Plants can detect cues associated with the risk of future herbivory and modify defense phenotypes accordingly; however, our current understanding is limited both with respect to the range of early warning cues to which plants respond and the nature of the responses. Here we report that exposure to volatile emissions from plant tissues infested with herbivore eggs promotes stronger defense responses to subsequent herbivory in two Brassica species. Furthermore, exposure to these volatile cues elicited an apparent shift from growth to reproduction in Brassica nigra, with exposed plants exhibiting increased flower and seed production, but reduced leaf production, relative to unexposed controls. Our results thus document plant defense priming in response to a novel environmental cue, oviposition-induced plant volatiles, while also showing that plant responses to early warning cues can include changes in both defense and life-history traits.

Larval performance bioassays

The effect of exposure to oviposition-induced volatiles on plant defenses in both Brassica species was assessed via larval performance assays. Ten plants from each priming treatment (E, Re, Rc and C) received 10 neonate P. brassicae. On days three and seven following the initiation of feeding, larval mass was measured on a microbalance (accuracy +/- 1μg; Mettler- Toledo AG, Greifensee, Switzerland) as described in Pashalidou et al. (2013, 2015a, c)

Volatile collection and analysis

For both Brassica nigra and Brassica oleracea, we collected volatiles from plants exposed to the four priming treatments (N=12 per treatment) and with or without larval damage. For damage treatments, ten L1 larvae were placed on E, Re and Rc plants. Due to logistical constraints (and because our previous assays showed no effect of priming treatment Rc on larval performance) we collected volatiles only from damaged Rc plants and used C plants as undamaged controls. Larvae were placed on the adaxial side of the 3^rd highest leaf. One damaged plant was excluded from the damaged Re treatment because of unrelated damage. Volatile collections were made one day prior to larval emergence and two hours after the initiation of larval feeding. Pots were wrapped in foil to minimize plastic contaminants. Two connecting metal plates were closed around the plant stem (with a hole for the stem to pass), and cotton was used to seal gaps. A 30 L glass dome was carefully placed over the leafy parts of the plant, with openings for incoming and outgoing air, which was filtered through activated charcoal, pulled through the chamber at a rate of 150 ml/min for 4 h, and collected in a stainless-steel cartridge containing 200 mg of Tenex TA (20/35 mesh; CAMSO, Houston, TX, USA). Due to space limitations, volatile collections were conducted in three blocks. After volatile collection, the aboveground parts of the plant were cut and weighed.

Volatile compounds were eluted from the filter using 150 μL of internal standard solution (2 ng/μL octane and 4 ng/μL nonyl acetate in dichlormethane) and the eluant was analysed by gas chromatography-mass spectrometry (GC-MS). Two μL of the eluant was injected with an automatic Agilent injector 7693 autosampler (Santa Clara, CA, USA) to an Agilent 7890B GC (Santa Clara, CA, USA) with a pulsed splitless inlet at 250°C, which was held for 2min and then analyzed on the connected MS Agilent 5977A. Compounds were quantified and identified as described in supplementary methods (Appendix S1). Volatile emissions per plant were calculated as mean peak area divide by both the fresh weight of foliage (in grams) and by 10⁴ the n of samples.

Testing effects of exposure to individual volatile compounds

Because the emission of cumene was significantly elevated on egg-infested plants for B. oleracea (Table S2), we also explored the defense priming effects of this compound on B. oleracea and B. nigra. Unfortunately, we were unable to similarly test the effect of β-thujene—a compound showing elevated emissions following egg infestation in B. nigra (Table S3)—as we could not obtain this compound. We made a cumene solution containing 156µg/ml of synthetic cumene (Sigma-Aldrich) in hexane, a concentration approximating the mean daily emission of an egg-infested plant with a fresh aboveground mass of 200g. Over a five-day period, 50µL of this solution was applied daily to sleeve-stopper septa (Sigma-Aldrich) placed at a distance of 15cm from focal plants (treatment Cu; Table S1). The septa were placed at the height of the receiver’s apical meristem to simulate elevated cumene emission from an egg-infested plant. Control plants were similarly exposed to 50µL hexane (treatment He). Each of the 10 replicate plants per treatment was infested with 10 neonate larvae after exposure to cumene for five days, and larvae were weighed three and seven days after placement.

Testing effects of egg-induced volatiles on plant growth and reproduction

To test whether priming by oviposition-induced volatiles altered plant reproductive output we focused on Brassica nigra, as this annual species has been previously shown to respond to egg infestation through changes in reproductive phenology. We produced new plants using six treatments described in previous sections (C, E and Re with and without larval damage; Fig. 1, Table S1), omitting Rc plants which were similar to C plants in previous assays. Larvae were allowed to feed freely until pupation, with the larval number reduced from ten to three at the third instar stage to avoid complete defoliation. When larvae neared pupation, plants were covered with a fine net (to prevent larvae from leaving the plant), which was removed following pupation (plant treatments without larvae were similarly covered). We recorded the number of leaves and flowers present three weeks after the first flower appeared on each plant. Once all plants were flowering, commercial bumblebees (Biobest, Switzerland) were introduced for three weeks to ensure pollination; previous work indicates that bumblebees do not discriminate between undamaged plants and those with either P. brassicae eggs or feeding damage on leaves/flowers (Lucas-Barbosa et al. 2013). After plants had completed their life cycle, ripe seeds were collected from each plant and measured with a seed counter (elmor c3 version 1.1, Switzerland). Germination rates were measured as in Pashalidou et al 2015b.