Plants respond to insect herbivores with changes in physical and chemical traits, both locally and systemically, in leaves and flowers. Such phenotypic changes may influence the behaviour of every community member that interacts with the plant. Here, we address effects of plant responses to eggs and subsequent herbivory by caterpillars on plant-mediated interactions with pollinators and consequences for plant fitness.
Using a common garden set-up, we have investigated responses of Brassica nigra plants to herbivore exposure from egg deposition onwards throughout larval development. We quantified effects of infestation by the specialist Pieris brassicae on: 1. behaviour of pollinators; 2. volatile emission and 3. timing and number of seeds produced.
Egg deposition and folivory did not influence visitation by pollinators to plots of infested or control plants. Effects of herbivore infestation on both pollinator visitation and volatile emission were observed only at a later stage, when caterpillars were feeding on the flowers.
Remarkably, before eggs had hatched, infested plants accelerated seed production. The caterpillars that developed from the eggs fed on flowers but not on seeds and thus seed production prior to herbivory on flowers safeguarded reproductive output.
The results of this study show that early plant investments in reproduction can successfully prevent consumption of expensive reproductive tissues. By accelerating seed production, plants prevented consumption of flowers and effectively defended themselves against the herbivores.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Plants respond to insect herbivores by changing physical and chemical traits (Kessler & Baldwin 2002; Erb et al. 2008; Dicke & Baldwin 2010). Induced responses can be expressed in leaves and flowers of damaged plants (Geervliet, Vet & Dicke 1994; Turlings & Fritzsche 1999; Dannon et al. 2010), and such phenotypic changes may influence the behaviour of every insect that interacts with the plant (Ohgushi 2005; Dicke 2009). Most attention has been paid, however, to the responses of arthropod predators and parasitoids foraging for host or prey (Vet & Dicke 1992; Kessler & Baldwin 2001; Dicke & Hilker 2003; D'Alessandro & Turlings 2006; Hilker & Meiners 2011). Yet, phenotypic responses by plants may also influence the behaviour of pollinating insects (Kessler & Halitschke 2009; Lucas-Barbosa, van Loon & Dicke 2011).
The enormous diversity of flower shapes, sizes, colours and odours has evolved among angiosperm plants to advertise the reward offered by the flowers to pollinating insects (Harder & Barrett 2006). Any phenotypic change in these traits may be associated by the pollinating insects with the quality of nectar and pollen offered as reward by the flowers (Weiss 1991; Raguso 2008; Rodriguez-Saona et al. 2011). Flower traits that change upon herbivore attack may positively or negatively influence foraging preferences of pollinators (Kessler & Halitschke 2009; Lucas-Barbosa, van Loon & Dicke 2011). Root herbivory, for instance, positively influenced pollinator behaviour, rendering plants more attractive to flower visitors (Poveda et al. 2003). In contrast, herbivore damage to leaves and flowers negatively affected foraging preferences of pollinators, according to results of most studies performed so far (Kessler & Halitschke 2009). A negative impact on pollinator preference has not always resulted in negative impact on plant reproductive success. Plants may compensate for herbivory despite the negative effects of herbivore damage on the behaviour of pollinators (Lehtilä & Strauss 1997; Strauss, Conner & Lehtila 2001; Steets, Hamrick & Ashman 2006).
In nature, plants that are flowering are exposed to pollinators and herbivores at the same time, and the interaction between plant and herbivore mostly starts at egg deposition (Hilker & Meiners 2011). In this context, we have used a common garden set-up to investigate responses of Brassica nigra plants to herbivore exposure, throughout larval development, since egg deposition. We were interested in the overall effects on plant fitness. We quantified effects of herbivore infestation by the specialist Pieris brassicae (Fig. 1) on (i) behaviour of pollinators; (ii) the emission of volatiles that may be used by pollinators and (iii) timing of seed set and number of seeds produced, including the contribution of day-active and night-active pollinators to seed production. As herbivores were exposed to naturally occurring predators and parasitoids, we also estimated dispersal and mortality of caterpillars during the experiments.
Material and methods
Insects and plants
Pieris brassicae L. (Lepidoptera: Pieridae) is a specialist herbivore of brassicaceous plant species. Female butterflies lay clusters of up to 100 eggs on the underside of leaves. When the eggs hatch, caterpillars will initially feed on the leaves of a flowering plant, but the second-instar larvae will move to the flowers and become florivorous (Smallegange et al. 2007). The fifth-instar larvae leave the plant to find a secluded pupation site. Brassica nigra L. (Brassicaceae) is considered to be an obligately out-crossing species (Conner & Neumeier 1995) and produce only hermaphroditic flowers. This plant species is a short-lived annual and a host plant of P. brassicae. In the Netherlands, P. brassicae caterpillars are regularly attacked by Cotesia glomerata L. (Hymenoptera: Braconidae) parasitoids.
In this study, P. brassicae was reared on Brassica oleracea var. gemmifera (Brussels sprouts) plants in a climate room (22 ± 1 °C, 50–70% r.h., L16:D8); the adults were provided with a 10% sucrose solution as food, in 5-mL vials with an opening in the centre of an artificial flower. Seeds of B. nigra were obtained from the Centre for Genetic Resources (CGN, Wageningen, the Netherlands) from an early-flowering accession (CGN06619) and have been multiplied by exposing plants to open pollination in the surroundings of Wageningen. Seeds collected from 25 plants were mixed to obtain seed batches for the experimental plants. Potted B. nigra plants were reared outside on tables protected by insect screens, in a location close to the field site.
Flowering plants with a few open flowers [growth stage 4·2, based on classification for Brassica napus (Harper & Berkenkamp 1975)], were infested with one egg clutch of P. brassicae by exposing plants to butterflies in an oviposition cage (100 cm × 70 cm × 82 cm). Flowers of the plant were covered with a mesh bag during exposure to butterflies to prevent flower visitation. The number of eggs on a plant was reduced to thirty by gently removing surplus eggs shortly after the plants had been removed from the oviposition cage. We infested four plants per day during four consecutive days, for each field trial. Egg-infested and control plants were transferred to the field right after egg deposition.
Effects of herbivore infestation on pollinator behaviour and consequences for plant fitness
Common garden experiment–field layout
Field layout consisted of 16 plots of B. nigra plants infested with P. brassicae and 16 control plots. Each plot (50 cm × 50 cm) was composed of five plants. The central plant of the plot was either infested with 30 P. brassicae eggs or was noninfested. The other four plants were all noninfested. Plots with an infested central plant are called ‘infested plots’ and plots with a noninfested central plant are called ‘control plots’ (Fig. 2). Equal numbers of control and infested plots were transplanted to the field on each of the 4 days. Plants that were infested on the same day were never planted in the same column or row in the field layout. Control and infested plots were planted alternately, and the distance between them was 1·5 m. Once plants had been transferred to the field, no attempt was made to prevent further infestation by any other herbivores. We carried out two serially repeated trials between 23 May and 23 July 2011, at an experimental field site in Wageningen, the Netherlands.
Dispersal and mortality of Pieris brassicae
The number of P. brassicae eggs on the egg-infested plants was counted at the end of egg development phase, as judged by darkening of the egg tips. If mortality of eggs or of the first-instar larvae appeared to be higher than 50% in a given plot, the egg-infested plant of the plot received fifteen first-instar caterpillars (i.e. 50% of initial number of eggs) obtained from the insect culture. To keep track of the actual amount and location of herbivore activity, we estimated dispersal and mortality of caterpillars by counting the number of individuals present on each of the five plants per plot, for each of the 16 infested plots in both trials. Number of caterpillars and their position on the plant were scored every other day. Caterpillar position was recorded in three categories, on (i) a leaf, (ii) a flower/bud or (iii) the stem. At the end of the experiments, when plants were harvested, caterpillars that were found on the plants were placed in Petri dishes to assess whether they had been attacked by a parasitoid. These caterpillars were subsequently kept in a greenhouse compartment (22 ± 1 °C, 50–70% r.h., L16:D8) and were fed with B. nigra flowers/leaves until pupation or until parasitoid larvae emerged from the caterpillar. Parasitism frequency was calculated as the number of parasitized caterpillars divided by the sum of hatched herbivore eggs and inoculated caterpillars.
Effects of herbivore infestation on visitation by pollinators
Pollinator visitation to infested and control plots was observed throughout P. brassicae egg and larval development. We observed the behaviour of pollinators at six time points spaced between the moments when the plants were infested with eggs until the time at which larvae had reached the final larval instar. The six time points were classified into three groups, as follows: (i) 24 and 144 h after egg deposition (egg infestation only); (ii) 24 and 120 h after caterpillars had hatched (mainly leaf feeding) and (iii) 216 and 312 h after caterpillars had hatched (on-going flower feeding). When a pollinator had contact with a flower, we recorded the pollinator's family [for Apidae, we distinguished bumblebees, Apis mellifera (honeybees) and ‘other Apidae’], number of flowers visited and time spent per flower. If other pollinator individuals happened to enter the plot under observation, we recorded only their visitation and family. If the same pollinator individual returned to the plot under observation after having visited a different plot, we scored that visit as a new visit. Each plot was observed for 10 min, and visitations were recorded using a handheld computer (Psion Workabout Pro™ 3, London, UK) programmed with The observer xt software (version 10; Noldus Information Technology, Wageningen, the Netherlands). Observation of pollinator visitation to infested and control plots was performed alternately and a maximum of eight plots were observed within a day. Observations were carried out whenever weather conditions were suitable (17–25 °C – wind speed ≤6 m s−1) between 9 a.m. and 4 p.m. We did not record number of flowers visited by pollinators for the first time point, that is, when infested plants carried only eggs, for trial 1.
Effects of herbivore infestation on plant growth and seed production
In the same field set-up as described previously, we investigated whether treatment affected growth and reproduction of B. nigra plants. At three time points per week, we scored which plants started to produce seeds. When caterpillars had reached the last larval instar, plants were harvested and siliques separated from shoots. We quantified seed production as a measure of reproductive success of B. nigra plants infested with P. brassicae compared with that of control plants. Seeds produced by infested and control plants were counted and weighed. When the total number of seeds was smaller than one hundred for an individual plant, all seeds were counted. When number of seeds was greater than one hundred seeds, the total number of seeds was estimated as total seed weight divided by the weight of one hundred seeds of that plant. To evaluate plant growth, we measured aboveground biomass of infested and control plants. Plant shoots were dried overnight (105 °C), and dry biomass was determined.
Contribution of day- and night-active pollinators to seed set of Brassica nigra plants
The contribution of day- and night-active pollinators to seed set of B. nigra plants was investigated in the same location as described previously. The experiment was carried out between 25 July and 19 August 2011. During this period in the Netherlands, the sun rises around 6:30 a.m. and sunset is around 8:00 p.m. Field layout was similar to the one used for the trials described previously. In total, twenty-four plots of five plants each were transplanted to the field; half of the plots were infested with P. brassicae eggs (infestation procedure described earlier) and half were noninfested control plots. Plants in twelve plots were exposed to pollinators during daytime (7:00–7:30 a.m. to 7:00–7:30 p.m.), and covered with a tent (BugDorm, 95 cm × 95 cm × 190 cm, white fine mesh in polyester 100 × 80 square/inch) during the night (7:00–7:30 p.m. to 7:00–7:30 a.m.). The other 12 plots were exposed to pollinators during the night and covered with a tent during daytime. Twenty-seven days after egg infestation, plants were harvested and siliques separated from shoots. Seed production and shoot dry biomass were quantified as described in the previous paragraph.
Headspace collection of plant volatiles in the field and analysis by GC-MS
We collected plant volatiles from aerial parts of infested and control plants. Headspace samples from egg-infested, leaf-infested, flower-infested, and similarly aged noninfested control plants for each of the three groups were collected between 25 July and 16 September 2011, whenever weather conditions were suitable, and in the same location, as described earlier. Aerial parts of the central plants of infested and control plots were enclosed with an oven bag (Toppits® Brat-Schlauch; polyester; 32 cm × 32 cm × 100 cm). Bags were closed around the stem and above the flowers with a strip of bag material. We pumped air from the environment into the bag at a flow rate of 300 mL min−1 (224-PCMTX8; air-sampling pump Deluxe, Dorset, UK; equipped with an inlet protection filter) by inserting Teflon tubing through an opening in the upper part of the bag. By inserting a second Teflon tube at the opening of each bag and connecting it to a glass tube filled with about 90 mg of Tenax-TA 25/30 mesh (Alltech, Breda, NL), air was sucked out and headspace volatiles were collected in the glass tube filled with Tenax for 1·5 h at a flow rate of 250 mL min−1 (224-PCMTX8; air-sampling pump Deluxe). Bags were discarded after use. Plant volatiles of egg-infested, leaf-infested, flower-infested, and control plants of the same stages were collected from in total 16 infested and 16 control plants, over two trials.
Headspace samples were analysed in a gas chromatograph with a thermodesorption unit (GC) (6890 series; Agilent, Santa Clara, CA, USA) connected to a mass spectrometer (MS) (5973 series; Agilent). Collected volatiles were desorbed from the Tenax in a thermodesorption trap unit (Gerstel, Mülheim, Germany) by heating from 25 to 250 °C (5 min hold) at a rate of 60 °C min−1 in splitless mode. Released compounds were focused in a cold trap (ID 1·80 mm) filled with glass beads (d 0·75–1·00 mm) at a temperature of −50 °C. By flash heating of the cold trap to 250 °C at 12 °C s−1, volatiles were transferred to the analytical column (60 m × 0·25 mm ID, 0·25 μm film thickness, DB-5; J&W, Folsom, CA, USA). Oven temperature programme started at 50 °C (1 min hold) and rose at a rate of 20 °C min−1–100 °C and subsequently increased at a rate of 4 °C min−1–280 °C (1·5 min hold) and finally rose up to 300 °C at a rate of 10 °C min−1. Column effluent was ionized by electron impact ionization at 70 eV. Mass scanning was carried out from 40 to 300 m/z with 5·36 scans s−1. Compounds were identified by comparison of mass spectra with those of NIST, Wiley libraries and the Wageningen Mass Spectral Database of Natural Products. Identity was confirmed by comparison of retention index described in the literature and the respective index calculated during this study. Compounds were quantified when detected in at least 50% of infested or control plants for each of the three stages.
Behaviour of day-active pollinators in infested and control plots was quantified when plants carried only eggs and when caterpillars were primarily feeding on leaves or flowers. Time spent per flower, mean number of flowers visited and mean number of flower visitors to infested and control plots were quantified for the main groups of flower visitors observed: bumblebees, honey bees, other Apidae, Syrphidae and Lepidoptera. Among the syrphid flies, Eristalis tenax was the main species observed, and Bombus lapidarius was the most abundant bumblebee species (Fig. 3). The cabbage white butterflies Pieris rapae and P. brassicae were the main lepidopterans observed visiting flowers of B. nigra plants, but their numbers were not large enough to be statistically analysed.
Bumblebees were by far the most abundant group of pollinator insects observed throughout trial 1 (results not shown) but were no longer flying during trial 2. In trial 1, flowers of B. nigra were more frequently visited by bumblebees than by honeybees (anova, Tukey post hoc test, P =0·025) and syrphid flies (anova, Tukey post hoc test, P =0·001), during the flower infestation period. Also, during the leaf infestation period, bumblebees visited a larger number of B. nigra flowers than did syrphid flies (anova, Tukey post hoc test, P =0·037).
Neither during the development of the butterfly eggs nor during the time when the caterpillars were feeding on the leaves did we observe effects of herbivore treatment on pollinator behaviour. However, effects of herbivory on pollinator behaviour were observed at the latest stage of infestation, when caterpillars were feeding on the flowers of infested plants. In trial 1, florivory influenced the behaviour of bumblebees and syrphid flies in different ways. When most caterpillars were feeding on flowers, bumblebees visited more flowers of infested plots than flowers of control plots (Fig. 4, Student's t-test, P =0·048). Syrphid flies spent more time on flowers of control plots than on flowers of infested plots (Fig. 4, Student's t-test, P =0·034). Behaviour of honeybees was not influenced by herbivore infestation in either of the two trials. In trial 2, infested plots received as many visits as control plots by all insects observed (See Fig. S1, Supporting information). No overall effect of treatment was observed (Tables S1 and S2, Supporting information, repeated measures anova). Number of pollinators increased throughout the field season, and, consequently, a larger number of flowers were visited later in the season (Tables S1 and S2, Supporting information).
Plant volatile emission
Volatile blends produced by infested and control plants changed quantitatively and qualitatively throughout the flowering stage and during seed development (Table 1 and Fig. S2, Supporting information). A Projection to Latent Structures Discriminant Analysis (PLS-DA) (SIMCA P + 12.0; Umetrics AB, Umeå, Sweden) including volatiles emitted by egg-infested, leaf-infested, flower-infested, as well as control plants at the same developmental plant stages resulted in a model with two significant principal components (explained variance 0·047) that separated the samples of the three plant developmental stages to a large extent, regardless of the herbivore treatment (Fig. S2, Supporting information). Monoterpenoids were characteristic of headspace samples of leaf-infested, flower-infested plants and their respective control plants but were not characteristic of headspace samples of egg-infested plants and their respective control plants (Table S3, Supporting information). Irrespectively of the treatment, the composition of the volatile blend emitted by plants quantitatively and qualitatively changed throughout the flowering stage, and these differences were more pronounced than the differences between treatments (Table S3, Supporting information). The total volatile emission by flower-infested plants was on average lower when compared with the respective control plants (Wilcoxon's signed-rank test, P =0·001). This was not the case when volatile blends from leaf-infested (Wilcoxon's signed-rank test, P =0·346) and egg-infested plants (Wilcoxon's signed-rank test, P =0·052) were compared with blends from control plants of the same developmental stages.
Dispersal and mortality of Pieris brassicae
The P. brassicae eggs hatched after about 10 days; mean temperatures during trials 1 and 2 were 15 and 17 °C, respectively. Survival of eggs per plot was higher during trial 1 (74%) than during trial 2 (27%) (Fig. 5). When eggs hatched, caterpillars initially fed on the leaves on which the eggs had been laid, and within 96 h, the caterpillars were found feeding on flowers (Fig. 6). Once caterpillars reached flowers they also dispersed to neighbouring plants of the same plot (Fig. 6).
Caterpillar survival was also higher during trial 1 (51%) than during trial 2 (29%) (Fig. 5). Eggs and early larval instars of P. brassicae were preyed upon mainly by ladybird beetles (Coccinellidae), whereas later larval instars were often predated by yellow jackets (Vespula spp.; D. Lucas-Barbosa, pers. observation). Of the caterpillars collected at the end of both trials, <1% was parasitized.
Plant growth and seed production
Brassica nigra plants that had been infested with P. brassicae eggs at the start of the experiment produced seeds sooner than control plants, at both plant and plot levels (Fig. 7). Elongation of siliques and seed formation was already observed at a very early stage of herbivore development in the infested plots, that is, when plants still carried eggs or when caterpillars had only just hatched (see Fig. 7b). For instance, in trial 1, we observed that 11 days after plants had been infested with P. brassicae eggs, a larger number of infested plants had produced seeds when compared to noninfested plants (Fig. 7, chi-square test, PA-plant= 0·001 and Pplot= 0·007). Moreover, in trial 1, infested plants produced more seeds than control plants, at both plant and plot levels (Fig. 8, student's t-test, PA-plant= 0·036 and Pplot= 0·012). In trial 2, infested plants and control plants produced equal numbers of seeds (Fig. 8, Student's t-test, PA-plant= 0·860 and Pplot= 0·740).
Infested plants produced as many seeds as did control plants, irrespective of having been exposed to pollinators during night or day. Plants that were exposed to night-active pollinators produced as many seeds as did plants that were exposed to day-active pollinators (results not shown). Shoot dry weights of infested and control plants were similar for all experiments carried out during this study (results not shown).
Our data show that B. nigra plants compensated for herbivory in terms of seed production, despite the changes observed in pollinator behaviour. Effects of herbivore infestation on pollinator behaviour were observed only when caterpillars were feeding on flowers of plants in infested plots. Egg deposition and folivory did not influence visitation by pollinators of infested plots compared with control plots of plants. Flower infestation influenced the behaviour of various pollinators in different ways. Bumblebees visited more flowers in infested plots than in control plots. Syrphid flies, however, spent less time on flowers of infested plots than on flowers of control plots. Comparable to our observations regarding pollinator behaviour, effects of herbivore infestation on volatile emission were also only observed in the final stage of the plant-herbivore interaction, that is, when caterpillars were feeding on the flowers. We collected headspace volatiles from egg-infested, leaf-infested, flower-infested, and control plants at the same stages, under field conditions, and observed that volatile emission by flower-infested plants was on average lower when compared with the respective control plants. Most studies investigating effects of herbivore infestation on the behaviour of pollinators have concluded that folivory and florivory mainly repel pollinators (Kessler & Halitschke 2009); that is, pollinators were less attracted to herbivore-infested plants when compared with control plants. Repellence of pollinators was in some instances associated with increased volatile emission (Kessler & Halitschke 2009; Zangerl & Berenbaum 2009). Indeed, plants in the vegetative stage usually respond to herbivory by increasing emission of volatiles (Mumm & Dicke 2010). Results of our study indicate that volatile emission can also be reduced in response to herbivore infestation. Thus, plants can increase or decrease volatile emission in response to herbivory, perhaps depending on their phenological state (Hare 2010; Diezel, Allmann & Baldwin 2011). Herbivore egg deposition can also result in reduced phytochemical responses, including reduced volatile emission, according to the studies performed so far (Bruessow et al. 2010; Fatouros et al. 2012). Pollination too may result in reduced emission of floral volatiles (Rodriguez-Saona et al. 2011). Irrespective of whether phytochemical responses to pollination and insect herbivores are increased or decreased, insects that interact with the plant may respond to these changes. The net effects on phytochemical responses, and ultimately plant fitness, will be influenced by plant-mediated interactions between herbivore and pollinators, starting from egg deposition by adult herbivores.
Surprisingly, in this study, B. nigra plants responded to egg deposition with accelerated seed production. Herbivore-infested B. nigra plants produced seeds sooner than did noninfested control plants. It is remarkable that this already occurred before the herbivore eggs had hatched. Eggs are the beginning of a new herbivore generation, and plants are known to respond to egg deposition (i) with responses that kill herbivorous larvae while developing in the eggs (Hilker & Meiners 2011), (ii) with changes that influence the performance of herbivorous larvae (Bruessow et al. 2010; Beyaert et al. 2012) and (iii) by inducing chemical cues that are exploited by egg parasitoids to locate their hosts and by butterflies during host plant selection (Blaakmeer et al. 1994; Fatouros et al. 2005). Our results show a novel response of plants to herbivore egg deposition. Early seed production in response to egg deposition by P. brassicae is likely to be advantageous for B. nigra. Caterpillars of P. brassicae are voracious, specialist leaf and flower feeders of B. nigra plants (Smallegange et al. 2007), and these plants considerably tolerate damage to leaves (Blatt et al. 2008). Pieris brassicae caterpillars actually prefer to feed on flowers of B. nigra plants rather than on leaves, consuming flowers entirely and in large numbers (Smallegange et al. 2007). These caterpillars do not feed, however, on seeds of B. nigra plants (D. Lucas-Barbosa, pers. observation). Thus, by accelerating seed production, B. nigra plants safeguard their reproduction before the caterpillars can consume the flowers.
In this study, accelerated seed production also resulted in compensation for herbivory, as infested plants produced as many seeds as noninfested control plants. Although plants compensated for herbivory, this was probably not mediated by modified interactions with pollinators. Brassica nigra is considered to be an obligately outcrossing species (Conner & Neumeier 1995) and thus dependent on pollinators for reproduction. The observation that herbivore-infested plants produced seeds sooner and in similar or larger numbers than noninfested control plants suggests that plots of infested plants received more visitation by pollinators than noninfested plots. Yet, we do not expect that the compensation in terms of seed production by the infested over control plants, during this study, could be the result of an efficient pollination service provided by bumblebees. Bumblebees indeed visited more flowers in infested plots than in control plots, but only during a later stage, when caterpillars were feeding on the flowers. An enhanced seed production rate was observed in the beginning of the experiments, when plants carried only eggs, and at this time point, egg-infested plots did not receive more visitation by bumble-bees than noninfested plots. This indicates that plant-mediated interactions between herbivores and pollinators do not explain how B. nigra plants compensated for herbivory.
The accelerated seed production recorded in trial 1 was also visible in trial 2, albeit that the effect was attenuated when compared to results from trial 1 (Fig. 7). The observation that egg mortality in trial 2 was much higher than in trial 1 (Fig. 5) reinforces our conclusion that plant responses to egg deposition triggered the accelerated reproduction in this system. Plants may compensate for herbivory by investing in self-pollination over outcrossing. Brassica nigra plants produce only hermaphroditic flowers; thus, we hypothesize that under stress, self-reproduction is favoured via autogamy. Plants can compensate for herbivory despite negative effects on the behaviour of obligate pollinators (Junker & Bluthgen 2010), and there is indeed evidence that direct plant responses to herbivore damage can increase plant reproductive success (Lehtilä & Strauss 1997; Steets, Hamrick & Ashman 2006; Wise, Cummins & Young 2008; Penet, Collin & Ashman 2009; Wise & Hebert 2010). Our data show that even before any herbivore damage had occurred, B. nigra plants responded to egg deposition and accelerated seed production.
Interestingly, early seed production was observed not only by the plants that were infested with P. brassicae eggs but also by the other four plants of the same plot that did not receive egg deposition (Fig. 7). The effects on the other four plants of the same plot may have been mediated either by pollinator attraction or by plant-plant communication. We did not observe that visitation rates by pollinators were higher in egg-infested plots than in control plots. Plant–plant communication may have occurred above- and below-ground (Heil & Karban 2010). Whether this has played a role remains to be investigated.
The annual plant B. nigra invested in seed production in response to egg deposition by P. brassicae, and not in increased volatile emission as typically observed after herbivory for plants in the vegetative stage. Our data show that accelerated investments in reproduction can successfully prevent consumption of expensive reproductive tissues and enable plants to compensate for herbivory. In other words, by investing in reproduction before herbivores attack reproductive organs, plants can effectively defend themselves against herbivorous insects.
We thank Nicolas Mairiniac and Yehua Li for helping with data collection in the field, Elbert van der Klift for the technical assistance and Maarten Posthumus for the fruitful discussions. This work was supported by Earth and Life Sciences council of the Netherlands Organisation for Scientific Research (NWO-ALW).