Manipulation of Aphid Behavior by a Persistent Plant Virus

This study demonstrates that a persistent plant virus can manipulate aphid behavior. This manipulation is in stark contrast to previously described effects of acute viruses on their hosts that facilitate their transmission. This study demonstrates a positive relationship between Pepper cryptic virus 1 and jalapeño pepper (Capsicum annuum) plants wherein the virus protects the plants from the vector of acute viruses and reduces aphid herbivory. This work reveals an important implication of persistent plant viruses for pest and pathogen management in agriculture.

ABSTRACT

Plants are frequently infected with cytoplasmic RNA viruses that persist for many generations through nearly 100% vertical transmission without producing any symptoms. Movement between plant cells and horizontal transmission have not been observed with these viruses; instead, they are distributed to all host cells through host cell division. Jalapeño peppers (Capsicum annuum) are all infected with Pepper cryptic virus 1 (PCV-1; family Partitiviridae). We compared the effect of odor cues from PCV-1-infected (J+) and virus-free (J−) jalapeño peppers on the aphid Myzus persicae, a common vector of acute plant viruses. Pairwise preference experiments showed a stark contrast to insect-plant interactions in acute virus infections—that is, the virus-infected plants deterred aphids. The acute plant virus Cucumber mosaic virus (CMV) manipulates its host's volatile emissions to attract aphid vectors and facilitate its transmission. We inoculated J+ and J− plants with CMV. Volatiles of J+ and J− CMV-infected plants were more attractive to aphids than those of J+ and J− mock-inoculated plants. However, in pairwise preference experiments with J+ CMV- and J− CMV-infected plants, aphids preferred the J− CMV volatile blend. Aphid reproduction on J+ and J− plants was measured as an indicator of the effect of PCV-1 on host quality for aphids. Aphid reproduction on J+ plants was more than 2-fold lower than that on J− plants.

IMPORTANCE This study demonstrates that a persistent plant virus can manipulate aphid behavior. This manipulation is in stark contrast to previously described effects of acute viruses on their hosts that facilitate their transmission. This study demonstrates a positive relationship between Pepper cryptic virus 1 and jalapeño pepper (Capsicum annuum) plants wherein the virus protects the plants from the vector of acute viruses and reduces aphid herbivory. This work reveals an important implication of persistent plant viruses for pest and pathogen management in agriculture.

INTRODUCTION

Based on their lifestyles, plant viruses are divided into two main groups: acute viruses and persistent viruses (1). Most well-studied plant viruses are acute viruses that cause disease in domestic plant hosts. In plants, infections with acute viruses can be resolved by recovery, death, or conversion to chronic infection. In contrast, persistent plant viruses replicate in their hosts for many generations, perhaps thousands of years. They do not cause obvious disease. The name “persistent” comes from their lifestyle in the host, which differs from that of viruses that are persistently transmitted by a vector (1–3). The persistent viruses are distributed to all plant cells through host cell division. They are not able to move between plant cells, and even grafting cannot transmit these viruses. There is no evidence for their horizontal transmission by a vector, whereas they are vertically transmitted to seeds at rates close to 100% (1, 4, 5). Unlike persistent animal viruses, persistent plant viruses cannot convert to an acute lifestyle because they lack the genes for movement.

Vectors play a crucial role in the transmission of acute plant viruses because plants are largely immobile (6). Plant-feeding insects are the most common plant virus vectors, and among insects, aphids are the most common vectors of agriculturally important plant viruses (7). Aphids probe any plant they land on to determine if it is a suitable food source (8), and during this probing process, many viruses can be transmitted. Because of the dependence of plant viruses on vectors for their transmission, viruses evolved ways to manipulate some host and vector features to facilitate transmission.

Studies on the impact of acute plant viruses on insect behavior began in 1951 and found that Aphis fabae population growth was higher on virus-infected sugar beets than on healthy plants (9). Studies on two viruses from the family Luteoviridae (Barley yellow dwarf virus [BYDV] and Potato leafroll virus) and their aphid vectors showed that virus infection changed aphid development time and reproduction (10). In addition, virus-induced changes in hosts of both luteoviruses encouraged aphid feeding on infected plants, which led to rapid population growth of aphids and subsequent dispersal of viruliferous aphids (11–14). These aphid behavioral changes were consistent with the transmission mode of luteoviruses by aphids. Luteoviruses are acquired by vectors feeding for hours to days on the phloem tissues of infected plants (6, 15, 16). Both viruses affected volatile organic compounds (VOCs) of their host plants, and vectors were responsive to host VOCs (12, 13, 17). The nonviruliferous aphid vector of BYDV preferred infected host plants, while viruliferous aphids preferred uninfected hosts; these differences promote pathogen spread. It has been demonstrated by in vitro feeding that aphid behavioral changes are mediated directly by virus acquisition (18). In soybean plants, the aphid Rhopalosiphum maidis remains on healthy plants longer than on plants infected with Soybean mosaic virus. Leaving infected plants shortly after probing increases the probability that an aphid will find another plant and transmit the virus (19).

Cucumber mosaic virus (CMV) is a widespread plant pathogen with more than 1,200 host species (20). Aphid vectors can acquire CMV during brief probing of infected plant cells and transmit the virus effectively if the vector disperses rapidly to a new host plant (6, 16, 21). CMV-infected squash plants, despite their reduced size, have elevated levels of VOCs that attract aphid vectors to the infected plants (22). CMV infection initially attracts aphids but then reduces the quality and palatability of plants so that aphids disperse quickly after virus acquisition (22). Salicylic acid- and jasmonic acid-mediated host defense responses are not involved in diminishing plant quality for aphids; instead, CMV infection induces changes in the concentrations of carbohydrates and free amino acids in the plant tissue that cause a reduction in plant quality for aphids. All of these virus-induced changes in the plants favor CMV transmission: they attract aphids to feed on the infected plant, which then disperse quickly to efficiently transmit the virus (23).

It was shown recently that although CMV infection induced quantitative and qualitative changes to the VOC emissions of tobacco plants, it did not change aphid preference for infected or uninfected tobacco plants (24). Inoculation of a squash isolate of CMV onto another host, pepper, did not result in the pattern observed on the native squash host. In addition, the virus-mediated changes in the new host were not consistent with CMV-vector interactions to aid its transmission (23). Hence, virus-host-vector interactions appear to be adaptive.

The most common persistent viruses in plants, and indeed the most common plant viruses overall in wild plants, belong to the family Partitiviridae (25, 26). The partitiviruses were first reported as cryptic viruses (5). The Partitiviridae family includes viruses from diverse hosts: plants, fungi, and oomycetes (27, 28). Phylogenetic analysis of partitivirus RNA-dependent RNA polymerase implies that these viruses have been transmitted between plant and fungi (1). Plants cannot be cured of these viruses using various treatments (29), although occasionally a virus-free plant will develop due to a rare lack of vertical transmission. Pepper cryptic virus 1 (PCV-1) belongs to the genus Deltapartitivirus and consists of two linear, monocistronic double-stranded RNAs (dsRNAs) that are encapsidated in separate isometric particles (30–32). PCV-1 was first reported from the Capsicum annuum cultivar Jalapeño M, and as with all persistent plant viruses, there is no evidence for its transmission by any vector, mechanical inoculation, or grafting, whereas ovules or pollen transmit the virus to over 98% of the next generation (30, 32).

Persistent viruses are very common in crop plants; hence, it seems likely that they were selected during domestication, and we hypothesize that these viruses provide some advantages to the plants. While there are many studies on the effects of acute viruses on plant volatiles that impact insect behavior, no studies have looked at the effects of persistent viruses. Here, we used three isogenic lines of jalapeño plants with and without PCV-1 to assess their attraction and plant quality traits for aphids.

RESULTS

Aphid preference for volatiles of plants with or without PCV-1.

To examine the response of Myzus persicae to VOC cues, we performed three pairwise preference experiments in which aphids were exposed to odors of (I) two empty cylinders, (II) an empty cylinder or a PCV-1-infected jalapeño plant (J+), and (III) an empty cylinder or a virus-free jalapeño plant (J−) (Table 1; Fig. 1a), without any visual cues. In experiment I, there was no significant difference between choosing either of two empty cylinders, while in experiments II and III, aphids selected cylinders containing plants. Figure 2a shows the mean percentages of aphids arrested in collection cages for each experiment, and the pairs involved in each choice test are shown bidirectionally. The 95% confidence intervals for the probability of selecting empty cylinder were 0.46 to 0.57 for experiment I, 0.23 to 0.34 for experiment II, and 0.22 to 0.33 for experiment III (Fig. 2b); only in the preference test with both empty cylinders (experiment I, empty versus empty) was 0.5 (i.e., no preference) within the 95% confidence intervals. Hence, aphid choice was not random, and they were responding to volatile cues. In experiments IV-1, IV-2, and IV-3, three isogenic lines of jalapeño with and without PCV-1 were used to discover how the virus affected aphid behavior. Similar patterns were observed in all three lines, and aphids demonstrated their preference for volatiles of virus-free plants (J−). The 95% confidence intervals for selecting J+ over J− plants were 0.36 to 0.49 (IV-1, J1+ versus J1−), 0.37 to 0.49 (IV-2, J2+ versus J2−), and 0.31 to 0.41 (IV-3, J3+ versus J3−), as shown in Fig. 2b. Aphids showed discrimination against volatiles of PCV-1-infected plants, and 0.5 was not contained in any of the 95% confidence intervals. Using three different jalapeño lines verified that aphid behavior is affected by manipulation of plant volatiles by PCV-1 infection.

TABLE 1.

Aphid preference experiment arrangements

Expt no.	Arrangement
I	Empty vs empty
II	Empty vs J+
III	Empty vs J−
IV-1	J1+ vs J1−
IV-2	J2+ vs J2−
IV-3	J3+ vs J3−
V	J− mock vs J− CMV
VI	J− mock vs J+ CMV
VII	J+ mock vs J+ CMV
VIII	J+ mock vs J− CMV
IX	J+ CMV vs J− CMV

FIG 1.

Experimental design for aphid preference and reproduction tests. (a) Schematic of experimental set up for aphid pairwise preference for plant volatiles. Wingless, nonviruliferous 3- to 5-day-old aphids were placed in the middle of the aphid box, which was covered to eliminate any visual cues affecting aphid behavior. Aphids could make a choice between two collection cages that were connected to plant cylinders containing treatments. Each collection cage had one layer of fine mesh to trap aphids in the cage, and the number of aphids in each collecting cage was recorded 45 min after releasing aphids. (b) Aphid reproduction was tested on 6- to 8-leaf-old jalapeño plants. Eight adult wingless aphids were confined in a small clip cage and placed on a caged plant (3 clip cages per plant). These adult aphids were removed after 24 h, and offspring were left in each clip cage. Offspring molted to adults and produced new offspring. Aphid reproduction in each clip cage was calculated by dividing the number of offspring produced by the number of adult aphids on day 14.

FIG 2.

Aphid preferences for volatiles of plants with or without PCV-1. The following control pairwise preference tests (I to III) were performed to test if aphids were responding to volatile cues: (I) empty versus empty; (II) empty versus J+; and (III) empty versus J−. In experiments IV-1 to IV-3, aphid preferences for volatiles of three different isogenic lines of jalapeño were tested, as follows: (IV-1) J1+ versus J1−; (IV-2) J2+ versus J2−; and (IV-3) J3+ versus J3−. (a) Numbers represent the mean percentages of aphids in collection cages at the end of each treatment, and pairs involved in each choice test are shown horizontally. In arrangement I, there is no significant difference between choosing either of two empty cylinders, while in arrangements II and III, cylinders containing plants were selected. In experiments IV-1, IV-2, and IV-3, three isogenic lines of J+ and J− were used to test aphid attraction to PCV-1-infected or virus-free plants. A similar pattern was observed in all three lines, and aphids demonstrated their preference for virus-free plants (J−). (b) The 95% confidence intervals for the probability of selecting the empty cylinder in arrangements I to III were 0.46 to 0.57 (I, empty versus empty); 0.23 to 0.34 (II, empty versus J+); and 0.22 to 0.33 (III, empty versus J−). The 95% confidence intervals for selecting J+ over J− plants were 0.36 to 0.49 (IV-1, J1+ versus J1−); 0.37 to 0.49 (IV-2, J2+ versus J2−); and 0.31 to 0.41 (IV-3, J3+ versus J3−).

Aphid preference for volatiles of plants with or without PCV-1 and CMV.

Several studies have shown that CMV manipulates its host by inducing elevated levels of volatile emissions that attract aphid vectors for its transmission (22, 23). In order to explore the effect of PCV-1 on aphid attraction to CMV-infected plants, we inoculated J+ and J− plants with CMV and performed aphid pairwise preference studies (Fig. 1a) in experiments V to IX. We compared aphid attraction to volatiles of mock-inoculated virus-free plants (J− mock) versus CMV-infected plants (J− CMV and J+ CMV) in experiments V and VI (Table 1). In both experiments, aphids preferred volatiles from CMV-infected plants (Fig. 3a). In arrangements VII and VIII, pairwise comparisons were between mock-inoculated PCV-1-infected plants (J+ mock) and CMV-infected plants (J+ CMV and J− CMV). A similar pattern was observed, and aphids significantly preferred the volatiles from CMV-infected plants (Fig. 3a). The 95% confidence intervals for the probability of selecting mock-inoculated plants in arrangements V to VIII ranged from 0.27 to 0.41 (Fig. 3b). However, in pairwise comparisons of aphid responses to volatile emissions from CMV-infected J+ and J− plants (IX, J+ CMV versus J− CMV), aphids preferentially selected the volatiles of J− CMV plants (Fig. 3a). The 95% confidence interval for choosing J+ CMV was 0.32 to 0.41 (IX, J+ CMV versus J− CMV) (Fig. 3b).

FIG 3.

Aphid attraction to volatiles of mock- or CMV-infected J+ and J− plants. Pairwise preference experiments were as follows: (V) J− mock versus J− CMV; (VI) J− mock versus J+ CMV; (VII) J+ mock versus J+ CMV; (VIII) J+ M versus J− CMV; and (IX) J+ CMV versus J− CMV. (a) Numbers represent the mean percentages of aphids in collection cages at the end of each treatment, and pairs involved in each choice test are shown horizontally. In arrangements V and VI, J− mock volatiles were less attractive to aphids than were those of J−CMV and J+CMV plants. In arrangements VII and VIII, J+ M volatiles were also less appealing to aphids than those of J+ CMV and J− CMV plants. In arrangement IX, aphids preferentially chose volatiles of J− CMV over those of J+ CMV. (b) The 95% confidence intervals for the probability of selecting J+ and J− mock in arrangements V to VIII were 0.32 to 0.0.41 (V, J− mock versus J− CMV); 0.30 to 0.40 (VI, J− mock versus J+ CMV); 0.32 to 0.41 (VII, J+ mock versus J+ CMV); and 0.27 to 0.36 (VIII, J+ mock versus J− CMV). The 95% confidence interval for choosing J+ CMV was 0.32 to 0.0.41 (IX, J+ CMV versus J− CMV).

Aphid reproduction on plants with or without PCV-1.

Virus infection results in biochemical and physical changes in the host, and aphid reproduction was used as an indicator of the impact of PCV-1 on plant quality for aphids. We compared the reproduction of nonviruliferous aphids that developed from day one to adulthood on virus-free and PCV-1-infected plants over 14 days. Since it took 8 to 9 days to complete the first reproduction cycle, only one generation of offspring was born within this period. Average reproduction on J1+ (74 adult aphids) was 7, whereas reproduction on J1− (71 adult aphids) was 13. For J2+ (69 adult aphids), reproduction was 9.5 compared to 21 for J2− (71 aphids). For J3+ (63 aphids), reproduction was 9.3 compared to 22 for J3− (61 aphids) (Fig. 4; Table 2). In all three lines, aphid reproduction on the PCV-1-infected plants (J+) was more than 2-fold lower than that on virus-free plants (J−). Statistical analysis indicated that the effect of the virus on aphid reproduction was significant (P value of <0.001). This showed the positive role of PCV-1 in protecting its host from increasing population growth of aphids. In line 1, aphid reproduction for both treatments (J1+ and J1−) was lower than in the other two lines. Although the conditions used were the same, the experiments were not all done at the same time, so comparisons are valid only within an experiment, although we cannot rule out distinct plant qualities in the different lines. In all three lines, we saw the same trend of higher reproduction on J− than on J+ plants. Thus, PCV-1 infection protects jalapeño plants from increasing aphid populations.

FIG 4.

Aphid reproduction on plants with and without PCV-1 tested on three isogenic lines of jalapeño plants. Reproduction was calculated by dividing the number of offspring by the number of adults on each clip cage after 14 days. Results are shown as averages ± standard error (SE) for 12 clip cages per treatment. Numbers of aphids and fecundity measures are shown in Table 2. P values: line 1, 2.485 × 10⁻⁶; line 2, 5.812 × 10⁻⁵; and line 3, 1.928 × 10⁻⁶. Statistical analysis indicated that the effect of the virus on aphid reproduction was significant (P value = 2.6 × 10⁻¹⁰). Asterisks indicate significant differences between control J+ and J− plants (***, P < 0.001).

TABLE 2.

Numbers of adults and offspring and aphid fecundity for three isogenic lines of jalapeño infected with PCV-1 (J+) and virus-free (J−) plants

Line	No. of aphids in:
	J+ plants			J− plants
	Adultsa	Offspringb	Fecundityc	Adultsa	Offspringb	Fecundityc
1	74	506	6.83	71	913	12.86
2	69	655	9.49	71	1,490	20.98
3	63	586	9.30	61	1,326	21.74

^aNumber of adults in clip cages 14 days after placement.

^bNumber of offspring found in clip cages. In 14 days, only one generation of aphids occurs.

^cFecundity is a measure of the ratio of offspring to adult aphids, as described in Materials and Methods.

DISCUSSION

Behavioral responses that impact insect dispersal are important because of their effect on viral disease spread. Several studies have shown that acute viruses manipulate their host volatiles so that infected plants are more enticing to vectors, in spite of the plants’ small size and the negative effects on apparent plant quality. Virus-induced changes are in favor of virus transmission, attracting aphids to feed on plants and acquire the virus for its dispersal (12, 13, 17, 22). These virus-induced changes in vector behavior are adaptive and vary among viruses, hosts, and vectors (23). Here, we showed that partitivirus infection also influences aphid behavior (Fig. 2, IV-1, IV-2, and IV-3), but this interaction is in stark contrast to insect-plant interactions in acute virus infection; that is, the presence of this partitivirus leads to deterrence of aphids and protection of plants from the vector of acute viruses, as well as from the damage of aphid herbivory itself. Based on these results, it seems that during thousands of years of virus replication in the plant host, the virus-plant interaction has evolved to be beneficial for the plants by discouraging the vector of acute viruses.

The acute plant virus CMV can manipulate its host’s volatile emissions to attract aphid vectors and facilitate its transmission, although these interactions are complex and dependent on specific relationships between virus strains and host plants. We inoculated J+ and J− plants with CMV; volatiles of J+ and J− CMV-infected plants were more attractive to aphids than those of J+ and J− mock-inoculated plants (Fig. 3, V, VI, VII, and VIII). This is consistent with other studies showing that the VOC emissions from CMV-infected plants are more attractive to aphids than those from uninfected plants (22, 23). PCV-1 did not counter the effect of CMV on the plant VOCs, perhaps because of a strong elevation of VOCs by CMV. However, in pairwise preference tests between J+ CMV- and J− CMV-infected plants, aphids preferred the J− CMV volatile blend (Fig. 3). Therefore, the presence of PCV-1 influenced aphid choice when both plants were CMV infected. This outcome verified our hypothesis that partitivirus infection decreases the likelihood of acute virus transmission, even in the presence of an acute virus.

Aphid reproduction on J+ and J− plants was measured as an indicator of the effect of PCV-1 on host quality for aphids. Aphid reproduction on J+ plants was more than 2-fold lower than that on J− plants (Fig. 4). There are several factors that can affect aphid reproduction. Turgor pressure can affect aphid feeding and therefore reproduction (33). However, in this study we minimized turgor pressure effects by watering plants regularly and similarly. Host plant quality during aphid growth and development (both nymph and adult stages) is another key factor of aphid reproduction (for a review, see reference 34). In poor host quality conditions, aphids resorbed their eggs for their own survival (35, 36). Nitrogen also plays an important role in plant quality and impacts aphid reproduction. Early in the growing season, phloem amino acid content is high, and aphid reproduction is higher than that when leaves mature and phloem amino acid levels drop (37–40).

Virus infections cause biochemical and physical changes in the host and consequently affect the reproduction of aphids. The populations of the aphids Schizaphis graminum, Rhopalosiphum padi, and Sitobion avenae increased on BYDV-infected plants in comparison with those on uninfected plants (11, 41, 42). Moreover, aphid reproduction was higher on wheat cultivars sensitive to BYDV than on a BYDV-tolerant wheat cultivar (42). Aphid population growth was correlated with the virus-vector relationship. The acquisition of BYDV requires continual aphid feeding on the phloem of infected plants. Thus, BYDV-induced changes in plants encourage sustained feeding and lead to aphid settling and rapid population growth, followed by dispersal of viruliferous aphids (11–14). Increased carbohydrate and amino acid contents of infected plants can lead to a higher reproduction rate in the vector. Total amino acid content was increased in spring wheat after infection with BYDV (43). Also, BYDV-infected plants had higher carbohydrate contents than healthy plants (44). On the other hand, CMV, which is transmitted in a nonpersistent manner, induced changes in the concentrations of carbohydrates and free amino acids in the plant tissue that lead to reduction of plant quality for aphids. When plant quality and palatability are reduced, aphids are dispersed quickly and efficiently transmit CMV to a healthy plant (6, 16, 21, 22).

The phytohormone ethylene is involved in the regulation of host defense responses against herbivores, necrotrophic pathogens, and nonpathogenic microbes (45, 46). In Turnip mosaic virus, the expression of nuclear inclusion a protease (NIa-Pro) is involved in decreasing host defenses and increasing vector reproduction. NIa-Pro is responsible for disruption of ethylene responses in plants and inhibition of aphid-induced callose production (47, 48). It is possible that PCV-1 infection directly or indirectly (by changing the plant microbiome) induces callose deposition, which is an important plant defense against aphid feeding, finally decreasing aphid reproduction on infected jalapeño plants.

Hence, while acute plant viruses manipulate the behavior of aphid vectors in their favor, enhancing their transmission, PCV-1, a persistent plant virus that is not horizontally transmitted, protects its plant host from aphids by reducing the attractiveness of the plant and the quality of the plant as an aphid host. This has the secondary effect of protecting the plants from aphid herbivory, which can also cause significant damage.

MATERIALS AND METHODS

Plants and insects.

Three isogenic lines of jalapeño peppers (Capsicum annuum) with and without PCV-1 (denoted as J1+/J−, J2+/J2−, and J3+/J3−, respectively) were obtained from Rodrigo Valverde (Department of Plant Pathology and Crop Physiology, Louisiana State University). Plants were sown in 10-cm plastic pots containing Sun-Gro horticulture soil and grown in an insect-free environmental room at 24°C under fluorescent light (a ratio of 16:8 for the light/dark photoperiod).

A colony of M. persicae was obtained from Gary Thompson (Penn State University) and maintained on turnips (Brassica rapa). To produce sufficient aphids for experiments, the aphid colony was transferred to fresh turnip plants, and colonies were maintained with a natural photoperiod at 25°C. Aphid colonies were reestablished by moving two aphid-infested leaves to a new turnip plant every 10 to 15 days.

CMV inoculation.

Two-week-old pepper plants were kept in the dark for 24 h and then mechanically inoculated using purified virus particles of Fny-CMV (49) diluted in 50 mM NaHPO₄ (pH 7.0). Leaves were dusted with carborundum to increase inoculation efficiency, followed by rubbing with acid-etched glass. Mock inoculation employed the same protocol, but with buffer only. Plants were used for aphid preference experiments 20 days after inoculation, when infected seedlings showed symptoms of virus infection.

Aphid preference for volatiles of plants with or without PCV-1.

Aphid preference tests of plant volatile cues were performed using the experimental setup shown in Fig. 1a. Two cylinders (13 × 28 cm) containing different treatments were connected to an aphid box (11 × 17 cm) via two aphid collection cages (2.5 × 4 cm). The aphid box was covered to eliminate any visual cues. Each collection cage had one layer of extrafine mesh at the bottom to allow aphids to respond to VOCs without any contact cues and to trap aphids in the cage.

For the pairwise preference tests, 50 wingless nonviruliferous 3- to 5-day-old aphids were collected using a fine brush and starved for 1 h, after which they were transferred to the middle of the aphid box (Fig. 1a). The number of aphids in each collection cage was recorded 45 min after releasing the aphids. The percentage of the aphids that made a choice was calculated for each treatment. Pairwise preference tests were performed in the following three different control arrangements: (I) empty versus empty, (II) J+ versus empty, and (III) J− versus empty. Aphid preferences for three isogenic lines of J+ and J− were tested: IV-1, J1+ versus J1−; IV-2, J2+ versus J2−; and IV-3, J3+ versus J3−. Plants used in these tests were matched for size and development stage (four- to six-leaf stage). The preference experiment for each plant pair was repeated at least twice, and each pairwise arrangement was done 12 times using four to six plants. All experiments were done at 25°C in daylight to ensure that plants released a consistent volatile blend.

Aphid preference for volatiles of plants with or without PCV-1 and CMV.

To investigate aphid preference for volatiles of jalapeño plants in the presence of CMV (an acute virus) and PCV-1 (a persistent virus), the same experimental setup as above was used (Fig. 1a). The pairwise preference tests were performed with the following five different plant arrangements: (V) J− mock versus J− CMV; (VI) J− mock versus J+ CMV; (VII) J+ mock versus J+ CMV; (VIII) J+ mock versus J− CMV; and (IX) J− CMV versus J+ CMV. All experimental conditions were as above.

Aphid reproduction on plants with or without PCV-1.

Experiments were done using 6- to 8-leaf-stage jalapeño plants in a pesticide-free greenhouse at 23 to 26°C supplemented with light for a 16-h day length. Eight adult wingless aphids were confined in a small clip cage (2.5 × 4 cm) and placed on a caged plant. Three clip cages were randomly attached to each plant, and for each line, four plants were used (Fig. 1b). The adult aphids were left to reproduce inside the cage for 24 h, and then adults were removed and offspring were maintained in clip cages to mature and reproduce. During 14 days, offspring molting to adults, reproducing, and producing new offspring were monitored in the clip cages. At day 14, the numbers of offspring and adults in each individual clip cage were recorded, and aphid reproduction was calculated by dividing the number of offspring produced by the number of adult aphids. This experiment was replicated 12 times for each jalapeño line, and in total, reproduction was measured for more than 60 aphids for each treatment.

$$\text{aphid reproduction}=\frac{\text{no. of offspring}}{\text{no. of adults}}$$

Statistical analysis.

Statistical analysis was performed using R version 3.3.3. A general linear model (GLM) with binomial error was used to analyze aphid preference data for all pairwise experiments, estimating the proportion of aphids that moved to the preferred option and 95% confidence intervals for each pairwise test. For statistical analysis of aphid reproduction, the lme4 package in R was used, and two models were defined. The null model for aphid reproduction included different plants and lines assumed as random effects. A second model included the presence of PCV-1 considered a fixed effect, with the same random effects as in the null model. A likelihood ratio test of the two models was performed (using analysis of variance [ANOVA] in R) to test the significance of the fixed effect.