Climate change, transgenic corn adoption and field-evolved resistance in corn earworm

Increased temperature anomaly during the twenty-first century coincides with the proliferation of transgenic crops containing the bacterium Bacillus thuringiensis (Berliner) (Bt) to express insecticidal Cry proteins. Increasing temperatures profoundly affect insect life histories and agricultural pest management. However, the implications of climate change on Bt crop–pest interactions and insect resistance to Bt crops remains unexamined. We analysed the relationship of temperature anomaly and Bt adoption with field-evolved resistance to Cry1Ab Bt sweet corn in a major pest, Helicoverpa zea (Boddie). Increased Bt adoption during 1996–2016 suppressed H. zea populations, but increased temperature anomaly buffers population reduction. Temperature anomaly and its interaction with elevated selection pressure from high Bt acreage probably accelerated the Bt-resistance development. Helicoverpa zea damage to corn ears, kernel area consumed, mean instars and proportion of late instars in Bt varieties increased with Bt adoption and temperature anomaly, through additive or interactive effects. Risk of Bt-resistant H. zea spreading is high given extensive Bt adoption, and the expected increase in overwintering and migration. Our study highlights the challenges posed by climate change for Bt biotechnology-based agricultural pest management, and the need to incorporate evolutionary processes affected by climate change into Bt-resistance management programmes.


Introduction
the soil, it overwinters in areas south of 40°N [38]. However, its overwintering and distribution ranges are predicted to expand due to warmer temperatures [11], leading to changes in arrival and intensity of infestations in northern latitudes [39].
For all Bt corn expressing Cry toxins, the high dose requirement for resistance management is not achieved for H. zea which is more tolerant to the Bt toxins. Owing to this 'moderate dose' expression, the risk of evolution of resistance increases in areas where Bt field corn is widely adopted, and H. zea overwinters successfully [40]. Recently, Dively et al. [25] described and characterized the field-evolved resistance in H. zea to multiple Cry toxins, including pyramided traits expressed in transgenic sweet corn. During 1996-2016, the rate of increase for the indicators of resistance was significantly different between Bt and non-Bt hybrids with steeper increases for Bt than non-Bt hybrids (see fig. 1a-d in Dively et al. [25]), clearly demonstrating the temporal trend of resistance accumulation. Many sweet corn farmers in Maryland, USA either have stopped growing these Bt hybrids or are applying more insecticide sprays to compensate for the reduced control efficacy. With resistant populations in Maryland, which historically represented northern range of overwintering [41], H. zea resistance to Bt corn may spread and increase.
Simulations predict that the regional deployment of Bt crops affects average gene frequencies, and therefore incomplete technology adoption can slow resistance evolution [42,43]. Conversely, extensive adoption of Bt technology increases selection pressure for H. zea resistance evolution. Contrastingly, models also predict that resistance could evolve rapidly in other Helicoverpa sp. when Bt crops are rare in the landscape, rather than common [44]. However, field evaluations of these simulation models quantifying the relationship between H. zea resistance to Cry proteins and Bt crop adoption in the landscape are not available. Such information is necessary to predict vulnerability of crops to increased damage due to the spread of Bt resistant H. zea under climate change scenarios. In this study, we use two decades of data on indicators of resistance to Bt sweet corn and examine the role of temperature anomaly and Bt adoption on field-evolved resistance in H. zea to Cry toxins.

Indicators of field-evolved resistance
We used the data provided by Dively et al. [25,45] that indicate field-evolved resistance of H. zea to Bt Cry proteins (measures of control efficacy and changes in susceptibility). Dively et al. [25] established sentinel plots of Cry1Ab expressing Bt sweet corn hybrids paired with non-Bt isogenic hybrids, and evaluated H. zea infestations and abundances during 1996-2016 in Maryland, USA. We used percentage of ears damaged by H. zea, mean kernel area consumed, mean instars and proportion of late instars (4th-6th instars) in Bt hybrids (henceforth damage, consumption, instar and late instars, respectively) as indicators of field-evolved resistance to Cry1Ab in sweet corn. See Dively et al. [25] for details on field protocols and data collection procedures. We used the mean nightly captures of H. zea moths in black light traps during 1996-2014 (henceforth counts) as measure of population abundance.

Transgenic corn acreage and temperature anomaly data
As predictors of field-evolved resistance, we used Bt corn (including field corn) acreage in the agricultural crop districts of the study sites in Maryland (https://www.nass.usda.gov/Charts_and_ Maps/Crops_County/boundary_maps/md.gif) as a measure of Bt adoption. The Bt events and Cry proteins expressed in field corn are also expressed in sweet corn (except for Cry1F). For e.g. Attribute (expressing Cry1Ab toxin, event Bt11) and Attribute II (expressing Cry1Ab and Vip3A, event MIR162) hybrids from Syngenta Seeds, and Performance Series (PS) hybrids (expressing the Cry1A.105 and Cry2Ab2 toxins, event MON89034) from Seminis Seeds. Therefore, Bt field corn acreage, which represented 81% of all corn acreage in the USA in 2015, is a good measure of the evolutionary selection pressure for resistance development in H. zea against Bt Cry proteins. Area under corn production differed among the study sites, representing a range of selection pressure for resistance development. Crops (particularly corn) are planted extensively around the sites on the Eastern Shore, while western Maryland sites include urban areas. We estimated the Bt corn acreage in the agricultural districts as the product of total area of corn planted each year [46] and the national average percentage of Bt corn for that year [47].
We compiled the temperature anomaly data for the respective climatic divisions (https:// www.ncdc.noaa.gov/monitoring-references/maps/us-climate-divisions.php). Temperature anomalies, particularly warmer temperatures, during the growing season (spring to summer) are related to increased number of generations [7], earlier arrival of migratory insect pests and increased feeding and crop damage [8]. We downloaded data on temperature anomaly during the growing seasons (spring-summer; April-September) of the study period , for each agricultural district, from NOAA-National Climatic Data Center (http://www.ncdc.noaa.gov/cag/time-series).

Statistical analysis
Linear mixed effects models (LMMs) were used to analyse the influence of temperature anomaly and Bt acreage on H. zea abundance and on indicators of field-evolved resistance. The LMMs contained each of the indicators (counts, damage, consumption, instar and large instar) as response variables, site as a random effect accounting for repeated measurement, and the interaction, additive, or individual effect of (log) Bt acreage and temperature anomaly as fixed effects. We tested the interactive influence of Bt acreage and temperature anomaly prior to analysing additive or individual influences. Where applicable, the response variables were square root transformed to conform to linear model assumptions.
We determined the statistical significance of the fixed effects through Wald F-tests with Kenward-Roger approximation. Also, a likelihood-ratio based pseudo-R 2 value [48] was calculated for each model as a measure of the proportion of total variance explained. We ensured model appropriateness through diagnostic plots of the models visualizing within-group residuals (standardized residuals versus fitted values, normal Q-Q plots, histograms of residuals) and estimated random effects (normal Q-Q plots and pairs-scatter plot matrix; see [49]; pp. 174-197). Package 'lme4' [50] was used to construct the LMMs, and pseudo-R 2 values were calculated using package 'MuMIn' [51]. Estimated coefficients were extracted from the LMMs and plotted using 'ggplot2' [52] and 'visreg' [53], all in R program [54].

Results
The interactive influences of temperature anomaly and log Bt acreage significantly influenced the abundance of H. zea over the past two decades (F 1,77 = 35.6, p < 0.001, pseudo-R 2 = 0.53; see table 1 for coefficients). Results show contrasting effects of the variables on abundance, with the effect of temperature anomaly decreasing with increasing Bt acreage. While the overall population trend was negative, abundance was highest at low Bt acreage and high-temperature anomaly (figure 1a).
For indicators of field-evolved resistance, damage was significantly positively associated with additive effects of temperature anomaly and log Bt acreage (temperature anomaly: F 1,89 = 20.3, p < 0.001; Bt acreage: F 1,87 = 15.9, p < 0.001, pseudo-R 2 = 0.49). Damage to corn ears increased with temperature anomaly (figure 2a) and log Bt acreage (figure 2b; table 1 for coefficients). Kernel consumption of Bt hybrids was significantly influenced positively by the interactive effects of temperature anomaly and log Bt acreage (F 1,78 = 7.06, p = 0.009, pseudo-R 2 = 0.20; table 1 for coefficients). Kernel consumption was highest with interactive effects of high-temperature anomaly and log Bt acreage (figure 1b). The effect of temperature anomaly on kernel consumption increased with Bt acreage. Mean instar was significantly positively associated with Bt acreage (F 1,86 = 13.7, p < 0.001, pseudo-R 2 = 0.15; figure 3; table 1 for coefficients). The proportion of late instars was significantly positively associated with the interactive effects of temperature anomaly and log Bt acreage (F 1,147 = 23.62, p < 0.001, pseudo-R 2 = 0.47; table 1 for coefficients). Effect of temperature anomaly on the proportion of late instars increased with Bt acreage (figure 1c).

Discussion
We analysed how temperature anomaly and Bt adoption affects insect biology, transgenic crop-pest interactions, and agricultural pest management through transgenic biotechnology. Our results suggest that temperature anomaly and its interaction with increased Bt adoption accelerated the field-evolved resistance in H. zea to Bt corn expressing Cry1Ab toxin. We provide field validations of predictions from simulation models for H. zea [42,43], demonstrating that the elevated selection pressure from extensive adoption of Bt corn in the mid-Atlantic region has enabled H. zea to develop resistance to Cry1Ab expressing Bt varieties. The 'moderate dose' delivered by Bt crops through Cry proteins provides incomplete control [40] and increases selection pressure for resistance development. Decreasing compliance with the 20% structured refuge requirement for single Cry protein expressing hybrids [55], and the shift to refuge-in-the-bag corn hybrids that contain a blend (e.g. 95 : 5 Bt : non-Bt seed) which is less effective than structured refuge [56], also increases selection pressure. While the evolution of   resistance with increased Bt adoption was expected, our study highlights the important interactive effect of climate change (temperature anomaly) for resistance development. Extensive regional adoption of Bt corn suppressed the population abundance of H. zea in Maryland, similar to other reports on regional suppression of pests by Bt crops [14,17,[21][22][23]. Although populations were reduced overall, H. zea abundance increased at low Bt acreage with temperature anomaly probably through increased voltinism [7,30], as they go through two to three generations in Maryland during the growing season. These interactive and/or additive influences affected the field-evolved resistance to Cry1Ab expressing Bt sweet corn. Previous anecdotal report alludes to hot weather and high Bt acreage as reasons for the failure of Bt crops (Bt cotton) against H. zea in Texas [57].
The mechanisms through which the interactive influences of increased Bt acreage and temperature anomaly influence evolution of H. zea resistance to Cry protein in Bt corn are complex as climate change impacts span spatial scales (local to global) and levels of biological organization (molecular to biosphere) [58,59]. Reduced toxicity at higher temperatures is observed for some conventional insecticides [60][61][62]. Also, physiological mechanisms such as increased rates of metabolic processes, excretion and the reactivity at site of action play an important role in insect resistance to conventional pesticides at warmer temperatures (e.g. Musca domestica L. resistance to pyrethroids; [60]). Regarding Bt crops, the activity and expression of alkaline phosphatase in the midgut binding sites is associated with resistance against Cry1F and Cry1Ac in many target lepidopteran pests including H. zea [63][64][65]. The cross-resistance between the Cry1A proteins (1Ab, 1Ac, Cry1A.105) in H. zea [66,67], and increased activity of alkaline phosphatase at higher temperatures [66][67][68] probably catalysed the field-evolved resistance to Cry1Ab particularly given high selection pressure from increased Bt adoption.
At the local spatial scale, increased feeding at high temperatures [7,8] resulted in high injury to Bt sweet corn. High temperatures may stress Bt crops and possibly degrade soluble proteins, resulting in decline of Cry1A toxin expression by the plant, and thereby reduced field efficacy [32]. Similarly, the accelerated phenology, population growth and voltinism [5,7,30]  We used temperature anomaly (compared to twentieth-century averages) during the growing season (April-September) of each year of the study period as a predictor. Contour plots depict the predictions from LMMs with each dependent variable scaled on the right axis, with lower values in blue and higher in red.
buffer against population reduction from Bt toxin during resistance evolution. Additionally, increased overwintering survival due to warmer temperatures, particularly reduced length of time spent at nearzero temperatures [39], affect the population dynamics of target species in Bt crops and can increase the resistant allele frequency that carries over to the following spring season [68]. Concurrently, at regional/continental spatial scales, the south-north migration of moths already selected for resistance on Bt corn and cotton, that also are Bt tolerant [69], probably enhanced the development of Cry1Ab resistance in Maryland populations. The high regional deployment of Bt corn in the mid-Atlantic region, as well as further south along the migratory route of H. zea, probably exacerbated the selection pressure for resistance development.
Our results have important implications for understanding the risk of resistance increasing and spreading, sustainability of Bt technology and regional insect pest management strategies for climate change preparedness. The expansion of overwintering range north of 40°N, increased migration due to climate change [11,25], and the selection pressure exerted by extensive Bt acreage in mid-Atlantic, north central and northern great plains regions in the US exacerbates the risk of resistance increasing and spreading. These results highlight the necessity for adapting existing regulatory policies on refuge requirement and resistance monitoring. Helicoverpa zea sampling for monitoring and testing for resistance is currently concentrated in the southern US regions where both Bt corn and Bt cotton are grown.    on Bt sweet corn, as indicator of field-evolved resistance to Bt corn. We estimated Bt corn acreage as the product of total area of planted corn for each year in each agricultural district and the national average percentage of Bt corn for that year. Points represent the raw data, blue line represents the predictions from LMMs, and dotted lines denote the upper and lower confidence levels (95% CI).
Expanded resistance monitoring for all Cry proteins registered against corn earworm in all high corn production regions would help detect resistance development as the overwintering range of corn earworm expands due to climate change.
Our study highlights risks and limitations of Bt technology for agricultural pest management in the context of climate change. Polyphagous, migratory species such as H. zea can benefit from warmer temperatures [7,8], and increased Bt exposure might lead to resistance development particularly in insects for which Cry proteins are moderately toxic. Hence there is a need to incorporate the evolutionary processes affected by climate change into resistance management programmes. Given the role of temperature and increased Bt adoption for resistance development, incorporating insect resistance management plans into broader integrated pest management strategies (biological control, crop rotation, etc.) could ameliorate selection pressure, and facilitate sustainability of Bt technology [24]. The evolutionary basis of H. zea resistance to Cry1Ab is driven by environmental conditions, and a complex genetic basis underlies insect resistance to Cry1A Bt toxins in Lepidoptera species [70]. We endeavoured to analyse the role of key environmental factors that also changed over time and may drive resistance evolution (namely Bt adoption and climatic conditions). While we highlight the statistical association of temperature anomaly and Bt acreage, we acknowledge the correlative nature of our study and the need for further research and detailed experiments. Detailed experiments relating toxicity and resistance development at varied Bt concentrations to temperature conditions are needed to clearly identify the mechanisms through which Bt adoption and climate change interactively influence field-evolved resistance.
Data accessibility. All the relevant data used for analysis are provided as electronic supplementary material, Dataset S1. Authors' contributions. P.D.V. conceived the study, compiled the data, carried out statistical analysis, produced the illustrations and drafted the manuscript; G.P.D. helped design the study, and draft the manuscript. Both authors gave final approval for publication.
Competing interests. Although this research and the preparation of this article were not financially supported by organizations that may gain or lose financially through its publication, one of the authors (G.P.D.) has received financial support for other research or consulting from companies including Syngenta, Monsanto, Bayer CropScience, Dow AgroScience and Pioneer/DuPont. Disclaimer. The views expressed in this document are solely those of the authors, and do not reflect those of either US EPA or AAAS. Products or commercial services mentioned in this publication are for reference purposes only, and do not represent endorsement by the authors, the US EPA, or AAAS.