Fungal volatiles influence plant defence against above-ground and below-ground herbivory

1. Plants have evolved resistance traits that negatively affect attackers, and tolerance traits that sustain plant growth despite herbivore damage. These mechanisms often co-occur in a mixed-defence strategy, balancing resistance and tolerance. These plant defences can be enhanced upon interaction with soil micro-organisms. emitted soil-borne fungi on plant defence to herbivory, As a proxy of plant resistance, we assessed the performance of Pieris brassicae , a caterpillar feeding on leaves and inflorescences, and of Delia radicum , an insect root herbivore. As a proxy of plant tolerance, we compared growth of volatile-exposed plants challenged with or without insects. Additionally, we assessed the effects on plant phe nology by recording bolting time and by counting the

plant colonisation by micro-organisms can affect these two components concurrently (Contreras-Cornejo, Macias-Rodriguez, del-Val, & Larsen, 2016;Hermosa et al., 2013). Microbial colonisation can condition plants to respond faster and stronger to a subsequent stress (Martinez-Medina et al., 2016;Mauch-Mani, Baccelli, Luna, & Flors, 2017) or can induce systemic plant resistance, for example, by negatively affecting the preference or performance of a herbivore or pathogen via plant metabolomic changes (Etalo, Jeon, & Raaijmakers, 2018;Pangesti et al., 2016;Pieterse et al., 2014;van de Mortel et al., 2012). These plant responses occur locally at the site of colonisation, but also systemically affect chemical and physical plant traits. As a consequence, root colonisation by soil micro-organisms can affect subsequent plant interactions above-ground and vice versa (Bezemer & van Dam, 2005;Pineda, Kaplan, & Bezemer, 2017;van Dam & Heil, 2010). Alongside resistance, tolerance, that is, the capability of plants to endure stresses that limit plant development, can be modulated upon microbial colonisation as well. Tolerance can be experimentally measured by determining the degree to which plant growth is affected by a given stress relative to its growth in the undamaged state (Strauss & Agrawal, 1999), and plant colonisation by micro-organisms can influence the degree of stress tolerance compared with uncolonised plants. For instance, micro-organisms can alleviate the negative effects of high salinity on plant growth by boosting photosynthetic rate (Han & Lee, 2005;Yang, Kloepper, & Ryu, 2009). These studies exemplify that plant colonisation by micro-organisms can enhance plant defence to biotic and abiotic stresses, and sustain plant fitness.
Yet, to our knowledge, it remains unknown whether microbial volatiles affect plant tolerance to insect herbivory and whether these responses are specific to the plant tissue(s) being attacked. Additionally, how plant resistance and tolerance to herbivory are concurrently modulated by microbial volatiles has not been addressed. Therefore, here, we investigated the effects of volatiles emitted by soil-borne fungi on plant tolerance and resistance to above-and below-ground insect herbivory, and on plant phenology. For this, we selected the brassicaceous plant species Brassica rapa and its natural herbivores: the cabbage root fly Delia radicum (Diptera: Anthomyiidae), whose larvae feed on B. rapa roots, and the large cabbage white butterfly Pieris brassicae (Lepidoptera: Pieridae), whose caterpillars feed on leaves and inflorescences of B. rapa plants. Furthermore, we selected four soil-borne fungi (Fusarium oxysporum, Rhizoctonia solani, Ulocladium atrum and Phoma leveillei) that all co-occur and interact with brassicaceous plants, and have a saprophytic phase in their cycle.
In the present study, all fungi were inoculated in ø 9 cm plastic Petri dishes containing one-fifth strength Potato Dextrose Agar (1/5th PDA). This medium was prepared with 7.8 g of PDA (Oxoid) and 14 g of Bacto TM Agar. The pH was set at 7. Fungi were incubated at 25°C in the dark for 7 days before the start of the exposure.
First instar caterpillars of P. brassicae feed on leaves, whereas later instars move to the inflorescence to feed on buds and flowers (Lucas- Barbosa, van Loon, Gols, van Beek, & Dicke, 2013

| Plant exposure to fungal volatiles
To expose B. rapa roots to fungal volatiles in vivo, we designed a two-compartment pot system ( Figure 1). One sterile B. rapa seed was sown in the top compartment (h = 20 cm, ø = 12.5 cm) filled with a sterile (i.e. autoclaved twice at 121°C for 20 min with 24 hr interval in between) soil mixture (1:1 v/v, ø 4 mm sieved Horticoop potting soil:sand), whereas the test fungus (F. oxysporum, R. solani, U. atrum or P. leveillei) was grown in a ø 9 cm Petri dish enclosed in the bottom compartment (h = 10 cm, ø = 12.5 cm). Both compartments were connected to each other by a cylinder (h = 12.5 cm, ø = 12.7 cm), and separated by a nylon membrane of 1 µm mesh width (ø = 14.5 cm) that allowed air exchange between the two compartments. Volatile exposure was initiated in a greenhouse compartment (21 ± 2°C; L16:D8; 70 ± 5% RH) with 7-day-old fungi as soon as B. rapa seeds were sown, and was maintained for 4 weeks, after which B. rapa plants had 6-8 fully developed leaves. Control plants were exposed to a Petri dish containing one-fifth PDA medium only.
Petri dishes containing the fungi and control were replaced weekly with Petri dishes containing fresh 7-day-old fungi or fresh one-fifth PDA medium. A total of 30 plant replicates was prepared for each fungal volatile exposure and divided in two experimental batches, with 1 week interval.

| Plant infestation with above-ground and below-ground herbivores
After the 4-week volatile exposure, all Petri dishes were removed permanently from the bottom compartments, and the B. rapa plants were either infested with one of the two insect herbivores or re-

| Effects of fungal volatile exposure on growth of uninfested plants
To assess the effects of fungal volatile exposure on plant growth, we measured the dry weight of 6-week-old uninfested B. rapa plants after the 4 weeks of exposure to fungal volatiles. Roots, leaves and inflorescences were harvested, dried at 105°C for 16 hr, and weighed. Effects of root exposure to fungal volatiles on the total plant dry weight as F I G U R E 1 Schematic representation of the two-compartment pot system used in this study to expose roots of Brassica rapa, growing in soil in the top compartment, to volatiles emitted by one fungus growing in a Petri dish in the bottom compartment or by a control Petri dish. Plants were subsequently infested with either caterpillars of the above-ground herbivore, Pieris brassicae, or with larvae of the below-ground herbivore, Delia radicum; or remained uninfested (no insects)

| Effects of fungal volatile exposure on plant growth upon above-ground and belowground herbivory
To assess the effects of fungal volatile exposure on plant tolerance to above-ground and below-ground herbivores, we measured the dry weight of B. rapa plants whose roots were previously exposed to fungal volatiles and then infested with either P. brassicae caterpillars or with D. radicum larvae, and compared it to that of volatile-exposed unin-  (Table S1). For each plant tissue and volatile exposure, the effect of herbivory was tested by comparing the above described differences to zero, using the t distribution (α = 0.05). For the visualisation of the data, we plotted the Cohen's D effect sizes by dividing the differences of least square means by the pooled standard deviation of each plant tissue. We ran the same analyses per batch.

| Effects of fungal volatile exposure on aboveground and below-ground herbivore performance
To assess the effects of fungal volatile exposure on plant resistance to above-ground and below-ground herbivores, we measured the performance of P. brassicae caterpillars and D. radicum larvae on plants whose roots were previously exposed to fungal volatiles. Individual fresh weight of P. brassicae caterpillars was assessed at 3 days post infestation (dpi) and at 7 dpi. At 3 dpi, all 20 P. brassicae caterpillars were recollected, and larval density was reduced by 50% to mimic natural to the method of Kenward and Roger (Kenward & Roger, 1997), followed by post hoc tests using the t distribution. We also tested the effects of fungal volatiles on caterpillar fresh weight per batch. At 14 dpi, we also scored the developmental stage (larvae or pupae) of recollected D. radicum, and individually weighed the recollected insects. A mixed model similar to that used for P. brassicae was used to analyse D. radicum fresh weight. Additionally, the number of recovered D. radicum (out of the 10 insects initially added) and the fraction of recovered D. radicum pupae were analysed using a beta-binomial (to handle binomial overdispersion) generalised linear model (GLM) and logit link function (Supporting Information; Moisan et al., 2020b). For this, we used the glmmTMB package in r (Brooks et al., 2017). In this model,

| Effects of fungal volatile exposure on plant bolting upon herbivory
To assess the effects of fungal volatile exposure on phenology of B. rapa plants upon above-ground and below-ground herbivory, we   Figure S1).

| Effects of fungal volatile exposure on plant growth upon above-ground and belowground herbivory
Overall, herbivory by P. brassicae negatively affected weight of above-ground tissues. This reduction was expected as P. brassicae caterpillars feed on above-ground tissues. Yet, the effects differed between tissues. When infested with P. brassicae, inflorescences of control plants (i.e. plants not exposed to fungal volatiles) and plants whose roots were exposed to R. solani volatiles weighed less than inflorescences of uninfested plants (Figure 3a; t tests; p control = 0.038; p R. solani = 0.005). Also, P. brassicae-infested plants exposed to U. atrum volatiles had lower leaf weight than uninfested plants (Figure 3a; t test; p = 0.020). However, leaf and inflorescence weights of plants whose roots were exposed to F. oxysporum and P. leveillei volatiles did not significantly differ between P. brassicae-infested plants and uninfested plants (Figure 3a and Table S1a; t tests; all p > 0.050).
Despite an overall reduction, infestation by D. radicum did not significantly impact root weight, irrespective of the fungal volatiles they had been exposed to (Figure 3b and Table S1b; t tests; all p > 0.050). However, infestation by D. radicum resulted in lower inflorescence weight upon exposure to R. solani volatiles (Figure 3b; t test; p = 0.023), and in lower leaf weight upon exposure to U. atrum volatiles (Figure 3b; t test; p = 0.039), compared with uninfested plants. We detected a significant interaction between batch and fungal volatiles (Table S2) thus, the effects of fungal volatiles varied between the two batches ( Figure S2).

| Effects of fungal volatile exposure on aboveground and below-ground herbivore performance
Plant exposure to fungal volatiles affected P. brassicae caterpillar weight at 3 and 7 dpi (Figure 4; LMM; p = 0.014 and 0.006, respectively). At 3 dpi, caterpillars feeding on plants whose roots were exposed to volatiles from R. solani or U. atrum were larger than those feeding on plants exposed to volatiles from F. oxysporum or P. leveillei Thus, the effects of fungal volatiles varied between the two batches ( Figure S3). Caterpillar weight at 7 dpi was neither correlated with leaf nor inflorescence weight of P. brassicae-infested plants (Table S3b; Pearson correlation tests; all p > 0.050), nor with the difference of F I G U R E 3 Cohen's D effect sizes [(differences of least squares means/ pooled SD) ± CI] in root, leaf and inflorescence dry weight of (a) 6-week-old Brassica rapa plants infested with Pieris brassicae and uninfested plants, and of (b) 6-week-old B. rapa plants infested with Delia radicum and uninfested plants, when exposed to volatiles of four different fungi (Fusarium oxysporum, Rhizoctonia solani, Phoma leveillei and Ulocladium atrum). Differences of least squares means were generated using a mixed model that allows for correlations and inequality of variances among plant tissues of the same plant, and were statistically tested per plant tissue and per fungal volatile exposure using the t distribution (*p < 0.050; **p < 0.010). 'N' indicates the number of plant replicates per treatment combination. Detailed information of the least squares means can be found in Plant exposure to fungal volatiles did not affect the number of D. radicum we recollected (Table S4;  indicates the total number of individuals (pupae and larvae) recollected. Each box-and-whisker shows the distribution of the dataset into quartiles: the minimum, first quartile, median, third quartile, and maximum. Dots show outliers. For the fraction of larvae and pupae, main effects of the volatile exposure, batch, and their interaction were tested using a generalised linear model with a beta-binomial distribution, and for the insect fresh weight we used a mixed model, with plant replicate as a random factor. Uppercase letters indicate pairwise differences in percentages of larvae and pupae between fungal volatile exposures using likelihood ratio post hoc tests (LRT). Each plant was infested with 10 D. radicum neonates, and each volatile exposure was replicated 7-10 times

| D ISCUSS I ON
We found that volatiles emitted by four soil-borne fungi differentially affected B. rapa resistance and tolerance to herbivory by P. brassicae caterpillars and D. radicum larvae. Effects on P. brassicae performance varied between the different fungi and between the batches, whereas the effects on D. radicum performance was predominantly negative, indicating an increased plant resistance.
Despite an overall reduction of root weight upon attack by the root herbivore D. radicum, B. rapa plants remained tolerant and compensated for the loss of root tissues. In contrast, attack by P. brassicae caterpillars led to an overall reduction of above-ground tissues and compensation varied between the tissues and between the fungal volatile exposures. Root exposure to R. solani or P. leveillei volatiles accelerated plant phenology, which resulted overall in more buds and flowers. Altogether, our data show that fungal volatiles can modulate plant mixed-defence strategy, balancing plant resistance and tolerance to above-ground and below-ground herbivory. These effects are variable and occur in a fungal-volatile-specific manner.

| Fungal volatiles influenced plant growth and tolerance upon herbivory
Exposure of B. rapa roots to fungal volatiles differentially affected growth of uninfested plants and insect-infested plants, thus influencing plant tolerance. Although D. radicum larvae feed intensively on roots, we did not observe a reduction of root weight compared to uninfested plants, suggesting that fungal volatileexposed plants remained tolerant and compensated for the loss of root tissues (Mesmin et al., 2019). Interestingly, upon certain volatile exposure, this compensatory growth occurred at the cost of undamaged above-ground tissues. In contrast, herbivory by P. brassicae caterpillars resulted in a reduction of B. rapa plant weight compared to uninfested plants. Nonetheless, these tissue losses differed between leaves and inflorescences and between fungal volatile exposures. These findings suggest that compensatory plant growth to herbivory may result from a reallocation of resources within the plant, for example, from above-ground tissues to roots (Núñez-Farfán, Fornoni, & Valverde, 2007;van Dam, 2009), and that plant exposure to fungal volatiles can specifically modulate this reallocation between tissues, sometimes at the cost of undamaged tissues. Yet, the effects of fungal volatiles on biomass of uninfested and insect-infested plants were variable between batches, suggesting that plant responses may be highly susceptible to the smallest variation in the fungal volatile emission. A thorough analysis of resource partitioning and allocation to storage and defence upon exposure to specific fungal volatiles will provide a better understanding of plant tolerance to herbivory following plant exposure to fungal volatiles.

| Effects of fungal volatiles on plant resistance differ between insect herbivores
Most fungal volatile exposures resulted in a reduced D. radicum development rate, indicating increased direct plant resistance, whereas the effects on performance of P. brassicae caterpillars differed between the fungi. For instance, P. brassicae caterpillars feeding on plants whose roots were exposed to R. solani volatiles were larger than those feeding on control plants and fungal volatileexposed plants, indicating higher plant susceptibility. This finding corroborates previous studies that reported larger Mamestra brassicae caterpillars when feeding on Arabidopsis thaliana seedlings exposed to VOCs from R. solani (Cordovez et al., 2017;Moisan et al., 2019). As the average insect fresh weights did not correlate with the difference of biomass between infested plants and uninfested plants, we conclude that the insect performances were not correlated with the amount of plant tissues consumed. Instead, slower development and lower body mass increase of the insect may result from changes in plant chemistry and morphological traits, which can lead to chemically or structurally more resistant roots upon fungal volatile exposure. For example, root exposure to fungal volatiles may alter architecture of primary and lateral roots (Casarrubia et al., 2016;Ditengou et al., 2015;Garnica-Vergara et al., 2015), which in turn, can negatively impact the performance of root herbivores (Felkl, Jensen, Kristiansen, & Andersen, 2005;Werner, Polle, & Brinkmann, 2016). Also, plant exposure to microbial VOCs can promote the accumulation of defensive secondary metabolites such as glucosinolates in leaves, which can diminish the performance of leaf caterpillars (Aziz et al., 2016). Levels of indole glucosinolates in the main roots can also slow down larval development (Van Dam & Raaijmakers, 2006).
For a comprehensive overview of plant resistance, it would be interesting to explore whether fungal volatiles affect behaviour of the insect herbivores. Plant exposure to fungal volatiles may have altered nutrient levels in some plant tissues, making the tissues repellent/attractive or unpalatable to the insect herbivores, thus positively or negatively impacting insect performance (Schoonhoven, Van Loon, & Dicke, 2005;Smallegange et al., 2007). As we did not monitor the position of caterpillar feeding over time, it is also plausible that fungal volatile exposure also influenced herbivore feeding preference by differentially altering the nutritional quality of leaves and inflorescences (Smallegange et al., 2007;Wetzel, Kharouba, Robinson, Holyoak, & Karban, 2016). Interestingly, the effects of fungal volatiles on P. brassicae performance were also influenced by the batches, which may be linked with the differential plant responses per batch, for example, plant biomass as discussed above.
A thorough analysis of the chemistry of the different tissues and a daily monitoring of the insect feeding sites will further improve the understanding of the specific modulation of plant resistance to above-ground and below-ground herbivory by fungal volatiles.

| Acceleration of bolting time by fungal volatiles suggests enhancement of plant fitness
Effects of fungal volatiles on plant phenology differed between the fungi but may increase reproductive success. Plant exposure to volatiles emitted by R. solani or P. leveillei accelerated overall plant bolting and enhanced production of buds and flowers. An acceleration of bolting could be disadvantageous for the plant as P. brassicae caterpillars prefer to feed on inflorescences (Smallegange et al., 2007), but it can also result from an escape strategy to reproduce faster (Lucas-Barbosa et al., 2013). By producing more buds and flowers and quicker, plants increase their chance of reproductive success, which would give a clear advantage for plants surrounded by potential pathogens and challenged with an insect herbivore. Yet, acceleration of flowering seems a common plant phenomenon in response to volatiles emitted by fungi of different lifestyles, including beneficial fungi (Cordovez et al., 2017;Moisan et al., 2019;Sánchez-López et al., 2016). Thus, we hypothesise that plant responses are specific to some fungal volatiles, which consequently affect plant defences to herbivores. To further assess the effects on plant fitness, it remains to be tested whether seed set is ultimately influenced by root exposure to fungal volatiles.

| CON CLUS IONS
Our results show that fungal volatiles can modulate plant mixeddefence strategies, balancing plant resistance and tolerance to above-and below-ground herbivory in a fungus-specific manner.
Yet, it remains to be investigated how these results obtained in controlled conditions with single fungal isolates can be extrapolated to natural ecosystems where plant roots are exposed to volatiles emitted simultaneously or in sequence by diverse fungal and other microbial communities. In such future studies, one should also address how these responses affect subsequent plant interactions with mutualists, for instance pollinators that are essential for reproduction of obligate outcrossing plant species such as B. rapa.