Preference of a Polyphagous Mirid Bug, Apolygus lucorum (Meyer-Dür) for Flowering Host Plants

Apolygus lucorum (Meyer-Dür) (Hemiptera: Miridae) is one of the most important herbivores in a broad range of cultivated plants, including cotton, cereals, vegetables, and fruit crops in China. In this manuscript, we report on a 6-year long study in which (adult) A. lucorum abundance was recorded on 174 plant species from 39 families from early July to mid-September. Through the study period per year, the proportion of flowering plants exploited by adult A. lucorum was significantly greater than that of non-flowering plants. For a given plant species, A. lucorum adults reached peak abundance at the flowering stage, when the plant had the greatest attraction to the adults. More specifically, mean adult abundance on 26 species of major host plants and their relative standard attraction were 10.3–28.9 times and 9.3–19.5 times higher at flowering stage than during non-flowering periods, respectively. Among all the tested species, A. lucorum adults switched food plants according to the succession of flowering plant species. In early July, A. lucorum adults preferred some plant species in bloom, such as Vigna radiata, Gossypium hirsutum, Helianthus annuus and Chrysanthemum coronarium; since late July, adults dispersed into other flowering hosts (e.g. Ricinus communis, Impatiens balsamina, Humulus scandens, Ocimum basilicum, Agastache rugosus and Coriandrum sativum); in early September, they largely migrated to flowering Artemisia spp. (e.g. A. argyi, A. lavandulaefolia, A. annua and A. scoparia). Our findings underscore the important role of flowering plays in the population dynamics and inter-plant migration of this mirid bug. Also, our work helps understand evolutionary aspects of host plant use in polyphagous insects such as A. lucorum, and provides baseline information for the development of sustainable management strategies of this key agricultural pest.


Introduction
Agricultural landscapes regularly consist of crop fields interspersed with uncultivated habitats, thus providing abundant food resources for generalist phytophagous insects [1,2]. Change in the phenology of certain host or food plants results in a constantly changing mosaic of habitats across the agro-landscape [1,3]. Most polyphagous plant-feeding insects ephemerally exploit suitable host plants and habitats, but equally engage in host plant switching to locate new, more suitable hosts [1,4,5]. One advantage of such periodic host switching is that it permits continuous exploitation of a nutrient-diverse diet, thereby improving survival and reproduction [1,6,7]. Additionally, polyphagous insect herbivores usually exhibit clear preferences for particular plant species or plant growth stages [7,8,9,10,11]. An in-depth assessment of host plant preferences of polyphagous insects is central to understanding their seasonal dynamics on a particular plant species and their movement between plants and habitats across the agricultural landscape.
Many polyphagous insects, such as butterflies and moths (Lepidoptera), show great preference for flowers [7,12,13,14]. Many species of mirid bugs (Heteroptera: Miridae) prefer to feed on the relatively energy-rich plant tissues in flowers and buds [15], giving this insect group the common name ''flower bugs'' [16]. For example, Lygus lineolaris (Palisot de Beauvois) typically feeds on leaf buds and reproductive structures such as flower buds and flowers [17]. This mirid bug usually tracks a succession of flowering plant species, with plant colonization initiating at the formation of floral buds or flowers [18], and maximum abundance attained during bloom [19]. Lygus hesperus Knight often attained its peak of adult abundance in alfalfa, when that crop was in the blooming stage [11]. Similar phenomena already have been described in many other mirid bugs [15].
The mirid bug Apolygus lucorum (Meyer-Dür) (Hemiptera: Miridae) has historically been regarded as a minor pest in cotton and many other crops in China [20,21]. However, the widespread adoption of transgenic Bt (Bacillus thuringiensis) cotton and subsequent reduction of insecticide sprays in Bt cotton has allowed A. lucorum to reach outbreak levels in cotton and several other agricultural crops [22]. A. lucorum adults and nymphs feed on vegetative and reproductive tissues of their host plants, causing stunted growth and the abscission or malformation of leaves, flowers and fruits [20]. As a polyphagous species, recorded from at least 242 different host species in 49 different families, A. lucorum has been found to switch intensively between habitats and host plants over time [20,23,24]. As early as 1958, A. lucorum were   reported to track locally available flowering plants over the course of a cropping season [25]. Lu et al. [22] found that A. lucorum adults preferred cotton plants over other major host crops in midto late June in northern China, and proposed that this was because cotton is one of the few flowering host crops locally present during this period. However, much remains to be investigated regarding the plant flower preference of polyphagous A. lucorum and the associated ecological mechanisms.
In this study, we related A. lucorum adult abundance of on a given plant species with plant phenology data. Our objectives were (1) to assess temporal differences in the extent of flower preference by A. lucorum adults, and (2) to assess the role of flower preference as the driver of A. lucorum host plant switching.

Ethics Statement
No specific permits were required for the described field studies.  (Table 1), including wild and cultivated plants commonly found in agro-ecosystems of northern China. These 174 species of plant species comprised 74.7% (174 of 233) of the known A. lucorum summer host plants. Each plant species was established in three separate 464 m plots, with all plots arranged randomly and separated by a 1 m space that was kept free of vegetation by hand weeding. Plots were embedded within a .5 ha cotton field. Plots were established in early May each year through direct seeding and managed using identical agronomic practices among years, while refraining from all insecticide use [26]. Wild plants that were not available commercially as seeds were transplanted as seedlings from nearby agricultural fields. Wild plant species were identified using regional weed guides [27] or with assistance from CAAS plant taxonomists.

Field Trials
Each year, we surveyed A. lucorum adult abundance within each field plot every 4-5 days from early July to mid-September, coinciding with times of high A. lucorum abundance in local agroecosystems [20]. Sampling consisted of visually inspecting plants  [28]. Four 161 m subplots were sampled within each plot. At each sampling event, we also recorded plant growth stage and presence of flowers for each plant species [22,26]. For a given plant species, sampling was restricted to times when live plant material was present.

Data Analysis
A chi-square test was performed to compare the extent to which A. lucorum adults visited flowering vs. non-flowering plants during a given specific 2-wk sampling window per year. Each sampling period comprised three or four field surveys. If flowers were found at one or more surveys, the plant species was regarded as ''flowering'' for the corresponding period. On the other hand, if no flowers were found during any of the surveys, the respective plant species was treated as ''non-flowering''.
We calculated the standard attraction (A) of a given plant species (p) to A. lucorum adults at a given sampling date as Ap = Pp*n, where Pp is relative attraction, defined as the percent abundance of A. lucorum adults on plant species p versus total adult abundance on all tested plant species, and n is a standardization factor, defined as the total number of plant species found with A. lucorum adults at the same date [22]. This algorithm eliminates the potential influence of temporal differences in A. lucorum population density and number or type of plant species tested between seasons in estimating degree of attractiveness to A. lucorum adults of a given plant at a specific sampling date. Each year, we analyzed the most important host plant of A. lucorum, cotton (Gossypium hirsutum L.) and all other host species with higher adult abundances (i.e.,   seasonal mean density) than cotton. Standard attraction data for a flowering or non-flowering plant at a given sampling date were considered as replicates in the analysis. Per year, statistical differences in standard attraction between flowering and nonflowering stages for each plant species were determined using analysis of variance (ANOVA) followed by Tukey's honestly significant differences (HSD) test after verifying the assumptions of normality, homogeneity of variance, and independence. All statistical analyses were performed using SAS/STAT, version 9.1 (SAS Institute, Inc., Cary, NC).

Results
Over the course of the experiment, the proportion of flowering plants with the presence of A. lucorum adults was significantly higher than that of non-flowering plants in each of the different periods (inc. early July, late July, early August, late August, and early September) (P,0.05) ( Table 2). More specifically, the proportions of flowering and non-flowering plants exploited by A. lucorum adults were 50.0-100.0% and 11.3-31.8% in early July, 48.7-95.8% and 10.1-58.3% in late July, 63.6-98.4% and 4.8-51.7% in early August, 71.0-96.4% and 10.9-45.0% in late August, and 73.9-96.3% and 18.2-63.2% in early September, respectively ( Table 2).
For a given plant species with high adult abundance, standard attraction during flowering periods was significantly higher than during non-flowering periods (P,0.05) (Figure 1, Table 3). The average standard attraction of all selected flowering plants at flowering stage was 9.3, 7.7, 19.5, 15.5, 12.9, and 12.3 times higher than that during non-flowering periods from 2007 until 2012, respectively. Seasonal fluctuations in A. lucorum adult abundance on each plant species and the relative standard attraction for a given plant species showed similar trends. The mean population level of the above plant species at flowering stage was 10.3, 17.8, 28.9, 18.6, 13.9, and 18.2 times higher than that during non-flowering periods from 2007 to 2012, respectively (Figure 2-7).
The use of flowering plant species by A. lucorum adults varied during the course of the sampling period. In early July, A. lucorum adults preferred a small number of species, such as Vigna radiata (L.) Wilczek., Gossypium hirsutum L., Helianthus annuus L. and Chrysanthemum coronarium L., which were in flower. In late July, adults dispersed more widely into other hosts (e.g. Ricinus communis L.,

Discussion
In earlier work, seasonal host switching of certain polyphagous mirid bugs (e.g. L. lineolaris, Pseudatomoscelis seriatus [Reuter]) has been related to their preference for flowering host plants [19,29,30]. In our study, A. lucorum equally exhibited a clear preference for flowering plants and switched food plants according to the succession of different flowering plant species in the local agro-ecosystem [22,25]. It provided important information for further understanding the interaction between A. lucorum and host plants, and exploring the patterns of population dynamics of this mirid bug in different host plants.
The polyphagous species A. lucorum prefers to feed on tender leaves, buds and flowers, which usually become scarce after flowering stage [20]. To locate suitable food, A. lucorum adults exhibit a clear preference for flowering plant species in the process of host plant switching. This strategy of host plant switching helps offset seasonal or year-to-year changes in host abundance [31] and also allows mirid bugs to avoid intra-and interspecific competition for host plants.  supporting our speculation that A. lucorum altered host plants primarily to find suitable food. Through host plant switching hemimetabolous insects, such as mirid bugs, possibly can increase their population growth [32]. For example, L. lineolaris shows different rates of reproduction on different hosts, and host switching thus can considerably increase its population growth and survival [33]. In a laboratory study, A. lucorum adults and nymphs had higher survival and fitness on mungbean (V. radiata) over cotton [34], and on flowering individuals of three plant species (G. hirsutum, R. communis and I. balsamina) [32]. However, it is unknown which parts of the flowers (e.g., pollen, flower nectars) are the main food sources for A. lucorum or which nutrients (e.g., sugars, amino acids) are the most important for the increase of its population fitness [32]. Additionally, A. lucorum preference-performance relationship for flowering plants needs to be assessed in field conditions, as other ecological factors such as natural enemy abundance, environmental conditions, and broader host plant availability can affect host plant choice [35]. At a given time, A. lucorum showed a clear preference for a limited number of plants species. As not all plant species are present in all agricultural landscapes of northern China, A. lucorum abundance is deemed highly dependent upon location and composition of local agricultural landscapes [36]. In China, there are different cropping patterns, including mixed plantations of food crops and cotton, fruit trees and cotton, pastures and cotton, and so forth [37]. In each cropping pattern, the dominant overwintering location and seasonal host plant range of A. lucorum vary considerably [24], which would lead to different patterns of host plant use (inc. seasonal dynamics, between-plant transfer).
Our work showed year-by-year fluctuations in general A. lucorum abundance (Figure 2-7), which affected its population levels on a given host plant at any specific time. Yearly differences in climatic conditions and associated plant germination and growth are thought to be the prime determinants of those seasonal patterns [32,38,39]. Computer models maybe help to simulate its population dynamics in the agro-ecosystem and then analyze the effects of various biotic factors (e.g., host plant selection, phenological relative survival) and abiotic factors (e.g. temperature, rainfall) on its seasonal occurrence [40].
For many phytophagous insects, host switching is guided by host plant volatiles [41,42]. Adults of A. lucorum are attracted to variable extent to different plant species in Y-tube olfactometer trials [43], with electro-antennogram (EAG) responses to (E)-2-hexenal and other plant volatiles [44]. Increase in A. lucorum abundance on flowering plants may hint that adults orient themselves to specific volatiles of flowering plants. Visual cues may further enhance their behavioral response to plant volatiles [45]. However, for A. lucorum as for many other mirid bugs, much remains to be learned about the exact chemical and non-chemical determinants of flower preference.
Recently, there has been increasing interest in the application of behavioral manipulation methods (e.g. trap cropping) as a component of integrated pest management (IPM) strategies [46,47,48,49]. Our elucidation of considerable variation in A. lucorum abundance among host plants and among different periods (Figure 2-7), will contribute to the development of sustainable management strategies for A. lucorum. Previous work has led to the use of V. radiata as a trap crop for A. lucorum in Bt cotton fields [26]. This work also provides several other potential trap plants of A. lucorum and aids in identify the attractive volatile compositions, all of which could be developed as new alternative methods of controlling this mirid bug [49,50].
Agricultural landscapes dominated by crops and uncultivated habitats may contribute in increasing or decreasing pest population density in the fields, therefore analyzing the temporal variability of source and sink effects is of importance for managing the placement of landscapes to promote pest control. For example, Ting [51] found that the population abundance of mirid bug complex (mainly including A. lucorum, and Adelphocoris suturalis (Jakovlev), Adelphocoris lineolatus (Goeze), Adelphocoris fasciaticollis (Reuter)) in alfalfa fields at middle April were positively correlative with that in cotton field at early July. Carrière et al. [2] reported that abundance of seed alfalfa and cotton flowering date were positively associated with Lygus density in cotton fields, whereas abundances of cotton and uncultivated habitats were negatively associated with Lygus density in cotton. Our present study provide an ability to explore the source/sink role of different plant species as factors affecting population dynamics of A. lucorum, and aiding the development of landscape-level pest management strategies.