Concurrent anthropogenic air pollutants enhance recruitment of a specialist parasitoid

Abstract

Air pollutants—such as nitrogen oxides, emitted in diesel exhaust, and ozone (O₃)—disrupt interactions between plants, the insect herbivore pests that feed upon them and natural enemies of those herbivores (e.g. parasitoids). Using eight field-based rings that emit regulated quantities of diesel exhaust and O₃, we investigated how both pollutants, individually and in combination, altered the attraction and parasitism rate of a specialist parasitoid (Diaeretiella rapae) on aphid-infested and un-infested Brassica napus plants. Individual effects of O₃ decreased D. rapae abundance and emergence by 37% and 55%, respectively, compared with ambient (control) conditions. When O₃ and diesel exhaust were emitted concomitantly, D. rapae abundance and emergence increased by 79% and 181%, respectively, relative to control conditions. This attraction response occurred regardless of whether plants were infested with aphids and was associated with an increase in the concentration of aliphatic glucosinolates, especially gluconapin (3-butenyl-glucosinolate), within B. napus leaves. Plant defensive responses and their ability to attract natural aphid enemies may be beneficially impacted by pollution exposure. These results demonstrate the importance of incorporating multiple air pollutants when considering the effects of air pollution on plant–insect interactions.

1. Introduction

Insects use a variety of stimuli when interacting with their environment but rely heavily upon olfactory stimuli and particularly volatile organic compounds (VOCs) to perceive and interact with other organisms. As such, VOCs are used during critical stages of the life cycles of many insects, such as for locating hosts, food or mates [1]. Common atmospheric pollutants, such as nitrogen oxides (NO and NO₂, collectively NO_x) that are released from diesel vehicle exhausts, and ozone (O₃), are capable of chemically altering many of the VOCs that insects use for communication [2]. The disruption of these VOC cues may have wide-ranging impacts on the important ecosystem services that insects provide (e.g. pest-regulation services or pollination), which are critical for the functioning of terrestrial ecosystems.

Parasitic wasps (or parasitoids) are a provider of critical pest-regulation services, both in natural ecosystems and in many horticultural and arable cropping systems [3]. Their larvae live as parasites on or in other host insects (which can be crop pests), feeding upon them until the host dies [4]. Parasitoids of insect herbivores locate their herbivore hosts using VOC cues released by (i) the herbivores and (ii) the herbivore's host plants, which are produced and released in response to herbivore feeding [5]. Air pollution has been shown, in laboratory assays, to negatively affect the ability of parasitoids to locate their insect hosts, because the pollutants: (i) react with and chemically change the VOCs released by plants and insects in the air, and (ii) cause physiological changes to the plants, altering the VOCs that the plants release [6,7]. While these laboratory studies have been critical for identifying the mechanisms by which air pollution affects herbivore–parasitoid interactions, little is known about the ecological impacts of air pollution on these odour-mediated interactions, and whether the efficiency of pest regulation (i.e. parasitoid recruitment) under field conditions would be affected by their disruption. Recent reviews [8,9] have called for studies to bridge the gap in our knowledge on how VOCs influence the process of host location by parasitoids at larger spatial scales. In addition, there have also been calls for more work on the combined effects of different air pollutants [10]; emissions of NO_x occur alongside elevations in O₃ ([11] and references therein), yet no studies to date have considered their combined effects on multi-trophic interactions.

The effects of air pollution on VOC-mediated interactions are of increasing concern because global tropospheric background O₃ concentrations are rising [12]. Concentrations of O₃ are typically higher in rural areas than urban areas because in urban areas there is more NO_x pollution from vehicles and industry, and O₃ and NO_x readily react with one another, reducing O₃ concentrations [13]. NO_x emissions remain a serious problem, with areas throughout the UK continually exceeding limits imposed by the EU Ambient Air Quality Directive [14]. Transportation accounts for the majority of NO_x emissions in the UK (e.g. 47% in 2017; [15]) and many of the latest ‘Euro 6’ diesel cars that have been approved for sale continue to exceed air pollution limits (RDE test data; [16]) so diesel exhaust pollution is likely to remain a problem for decades to come [17]. Furthermore, as urbanization and traffic congestion increases, those critical ecosystem services delivered by insects that rely upon odour-mediated interactions, such as pest regulation and pollination, may be at increased risk [13,14,18]. Long-term, this risk may diminish as diesel exhaust emission sources reduce, at which point urban environments and polluted rural areas (e.g. those next to major roads) will have to contend with relatively higher levels of O₃ (because less O₃ is quenched by NO_x [19,20]) and its effect on the natural ecosystem services that we rely on [10].

Using a unique set of Free-Air Diesel and Ozone Exposure (FADOE) rings over 2 years, this study aimed to determine how changes in diesel exhaust and O₃ pollution, individually and in combination, shape populations of insect herbivore pests (cabbage aphids, Brevicoryne brassicae L.) and their parasitoids (specifically, Diaeretiella rapae MacIntosh). We used oilseed rape (or OSR; Brassica napus L.) as our model plant, which is commonly attacked by B. brassicae and is the most economically important brassica species in Europe [21]. The native European B. brassicae, now distributed throughout the world, are specialist feeders on brassicaceous crops (e.g. OSR, cabbage, broccoli, cauliflower, etc.). The parasitoid D. rapae commonly targets B. brassicae and has been shown to be attracted to aphid- and plant-released glucosinolate hydrolysis products that act as indicators of host presence [22–24], making these insects ideal model species for investigating the mechanisms underpinning the effects of diesel exhaust and O₃ on tri-trophic interactions. We identified the total number of parasitoids attracted to aphid-infested and non-infested plants under each pollution scenario (diesel exhaust, O₃, diesel exhaust plus O₃ and control) and recorded the abundance and oviposition success of the parasitoid, D. rapae. As such, we hypothesized that air pollution would result in reduced abundances of all species of parasitoid, including D. rapae, by reacting with and depleting the VOCs that parasitoids use to locate their aphid hosts, and as a result decreased parasitism rates of their aphid hosts. Phloem-feeding aphids tend to respond positively, in terms of growth/reproductive rate, to pollution-mediated changes in plant quality due to stress-related increases in nitrogen-containing compounds and/or decreases in plant defensive compounds [25]; consequently, reductions in parasitism may act to further increase their pest status [18,26]. We therefore hypothesized that pollution-mediated increases in aphid abundance (resulting in increases of aphid-emitted VOCs) would counteract the negative effects of air pollution on parasitoid abundance. We also examined whether the concentrations of specific glucosinolates in leaf tissue changed as a result of exposure to pollution and, if so, whether they were correlated with any changes in parasitoid abundance. Understanding how air pollution could modify pest–parasitoid interactions in the field at temporal and spatial scales that are relevant to the insects could provide greater insight into how current or future levels of air pollution may mediate and influence insect pest outbreaks.

2. Materials and methods

(a) . Insect cultures and plant material

Four Brevicoryne brassicae aphid cultures were established from a single parthenogenetic adult female collected from an OSR field at Sonning Farm (latitude 51.480330, longitude −0.899504). Cultures were maintained at 20°C on propagated OSR (cv. Tamarin, sourced from Senova, Cambridge, UK) for at least six generations (ca 8 weeks) prior to the experiment. For the experiment, OSR plants (cv. Tamarin) were grown from seed in 100 ml round cell trays in glasshouse rooms receiving natural light. After four weeks, plants were transplanted to 18 cm diameter pots containing ca 2.7 kg of vegetable topsoil (Quality Garden Supplies Ltd, Staffordshire, UK) and white mesh (organza) nets (55 × 75 cm) were placed over the plants and attached tightly around the rim of all pots to prevent any insect damage under field conditions. Bamboo sticks were placed inside the nets to prevent contact between the leaves and the mesh.

(b) . Field conditions and experimental procedures

In 2018, eight FADOE octagonal rings (8 m in diameter) were constructed at the University of Reading's Sonning farm within a field of winter wheat (Triticum aestivum cv. Skyfall), which maximized weed control (i.e. prevented weeds from growing, which themselves could have emitted different VOCs that may have altered the odour landscape inconsistently across the field). The centre of each ring was positioned 46 m from the centre of the field (latitude 51.482853, longitude −0.897749) in an octagonal formation, such that each ring was separated by a distance of at least 30 m. Full details of the FADOE configuration and layout are reported in [11]. Two rings were assigned to each of four treatments: (i) diesel exhaust (D), (ii) O₃, (iii) diesel exhaust and O₃ combined (D + O₃) and (iv) ambient air control. Concentrations of nitric oxide (NO), nitrogen dioxide (NO₂), nitrogen oxides (NO_x = NO + NO₂) and O₃ were monitored continuously and automatically maintained at field-realistic levels. The target concentrations were 120 ppb NO_x (based on average concentrations adjacent to major UK roadways and urban areas; [27]) and 90 ppb O₃ (based on peak concentrations recorded in rural European sites in 1990–2012; [28]) but average concentrations achieved within the rings were significantly lower than these, as described in the results. Diesel and O₃ generators were turned on for up to 17 h of the day (between 04.30 and 21.30), during which oviposition rates of D. rapae parasitoids are highest (females oviposit over 96% of their total eggs during the photophase; [29]). In 2019, the FADOE rings were moved to an adjacent field of wheat (latitude 51.482374, longitude −0.895855) and rotated within the field to account for the effects of pseudo-replication. A total of three experimental runs (described below) were undertaken, one in September–October 2018 and two in September–October 2019. Natural environmental conditions, including air temperature, wind speed and wind direction were monitored continuously throughout the experiment.

Netted OSR plants (16 plants in the first and third experimental runs and 28 plants in the second experimental run) were placed in each of the eight FADOE rings in four random groups (four plants per group in the first and third experimental runs and seven plants per group in the second experimental run; figure 1a). When plants were five-weeks old (i.e. one week after plants were transplanted into pots within the rings), each plant in two groups (aphid treatments A10 and OPEN) was inoculated with 10 teneral adult B. brassicae aphids. A further group was inoculated with 50 aphids (A50) and the final group remained insect free (CON). The aphids were left to establish for one week on the netted plants before placing a sticky trap (22 cm × 10 cm) on a stake in the centre of each group (outside of the nets), in order to capture naturally occurring parasitoids. After a further week, the first set of sticky traps was stored at −20°C and replaced in all but the OPEN group. The nets from the OPEN group were removed so that the aphids were exposed (figure 1a). After a further week, the plants in the OPEN group were re-netted and a sticky trap was positioned inside the net of each plant to catch emerging parasitoids. The second set of sticky traps was collected from the three other treatments (A10, A50 and CON) on the same day and stored at −20°C until required. Aphids were removed from each plant within these three groups using a pooter and stored in 60 ml pots at −20°C before being freeze-dried for 72 h and weighed. Plants were oven-dried at 70°C and weighed. The sticky traps from inside each of the nets of the OPEN plants were collected 10 days later and stored at −20°C. Diaretiella rapae, as well as all other parasitoids, were identified and counted from all sticky traps, from which D. rapae was easily identified by its distribution and wing structure [30,31]. Therefore, parasitism rate (i.e. parasitoid emergence of D. rapae) was recorded from the OPEN treatment only, and parasitoid abundance (i.e. total parasitoid abundance and the abundance of D. rapae) was recorded from the three other treatments (A10, A50 and CON). The timeline of experimental events is visualized in figure 1b.

Figure 1.

Experimental layout and timeline of events. Plants and aphid treatments within the eight individual rings are shown for the second experimental run (a), with ‘B. napus—GLS’ referring to OSR plants that were selected for glucosinolate analysis. The first and third experimental runs included four plants for each aphid treatment. Therefore, circles with dotted lines indicate plants that were included in the second experimental run only. The timeline of events (b) was the same for all three experimental runs. ‘ST’ refers to sticky trap. Parasitoid numbers on ST1 and ST2 (yellow diamonds) were pooled, and ST3 (green diamonds) was used to trap emerging parasitoids from OPEN plants only. (Online version in colour.)

(c) . Chemical analysis

For the second experimental run, three additional plants were added to each treatment within each ring for purposes of chemical analyses. Therefore, all seven plants in the A10, A50 and OPEN treatments were inoculated with aphids (described above). Upon harvest, three plants were selected from each of the A50 and CON treatments (figure 1a), from which four intact leaves were removed, freeze-dried and ground for glucosinolate analysis using liquid chromatography–mass spectrometry, following the protocol set out by [32]. Seven glucosinolates were detected using this method. Based on their side-chain structure and amino acid precursors [33], these glucosinolates were classified as aliphatic (glucoalyssin, progoitrin, glucobrassicanapin and gluconapin) and indolic (glucobrassicin, neoglucobrassicin and 4-methoxyglucobrassicin).

Solid phase micro-extraction gas chromatography–mass spectrometry was used to determine VOC relative abundances in fresh OSR leaf samples collected from three plants in each of the A50 and CON (i.e. no aphids) treatments within each ring. Sample preparation, headspace extraction from macerated leaves, chromatography and mass spectrometry conditions were as presented by [34]. VOCs were identified or tentatively identified by comparison of each mass spectrum with authentic compounds, or the NIST mass spectral database (NIST/EPA/NIH Mass Spectral database, 2014). A spectral quality value greater than 80 was used alongside linear retention index (LRI) to support the identification of compounds where no authentic standards were available. All peak areas were normalized. LRI was calculated for each VOC using the retention times of a homologous series of C₆–C₂₅ n-alkanes and by comparing the LRI with those of authentic compounds analysed under similar conditions.

(d) . Statistical analysis

Air pollution and aphid treatment effects on plant mass, final aphid population mass, parasitoid abundance and emergence, and glucosinolate concentrations were analysed using mixed models in the lme4 statistical package [35] within the R statistical interface v4.1.1. The fixed effects included air pollution treatment (control, O₃, D and D + O₃) and aphid treatment (no aphids, A10 and A50) as well as the two-way interaction between these terms. The random terms included year, run and ring location to account for seasonal and spatial differences that could confound any treatment effects and their inclusion was confirmed by model reductions using AIC and QQ plots. Negative binomial models were used for dependent variables with count data (numbers of D. rapae and other parasitoids and numbers of D. rapae that emerged from parasitized aphids) based on model deviance and critical chi-squared values. Where appropriate, response variables were transformed before analysis (electronic supplementary material, table S1) to standardize residuals, which were confirmed with AIC and QQ plots. The model d.f., total number of observations not accounting for replication (N_obs) and group N (i.e. N_group; Year/Run/Ring = 24) are included in all reported model statistics. Parasitoids captured in sticky traps within the OPEN group prior to net removal were not included in these analyses. Pairwise comparisons of means for treatment effects were made with Tukey's post hoc tests using the glht function in R's ‘multcomp’ package [36]. Pearson's correlation tests using the R base function cor.test were used to determine whether indolic, aliphatic and total glucosinolate concentrations were correlated with average D. rapae abundances for each ring. Normalized VOC abundances were analysed using XLSTAT (Addinsoft, Paris, France) protected analysis of variance (ANOVA) with post hoc Tukey's honestly significant difference pairwise comparison (p < 0.05). As such, if the ANOVA model is not significant, post hoc pairwise comparisons are ‘protected’ from overinterpretation (e.g. through the generation of type I statistical errors) by not being calculated, and simply stated as being non-significant. This approach has been used previously in relation to glucosinolate and VOC data in Brassicaceae plants [37].

3. Results

(a) . Air pollutant concentrations

Concentrations of NO_x and O₃ recorded during the experimental period of September 2018 and 2019 are reported and analysed in [38]. In short, individual pollution rings averaged low to moderate concentrations (as defined by DEFRA Air Quality Index; [39]) of 48.63 (± 1.39 s.e.) ppb NO_x and 38.85 (± 1.30 s.e.) ppb O₃ in the individual D and O₃ treatment rings, respectively, compared with 7.35 ± 0.18 ppb NO_x and 21.74 ± 0.28 ppb O₃ in the ambient (control) rings. In the combined (D + O₃) pollution treatment, NO_x concentrations (34.14 ± 1.14 ppb) were significantly lower than those in D, associated with the interaction between atmospheric NO_x and O₃ and the conversion of NO₂ to NO (see [11,38] for details). Moreover, the concentrations of O₃ in the D + O₃ treatment (20.38 ± 0.27 ppb) were 48% lower than the O₃ treatment, decreasing to levels equivalent to those in the control treatment.

(b) . Aphid population and plant mass under air pollution and aphid treatments

Aphid population mass was not significantly affected by air pollution but was significantly higher in the A50 treatment compared with the A10 treatment (figure 2). Air pollution and aphid treatments, individually and in interaction, had no significant effects on plant mass (electronic supplementary material, table S1).

Figure 2.

Aphid (B. brassicae) final population mass under air pollution (control, ozone, diesel exhaust or diesel exhaust and ozone) and aphid treatments (either 10 aphids or 50 aphids added as a parent population at the start of the experiment to B. napus plants) three weeks after aphid inoculation. Values are means ± s.e. Statistical effects of treatments, and their interaction, on aphid population mass shown. N_obs = total number of observations. N_group = group number associated with the random effects of Year/Run/Ring. (Online version in colour.)

(c) . Parasitoid responses to air pollution and aphid treatments

Diaeretiella rapae parasitoid abundance was significantly affected by air pollution (${\chi }_{3,10}^2=20.00$, p < 0.001, N_obs = 72, N_group = 24), whereby D. rapae decreased under O₃ but increased under D + O₃ compared with ambient (control) conditions (figure 3). The abundance of other parasitoids, in contrast, significantly decreased under all three pollution treatments relative to the control treatment (${\chi }_{3,10}^2=8.35$, p = 0.039, N_obs = 72, N_group = 24). Air pollution had a significant effect on the percentage of all parasitoids that were D. rapae (${\chi }_{3,10}^2=25.68$, p < 0.001, N_obs = 72, N_group = 24), which increased significantly under both diesel treatments (D and D + O₃) but not under O₃ (figure 3).

Figure 3.

Parasitoid abundance (left axis) and the percentage of total parasitoids that were D. rapae (right axis, blue) under air pollution (control, ozone, diesel exhaust or diesel exhaust and ozone) and aphid treatments (i.e. plots with no B. brassicae aphids, and either 10 aphids or 50 aphids added as a parent population to B. napus plants) after two weeks. Values are means ± s.e. Asterisk (*) indicates significant effects (p > 0.05) for all three dependent variables. Pairwise comparisons for the individual effects of air pollution and aphid treatments are shown in the electronic supplementary material, figure S1. (Online version in colour.)

The abundance of D. rapae significantly increased when aphids were present (i.e. plants that were initially infested with 10 and 50 aphids) compared with plants that were not inoculated with aphids (${\chi }_{2,10}^2=16.60$, p < 0.001, N_obs = 72, N_group = 24). The abundance of other parasitoids increased when plants were infested with 50 aphids but not when plants were infested with 10 aphids. There were no interactive effects of air pollution and aphid treatment on parasitoids. Full statistical results are shown in the electronic supplementary material, table S1.

The number of D. rapae that successfully emerged from parasitized aphids (OPEN plants) was significantly affected by air pollution (${\chi }_{3,8}^2=25.39$, p < 0.001, N_obs = 120, N_group = 24), whereby D. rapae emergence decreased under O₃ and increased under D + O₃ (figure 4).

Figure 4.

The effects of air pollution treatment (control, ozone, diesel exhaust or diesel exhaust and ozone) on the number of D. rapae parasitoids that emerged from parasitized aphids on exposed (OPEN) B. napus plants. Values are means ± s.e. Statistical effects of air pollution on parasitoid emergence shown. Bars with the same letters were not significantly different (p < 0.05). N_obs = total number of observations. N_group = group number associated with the random effects of Year/Run/Ring. (Online version in colour.)

(d) . Plant glucosinolate concentrations

Total glucosinolate concentrations in leaves of OSR increased in the combined D + O₃ treatment (${\chi }_{3,8}^2=11.89$, p = 0.008, N_obs = 133, N_group = 24), which was driven by increases in aliphatic glucosinolates (${\chi }_{3,8}^2=14.57$, p = 0.002, N_obs = 133, N_group = 24), especially gluconapin (3-butenyl-glucosinolate: figure 5). Concentrations of indolic glucosinolates did not vary significantly between pollutants or aphid treatments. Full statistical results are shown in the electronic supplementary material, table S1. Concentrations of aliphatic glucosinolates were positively correlated with the abundance of D. rapae that were attracted to aphid-infested plants, which was driven by the positive association between gluconapin and D. rapae abundance (figure 6).

Figure 5.

Indolic (highlighted blue) and aliphatic (highlighted red) glucosinolate concentrations under air pollution treatments (control, ozone, diesel exhaust or diesel exhaust and ozone) and aphid treatments (control B. napus plants with no aphids and A50 plants with aphids). Black letters indicate significant differences between air pollution treatments for total glucosinolate concentrations; bars with the same letters were not significantly different (p < 0.05). (Online version in colour.)

Figure 6.

Correlations between the mean number of D. rapae parasitoids counted on sticky traps within each ring and mean glucosinolate concentrations (GLS) per ring. N = 8. Correlation test statistics shown. Significant correlations between individual aliphatic GLS and parasitoid abundance (i.e. gluconapin) are displayed. Other individual GLS were not significantly correlated with parasitoid abundance. (Online version in colour.)

(e) . Plant volatile organic compound relative abundances

Pollution treatment and aphid infestation had a significant effect on 21 of the 44 VOC compounds identified (electronic supplementary material, table S2), and there was a two-way interactive effect of air pollution and aphid treatment on six VOC compounds (electronic supplementary material, table S3). In general, VOCs from plants that were infested with aphids and subjected to both O₃ and diesel exhaust combined were significantly higher than other aphid and/or pollution treatment combinations. For example, normalized peak areas of four methyl esters (methyl hexanoate, 2-hexenoic acid, methyl octanoate and (Z)-3-hexenyl isobutyrate), one alcohol (2-hexen-1-ol), two aldehydes ((E)-tiglaldehyde and (E,E)-2,4-heptadienal), dimethyl disulfide and hexanoic acid were significantly higher in aphid-infested D + O₃-fumigated plants compared with un-infested D + O₃-fumigated plants. Furthermore, normalized peak areas of three methyl esters (methyl hexanoate, 2-hexenoic acid and methyl octanoate), two ketones (3-pentanone and β-ionone), two aldehydes ((E)-tiglaldehyde and (E,E)-2,4-heptadienal) and dimethyl disulfide were significantly higher in aphid-infested D + O₃-fumigated plants compared with aphid-infested plants under ambient (control) conditions (electronic supplementary material, table S2).

4. Discussion

Exposure to diesel exhaust and O₃ pollutants had no clear effect on the population mass of aphids that were not exposed to natural enemies, yet these pollutants, in isolation, had opposing effects on the parasitism rate of D. rapae and their attraction to aphid-infested plants. In particular, we demonstrated negative effects of O₃ on parasitoid recruitment, mirroring effects which have been reported by others [18,40,41] and that are generally considered to be a result of the degradation of behaviourally important plant-released VOCs [42]. However, it is also possible that air pollution could physiologically alter VOC perception and directly impair insect health or motility [43–46]. In the current study, plants exposed to diesel exhaust, alone and in combination with O₃, were generally more attractive to D. rapae, and those aphids on plants exposed to both pollutants experienced higher rates of parasitism. This change in attraction, which was especially pronounced under the combined pollution treatment, occurred regardless of whether the plants were aphid-infested or not, indicating that changes in plant-released VOCs, as opposed to insect-emitted VOCs (i.e. those released from aphids directly), are more likely be responsible for the increased attraction of D. rapae. It is also possible that compounds within the pollution mix attracted D. rapae directly and, as such, further studies examining the direct impacts of concurrent air pollutants on the behaviour of parasitoids will be essential for mechanistically determining how parasitoids will respond to changes in the atmosphere as we shift away from fossil fuel dependence.

The combined pollution treatment significantly increased plant glucosinolate concentrations, especially of the aliphatic glucosinolate gluconapin (3-butenyl-glucosinolate). These changes in leaf tissue concentrations suggest that the plants modified their glucosinolate production as a stress-induced response to their exposure to both pollutants simultaneously [47,48]. Increases in concentrations of gluconapin were positively correlated with increases in abundance of D. rapae. The hydrolysis product of gluconapin, 3-butenyl isothiocyanate, has been shown to act as an attractant for D. rapae in previous studies [23,24]. We did not identify any glucosinolate hydrolysis products in the headspace of macerated OSR leaves, although it is possible that parasitoids may be attracted to hydrolysis products (i.e. isothiocyanates) that were below the detection threshold of instrumentation. OSR has typically been bred to contain low concentrations of glucosinolates and isothiocyanates [49], making them challenging to detect; however, further studies using non-destructive headspace sampling from whole living tissue (e.g. [5,50]) could more effectively mimic the VOC emissions that the parasitoids would be exposed to and provide the identities of additional VOCs that may contribute to parasitoid recruitment [49]. The increase in some VOCs within the leaf tissue of OSR when aphids were present and when plants were fumigated with both pollutants provides further evidence to suggest a stress-induced systemic impact on the secondary metabolism of OSR.

Air pollution-mediated changes in the proportion of D. rapae in the parasitoid assemblage has the potential to impact the structure of insect communities associated with OSR and other brassica species via changes in pest regulation [51]. Parasitoids other than D. rapae differed in their response to air pollution and tended to decrease in all pollution treatments. These contrasting responses of D. rapae, a specialist parasitoid of Brassica-feeding aphids [23,52], when compared with responses of other parasitoids to air pollution suggests that differences could be species-specific or a function of their diet (i.e. whether they are specialist or generalist feeders that target few or multiple hosts, respectively). As specialist parasitoids (i.e. those with preferred prey limited to one or a few related host species [53,54]), D. rapae may be more likely to respond to air pollution-mediated changes in VOCs because they tend to respond to a restricted set of stimuli, specific to their aphid hosts and the plants they feed upon [41,52,55]. Generalist parasitoids that target multiple hosts that may feed on many different plants, in contrast, may be less likely to rely on specific stimuli [38,56]) but instead use a range of different VOCs that may be more prone to being degraded by oxidizing air pollutants. It is also possible that specific VOCs are induced by host herbivory, therefore enhancing the signal that conveys host presence for D. rapae specifically.

Regardless of the specific mechanism used, parasitoids may be able to adapt to forage in polluted atmospheres by learning to associate altered VOC emission profiles with their target host [57]. As such, studies comparing parasitoids originating from polluted and unpolluted environments could quantify their ability to adapt, which is also likely to differ according to their diet specialization. Comparing generalist and specialist parasitoids originating from urban environments with those originating from more rural environments would be a useful next step. In general, mechanistically identifying how groups of generalist and specialist parasitoid species will respond to air pollution, using a combination of controlled laboratory studies and long-term population studies, could contribute to the targeted formation and/or release of specific compounds that effectively attract natural enemies and reduce herbivore populations. This would be especially valuable for protected (i.e. crops grown under glass or plastic) and high-value crops that more often rely on biological control for herbivore pest management [58,59]. From a wider ecological perspective, declines in the abundance of parasitoids other than D. rapae within polluted environments is likely to enhance the pest status of aphids and other plant pests in general, with negative consequences for food security.

5. Conclusion

Both diesel exhaust and O₃ pollution, individually and in combination, had deleterious effects on the abundance of parasitoid species other than D. rapae. By contrast, we demonstrated significant increases in D. rapae parasitoid recruitment under the combined effects of diesel exhaust and O₃ pollution, which contrasted with our hypothesis that both pollutants would deplete the VOCs that these parasitoids use to find their aphid hosts. This attraction response by D. rapae to diesel exhaust- and O₃-polluted environments is likely associated with an increase in the aliphatic glucosinolate, gluconapin, in OSR leaves, which is the precursor of an isothiocyanate which is attractive to D. rapae. These results stress the importance of studies incorporating the effects of multiple pollutants occurring in tandem in the natural environment. Concentrations of NO_x and O₃ emitted from the FADOE rings were lower than those considered safe under current air quality standards, emphasizing how only moderate levels of air pollution can have significant impacts on plant–parasitoid dynamics. Shifting to sustainable energy generation and electrifying the fleet of diesel vehicles within the next two decades will again significantly alter the levels of atmospheric pollutants at times of peak daily activity of important ecosystem service providers, including those providing pest-regulation services. A mechanistic understanding of how these service providers will respond to air pollution is, therefore, a vital, but hitherto neglected, component required to aid understanding and prediction of pest outbreaks.

Data accessibility

Data are deposited in the EIDC Digital Repository: https://doi.org/10.5285/b28c13f2-7355-4c4b-8c34-10470e1772d1 [60].

Supplementary material is available online [61].

Authors' contributions

J.M.W.R.: conceptualization, data curation, formal analysis, funding acquisition, investigation, methodology, resources, visualization and writing—original draft; L.M.B.: data curation, investigation and writing—review and editing; L.B.: data curation, formal analysis and writing—review and editing; J.J.: data curation and writing—review and editing; N.J.M.: methodology, resources, software and writing—review and editing; J.D.B.: conceptualization, supervision and writing—review and editing; R.D.G.: conceptualization, funding acquisition, supervision and writing—original draft.

All authors gave final approval for publication and agreed to be held accountable for the work performed therein.

Conflict of interest declaration

We declare we have no competing interests.

Funding

This work was supported by a British Ecological Society Large Research grant (grant no. LRB18/1009) and Leverhulme Trust Early Career Fellowship (grant no. ECF-2020-017) awarded to J.M.W.R. and a University of Reading Research Endowment Trust Fund Open Call awarded to R.D.G. Construction of the FADOE facility was funded by a Natural Environment Research Council grant (grant nos. NE/P002404/1 and NE/P001971/2).