Unravelling mycorrhiza-induced wheat susceptibility to the English grain aphid Sitobion avenae

Arbuscular mycorrhizal (AM) fungi are root symbionts that can increase or decrease aphid growth rates and reproduction, but the reason by which this happens is unknown. To investigate the underlying mechanisms of this interaction, we examined the effect of AM fungi on the English Grain aphid (Sitobion avenae) development, reproduction, attraction, settlement and feeding behaviour on two naturally susceptible varieties Triticum aestivum (L.) variety Solstice and T. monococcum MDR037, and two naturally resistant lines, T. monococcum MDR045 and MDR049. Mycorrhizal colonisation increased the attractiveness of T. aestivum var. Solstice to aphids, but there was no effect on aphid development on this variety. Using the Electrical Penetration Graph (EPG) technique, we found that mycorrhizal colonisation increased aphid phloem feeding on T. monococcum MDR037 and MDR045, colonisation also increased growth rate and reproductive success of S. avenae on these varieties. Mycorrhizas increased vascular bundle size, demonstrating that these fungi can influence plant anatomy. We discuss if and how this could be related to an enhanced success rate in phloem feeding in two varieties. Overall, we present and discuss how mycorrhizal fungi can affect the feeding behaviour of S. avenae in wheat, inducing susceptibility in a resistant variety.

MDR037, MDR045 and MDR049 decreased over time within the root system (F 15, 92 = 2.574, P < 0.05). The T. aestivum variety Solstice showed a lower affinity towards the mycorrhizas than the T. monococcum varieties (F 3, 92 = 11.049, P < 0.05). The %RLC of Solstice remained the same throughout the experiment despite the increasing root length.
Mycorrhizal quantification carried out after experiments showed that the %RLC was lower in the control plants than the mycorrhizal plants (F 1, 247 = 15.87, P < 0.001) with the majority of control plants exhibiting no colonisation (Table 1).
Aphid development and fecundity. The growth rate of S. avenae was lowest on MDR045 and MDR049 and highest on Solstice and MDR037 (F 3, 94 = 25.797, P < 0.001), this was especially noticeable in the non-mycorrhizal plants (Fig. 2a). Colonisation by mycorrhizal fungi increased the growth rate of aphids raised on MDR037 and MDR045 plants (F 1, 94 = 16.797, P < 0.001) (Fig. 2a). Furthermore, there was an interaction between the treatment and variety (F 3, 94 = 3.912, P < 0.05) as mycorrhizas increased the growth rate of aphids on MDR037 and MDR045, but there was no effect on Solstice and MDR049.
There was a significant difference in the time aphids took to produce their first nymph. Aphids raised on MDR037 and MDR045 took the most number of days, whilst those raised on Solstice and MDR049 took the least number of days (F 3, 80 = 48.636, P < 0.001) (Fig. 2b). Importantly, aphids on MDR049 and Solstice took the same number of days to produce their first nymph. Mycorrhizal colonisation greatly reduced the number of days before aphids produced their first nymph (F 1, 80 = 48.636, P < 0.001) across all of the varieties (Fig. 2b). There was a strong interaction between the variety and treatment (F 3, 80 = 4.606, P < 0.01) as AM fungi reduced the time aphids on MDR045 produced their first nymph to a much greater extent than the other three varieties.
Although there was no difference in the time aphids on MDR049 and Solstice took to produce their first nymph, this did not extend to the lifetime reproductive success as MDR045 and MDR049 had the lowest reproductive success and aphids raised on Solstice and MDR037 had the highest (F 3, 91 = 52.153, P < 0.001) (Fig. 2c). The reproductive success was directly related to the growth rate results (Fig. 2a). The effects of AM fungi on the number of days to produce first nymph were similar to those observed for the reproductive success. Across all of the varieties, aphids on mycorrhizal plants had more reproductive success than those on non-mycorrhizal plants (F 1, 91 = 29.790, P < 0.001) (Fig. 2c).
Leaf anatomy. Vascular bundle width was greater in MDR037 plants than in MDR045 plants (F 1, 36 = 14.68, P < 0.001) ( Table 2). Mycorrhizal colonisation increased the bundle width consistently in both varieties (F 1, 36 = 6.77, P < 0.05) ( Table 2) so there was no interaction between the variety and the effect of colonisation.

Aphid feeding behaviour.
Overall, the addition of AM fungi had a large effect on the feeding behaviour which is apparent from the analysis of the feeding phases (Table 3).
Probing (tissue penetration). The duration of the first probe was on average 15 minutes longer in aphids feeding on MDR045 plants than MDR037 (F 1, 61 = 4.590, P < 0.05). Mycorrhizal colonisation had no effect on the duration of the first probe.
Pathway phase. The pathway (C) phase is defined as when the stylet is moving through the mesophyll towards the phloem. The number of pathway periods and the average time spent in the pathway phase did not differ between the varieties. Colonisation by mycorrhizas reduced the number of C phases that occurred (F 1, 61 = 8.39, P < 0.01), but increased the average time spent within each C phase (F 1, 61 = 6.62, P < 0.05); this was shown in both MDR037 and MDR045 varieties.
Salivation and phloem feeding. Aphids on MDR037 exhibited more phloem feeding phases (E2) than those feeding on MDR045 (F 1, 61 = 4.18, P < 0.05). The addition of AM fungi had no effect on the number of E2 phases. However, aphids on mycorrhizal plants fed on average for longer (F 1, 61 = 5.28 P < 0.05) with a longer maximum phloem feeding period (F 1, 61 = 4.333, P < 0.05). In both of these variables, the difference between the treatment and control was more pronounced in MDR045 than MDR037. Colonisation by AM fungi caused aphids on MDR045 to experience more sustained E2 phases, which lasted for longer than 10 minutes with the absence of salivation, (F 1, 61 = 5.471, P < 0.05). The mycorrhizal fungal colonisation also caused aphids to spend a longer proportion of the overall time within the phloem (F 1, 61 = 3.953, P ≤ 0.05).

Penetration difficulties.
There was no difference in the number of penetration difficulty events between aphids feeding on MDR037 and MDR045; however, aphids feeding on mycorrhizal plants of both varieties experienced approximately half the number of penetration difficulties than those feeding on control plants (F 1, 61 = 16.71, P < 0.001).  (Fig. 4a) showing an interaction between the treatment and hours post release (F 1, 12 = 3.819, P < 0.05). The colonisation of mycorrhizal fungi had no effect on the attraction and settlement of aphids in the other varieties investigated (Figs 3 and 4).

Discussion
We have combined EPG feeding behavioural analysis with studies on aphid reproductive success, pre-and postalighting preferences, and leaf anatomy to unravel the effect of mycorrhizal colonisation on modern and ancestral   wheat with different susceptibility to these destructive herbivores. Conflicting reports of the effects of AM fungi on plants are often based on a narrow range of experiments, but here we have shown that mycorrhizal colonisation differs between ancestral and modern wheat varieties, has a positive effect on aphid's ability to phloem feed, induces susceptibility in a resistant wheat variety, and that this is likely associated with factors involving an increase in the size of the vascular bundle. These results demonstrate the potential of AM fungi to affect plant resistance to aphids and induce susceptibility by interfering with the plants natural defence mechanism. The increased size of the sieve element could be indicative of a healthier and more nutritious plant. The higher level of salivation on resistant lines without AM fungi is indicative of problems with establishing phloem feeding, this suggests that the AM fungi are affecting the plants ability to block the sieve elements. Therefore, it is likely that there are multiple factors at play including plant anatomy, defence chemistry and plant health which we discuss herein.
Mycorrhizas can benefit their host plant by increasing their yield and nutrient acquisition 37 especially those nutrients which are poor at diffusing through the soil such as P and Zn. Mycorrhizas do this by extending hyphae through the soil, thereby accessing these diffuse-poor nutrients 38 . Although mycorrhizas can colonise wheat varieties, the extent of symbiosis that occurs between the different cultivars and varieties is varied due to the ability of the fungus to absorb soil nutrients as well as the amount of carbon the plant will supply to the fungus 39 . Our work shows that there are differences in the mycorrhizal affinity depending on the wheat cultivar. The T. aestivum hexaploid susceptible variety Solstice was the most modern variety investigated in this study and it had the least affinity to the mycorrhizas. This is in keeping with previous experiments which showed that older crop varieties of current cultivars as well as ancestral cultivars have a higher level of symbiosis with AM fungi than more modern varieties 39 . The AM fungal colonisation in the T. monococcum varieties decreased over time, possibly due to the fungus proliferating at a slower rate than the root length increased, therefore creating a dilution effect. The AM fungi in the Solstice root system maintained a steady level of colonisation, suggesting that although Solstice has a lower affinity towards AM fungi, the fungi are able to proliferate more within the root system than in the T. monococcum varieties.
The multi-trophic interactions between mycorrhizas and herbivorous insects have been widely studied. The effect of mycorrhizal fungi on higher trophic levels is dependent on many biotic and abiotic factors, including soil phosphorus limitation as well as the variability that comes with different host species 26 . Here we have shown that the level of mycorrhizal colonisation is dependent on the Triticum species as T. aestivum var. Solstice had less AM fungi colonisation than T. monococcum varieties.
We investigated how the mycorrhizal colonisation affects aphid development and reproductive success. Results show variation within the diploid varieties and that the varieties MDR045 and MDR049 hold some resistance towards S. avenae with a slower aphid growth rate and a lower reproductive success. These results are in line  Table 3. Feeding behaviour of Sitobion avenae feeding on mycorrhizal treated and control Triticum monococcum varieties MDR037 and MDR045. Mean (± SE) EPG variables; total duration (in seconds), frequency and average duration (predicted means) from 8 hour recordings of feeding.
with previous investigations of T. monococcum susceptibility to herbivorous insects. di Pietro et al. 23 have shown that S. avenae aphids raised on T. monococcum varieties had lower reproductive success than on the susceptible T. aestivum variety Arminda. Arbuscular mycorrhizal fungal colonisation had less of an effect on aphids raised on Solstice as the AM fungi reduced the time to first nymph being produced and their reproductive success, but had no effect on growth rate. Solstice had the least affinity towards the commercial mycorrhizas so the effect of the treatment may be too low to be noticed in these experiments. Mycorrhizal root colonisation increased the growth rate of aphids on MDR037 and MDR045, as well as reproductive success of aphids on MDR037, MDR045 and MDR049. Mycorrhizal colonisation is known to increase life history traits of aphids including growth rate and reproductive success 27 . These findings are in keeping with Koricheva et al. 25 showing that AM fungi affects S. avenae on ancestral plants in the same way as other phloem feeders on modern plants. To date, the mechanism of this interaction was unknown; it could be due to many factors including increased leaf size 40 , plant height and biomass (Supplementary material Fig. 1) in which there would be more area for potential feeding. However, we have shown that the internal anatomy of the leaf can be influenced by mycorrhizas, and factors related to this are likely to be a significant reason why aphid feeding success (and thence reproduction) is greater on mycorrhizal plants. Aphid sap ingestion takes place through phloem sieve elements which are transport channels of vascular bundles 41 and one previous study has also shown that mycorrhizas can increase the size of vascular bundles 35 . This could structurally aid aphid sap ingestion, by providing a larger target and thus increased chance of phloem location, resulting in increased susceptibility to attack 42 . The vascular bundle width of MDR037 was greater than MDR045, and AM fungi caused an increasein bundle width of both MDR037 and MDR045. The behavioural studies complement this, showing that aphids on mycorrhizal plants spent more time ingesting phloem and experienced longer sustained ingestion events. The effect of AM fungi on vascular bundles and other physical defence mechanisms including the role of callose and proteins 43 on aphid behaviour would need to be studied in future experiments. Therefore, we make the link between aphid feeding behaviour and plant vascular bundle size; which are both positively influenced by mycorrhizal colonisation. Our EPG results support other studies where aphids are able to locate the phloem on resistant varieties, but with reduced phloem ingestion or feeding periods and increased salivation 22,44 . These results suggest that there could be further aspects of resistance other than leaf anatomy, as they occurred in the absence of mycorrhizas. A number of possible mechanisms have been suggested, such as; the sap is unable to flow easily within the phloem due to blocked sieve elements 41 , the nutritional quality of the sap is altered by differences in the amino acid, sucrose and sugar: amino acid ratio 45 or that the wound created in the phloem alerts the plant defences faster 45 . It is unlikely, however, that a nutritional change in the sap is key to the resistance as studies investigating sap quality 46,47 show that it is only loosely linked to aphid feeding and that there are other factors involved in aphid feeding behaviour. Metabolites may also play a role in resistance; Greenslade et al. 22 showed distinct metabolic phenotypes for resistant T. monococcum lines in comparison to susceptible lines. The expression of trehalose was shown to be elevated in T. monococcum MDR049 and MDR657 and T. aestivum Solstice after aphid infestation 22 . The disaccharide sugar trehalose has been shown to increase plant resistance to the Powdery mildew pathogen (Blumeria graminis) by accelerating the stress signalling response 48 . During the probing phase, the aphids salivate which aids the stylet as it passes through the mesophyll, and that the saliva will prevent the plant responding to the phloem wound created as well as preventing the sieve elements from blocking 47 . The germination of AM fungal spores has been shown to decrease trehalose 49 and mycorrhizas have been shown to increase the enzyme trehalase activity in plant leaves and roots 50 . Trehalase acts independently from trehalose 51 breaking its substrate down into two glucose molecules 52 . The feeding behaviour investigations showed that aphids were able to feed for longer on the mycorrhizal plants. The higher activity of trehalase in mycorrhizal plants would reduce the activity of trehalose and in turn the stress signalling response would be slower in the mycorrhizal plants, allowing the aphid to feed more. Further investigations are needed to determine whether the addition of mycorrhizal fungi increases trehalase and decreases trehalose within T. monococcum leaves and if there is any associated effect on aphid feeding behaviour.
Mycorrhizas can manipulate the emissions of VOC's from plants which can increase the attraction of the plant to insects 28,53 . Our results showed that the addition of AM fungi only affected the attraction of aphids to T. aestivum variety Solstice, with alate S. avenae being attracted to and settling more on the mycorrhizal plants. Aphids on these plants had increased reproductive success, but there was no effect on the growth rate. We would expect this to translate into increased aphid infestation on mycorrhizal compared to non-mycorrhizal plants of this variety. Therefore, it is encouraging that Solstice had the lowest affinity towards AM fungi in the varieties studied. Babikova et al. 28 showed that AM fungi inhibited insect repellent VOC sequiterpenes and these are known to be present in the volatile profile of T. aestivum 54 . To our knowledge, the levels of sesquiterpenes in ancestral Triticum varieties and to what level AM fungi can affect them are unknown. This would be a most useful avenue for future research.
In summary, we conclude that AM fungi have a higher affinity to ancestral than modern wheat varieties and that any associated benefit to the plant can be compromised due to associated effects on the plant defences against aphids. The ability to alter the resistance level of a plant through the addition of AM fungi opens up avenues for understanding the mechanisms underlying plant resistance to aphids, and aphid-plant interactions, which have long eluded researchers. The commercial AM fungal product "Rootgrow" was provided by PlantWorks (Sittingbourne, Kent) and stored at 4 °C. Rootgrow is made up of a mixture of arbuscular mycorrhizal fungi which are of UK origin.

Materials and Methods
Soil used for growing all plants was provided by Petersfield Products (Cosby, Leicester). The soil composition was 75% medium grade peat, 12% screened sterilised loam, 3% medium grade vermiculite and 10% grit. The N content of the soil was 14% and the P 2 O 5 content was 16%.

Arbuscular Mycorrhiza inoculation and detection.
Inoculation occurred at the time of sowing seeds; pots of 4 cm diameter were filled with approximately 40 g of soil. Then 2.5 ml of inoculum was mixed in with the top layer of soil, the seed was added on top of this inoculum mixed soil and approximately 3 g soil added on top of the seed. The control was treated identically but used inoculum that had been autoclaved and allowed to dry for 12 h before sowing. Elimination of bacteria in the blending process of the inoculum meant that a microbial filtrate was not required in control treatments. At the end of each experiment (see below), the roots were extracted from the soil and stored in 70% ethanol at room temperature until they were stained for visualisation. The AM fungi visualisation method followed that of Vierheilig et al. 55 , root clearing was carried out in 10% potassium hydroxide in a water bath at 65 °C, and the time for the roots to clear was changed to 4 minutes. After this, the roots were rinsed with tap water and placed in a staining solution (84.4:15:0.6, dH 2 O: 1%HCl: Quink) in a water bath at 75 °C for 15 minutes. Stained roots were mounted onto microscope slides with coverslips. AM fungi root structures were quantified by the cross-hair eye piece method 56  Three mature apterous aphids were placed on each plant within a whole plant cage which consisted of insect proof netting with a plastic support structure. These mature aphids were allowed to larviposit overnight. The following morning the mature aphids were removed and the number of nymphs produced recorded. The neonate nymphs (< 1 day old) were weighed and transferred back to the plant of the same treatment and variety and left undisturbed for seven days. After seven days, the number of survivors was recorded and survivors re-weighed to determine the mean Relative Growth Rate (mRGR). This was calculated 57,58 as: After re-weighing, one of the nymphs was chosen at random and transferred back to their original plant. These aphids were left undisturbed to develop and were monitored every day; all of the aphids developed into apterous aphids. The time taken to produce their first nymph (FD) and the number of nymphs they produced over their lifetime (D) were recorded to calculate the intrinsic rate of increase using a constant from the mean pre-reproductive time for aphid species 61 . Subsequent nymphs were removed to prevent overcrowding. The intrinsic rate of increase was calculated as: Leaf anatomy. Mature leaves from 7 day old T. monococcum MDR037 and MDR045 were selected at random. Each leaf was fixed in a mixture of formalin, ethyl alcohol and acetic acid mixture and embedded in paraffin wax, as described by El-Afry et al. 59 . Sections were taken by rotary microtome to a thickness of 8 μ m and stained with safranine and light green. Sections were examined under light microscopy and width of vascular bundles measured, using an eye piece graticule. Ten replicate mycorrhizal and non-mycorrhizal plants of each variety were used.

Electrical Penetration Graph recording of aphid feeding behaviour. The Direct Current Electrical
Penetration Graph (EPG) method 36 was used to investigate the feeding behaviour of S. avenae on mycorrhizal and control T. monococcum MDR037 and MDR045. Apterous S. avenae were collected from T. aestivum cv Tybalt and starved for 1.5 h before attaching 20 μ m gold wire of 1.5-2 cm length to the dorsum of the aphid using water based adhesive containing silver paint. This was also used to connect the gold wire to a 3 cm copper wire which in turn was connected to a brass pin. The brass pin was connected to an EPG probe which was connected to an 8 channel Giga 8 Direct Current amplifier (EPG systems, Wageningen, The Netherlands). A ground electrode was inserted into the soil of each of the potted plants and connected to the amplifier. This experimental set-up was encaged within a grounded Faraday cage. Each aphid was placed on the first leaf of an individual seven-day old T. monococcum plant which was placed upon an upside down 100 ml beaker. A petri dish filled with water was placed underneath each potted plant to ensure that the soil stayed moist for electrical conductivity. The recordings were collected continuously for 8 hours using Stylet + data acquisition software (EPG systems, Wageningen, The Netherlands). During recordings, plants were kept at room temperature and under constant light. Two replicates of each of the varieties and treatments were carried out each day with the plants placed in a randomised order. A total of 104 aphids were investigated. Data processing of EPG aphid feeding behaviour experiments. The waveform patterns were identified, interpreted and annotated using Stylet + analysis software (EPG systems, Wageningen, The Netherlands) to correlate waveforms to aphid behaviour. The waveforms were placed into the following categories: non-probing, phloem sieve element salivation (E1), phloem sieve element ingestion (E2), penetration difficulties (F), xylem drinking (G) and stylet pathway phase (C) 36,60 . Responses were valid if a waveform was present within the first hour, and if waveform activity lasted for at least 30 minutes in the last hour of recordings. Annotated results were then imported into EPG analysis Microsoft Excel macro (EPG systems, Wageningen, The Netherlands) which calculated the variables from the annotated feeding behaviour.
Data analysis. The mycorrhizal colonisation of the different wheat varieties was compared with a one-factor analysis of variance in which the variety of Triticum was the independent factor. Data were subject to the angular transformation before analysis. Mycorrhizal colonisation of control and treatment plant roots after the completion of each experiment were compared using a two sample student's T-Test with a group factor. Data were subject to square root transformation before analysis. The differences in aphid mean relative growth rate (mRGR), average number of days to produce their first nymph and the intrinsic rate of increase (r m ) were examined with a two-factor analysis of variance (ANOVA) with the wheat variety and the mycorrhizal treatment as factors. The development data were subject to square root transformation.
Scientific RepoRts | 7:46497 | DOI: 10.1038/srep46497 The effect of mycorrhizal treatment on both the attraction and settlement of aphids over the course of 72 hours were determined by carrying out a multiple regression analysis in which the hours post release and the treatment were the independent factors.
The effect of mycorrhizal colonisation on vascular bundle width was analysed using a two-factor ANOVA with the variety and mycorrhizal treatment being the factors.
All of these comparative analyses were carried out in SPSS Statistics ® for Windows (2012, 21 st Edition, © IBM Corp, Armonk, New York, USA).
Electrical Penetration Graph results were analysed by a linear mixed model fitted by restricted maximum likelihood (REML) in Genstat ® (2013, 16 th Edition, © VSN International Ltd, Hemel Hempstead, UK). Due to the validity process the number of replicates was uneven across varieties and treatment; MDR037 control n = 15, MDR037 treatment n = 12, MDR045 control n = 16 and MDR045 treatment n = 19. Each of the 37 variables recorded were analysed separately and where necessary, the variable was subject to log, square root or logit transformations so the results conformed to the assumed variance homogeneity and residual normality. Variables that contained values of zero were off-set by adding half of the lowest value above zero that was recorded.