Feeding Behavior Comparison of Bean Bugs, Riptortus pedestris and Halyomorpha halys on Different Soybean Cultivars

Simple Summary This study compared the feeding behavior of Riptortus pedestris and Halyomorpha halys on six soybean cultivars. In addition, the damage caused by R. pedestris and H. halys was field-surveyed. The non-penetration waveform was the shortest in Pungsannamul and the longest in Daepung-2ho. In the case of feeding waveforms of the xylem and phloem, it was the longest in Pungsannamul and the shortest in Daepung-2ho, respectively. When six soybean cultivars were planted in two fields, and the seed damage caused by hemipterans was investigated; in both fields, the proportions of damage type B and damage type C were the highest in Pungsannamul and the lowest in Daepung-2ho. The above results will be used as basic information in the control of hemipteran insect pests. Abstract Riptortus pedestris (Fabricius) and Halyomorpha halys (Stål) are the major pests that feed on soybean pods, seeds, and fruits. Higher populations and damage occur during the soybean maturity stages (podding to harvest). To compare the feeding behavior of R. pedestris and H. halys, we used the six most cultivated cultivars (Daepung-2ho, Daechan, Pungsannamul, Daewon, Seonpung, and Seoritae) in Korea using the electropenetrography (EPG) technique. Both R. pedestris and H. halys, the NP (non-penetration), a non-probing waveform, was the shortest in the Pungsannamul (298 and 268 min) and the longest in the Daepung-2ho (334 and 339 min), respectively. The feeding waveforms Pb (phloem feeding: E1-Salivation and E2-Sap feeding) and G (xylem feeding) were the longest in Pungsannamul and the shortest in Daepung-2ho. In addition, as a result of investigating the damage rate by planting six cultivars of beans in the field, as expected, the proportions of damage types B and C were highest in Pungsannamul and lowest in Daepung-2ho. These results reveal that both bug species ingest xylem sap from leaflets and stems using a salivary sheath strategy to acquire water and nutrients from soybean pods/seeds with cell-rupture tactics. This study provides perceptive information to understand the feeding behavior, field occurrence, and damage patterns of R. pedestris and H. halys, which may have key implications for the management of hemipteran pests by determining the specificity and susceptibility of host plants.


EPG Recordings
The probing and feeding behaviors of R. pedestris and H. halys were recorded using the DC-EPG Systems with 1 GΩ of resistance (Wageningen, Güeldres, The Netherlands) by Tjallingii [31,37]. Second instar nymphs (4 hours starved) of both species were used in the experiment. Gold wire (AlfaAesar, Waltham, MA, USA) with a diameter of 50 µm and a length of 3-4 cm was attached to the center of the chest and dorsal side of each insect using a dissecting microscope (Olympus Co., Tokyo, Japan). For attachment, conductive electroconductive silver paste (P-100, JP) was used as an adhesive. Insects with gold wires were then connected to the Giga-8 DC EPG amplifier (EPG Systems, Wageningen, The Netherlands). They were then placed into bean stems (14 days old) of each soybean cultivar, and the output voltage was recorded for 6 hours. The voltage changed due to the difference in electrical resistance resulting from the stylet penetration movement when the stylet of an insect was inserted into the plant. Stink bugs were tested individually, using a new adult for each replicate. The recording was repeated 49-71 times for each soybean cultivar. All experiments were performed in a Faraday cage surrounded by copper mesh to minimize noise disturbing electrical signal recording.

Analysis of EPG Waveforms
The total duration of EPG waveforms for each soybean cultivar was measured. EPG signals were acquired and analyzed using Stylet+ software for Windows (EPG Systems). EPG variables were processed using the EPG-Excel Data Workbook developed by Sarriá et al. (2009) [38]. The electrical signals and their correlations with stink bug behavior were scored based on the categories described by Tjallingii (1990): non-probing (stylets external to the plant, NP), xylem ingestion (waveform G), and phloem activities (pooling together waveforms E1 and E2 reflecting salivation into sieve elements and phloem ingestion, respectively).
To compare the probing behavior of R. pedestris and H. halys on soybean leaves, a selected set of EPG variables was calculated as follows: the total waveform duration for each insect (the total duration of a waveform summed over all occurrences of the waveform for each insect) and the mean duration of waveform events for each insect (the total waveform duration divided by the number of waveform events for each insect). Then, the data of the total duration of the waveforms were used to calculate the mean percentage of time spent on each stylet activity (NP, G, E1, and E2) per soybean cultivar.

Field Experiment
On 30 May 2021, two fields (CJ-1 and CJ-2) of soybean cultivation were established on a farm located at Chungbuk National University, Cheongju-si, Chungbuk province, South Korea. Six soybean cultivars (Daepung-2ho, Daechan, Pungsannamul, Daewon, Seonpung, and Seoritae) recommended by the Korea Rural Development Administration (Jeonju, South Korea) were sown. The test plots were arranged in a randomized block design with 6 treatments and 3 replications. Two beans were sown in one row at an interval of 60 × 15 cm in each test plot. To check whether they were being damaged by the stink bug, the level of stink bug occurrence was investigated at 7-day intervals. The survey was conducted from August 31 to October 19 (grain filling period), the date when the stink bug moves to the soybean field. We walked slowly through the soybean fields, "searching and netting" the insect pests. Succulent insects other than stink bugs were not found; only a few moth larvae were found, but they were removed.

Damage Measurement and Analysis
The seed damage rate was determined by the soybean development stage classification method [39] and the Criteria for Research and Analysis of Agricultural Science and Technology [40]. Harvested soybeans were peeled off all pods, and the damage to the seeds was visually inspected. Seed damage was reported according to the classification criteria developed by Jung et al. [41] (Table 1): type A: normal seeds without feeding marks; type B: seeds that are normal in shape but show feeding marks on the surface; type C: seeds that are deformed in size or shape; and type D: seeds that are immature and have little shape. The number of pods and seeds harvested for each cultivar was investigated. The ratio was also analyzed for each type. Table 1. Classification of injury types of soybean seeds caused by hemipteran bugs.

Type Injury Phenomena
A seed Normal shape, non-injury B seed Normal shape but distinct injury marks C seed Deformed shape or diseased seed D seed Immature and undeveloped seed

Data Analysis
The two sting bugs were studied independently. First, the raw data of R. pedestris and H. halys were checked for normality and homogeneity of variance using Shapiro-Wilk W and were transformed with sqrt (x + 1) and ln (x + 1) if needed. According to a previous report by Vázquez et al. (2020) [42], most EPG variables did not follow a normal distribution; therefore, they were compared by Student's t-test for parametric variables at a 0.05 significance level using SPSS V26.0 Statistics software (IBM ® ) for R. pedestris data. Additionally, the comparisons between the proportions of individuals that produced a given type of wave were analyzed using Tukey's HSD (p < 0.05) [43]. To determine the preference of soybean cultivars for stink bug probing and feeding behaviors, nonparametric analyses were performed using Tukey's test. In addition, the seed damage rate by stink bugs was also analyzed using Tukey's HSD (p < 0.05).

The EPG Waveforms of R. pedestris and H. halys on Six Soybean Cultivars
The changes in the feeding behavior of the R. pedestris and H. halys were analyzed as waveforms (voltage changes due to the differences in electrical resistance) using the EPG system. Figure 1a shows the entire waveform measured 1 hour after the stink bug was placed on the host plant. A close analysis of the entire waveform shows the NP (non-penetration) waveform, as shown in Figure 1b. NP is a non-probe waveform for which the stylet and the host surface are completely separated. It is the waveform shown when the stink bug is resting before or after eating its host or when it does not prefer its plant. G (Xylem phase) is the waveform when the stink bug feeds on the xylem of the plant ( Figure 1c). The phloem phase (when the stink bug feeds on the host phloem) was analyzed by dividing it into E1 (Salivation), which secretes saliva, and E2 (Sap feeding), which sucks the phloem (Figure 1d).
After recognizing each waveform, we compared and analyzed the feeding behavior of two species of stink bugs on six soybean cultivars ( Table 2) for 6 hours. The response variable values were the total waveform duration (min) of NP, G, E1, and E2. For both stink bugs, the time spent feeding for 6 hours in the six soybean cultivars was the longest for the NP without stylet insertion, and the time spent for phloem feeding (E1, E2) was shorter than for xylem feeding (G). First, the results revealed that the duration of the NP waveforms of R. pedestris (332.47 ± 10.8 min) was longer than that of H. halys (303.83 ± 40.2 min) only in the Daechan cultivar (p = 0.0095). The G waveform did not show any significance between the R. pedestris and H. halys in any of the six soybean cultivars. The duration of the E1 waveform of R. pedestris was longer in Daepung-2ho (p < 0.0001), Daechan (p = 0.0273), Seonpung (p < 0.0001), and Seoritae (p < 0.0001) than that of H. halys. In the E2 waveform, the time of R. pedestris was significantly longer than that of H. halys in Daepung-2ho (9.84 ± 8.9 min) (p < 0.0001) and Seoritae (16.82 ± 13.1 min) (p < 0.0001). However, H. halys showed the E2 waveform for 0 min in Daepung-2ho and Seoritae, unlike R. pedestris. After recognizing each waveform, we compared and analyzed the feeding behavior of two species of stink bugs on six soybean cultivars ( Table 2) for 6 hours. The response variable values were the total waveform duration (min) of NP, G, E1, and E2. For both stink bugs, the time spent feeding for 6 hours in the six soybean cultivars was the longest for the NP without stylet insertion, and the time spent for phloem feeding (E1, E2) was shorter than for xylem feeding (G). First, the results revealed that the duration of the NP waveforms of R. pedestris (332.47 ± 10.8 min) was longer than that of H. halys (303.83 ± 40.2 min) only in the Daechan cultivar (p = 0.0095). The G waveform did not show any significance between the R. pedestris and H. halys in any of the six soybean cultivars. The duration of the E1 waveform of R. pedestris was longer in Daepung-2ho (p < 0.0001), Daechan (p = 0.0273), Seonpung (p < 0.0001), and Seoritae (p < 0.0001) than that of H. halys. In the E2 waveform, the time of R. pedestris was significantly longer than that of H. halys in Daepung-2ho (9.84 ± 8.9 min) (p < 0.0001) and Seoritae (16.82 ± 13.1 min) (p < 0.0001). However, H. halys showed the E2 waveform for 0 min in Daepung-2ho and Seoritae, unlike R. pedestris.
The percentage of time that sting bugs spent conducting each probing activity is represented in Table 3. When R. pedestris was placed on Pungsannamul, the percentage of the total time spent on non-probing events (np) was shorter (82.71%) when compared with that on Daepung-2ho (92.97%) (p = 0.0001). Additionally, the percentage of time spent on phloem ingestion (E2) was significantly (p = 0.001) lower on Daepung-2ho (2.73%) than on Pungsannamul (7.91%). In other words, this result indicates that R. pedestris prefers Pungsannamul to Daepung-2ho. No differences were found between treatments for salivation in phloem sieve elements (E1) (p = 0.147). In H. halys, NP was the shortest (74.55%) in Pungsannamul and the longest (94.23%) in Daepung-2ho (p = 0.0001), similar to that of the R. pedestris. Additionally, E2 was the longest (11.2%) in Pungsannamul and 0% in Daepoong-2ho and Seoritae (p = 0.023). These results indicate that H. halys does not prefer Daepung-2ho and Seoritae cultivars.  Means followed by the same letter (lowercase), in the same row, do not differ significantly (p > 0.05) using Student's t-test. Highlighted in bold are those variables that showed striking significant differences between insect species.

Occurrence Density of Stink Bugs in Soybean Fields
First, to analyze the occurrence density of stink bugs in a soybean field, we planted six soybean cultivars (Daepung-2ho, Daechan, Pungsannamul, Daewon, Seonpung, and Seoritae) in fields A and B of Cheongju, Chungbuk province. The distribution of the stink bugs in the soybean fields was investigated for approximately one month (31 Aug-19 Oct) (Figure 2). To survey, we walked slowly through the soybean fields (CJ-1 and CJ-2), "searching and netting" the insect pests. Succulent insects other than stink bugs were not found; only a few moth larvae were found, but they were removed. As a result, six species (R. pedestris, H. halys, Nezara antennata, Homalogonia obtusa, Dolycoris baccarum, and Acanthocoris sordiduswere) were mainly caught, and the number was the highest in September. In both fields, the dominant species was R. pedestris

Seed Damage Assessment
The number of seeds compared to the yield of soybean pods ranged from 1 (Table 4). Soybean seed damage types (types B and C and combined (B + C)) were found to vary according to soybean cultivars. The number of type-B and combined (B + C) seeds was significantly different among soybean cultivars in field CJ-1 (F = 19.56, df = 5, p < 0.0001 for type B, and F = 11.60, df = 5, p = 0.0003 for combined type); however, type-C seeds were not different among soybean cultivars (F = 2.48, df = 5, p = 0.092). The least damaged soybean cultivar was Daepung-2ho (17.9%), while the most damaged soybean cultivar was Pungsannamul (53.5%). The pattern of the seed damage rate of soybean cultivars in field CJ-2 was the same as that in field CJ-1. The numbers of type-B and combined (B + C) seeds were significantly different among soybean cultivars (F = 11.93, df = 5, p = 0.0003 for type B, and F = 9.01, df = 5, p = 0.0009 for combined type); however, type-C seeds were not different among soybean cultivars (F = 1.75, df = 5, p = 0.199). The least damage was recorded for Daepung-2ho (22.2%), and the most damage was recorded for Pungsannamul (45.7%). Therefore, Daepung-2ho can reduce damage by stink bugs and is more resistant than other soybean cultivars.

Seed Damage Assessment
The number of seeds compared to the yield of soybean pods ranged from 1.  (Table 4). Soybean seed damage types (types B and C and combined (B+C)) were found to vary according to soybean cultivars. The number of type-B and combined (B+C) seeds was significantly different among soybean cultivars in field CJ-1 (F = 19.56, df = 5, p < 0.0001 for type B, and F = 11.60, df = 5, p = 0.0003 for combined type); however, type-C seeds were not different among soybean cultivars (F = 2.48, df = 5, p = 0.092). The least damaged soybean cultivar was Daepung-2ho (17.9%), while the most damaged soybean cultivar was Pungsannamul (53.5%). The pattern of the seed damage rate of soybean cultivars in field CJ-2 was the same as that in field CJ-1. The numbers of type-B and combined (B+C) seeds were significantly different among soybean cultivars (F = 11.93, df = 5, p = 0.0003 for type B, and F = 9.01, df = 5, p = 0.0009 for combined type); however, type-C seeds were not different among soybean cultivars (F = 1.75, df = 5, p = 0.199). The least damage was recorded for Daepung-2ho (22.2%), and the most damage was recorded for Pungsannamul (45.7%). Therefore, Daepung-2ho can reduce damage by stink bugs and is more resistant than other soybean cultivars.  Values represent the mean ± SD. Means followed by the same letters in a column are not significantly different among soybean cultivars (ANOVA, Tukey's HSD test, p < 0.05).

Discussion
In this study, we compared the feeding behavior of R. pedestris and H. halys on six soybean cultivars. The EPG analysis revealed that the NP waveform was the shortest in Pungsannamul and the longest in Daepung-2ho. In the case of feeding waveforms of the xylem and phloem, it was the longest in Pungsannamul and the shortest in Daepung-2ho. From these results, it seems that the two stink bugs prefer Daepung-2ho less than the Pungsannamul cultivar. By examining the total feeding pattern recording time and non-probe waveform recording time for each soybean cultivar, it is possible to know the bean cultivars preferred by the stink bug [44,45]. In general, insects of Hemiptera use two feeding strategies [26]. One is the salivary sheath strategy, which gels saliva around the stylet to form an envelope. The other is cell rupture [30,46]. It secretes saliva and destroys cells while moving in and out after inserting stylet into host tissue [30,47,48]. In addition, cell rupture is divided into the salivation phase and the feeding phase. Stink bugs use only one strategy or both, depending on the species [49]. The two species of stink bug used in this experiment showed a pattern of dividing the salivation phase and the feeding phase, so they seem to use cell rupture as a feeding strategy. In the process of EPG analysis, the sucking of xylem by stink bugs was performed only once, and no more than two suckings were observed during the study period. The feeding pattern we recorded in this study is in line with that of Lucini et al. [35], who also found a similar pattern in Piezodorus guildinii. Sucking the xylem is considered a strategy used by Aphidoidea and Psyllidae to avoid dehydration and maintain water balance [31,50,51]. Similarly, in both stink bugs, sucking the xylem seems to be a means to avoid dehydration and maintain water balance, so it is thought that they did not suck the xylem more than two times. Interestingly, no waveforms of E1 and E2 were observed in Daepung-2ho and Seoritae. In the results of the development and reproduction experiment by soybean cultivar, the development period was slow and the survival rate was low in Daepung-2ho and Seoritae (data not shown). In addition, analyzing the total duration of G, E1, and E2 waveform, except for Daepung-2ho and Seoritae, they feed more xylem but feed similarly to sap. In the case of feeding only the xylem, it seems that the stink bug did not eat the phloem because it is not a preferred soybean cultivar. It is possible that the longer the experiment time, the higher the sap-feeding rate. However, since the EPG experiment was conducted with highly active insects, they may have been under a lot of stress. When comparing the feeding waveform of stink bugs with the feeding waveform of cotton aphids (Aphis gossypii Glover, Hemiptera: Aphididae) [52], we recorded a threefold higher NP time for stink bugs than for cotton aphids. In our previous study, the feeding pattern of the NP waveform in cotton aphids is usually 70-80 min, the G waveform is about 20 min, the E1 waveform is about 25 min, and the E2 waveform is about 200 min. Interestingly, resistant cotton aphids were found to be more active in finding suitable feeding sites on insecticide-treated hosts than susceptible cotton aphids. Therefore, the feeding patterns of stink bugs and cotton aphids were different. Our study also shows that there are differences among stink bugs. In addition, Daepung-2ho and Seoritae can be resistant cultivars against H. halys. Overall, for both stink bugs, the feeding time was the shortest in Daepung-2ho and the longest in Pungsannamul. In the other four soybean cultivars, the feeding patterns were different according to the stink bug species. Host plant suitability and feeding preferences of insects are determined by various factors such as taste, smell, leaf color, leaf hair, and leaf thickness.
The dominant species of the stink bug on six soybean cultivars were R. pedestris, followed by H. halys and N. antennata. The number was the highest in September. In addition, the least damaged soybean cultivar was Daepung-2ho (17.9%, 22.2 %), while the most damaged soybean cultivar was Pungsannamul (53.5%, 45.7%) in fields A and B, respectively. The number of seeds compared to the yield of soybean pods, Daepung-2ho showed the highest value. Daepung-2ho is currently a cultivar of the highest yield (345 Kg/10a) in South Korea. This cultivar is also tolerant to lodging, fire blight, and seed shattering. Therefore, Daepung-2ho has a good yield and can reduce damage to stink bugs. EPG waveform analysis also confirmed that Pungsannamul had a high preference for stink bug feeding. However, the longer feeding time of stink bugs and the higher damage rate of soybeans did not necessarily indicate the preference of stink bugs. In the EPG experiment, it may be that the nutritional level of soybean could not be satisfied, so it needs to be fed for a long time. In the field experiment, it may be that soybeans are less resistant due to being more responsive to feeding or because of long-time feeding. The findings of this study are similar to those of previous studies showing that stink bug damage varies according to soybean cultivar [10,53]. Oh et al. [54], however, reported that Pungsannamul has less stink bug pod damage but higher yield reduction, which contradicts our estimates. This discrepancy may be due to the availability of other palatable plant hosts and crops/fruits near soybean fields planted with Pungsannamul and biotic environmental variability. A previous study indicated that plant species and genotype and abiotic factors (humidity, temperature, and photoperiod) are significant factors influencing the cultivar of volatiles released by host plants [55]. In this experiment, the stink bugs showed significantly higher preferences toward the Pungsannamul soybean cultivar and appeared to prefer the Pungsannamul bean the most, findings that contradict those of Choi et al. [6], who suggested that Pungsannamul would have insect-resistant characteristics when considering the growth and weight change of stink bugs. This discrepancy might be influenced by the passage of time or environmental changes.
In conclusion, both R. pedestris and H. halys did not prefer Daepung-2ho as food, among the six soybean cultivars, and there was little damage from them. Further studies need to determine the effects on the growth, survival rate, and oviposition numbers of R. pedestris and H. halys. Therefore, we are currently conducting research in our laboratory.