Mirid Bug Outbreaks in Multiple Crops Correlated with Wide-Scale Adoption of Bt Cotton in China

See allHide authors and affiliations

Science  28 May 2010:
Vol. 328, Issue 5982, pp. 1151-1154
DOI: 10.1126/science.1187881


Long-term ecological effects of transgenic Bacillus thuringiensis (Bt) crops on nontarget pests have received limited attention, more so in diverse small holder–based cropping systems of the developing world. Field trials conducted over 10 years in northern China show that mirid bugs (Heteroptera: Miridae) have progressively increased population sizes and acquired pest status in cotton and multiple other crops, in association with a regional increase in Bt cotton adoption. More specifically, our analyses show that Bt cotton has become a source of mirid bugs and that their population increases are related to drops in insecticide use in this crop. Hence, alterations of pest management regimes in Bt cotton could be responsible for the appearance and subsequent spread of nontarget pests at an agro-landscape level.

Genetically engineered crops that express δ-endotoxins (Cry proteins) from Bacillus thuringiensis (Bt) can successfully control several insect pests. The adoption of Bt crops increases yield and causes vast reductions in insecticide use (15). With Bt crops presently adopted in over 20 countries (6), the ecological risks of their commercial cultivation have received considerable scientific scrutiny (711). In China, Bt cotton was approved in 1997 for commercial use to control cotton bollworm, Helicoverpa armigera, and has steadily been adopted by the bulk of Chinese cotton growers (i.e., presently 95% adoption in northern China). Bt cotton controls H. armigera larvae very effectively and acts as a dead-end trap crop for regional populations of this pest in local agricultural landscapes (12); that is, a large percentage of the pest moths lay their eggs in cotton, where the hatching larvae are killed and do not subsequently infest other crops as adults. Hence, Bt cotton controls a key target pest not only within cotton fields but also on multiple other non-Bt host crops (i.e., corn, peanuts, soybeans, and vegetables), reducing the overall need for insecticide sprays (3, 12). Nevertheless, long-term impacts of Bt cotton on nontarget arthropods, such as polyphagous insect pests, in local agro-ecosystems remain to be quantified (1315).

Mirid bugs (Heteroptera: Miridae) are herbivores in a broad range of cultivated plants, including cotton, cereals, vegetables, and fruit crops. In China, mirid bugs have historically been considered occasional or minor pests in most crops, occurring at relatively low population densities and only sporadically requiring pest management intervention (16). Nevertheless, mirid bugs can easily attain outbreak densities, switch host crops, or experience geographic spread because of their environmental adaptability (1618), high population growth rate (16, 17), and strong dispersal capacity (19, 20). In this study, we determined whether mirid bug outbreaks are more likely to occur in Bt cotton than in conventional cotton and to what extent these insects could cause unintended ecological impacts in the broader agro-ecosystem (21). Research was conducted in six major cotton-growing provinces (i.e., Henan, Hebei, Jiangsu, Anhui, Shandong, and Shanxi) of northern China (Fig. 1), where 3 million ha of cotton and 26 million ha of alternative crops, potentially susceptible to mirid bug attack, are cultivated annually by >10 million small-scale farmers.

Fig. 1

Survey locations in northern China. Mirid bug population dynamics on Bt and non-Bt cotton were monitored at Langfang site (star). Population densities were surveyed on cotton at 38 sites (red), and infestation densities were determined on other host crops at 77 sites (red and blue).

Mirid bugs were sampled from 1998–2009 in Bt and non-Bt cotton plots at the Langfang Experiment Station in Heibei province (21). Mirid bug abundance did not differ between cotton varieties with similar management regimes (P > 0.05), and calendar-based insecticide sprays for H. armigera control in non-Bt cotton significantly lowered mirid bug infestation densities (P < 0.05) (Fig. 2 and table S3). Hence, Bt cotton per se does not affect mirid bug infestation densities, and mirid bug populations prove susceptible to broad-spectrum insecticides for H. armigera.

Fig. 2

Mirid bug population dynamics in Bt and non-Bt cotton with different management regimes from 2002–2009. Bt and Non-Bt indicate Bt cotton and non-Bt cotton, respectively, without insecticide sprays, whereas Non-Bt-CC represents non-Bt cotton with H. armigera insecticide sprays.

We simultaneously monitored mirid bug abundance and insecticide use in cotton at 38 locations throughout the study region during 1997–2008 and 1992–2008, respectively (21). Mirid bug population levels gradually increased over time and were significantly related to Bt cotton planting proportion (P < 0.05) (Fig. 3). Insecticide use patterns also changed with Bt cotton adoption (fig. S1). After introduction of Bt cotton, the number of insecticide sprays against H. armigera and all insect pests was evidently lower than during 1992–1996. On the other hand, the number of sprays against mirid bugs increased over time, in line with Bt cotton planting proportion (fig. S1). Meanwhile, H. armigera insecticide use was a highly explanatory variable for annual mirid bug population levels and insecticide use for mirid bug control (P < 0.05) (Table 1).

Fig. 3

Association between mirid bug population density (top) or number of mirid bug insecticide sprays (bottom) and Bt cotton planting proportion in northern China during 1997–2008. Linear model for population density: y = 6.81x + 0.54, F 1,10 = 88.65, P < 0.0001, coefficient of determination (R2) = 0.90. Nonlinear model for number of mirid bug insecticide sprays: y = 0.89 + 0.08*exp(4.21x), F2,9 = 126.46, P < 0.0001, R2 = 0.97.

Table 1

Multivariate assessment of determinants of mirid bug population density and insecticide spray frequency in northern China during 1997–2008. For stepwise regression, only variables with significance P < 0.05 were entered into the analysis. “Other insecticide sprays” target insect pests different from H. armigera or mirid bugs. r, regression coefficient. R2, coefficient of determination. A slash indicates no analysis.

View this table:

Additionally, mirid bug infestations were recorded in alternative host crops: Chinese date, grapes, apple, peach, and pear (21). Mirid bug infestation severity increased in alternative host crops, and respective infestation severities were significantly correlated with regional proportion of Bt cotton planted (P < 0.05) (Fig. 4).

Fig. 4

Association between mirid bug infestation severity in either cotton or key fruit crops and Bt cotton planting proportion. The measure of mirid bug infestation was assigned a score ranging from 1 (no infestation) to 5 (extreme infestation). Nonlinear model for cotton, y = 1.23 + 0.04*exp(4.24x) (F2,9 = 346.40, R2 = 0.99, P < 0.0001); apple, y = 0.64 + 0.40*exp(1.76x) (F2,9 = 107.77, R2 = 0.96, P < 0.0001); grape, y = 1.12 + 0.02*exp(4.15x) (F2,9 = 118.05, R2 = 0.96, P < 0.0001); peach, y = 0.97 + 0.04*exp(3.61x) (F2,9 = 46.07, R2 = 0.91, P < 0.0001); Chinese date, y = 1.06 + 0.20*exp(2.74x) (F2,9 = 166.88, R2 = 0.97, P < 0.0001); and pear, y = 1.39 + 0.11*exp(2.88x) (F2,9 = 153.39, R2 = 0.97, P < 0.0001).

Over the study period, mirid bugs gradually increased population densities and damage in cotton and multiple other crops. Mirid bug attraction to flowering plants and associated seasonal host alternation partially explains the observed pattern (16, 22). In mid- to late June, mirid bugs largely move from early-season host plants to crop fields, where they build up their initial populations. Results of 2006–2009 field-plot trials indicate that mirid bugs greatly prefer cotton over other major host crops in mid- to late June (fig. S2), because cotton is locally one of few flowering host crops during this time period. Before Bt cotton adoption, broad-spectrum H. armigera insecticide use reduced early mirid bug populations, with cotton acting as a dead-end trap crop. Current absence of insecticide sprays in Bt cotton permits unrestrained mirid bug population buildup and subsequent (active) spread or (passive) spillover to a multitude of other (flowering) crops. Hence, a reduction in insecticide use for H. armigera control in Bt cotton correlates with mirid bug outbreaks in cotton and various fruit crops in the broader agro-landscape.

Most polyphagous insects exhibit clear preferences for one or few host plants and may seasonally concentrate in patches of these plants. Consequently, management actions in these patches can greatly determine population dynamics of such insects at the landscape level (23). Our work shows that a drop in insecticide use in Bt cotton fields leads to a reversal of the ecological role of cotton: from being a sink for mirid bugs in conventional systems to being an actual source for these pests in Bt cotton–growing systems. This perspective should be instrumental in developing regionwide management strategies for these polyphagous pests in northern China and elsewhere in the world.

Pest resurgence and replacement are usually ascribed to alterations in pest management regimes (24). For example, cotton aphid, Aphis gossypii, evolved as a primary pest of cotton in the mid-1970s because of intensive insecticide use for H. armigera control (25, 26) but currently occurs at low population densities in Bt cotton (27). In a similar fashion, mirid bugs were previously suppressed by H. armigera insecticide use and only acquired pest status in cotton and other crops after Bt cotton adoption. Hence, areawide cultivation of transgenic crops may bring various (direct and indirect) effects on ecological status of different organisms, which should be assessed or anticipated in a comprehensive fashion.

In many parts of the world, transgenic crops such as Bt crops have come to dominate agricultural landscapes, and their landscape-level impact has been quantified on target pests (12, 28), nontarget organisms such as natural enemies (7, 29, 30), or charismatic species such as the monarch butterfly Danaus plexippus (31). However, few studies have described the impact of transgenic crops on nontarget insect pests (2, 13) or, more specifically, assessed their landscape-level effects (15, 32, 33). This study confirms reports of a landscape-level emergence of nontarget pests with the adoption of Bt crops resulting from reductions in insecticide applications. Our work highlights a critical need to predict landscape-level impacts of transgenic crops on (potentially) pestiferous organisms in future ecological agricultural risk assessment. Such more-comprehensive risk management may be crucial to help advance integrated pest management and ensure sustainability of transgenic technologies.

Supporting Online Material

Materials and Methods

Figs. S1 and S2

Tables S1 to S3


References and Notes

  1. Materials and methods are available as supporting material on Science Online.
  2. This work was supported by the National Key Basic Research Program (2006CB102004), the National Natural Science Foundation of China (30625028), the Key Project for Breeding Genetically Modified Organisms (2008ZX08012-004), and the Commonwealth Agricultural Scientific Research Project (200803011). We thank two anonymous reviewers for helpful comments and suggestions.
View Abstract

Stay Connected to Science

Navigate This Article