Host plant use of a polyphagous mirid, Apolygus lucorum: Molecular evidence from migratory individuals

Abstract While the host plant use of insect herbivores is important for understanding their interactions and coevolution, field evidence of these preferences is limited for generalist species. Molecular diet analysis provides an effective option for gaining such information, but data from field‐sampled individuals are often greatly affected by the local composition of their host plants. The polyphagous mirid bug Apolygus lucorum (Meyer‐Dür) seasonally migrates across the Bohai Sea, and molecular analysis of migrant bugs collected on crop‐free islands can be used to estimate the host plant use of A. lucorum across the large area (northern China) from where these individuals come. In this study, the host plant use of A. lucorum adults was determined by identifying plant DNA using a three‐locus DNA barcode (rbcL, trnH‐psbA, and ITS) in the gut of migrant individuals collected on Beihuang Island. We successfully identified the host plant families of A. lucorum adults, and the results indicated that captured bugs fed on at least 17 plant families. In addition, gut analyses revealed that 35.9% of A. lucorum individuals fed on multiple host plants but that most individuals (64.1%) fed on only one plant species. Cotton, Gossypium hirsutum L., DNA was found in 35.8% of the A. lucorum bugs examined, which was much higher than the percentage of bugs in which other host plants were found. Our work provides a new understanding of multiple host plant use by A. lucorum under natural conditions, and these findings are available for developing effective management strategies against this polyphagous pest species.

1987; Schlein & Muller, 1995;Zhang et al., 2019). Generalist herbivores have a wide range of host plant species and rarely show specific adaptations to particular plants (Barros, Torres, Ruberson, & Oliveira, 2010;Franzke, Unsicker, Specht, Köhler, & Weisser, 2010;Hereward & Walter, 2012;Joern, 1979). However, not all the plant species found in the habitats of generalist herbivores can be utilized, and the diets of these herbivores, while diversified, are still somewhat selective (Ibanez et al., 2013). Direct observations of herbivory in the field are problematic in habitats that are difficult to access, such as the forest canopy or underground, and are also greatly limited by the ability of the researcher to correctly identify the species involved in the interactions. Since the observation of feeding behavior cannot produce a clear picture of a generalist herbivore's entire host plant range, a more accurate method for determining the feeding history and alternative (noncrop) host plants of generalist herbivores is needed.
DNA barcoding uses short DNA sequence markers for the taxonomic identification of species (Hebert, Penton, Burns, Janzen, & Hallwachs, 2004;, which can overcome the problems associated with more conventional methodologies, as it can enable rapid, sensitive, and accurate plant species identification by detecting host plant-specific DNA extracted from herbivorous insects (Traugott, Kamenova, Ruess, Seeber, & Plantegenest, 2013;Valentini, Pompanon, & Taberlet, 2009). For these reasons, this technique has attracted increasing attention in the past several years as a method for determining the dietary composition of herbivores (Erickson et al., 2017;García-Robledo et al., 2013;Heise et al., 2015;Jurado-Rivera, Vogler, Reid, Petitpierre, & Gomez-Zurita, 2009;Navarro, Jurado-Rivera, Gómez-Zurita, Lyal, & Vogler, 2010;Staudacher, Wallinger, Schallhart, & Traugott, 2011). In these studies, specific plant barcode regions (e.g., rbcL and trnL) were amplified and compared with known DNA sequences in GenBank using BLAST (Altschul, Gish, Miller, Myers, & Lipman, 1990), which could allow for the identification of unknown ingested host plant species (Jurado-Rivera et al., 2009;Navarro et al., 2010). Molecular markers have shown great potential for identifying the diets of insect herbivores at the taxonomic levels of family and genus (Jurado-Rivera et al., 2009;Navarro et al., 2010) and even at the species level (García-Robledo et al., 2013). In species-level identification, a comprehensive DNA sequence database of the target community is required, and improved DNA extraction techniques and multiple molecular markers will help increase the efficiency of species discrimination. For example, García-Robledo et al. (2013) accurately identified the dietary breadth of leaf-rolling beetles in a tropical rain forest in Costa Rica by three DNA barcode loci (i.e., rbcL, ITS2, and trnH-psbA). Hereward and Walter (2012) used a trnL-trnF fragment to identify the plant species fed on by the green mirid Creontiades dilutus in northeastern Australia and found that the mirid individuals frequently fed on more plants than the species from which they were collected. This DNA-based technique allows us to better understand the feeding activities of insect herbivores instead of needing to make direct feeding observations (Kiston et al., 2013;La Cadena, Papadopoulou, Maes, & Gómez-zurita, 2015;Wang, Bao, Zeng, Yang, & Lu, 2016). Moreover, as DNA barcoding techniques are less targeted, they can reduce the risk of overlooking the trophic relationships of generalist herbivores (Kishimoto-Yamada et al., 2013). Many unexpected trophic associations have been discovered with the application of molecular methods (Jurado-Rivera et al., 2009;La Cadena et al., 2015). Jurado-Rivera et al. (2009) (Lu, 2008;Lu, Wu, Jiang, et al., 2010). A. lucorum nymphs and adults feed on multiple vegetative and reproductive tissues of their host plants via piercing and sucking mouthparts (Jiang, Lu, & Zeng, 2015;Zhang, Lu, & Liang, 2013).
They use stylets to lacerate the plant cells while secreting a watery saliva (including a high diversity of digestive enzymes) into the ruptured cell and then ingest the resultant lacerated/macerated "soup" (Backus, Cline, Ellerseick, & Serrano, 2007). This feeding strategy usually leads to the necrosis and discoloration of plant tissue, the formation of bushy plants, the abscission of flower buds, and the distortion of mature fruits (Jiang et al., 2015;Shackel et al., 2005), which often greatly reduces yield and quality when the population of A. lucorum is large (Lu & Wu, 2008). Damage symptoms usually appear approximately one week after mirid bug feeding (Jiang et al., 2015;Zhang et al., 2013), and adults frequently move between different host plants (Pan, Lu, Wyckhuys, & Wu, 2013;Wang, Bao, Yang, Yang, & Lu, 2018). The relatively cryptic feeding habits and high mobility of this species make it difficult to precisely assess its host plant use with field population surveys. However, plant identification using plant DNA barcode loci and the well-studied plant-herbivore system allows us to accurately identify insect diets (Kress & Erickson, 2007;Li et al., 2011).
In molecular dietary analysis of herbivorous insects, the information on host plant use obtained from field-sampled individuals is likely to vary greatly among different sampling locations, which usually differ in host plant composition (Kishimoto-Yamada et al., 2013;Wang et al., 2016). Hence, the design of the sampling program is vital and plays an important role in lessening the possible overrepresentation of particular locally abundant hosts in data from field-collected insect individuals (e.g., Hereward, DeBarro, & Walter, 2013).
For adult A. lucorum, 10-day-old mated females showed a maximum flight distance of 111.4 km during a 24-hr period in flight mill assays, indicating that A. lucorum adults possess strong potential for longdistance flight (Lu, Wu, & Guo, 2007). An 11-year searchlight trapping and radar observation study on an isolated island (Beihuang) in the center of the Bohai Gulf found that A. lucorum, a migratory species, travels at least 40-60 km from land (Fu et al., 2014). As almost no crops are grown on Beihuang Island, it is an ideal site to collect In this study, we first collected migrant A. lucorum adults using light traps on the island of Beihuang, sequenced short stretches of plant-specific genes (i.e., rbcL, ITS, and trnH-psbA) from the gut contents of each A. lucorum adult, and then compared the resultant DNA sequences with GenBank sequences to confirm the host plant species.

| Insect DNA extraction
DNA was extracted from whole adult of A. lucorum following a previously described CTAB-based protocol (Wallinger et al., 2013).
Before DNA extraction, each adult was cleaned of plant material potentially adhering to its body surface following a modified method (Greenstone, Payton, Weber, & Simmons, 2014;Remén, Krüger, & Cassel-Lundhagen, 2010;Wallinger et al., 2013). Specifically, we placed each A. lucorum in 1 ml of 1%-1.5% sodium hypochlorite (Beijing Chemical Works) for 5 s and then rinsed it twice with molecular analysis-grade water (Wang, Bao, Wu, Yang, & Lu, 2017). To check for cross-sample contamination, two extraction-negative controls were included in each batch of 24 samples.

| PCR assays
Three plant DNA barcode loci (i.e., rbcL, ITS, and trnH-psbA) were sequenced for each sample to increase the recovery of intact sequences from potentially highly degraded plant DNA from insect gut contents (Kress & Erickson, 2007;Kress et al., 2009;Li et al., 2011).
The nucleotide sequences (5′ to 3′) of the primers are listed in Table   S1.

| Cloning and DNA sequencing
PCR products were purified with a gel extraction kit (Tiangen) and ligated into pGEM-T cloning vector (Promega).

| Identification of A. lucorum diets using molecular markers
Apolygus lucorum gut content DNA sequence identifications were performed using BLAST against GenBank using the default search parameters (Altschul et al., 1990). Each unknown DNA sequence from the gut contents was identified to the species level only when it was nearly completely consistent with the best hit of the query sequences (percent identity > 99%). In cases where top BLAST scores were equal for species from different genera within the same genus, we identified such interactions to the genus level. Identification of DNA sequences at the family level was similar to the method used for genus identification. Sequences from gut contents that did not match any of the plant DNA sequences in the DNA barcode library were scored as unidentified.

| Data analysis
Differences in the detected host plants of A. lucorum in different years and months were compared via two-factor nonrepetitive variance analysis via the GLM (proc glm) process step in SAS 9.30 software (SAS Inc). Before the analysis, the detection rate data were subjected to inverse sine transformation to improve normality.

| Inferred plant families
Two hundred and seventy-eight high-quality sequences were detected among the 156 A. lucorum individuals, including 29 rbcL sequences, 137 ITS sequences, and 112trnH-psbA sequences, which were discriminated into 33 OTUs that were assigned to at least 17 families (Table 1)

| Feeding activity during different time periods
Our analyses of the gut contents of adult individuals revealed that 35.9% of the oversea migratory A. lucorum were detected with the plant DNA from multiple hosts (n = 156), while the rest were found with that of only one host plant (  Figure 2). In 2015, a total of 7, 9, and 8 families of host plant DNA were detected in A. lucorum adults in June, July, and August, respectively.

| D ISCUSS I ON
In this study, we identified host plant families, genera, and species used by the oversea-migrating adults of A. lucorum using DNA barcoding. We found that A. lucorum adults fed on a wide range of host plants, including at least 17 families. We also documented the simultaneous use of multiple host species by A. lucorum individuals.
The rapidly evolving sequences of the chloroplast genome region make them appropriate DNA barcodes for identifying plants (Valentini et al., 2009). The Consortium for the Barcode of Life (CBOL) working group has proposed the rbcL + matK combination as the best plant barcode because of its universality, sequence quality, and species discrimination (CBOL Plant Working Group, 2009). However, the success rate of plant DNA amplification in these mirid bugs was relatively low for the chloroplast rbcL intron (599 bp) in this study, probably due to degradation by extraoral digestion that reduced the number of larger DNA fragments remaining in the gut. Deagle, Eveson, and Jarman (2006) found that the number of template molecules of degraded DNA declined rapidly with increasing fragment size during the digestion period. Hereward and Walter (2012) suggested that the chloroplast trnL intron was not successfully amplified from target plant DNA in the green mirid bug C. dilutus because of degradation by extraoral digestion. A. lucorum resembles C. dilutus in feeding behavior, performing extraoral digestion and lacerating and macerating plant cells with a stylet-probing movement and watery salivary discharge (Backus et al., 2007). In this study, we therefore selected  (Wang, Bao, Yang, Xu, & Yang, 2017;Wang et al., 2018). A previous study found that A. lucorum adults were most active from 16:00 to 24:00 in crop fields (Geng, Lu, & Yang, 2012). Therefore, we speculated that the time of A. lucorum adult flight from host plants was at dusk. As we collected A. lucorum adults from the light traps at 6:00 every morning, DNA analysis took place approximately 6-12 hr after the last time of plant feeding of A. lucorum before it began its migration over the sea.
The number of template molecules of the degraded DNA declined rapidly with increasing fragment size during the digestion period (Deagle et al., 2006;Hereward & Walter, 2012;Wallinger et al., 2013;Wang, 2017;Wang et al., 2018). Hence, we targeted short DNA fragments of multiple-copy genes to increase the probability of successful DNA detection .  (Kajtoch, Kubisz, Heise, Mazur, & Babik, 2015). These samples were immediately preserved in the field in ethanol to minimize DNA degradation. Our study demonstrates that it is possible to determine the host use and ultimately dietary breadth of migratory insects from herbivore tissue by DNA-based plant identification.
In this study, a significant proportion of A. lucorum individuals were found to have fed on multiple host plants. Fragments of the length that we amplified from the mirid gut contents can evidently be detected only within 48 hr postingestion (Fournier et al., 2008;Gariepy et al., 2007;Hoogendoorn & Heimpel, 2001;Muilenburg et al., 2008). Therefore, individual mirid adults frequently move between hosts. Similarly, A. lucorum individuals moved frequently between cotton and mungbean fields when these crops were planted nearby (Wang, 2017). Moreover, Creontiades dilutus (Hemiptera: Miridae) often feeds on several host plant species other than the one it has been collected from, based on molecular gut content analyses (Hereward, 2012;Hereward & Walter, 2012), indicating potential movement and the utilization of multiple host plants by this mirid bug. Nezara viridula (Hemiptera: Pentatomidae) showed similar feeding habits, moving from one plant species to another during the feeding process (Todd, 1989), while host switching enhanced its survival and reproduction (Velasco & Walter, 1993). For A. lucorum, Pan, Liu, and Lu (2018) found that the combination of feeding nymphs on maize and adults on green bean resulted in the fastest population growth rate in the laboratory, indicating that host food switching between stages was beneficial. This potential benefit warrants further investigation under natural conditions to determine whether the ecological significance of A. lucorum movement resembles that of N. viridula.
As a polyphagous species, A. lucorum has been recorded on at least 288 different host species in 54 different families (Jiang et al., 2015). Based on our analyses of the gut contents of individual adults,  (Lu, Wu, Wyckhuys, & Guo, 2010;Pan et al., 2013)). Our results are consistent with previous findings.
In summary, we identified the diets of migratory mirid bugs by multiple DNA barcode loci at the plant family, genus, and species levels. Our findings suggest that A. lucorum individuals feed on multiple host plants. This is a significant step in studying the feeding ecology of A. lucorum under natural conditions and developing landscapelevel pest management strategies for this mirid bug.

ACK N OWLED G M ENTS
The authors thank the graduate trainees at Changdao and Langfang Experimental Station, Institute of Plant Protection of the Chinese Academy of Agricultural Sciences, for assistance with sample collection and laboratory work during the study period.

CO N FLI C T O F I NTE R E S T
None declared.

AUTH O R CO NTR I B UTI O N S
YHL and YZY conceived the idea and designed the methodology; XWF collected the samples; QW, WFB, and QZ performed the laboratory work; QW analyzed the data; and QW and YHL wrote the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
Sequence files have been deposited in the Dryad data repository (https ://doi.org/10.5061/dryad.9cp7219).