Elevated CO2 impacts bell pepper growth with consequences to Myzus persicae life history, feeding behaviour and virus transmission ability

Increasing atmospheric carbon dioxide (CO2) impacts plant growth and metabolism. Indirectly, the performance and feeding of insects is affected by plant nutritional quality and resistance traits. Life history and feeding behaviour of Myzus persicae were studied on pepper plants under ambient (aCO2, 400 ppm) or elevated CO2 (eCO2, 650 ppm), as well as the direct impact on plant growth and leaf chemistry. Plant parameters were significantly altered by eCO2 with a negative impact on aphid’s life history. Their pre-reproductive period was 11% longer and fecundity decreased by 37%. Peppers fixed significantly less nitrogen, which explains the poor aphid performance. Plants were taller and had higher biomass and canopy temperature. There was decreased aphid salivation into sieve elements, but no differences in phloem ingestion, indicating that the diminished fitness could be due to poorer tissue quality and unfavourable C:N balance, and that eCO2 was not a factor impeding feeding. Aphid ability to transmit Cucumber mosaic virus (CMV) was studied by exposing source and receptor plants to ambient (427 ppm) or elevated (612 ppm) CO2 before or after virus inoculation. A two-fold decrease on transmission was observed when receptor plants were exposed to eCO2 before aphid inoculation when compared to aCO2.

. This would imply that the larger amount of biomass could constitute a reservoir of infected material with subsequent higher risk of virus transmission by insect vectors 24 . Whilst, symptomatology of vegetable viruses can also be affected by rising CO 2 , enhancing earlier or more pronounced phenotypic differences between healthy and infected plants and making infected hosts more attractive to vectors 26,27 .
Investigating plant-insect interactions under increasing CO 2 is crucial to evaluate their consequences on environment and ecosystems, and to develop new crop management strategies under future climate change scenarios. Therefore, the objective of our work was to study the life history parameters and feeding behaviour of M. persicae on pepper plants under aCO 2 (400 ppm) and eCO 2 (650 ppm). We also analysed the direct impact of increasing CO 2 on plant growth and leaf chemistry. Links between plant tissue quality and aphid responses are discussed. Lastly, the cuticula-borne, non-persistent virus CMV was included in this investigation. We studied CMV transmission by M. persicae on ambient (427 ppm) and elevated CO 2 (612 ppm) conditions with source and receptor plants exposed to the two CO 2 regimes. Receptor plants were exposed to CO 2 conditions either immediately after aphid introduction (direct exposure), or two weeks prior, in order to study previous acclimation (indirect exposure).
CMV transmission by aphids. To understand the potential effects of increased CO 2 on virus transmission/ acquisition efficiency, we used CMV and the same aphid/plant combination as per previous experiments. No differences were found if CO 2 was applied to receptor plants straight after aphid introduction (direct exposure), when the CMV source inoculum had been grown either under aCO 2 or eCO 2 (Fig. 4). If the receptor plants had been previously grown under one of the CO 2 regimes before CMV inoculation by aphids (indirect exposure), a two-fold decrease on CMV transmission was observed when source and receptor plants were grown under eCO 2 conditions compared to aCO 2 (χ 2 = 4.432, p = 0.035) (Fig. 4).

Discussion
Current carbon dioxide (CO 2 ) concentration has exceeded 400 ppm and the concentration will continue increasing at an accelerating rate from decade to decade. This rising of CO 2 levels in the Earth's atmosphere causes general climate change and global warming 46 . Among the observed effects on plants, increasing CO 2 leads to changes in plant growth, physiology and metabolism 7,13,15 . At the same time, these changes have an indirect impact on insect pest biology and behaviour, altering their population growth or feeding habits, as well as on the symptomatology and incidence of associated pathogens such as plant viruses 24,26,27,35 . To our knowledge, this is the first study to demonstrate the variety of effects of increasing CO 2 on pepper plants and the subsequent indirect response on M. persicae fitness and feeding behavior. Moreover, these experiments have been coupled with research on CMV transmission with receptor plants exposed to ambient and elevated CO 2 either before or after aphid introduction.
Overall M. persicae fitness was reduced on pepper plants grown under eCO 2 , being consistent with previous results for this aphid on Brassicaceae species 10,15,21,33 and other aphid species such as A. pisum on broad bean or B. brassicae on Brussels sprout 6,13,18 . Not only was the time from birth to adulthood longer, but they also produced less offspring, which resulted in lowered growth rates. Again, this highlights the specificity of responses to hosts, as eCO 2 has been reported to increase the abundance of the green peach aphid on herbaceous species in the past 6,39 . The response to eCO 2 also differs among insect guilds. It is generally agreed that the decline in the foliar nitrogen content of host plants under eCO 2 prolongs the development of chewing insects, Statistical differences are calculated according to Student t-test for Gaussian variables or Mann-Whitney U-test for non-Gaussian variables (p ≤ 0.05). No significant differences are labelled as "ns", and significant differences are indicated by stars with ***p ≤ 0.001, **p < 0.01 and *p < 0.05. d is the time (days) from birth to the onset of reproduction; Md is the reproductive output per aphid that represents the duration of d; M 10 is the mean offspring number per female over the 10 day period; Td is the mean generation time; r m is the intrinsic rate of natural increase and RGR is the mean relative growth rate. and that they compensate for decreased nutritional quality by consuming more foliage 5,22,47 . By contrast, eCO 2 has species-specific effects on phloem-sucking insects 22 .
The amount of foliar nitrogen in pepper was significantly lower under eCO 2 when compare to ambient conditions, which may well explain the poor aphid performance we observed. Aphids which fed on plants grown under eCO 2 had a reduced efficiency, with an 11% longer pre-reproductive period and 37% fewer nymphs. This is probably mediated by reduced nitrogen content in the pepper source as aphids tend to be limited by the amount of nitrogen rather than carbohydrates 47 . Additionally, it has been shown that this aphid species is particularly sensitive to low nitrogen contents 48,49 . However, it has to be also pointed out that we measured whole foliar tissue nitrogen rather than phloem nitrogen but it is likely that our results could be extended to sap composition 11,42,50 . Chlorophyll content of leaves was also significantly lower under eCO 2 , this result being consistent with significantly lower nitrogen 7,15,17,18 . Also, it has been reported that C:N ratio of leaves can be more affected by increasing CO 2 than that of the plant as a whole 51 .
Carbon levels were similar in both treatments, as opposed to other plant species under eCO 2 , in which carbon content increased [14][15][16] . As expected, the increased height and above-ground dry weight of peppers are in agreement with previous studies, reporting enhanced growth under eCO 2 in several plant species 3,[11][12][13]21,33,35,36 . Finally, using infrared imagery an increase of 1.2 °C in plant canopy temperature was measured for peppers grown under eCO 2 . Cotton and wheat also show an increase ranging from 0.6 to 1.1 °C, and closure of leaf stomata appears to be the main driver behind this response 52 . Increased plant canopy temperature can further modify the microclimate for the aphid and higher temperature has been shown to increase virus titer in wheat 53 .
In this study, EPG recordings showed decreased aphid salivation into sieve elements and shorter non-probe phase under rising CO 2 , similar to the effects found for A. pisum 31,42 . However, no differences were found for the duration and number of events of phloem ingestion, suggesting that eCO 2 was not a factor impeding sap feeding by M. persicae, as opposed to A. pisum 31 . This result may also indicate that the diminished fitness observed could be mostly due to poorer tissue quality. What analysed parameters seem to stress is that there were no differences in feeding behaviour explaining a worse performance in eCO 2 plants. That is, aphids did not reduce or increase feeding, but simply ingested sap that was less nutritious. Therefore, with equal ingested sap, aphids grew slower under eCO 2 because they acquired a much more unfavourable N content diet.
Furthermore, the number of single E1 waveforms (phloem salivation) not followed by E2 events (phloem ingestion) and the percentage of time spent on E1 were lower under eCO 2 . Altogether, these results might imply that the ingestion success rate once aphids reached the phloem was higher under eCO 2 . An explanation could be found in the natural defense responses triggered under herbivory pressure, resulting in a rejection of sustained feeding under ambient conditions. Indeed, eCO 2 is likely to reduce the resistance ability of hosts against aphids by the down-regulation of genes involved in plant defence 31,39,54 . Without these barriers, aphids might be able to ingest phloem more easily under eCO 2 . Several mechanisms have been proposed, and include but are not limited to: 1) decreased activity of enzymes mediating the accumulation of reactive oxygen species, 2) reduced polymerization of phenolics absorbed by aphids that causes browning of cells in contact with the saliva, a disadvantage for aphid feeding, 3) down-regulation of ethylene signalling pathways 31 , and 4) decreased jasmonic and salicylic acid defences 39,54 .
Plants used for EPG recordings were grown inside CO 2 chambers for four weeks, whereas those used for the life history experiment were exposed to CO 2 during the aphid cycle as well, reaching a total of seven weeks, so it is likely that the nutritional profile of these plants declined with time. Additional work testing different periods of CO 2 exposure should be addressed to evaluate if aphid response changes with time.
Work on the interactions between pepper plants and aphids was complemented with the study of CMV transmission by M. persicae when source and receptor plants were either previously exposed to ambient and elevated CO 2 before aphid introduction (indirect effect) or straight after insect inoculation (direct effect). Elevated CO 2 lessened virus transmission risk if receptor plants had been previously grown under eCO 2 , but not if CO 2 was applied directly after pest introduction, which suggests that this difference may be associated with early exposure to elevated CO 2 . CMV is transmitted in a non-circulative manner during brief probes in the host epidermis 55 . As we observed a similar number of stylet punctures (pd) under both treatments, but shorter duration under eCO 2 ,

Table 1. Mean ± SEM (ranges in parenthesis) values of non-sequential and sequential EPG variables for the probing behaviour of Myzus persicae apterae adults on pepper plants grown under aCO 2 (400 ppm) and eCO 2 (650 ppm). P-values according to Mann Whitney U-test for non-Gaussian variables.
Bold-type indicates significant differences (p ≤ 0.05). PPW, proportion of individuals that produced the waveform type; NWEI, number of waveform events per insect; WDI, waveform duration (sec) per insect; WDE, waveform duration (sec) per event. Probe: probe activity. Waveforms: pd, short intracellular punctures; C, intercellular stylet pathway; E1, phloem salivation; single E1, E1 not followed by E2.
the decrease in CMV transmission might be due to plant resistance mechanisms once the virus infected the host, rather than an altered aphid feeding behaviour during the stages involved in virus inoculation. Under ambient CO 2 conditions, it has been shown that plant pathogens are responsible for triggering changes in nutritional quality or mediating attractiveness to increase the chances of vectors spreading viruses 56 . In our experiments, the decreased CMV transmission found under eCO 2 may be associated with lower aphid attraction to virus-infected plants, resulting in fewer viruliferous aphids, which would ultimately reduce the spread of CMV under future climate conditions. Consecutive rising of CO 2 concentration and global temperature are a reality of future scenarios. If disease severity changes under forthcoming environmental conditions, then additional research becomes imperative to understand the impact of viral infections on crop production and to minimise losses due to vector infestation.  Additional studies could add to knowledge on insect-plant interactions under elevated CO 2 . Further studies with other insect vectors and beneficials are required to identify if the causal factors for the observed changes are conserved among different insect guilds. With regard to subsequent virus transmission by vectors, studying the causes of a delay in symptom expression after CMV infection and the decrease in CMV transmission under indirect eCO 2 conditions may benefit the overall knowledge on the interactions among pathogens, host plants and vectors and help predict potential outbreaks.

Methods
Plants and CO 2 growth chambers. Pepper seeds (Capsicum annum L.) cv. 'California Wonder' (D. T. Brown & Co Ltd., South Windsor, Australia) were sown in plastic pots filled with potting mix containing slow release fertilizer, additional trace elements, iron and lime. At two true-leaf stage, pots were placed inside growth chambers (internal dimensions 1.2 × 0.7 × 1.5 m, 1260 L) with five 400W high pressure sodium lights (Lucagrow GE Lighting, Richmond, Australia) and four 77W halogen lights (Osram Pty. Ltd., Sydney, Australia) at two different CO 2 regimes, aCO 2 (400 ppm) and eCO 2 (650 ppm) (CLIMATRON-1260, Thermoline Scientific, Smithfield, Australia). Growing conditions were 24 °C, 70% RH, 16-hour photoperiod and 1000 μ mol m −2 s −1 light intensity at canopy level. Water in trays was maintained at a similar level between treatments and plants were rotated within the chambers to ensure there was no positional influence. CO 2 regime was switched between chambers once a week to ensure there was no potential chamber effect.

Aphid population.
A clonal source population of M. persicae was established from a virus free female naturally infesting marshmallow (Malva parviflora L.) at Grains Innovation Park (Horsham, Australia). Aphids were continuously reared on pepper plants in a chamber at 22 °C, 70% RH, 16-hour photoperiod and 800 μ mol m −2 s −1 light intensity. Individuals were synchronised prior to assays to ensure they were the same age.
Aphid life history. Peppers were previously grown under aCO 2 (400 ppm) and eCO 2 (650 ppm) for four weeks since two-leaf stage. A single wingless M. persicae adult was placed in a clip-cage on the adaxial side of the youngest fully developed leaf of each pepper and allowed to produce nymphs for 24 hours. Surplus nymphs were then removed leaving one nymph per plant, which was monitored until adulthood. Offspring were counted by removing nymphs daily for 10 days after the onset of reproduction. Duration of the four nymphal instars, pre-reproductive period (d), effective fecundity (offspring for a period equal to the pre-reproductive period (Md)), offspring for 10 days, mean generation time (Td = d/0.738), intrinsic rate of natural increase (r m = 0.738*(log e Md)/d) and mean relative growth rate (RGR = r m /0.86) were calculated (n = 25).

Feeding behaviour by Electrical Penetration Graphs. Myzus persicae probing and feeding behav-
iour was monitored by EPG, a tool for determining the activities of the aphid's stylets, including probing, salivation into sieve elements and passive uptake of phloem in real time 40,57 . Recordings were conducted inside a Faraday cage to avoid noise and interference for eight hours on peppers previously grown under aCO 2 (400 ppm) and eCO 2 (650 ppm) for four weeks from the two-leaf stage. Each recording was conducted using a new plant and aphid. Recordings were made simultaneously over an eight-week period (n = 42 for aCO 2 and n = 40 for eCO 2 ). Young adult aphids were immobilized using a vacuum device 57 . A thin gold wire (12.5 μ m diameter, 3 cm length) was attached to the dorsum of the insect with a small droplet of water-based silver glue (EPG-Systems, Wageningen, The Netherlands) using an entomological pin. The gold wire was glued to a copper wire (0.2 m diameter) attached to a brass pin, which was inserted into the input connector of the first-stage amplifier. The output electrode was a copper post (10 cm × 2 mm), which was inserted into the pot. Recordings were performed using a Giga-8 DC amplifier with a 1 Giga Ω input resistance (EPG-Systems, Wageningen, The Netherlands). EPG output was set to 50x gain and data was acquired at 100 Hz using a DATAQ Di700 A/D data acquisition USB device card (Dataq Instruments, Ohio, USA). Data was analysed with Stylet + a software (EPG-Systems, Wageningen, The Netherlands). All behavioural variables were then processed using the MS Excel Workbook for automatic EPG data calculation 58 .
Selected variables were compared between aCO 2 and eCO 2 59 : PPW, proportion of individuals that produced a specific waveform type; NWEI, number of waveform events per insect, that is the sum of the number of events of a particular waveform divided by the total number of insects under each treatment; WDI, waveform duration per insect, that is the sum of durations of each event of a particular wave-form made by each individual insect that produced that waveform divided by the number of insects that performed that particular waveform under each treatment; and WDE, waveform duration per event, that is the sum of the duration of the events for a particular waveform divided by the total number of events of that particular waveform under each treatment.
Pepper growth and physiology. Peppers grown under aCO 2 and eCO 2 were monitored for height, number of leaves and leaf chlorophyll content before insect introduction. Chlorophyll was measured in SPAD (Soil Plant Analytical Development) units, commonly used as an indirect indicator of nitrogen foliar content (Chlorophyll meter SPAD-502Plus, Konica Minolta, Japan). This device measures leaf transmittance at two wavelengths, 660 and 940 nm. SPAD readings were taken on every leaf and averaged per plant.
Canopy temperature was recorded by infrared thermal imaging (ThermaCAM P40, Flir Systems Pty. Ltd., Notting Hill, Australia) at the end of the fitness experiment. Thermal imaging was taken within five minutes after plants were removed from the chambers and placed in a climate controlled room (matching chamber temperatures) and against a contrast temperature background. Images were processed, and minimum, maximum and average temperature values were recorded for each treatment. Using box area tool, an equal sized rectangle was overlaid over each treatment, which covered around 50% of the plant canopy (ThermaCAM Researcher Professional 2.8 software, Flir Systems Pty. Ltd., Notting Hill, Australia).
Plants were harvested and leaf area recorded (Portable area meter LI-300C, LI-COR Biosciences, Nebraska, USA). Leaves and stems were dried at 65 °C for 48 hours (Dehydrating oven TD-150FTS, Thermoline Scientific, Smithfield, Australia). Specific leaf area (SLA = leaf area/leaf dry weight) was calculated. Leaves were finely ground (< 0.5 mm) (Tissuelyser MM300, Quiagen Retsch, Haan, Germany). Total carbon (C) and nitrogen (N) concentration of leaf tissue was determined by the Dumas combustion method using a CHN analyser (CHN 2000, LECO, St. Joseph, USA) at the University of Melbourne, Creswick. C:N ratio was calculated by dividing the concentration of C by the concentration of N for each leaf sample. CMV source plants. Seedlings were infected with CMV strain M6, subgroup IA at two-true leaf stage using a non-persistent virus type protocol. Fifteen M. persicae with one-hour starvation were allowed to feed for five minutes (acquisition access period) on a CMV positive leaf and were transferred to the seedlings for 24 hours (inoculation access period). These virus sources were then maintained either under aCO 2 (427 ppm) or eCO 2 (612 ppm) conditions for four weeks until symptom development inside climate chambers with 24:20 °C (D:N), 70% RH and 14-hour photoperiod. After four weeks, these plants were used as CMV inoculum for the transmission experiments.

Aphid ability to transmit CMV and systemic infection.
CMV transmission experiments. Two types of experiments were designed based on the moment of CO 2 exposure of receptor plants: 1) an experiment where receptor plants were exposed to the CO 2 regimes, either aCO 2 or eCO 2 , straight after virus inoculation by aphids, with the objective of studying the direct effect of CO 2 on the development of the viral disease, and b) an experiment where receptor plants had been previously grown for two weeks under each of the two CO 2 regimes before aphid introduction, to study the effect of the indirect acclimation of receptor plants on CMV transmission. In both experiments, plants providing CMV inoculum were the virus sources obtained in section 2.6.1, grown either under aCO 2 or eCO 2 .
CMV inoculation of receptor plants for both experiments was done with the same procedure mentioned above, placing five M. persicae which had previously fed on the CMV source plants for five minutes onto each receptor plant (n = 56). After 24 hours, receptor plants were sprayed with imidacloprid (Confidor ® 20 LS, Bayer CropScience, Valencia, Spain) and maintained either under aCO 2 or eCO 2 conditions inside climate chambers with 24:20 °C (D:N), 70% RH and 14-hour photoperiod for four weeks until visual inspection of symptoms. Statistical analysis. Data was transformed with either √(x + 0.5), x 2 , Ln(x + 1) or 2*arcsin√x if needed to reduce heteroscedasticity. Differences in parameters of aphid life history, EPG recordings and plant physiology between CO 2 treatments were assessed with Student t-test (p ≤ 0.05) using IBM Statistics SPSS 21.0 software (SPSS Inc.). When data did not follow the ANOVA assumptions, a non-parametric Mann-Whitney U-test (p ≤ 0.05) was performed. Daily offspring for 10 days over the fitness assay was assessed with GLM repeated measures analysis (p ≤ 0.05) (SPSS Inc.). CMV transmission was compared by a χ 2 goodness of fit test (p ≤ 0.05) to check if the observed frequency distribution was related to the expected frequency distribution using Statview 4.01 software (Abacus Concepts Inc.).