Airborne plant communication is a widespread phenomenon in which volatile organic compounds (VOCs) from damaged plants boost herbivore resistance in neighbouring, undamaged plants. Although this form of plant signalling has been reported in more than 30 plant species, there is still a considerable knowledge gap on how abiotic factors (e.g., water availability) alter its outcomes.

We performed a greenhouse experiment to test for communication between potato plants (Solanum tuberosum) in response to herbivory by the generalist insect Spodoptera exigua and whether communication was affected by water availability. We paired emitter and receiver potato plants, with half of the emitters damaged by S. exigua larvae and half serving as undamaged controls. Both emitter and receiver plants were subjected to one of two water availability treatments: high (i.e., well-watered) vs. low (i.e., reduced watering) availability, thus effectively teasing apart water availability effects on the emission and reception components of signalling. After four days of herbivore feeding, we collected emitter VOCs and receivers were subjected to feeding by S. exigua to test for effects of signalling on induced resistance.

Herbivory by S. exigua led to increased VOCs emissions as well as changes in VOCs composition in emitter plants. Furthermore, emitters subjected to low water availability exhibited a weaker induction of VOCs in response to herbivory relative to well-watered emitters. Results from the feeding bioassay indicated that receivers exposed to VOCs from herbivore-induced emitters showed lower S. exigua damage (i.e. higher induced resistance) compared to receivers exposed to undamaged emitters. However, we did not observe a significant effect of water availability in either emitters or receivers on plant communication.

Overall, our study contributes to the understanding of how the abiotic context affects plant communication by providing evidence of water availability effects on the induction of VOCs that may act as airborne signals between plants. The observed changes in induced VOCs had no visible consequences for plant communication. These findings thus suggest that the induction of key compounds mediating communication was not compromised by our experimental conditions.

In April 2021, we sowed 168 tubers from three different Solanum tuberosum varieties (cv. Baraka, cv. Desiree, and cv. Monalisa) in 4-L pots containing potting soil and peat (Gramoflor GmbH & Co. KG Produktion, Vechta, Germany). We grew plants in a glasshouse under controlled light (minimum 10 h per day, Photosynthetically Active Radiation = 725 ± 19 μmol m-2 s-1) and temperature (10°C night, 25°C day), and watered them twice a week up to field capacity. Five weeks after germination, we assigned half of the plants to one of two water availability treatments: high (i.e., well-watered) vs. low (i.e., reduced watering) water availability (Fig. 1). We watered plants in the high water availability treatment every three days to replenish the 100% of their water demand, whereas for plants in the low water availability treatment watering was reduced to meet the 25% of the total water demand. We estimated water demand gravimetrically.To corroborate that plants in the low water availability treatment were under higher physiological stress in comparison to well-watered plants, two weeks after the start of the treatments (right before applying the herbivory treatment, see below) we used a subset of 24 plants (half high and half low water availability; four of each potato variety) to measure stomatal conductance and photosynthesis. We measured stomatal conductance and net photosynthetic rate on a leaflet of a young, fully-expanded leaf from 11:30 to 12:30 am at an irradiance of 1500 μmol m^-2 s^-1 and CO₂ concentration of 400 μmol mol-1 with a portable photosynthesis system Li-6400XT (Li-Cor Inc., Lincoln, NE, USA). Plants in the low water availability treatment exhibited significantly lower stomatal conductance (F_1,22 = 22.1, P < 0.001) and photosynthetic rates (F_1,22 = 31.5. P < 0.001) compared to well-watered plants. Specifically, reduced watering resulted in a 90 % and an 80% decrease in stomatal conductance (high water availability: 0.073 ± 0.01 mol H₂O m^-2 s^-1; low water availability: 0.008 ± 0.01 mol H₂O m^-2 s^-1) and photosynthesis rates (high water availability: 9.31 ± 0.94 µmol CO₂ m^-2 s^-1; low water availability: 1.88 ± 0.94 µmol CO₂ m^-2 s^-1), respectively (Fig. S1a, b).

Seven weeks after germination (two weeks after establishing the water availability treatments), we paired 144 potato plants in 37.5 × 37.5 × 96.5 cm plastic cages to prevent VOCs cross-contamination between replicates. One plant of each pair (i.e., replicate) acted as the emitter (average height ± SE = 51.17 ± 0.64 cm) and the other served as the receiver (48.52 ± 0.70 cm). Within each cage, emitter and receiver plants were placed 20 cm apart so that they did not touch each other. Plants in each water stress treatment were randomly selected as either receiver or emitter plants resulting in a factorial design consisting on four combinations of water availability treatment in the emitter (two levels; high vs. low) and water availability treatment in the receiver (two levels; high vs. low) (Fig. 1).  In addition, we randomly assigned emitter plants within each cage to one of the following herbivory treatments: (1) subjected to S. exigua feeding (“herbivore-induced plants” hereafter) or (2) control (intact; no herbivory) plants (Fig. 1). Overall, the experiment consisted in 72 replicate cages, namely 36 for the herbivore-induced treatment (nine per emitter vs. receiver water availability combination) vs. 36 for the control (nine per emitter vs. receiver water availability combination). Emitter and receiver plants were always of the same variety and varieties were similarly distributed across treatment combinations. For herbivore-induced emitters, we placed two third-instar larvae of S. exigua on each of three fully expanded leaves per plant using a fine paintbrush and covered these leaves with a nylon bag to prevent herbivore dispersal. For control plants, we covered three fully expanded leaves with a nylon bag but without adding the larvae to control for a possible bagging effect. After four days of herbivore feeding, we removed all emitter plants from cages and collected VOCs from each emitter (see below). After VOCs sampling, we collected leaves subjected to larvae feeding and photographed them with a Samsung Galaxy A30s (25 effective megapixels, 4× digital zoom). We estimated the percentage of leaf area consumed using the mobile application BioLeaf - Foliar Analysis™ (Brandoli Machado et al., 2016). Average percentage leaf area consumed by S. exigua for herbivore-induced emitters was 77.58% (± 3.72) and was homogeneously distributed among plants in the high (80.46% ± 3.50) vs. low (73.63% ± 4.25) water availability treatments (F1,33 = 0.7; P = 0.399).

We collected aboveground VOCs produced by emitter plants following Rasmann et al. (2011). Briefly, we bagged plants with a 2-L Nalophan bag and trapped VOCs on a charcoal filter (SKC sorbent tube filled with anascorb CSC coconut-shell charcoal) for two hours at a rate of 0.25 L min-1. We eluted traps with 150 μL dichloromethane (CAS#75-09-2, Merck, Dietikon, Switzerland) to which we had previously added one internal standard (tetralin [CAS#119-64-2], 200 ng in 10 μL dichloromethane). We then injected 1.5 μL of the extract for each sample into an Agilent 7890B gas chromatograph (GC) coupled with a 5977B mass selective detector (MSD) fitted with a 30 m × 0.25 mm × 0.25 μm film thickness HP-5MS fused silica column (Agilent, Santa Clara, CA, USA). We operated the GC in pulsed splitless mode (250 ºC, injection pressure 15 psi) with helium as the carrier gas (constant flow rate 0.9 mL min-1). The GC oven temperature programme was: 3.5 min hold at 40ºC, 5ºC min-1 ramp to 230ºC, then a 3 min hold at 250ºC post run. Transfer line was set at 280 ºC. In the MS detector (EI mode), a 33-350 (m/z) mass scan range was used with MS source and quadrupole set at 230ºC and 150ºC, respectively. We identified volatile terpenes using the NIST MS Search Program v.2.3 and by comparison with the terpenes reference database developed at the University of Neuchâtel and based on pure standards. We quantified total emission of individual VOCs using normalized peak areas and expressed it as nanograms per hour (ng h-1). We obtained the normalized peak area of each individual compound by dividing their integrated peak area by the integrated peak area of the internal standard (Abdala-Roberts et al., 2022). The total emission of VOCs was then calculated as the sum of individual VOCs.

The same day after collecting emitter VOCs, we set up an herbivore bioassay for receiver plants to test whether prior exposure to VOCs from herbivore-induced emitters increased herbivore resistance. For this, we placed one third-instar S. exigua larvae on each of two fully expanded leaves per plant following the same procedure described above for emitter induction. We kept larvae on receivers for three days and then estimated the percentage of leaf area consumed by S. exigua (‘leaf damage’ hereafter) using the same procedure described above for emitter plants.