We deployed >50,000 Helicoverpa armigera eggs in maize fields to assess the rate of parasitism by Trichogramma chilonis across 33 sites during a three-year span (2012–2014) in northern China. Subsequently, we used a partial least squares (PLS) regression approach to assess the relationship of landscape diversity with composition and parasitism potential. The parasitism rate of H. armigera eggs by T. chilonis ranged from 0–25.8%, with a mean value of 5.6%. Landscape diversity greatly enhanced parasitism at all four different spatial scales (0.5, 1.0, 1.5 and 2.0 km radius). Both the proportion of arable area and the total planting area of two major crops (cotton and maize) had a negative correlation to the parasitism rate at each scale, whereas parasitism was positively correlated to the proportion of host crops of H. armigera other than cotton and maize at the 0.5 to 2.0 km radius scales as well as to that of non-crop habitat at the 0.5 and 1.0 km radius scales. The study indicated that maintaining landscape diversity provided an important biocontrol service by limiting H. armigera through the egg parasitoid T. chilonis, whereas rapid agricultural intensification would greatly reduce the presence and parasitism of T. chilonis in China.
Citation: Liu B, Yang L, Yang Y, Lu Y (2016) Influence of Landscape Diversity and Composition on the Parasitism of Cotton Bollworm Eggs in Maize. PLoS ONE 11(2): e0149476. https://doi.org/10.1371/journal.pone.0149476
Editor: Erjun Ling, Institute of Plant Physiology and Ecology, CHINA
Received: November 6, 2015; Accepted: January 31, 2016; Published: February 16, 2016
Copyright: © 2016 Liu et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper.
Funding: This research was supported by the National Natural Science Funds (No. 31222046, 31321004), and the Jiangsu Province graduate student research and creative projects (No. CXZZ13_0907).
Competing interests: The authors have declared that no competing interests exist.
Agricultural sustainability depends largely on the ability of natural enemies to suppress pest populations . The biocontrol services provided by natural enemies can not only control crop pests  but also reduce the need for chemical pesticides in cropping practices [3, 4], and decrease the potential for pesticide residue in plant-derived foods and the environment, slow the development of pest resistance to pesticides, and control secondary pest resurgence . However, this important ecosystem services has received limited attention from researchers and the public in many parts of developing countries, including China .
Previous studies indicate that biocontrol services are greatly affected by landscape contexts, especially by landscape composition [7–9]. Heterogeneous habitats possess diverse ecosystem functions . Among these functions, primary crops usually provide the major insect source of prey and act as hosts for natural enemies , whereas non-crop habitats or perennial habitats often support floral nectars and alternative prey species, and provide temporary refuges and overwintering sites [12, 13]. Biocontrol services also vary at multiple spatial scales [14–17], which are usually related to the dispersal capability of various natural enemies . Generally, predators such as ladybeetles fly longer distances than parasitoid wasps [19, 20]; thus, parasitism has stronger spatial scale effects than does predation . Recently, rapid agriculture intensification has led to an obvious simplification of agricultural landscapes all over the world, resulting in a marked reduction in the sustainability, stability and diversity of ecosystems [22–24] while simultaneously increasing the instability of relationships among food webs of crops, pests and natural enemies and their associated biocontrol services [25, 26]. For example, increasing the amount of corn planted for biofuel production reduced ecosystem services of natural enemies for controlling Aphis glycines in soybean fields in the United States . In China, the cropping structures and ecological conditions in agri-ecosystems are changing rapidly; however, the effect of this agricultural intensification on biocontrol services has not been widely assessed at landscape levels in China [27, 28].
During the early 1990s, the cotton bollworm Helicoverpa armigera was the most serious insect pest on cotton, maize, peanut, soybean, vegetables and many other crops in China [29, 30]. After the widespread commercial application of transgenic cotton expressing Bt (Bacillus thuringiensis) toxin, the bollworm population density in cotton and other neighboring crops dramatically decreased . However, H. armigera remains a common pest on non-Bt crops in China, including maize, peanuts, and vegetables. In this study, we estimated the effect of landscape context on the parasitism of H. armigera eggs in maize fields to address the following three questions. (1) Does landscape diversity and composition affect parasitism? (2) Do biocontrol services have spatial scale effects? (3) How will biocontrol services change as agriculture intensifies? We found that high landscape diversity and high proportion of host plants other than main crops (cotton and maize) greatly enhanced the parasitism of H. armigera eggs.
Materials and Methods
This field trial was approved by the local farmers. No specific permits were required for the described field studies, and our study did not involve endangered or protected species.
The study was conducted at a total of 33 field sites distributed around the city of Langfang in Hebei Province and around the cities of Wuqing and Jinghai in Tianjin Province of northern China (11 sites in 2012, 6 sites in 2013 and 16 sites in 2014). All selected sites were planted with maize along a gradient of proportion between pure maize and landscape diversity. In each year of the study, a minimum distance of 4 km separated the selected field sites to prevent any disturbance to the studied maize fields from affecting adjacent sites. We also avoided overlapping sites during the study. To simplify parasitism assessment, the field experiments were conducted from middle-to-late August, when the 4th generation of H. armigera emerged, during each of the three years (2012–2014). We ensured that no insecticide was sprayed during the trials, and we reimbursed farmers for yield losses at all sites.
The geographical coordinates of each focal maize field site were obtained using a handheld GIS product (Model “MG768”, Beijing UniStrong Science & Technology Co., Ltd., China). We defined our study regions at radii of 0.5–2.0 km with a 0.5 km interval around each site by using open-access satellite imagery from Google Earth. The land-cover map within the 2.0 km radius was digitized using ArcGIS 10.0 software  based on ground verification. We dropped small polygons such as ditches or very small features that were smaller than 5 m2 and could not be located during ground verification; the total area of all the unidentified field polygons was less than 0.1% of each landscape.
Land cover was assigned to one of 15 categories that were combined into three primary classes for landscape analysis: (1) Arable crops (the dominated crops cotton and maize as well as other host crops such as peanut, soybean, sweet potato, vegetables and fruit trees) and (2) Non-crop habitats (included forests, shrublands and grasslands), and (3) Urban which included buildings, roads, abandoned land, other impervious surfaces, and water. The digitized polygon land cover layers for each site were converted to a grid format in ArcGIS, and then the area and proportion of each selected site were determined with Fragstats 4 software  at each spatial scale.
We quantify landscape heterogeneity by using Simpson’s diversity index (D) which can effectively describe the variance in the proportion of area covered by different land-cover categories and can be calculated for each spatial scale . Simpson’s diversity, D, is calculated by D = 1/Σ(pi)2, where pi is the proportion of the ith land-cover category based on the 15 landscape categories.
We used H. armigera eggs as the sentinel host for parasitoid T. chilonis. Adult H. armigera were collected at night using artificial lights. We established a laboratory colony at the Langfang Experimental Station (Institute of Plant Protection of the Chinese Academy of Agricultural Sciences, Hebei province, China). The moths in the colony were maintained at 25 ± 1°C, 60 ± 10% RH and 16:8 h (L:D). Larvae were reared on an artificial diet, the main components of which were wheat germ, casein and sucrose. Adult moths were fed a 20% honey solution after emergence. In the laboratory, gauze is used to collect eggs because the moths prefer to lay eggs on it . New gauze was placed in an oviposition cage containing female and male adults at 18:00 pm when moths were ready to begin oviposition. The next morning (6:00 am) we collected the gauze with sentinel eggs attached. This “egg gauze” was cut into small pieces (approximately 3 cm wide and 5 cm long), and any flat (unfertilized) eggs on it were removed using an insect pin under a stereomicroscope. Each piece of “egg gauze” contained 50–100 eggs were used in field studies between 8:00–12:00 am on the same day. We randomly deployed 10 pieces of selected “egg gauze” as 10 replications at each study site for each deployment time. Due to low background pest density , the number of sentinel eggs was greatly higher than that naturally occurred in focal field, and ensured that it was sufficient to assess parasitism.
We selected a large plot that ranged from 3000–10000 m2 in the center of each maize field site and then randomly selected a total of 10 maize plants with a minimum 10 m interval between each plant and the field border. For each selected plant, one piece of “egg gauze” was attached to a young leaf near the middle of the maize stalk using an insect pin for each trial. The egg gauze was carefully collected from the field after 48 h of exposure and maintained within 10 ml glass dactylethrae. Then, a new egg gauze was deployed onto the same plant and similarly collected after another 48 h in the field. Three consecutive egg gauze deployments per site were conducted every year. During the summer, the developmental period for H. armigera eggs is approximately 72 h ; thus, after collecting the gauze from the field, we immediately recorded the number of eggs remaining on each piece of gauze prior to egg hatching because some of the eggs might have fallen off during the experimental operation or been damaged by predators. By 24 h after collection, nearly all the unparasitized eggs hatched successfully. By this point, all parasitized eggs darken to black and any remaining dead eggs shrivel due to water loss. At this time, we counted the number of parasitized eggs present per egg gauze. The collected “egg gauzes” were kept at 25 ± 1°C, 60 ± 10% RH and 16:8 h (L:D) and observed once per day. As parasitoids emerged, they were collected and placed into a 1.5 ml centrifuge tube containing 95% alcohol. The parasitoid samples were stored at -20°C for morphological species identification.
We used a partial least squares (PLS) regression approach to model variation between landscape variables and parasitism [34, 35]. The PLS model can minimize sample response prediction error and works by extracting successive linear combinations of the predictors to explain both predictor and response variation, especially when the predictors are highly correlated . The PLS model chooses the number of extracted factors by calculating cross validation (Qh2). In general, when Qh2≥0.0975, adding a factor to the model will be significant ; however, extracting too many factors can cause overfitting; in other words, tailoring the model too much will be detrimental to future predictions. Here, we used a one-at-a-time validation method to select the number of extracted factors by cross validation; the number of factors that minimize the predicted residual sum of squares (PRESS) was considered the best.
After selecting the adapted factors, we assessed the validity of the model by examining the amount of variation in the matrix of predictor and response variables using three indices (R2X, R2Y and Q2). R2X is the proportion of variance in the matrix of predictor variables used in the model, R2Y is the proportion of variance in the response variable explained by the model, and Q2 is the proportion of variance in the response variable that can be predicted by the model.
Next, the relative influence of each predictor variable was estimated using the Variable Importance for Projection (VIP), which is the sum of the influence of the variable over all the model dimensions divided by the total variation explained by the model. The VIP represents the contribution of each predictor in fitting the PLS model for both predictors and responses. In fact, variables are significantly explained when VIP ≥0.8 [36, 38]. Because predictor and response variables were standardized, the coefficients (B) of this equation may be interpreted as the influence of a particular predictor variable on the response variable, providing a direct indication of which predictors are most useful for predicting the dependent variable. If a predictor has a relatively small coefficient (|B|) and VIP (<0.8), it is a prime candidate for removal , after which the PLS procedures should be re-run to obtain a better model. The correlations loading plot shows how much variation in each variable is accounted for by the first two factors, jointly by the distance of the corresponding point from the origin and individually by the distance for the projections of this point onto the horizontal and vertical axes .
Before implementing this PLS regression (PROC PLS), we first tested the normality of the data and determined whether the variables were symmetrical. When they are not, an arcsine square root transformation of the predictors (percentage data) and response variable (parasitism rate) should be conducted to achieve data standardization. Ultimately, eight predictor variables (Simpson, Cotton, Maize, CM, OHC, Arable, NCH, Urban; see variable description in Table 1) were used in the model. We calculated the mean value of the parasitism rate on H. armigera eggs per site for each of the three deployment times, which served as response variables in the analysis. The PLS analysis was performed using SAS 9.3 software .
Parasitism on sentinel eggs in maize fields
Over the three years of the study, we deployed more than 50,000 sentinel eggs to investigate parasitism in the center of maize fields in all 33 study sites. A total of 2,000 wasps of T. chilonis emerged from the parasitized sentinel eggs. The parasitism rate of sentinel eggs varied from 0–25.8%, with a mean value of 5.6% at each site.
The term “landscape” in this study represented a gradient of agriculturally dominated land; the landscape varied at different spatial scales. Arable crop was the primary landscape type across all scales, varying from 27.9–96.9%. Maize and cotton were the main host crops for H. armigera with a range from 20.7–96.9%; other host crops were less prevalent, ranging from 0–27.6%. Moreover, the percent of non-crop habitat ranged from 0.4–57%. Urban land is an abiotic variable in the landscape system, with a range of 0.7–61.2%. We used Simpson’s diversity as the indication of landscape composition, the range of the Simpson value was 1.27–3.15 (Table 2).
Influence of landscape composition on parasitism services
The first two-factor model explained the cumulative variations of 65.2%, 59.9%, 54.9% and 55.5% in landscape predictor variables as well as the 35.6%, 25.7%, 40.7% and 42.8% variations in parasitism rate at the 0.5, 1.0, 1.5 and 2.0 km radius scales, respectively. Factor 1 explained 40.7%, 42.6%, 44.4% and 44.7% of the variation in landscape predictor variables (R2X) and 27.5%, 21.1%, 27.8% and 29.0% of the variation in parasitism rate (R2Y), whereas Factor 2 explained 24.5%, 17.3%, 10.5% and 10.8% of the variation in landscape predictor variables (R2X) and 8.1%, 4.6%, 12.9% and 13.8% of variation in parasitism rate (R2Y) at the four spatial scales, respectively (Table 3).
Furthermore, the four common landscape variables had a significant VIP score ≥0.8 at all four radius scales (Fig 1a and 1d), in which the parasitism rate was positively related to two variables: host crops of H. armigera other than the two major ones (cotton and maize) (OHC) and the landscape diversity index (Simpson). In contrast, the parasitism rate was negatively related to the total area percentage of cotton and maize (CM) as well as to the arable area (Fig 2a and 2d). The coefficients (B) of the PLS model also described the importance each predictor had in predicting the response (Table 4). The other three landscape variables (Urban, Cotton and Maize) had no significant correlation to the parasitism rate; they had VIP scores of <0.8 (Fig 1a and 1d) and so had low coefficients (B, Table 4) across all four radius scales. Another landscape variable, non-crop habitat (NCH), had high VIP values (VIP ≥0.8, Fig 1a and 1b) and was also positively related to the parasitism rate at the 0.5 and 1.0 km scales (Fig 2a and 2b) but had small VIP values (VIP <0.8, Fig 1c and 1d) and no significant correlation to the response variable at the 1.5 and 2.0 km radius scales (Fig 2c and 2d).
Radius (a): 0.5 km, (b): 1.0 km, (c): 1.5 km, (d): 2.0 km. The line parallel to the X axis indicates Wold’s Criterion (VIP = 0.8).
Radius: (a) 0.5 km; (b) 1.0 km; (c) 1.5 km; (d) 2.0 km. The four dotted line circles (25%, 50%, 75% and 100%) indicate the percentage range of variation explained by the first two factors. The distance of marked points from the origin and from the horizontal and vertical axes jointly show the amount of variation in each variable accounted for by the first two factors.
Although T. chilonis has been recorded as the main parasitoid on H. armigera eggs in northern China, limited data are available concerning its control efficiency under field conditions . In this study, parasitism was assessed at 33 sites in northern China, with the highest parasitism rate of 25.8% occurring at a mean value of 5.6% at each site. These results are similar to those obtained by a recent parasitism survey in cotton fields in the same region with the parasitism rate ranging from 0 to 38.8% and at a mean value of 2.5% . It showed that this parasitoid can provide an important suppressive effect on H. armigera.
Our data confirmed that landscape diversity can significantly improve the parasitism rate on H. armigera by T. chilonis in maize fields. The parasitism rate for the sentinel eggs was significantly positively related to the landscape diversity index (Simpson’s D) at four spatial scales with radii of 0.5, 1.0, 1.5 and 2.0 km. Many previous studies have demonstrated that landscape diversity (complexity) has a positive correlation to parasitoid wasp abundance and their biocontrol services. The results here are also similar to those of our previous study, which indicate that highly heterogeneous landscapes can effectively enhance the rates of parasitism on H. armigera eggs in cotton fields . However, the previous study confirmed high proportion of urban and water could enhanced parasitism, but in this study we found other host crops and non-crop habitats can improve it. Roschewitz et al.  and Plećaš et al.  also found that in contrast to simple landscapes, complex landscapes with higher diversity enhanced the abundance of parasitoids and parasitism on cereal aphids. In total, diverse habitats in the agricultural landscape always provide varied food sources and refuges for parasitoid wasps, thus favoring their population increase and improving their biocontrol services.
In this study, parasitism was not significantly correlated with the proportion of cotton or maize but was negatively correlated with the total proportion of both major crops as well as with arable land at all four radius scales. Another meaningful finding was that the presence of small host crops for H. armigera other than cotton and maize significantly improved parasitism rates at the four spatial scales tested. H. armigera is a polyphagous insect with more than 200 host plants in China . In multiple-cropping systems, the various crops have diverse planting seasons, growth stages and agricultural practices, but the agricultural practices of major crops (cotton and maize) are relatively consistent. Hence, crops other than cotton and maize are more available to support populations of H. armigera through all generations and function as an important source habitat for this parasitoid wasp.
Many studies have reported a positive effect of non-crop habitats on parasitoid abundance  and on their biocontrol services , but the effect depends on the spatial scales . We found that non-crop habitat surrounding a maize field could significantly enhance the parasitism rate by T. chilonis from 0.5 to 1.0 km scales, but the effect was not significant at larger scales (such as the 1.5 and 2.0 km radii). Non-crop habitat can provide nectar, pollen sources and refuges with limited human disturbance, all of which naturally enhance the occurrence of parasitoids in agro-ecosystems . Along with many crops, weeds and other vegetation in non-crop habitat also function as host plants for H. armigera , the presence of which also helps support the parasitoid population.
As the dispersal ability of parasitoids is relatively weak, their efficiency as biocontrol services usually changes between small and large spatial scales . In the present study, the Variable Importance for Projection (VIP) scores of small crops (other host crops, OHC) were higher than the VIP scores for non-crop habitat (NCH) at all four spatial scales (Fig 1a and 1d). Parasitism was correlated to the proportion of small crops at 0.5 to 2.0 km radius scales and to that of non-crop habitat at 0.5 and 1.0 km scales. This finding indicated that small crops other than cotton and maize played a more important role in promoting the population suppression of H. armigera by T. chilonis. In general, the proportion of other host crops is larger than the proportion of non-crop habitat nearby the main crop field. The poplar, the dominant tree species in non-crop habitat, is not a host plant for H. armigera and cannot provide ample nectar and pollen sources. Hence, the ecological effect for conservation biological control of non-crop habitat is less than that of small crops.
Based on the above findings, as agriculture has intensified, landscape diversity has significantly decreased. In particular, a reduction in planting of small crops impairs the ecosystem services function of natural enemies (e.g., the parasitism of H. armigera by T. chilonis). Our study provides significant knowledge on pest management practices that utilize ecosystem services of natural enemies to suppress pests and reduce insecticide spraying in China.
We thank the graduate trainees at Langfang Experimental Station, Institute of Plant Protection of the Chinese Academy of Agricultural Sciences during the period 2012–2014 for their assistance with the field trials.
Conceived and designed the experiments: YL YY. Performed the experiments: BL LY. Analyzed the data: BL. Contributed reagents/materials/analysis tools: YL. Wrote the paper: BL LY.
- 1. Power AG. Ecosystem services and agriculture: tradeoffs and synergies. Philos T R SOC B. 2010;365(1554):2959–71.
- 2. Landis DA, Gardiner MM, van der Werf W, Swinton SM. Increasing corn for biofuel production reduces biocontrol services in agricultural landscapes. Proc Natl Acad Sci U S A. 2008;105(51):20552–7. pmid:19075234; PubMed Central PMCID: PMC2603255.
- 3. Ostman O, Ekbom B, Bengtsson J. Landscape heterogeneity and farming practice influence biological control. Basic Appl Ecol. 2001;2(4):365–71. WOS:000172713900008.
- 4. Naranjo SE, Ellsworth PC, Frisvold GB. Economic Value of Biological Control in Integrated Pest Management of Managed Plant Systems. Annu Rev Entomol. 2015;60:621–45. pmid:25423598
- 5. Roubos CR, Rodriguez-Saona C, Isaacs R. Mitigating the effects of insecticides on arthropod biological control at field and landscape scales. Biol Control. 2014;75:28–38.
- 6. Wyckhuys KA, Lu Y, Morales H, Vazquez LL, Legaspi JC, Eliopoulos PA, et al. Current status and potential of conservation biological control for agriculture in the developing world. Biol Control. 2013;65(1):152–67.
- 7. Gardiner MM, Landis DA, Gratton C, Schmidt N, O'Neal M, Mueller E, et al. Landscape composition influences the activity density of Carabidae and Arachnida in soybean fields. Biol Control. 2010;55(1):11–9. WOS:000281060500002.
- 8. Gardiner MM, Landis DA, Gratton C, Schmidt N, O’Neal M, Mueller E, et al. Landscape composition influences patterns of native and exotic lady beetle abundance. Divers Distrib. 2009;15(4):554–64.
- 9. Steckel J, Westphal C, Peters MK, Bellach M, Rothenwoehrer C, Erasmi S, et al. Landscape composition and configuration differently affect trap-nesting bees, wasps and their antagonists. Biol Conserv. 2014;172:56–64.
- 10. Vasseur C, Joannon A, Aviron S, Burel F, Meynard J-M, Baudry J. The cropping systems mosaic: How does the hidden heterogeneity of agricultural landscapes drive arthropod populations? Agric, Ecosyst Environ. 2013;166:3–14.
- 11. Benton TG, Vickery JA, Wilson JD. Farmland biodiversity: is habitat heterogeneity the key? Trends Ecol Evol. 2003;18(4):182–8.
- 12. Chaplin-Kramer R, O'Rourke ME, Blitzer EJ, Kremen C. A meta-analysis of crop pest and natural enemy response to landscape complexity. Ecol Lett. 2011;14(9):922–32. WOS:000293628300012. pmid:21707902
- 13. Landis DA, Wratten SD, Gurr GM. Habitat management to conserve natural enemies of arthropod pests in agriculture. Annu Rev Entomol. 2000;45(1):175–201. WOS:000086173900008.
- 14. Gardiner MM, Landis DA, Gratton C, DiFonzo CD, O'Neal M, Chacon JM, et al. Landscape diversity enhances biological control of an introduced crop pest in the north-central USA. Ecol Appl. 2009;19(1):143–54. pmid:19323179.
- 15. Plećaš M, Gagić V, Janković M, Petrović-Obradović O, Kavallieratos NG, Tomanović Ž, et al. Landscape composition and configuration influence cereal aphid–parasitoid–hyperparasitoid interactions and biological control differentially across years. Agric, Ecosyst Environ. 2014;183:1–10.
- 16. Thies C, Steffan-Dewenter I, Tscharntke T. Effects of landscape context on herbivory and parasitism at different spatial scales. Oikos. 2003;101(1):18–25. WOS:000182106600003.
- 17. Thies C, Tscharntke T. Landscape structure and biological control in agroecosystems. Science. 1999;285(5429):893–5. pmid:10436158.
- 18. Sivakoff FS, Rosenheim JA, Hagler JR. Relative dispersal ability of a key agricultural pest and its predators in an annual agroecosystem. Biol Control. 2012;63(3):296–303.
- 19. Rand TA, Tylianakis JM, Tscharntke T. Spillover edge effects: the dispersal of agriculturally subsidized insect natural enemies into adjacent natural habitats. Ecol Lett. 2006;9(5):603–14. pmid:16643305
- 20. Tscharntke T, Bommarco R, Clough Y, Crist TO, Kleijn D, Rand TA, et al. Conservation biological control and enemy diversity on a landscape scale. Biol Control. 2007;43(3):294–309.
- 21. Briggs CJ, Hoopes MF. Stabilizing effects in spatial parasitoid–host and predator–prey models: a review. Theor Popul Biol. 2004;65(3):299–315. pmid:15139366
- 22. Bianchi FJ, Booij CJ, Tscharntke T. Sustainable pest regulation in agricultural landscapes: a review on landscape composition, biodiversity and natural pest control. Proceedings Biological sciences / The Royal Society. 2006;273(1595):1715–27. pmid:16790403; PubMed Central PMCID: PMC1634792.
- 23. Jonsson M, Buckley HL, Case BS, Wratten SD, Hale RJ, Didham RK. Agricultural intensification drives landscape-context effects on host–parasitoid interactions in agroecosystems. J Appl Ecol. 2012;49(3):706–14.
- 24. Thies C, Haenke S, Scherber C, Bengtsson J, Bommarco R, Clement LW, et al. The relationship between agricultural intensification and biological control: experimental tests across Europe. Ecol Appl. 2011;21(6):2187–96. pmid:21939053
- 25. Cardinale BJ, Srivastava DS, Duffy JE, Wright JP, Downing AL, Sankaran M, et al. Effects of biodiversity on the functioning of trophic groups and ecosystems. Nature. 2006;443(7114):989–92. WOS:000241523400054. pmid:17066035
- 26. Tscharntke T, Tylianakis JM, Rand TA, Didham RK, Fahrig L, Batary P, et al. Landscape moderation of biodiversity patterns and processes—eight hypotheses. Biol Rev Camb Philos Soc. 2012;87(3):661–85. pmid:22272640.
- 27. Zhao ZH, Hui C, Hardev S, Ouyang F, Dong ZK, Ge F. Responses of cereal aphids and their parasitic wasps to landscape complexity. J Econ Entomol. 2014;107(2):630–7. pmid:24772543
- 28. Zhou K, Huang JK, Deng XZ, van der Werf W, Zhang W, Lu YH, et al. Effects of land use and insecticides on natural enemies of aphids in cotton: First evidence from smallholder agriculture in the North China Plain. Agric, Ecosyst Environ. 2014;183:176–84.
- 29. Wu KM, Lu YH, Feng HQ, Jiang YY, Zhao JZ. Suppression of cotton bollworm in multiple crops in China in areas with Bt toxin-containing cotton. Science. 2008;321(5896):1676–8. pmid:18801998
- 30. Guo YY. Research of cotton bollworm: Beijing: China Agriculture Press; 1998.
- 31. ESRI. ArcGIS 10, ArcMap Version 10.0. Environmental Research Systems Institute Incorporated Redlands, CA; 2010.
- 32. McGarigal K, Cushman S, Maile N, Ene E. FRAGSTATS v4: Spatial Pattern Analysis Program for Categorical and Continuous Maps. Computer software program produced by the authors at the University of Massachusetts, Amherst. Available: http://www.umass.edu/landeco/research/fragstats/fragstats.html. 2012.
- 33. Liang GM, Tan WJ, Guo YY. Improvement of the artificial rearing techniques for cotton bollworm Helicoverpa armigera (Hübner). Plant Protection. 1999;25(2):15–7.
- 34. Grez AA, Zaviezo T, Gardiner MM. Local predator composition and landscape affects biological control of aphids in alfalfa fields. Biol Control. 2014;76:1–9.
- 35. Woltz JM, Landis DA. Coccinellid response to landscape composition and configuration. Agric For Entomol. 2014;16(4):341–9.
- 36. Wold S, Sjöström M, Eriksson L. PLS-regression: a basic tool of chemometrics. Chemometrics Intellig Lab Syst. 2001;58(2):109–30.
- 37. Tenenhaus M, Pages J, Ambroisine L, Guinot C. PLS methodology to study relationships between hedonic judgements and product characteristics. Food quality and preference. 2005;16(4):315–25.
- 38. Wold S. PLS for multivariate linear modeling. Chemometric methods in molecular design. 1995;2:195.
- 39. SAS Institute. SAS/STAT® 9.3 User’s Guide. Cary, North Carolina: SAS Institute Inc. 2011.
- 40. Zhang YJ. Investigations and application of Trichoghramma, the parasitisod of the cotton bollworm, Helicoverpa armigera. Nat Enemies Insects. 1981;3(3):14–8.
- 41. Liu B, Yang L, Yang F, Wang Q, Yang YZ, Lu YH, et al. Landscape diversity enhances parasitism of cotton bollworm (Helicoverpa armigera) eggs by Trichogramma chilonis in cotton. Biol Control. 2016;93:15–23.
- 42. Roschewitz I, Hucker M, Tscharntke T, Thies C. The influence of landscape context and farming practices on parasitism of cereal aphids. Agric, Ecosyst Environ. 2005;108(3):218–27. WOS:000229684400005.
- 43. Bianchi F, Van Wingerden W, Griffioen A, Van Der Veen M, Van Der Straten M, Wegman R, et al. Landscape factors affecting the control of Mamestra brassicae by natural enemies in Brussels sprout. Agric, Ecosyst Environ. 2005;107(2):145–50.
- 44. Alignier A, Raymond L, Deconchat M, Menozzi P, Monteil C, Sarthou J-P, et al. The effect of semi-natural habitats on aphids and their natural enemies across spatial and temporal scales. Biol Control. 2014;77:76–82.
- 45. Ramsden MW, Menéndez R, Leather SR, Wäckers F. Optimizing field margins for biocontrol services: The relative role of aphid abundance, annual floral resources, and overwinter habitat in enhancing aphid natural enemies. Agric, Ecosyst Environ. 2015;199:94–104.