Key message

  • Drought stress alters multitrophic interactions between sugar beet, aphids and their parasitoids

  • Aphids benefit from drought stressed plants through bottom-up and top-down effects

  • Drought enhances host plant quality for aphids (bottom-up)

  • Drought reduces emission of herbivore-induced plant volatile that attract parasitoids (top-down)

  • Drought reduces parasitoid reproduction in aphid hosts (top-down)

Introduction

Climate change models predict altered rainfall patterns and an increased number of extreme drought events, as experienced recently in Central Europe (IPCC 2014; Boergens et al. 2020). Prolonged water deficit can have a broad range of impacts on arable cropping systems and their arthropod communities by altering the interactions among species and thus changing multitrophic networks (Jamieson et al. 2012; Walter 2018). Drought negatively affects plant growth and morphology (Zhang et al. 2018; Grzesiak et al. 2019; Yang et al. 2021), and the prime symptoms are decreased plant height (Anjum et al. 2017; Patmi et al. 2020; Misra et al. 2020; Li et al. 2020), shoot and root biomass (Benjamin and Nielsen 2006), number and area of leaves (Khaleghi et al.; 2019; Mishra et al. 2018), as well as wilting and rolling of leaves (Willick et al. 2018).

Drought also alters phytohormonal signaling (Gupta et al. 2020; Jogawat et al. 2021; Mubarik et al. 2021) and the biochemical composition of plant tissues (Bettaieb Rebey et al. 2012) which may improve (Khan et al. 2010) or decrease herbivore performance (Xie et al. 2020). The effects of drought on herbivore performance can depend on the magnitude of the experienced stress (Mody et al. 2009; Tariq et al. 2012). Aphids, such as Rhopalosiphum padi, respond differently to moderate and severe levels of drought which can alter aphid-parasitoid interactions (Kansman et al. 2021). Drought stress further has bottom-up effects on the third trophic level by altering the behavior and performance of parasitoids (Tariq et al. 2013; Shehzad et al. 2020).

However, reported bottom-up effects of drought stress on aphid performance are inconsistent and may depend on the level of drought stress as well as the plant and aphid species studied (Mewis et al. 2012; Shehzad et al. 2020; Xie et al. 2020; Cui et al. 2021; Luo and Gilbert 2022). Studies on aphid parasitoids, representing the third trophic level, are scarce and mainly focus on parasitization rates and mummification success (e.g. Ahmed et al. 2017; Kansman et al. 2021). Parasitoid wasps specialized on aphids develop within the body of their hosts. Thus, their performance is significantly influenced by changes in the physiology or behavior of aphids (Brodeur and Boivin 2004; Kaplan et al. 2016). Several studies showed indirect effects of plant drought stress on parasitoid performance, but the responses range from positive (Romo and Tylianakis 2013) to negative (Johnson et al. 2011; Ahmed et al. 2017; Nguyen et al. 2018).

Another essential aspect is the ability of parasitoids to find host-infested plants under drought-stress conditions. Parasitoids use herbivore-induced plant volatiles (HIPVs) to detect their hosts (e.g. Turlings and Erb 2018); because HIPVs provide information to parasitoids that help suppressing the herbivore attack, HIPVs can be considered as a plant's information-mediated indirect defense (Kessler and Heil 2011). Some studies have assessed the effects of drought on HIPV emission, parasitoid attraction and the underlying mechanisms (Gouinguené and Turlings 2002; Weldegergis et al. 2015; Martini and Stelinski 2017; Salerno et al. 2017; Lin et al. 2022b) but only one other has focused on aphids (Tariq et al. 2013). Hence, more research is needed before a comprehensive understanding can be obtained on how drought stress modulates the plant’s indirect defense.

Sugar beet (Beta vulgaris spp. vulgaris) is an important crop and a significant source of sugar in temperate regions, contributing to a third of the world's annual sugar production (Dohm et al. 2014). Since neonicotinoid seed treatment in sugar beet has been phased out in many European countries in 2018, increasing pest pressure has been observed (Viric Gasparic et al. 2020). Naturally occurring parasitoids and their conservation are therefore likely to play a more prominent role as biocontrol agents in the future by reducing aphid populations in sugar beet fields. How the tritrophic system, consisting of sugar beet, the important herbivore A. fabae, and its parasitoid A. colemani, is affected by water limitation is entirely unknown. In this comprehensive study, we specifically addressed the effects of different levels of drought on (i) sugar beet plant growth and morphology (ii) aphid performance on drought-stressed sugar beets (iii), parasitoid performance on aphids reared on drought-stressed plants, (iv) volatile emission from aphid-infested and uninfested sugar beets and (v) parasitoid attraction towards aphid-infested and uninfested plants.

Materials and methods

Plant and insects

Seeds of B. vulgaris subspec. vulgaris cultivar 'Vasco' (SESVanderHave, Belgium) were sown in plastic trays (54 holes, each 3.5 cm diameter) filled with quartz sand (0.2–0.8 mm). Trays were placed in a controlled climate room with high-pressure sodium vapour lamps (16L:8D photoperiod, light intensity: 130 ± 10 µmol/(s m2), relative humidity: 65 ± 5%, temperature: 20 ± 1 °C) and were supplied with tap water. Twelve days after sowing (DAS), seedlings were supplied with half-strength modified Hoagland solution (HS) (Hoagland and Arnon 1938) for up to 23 DAS to provide nutrients. At 24 DAS, seedlings were transferred to the drought system with full strength HS. For preparing HS, Na2MoO4.2H2O was used instead of H2MoO4.H2O, and C10H12FeN2NaO8.3H2O replaced C12H12Fe2O18 with concentrations of 0.12 mg/l and 22.5 mg/l, respectively.

Colonies of Aphis fabae (Order: Hemiptera, Family: Aphididae) were reared on the same genotype of sugar beet as used in the experiments and grown in a separate insect rearing room. A. colemani was reared on A. fabae, and sugar syrup (10% sucrose solution) was supplied as food source for adult parasitoids.

Drought system and experimental setup

As described by Marchin et al. 2020, a modified capillary action-based drought system was used for all drought stress experiments. PVC-U (polyvinyl chloride without plasticizers) cylinders (diameter: 6.5 cm; height: 12.5 cm) with their bottom part fitted with fine nylon mesh with four fine holes were filled with quartz sand (0.2–0.8 mm) and used as a plant holder. Porous floral foam (length: 23 cm, width: 11 cm, height: 8 cm) (BIG-mosy, Mosy GmbH, Germany) was cut into three equal pieces and placed in black plastic box boxes (length: 46 cm, width: 30 cm, height: 16 cm) (Iris Ohyama, Germany). Six plants were placed in each box on top of the floral foam, and HS was added to the box with an adjusted water level (Figure S1). By maintaining the depth of the water level, three treatments were imposed as- i) control: 40% volumetric water content (VWC), ii) moderate drought: 16% VWC, and iii) high drought: 10% VWC. Prior to the experiment, the maximum water holding capacity of the specified volume of quartz sand (40% VWC) and the permanent wilting point (below 9%) of sugar beet plants were determined in a preliminary experiment. Each morning at 10 a.m., the VWC was checked with a digital soil moisture meter (ThetaProbe type ML3, Delta-T, Cambridge, UK), and water levels were adjusted by applying HS as needed. From 24 to 26 DAS, all plants were kept at 40% VWC with full strength HS for acclimation. Subsequently, plants were exposed to the drought treatments described above.

To answer the first three questions regarding plant, aphid, and parasitoid performance, three separate experiments were conducted in the same manner, where three drought boxes (each box contained 6 plants and was considered a block) were used for each treatment (control, moderate drought, high drought). This resulted in a total of n = 18 plants per treatment.

A fourth experiment was conducted to assess parasitoid olfactory behavior to plant volatiles; in this case, six boxes were used to prepare the six different odor sources: (i) control, (ii) control + aphid, (iii) moderate drought, (iv) moderate drought + aphid, (v) high drought, and (vi) high drought + aphid. A total of n = 6 plants per treatment were tested.

Monitoring plant performance

To test the effect of drought stress on sugar beet, plant height was measured every seven days, starting at the two true leaf stage (26 DAS), and prior to drought application, until 54 DAS. Plants were harvested at 73 DAS, and total biomass, root and shoot weight, root and shoot length, and the number of leaves were recorded.

Aphid performance

To explore the influence of drought stress on the reproductive performance of A. fabae, three wingless adults were placed on the second leaf of each sugar beet plant at 30 DAS (day 4 of drought treatment). A perforated polypropylene bag (15 × 25 cm) (Nette GmbH, Germany) was carefully placed over each aphid-infested leaf. After 24 h, all aphids except two neonate nymphs were removed from the leaf. These nymphs (F0 generation) were left undisturbed for 8–9 days. Before reproduction began, one aphid was selected as the F0 mother and the other was removed. The F0 mother was allowed to produce its first progeny (F1) before being transferred to the third leaf of the same plant. Two first F1 offspring were kept undisturbed on leaf two until either one started to reproduce. In the meantime, the number of offspring of the F0 mother (now on leaf three) was recorded and nymphs were removed daily until the death of the F0 mother. The number of offspring until their own first offspring (F1) started to reproduce (Nd), the total number of offspring, the duration of the reproductive period, and the average number of offspring per day were calculated for the F0 mother. After the F1 aphids had started to reproduce, one was randomly selected as the F1 mother, while the other was removed (see also Figure S2). The pre-reproductive period and longevity of the F1 mother were recorded. The following equation was used to calculate the intrinsic rate of increase (Rm): Rm = 0.738 (loge Nd)/d (d = reproductive periods; Nd = number of offspring of the F0 mother until her own first offspring reproduces) (Wyatt and White 1977).

Parasitoid performance

Five wingless adult aphids were placed on the second leaf of a sugar beet plant (30 DAS), and the entire plant was covered with a perforated polypropylene bag (15 × 25 cm, Nette GmbH, Germany). 18 plants were prepared per treatment. After 24 h, the adults and all offspring except five neonate nymphs were removed. The nymphs were left undisturbed for nine days (pre-reproductive stage). At 40 DAS, one naïve female A. colemani was released into each polypropylene bag for five hours to parasitize the aphids. After removal of the parasitoids, aphids were checked daily to record the following parameters: proportion of (1) mummified, surviving, and dead aphids, (2) time from parasitization to mummification, (3) time from mummification to adult emergence, (4) total parasitoid development time, (5) emergence rate, (6) sex ratio and (7) hind tibia length of the emerged parasitoids. A. colemani adults were preserved in 96% (v/v) ethanol and hind tibia length of both legs was measured using a stereomicroscope (Leica WILD M3Z) equipped with an ocular micrometer.

Parasitoid host finding behavior and plant volatile analysis

A six-arm olfactometer (Turlings et al. 2004) was used to trap plant volatiles and to simultaneously measure the attraction of female A. colemani to six odor sources. Aphid-infested and non-infested plants subjected to the three drought treatments (control, moderate drought, high drought) were used as the six odor sources. All plants were tested at the same age (40 DAS) and had been exposed to 13 days of drought stress (start at 27 DAS). In the case of aphid treatments, 40 A. fabae (mixture of all ages) were released on each plant at 30 DAS and were allowed to feed and reproduce for 10 days. A single plant from each treatment was randomly assigned to each arm of the olfactometer. Instead of placing each sugar beet plant into an odor source glass vessel of the olfactometer, the plant was bagged using an inert oven bag (Bratschlauch, Toppits, Germany).

Collections of volatiles were carried out for 24 h (from 09:00 to 09:00 next day), and within this time, parasitoid behavioral assays were conducted (from 10:00 to 16:00). Ten fluorescent lamps (PAR inside odor source oven bags: 130 µmol/(s m−2) at 3 cm distance from lamps) were illuminating the plants during VOC collection for 16 h during the day and were switched off for 8 h during night conditions. Volatile collection traps (7 cm glass tube) containing 30 mg of 80–100 mesh Porapak Q (Volatile Collection Trap LLC, FL, USA) were kept in place by two fine mesh metal screens and attached to each oven bag. Two activated charcoal filters (400 ccs, Alltech, Deerfield, IL, USA) were used to filter the air originating from a central in-house compressor. Then filtered and humidified (relative humidity around 56%) by passing through deionized water in a glass bottle (1L), the air went through an airflow meter before it was pushed into each oven bag at a rate of 1.0 l/min. With a vacuum pump (N816.3KN.18, Laboport®, Germany), 0.5 l/min of air was pulled through the trapping filter.

Every day, a total of 30 naïve females of A. colemani (2–5 days old) were tested in five experimental rounds. In each experimental round, six parasitoids were released simultaneously. After 60 min, the choices made by the parasitoids were recorded, and all parasitoids were removed before the next group was released. After every experimental day, the glass and Teflon parts of the olfactometer were first cleaned with demineralized water and then rinsed with 99.5% acetone. After evaporating the solvents, all cleaned glassware was placed in an oven for two hours at 180 °C. This procedure was repeated on six experimental days, with a fresh set of plants used each day. Each day after the volatile collection, the above-ground biomass was measured with a balance (KERN PEJ 4200-2 M, Kern & Sohn GmbH, Germany), leaf areas were measured by an automatic leaf area meter (LI-3100C, LI-COR® Biosciences GmbH, Germany) to calculate volatile emission as ng/h/g Fresh Weight (FW) and ng/cm2 leaf area.

After headspace collection, the trapped volatiles were eluted with 150 µl dichloromethane (DCM) into a 1 ml glass vial and stored at -80 °C for further analysis. Before the analysis, 200 ng of tetralin (1,2,3,4 tetrahydronaphthalene, Sigma-Aldrich, Taufkirchen, Germany) was added to each sample as an internal standard. To calculate the retention index, n-alkanes (C8-C20) (Sigma Aldrich) with a concentration of ~ 40 mg/L each, in hexane were purchased and run on the GC–MS. Each sample's qualitative and quantitative VOC composition was analyzed by gas chromatography—mass spectrometry (5977B HES MSD, Agilent Technologies). A 2 µl sample was injected in pulsed splitless mode with an automated injection system. The oven temperature was held at 40 °C for 3 min and then increased gradually to a final temperature of 220 °C, which was held for 10 min. Helium (1.5 ml/min) was used as the carrier gas. The software MSD ChemStation, coupled with the NIST17 and Wiley11 mass spectral libraries, was utilized to tentatively identify compounds based on their mass spectra, ensuring a reverse match accuracy of over 90%. Experimental retention index (RI) was calculated by the following formula:

$$\begin{aligned} {\text{RI}} & = \left( {\left( {\frac{{Retention\,Time\,\left( {RT} \right)\,of\,comp - RT\,of\,previous\,alkane}}{{RT\,of\,later\,alkane - RT\,of\,previous\,alkane}}} \right)} \right. \\ & \quad + No.\,of\,carbon\,of\,previous\,alkane\bigg)*100 \\ \end{aligned}$$

Following the initial identification process, if the experimental retention index (RI) of the suggested molecules aligns with the RI values listed in the Van Den Dool and Kratz RI Table on the website https://webbook.nist.gov/chemistry/name-ser/, with a similarity threshold of ± 5 RI units, then the compound was considered identified. Subsequently, compound quantification was conducted by comparing the peak areas to the peak area of the internal standard.

Statistical analysis

All data was analyzed by using the statistics package R-version 4.1.2 integrated with R-Studio Desktop-version 2021.09.1 + 372. Prior to the analysis of each dataset, different models were tested with different family distributions, depending on the type of dataset. Models were simplified when necessary, model assumptions were checked, performance of mixed-effects models was compared based on AIC values fitted to the same data set, and finally the model was simulated and compared using the Performance (Lüdecke et al. 2021) and DHARMa (Florian Hartig 2021) packages to find the best model for each data set.

Mixed effects models were used to account for the fact that three or six plants were placed together in one plastic box to manipulate water availability. Each plastic box was considered as one block and the effect of a block was included as a random factor in the mixed effects models. The effect of drought on plant height at different times (days) was analyzed by a generalized linear mixed effects model (GLMM) using the glmmTMB function (Brooks et al. 2017) with a log link to Gaussian to account for repeated measurements. Drought stress and time were used as fixed factors, but the interaction between drought stress and time was not included. Plant identity, nested in blocks, was used as a random factor in this analysis. Linear mixed effects models (LMMs) were used to analyze total biomass, root weight, shoot weight, and shoot length. Root-shoot ratio by weight was analyzed using GLMM with log link to Gaussian. Root length, root-shoot ratio by length, reproductive rate, intrinsic growth rate of aphids were analyzed by GLMM with Gaussian family.

Male and female hind tibia length of parasitoids were analyzed by linear mixed effect models (LMM) with block as a random factor. Pre-reproductive period, reproductive period, longevity, number of offspring until own first offspring reproduce, cumulative aphid count data from the last day of the experiment, mummification to oviposition time of parasitoid, oviposition to emergence time of parasitoid, total development time of parasitoid, and number of leaves were analyzed by GLMM with Conway-Maxwell Poisson distribution (family = compois) (Huang 2017), considering block as a random factor.

Parasitoid responses to the six different odor sources were analyzed using a general linear model (GLM) with binomial family distribution, considering number of successes (= all parasitoids that went to each odor source) and number of failures (= all parasitoids that went to the other five odor sources) as two-vector response variables. To account for the fact that the choices of parasitoids that were release as one group are not independent from each other and that multiple release rounds were done on the same day with the same set of plants, we first tried to glmm with release round and experimental day as random factors, but it turned out that the random factors had no influence and we thus continued with a glm.

Mummification rate was analyzed by GLMM with beta binomial family distribution. Proportion of surviving and dead aphids that were exposed to parasitoids, parasitoid emergence rate, and the proportion of male and female parasitoids did not meet assumptions for GLMM, thus a non-parametric Kruskal–Wallis test was used, followed by Dunn's test with Holm's method for post-hoc analysis.

Total volatile emissions were analyzed by two-way ANOVA with drought stress and aphid presence as explanatory variables, followed by Tukey HSD post-hoc test for multiple comparisons. Hierarchical clustering was calculated for VOCs based on the length of the straight line drawn, followed by Euclidean distance with complete linkage. The Pearson correlation coefficient was calculated to determine the relationship between total VOC emission and parasitoid response. The Pearson correlation coefficient was computed by first standardizing both the VOCs and parasitoid response variables by z-score normalization to ensure comparability. The significance of the correlation coefficient was assessed using Student’s t-test, with a p-value threshold of 0.05 indicating statistical significance. One-way ANOVA was performed on the emission rates of individual VOC compounds, but for those individual VOCs that did not meet the model assumptions, we used the Kruskal–Wallis test followed by the Dunn test with the Holm method for post hoc analyses.

Results

Effect of drought stress on the performance of sugar beet plant

Drought stress had a negative effect on sugar beet height over time (GLMM: Drought: χ2 = 57.76, p < 0.001; Time: χ2 = 510.26, p < 0.001; Figure S3; Table S1). Before applying drought at 26 DAS, plant height was not significantly different between the treatments (Figure S3). After one week of drought treatments (33 DAS), the height of plants exposed to medium or severe drought was strongly reduced (p < 0.001) compared to control plants. Plants exposed to severe drought were the smallest and height differences remained constant throughout sampling times. After four weeks of drought (54 DAS), moderately and highly stressed sugar beet plants were 15.2% and 30.9% shorter than controls.

Similarly, moderate and high drought significantly reduced total plant biomass (g FW) (LMM: F2,15 = 224.16, p < 0.001), root weight (g FW) (LMM: F2,15 = 16.574, p < 0.001), and shoot weight (g FW) (LMM: F2,15 = 139.71, p < 0.001) (Table S2) compared to control plants. Plants exposed to severe drought showed significantly reduced weights compared to plants from the moderate drought treatment. Moderately and highly stressed plants had 29.1% and 58.4% reduction in total plant biomass; 22.6% and 36.2 reductions in root weight; and 31.7% and 67.52% reduction in shoot weight, respectively. Shoot length (cm) was also higher in control plants compared to stressed plants (LMM: F2,15 = 70.287, p < 0.001); however, root length (cm) was lowest in control plants (GLMM: χ2 (df 2) = 40.41, p < 0.001) (Table S2). Root to shoot ratio by weight and by length was lowest in control plants and highest in highly stressed plants (p < 0.001) (Table S2). Number of leaves on plants grown in moderate and high drought conditions were significantly reduced compared to control plants (χ2 (df 2) = 39.54, p < 0.001) (Table S2).

Effect of drought stress on the performance of Aphis fabae

There were significant differences in the pre-reproductive period (GLMM: χ2 (df 2) = 18.69, p < 0.001) and reproductive period (GLMM: χ2 (df 2) = 12.47, p = 0.0019) of A. fabae feeding on drought-affected plants, (Fig. 1A). Aphids developing on highly drought-stressed plants matured faster by shortening their pre-reproductive period (control vs high: p < 0.001; control vs moderate: p = 0.975; moderate vs high: p < 0.001) and extended their reproductive phase (control vs high: p < 0.001; control vs moderate: p = 0.14; moderate vs high: p = 0.061) without changing their total life span (GLMM: χ2 (df 2) = 0.079, p = 0.961).

Fig. 1
figure 1

Effect of drought stress (~ 40% volumetric water content (VWC), ~ 16% VWC, ~ 10% VWC) on the performance of A. fabae. Bar graphs represent the average duration of pre-reproductive and reproductive period as well as longevity of A. fabae (A) and line graphs represent the cumulative offspring number of A. fabae (B). Data points in bar graphs represent individual replicates. Asterisks (* p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001) and different letters (p ≤ 0.05) indicate significance among treatments, n = 18

Full size image

The fecundity of A. fabae differed significantly depended on drought treatments (GLMM: χ2 (df 2) = 26.02, p < 0.001). Aphids on moderately and highly drought-stressed plants produced more offspring than aphids on control plants (Fig. 1B). Similar patterns were observed in the number of offspring that were produced by the F1 generation until their own first progeny reproduced (Nd) (GLMM: χ2 (df 2) = 23.45, p < 0.001), the average number of offspring produced per day (GLMM: χ2 (df 2) = 7.20, p = 0.027) and the intrinsic rate of increase (Rm) (GLMM: χ2 (df 2) = 30.58, p < 0.001) (Figure S4).

Effect of drought stress on the performance of Aphidius colemani

Significantly more aphid mummies formed on highly drought-stressed plants than on control plants (GLMM: χ2 (df 2) = 10.40, p = 0.005, control vs moderate: p = 0.388, control vs high: p = 0.0017, moderate to high: p = 0.056) and significantly less aphids survived (Kruskal–Wallis: χ2 (df 2) = 7.03, p = 0.029), even though the percentage of aphids that died a few days after parasitization was the same in all treatments (Kruskal–Wallis: χ2 (df 2) = 2.46, p = 0.29) (Fig. 2A). The highest mummification rate (57.8%) was found in drought-stressed plants, while control plants showed the lowest mummification rate (42.2%) (Fig. 2A).

Fig. 2
figure 2

Effect of drought stress (~ 40% volumetric water content (VWC), ~ 16% VWC, ~ 10% VWC) on the performance of the parasitoid A. colemani. Stacked bar graphs represent the percentage of aphids that formed mummies, died or survived (A); time from oviposition to mummification and mummification to emergence (B); and percentage of male and female parasitoids (C). Bars show the hind tibia lengths of the emerged male and female A. colemani (D). Asterisks and different letters indicate significant differences among treatments (p ≤ 0.05). * p ≤ 0.05, ** p ≤ 0.01, *** p ≤ 0.001, different letters indicate significant differences between treatments, small letters refer to the respective part of the bar plot, capital letters refer to the total bar plot, e.g. total development time in (B), n = 18 plants with 5 aphids each, resulting in 90 aphids that were evaluated per treatment, N = exact number of aphids or mummies per category, depending on survival and gender of the emerging parasitoid

Full size image

Total developmental time (oviposition to adult emergence) of A. colemani was shorter on drought-stressed than on control plants (GLMM: χ2 (df 2) = 25.57, p < 0.001; control vs high: p < 0.001; control vs moderate: p = 0.0007; moderate vs high: p = 0.39) (Fig. 2B). Specifically, within the developmental period, drought boosted the oviposition to mummification time (GLMM: χ2 (df 2) = 60.51, p < 0.001) but not the time from mummification to adult emergence (GLMM: χ2 (df 2) = 4.25, p = 0.119) (Fig. 2B).

Drought had negative effects on the adult emergence rate of A. colemani. Adults emerged successfully from 84.3% of the mummies on control plants, from 53.7% of mummies on moderately drought-stressed plants and from 40.3% of mummies on highly drought stressed plants (Kruskal–Wallis: χ2 (df 2) = 22.82, p < 0.001, control vs moderate: p = 0.005, control vs high: p < 0.001, moderate to high: p = 0.091) (Fig. 2C). Sex ratio was also altered by drought stress (Kruskal–Wallis: male: χ2 (df 2) = 6.77, p = 0.033, and female: χ2 (df 2) = 10.31, p = 0.005). The proportion of mummies from which males emerged increased under drought stress (Kruskal–Wallis: χ2 (df 2) = 6.77, p = 0.033), while the proportion of mummies from which females emerged was significantly lower in moderately and highly drought-stressed plants compared to controls (Kruskal–Wallis: χ2 (df 2) = 10.31, p = 0.005, Fig. 2C).

Hind tibia length of male (LM: F2,36 = 19.91, p < 0.001) and female (LM: F2,31 = 8.66, p < 0.001) A. colemani showed significant differences depending on the intensity of drought stress. Parasitoids that emerged from aphids on drought-stressed plants had shorter hind tibia lengths compared to aphids on control plants. This was the case for both male (control vs high: p < 0.001; control vs moderate: p = 0.27; moderate vs high: p < 0.001) and female parasitoids (control vs high: p = 0.013; control vs moderate: p = 0.004; moderate vs high: p = 0.87) (Fig. 2D).

Analysis of VOCs and six-arm olfactometer bioassay

Strong differences in volatile emission were observed between the different treatments. While well-watered plants without aphids emitted 21 VOCs, aphid-induced well-watered plants emitted 29 different VOCs. Seven of these compounds were not detected in any other treatment (Fig. 3). Aphid-induced highly drought-stressed plants emitted 14 VOCs, and highly drought-stressed plants without aphids emitted only six different compounds. However, non-infested, moderately drought-stressed plants released more (13 VOCs) compounds than aphid-induced moderately drought-stressed plants (8 VOCs) (Fig. 3). Hierarchical clustering of the emitted plant volatile blends revealed that the blend from non-infested, well-watered controls was distinct from all other treatments (Fig. 3). Among the different VOCs, emission of 3-hexen-1-ol; p-menthane, 6-methyl-5-hepten-2-one, and 3-carene were significantly different among treatments when emission rates were analyzed per plant (Fig. 3, Table S4). When VOCs emission rates were standardized by leaf biomass (ng/h/g FW), 6-methyl-5-hepten-2-one, p-menthane, o-cymene and 3-hexen-1-ol were significantly different among different treatments (Figure S5, Table S5). After standardized by leaf area (ng/cm2), emissions rates of 6-methyl-5-hepten-2-one, p-menthane, p-cymen-7-ol, o-cymene and 3-hexen-1-ol were significantly different (Figure S5, Table S6). For all emitted compound, compound classes, retention time, and indices from experiments and literature are summarized in Table S3.

Fig. 3
figure 3

Effect of aphid infestation and drought stress on volatile emission. The hierarchical clustering indicates how similar the blends of the different treatments are, taking into account all listed compounds. Heat map represents emission rates of specific volatile organic compounds (VOCs) and the boxplot represents the total amount of VOCs emitted from differently treated sugar beet plants. Compound names and P values in bold indicate significant differences among treatments. Different letters show statistically significant differences in VOC emission between treatments (p ≤ 0.05, n = 6). N = total number of different compounds detected in the respective treatment

Full size image

Overall, 80% of the released Aphidius colemani females were attracted by sugar beet VOCs and were found in one of the arms of the six-arm olfactometer at the end of the trial. Drought stress and aphid presence significantly influenced attraction of parasitoids to VOCs emitted by the sugar beet plant (binomial GLM:χ2 (df 5) = 220.68, p < 0.001) (Fig. 4A). Aphid-infested well-watered plants attracted the highest number of parasitoids (46.1%), while non-infested well-watered controls attracted the second highest number of parasitoids (18.9%), but significantly less than the aphid-infested plants. Plants subjected to moderate drought stress in the absence of aphids attracted similar numbers of parasitoids as the non-infested well-watered plants. However, aphid-infested plants exposed to moderate drought stress attracted significantly fewer parasitoids (0.6%) than the non-infested plants exposed to moderate drought stress (10.6%). Only few parasitoids were attracted to highly drought-stressed plants with and without aphids (1.1% and 3.3%, respectively). These numbers did not differ significantly from the number of parasitoids attracted to the aphid-infested plants exposed to moderate drought stress.

Fig. 4
figure 4

Effect of aphid infestation and watering regime on volatile preference of female A. colemani. Bar graph depicting the proportion of active females that chose the respective odour source. Ring represents the total percentage of female A. colemani, which made a choice among the treatments. Different letter indicate significant differences among treatments (p < 0.05). In total 6 different groups of plants were tested on 6 days. Each day, 5 groups of 6 parasitoids each were released in the six-arm olfactometer to choose among the six different odour sources (A) and correlation between parasitoid preferences and total VOCs emission per plant (B)

Full size image

A positive correlation (R = 0.9, p < 0.001) was found between total VOC emission per plant and the number of attracted parasitoids (Fig. 4B).

Drought and aphid infestation had a significant effect on total VOC emission from the whole plant (LM: Drought: F2,30 = 136.0, p < 0.001; Aphid: F1,30 = 3.39, p = 0.075; Drought × Aphid: F2,30 = 7.86, p = 0.0017) (Fig. 3), per gram fresh weight (LM: Drought: F2,30 = 36.45, p < 0.001; Aphid: F1,30 = 6.97, p = 0.013; Drought × Aphid: F2,30 = 11.10, df = 2, p < 0.001) and also when standardized by leaf area (cm2) (LM Drought: F2,30 = 35.46, df = 2, p < 0.001; Aphid: F1,30 = 2.42, p = 0.13; Drought × Aphid: F2,30 = 8.26, p = 0.0013) (Figure S5, Table S5, Table S6). Aphid-induced control plants emitted the highest amounts of VOCs per plant, followed by control plants without aphids. The emission was significantly lower in moderately and highly drought-stressed plants regardless of aphid infestation (Fig. 3). After standardization of VOC emission per gram leaf biomass or leaf area (cm2), this pattern changed due to drought-stressed plants having significantly lower biomass and smaller leaf area than well-watered plants (Table S7). When corrected for plant biomass or leaf area, no differences were found in emission rates from highly drought-stressed plants with aphids and well-watered plants (with and without aphids) (Figure S5).

Discussion

Our study demonstates that drought stress differentially affects the three trophic levels associated with sugar beet. While drought stress had direct negative effects on sugar beet plants by reducing their size and biomass, it benefited A. fabae, as these aphids developed faster and produced more offspring on drought-stressed plants. Even though more parasitoid mummies were formed on drought-stressed plants, fewer adult parasitoids emerged from the mummies on drought-stressed plants. In addition, HIPV emission was drastically reduced by drought stress and parasitoids were less attracted to drought-stressed plants, with and without aphids, when compared to well-watered plants. This suggests that biological pest control by parasitoids might be severely impaired by drought stress in this system.

Drought stress severely alters plant growth and morphology and leads to many physiological and biochemical responses. Water limitation triggers a phytohormonal signaling cascade involving abscisic acid (ABA) and induces stomatal closure, resulting in reduced gas exchange and ultimately in reduced photosynthesis (Farooq et al. 2009; Ding et al. 2018). Drought also interferes with mitosis and the loss of turgor inhibits cell elongation (Farooq et al. 2009; Fahad et al. 2017). In addition, it disturbs the water balance, membrane permeability, mineral nutrition, and enzyme activities in the plant (Dubey et al. 2021). Taken together, the effects of drought stress on plant physiology typically result in reduced growth and lower biomass production, as was observed in this study and in other crops, such as maize (Anjum et al. 2017), sugarcane (Misra et al. 2020), wheat and rice (Zhang et al. 2018; Patmi et al. 2020). The increase in root:shoot ratio reported here is also a typical plant response that enhances water uptake to mitigate drought effects (Fang and Xiong 2015; Kurepa and Smalle 2022). In our study, drought stress symptoms were already visible at moderate drought levels and were even more pronounced in the high drought treatment. Plant stress caused by water limitation can be further exacerbated if it makes plants more susceptible to herbivores, either directly or indirectly by reducing their ability to attract natural enemies for defense.

In this study we found that the performance and fecundity of individual A. fabae were highest on plants receiving high drought treatment, intermediate on plants experiencing moderate drought and lowest on well-watered plants. Effects of drought stress on aphid performance are variable (Leybourne et al. 2021), indicating that there is no general response of aphids towards drought-stressed host plants. The effects of drought stress on aphids can further depend on the timing and magnitude of the water limitation experienced (Tariq et al. 2013; Luo and Gilbert 2022), on specific interactions between aphid species and host plant species (Mewis et al. 2012; Leybourne et al. 2021) and on the host-specificity of the aphid (Nguyen et al. 2023). Reasons for enhanced aphid performance on drought-stressed plants could be an increase in the nutritional quality of the drought-stressed plant (Smith et al. 2019), which includes sugar and or/ amino acid concentrations in the phloem sap (Hale et al. 2003; Fàbregas and Fernie 2019). Phytohormonal crosstalk between different stress-related phytohormones such as jasmonic acid, salicylic acid and abscisic acid, further affects inducible defenses against aphids (Guo et al. 2016; Kansman et al. 2022) and may benefit the aphids on drought-stressed plants.

The emission of VOCs can vary both in quantity and quality, depending on the biotic and abiotic stress factors involved, and these changes can influence the attraction of natural enemies to herbivore-infested plants (Dicke and Baldwin 2010; Kugimiya et al. 2010). Drought has been shown to affect the foraging success of a parasitoid by altering plant volatile emissions. This resulted in less attractive (Tariq et al. 2013) or even unrecognizable cues (Martini & Stelinski 2017), but positive (Salerno et al. 2017) or neutral effects of drought stress on parasitoid attraction have also been reported (Weldegergis et al. 2015). In our study, aphid infestation on well-watered plants resulted in a strong increase in total VOC emission and in the release of eight VOCs that were not detected in the well-watered controls without aphids. Not all plant VOCs are perceived by parasitoids (Goelen et al. 2021; Li et al. 2022), but in our study we found a strong correlation between the total amount of emitted VOCs and attraction of A. colemani. Well-watered aphid-infested plants emitted the highest amount of volatiles and attracted most parasitoids. Well-watered plants without aphids ranked second for to total volatile emission, number of emitted compounds and parasitoid attraction. In the drought-stressed plants, the number of compounds detected and the total emission of volatile compounds per plant were greatly reduced, resulting in reduced attraction of parasitoids. Interestingly, volatiles from drought-stressed plants with aphids did not attracted more parasitoids than drought-stressed plants without aphids, suggesting that parasitoids may face difficulties in finding aphids on drought-stressed sugar beet.

All studies on HIPV emission in relation to drought stress report changes in the emission of some VOC compounds, but the effects of altered HIPV bouquets on parasitoid attraction are variable and the mechanisms leading to altered behavior are not well understood (Gouninguene & Turlings 2002, Tariq et al. 2013; Weldegergis et al. 2015; Salerno et al. 2017; Catola et al. 2018; Pagadala Damodaram et al. 2021; Lin et al. 2022b; Sun et al. 2023). Similar to our results, A. colemani and Diaeretiella rapae preferred aphid-infested well-watered Brussels sprouts to aphid-infested drought-stressed plants (Tariq et al. 2013). In this study, emission rates of allyl isothiocyanate, limonene and β-phellandrene from drought-stressed aphid-infested plants were more similar to the undamaged well-watered controls than to aphid-infested well-watered plants, suggesting that reduced differences in emission rates of behaviorally active compounds from infested and undamaged plants under drought stress may reduce parasitoid attraction. Similarly, drought-stress alone may increase the attractiveness of non-infested plants for parasitoids and predators if drought stress induces the emission of similar VOC blends as the presence of hosts (Takabayashi et al. 1994; Salerno et al. 2017; Catola et al. 2018). The physiological mechanisms that lead to changes in VOC emission upon drought stress are also not fully understood and may depend on the exact drought stress treatment (e.g. Salerno et al. 2017). The reduced VOC emission of drought-stressed plants in our study can partially be explained by the reduced biomass of drought-stressed plants. After standardizing VOC emission by plant biomass or leaf area, severely drought-stressed plants with aphids emitted similar amounts of VOCs as well-watered plants with aphids, indicating that VOC emission was not affected by severe drought stress. However, moderately drought-stressed plants with aphids emitted significantly less VOCs than the well-watered plants with aphids, and reduced emission during drought stress was found in uninfested plants (Figure S5). Contrary to our results, Lin et al. (2022b) found no effect of drought stress on total VOC emission per tomato plant after 1 day of aphid feeding, but a strong increase when emission rates were standardized by biomass.

VOCs are mainly released through the stomata and the reduced emission of VOCs from drought-stressed plants may be partly the result of low stomatal conductance (Niinemets et al. 2004; Harley 2013; Seidl-Adams et al. 2015; Lin et al. 2022a). Stomatal closure should have strong effects on the emission rates of compounds with low Henry’s law volatility constant (Hpc, for definition see Sander 2015) such as alcohols, carbonyls, aldehydes and oxygenated monoterpenes. Emission rates of compounds with a high Hpc such as non-oxygenated monoterpenes should not be affected by stomatal conductance (Lin et al. 2022a). Most non-oxygenated monoterpenes emitted by sugar beet (e.g. β-pinene, p-menthane, β-myrcene, 3-carene, D-limonene, o-cymene etc.) were released in higher quantities from aphid-infested plants than from undamaged plants belonging to the well-watered and the high drought treatments. Thus, these compounds may be reliable cues for parasitoids to indicate host presence even under drought stress conditions. Notably, most of these compounds were absent or emitted only in low amounts from the aphid-infested plants in the moderate drought treatment, coinciding with low parasitoid attraction. Emission rates of p-menthane, β-myrcene, 3-carene, D-limonene and o-cymene standardized by plant biomass were highest from severely drought-stressed aphid-infested plants, suggesting that the production of these potentially important compounds might be upregulated in highly drought-stressed plants upon aphid feeding to compensate for the decrease in plant biomass. Moreover, γ-terpinene and trans-β-ocimene, two monoterpenes with a high Hpc that should not be effected by stomata conductance, were absent from the blends emitted by drought-stressed plants. This may indicate that other regulatory processes such as phytohormonal signaling may be responsible for altered VOC emission upon drought stress. It has been suggested that plants may downregulate indirect defenses and rely more strongly on direct defenses upon drought stress (Lin et al. 2023), but the increased aphid performance on drought-stressed plants suggests that this has not been the case in our study. Alternately, changes in VOC emission may be the result of phytohormonal cross talk and dual roles of phytohormones in biotic and abiotic stress signaling (Salvi et al. 2021; Aerts et al. 2021; Singh & Roychoudhury 2023).

Aphidius colemani may actually benefit from a reduced attraction to aphids on drought-stressed sugar beet plants, because emergence rate and adult size were strongly reduced on drought-stressed plants. This may be due to the smaller body size of aphid hosts on drought-stressed sugar beets (Rahman et al. unpublished). Aphid body size can be affected by drought stress (Ahmed et al. 2017; Kansman et al. 2021). Aphid body size and the size of emerging parasitoids are usually positively correlated, and parasitoid emergence rate can be related to host size and quality (e.g. Garratt et al. 2010; Tariq et al. 2013; Yasir Ali et al. 2022). In our study, emerging adults of A. colemani were indeed smaller when developing in aphids on drought-stressed plants. Interestingly, only the emergence of females was negatively affected by drought stress in our experiment, with just four females emerging from aphids on highly drought-stressed plants compared to 16 males. Under moderate drought, the sex ratio was almost equal, while on well-watered plants, 2/3 of the emerging parasitoids were females. A lower emergence rate and a shift from female-biased to a male-biased sex ratio of A. colemani and D. rapae under severe drought stress was also observed by Tariq et al. (2013) and Shehzad et al. (2020). Aphid parasitoids can actively control sex allocation by laying more unfertilized eggs, resulting in more male offspring on poor quality hosts (Cloutier et al. 1991; Pandey & Singh 1999). Alternatively, female parasitoids, with their larger body size, may require larger hosts to develop successfully, and female larvae may not have been able to develop successfully in small hosts on drought-stressed plants. Indeed, smaller males emerged from aphids on highly drought-stressed plants than from aphids on moderately drought-stressed plants, while the body sizes of female parasitoids did not differ between the two drought treatments. This suggests that female A. colemani may have already reached the lower limit of their body size on moderately drought-stressed plants. Tariq et al. (2013) also found no effect of drought stress on the size of female A. colemani, but female D. rapae were smaller on drought-stressed plants. In contrast, Shehzad et al. (2020) reported negative effects of severe drought and positive effects of mild drought on the body size of both parasitoid species. Reduced attraction to drought-stressed, aphid-infested sugar beet plants as a result of reduced VOC emissions may thus help A. colemani to avoid oviposition on low quality hosts, especially when plants without water deficit are available in close proximity.

More mummies formed on highly drought-stressed plants when compared to the well-watered controls. The higher mummification rate could have been the result of higher attack rates on highly drought-stressed plants, because smaller aphids require less handling time by the parasitoid (Wu et al. 2011). Alternatively, immune responses that kill some of the parasitoid eggs or larvae may have been compromised in aphids that developed on highly drought-stressed plants. Immunity to parasitoids is encoded in the aphid genome (Martinez et al. 2014) and provided by defensive endosymbionts such as Hamiltonella defensa (Rouchet & Vorburger 2014). Heat stress negatively affects the immunity provided by H. defense, but has no negative effect on PAXS (pea aphid X-type symbiont), another defensive endosymbiont (Benjamin & Nielsen 2006; Guay et al. 2009). It remains to be tested whether plant-mediated effects of abiotic stress, such as drought, also influence aphid immune responses against parasitoids. We further found that parasitoid larvae needed less time to complete larval development on highly drought-stressed plants compared to well-watered controls, which coincides with faster aphid development in this treatment. In contrast, parasitoid development time increased with increasing drought stress in the study by Ahmed et al. (2017). Changes in the duration of development might alter the predation risk of parasitoid larvae, which often become intraguild prey of aphid predators (Mottaghinia et al. 2018) and thus affecting the size of the parasitoid population. However, these effects might be minor compared to the reduced female emergence on drought-stressed plants observed in our study.

Due to the intimate relationship between aphid parasitoids and their hosts, there are multiple ways in which drought stress can affect parasitoid performance and foraging behavior, and it is difficult to predict how this would affect pest control. Suppression of aphid populations by A. colemani in a cage experiment was strongest under mild drought stress and moderate under severe drought stress (Kansman et al. 2021). In our study, the effects of drought stress on parasitoids were mainly negative, due to reduced numbers of emergent females and a reduced attractiveness of VOCs emitted by drought-stressed plants. A reduced ability of parasitoids to find aphid-infested plants combined with a low female emergence rate can have devastating effects on pest control. Moreover, drought stress often coincides with high temperatures and low atmospheric humidity and these additional factors may exacerbate the effects of drought stress. For example, low atmospheric humidity negatively affects parasitoid activity and host-finding success in the field (Vosteen et al. 2020). Drought stress combined with high temperatures negatively affected the ability of parasitoids to control aphid populations, despite the positive effects of these factors when tested alone (Romo & Tylianakis 2013). Overall, our findings indicate that drought stress exerts positive plant-mediated, bottom-up effects on aphids infesting drought-stressed sugar beets, resulting in a positive impact on the A. faba population while negatively affecting its parasitoid, A. colemani. Consequently, we anticipate that with an increased frequency of drought events projected under future climate change conditions, aphid outbreaks could be facilitated in this system.

Author contributions

S.R., M.R. and I.V. conceptualized this study. S.R. conduced the experiments, analyzed the data and prepared the figures. S.R. and I.V. wrote the main manuscript. All authors reviewed and edited the manuscript and approved the final version.