Identification of host plant use of adults of a long–distance migratory insect, Mythimna separata

Adults of many insect species often become contaminated with pollen grains when feeding. Identification of plant hosts for M. separata moths could increase our understanding of their geographic origin and the coevolution of M. separata moths and their host plants. However, identifying the diet of noctuid moths using traditional direct observation is limited by their nocturnal and flight habits. In this study, we used core barcode markers and pollen morphology to identify pollen species. We found pollen from 13 plant species belonging to nine families on trapped M. separata moths, mainly from Angiosperm, Dicotyledoneae. Pollen was found on 14.4% and 12.3% of females and males, respectively, and the amount of pollen transported varied with the body part, with the most pollen on the proboscis. We were able to determine from this that the moths visited woody plants more than herbaceous plants, but not significantly so, and that they carried more pollen earlier in the migration season. In this study, we clarified the species and frequencies of pollen deposition on M. separata moths. These findings improve our understanding of the coevolution of the moths and their host plants. Identification of plant hosts for adult moths provides a new means of studying noctuid moth-host plant interactions, and informs the development of more efficient management practices for M. separata.


Introduction
Plant and insects have co-evolved all over the world into patterns of interactions that are often mutually beneficial [1,2]. Plants depend on insects for pollination and simultaneously provide pollen, nectar and other plant exudates for insects [1]. Understanding the interactions between plants and insects is critical to interpreting ecological and evolutionary phenomena, and the first step is to understand the range of host plants used by specific herbivorous insects.
Lepidoptera (butterflies and moths) are probably the most diverse group of phytophagous insects [3], encompassing more than 155,000 recognized species [4,5]. Noctuidae is the largest family of Lepidoptera, containing over 40,000 currently described species [6]. Noctuids are a prominent element of terrestrial ecosystems, functioning as herbivores and pollinators, as well a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 as one of the most damaging groups of pests to agriculture. They feed on plants as larvae and on nectar as adults [7]. Although direct laboratory and field observations can provide insight into the larval diets [8][9][10][11], ascertaining the host plants of adults using these methods is subject to several compounding limitations, including their nocturnal and flight habits in the field. Therefore a new method is needed, using a combination of DNA barcoding and pollen morphology.
Adults often become contaminated with pollen during the process of nutritional supplementation, and this pollen can be identified to determine the insect's host plants [1]. Although conventional pollen morphology is widely used for plant identification, in some cases, the pollen of closely related plants has very similar morphology [12][13]. Alternatively, DNA bar-coding has performed well for the determination of the dietary composition of various organisms [14][15][16][17]. For this purpose, a partial region of plant DNA sequence is amplified and compared to a reference database such as GenBank [14][15]. Previous studies have demonstrated that such molecular markers can potentially identify plants in the diet of insect herbivores to the family or genus level [14][15]. However, an unambiguous, reliable identification of the host plants to the species level using molecular markers is not yet possible [17]. Therefore, in the present study, DNA barcoding was used along with pollen morphology to identify the pollen species.
Mythimna separata is an important migratory pest in eastern Asia that undertakes a seasonal migration in China [18][19][20]. However, its geographic origin has still not been confirmed. The identification of pollen found on an insect's exterior provides evidence of its origins, as some plants grow only in certain ecological zones or geographic locations. The objective of this study was to determine the host relationship and geographic origin of M. separata moths by identifying the pollen that adhered to them during their long-distance migration. We also investigated the quantities of pollen attached to various parts of the bodies of M. separata moths.

Ethics statement
No specific permits were required for the collection of Mythimna separata.

Collection of Mythimna separata moths
Mythimna separata moths were collected using light traps. Collection was conducted every night from April to October, 2013-2015 on the 2.5 km 2 island of Beihuang (BH, 38˚24 0 N; 1205 5 0 E) in the Bohai Strait, which is a major pathway for the seasonal migration of many insect species. BH is approximately 40 km from mainland China to the north and approximately 60 km from land to the south [21][22][23]. A vertical-pointing searchlight trap (model DK.Z.J1000B/ t, 65.2 cm in diameter, 70.6 cm in height, and 30 in spread angle), equipped with a 1,000-W metal halide-lamp (model JLZ1000BT; Shanghai Yaming Lighting Co. Ltd., Shanghai, China) was suspended from the top of a house (at 500 m elevation). Twenty M. separata (or all individuals if the total captured was <20) were removed from the bags of the nylon net (60 mesh) each morning and killed by crushing their thorax, and then each moth was transferred to a 2-ml tube and stored in a freezer (-20˚C) for later microscopic examination. PA) to make a preliminary identification of the attached pollen. As in Bryant et al. [24], pollen was sought on only the proboscis and antenna, which were the parts that most frequently carried the pollen. To prevent contamination, a piece of paper towel (9 × 9 cm) was placed on the microscope stage and changed with each new sample, and the forceps were cleaned after each sample. The pollen grains were isolated from the proboscis or antennae, mounted on aluminum stubs, coated with gold in a sputter coater, and immediately photographed using a Hitachi S-4800 SEM (Hitachi High-Technologies Co., Tokyo, Japan).

Analysis of DNA from the pollen grains
Single pollen grains were isolated from the aluminum stubs using a micropipette with a 3-5 μm diameter tip made by the Micropipette Puller (micropipette puller, Sutter Instruments, USA) and soaked in 1 μL of lysis solution (0.1 M NaOH)(Beijing Chemical Reagent Co. Ltd., Beijing, China) plus 2% Tween-20 (Beijing Chemical Reagent Co. Ltd.) in individual PCR tubes. The tubes were spun briefly to precipitate the pollen grains, after which 5 μL of mineral oil was added to prevent a volume loss of the solution during lysis at 95˚C. The samples were spun briefly (1 min at 1,000 g), and after a 17 min and 30 s preheating treatment (95˚C), the pollen grains were lysed. Equimolar aliquots of 1 μL TE buffer were added to neutralize the samples, which were spun briefly [25]. The resulting solution was used as the template for Whole Genome Amplification (WGA) using an illustra GenomiPhi V2 DNA Amplification Kit (GE Healthcare UK Ltd.). The WGA products of single pollen grains were used to amplify the plant plastid DNA.
Following PCR, the rbcL amplicons were purified using a Gel Extraction Kit (Tiangen, Beijing, China) and cloned using the pGEM-T East Vector System (Promega, Madison, WI). Each DNA-containing plasmid was isolated from cultured E. coli cells by an alkaline iniprep method. Successful insertion was verified by PCR using M 13 forward (5 0 -GTTTTCCCAGT CACGAC-3 0 ), and reverse primers (5 0 -CAGGAAACAGCTATGAC-3 0 ) and Sanger sequencing was done by the Biomed company (Beijing, China).

Pollen identification and characteristics of the pollen source plants
Identification of the rbcL sequences was performed individually through similarity BLAST searches against GenBank [28]. An unknown sequence was considered a member of the best hit of the query sequences when it was completely consistent with them, and the unknown species sequence was considered to be the same genus as indicated by its top hits if there were differences between the sequences. The species were corrected according to their morphological features. The pollen's morphological features were identified using the published SEM images in the atlas of pollen flora of China and pollen flora of China woody plants by SEM [29][30].
Pollen grains that could be classified to genus or species level were used to identify the source plants of pollen.

Data analysis
One-way ANOVA and Tukey's Honestly Significant Difference (Tukey's HSD) were used to evaluate differences in the frequency of pollen deposits on M. separata during different migration phases. Student's t-test was used to compare differences in the annual mean frequency of pollen occurrence on female and male body parts (proboscis, antennae, or the total [combined proboscis and antennae]) of M. separata moths, and differences in the annual mean frequency of pollen deposits on the proboscis or antennae of female, male, and total (female and male) M. separata moths. A chi-square test was used to compare differences in the frequency of pollen deposits on female and male body parts (proboscis, antennae or the total [proboscis and antennae]) of M. separata moths in each year, the differences in the rates between the proboscis and the antennae of female, male, and total (female and male) M. separata moths in each year, and the characteristics of the pollen source plants. All statistical computations were performed using SPSS 13.0 (SPSS Inc., Chicago, IL, USA) [31] and all proportion data were logit transformed before being analyzed.

Plant hosts inferred from pollen
Most of plant hosts were identified from single moths and only several moths harbored more than one plant species. Thirteen pollen-source species were identified, from nine families on M. separate moths using a combination of DNA rbcL data and pollen morphology. Six of the 13 pollen-source samples were identified to the species level: Melia azedarach Linn., Castanea mollissima Blume, Amorpha fruticosa Linn., Styphnolobium japonicum (L.) Schott., Chenopodium album L. and Flueggea virosa (Roxb. Ex Woigt) Voigt, and four to the genus level: Citrus, Adenophora, Aster, and Artemisia (Table 1, Fig 1). The geographic distribution of the pollen source plants is shown in Table 1 [32][33]. The success rates for a combination of pollen morphology, DNA barcoding, and distribution data identifying pollen to the species-and genus levels were 46.2 and 76.9%, respectively, while DNA barcoding alone had a success rate of 15.4 and 61.5%, and pollen morphology alone had a success rate of 7.7 and 69.2%.

Pollen detection rates by body part
No significant sex-related differences were seen in the annual mean frequency of pollen occurrence on the proboscis, antennae, or the total (combined proboscis and antennae) of M. separata moths. This was true both when years were grouped for analysis (2013-2015) ( Table 2, Fig  2) (with 14.2, 1.2 and 14.4% of females and 12.1, 0.9 and 12.3% of male moths contaminated with plant pollen on the proboscis, antennae, or both of proboscis and antennae) and when the years were analyzed individually ( Table 2, S1A, S1B and S1C Fig).
The pollen detection rates were higher on the proboscis (13.2%) than on the antennae (0.9%) for 2013-2015 as a group (Table 2, Fig 2), and this difference was significant for individual years ( Table 2, Table 2, S1D and S1E Fig).
With respect to the phenology of migration, the frequency of pollen occurrence on the bodies of M. separata was significantly higher (39.2%) in the early part of the migration period (May and June) than in the middle period (July and August) (3.4%) of the migration season, and no difference was detected between the late period (September and October) and either the early or the middle period in any of the years (F = 5.85, df 1,6 , P = 0.039) (Fig 3).

Discussion
We used DNA bar coding, pollen morphology, and the known distributions of plants to identify the major nectar plants of M. separata moths, including 13 species from nine families that were primarily Angiosperms in the Dicotyledoneae. Our study showed that M. separata moths visited woody plants more frequently than herbaceous plants, which was a finding that was similar to that of a previous study on Agrotis ipsilon (Hufmagel) [34]. The rate of population increase of insects is closely related to their host plant species [35], since the quality and quantity of available food plays an important role in insect development and fecundity [36]. It is Table 1 likely that noctuid moths tend to choose woody plants as a food sources, but this assumption requires further study. The migratory behavior of some insects plays an important role in their life-history [37]. Population dynamics, capture-release-recapture experiments and radar observation have  shown that M. separata is a long-distance migratory insect. The availability of nectar, which provides energy for flight, is a crucial factor that affects the duration of flight [37][38]. Host plant species for M. separata adults can be sufficiently identified by identifying the pollen species adhering to migratory M. separata because the nectar was not only collected from plants in the source area, but also from transit areas that could have been visited by the migratory M. separata.

Number Identified plants rbcL-molecular identification Pollen morphology identification Geographic distribution in China
Adult feeding is one of the most important behavioral characteristics of M. separata and has a significant influence on female fecundity [39]. Previous studies have found that when deprived of access to supplementary food, M. separata females could not regularly oviposit  [40]. Although such supplementary nutrition is not directly involved in vitellogenic synthesis, it is necessary for such synthesis to be successful. In addition, the type of nutrients also affects M. separata fecundity [41], as demonstrated by Guo and Liu (1964), who found that the fecundity of M. separata moths fed Brassica campestris L. nectar was significantly higher than that of moths fed Astragalus sinicus L. nectar [42]. Whether the adult host plants identified in this study can improve the fecundity of M. separata females remains to be studied. Moreover, adult feeding has a significant effect on the flight capacity of M. separata [43], and adequate nutrition after adult emergence is necessary for successful migration [44][45][46]. The flight potential of M. separata adults decreases significantly after adult emergence, as migratory moths became residents within 24 h when the adults were starved [47][48]. Therefore, the pollen source plants may have a direct effect on the trans-regional migration of M. separata by providing the moths with an energy supply.  We found pollen from M. azedarach, C. mollissima, Citrus L., S. japonicum,and A. fruticosa on M. separata adults captured in Beihuang from late May to June. Because the Yangtze River Basin is their common distribution area [32], these adults must have visited these flowers there, and the Yangtze River Basin may thus be the natal origin of M. separate captured in Beihuang in the spring. In addition, Since the Yangtze River Basin is located in southern Shandong province, M. separata migrated to Beihuang, which is evidence for the northward spring migration of M. separata.
Traditionally, pollen identification, which is highly dependent on human expertise and vulnerable to human error, often has limited taxonomic precision and is prohibitively time-consuming for large-scale studies [49]. The novel approach of DNA-based pollen analysis has great potential for the study of plant-herbivore interactions [14]. The Consortium for the Barcode of Life-plant Working Group (CBOL) has recommended the two-locus combination of rbcL + matK as the best plant barcode due to its universality, sequence quality, and species discrimination [30]. Unfortunately, matK is difficult to amplify universally using currently available primer set [31]. In this study, we chose the rbcL intron because it has the highest level of coverage in GenBank of potential barcoding markers. The species-level identification of host plants had better have a full DNA reference library of host plant species in the target community, however, it is not possible for M. separata because of their migratory habits, so the Gen-Bank was used in this research. RbcL is a good DNA barcoding region for plants at the family and genus levels, but for identification at the species level, a combination of pollen morphology and DNA barcoding is more precise. Meanwhile, other three markers (ITS, ITS2 and psbA-trnH) were recommended as primary DNA barcodes for plants at the Third International Barcoding Conference [50], and the trnL (UAA) intron chloroplast marker often used as a plant 'DNA barcode' for plant identification [51]. The species resolution can be increased by the combination of several barcoding [52][53], for better improvement the species discrimination, several barcoding should be used simultaneously in the future.
Adult Lepidoptera feed primarily on sugar sources [54]. We showed that some female and male moths carried pollen, which might have been contaminated during the process of nutritional supplementation. When pollen was found on the proboscis, many grains were typically present, which suggested active contact through feeding rather than casual contact through wind-blown contamination and provided evidence that the presence of nectar sources along the migration pathway of the population are a prerequisite for moths to reach their target habitat and lay eggs [55]. Mythimna separata carry more pollen in the early stage of the migration season, similar to A. ipsilon [37], which may be caused by differences in the abundance of nectar plants or the nutritional requirements of the moths. This topic requires further investigation. Nectar-feeding moths are attracted to the odors of their floral hosts [56][57]. Floral volatiles play a major role in plant-insect communication [58]. The use of host volatiles has been proposed as a potential lure for both male and female insects and as a means of monitoring and forecasting populations [59][60]. We showed that M. separata adults were effective pollinators of M. azedarach, C. mollissima, A. fruticosa, S. japonicum, C. album, F. virosa and other plants.
The flowers of these plants may contain specific attractant volatile components. The identification of these volatiles may allow the use of floral attractants for the management of M. separata.
Each year, M. separata undertakes a seasonal, long-distance, multigenerational roundtrip migration between southern and northern China [55]. This migration facilitates genetic exchange via pollen among plant populations across a large areas that is facilitated by adult M. separata [61]. The interactions between plants and insect herbivores are among the most important processes in terrestrial ecosystems [62][63][64][65]. Adult herbivores require nutrients or energy supplements from flowers for reproduction or flight, and flower-visiting herbivores are important pollinators [66]. The yields of many crops can be increased with the help of pollinators [67]. However, some studies confirm that the presence of adult food can enhance the herbivore population density in a range of agricultural systems [68][69]. Larvae of M. separata are a pest of millet and wheat [70] and can also occur in maize, rice and other crops [55].  suggested that the unprecedented increase in the geographic distribution of milk vetch from central China into south China was a key factor in the M. separata outbreaks from 1966-1977 [55]. Ultimately, more knowledge about M. separate will allow a better understanding of the regulation of the M. separata migration, allowing for the development of more efficient management practices.