Effectiveness of the natural resistance management refuge for Bt-cotton is dominated by local abundance of soybean and maize

Resistance of H. zea populations to Cry1Ac

To measure variation in resistance of H. zea populations across field locations during 2017 and 2018, larval offspring of insects collected from non-Bt maize were subjected to a diet-overlay bioassay containing a diagnostic concentration of Cry1Ac (29 µg/cm²) corresponding to the mean LC₉₅ of four Cry1Ac susceptible H. zea populations. Overall, larval survivorship varied significantly among years (Fig. 1).

The wide range in survivorship (1 to 96%) among locations reveals high levels of spatial variation in resistance of local H. zea populations to the Cry1Ac toxin (Fig. 2). Because the populations included in our study were collected as larvae from ears of non-Bt maize and their offspring were subjected to the bioassay, the response did not reflect effects of selection for resistance on the parents of larvae used in our bioassays.

Effects of prior year cotton, maize, and soybean abundance on Cry1Ac resistance

In our bioassay, higher survival indicates higher resistance to Cry1Ac. Selection for resistance to Bt toxins in H. zea occurs almost exclusively in maize and cotton, and selection is reduced at higher levels of relative abundance of non-Bt hosts in the landscape. The relationships are complex, reflecting how differences in the relative abundance of each of these crops affect H. zea populations and their associated inter-crop source-sink dynamics. These dynamics, in turn, influence the intensity of selection on the local populations.

The main effects of proportional areas of cotton and maize on H. zea survival in the bioassay are highly significant (P < 0.0003 and P < 0.0001, respectively) but neither the main effect of soybean nor the cotton * soybean interaction effect is significant (P = 0.2429 and P = 0.0967, respectively). Importantly, the cotton * maize and the maize * soybean interaction effects are both highly significant (P < 0.0001; Supplementary Table S1).

To examine the cotton * maize interaction, probability of survival was fit to the abundance of cotton at different levels of maize abundance and at the mean abundance of soybean within a 1-km radius of the collection site during the prior year (Fig. 3). The result shows a small increase in larval survival in response to increases in abundance of cotton when maize abundance is very low (0.05), but a negative relationship between larval survival and cotton abundance that becomes increasingly strong as maize abundance increases. The latter can be interpreted to indicate that effectiveness of the local natural refuge increases with increasing abundance of maize. Ineffectiveness of the local refuge when proportion of maize was low (0.05) likely results from low maize abundance limiting the local H. zea population early in the season, resulting in an increase in the relative proportion of the population that immigrated to the study area after being subjected to selection elsewhere.

The role of soybean as the primary constituent of the natural refuge is revealed by the maize * soybean interaction in which larval survival is largely independent of soybean abundance in the local landscape at low levels of maize abundance but shows an increasingly strong negative response as maize abundance increases (Fig. 4). Together these results indicate that the effect of proportional area of any one of either maize, cotton, or soybean on the resistance level of the local H. zea population is dependent on the proportional areas of the other two.

Maize abundance strongly influences the size of the dispersing H. zea populations that infest subsequent hosts based on their relative attractiveness and availability in the local landscape^25,26,30. Cotton and soybean are most attractive to H. zea when flowering, which typically coincides with moth dispersal from maize. Flowering soybean is more attractive than flowering cotton³¹. Although populations that develop on soybean are influenced by numerous factors³², larval populations infesting soybean in North Carolina are much higher than those infesting cotton²⁶. Hence, the number of susceptible moths produced per unit area on soybean is expected to exceed that of resistant moths completing development on cotton²⁰. Populations infesting cotton are not only selected for resistance to Bt toxins but are also increasingly targeted by insecticide applications to reduce damage by Bt-resistant larvae¹⁴. To the extent that these applications reduce the size of H. zea populations under selection for Cry toxin resistance in cotton, they can be expected to increase the effective size of the natural refuge. The contribution of soybean to the natural refuge for Bt Cry toxins in cotton varies among locations depending on the relative abundances of maize, cotton, and soybean in the local landscape. We did not specifically consider other crop and non-crop hosts of H. zea in the landscape, but other hosts contributing to the natural refuge are represented in the proportional abundance within the 1-km buffer not occupied by maize, cotton, and soybean. Our results indicate that although soybean is only one component of the natural refuge, variation in soybean abundance is an important determinant of effectiveness of the natural refuge across the sites studied.

The effect of maize abundance on natural refuge efficacy may be explained by its dual role as a site for selection for cross resistance among Cry1 toxins^{11,12,21,28,29} and as a nursery producing moths that disperse from maize fields to infest cotton, soybean, and other plants comprising the natural refuge^25,26,30. Helicoverpa zea moths produced on maize include those subjected to selection on Bt varieties, as well as susceptible moths that developed on non-Bt varieties planted to meet structured-refuge requirements for Bt maize. Increasing effectiveness of the natural refuge for Bt cotton with increasing maize abundance points to the importance of the structured refuge for Bt maize as a key source of Cry-toxin susceptible moths dispersing into the natural refuge for Bt cotton. Our results suggest that poor compliance by growers with the structured refuge requirement for maize, which is widespread^12,27, not only compromises its effectiveness, but also undermines effectiveness of the natural refuge for Bt-cotton.

Before maize and cotton expressing pyramided toxins were adopted beginning in 2007, an unstructured natural refuge was considered inadequate for single-toxin cotton, and a structured refuge was required. The current US EPA approved resistance management plan for Bt cotton relies on expression of multiple toxins, which are not high dose against H. zea, and the presence of an unstructured natural refuge^12,21. Implicit in the rationale for a natural refuge is that the total abundance and diversity of non-Bt crops and natural hosts remain sufficiently stable over time and space to ensure the refuge is functional. A Scientific Advisory Panel, convened in 2006 by the US EPA¹⁹, critically examined the feasibility of a natural refuge for Bt-cotton and recognized the potential influence of variation in abundance of non-Bt crop hosts within the natural refuge on the effective size of the refuge for H. zea, but empirical studies documenting this relationship have been lacking. The Panel also recognized that this variation might differ in importance among cotton production regions. Current levels of sensitivity of H. zea populations to Cry1Ac have likely been influenced by selection over many years, including that by single-gene Cry1Ac cotton and Cry1Ab maize, and more recently by selection and cross-resistance among related pyramided toxins in maize and cotton. Our findings indicate that on-going selection is important and that despite the capacity of H. zea for long-distance dispersal²⁴, the effects of local abundance of soybean in relation to maize and cotton, acting within the larger natural refuge, influence the resistance levels of local H. zea populations at least two generations later in the following growing season.

Because abundance of soybean as a component of the natural refuge varies among US cotton production regions²⁶, the relationships we observed for soybean are expected to vary in importance among regions as well. However, we believe the implications of our findings are general. They indicate that effectiveness of natural refuges can be expected to vary among locations and years in response to differences in local abundances of relevant Bt and non-Bt crop hosts of H. zea, and that maize as a source of H. zea infesting cotton and non-Bt hosts in the landscape can be especially influential in determining natural refuge effectiveness. Examination of the proportional abundances of maize, cotton, and soybean in a 1-km radius surrounding our sample sites over a 5-year period encompassing our study (2014–2018) reveals considerable variation in abundances among years as well as among locations (Supplementary Table S3). Given the availability of geospatial crop production data, trends in relative crop composition could be leveraged to better understand where crop components of unstructured natural refugia could be manipulated to improve Bt toxin resistance management in economically important polyphagous pests, like H. zea.

These findings have important implications for commercial and regulatory decisions regarding potential future deployment of Bt Cry toxins in soybean and Bt Vip3 toxins targeting H. zea in maize, cotton, and potentially other crops^12,21, as well as for assumptions regarding the efficacy of natural refuges for resistance management. They provide evidence that in the North and South Carolina field crop production systems studied, effectiveness of the natural refuge for cotton depends on the abundance of soybean as the dominant agricultural component of the natural refuge and on the relative abundance of maize in the local landscape. This dependency suggests that refuge requirements for Bt maize targeting H. zea in cotton production areas should reflect the important role of maize as a source for the H. zea populations that are critical to the functioning of the natural refuge.

Source: Ecology - nature.com