SEARCH

SEARCH BY CITATION

Keywords:

  • amino acids;
  • aphids;
  • plant-mediated interactions;
  • semi-persistent transmission;
  • volatiles;
  • Hemiptera;
  • Amphorophora idaei;
  • Aphididae;
  • Rubus idaeus;
  • raspberry leaf mottle virus;
  • black raspberry necrosis virus

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Many insect herbivores and plant pathogens influence each other via plant-mediated mechanisms. Although there is speculation that these interactions may be important in structuring terrestrial food webs, few studies have empirically demonstrated the mechanisms by which pathogens manipulate the behaviour of their insect vectors. We investigated how infection of red raspberry, Rubus idaeus L. (Rosaceae), with two viral pathogens, black raspberry necrosis virus (BRNV) and raspberry leaf mottle virus (RLMV), affected the behaviour of their vector, the large raspberry aphid, Amphorophora idaei Börner (Hemiptera: Aphididae: Macrosiphini). As semi-persistently transmitted viruses, comparatively little is known about how such viruses affect vector biology. We also examined the effect of infection on plant volatile emissions and amino acid content, which could drive changes in aphid behaviour and performance. Virus-infected plants were initially more attractive to the aphid and the insects remained on infected plants for 30 min, but were found equally on uninfected plants 12 h after inoculation. Twenty-seven volatile compounds were identified. Two green leaf volatiles were emitted at higher concentrations by infected plants: (Z)-3-hexenyl acetate and 2-hexenal. In dose-response assays, (Z)-3-hexenyl acetate was attractive to the aphid at concentrations of 50 ng ml−1. When reared on infected plants, aphids took more than 3 days longer to reach adulthood compared with those on uninfected plants, although the number of offspring remained the same. Soluble amino acid (essential and non-essential) concentrations in raspberry leaves increased more than two-fold with virus infection. Amino acid composition was dominated by glutamate, accounting for 64 and 77% of the total in uninfected and infected leaves, respectively. Excessive glutamate may have underpinned the negative effects of viral infection on aphid performance. These results demonstrate the capacity of viruses to alter their host plant to manipulate vector behaviour, which may have evolved to be consistent with the transmission requirements of the virus.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Plants come under simultaneous attack from many damaging organisms, including insects and pathogenic microbes, such as bacteria and viruses (Stout et al., 2006). Hence, there is the potential for host manipulation by one enemy to indirectly affect the success of another. Natural selection predicts that viruses will evolve to maximise their fitness, which may also involve changes in the fitness and behaviour of their vector, most notably hemipteran vectors (Fereres & Moreno, 2009; Bosque-Pérez & Eigenbrode, 2011). In some cases, plant viruses make their host plants more attractive to the insects that transmit them (Khan & Saxena, 1985; Musser et al., 2003; Maris et al., 2004) and these vectors are often rewarded through improvements to their survival and fecundity mediated by chemical changes in the host plant (Castle & Berger, 1993; Belliure et al., 2005, 2008). Although manipulation of host plants and vectors by viruses is becoming increasingly recognised, the plant-mediated physiological processes underpinning the increased attraction of the insect vector to infected plants and, crucially, the mechanistic basis for their altered performance, are far less studied and often remain unidentified.

Estimates suggest that aphids are responsible for the transmission of 25–55% of all plant viruses, which are transmitted by insects (Nault, 1997; Ng & Perry, 2004; Hogenhout et al., 2008; Fereres & Moreno, 2009). Viruses vary in the way they are transmitted by aphids, from those which are carried in the insect’s mouthparts (non-persistent, non-circulative viruses; see Ng & Falk, 2006), to those which are ingested during phloem feeding and which circulate within the body before they are transferred to a new host (persistent, circulative viruses; see Hogenhout et al., 2008). Semi-persistently transmitted viruses are those which require longer acquisition periods than non-persistent viruses, but after successful acquisition, do not circulate in the aphid’s body (Ng & Falk, 2006). These different transmission strategies are likely to impose different selection pressures on both virus and vector because they rely on different feeding behaviours of the vector. For example, non-persistent, non-circulative viruses require only very short exploratory probes from the aphid mouthparts to be successfully acquired, therefore for optimal transmission, the virus should not induce any changes in host plant physiology that cause the aphid to remain on the plant for prolonged periods of feeding. Rather, the virus benefits from inducing the aphid to relocate to a more suitable (i.e., an uninfected) host, transferring the virus as it does so. This type of ‘deceptive’ interaction has only recently been described (see Mauck et al., 2010), identifying an important transmission strategy previously only characterised in vectors of animal diseases, e.g., malaria-transmitting mosquitoes (Hurd, 2003). In contrast to non-persistent viruses, persistent viruses have a requirement for prolonged feeding (hours to days) by their aphid vectors to ensure they are acquired from the plant, hence the impact of infection on aphid behaviour can be very different to that of non-persistent viruses. For example, plants infected with Potato leaf roll virus are more attractive to aphids (Eigenbrode et al., 2002; Alvarez et al., 2007) and promote their performance (Castle & Berger, 1993). Previous studies have focussed on persistent or non-persistent viruses and, to our knowledge, no similar detailed behavioural and physiological experiments have been carried out using semi-persistently transmitted viruses.

Aphids are often limited by nitrogen availability due to the low concentrations of amino acids in the phloem sap (Douglas, 1993; Dixon, 1998), so these nutrients may be important in terms of insect performance on infected and non-infected host plants. For example, Fiebig et al. (2004) reported a significant reduction in total amino acids in response to wheat infection with Barley yellow dwarf virus, with subsequent adverse effects for the cereal aphid, Sitobion avenae (Fabricius). Conversely, Johnson et al. (2003a) showed that infection of birch leaves with a fungal pathogen, Marssonina betulae (Libert), led to a higher concentration of free amino acids in symptomatic leaves where performance of the birch aphid, Euceraphis betulae (Koch), was enhanced. This led to higher abundances of aphids on infected plants in field populations (Johnson et al., 2003b).

The aim of this study was to investigate whether there was a preference of Amphorophora idaei Börner (Hemiptera: Aphididae: Macrosiphini) for red raspberry plants (Rubus idaeus L.) with two viral pathogens, black raspberry necrosis virus (BRNV) and raspberry leaf mottle virus (RLMV). As a vector of both these diseases A. idaei is a major pest of red raspberry (McMenemy et al., 2009), which is predicted to become more problematic with changes in climate (Martin & Johnson, 2011) and management practices (Johnson et al., 2012). RLMV and BRNV are both single-strand RNA plant viruses; RLMV is classified in the genus Closterovirus and has a single long, flexuous rod-shaped particle (McGavin & MacFarlane, 2010), whereas BRNV is classified in the family Secoviridae and has similarities with members of the genus Sadwavirus that have two spherical virus particles each containing a different viral RNA (Halgren et al., 2007). Aphid transmission of both RLMV and BRNV requires short acquisition and inoculation access periods, in the order of minutes only, suggesting that the virus particles do not circulate in the aphid during transmission. In addition, although the viruses do not achieve high titres in raspberry plants there is no evidence that they are phloem limited. Epidemiological field studies have shown that a newly planted raspberry crop can become 100% infected by BRNV after a single year (Stace-Smith, 1987) and then very rapidly become infected by other viruses, such as RLMV suggesting that the virus-infected plants may attract further feeding by viruliferous aphids. Because both viruses normally co-occur, we considered them as a dual infection in this study to ensure ecological realism. We hypothesised that, in line with several other plant virus systems, A. idaei would exhibit a preference for virus-infected host plants over uninfected plants. Furthermore, as both viruses are semi-persistent in their manner of transmission, we hypothesised that virus-infected plants would provide a poorer host plant for aphid development and that initial attraction to infected plants would be short lived, but enough for successful virus acquisition (McMenemy et al., 2009).

In addition, we investigated potential virus-induced changes to the host plant, which may be responsible for differential performance and attraction of aphid vectors. Specifically, we quantified leaf soluble amino acid concentration, an indicator of the amino acid composition of the phloem sap (Winter et al., 1992), and therefore of the aphid diet, to identify a causal link between aphid performance and host plant virus infection. We also analysed plant volatile emissions to identify the mechanism of aphid recruitment to raspberry plants.

Materials and methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Plants and insects

Mature raspberry plants (cv. Glen Ample) were infected with BRNV and RLMV by bottle grafting of virus-infected scions the summer prior to the commencement of experimental work. Vernalised root from mature ‘infected’ and ‘uninfected’ plants was cold stored at −80 °C for at least 6 months before propagation. Root samples were sown in hotbox propagators at 20 ± 1 °C in small trays using Bulrush bedding compost (Bulrush Horticulture, Bellaghy, UK). Seedlings were transplanted to individual pots (12 cm diameter) after 4 weeks and allowed to grow for a further 4 weeks before use in all experiments. Plants did not typically display chronic symptoms or discoloration. All experiments were conducted using the same temperature and light regime as above with all plants being screened for the presence or absence of virus by extracting total RNA from an excised leaf and using it in a reverse transcription polymerase chain reaction (RT-PCR) with virus-specific primers 1153f (5′-GCGCAATGAACCCAAGTTA-3′) and 1154r (5′-CAACATCGAATCCCTCAAGC-3′) for detection of BRNV and 1291f (5′-GTCCGACTTAGTGATGACGTATCG-3′) and 1295r (5′-CCTCGGATGGAGTAAGCCCACTG-3′) for detection of RLMV.

A clone of A. idaei (biotype 2; see McMenemy et al., 2009) was maintained in Perspex cages within a controlled environment in the laboratory at 19 ± 1 °C and L16:D8 photoperiod. The insects were reared on virus-free raspberry plants of cultivar Malling Landmark.

Aphid behavioural assays

Choice tests that assessed initial attraction of aphids to infected (BRNV and RLMV) and uninfected plants were conducted using two plants (described elsewhere) connected by a wooden ‘bridge’ (30 × 3 cm). Each bridge had a 15-ml universal tube bored through the centre such that the lid of the tube sat flush with the wood when closed. The bridge was balanced between the two plant pots ensuring that the central universal tube was positioned equidistantly from each plant and final positions were checked using a spirit level to ensure the bridge was parallel to the laboratory bench. Three apterous A. idaei adults were transferred to the central tube, left for 30 min then released by opening the lid of the tube. Aphids were assumed to have chosen a host plant after they alighted from the wooden bridge. The experiment was repeated in darkness and aphid positions were checked with a dim red light. In total 20 replicates were obtained for the light experiment and 10 for the dark experiment using different aphids and plant pairs on each occasion.

Two further behavioural assays were conducted over 30 min and 7 days. These experiments differed from the first in that they allowed aphids to move freely between plants after the initial selection response. The 30-min assay presented an uninfected and infected leaf within a square transparent arena (23.5 × 23.5 cm) to a single apterous A. idaei adult. Leaves remained attached to the plant through small apertures at each side of the arena. The position of the aphid was monitored every 1 min for 30 min, using 43 replicates (new aphids and plant pairs on each occasion). The 7-day assay consisted of two plants, one uninfected and infected with viruses, positioned at opposite sides within a wire-mesh insect cage (as described in Clark et al., 2011). Five apterous A. idaei adults were released from a dish between the two plants. Aphid positions were recorded after 12 and 24 h, then at 24-h intervals thereafter for 7 days. In total 28 replicates were obtained in this way.

Volatile collection and analysis

Volatiles were entrained simultaneously from three pairs of 8-week-old uninfected and virus-infected plants grown in identical conditions to those used in aphid bioassays, using carboxen-PDMS solid phase microextraction (SPME) fibres (Supelco, Sigma–Aldrich, Gillingham, UK). Entrainments were made within a leaf-shaped copper wire cage fitted around a fully expanded leaf and loosely enclosed and sealed within a multi-purpose cooking bag (Sainsbury’s, London, UK) made from transparent PET (polyethylene-terephthalate) sheet (Stewart-Jones & Poppy, 2006). Cages were suspended from a support frame placed above the plant to be sampled, to minimise physical stress to the leaf. Prior to collection of volatiles, the sampling cage was flushed out with air filtered through molecular sieves and activated carbon at 150 ml per min for 20 min, using an oil-free air pump. SPME fibres used in the experiments were new and had been pre-conditioned prior to use by exposure in a flow of dry nitrogen at 300 °C for 1 h as recommended by the manufacturer, using a fiber conditioning station attached to the CTC CombiPal GC–MS autosampler (CTC Analytics, Zwingen, Switzerland). An SPME fiber was attached to a fiber holder (Supelco) for use with the GC–MS autosampler and positioned using laboratory clamps allowing positional adjustments in three dimensions. The fiber within its protective sheath was pushed through a small perforation made in the PET film and the fiber was exposed 5 mm above and parallel to the leaf surface for 90 min on anatomically similar locations of each plant. On completion of entrainment, the SPME fiber was withdrawn into its protective sheath, and the fiber holder assembly was removed and fitted to the GC–MS autosampler.

Samples were analysed using a Trace DSQII GC–MS (Thermo Scientific, Hemel Hempstead, UK) fitted with a CombiPal autosampler configured for use with SPME fibres. Volatiles were desorbed from the SPME fiber isothermally at 280 °C for 2 min within a programmable temperature vapourising (PTV) injector operating in splitless mode and fitted with a Merlin MicrosealTM high pressure septum (Agilent Technologies, Wokingham, UK). Compounds were separated on a DB 1701 GC column (30 × 0.25 mm i.d. × 0.25 μm; Agilent Technologies) using helium at 1.5 ml per min in constant flow mode. The GC oven temperature was held for 2 min at 40 °C followed by a 10 °C per min temperature increase up to 240 °C with a further 10 min hold at that temperature. The GC–MS interface temperature was 250 °C and the MS was used in electron impact (EI) mode at 70 eV over a mass range of 25–400 amu with a source temperature of 200 °C. Data were acquired at four scans per second and analysed using the XCALIBURTM software package V. 2.07 (Thermo Fisher, Loughborough, UK).

Several XCALIBURTM raw data files were used to verify the presence of individual volatile analytes. For each compound, several specific and characteristic MS fragmentation ions were used for compound detection and quantification in a processing method created using XCALIBURTM (listed in Table 1). A time window was defined for each component, centered on the retention time of the appropriate chromatographic peak. Summed selected ion chromatograms (SIC) for the chosen ions were generated for each metabolite within the appropriate time window, and the SIC peak area was calculated automatically. Suitable ions for compound identification were selected on the basis, where possible, of having a high relative abundance and being unique to the compound and/or being well resolved from other ions with the same ‘m/z’ in the defined time window. Processed data were checked and corrected where required, before being subject to further data analysis. Compounds were identified by comparison of their mass spectra with entries in MS spectral libraries (NIST, Wiley, and Pal600K), by comparison of mass spectral data and retention behaviour with authentic standards and by extrapolation from data for known compounds.

Table 1. Composition of volatiles released from uninfected and virus-infected raspberry leaves, entrained using SPME fibres and analysed by GC–MS
ComponentIons used for identification and quantification1Rt2 (min)RRI3Volatiles GC–MS signal level (× 105)% of total plant volatile signal
Uninfected controlBRNV + RLMVUninfected controlBRNV + RLMV
  1. BRNV, black raspberry necrosis virus; RLMV, raspberry leaf mottle virus; SPME, solid phase microextraction; GC–MS, gas chromatography–mass spectrometry.

  2. Values for individual components are expressed as the means (± SEM; n = 3) of their raw GC–MS signal levels and as a percentage of the total signal for all components. Compounds with non-overlapping standard errors for raw signal levels are highlighted in bold.

  3. 1Ions used for automated compound identification and quantification using the XCALIBUR™ software.

  4. 2Retention time.

  5. 3Relative retention index (RRI) values are calculated for each analyte, based on comparison with the retention times of a mixture of n-alkanes in the range from C5–C16. Each alkane of carbon number Cn is assigned a RRI value 100 n. The RRI value of a given analyte is calculated based on linear interpolation of the spacing of its retention time between those of the two nearest adjacent retention index marker alkanes.

  6. 4Identity of compounds was confirmed by comparison of retention characteristics and mass spectra with those of authentic standards, and by comparison with mass spectral database entries.

  7. 5Compounds were identified by comparison with mass spectral database entries.

Acetic acid443, 45, 603.937711815.0 ± 390.91921.9 ± 350.049.88 ± 3.9245.88 ± 8.91
Hexanal444, 56, 57, 67, 72, 825.7588353.8 ± 16.591.3 ± 23.71.52 ± 0.412.02 ± 0.26
α-Pinene477, 79, 91, 93, 105, 121, 1366.90949765.4 ± 312.51153.8 ± 679.618.58 ± 2.8419.82 ± 7.92
2-Hexenal441, 55, 69, 83, 987.119622.00 ± 0.2911.59 ± 8.810.06 ± 0.0080.19 ± 0.11
Camphene467, 79, 93, 107, 121, 1367.2797125.9 ± 12.632.8 ± 17.50.65 ± 0.180.67 ± 0.20
Hexanal443, 44, 55, 57, 70, 81, 86, 1147.5398614.3 ± 5.916.7 ± 1.70.38 ± 0.100.43 ± 0.12
β-Pinene441, 69, 77, 79, 93, 121, 1367.80100174.6 ± 40.087.9 ± 52.51.73 ± 0.471.64 ± 0.56
δ-3-Carene477, 79, 91, 93, 105, 121, 1368.301033175.0 ± 87.9370.4 ± 292.54.00 ± 1.086.14 ± 3.85
Limonene441, 53, 67, 68, 79, 93, 107, 1368.70105774.8 ± 26.380.6 ± 24.62.03 ± 0.481.93 ± 0.58
β-Phellandrene441, 77, 79, 91, 93, 121, 1368.83106510.9 ± 4.311.8 ± 7.40.28 ± 0.050.23 ± 0.09
Eucalyptol443, 55, 71, 81, 84, 111, 139, 1549.021077118.9 ± 50.366.7 ± 19.42.90 ± 1.181.42 ± 0.02
(Z)-3-Hexenyl acetate443, 67, 829.07108057.1 ± 18.7179.2 ± 48.92.00 ± 1.153.87 ± 0.36
Benzaldehyde450, 51, 77, 105, 1069.151085134.3 ± 42.8152.5 ± 33.43.68 ± 0.803.50 ± 0.48
Octanal441, 57, 69, 849.23109032.3 ± 19.642.3 ± 8.40.75 ± 0.271.06 ± 0.33
Butyrolactone541, 42, 56, 869.861129233.9 ± 97.4213.0 ± 143.16.30 ± 2.744.62 ± 2.77
5-Ethyl-2[5]H-furanone555, 83, 1129.93113412.7 ± 3.010.8 ± 3.10.34 ± 0.030.24 ± 0.05
Linalool455, 69, 71, 93, 121, 13610.84119335.9 ± 22.142.8 ± 9.50.83 ± 0.311.03 ± 0.25
Nonanal457, 70, 82, 98, 11410.84119334.5 ± 22.635.4 ± 7.30.79 ± 0.330.86 ± 0.23
Camphor441, 55, 69, 81, 95, 108, 109, 15211.8912669.7 ± 1.511.8 ± 2.00.28 ± 0.040.31 ± 0.12
δ-Valerolactone542, 56, 70, 10012.2512908.8 ± 2.27.9 ± 1.90.24 ± 0.040.19 ± 0.04
Decanal443, 57, 70, 82, 95, 11212.32129717.5 ± 11.919.3 ± 4.00.39 ± 0.170.51 ± 0.21
Bornyl acetate555, 67, 93, 95, 121, 136, 15413.46137834.0 ± 9.749.1 ± 20.10.96 ± 0.271.69 ± 1.21
UK sesquiterpene 141, 55, 69, 79, 93, 105, 119, 133, 149, 161, 189, 20413.7613990.47 ± 0.080.48 ± 0.140.01 ± 0.0020.01 ± 0.001
α-Copaene441, 55, 69, 79, 93, 105, 119, 133, 149, 161, 189, 20413.8714072.0 ± 0.91.7 ± 0.60.050 ± 0.010.04 ± 0.005
UK sesquiterpene 241, 55, 69, 79, 93, 105, 119, 133, 149, 161, 189, 20414.4414545.5 ± 0.89.5 ± 1.60.16 ± 0.030.25 ± 0.08
β-Caryophyllene441, 55, 69, 79, 93, 105, 119, 133, 149, 161, 189, 20414.7014751.0 ± 0.37.3 ± 5.20.03 ± 0.010.16 ± 0.10
UK sesquiterpene 341, 55, 69, 79, 93, 105, 119, 133, 149, 161, 189, 20415.6115500.35 ± 0.050.41 ± 0.150.01 ± 0.0020.01 ± 0.001
Total   3765.1 ± 1066.74669.8 ± 1409.3  

Volatile behavioural assays

To test aphid responses to (Z)-3-hexenyl acetate, two filter paper discs (Whatman grade 1, 5.5 cm diameter) were positioned in opposing sides of a rectangular glass arena (21.5 × 16.5 cm). One of the discs was treated with 100 μl of paraffin oil (Sigma–Aldrich, Dorset, UK) and the other with 100 μl of (Z)-3-hexenyl acetate dissolved in paraffin oil. An apterous adult A. idaei, which had been starved for 1 h was released into the centre of the arena and its position was recorded every 1 min for a total of 20 min. Aphid responses were tested using a range of concentrations obtained by serial dilution: 10, 50, 100, and 250 ng ml−1 (Z)-3-hexenyl acetate in paraffin oil. The bioassays were conducted in a darkened laboratory where arena temperatures were maintained at a constant 23 ± 1 °C. Arenas were lit from above by a single white light emitting diode (LED), which was positioned to ensure an even distribution of light (350 ± 1 lux). Between 20 and 24 replicates were obtained for each volatile concentration tested using different aphids and paper discs for each test.

Aphid performance assay

One uninfected and one infected plant were positioned on a raised platform within a water-filled plastic tray, which served as a ‘moat’ barrier to aphid movement between plants and avoided the use of aphid clip-cages, which are known to affect plant chemical composition (Crafts-Brandner & Chu, 1999). This approach meant that we did not select feeding/birth sites on the plant for the aphids (introducing bias), but could only reliably monitor the less mobile larval stages. Eight-week-old plants were inoculated with one apterous adult aphid, which was then monitored daily for production of offspring. At the onset of reproduction, all but one first-instar aphid was removed. The remaining nymph was monitored daily to the onset of reproduction. The aphid and any offspring produced remained on the plant for 7 days before final counts of nymphs were made. In total 10 replicates were obtained for both uninfected and virus-infected plants.

Leaf chemical analyses

Soluble amino acids from 50 mg freeze-dried and milled leaf material were extracted in 2 ml of 80% high-performance liquid chromatography (HPLC) grade methanol (Fisher Scientific, Leicestershire, UK), which was subsequently removed by evaporation under vacuum. The residue was re-dissolved in ultrapure water and amino acids were separated by reverse phase HPLC after derivatisation with o-phthaldialdehyde (Jones et al., 1981) using a Hewlett Packard auto-sampling LC system with a Zorbax XDB-C18 column and fluorescence detection. Amino acids were quantified by comparison with AA-S-A18 (Sigma–Aldrich, St. Louis, MO, USA) reference mixture supplemented with asparagine, glutamine, and tryptophan. All protein amino acids, with the exception of proline and cysteine, could be quantified using this method. All analyses were conducted on 10 of each uninfected and virus-infected plants (controls), plus six plants of each type inoculated with aphids during the performance experiment. Control plants were subjected to the same growing conditions as those used in the experiment and were maintained on a bench in the same room as the experiment.

Statistical analysis

The proportion of aphids initially choosing virus-infected plants in the light and dark experiments was analysed using a generalised linear model (GLM) fitted using a binomial error structure with a logit-link function in R version 2.12.1 (R Foundation for Statistical Computing, Vienna, Austria). The proportion of aphids on virus-infected plants was fitted as the y-variable and light treatment (light or dark) was initially fitted as the x-variable. Light treatment was subtracted from the model as it was deemed to have no effect on aphid preference and its removal led to no significant increase in deviance from the simplified model. The minimum adequate model was compared with a null model, which assumed no aphid preference (proportion of aphids on infected plants = 0.5). Significant deviations between these two models were therefore indicative of an aphid preference for virus-infected plants and the result is reported as the chi-square value generated and associated probability. We report the effect of light treatment as the increase in deviance on addition to the minimum adequate model.

Due to the repeated measures on the same aphid individuals over time, the 30-min, 7-day, and volatile bioassays were analysed using a generalised linear mixed effects model (GLMM) assuming a binomial distribution and utilising a logit-link function. In each analysis, the proportion of aphids on infected plants was fitted as the y-variable and time was initially fitted as the x-variable. Cage or arena nested within time was initially fitted as the random term. Terms were subtracted from the model until any further removal led to significant increases in deviance and higher value of Akaike’s Information Criterion (AIC; see Akaike, 1974), providing a minimum adequate model. Probabilities are reported based on this model for each experiment. Aphids on the side of the arena or cage were assumed to be non-responsive and were excluded from the analyses. All mixed models were run using the lme4 package in R version 2.12.1 following the methods of Crawley (2007) to eliminate temporal pseudoreplication in the dataset.

Individual components of plant volatile blends were not subject to specific statistical tests but non-overlapping standard errors were considered evidence of a change in relative concentration between uninfected and virus-infected plants (sensu Eigenbrode et al., 2002). Aphid performance responses were analysed using one-way ANOVA after log-transformation to meet assumptions of normality for the test (Shapiro–Wilk test). To investigate potential interactions between plant virus infection and aphid feeding, leaf soluble amino acid concentrations were analysed using a two-way ANOVA. Some amino acid data (asparagine, serine, glycine, arginine, histidine, lysine, methionine, tryptophan, and valine) required log-transformation to meet assumptions of normality (as elsewhere). All statistical analyses were conducted in Genstat version 13.0 (VSN International, Hemel Hempstead, UK) unless otherwise stated.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Aphid behavioural assays

All aphids had selected a host plant 30 min after being released. A higher proportion of adult A. idaei initially selected plants infected with BRNV and RLMV, compared with those selecting uninfected plants, both in light and in dark conditions (Figure 1A and B). Light treatment had no effect on aphid preference for virus-infected plants (χ2 = 0.10, d.f. = 1, P = 0.76), so the data were pooled. The proportion of aphids selecting virus-infected plants was significantly different from 0.5 (χ2 = 7.62, d.f. = 1, P = 0.006). In the behavioural assay conducted over 30 min, consistently more aphids were present on virus-infected plants than on uninfected plants after 1 min (Figure 1C). There was, however, no difference (z = 1.064, P = 0.29) between plants in the 7-day assay, with aphids becoming equally abundant on both types of plant 12 h after introduction and throughout the course of the assay (data not shown).

image

Figure 1.  Aphid mean (± SEM) initial preference for uninfected and virus-infected plants in (A) light and (B) dark conditions. (C) Aphid preference over 30 min (z = 4.09, P<0.001; n = 43). BRNV, black raspberry necrosis virus; RLMV, raspberry leaf mottle virus.

Download figure to PowerPoint

Volatile collection and analysis

The sampling enclosure used for entrainment of volatiles was designed to provide an undisturbed airspace immediately above the leaf surface to minimise rapid dilution of leaf-derived volatiles with ambient air. This, in conjuction with the fine control of the positioning of the SPME fibre close to the leaf surface, maximised the efficiency of interception and entrainment of leaf-derived volatiles. Preliminary tests had shown that the levels of volatiles emitted from leaves were well within the loading capacity of the Carboxen-PDMS fibres used. Although the SPME technique used does not provide a measure of the actual concentration of volatiles present, the raw GC–MS signal levels generated for a total of 27 individual volatiles identified from uninfected and virus-infected plants (Table 1) were considered to be representative of their local abundance close to the leaf surface, allowing direct comparisons to be made between treatments. Abundances for individual volatiles listed in Table 1 are expressed as a proportion of the total GC–MS signal, although it should be noted that SPME fibre adsorption is dependent on saturation of the fibre, so this is indicative rather than absolute. Volatiles simultaneously trapped from pairs of plants using SPME fibres consisted largely of straight chain aldehydes in the range C6–C10, several monoterpenes and sesquiterpenes, and a number of characteristic ‘green leaf’ volatiles. Consistent differences in the volatile blends released in response to virus infection were observed. In particular, the green leaf volatiles 2-hexenal and (Z)-3-hexenyl acetate were elevated in virus-infected plants.

Volatile behavioural assays

Behavioural assays that exposed A. idaei to paper discs treated with (Z)-3-hexenyl acetate showed that aphids moved towards the treated disc more frequently than to the control disc at concentrations of 50 ng ml−1 (z = 2.661, P = 0.008), but not at the other concentrations of 10 (z = −0.7685, P = 0.19), 100 (z = 1.042, P = 0.30), or 250 ng ml−1 (z = 0.981, P = 0.33; Figure 2).

image

Figure 2.  Volatile behavioural assays, indicating mean (± SEM) response of aphids to paper discs treated with paraffin oil only (control) or with (Z)-3-hexenyl acetate dissolved in paraffin oil (n = 20–24). Asterisk denotes significant (P<0.05) difference (see text for details).

Download figure to PowerPoint

Aphid performance assay

When aphids were confined to a single test plant, development time to adulthood was more than 3 days longer on virus-infected plants compared with those feeding on uninfected plants (F1,18 = 4.75, P = 0.043; Figure 3). Aphids on infected plants also had significantly longer pre-reproductive periods (i.e., time taken to produce offspring; F1,18 = 6.15, P = 0.023). However, the number of offspring produced over 7 days on infected and uninfected plants was not significantly different (F1,18 = 0.04, P = 0.85).

image

Figure 3.  Mean (± SEM) aphid performance (development time, pre-reproductive period, and daily number of nymphs produced) on uninfected and virus-infected plants (n = 10). Asterisks denote significant differences between infected and uninfected plants (ANOVA: P<0.05). BRNV, black raspberry necrosis virus; RLMV, raspberry leaf mottle virus.

Download figure to PowerPoint

Quantification of leaf amino acids

HPLC quantification of soluble amino acids in the leaf tissue revealed that the total amino acid concentration of plants infected with BRNV and RLMV was significantly affected by virus infection, but not by aphid feeding (Figure 4A). This effect included both the concentration of essential and non-essential amino acids (Figure 4B and C). Virus infection had a significant effect on the concentration of 6 of the 8 non-essential amino acids quantified – asparagine, glutamine, aspartate, tyrosine, alanine, and glutamate (Figure 4D–I) – and on 3 of the 10 essential amino acids – isoleucine, methionine, and arginine (Figure 4J–L). There was an interactive effect of virus infection and aphid feeding on the concentration of arginine and methionine. No significant effect of virus infection and/or aphid feeding was found for serine, glycine, alanine, histidine, leucine, lysine, phenylalanine, threonine, tryptophan, or valine (data not shown). The dominant amino acid in both uninfected and virus-infected raspberry leaf tissue was glutamate, which accounted for 64% of the total amino acid molar concentration in uninfected plants. This proportion was elevated in response to infection with viruses (77%) and when aphids fed on virus-infected leaves (82%).

image

Figure 4.  Mean (± SEM) amino acid concentrations in uninfected and virus-infected leaves in presence (+) or absence (−) of aphids (n = 6–10). (A) Total amino acids, (B) essential amino acids, (C) non-essential amino acids, (D) asparagine, (E) glutamine, (F) aspartate, (G) tyrosine, (H) alanine, (I) glutamate, (J) isoleucine, (K) methionine, and (L) arginine. Asterisks denote essential amino acids. BRNV, black raspberry necrosis virus; RLMV, raspberry leaf mottle virus.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The results of this study strongly suggest that BRNV and RLMV manipulate the behaviour of their vector A. idaei by inducing a ‘deceptive’ attraction to plants, which in fact are nutritionally poor hosts for the aphid. Mathematical simulations of optimum virus transmission suggest that transmission peaks when aphids select infected plants and remain long enough to acquire the virus, before moving to a new host (Sisterson, 2008). The present study suggested that, after the initial attraction of the A. idaei to plants infected with BRNV and RLMV, there was no discernible difference in aphid numbers on infected or uninfected plants after 12 h. Although this does not equivocally demonstrate migration from infected to uninfected plants (as reported by Mauck et al., 2010) it suggests that attraction is short-lived and may result in some aphids that have acquired the virus moving onto previously healthy plants. Although the time period we observed is highly conducive to successful acquisition of both viruses (McMenemy et al., 2009), a longer experimental period with more frequent sampling is required to further investigate aphid migration and virus transmission.

Newly produced first instars of A. idaei took longer to develop to adulthood on plants infected with viruses than on uninfected control plants and thus the time taken to reach reproductive maturity and to begin to produce offspring was also longer. In spite of this, no difference was observed in the total number of offspring produced on either type of plant, but the prolonged development times would undoubtedly affect the rate of increase of aphid numbers on infected plants. Our results may represent another example, in this case for a semi-persistent virus, of the phenomenon noted by many studies of non-persistent viruses [Mauck et al. (2010) and references therein], namely that aphids are attracted to a host, upon which they subsequently perform poorly, at least in terms of development time. There remains a clear conflict between viruses and vector, with viral-induced changes in the plant phenotype benefitting the viruses at the expense of the aphid. Similar antagonistic effects of plant viruses on insect vectors have been found previously. For example, Donaldson & Gratton (2007) demonstrated that the population growth rate of Aphis glycines Matsumura on soybean infected with Alfalfa mosaic, Soybean mosaic, or Bean pod mottle viruses, was reduced by an average of 20% and Mauck et al. (2010) reported that squash plants infected with Cucumber mosaic virus supported smaller populations of Aphis gossypii Glover in the field.

Generally, plant stress can impair protein synthesis and lead to an increase in tissue concentrations of amino acids (Brodbeck & Strong, 1987). The two stressors, herbivory and virus infection, examined in this study might have been expected to exacerbate increases in leaf concentrations of soluble amino acids. However, almost all of the individual amino acids exhibited a less pronounced increase when aphids fed on plants already stressed by virus infection, which may reflect the direct removal of amino acids from the leaf pool by aphids during phloem feeding. The exception to the general trend was the essential amino acid methionine, which was most elevated in plants challenged by both virus infection and aphid feeding. Methionine has previously been suggested as a feeding stimulant for several aphid species (Mittler, 1967; Harrewijn & Noordink, 1971) and it could be hypothesised that the elevated levels in virus-infected raspberry tissue is a pathogen-induced mechanism to facilitate aphid acquisition of virions from the phloem. Further studies using electrical penetration graphs to detect the position of the stylets in the leaf would confirm this.

Aphid reproduction and growth is generally limited by amino acid availability. For example, Fiebig et al. (2004) reported an overall decrease in leaf amino acid concentrations in Barley yellow dwarf virus-infected wheat plants where the cereal aphid, S. avenae, performed poorly. However, in the present investigation an overall increase in leaf amino acid concentration, which is likely to reflect changes in phloem amino acid availability (Winter et al., 1992), did not promote aphid development. Furthermore, the non-essential amino acid glutamate, which was the dominant amino acid in this study (64% of total amino acids in uninfected plants and 77% in infected plants) has been implicated in a reduction in host plant suitability when present in high relative concentration (Douglas, 1993; Karley et al., 2002), as it is in a raspberry leaf. Furthermore, high relative levels of glutamate have been shown to be present in aphid-resistant plant cultivars and have been suggested as contributing to poorer aphid performance on these plants (Weibull, 1988; Chen et al., 1997). It seems that glutamate may act similarly as an indicator amino acid of host plant suitability for A. idaei in raspberry.

Our study has shown that it is detrimental for A. idaei to feed on virus-infected tissue when uninfected raspberry plants are available. Although this study did not measure longevity and fecundity of A. idaei for the entirety of the aphid’s lifespan, the prolonged development time is likely to have a direct impact on overall population growth, and could also have indirect implications for vector survival through an increased period of susceptibility to natural enemies and predators. It may, in fact, be less costly to the aphid to remain on these plants rather than investing energy in moving to a more nutritionally suitable host where it would potentially expose itself to an increased number of predators. As aphid morph determination can be influenced by plant virus infection (Blua & Perring, 1992; Fiebig et al., 2004), the negative effects on development may be counteracted later in the season with production of more winged morphs. In addition, the juvenile developmental stages of A. idaei are less effective than adults at evading attack by the parasitic wasp Aphidius ervi Haliday (Mitchell et al., 2010) and it is therefore beneficial for the aphid to reach adulthood quickly. Indeed, the slow-growth, high-mortality hypothesis has been reported across taxa (Feeny, 1970; Clancy & Price, 1987; Williams, 1999).

Although the plants used in this study did not yet exhibit the leaf chlorosis normally associated with infection with BRNV and RLMV, there were distinct differences in the wavelengths of light reflected from the leaves, particularly in the regions of light detected by aphids (see McMenemy, 2011). Nonetheless, similar aphid behaviours in both light and dark conditions suggest that plant volatiles play a major role in aphid attraction to virus-infected plants and subsequent analysis of raspberry volatile emissions indicated that compositional changes to the plant headspace in response to virus infection may be implicated. In particular, both 2-hexenal and (Z)-3-hexenyl acetate are green leaf volatiles [(Z)-3-hexenyl acetate is the end product of the metabolism of 2-hexenal], and have been shown to elicit electroantennogram responses from other aphid species, including the black bean aphid, Aphis fabae Scopoli, and the peach-potato aphid, Myzus persicae (Sulzer; Visser et al., 1996). These two compounds have been detected in volatile emissions from potato (Eigenbrode et al., 2002), faba bean (Nottingham et al., 1991), wheat (Jimenez-Martinez et al., 2004), and squash (Mauck et al., 2010). The bioassays demonstrated that individual A. idaei were attracted to (Z)-3-hexenyl acetate at concentrations of 50 ng ml−1. Under our experimental conditions, this falls within typical emission rates of green leaf volatiles from host plants also reported to be attractive to aphids (e.g., Ngumbi et al., 2007; Medina-Ortega et al., 2009). It should be noted that the volatile attractants may comprise a more complex blend of volatiles (see study by Ngumbi et al., 2007), but our results (together with the examples elsewhere) suggest that (Z)-3-hexenyl acetate is at least implicated in this attraction. In this respect, our results appear to support the hypothesis advanced by Mauck et al. (2010), that in cases where vector-borne pathogens increase attractiveness of plants to vectors, they do so by increasing or exaggerating the concentration of existing volatile cues used for host location, rather than eliciting the production of novel compounds. We did not detect any novel volatiles from infected plant, only changes in the relative concentrations of existing compounds.

In conclusion, we have shown that virus-induced changes to raspberry plant biochemistry results in the alteration of A. idaei behaviour. Specifically, this behaviour appears to be detrimental to the aphid, but it would likely be of benefit to the viruses by increasing the probability of transmission. Our results further illustrate the capacity of plant viruses to alter the host plant and manipulate vector behaviour. This represents a potentially important mechanism driving the evolution of plant virus transmission. Our study supports the emerging pattern of transmission mechanisms being one of the major factors shaping the evolution of pathogen-induced changes in host phenotypes.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The authors would like to thank Sheena Lamond for assistance with insect cultures, Wendy McGavin for advice on virus testing, and Fiona Falconer for her help with amino acid extractions and HPLC analysis. This work was funded by a joint PhD studentship scheme between the James Hutton Institute and the University of Sussex.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Materials and methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References