Field evaluation of sunflower as a potential trap crop of Lygus pratensis in cotton fields

The mirid bug Lygus pratensis is an important pest of cotton, and is primarily managed through insecticide application. In this study, conducted in Xinjiang (China), we assessed the relative attractiveness of sunflower (Helianthus annuus) to L. pratensis adults in local cotton plots from 2015–2016 and quantified the associated contribution of inter-planted sunflower strips to suppress field-level L. pratensis populations from 2016–2017. Field-plot trials showed that among six combinations of two sunflower varieties (XKZ6 and SH363) and three planting dates (early-, middle- and late-planted), adult abundance of L. pratensis was highest on early-planted XKZ6 and attained 3.7–5.8 times higher levels than in neighboring cotton plots. In commercial cotton fields, the combined deployment of sunflower strips at field edges and the periodic application of insecticides directed to those strips was found to (1) reduce the mean abundance of L. pratensis population on cotton by 41.9–44.0%, (2) lower the rate of cotton leaf damage by 27.3–30.6% and boll damage by 44.8–46.0%, and (3) increase the number of mature bolls by 7.5%-8.0%. Our work emphasizes how sunflower can be an effective trap crop for L. pratensis and that the establishment of sunflower strips could contribute to its effective and environmentally-sound management in cotton crops.


Introduction
Lygus pratensis (L.) (Hemiptera: Miridae) is a common pest on cotton (Gossypium spp.), alfalfa (Medicago sativa), Chinese date (Ziziphus jujuba), grape (Vitis vinifera), pear (Pyrus spp.), apple (Malus pumila) and many other crops. Aside from its broad host range, this mirid bug has a wide geographic distribution and is found throughout Europe, north Africa, the Middle East, northern India, China, and Siberia [1,2]. In China, L. pratensis is found mainly in the Xinjiang Uyghur Autonomous Region, and was locally regarded as a major pest on cotton in the 1950s and 1960s [3][4][5]. Both nymphs and adults of L. pratensis feed on vegetative and reproductive parts of cotton plants, thus causing leaf damage, plant stunting and abscission of squares and bolls [6][7][8][9]. Occasionally, L. pratensis outbreaks caused 100% damage to cotton plants and resulted in significant loss of crop quality and yield [10]. Over the past decades, insecticide-based approaches have suppressed L. pratensis population levels and safeguarded cotton yields [9]. Yet, L. pratensis population levels in Xinjiang have increased in recent years after wide-scale adoption of Bt (Bacillus thuringiensis) cotton, and this mirid bug has now reemerged as a major pest of cotton and multiple other crops (e.g., stone fruits) in cotton agrolandscapes [1,[11][12][13][14]. As chemically-synthesized insecticides continue to be the cornerstone of L. pratensis mitigation programs [1,14], non-chemical technologies need to be developed, validated and implemented [15]. As a polyphagous pest, L. pratensis disperses from one host plant to another under field conditions, and exhibits variable feeding responses and a marked preference for plants at flowering stage [16,17]. This type of foraging behavior lends itself to devise trap cropping systems; a method of behavioral manipulation based on the pest's host plant preferences [18,19]. By establishing the pest's preferred host plant (i.e., trap plant) close to the main crop, the target pest is diverted from the main crop and pest damage is lowered. Trap cropping been successfully used against multiple economically-important pests, and been successfully applied for the management of many important pests, including several species of mirid bugs (Hemiptera: Miridae) [20]. For instance, the cotton-alfalfa trap crop system relies on alfalfa strips sown within cotton fields to control Lygus hesperus Knight in the United States [21][22][23]. A similar intercropping pattern has been used in Australia to lower infestation levels of Creontiades dilutus (Stål) in Australia [24]. Alfalfa has also been used as a trap crop for several species of Lygus (e.g., L. hesperus and L. rugulipennis Poppius) in strawberry (Fragaria ananassa) fields in the United States and in Italy [25][26][27]. In the UK, German chamomile, Matricaria recutita, acts as a trap crop for L. rugulipennis on strawberry [28] Furthermore, other plant species, such as red clover (Trifolium pretense), mugwort (Artemisia vulgaris) and sunflower (Helianthus annuus), can be used for trapping species of Lygus spp. in lettuce (Lactuca sativa) and cucumber (Cucumis sativus) [29,30]. In Chinese cotton fields, mungbean (Vigna radiata) and cowpea (Vigna unguiculata) are used as trap crops for Apolygus lucorum (Meyer-Dür) and Adelphocoris suturalis (Jakovlev), respectively [31,32]. Overall, trap cropping has proven to be an effective means of mirid bug management in various agro-ecosystems worldwide and thus carries ample promise to reduce the current reliance upon chemically-synthesized insecticides to control L. pratensis.
In our field survey, L. pratensis was found to attain high population levels on sunflower, hinting its potential value as a trap crop for this mirid bug. In this study, we 1) compared L. pratensis infestation levels in field plots established with cotton and different varieties and planting dates of sunflower, 2) assessed whether L. pratensis' infestation pressure and damage was lowered in cotton fields inter-planted with sunflower strips.

Plants and field sites
In our trials, we used two sunflower varieties: (1)

Lygus pratensis occurrence in cotton and sunflower plots
In 2015, cotton was planted on April 1, and sunflower was sown at three different times: April 1 (early-planted), April 10 (middle-planted), and April 20 (late-planted). In 2016, cotton was sown on April 3, and sunflower was sown on April 3 (early-planted), April 13 (middle-planted), and April 23 (late-planted). During each year, the following treatments were established: two sunflower varieties (i.e., XKZ6, SH363) each with three planting dates (6 treatments), and cotton with a single planting date (control treatment).
Each of the above treatments was established in 3 field plots (i.e., replicates), totaling 21 plots for the entire experiment. Each plot was 100 m 2 (i.e., 10 m wide by 10 m long), and all plots were randomly arranged with 2 m wide vegetation-free aisles between neighboring plots. The seeding density was 2.25×10 5 seeds per ha for cotton, and 6.5×10 4 seeds per ha for sunflower. All plots were under flood irrigation with an identical fertilization and watering regime, and no chemical pesticides were applied during the entire growing season. A field survey of L. pratensis population was conducted every five days from early June to middle August during both years. In each plot, using five-point sampling method and 20 consecutive plants were checked each point (Fig 1A), the number of L. pratensis adults was visually determined and recorded.
Beginning in early June, as these first generation of L pratensis nymphs become adults on spring host plants, they disperse to other host plants and then produce second generation there. In other words, before they attack the sunflower and cottons, no L. pratensis' infestation were occurred on these two crops. Therefore, we only investigate the relative attractiveness of sunflowers and cottons to adults from entering period, ignore the nymphs.

Lygus pratensis occurrence in cotton with and without sunflower border strips
Cotton and sunflower were simultaneously planted on April 5, 2016 and April 9, 2017. Seeding densities of cotton and sunflower were 2.25×10 5 seeds per ha and 6.5×10 4 seeds per ha, respectively. During each year, a total of eight cotton fields (each approximately 30 m wide and 30 m long) were set up with >10 m wide vegetation-free areas between neighboring fields. Four cotton fields with sunflower strips were established as trap cropping treatments, and the other four without sunflower strips were designated as control treatment.
For each trap-cropping field, two 1-m wide strips of sunflower (accounting for 6% crop surface) were planted on the opposite borders of the field, parallel to the cotton rows. Cotton fields with and without sunflower strips were managed under identical irrigation, fertilization and crop management schemes. No pesticides were used on cotton plants in any of the fields. To control L. pratensis populations on sunflower strips, 40% acetamiprid water dispersible granules at 120g/ha or 25% imidacloprid wettable powder at 120g/ha with water 600 kg/ha were applied in alternation every 10 days following the first detection of the pest.
Sampling of L. pratensis in cotton fields was done every five days, from early June to mid-August, using the same method as above. On each sampling date, the number of L. pratensis adults and nymphs (per plant) was recorded, and L. pratensis feeding damage on the top five cotton leaves was assessed in late June. Also, the number of mature cotton bolls (diameter >2.5 cm) was determined and the damage rate for bolls was determined in late August. For both boll number and damage level, 100 plants from five selected points (20 plants per point) were examined in each field (Fig 1B).

Statistical analysis
Differences in the abundance of L. pratensis (including adults, nymphs, and total population, as three counts) in cotton and sunflower plots (2015-2016) or in cotton fields with/without sunflower strips (2016-2017) were analyzed by a repeated-measures analysis of variance (ANOVA) using SAS PROC MIXED, and the means were compared by the least significant difference test (LSD). The data were log 10 (n+1)-transformed before analysis. The mean abundances of L. pratensis adults in plots of cotton and sunflower during 2015-2016 were compared using one-way ANOVA followed by Tukey's HSD for multiple comparisons. Mean abundances of L. pratensis population, the damage rate of leaves and bolls (arcsin-transformed), and the number of mature bolls in cotton fields with/without sunflower strips during 2016-2017 were compared through non-paired t-tests. In 2016 and 2017, L. pratensis abundances in both cotton fields were below the existing economic thresholds for this pest (i.e., 20 bugs per 100 cotton plants at blossoming stage; Wang et al. [14]) after middle July. Hence, only population abundance records from early June to mid-July were included in the repeated-measures ANOVA and non-paired t-tests.

Lygus pratensis occurrence in cotton and sunflower plots
In 2015 and 2016, L. pratensis adults attained peak abundance between early June and late July in cotton fields. In June, adult abundance was high on early-planted sunflower while in July, abundance of L. pratensis adults was relatively higher on early-and middle-planted sunflower. In plots of late-planted sunflower, adult abundance remained low throughout the whole period (Figs 2 and 3).
Population levels of L. pratensis adults differed significantly among all seven treatments in 2015 (repeated measures ANOVA; F (6,12) = 70.88, P <0.0001) and 2016 (F (6,12) = 43.59, P <0.0001) (Fig  2). Also, mean abundance of L. pratensis adults on early-planted XKZ6 was significantly higher than on the other six treatments during both years (one-way ANOVA; 2015: F (6,12) = 66.37, P <0.0001; and 2016: F (6,12) = 107.58, P <0.0001). More specifically, adult abundance on earlyplanted XKZ6 was 3.73 times and 5.81 times higher than that on cotton in 2015 and 2016, respectively (Fig 3). Cotton as control treatment, early-plant sunflowers XKZ6 have the highest attraction for L. pratensis, so it have more potential as trap crop in cotton fields. Therefore, comparing the abundance of L. pratensis between sunflowers and cotton is requisite and meaningful.

Lygus pratensis occurrence in cotton with and without sunflower strips
As results above showed, early-plant sunflowers have more attractiveness to L. pratensis adults, so used early-plant sunflower strips in cotton fields as trap crop. Because L. pratensis adults produced second and three generations in cotton fields, to assess the role of sunflower strips on population dynamics of L. pratensis in cotton fields, the abundance including adults and nymphs were investigated.
As compared with control treatments, L. pratensis population levels were markedly lower in cotton fields with sunflower borders during 2016 and 2017 (Fig 4). More specifically, significant differences were recorded between fields with and without sunflower strips for population levels of L. pratensis nymphs (repeated measures ANOVA;  Table 1).

Discussion
The mirid bug Lygus pratensis is a polyphagous, economically-important pest of multiple crops in Europe, central Asia and north Africa, and is a common target of insecticide applications in cotton fields in China. In this study, we revealed how L. pratensis adults greatly preferred sunflower plants sown in early season, and showed how sunflower strips clearly suppressed L. pratensis population levels and crop damage in cotton. Our work can constitute the basis for further development of integrated pest management schemes against L. pratensis in China's cotton crop, and eventually permit significant reductions in insecticide use against this pest. Aside from its role in lowering L. pratensis population levels in China's cotton crop, sunflower has also shown potential as a trap crop in various other crop x pest systems e.g., for control of the stalk borer Dectes texanus LeConte (Coleoptera: Cerambycidae) in soybean [33], the brown marmorated stink bug, Halyomorpha halys Stål (Hemiptera: Pentatomidae), in pepper [34,35], and the European tarnished plant bug, L. rugulipennis, in glasshouse cucumber [30]. Also, multi-species strips of sunflower, Chinese cabbage, marigolds and rapes can help suppress the pollen beetle, Melighetes aeneus F. (Coleoptera: Nitidulidae), in cauliflower fields [36]. Hence, considering the importance of L. pratensis as a key pest of alfalfa, vegetables, fruit trees, and multiple other crops [1], sunflower strips potentially could also serve as a trap crop and help alleviate insecticide application pressure in these other crops.
The success of trap cropping depends upon the presence of a highly-attractive trap crop, during times when the population of the target pest is high [18]. Hence, the spatio-temporal presence of a trap crop (e.g., planting date, growth dynamics and flowering time) is an important selection criterion when devising a trap-cropping strategy aimed at a particular pest [20]. In certain cases, the target pest can sustain high population levels during prolonged periods of time, and a staggered establishment of a trap crop might thus be needed to sustain its attractiveness [31,37,38]. In our study, the 2015-2016 survey showed comparatively high L. pratensis adult infestation levels in cotton fields during June, which were similar to those in plots with early-planted sunflower. The attractiveness of early-planted sunflower also appeared to be superior to that of middle-and late-planted ones, further accentuating its potential as a trap crop for L. pratensis in local cotton agro-ecosystems. As considerable intra-specific variability may occur in the degree of attractiveness of a particular trap crop to a given pest [39,40], field work is often required to carefully select the most attractive varieties. Furthermore, yield and marketability of a given trap crop variety could also be determining factors in securing its wider adoption by farmers [41]. This study compared two common varieties of sunflower and revealed the superior attractiveness of early-planted XKZ6. Yet, no intra-specific differences in in-field L. pratensis attraction were reported for middle-and late-planted sunflower, possibly related to particularities of L. pratensis adult host plant selection and its marked preference for plants at the budding and blossoming stage [1,16]. Early-planted sunflower generally entered into bud stage by late May, at which time high L. pratensis adult abundance was recorded in the field. Also, XKZ6 is characterized by a comparatively slower development from bud to florescence, which might have further enhanced its attractiveness to L. pratensis over the course of June.
Periodic suppression of target pests within the trap-crop strips can reduce the potential of re-colonization of the main crop and boost its pest control efficiency [18]. In our study, insecticide sprays in sunflower strips at a 10-day frequency significantly lowered L. pratensis abundance in the neighboring cotton field. As sunflower strips solely occupied 6% of the cotton area, the current 6 insecticide sprays directed to the trap-crop strips amounted to a mere 0.36 applications in terms of total cotton area. Thus, trap cropping might greatly reduce the amount of chemical insecticides for L. pratensis control in cotton and lower the environmental burden of commercial cotton production in Xinjiang, China.
Sunflower is an important crop in its own right in southern Xinjiang, and socio-economic factors possibly can further facilitate its field-level establishment by local cotton growers. Yet, given the proven environmental impacts of neonicotinoid insecticides such as acetamiprid and imidacloprid [42], and the unmistakable risks those products pose to pollinators, insect natural enemies and other wildlife, future studies should investigate biopesticides and more environmentally-friendly alternatives for L. pratensis control. A number of other control options of L. pratensis, including the use of sex pheromones and light traps [1], have the potential to be coupled with trap cropping and might further boost control efficacy of sunflower strips. Also, more scientific attention can be paid to the exact size and physical placement of the trap crop, and how such relates to its trapping efficiency and field-level L. pratensis pest control [18,43].
During 2016-2017 trials, L. pratensis nymphal abundance of subsequent generations was decreased on trap-cropped cotton fields and the associated leaf and boll damage rates of leaf and boll were lower than in control plots. Yet, L. pratensis abundance in trap-cropped fields still exceeded existing economic thresholds for this pest (i.e., 12, 20, 41 bugs per 100 plants at budding, blossoming and bolling stages, respectively) in local cotton fields [14]. It is not uncommon for trap cropping to be unable to fully suppress the target pest below economically-significant levels [18,44]. For instance, insecticide application was used in cotton fields with mungbean strips when A. lucorum surpassed certain population levels [31]. Hence, trap cropping for L. pratensis may need to be complemented with other control tactics, such as targeted applications of selective insecticides [1,15], pheromone-based and light-trap mass-trapping schemes (see above), augmentative biological control, or ecological engineering measures.
Our study validates the usage of sunflower as a trap crop for L. pratensis in cotton fields in southern Xinjiang, and describes the particular role of early-planted XKZ6 sunflower strips (coupled with targeted insecticide sprays) to suppress field populations of this mirid bug. Our work provides an effective L. pratensis pest management tool which can now be adapted and integrated with other tactics, in order to advance sustainable, environmentally-sound crop protection in China's commercial cotton production.