles from their vegetative parts, and there is an overlap in the profile of volatiles emitted after damage between cotton and alfalfa (Blackmer et al. 2004; Kigathi et al. 2009). Resistance against attacking herbivores by volatiles from damaged clover has been recorded (Jiang, RidsdillSmith & Ghisalberti 1997). It is difficult to conclude whether these similarities indicate a cross-plant species set of volatiles indicating herbivore damage, since we do not know which compounds are behaviourally active. However, in our study, HIPVs from cotton affected oviposition behaviour in S. littoralis more strongly than HIPVs from alfalfa and clover. In terms of the volatile emissions from undamaged and herbivore-damaged plants, cotton generally shows larger qualitative and quantitative differences than clover (McCall et al. 1994; Kigathi et al. 2009). It is possible that the induced volatile emissions from cotton have a larger active range than the emissions from the other plants tested and thereby larger effects on neighbouring plants. Furthermore, the qualitative differences in induced cotton volatiles may more clearly signal herbivore induction in the plant and may serve as a better indicator of reduced plant quality for the ovipositing female.

Our experiments provide support for a mechanism of associational resistance involving HIPV emissions from plants and their effect on herbivore host plant selection, as well as for extended ecological significance of these compounds. Further studies are needed to identify the volatiles involved and to study olfactory-based interactions affecting the spatial distribution and fitness of S. littoralis and its host plants. This could also increase our understanding of intra- and interspecific interactions based on HIPVs between plants within the same patch and the costs and benefits of HIPV emission.


We thank Elisabeth Marling for her help with the insect rearing and Wael Samir for his assistance with field tests in Egypt. Peter Hambäck, Teun Dekker and the members of Chemical Ecology Group in Alnarp are gratefully acknowledged for comments on an earlier version of the manuscript. This study was supported by the Linnaeus initiative ‘Insect Chemical Ecology, Ethology and Evolution’ IC-E3 (Formas, SLU), the Swedish Research Council (VR), the European Science Foundation and Higher Education Commission (HEC) of Pakistan.