Different Plant Viruses Induce Changes in Feeding Behavior of Specialist and Generalist Aphids on Common Bean That Are Likely to Enhance Virus Transmission

Abstract

Bean common mosaic virus (BCMV), bean common mosaic necrosis virus (BCMNV), and cucumber mosaic virus (CMV) cause serious epidemics in common bean (Phaseolus vulgaris), a vital food security crop in many low-to-medium income countries, particularly in Sub-Saharan Africa. Aphids transmit these viruses "non-persistently," i.e., virions attach loosely to the insects' stylets. Viruses may manipulate aphid-host interactions to enhance transmission. We used direct observation and electrical penetration graph measurements to see if the three viruses induced similar or distinct changes in feeding behaviors of two aphid species, Aphis fabae and Myzus persicae. Both aphids vector BCMV, BCMNV, and CMV but A. fabae is a legume specialist (the dominant species in bean fields) while M. persicae is a generalist that feeds on and transmits viruses to diverse plant hosts. Aphids of both species commenced probing epidermal cells (behavior optimal for virus acquisition and inoculation) sooner on virus-infected plants than on mock-inoculated plants. Infection with CMV was especially disruptive of phloem feeding by the bean specialist aphid A. fabae. A. fabae also experienced mechanical stylet difficulty when feeding on virus-infected plants, and this was also exacerbated for M. persicae. Overall, feeding on virus-infected host plants by specialist and generalist aphids was affected in different ways but all three viruses induced similar effects on each aphid type. Specifically, non-specialist (M. persicae) aphids encountered increased stylet difficulties on plants infected with BCMV, BCMNV, or CMV, whereas specialist aphids (A. fabae) showed decreased phloem ingestion on infected plants. Probing and stylet pathway activity (which facilitate virus transmission) were not decreased by any of the viruses for either of the aphid species, except in the case of A. fabae on CMV-infected bean, where these activities were increased. Overall, these virus-induced changes in host-aphid interactions are likely to enhance non-persistent virus transmission, and data from this work will be useful in epidemiological modeling of non-persistent vectoring of viruses by aphids.

Keywords: aphid; cucumovirus; electrical penetration graph; legume; non-persistent transmission; potyvirus.

Figure 1

Electrical penetration graph (EPG) monitoring showed that feeding behavior of specialist and non-specialist aphids was modified on virus-infected bean plants. For both Aphis fabae (A) and Myzus persicae (B), the combined time engaged in epidermal cell probing and pathway activity accounted for most of the activity recorded over 8 h and this was not markedly altered by the virus infection status of plants, except in the case of A. fabae placed on plants infected with cucumber mosaic virus (CMV), where these activities were increased (A). Phloem ingestion accounted for a substantial proportion of A. fabae activity on mock-inoculated plants (A) but this was not the case for M. persicae where mechanical stylet difficulties and drinking from the xylem occurred, indicating that common bean is a poor host for M. persicae (B), which is a generalist aphid, rather than a legume specialist like A. fabae. Phloem ingestion by M. persicae was not markedly affected on plants infected with bean common mosaic virus (BCMV), bean common mosaic necrosis virus (BCMNV), or CMV but mechanical stylet difficulties increased, especially on CMV-infected plants (B) and for A. fabae phloem ingestion declined on virus-infected plants (A). Data was collated from EPG recordings of 240 aphids comprising 15 aphids per treatment for each aphid species, i.e., n (M. persicae) = 120, and n (A. fabae) = 120.

Figure 2

Aphids placed on virus-infected plants initiate probing behavior sooner than on mock-inoculated common bean plants. Direct observations showed that both the legume specialist Aphis fabae and generalist Myzus persicae began probing on virus-infected plants sooner than on mock-inoculated plants. Bar charts show the mean times from placement to first probe for 40 aphids per treatment group to begin probing. All experiments were done using single aphids placed on separate plants (virus-infected or mock-inoculated plants at 10 days post-inoculation/mock inoculation). Error bars represent the standard error of the mean. The decreased times-to-probe for aphids on virus-infected plants compared to mock-inoculated plants were statistically significant in all cases [survival analysis, Kaplan-Meier: p < 0.0001 (A. fabae) and p = 0.00012 (M. persicae)]. Pairwise comparisons between aphid behavior on mock-inoculated and infected plants showed significant differences [Peto-Peto: BCMNV vs. Mock p = 0.00018, BCMNV vs. Mock p = 4.88.10⁻⁵, CMV vs. Mock p = 2.45.10⁻⁷ (A. fabae)] and [Peto-Peto: BCMNV vs. Mock p = 0.005, BCMNV vs. Mock p = 0.00093, CMV vs. Mock p = 0.00026 (M. persicae)]. Mock, mock-inoculated plants; CMV, cucumber mosaic virus-infected plants; BCMV, bean common mosaic virus-infected plants, and BCMNV, bean common mosaic necrosis virus-infected plants.

Figure 3

Analysis of differences in aphid salivation into the phloem and phloem sap ingestion for Aphis fabae and Myzus persicae on mock-inoculated and virus-infected plants. EPG data was analyzed for changes in salivation into the phloem (E1 waveform: A, B) and subsequent phloem sap ingestion (E2 waveform: C, D) by the legume specialist A. fabae (left panels) and the generalist M. persicae (right panels) on common bean plants that had been mock-inoculated or infected with bean common mosaic virus (BCMV), bean common mosaic necrosis virus (BCMNV), or cucumber mosaic virus (CMV). A. fabae placed on CMV-infected plants spent significantly longer (p = 0.0161) salivating into the phloem than on plants that were mock-inoculated (A), and this was consistent with a corresponding significant difference for the mean duration of incidents of sap ingestion (p = 0.0236) (B). The total time A. fabae spent phloem feeding was significantly reduced on CMV-infected plants (p = 0.00059) (C) as was the mean duration of phloem ingestion bouts (p = 8.0.10⁻⁵) (D). A. fabae on plants infected with BCMV showed a similar trend of decreased phloem activity, although this was not statistically significant (A–D). On mock-inoculated plants M. persicae and A. fabae spent similar periods of time salivating into the phloem (A, B). However, on mock-inoculated plants, the phloem ingestion bouts of M. persicae averaged 5 min compared to 70 min for A. fabae (D), which was reflected in the overall times spent feeding from the phloem by the two species (C). The only statistically significant effect seen for M. persicae was a decrease in overall time spent in salivation on BCMNV-infected plants (p = 0.0427) (A). Asterisks denote values significantly different from mock treatments (Dunnett p < 0.05). The error bars represent the standard error of the mean.

Figure 4

On common bean Myzus persicae is less likely than Aphis fabae to ingest phloem sap and its phloem feeding activity is less affected by plant infection status. (A) Using EPG data the likelihood (as percentage odds) of an aphid transitioning to sustained phloem sap ingestion (the potential E2 index, PEI) was determined by beta GLM with the multcomp package in R (Hothorn et al., 2009) for A. fabae and M. persicae placed on mock-inoculated plants or plants infected with bean common mosaic virus (BCMV), bean common mosaic necrosis virus (BCMNV), or cucumber mosaic virus (CMV). Sustained phloem feeding in this case is indicated by periods of E2 waveform activity of >10 min. (A) As shown by the PEI, A. fabae was markedly more likely to ingest phloem sap from mock-inoculated plants than M. persicae (which appears to find common bean an unsuitable host) but the likelihood of phloem feeding was diminished for A. fabae on virus-infected plants and was significantly decreased on CMV-infected plants (p = 0.040). Virus infection status had no effect on the already low likelihood of phloem feeding by M. persicae. (B) On plants infected with BCMV and CMV but not with BCMNV, the proportion of time spent by A. fabae in salivation into the phloem (waveform E1) relative to sap ingestion (waveform E2) increased markedly. The proportion of time M. persicae spent salivating into or feeding from the phloem appeared unaffected by plant infection status. (C) EPG recordings of A. fabae on mock-inoculated plants showed sustained phloem feeding began by 4-h post-placement while A. fabae placed on plants infected with BCMV, BCMNV, or CMV took on average 5.8, 5.7, and 6.9 h, respectively. In all panels error bars represent standard error of the mean. The asterisk denotes a significant difference in feeding behavior between insects on virus-treated and mock-inoculated plants (Dunnett test, p < 0.05: see Supplementary Table 3 for analyses).

Figure 5

The rate at which Aphis fabae engage in repeated phloem sap ingestion is decreased on virus-infected plants, whereas Myzus persicae is reluctant to ingest phloem sap regardless of plant infection status. The EPG E1 (saliva secretion into the phloem) and E2 (ingestion of phloem sap) waveform data for both A. fabae and M. persicae were subjected to survival analysis, which is an application of an actuarial method for assessing the probability that a specific event will occur during a given period of observation time. Curves representing the probability that aphids had re-entered the phloem and were producing waveforms related to phloem feeding (i.e., survival curves, the probability that the aphid did not feed as a function of time, S(t), denoted as 1-S(t) on the Y-axes) were computed for each waveform. (A). On mock-inoculated plants, the likelihood of aphids initiating saliva secretion (E1 activity, left panel) was similar for the legume specialist A. fabae and for the generalist M. persicae. However, the probability of transitioning back to phloem sap ingestion (E2 activity, right panel) was significantly lower for M. persicae than for A. fabae. When feeding on virus-infected plants (infected with either BCMV, BCMNV, or CMV), there was no significant decrease in the probability of either A. fabae (B) or M. persicae (C) initiating saliva secretion into the phloem (E1 activity, left panels). However, A. fabae was significantly less likely to initiate phloem sap ingestion (E2 activity, right panel of B) on virus-infected plants than on mock-inoculated plants. The infection status of plants made no significant difference to the already very low probability with which M. persicae re-initiates phloem sap ingestion (E2 activity, right panel of B). The survivorship functions, S(t), were calculated as Kaplan-Meier estimates and Kaplan-Meier p-values are shown (see Supplementary Information Excel File 3 for pairwise curve comparisons).

Figure 6

Mechanical stylet difficulty during aphid feeding is more frequently encountered by aphids placed on virus-infected plants. EPG data was analyzed to reveal: (A) the overall duration of time over the recording period in which aphids (the legume specialist, Aphis fabae, or the generalist, Myzus persicae) encountered stylet mechanical feeding difficulties (Waveform F: Table 1 ); (B) the mean duration of instances of feeding difficulty; and (C) the frequency of these feeding difficulty incidents over the recording period on mock-inoculated plants or plants infected with BCMNV, BCMV, or CMV. In contrast to their feeding on mock-inoculated plants A. fabae encountered more feeding difficulties on virus-infected plants (A, left panel) that were, in addition, more protracted (B) and more frequent (C). The non-specialist aphid M. persicae experienced feeding difficulties even on mock-inoculated common bean plants and there was no overall increase in time spent experiencing feeding difficulties or in the frequency of these events on virus-infected plants as was the case for A. fabae (A, C). However, M. persicae experienced more prolonged incidents of feeding difficulty on plants infected with BCMV and CMV (B, right panel). Asterisks denote significant differences in stylet difficulty between insects on virus-infected and mock-inoculated plants (Dunnett test, p < 0.05: see Supplementary Table 1 for analyses). Error bars represent the standard error of the mean.