Plant- and ground-active arthropods were surveyed over two years from a two-acre cotton farm in Athens, GA supporting whole plot treatments of conventional- (CT) and no-tillage (NT) with subplot treatments of Bt and non-Bt cotton and a winter cover crop of rye or clover. Although neither Bt cotton nor its residue affected total abundance, richness or diversity, three families were significantly affected. Conversely, richness and diversity of non-target arthropods were both significantly higher in NT than in CT plots and a clover cover crop supported higher levels of abundance, richness and diversity during 2005. An interaction was detected between cover crop and cotton residue type with higher abundance, richness and diversity occurring in clover plots with Bt cotton residue than in clover plots with non-Bt residue. Arthropods contributing to significant differences at the community level include plant- and soil surface-active predators, detritivores, pollen feeders, and non-target herbivores. While this study demonstrates slight variations in the impact of Bt cotton under various agricultural management schemes it also reveals that cotton type is the least important factor shaping non-target arthropod communities relative to the other management strategies employed at this particular site.
The effects of genetically modified crops on non-target organisms have been the topic of much recent research, particularly crops modified to express toxins of the bacterium Bacillus thuringiensis Berliner (Bt) (O’Callaghan et al. 2005). The potential effects include toxicity of pest species to their natural enemies due to secondary ingestion of the Bt endotoxin, loss of prey/hosts for predators/parasitoids, decreased prey/host suitability due to endotoxin effects, and effects on soil meso- and macrofauna due to ingestion of either decomposing transgenic plant material or root-exuded endotoxins. O’Callaghan et al. (2005) report that results from both laboratory and field studies reveal mixed effects of Bt cotton on non-target arthropods. This suggests that if Bt cotton does affect non-target arthropods, the impact may be both subtle and context specific.
Most studies addressing the effects of Bt crops on non-target arthropods are limited to simple comparisons between Bt and non-Bt crops. Given the mixed nature of results from past experiments, an examination of the effects of Bt cotton relative to other important agronomic factors that shape arthropod communities may shed light on the true ecological impact of these crops in the context of the overall cropping system. Beyond the context of simple experiments, Bt crops are grown in a complex landscape matrix of agronomic management strategies that include variation in tillage regimes and in cover crops. To begin addressing potential interactions among crop choices and management strategies, I examined the relative contributions of tillage and cover crops to the foliar and ground-dwelling arthropod communities in Bt and non-Bt cotton fields.
The purpose of this study is to compare the effects of Bt and non-Bt cotton on the arthropod communities that are active on plants and at the soil surface as a function of the tillage and cover crop practices in which the respective crops are embedded. By doing so, we can assess context-specific effects of the cotton types on the communities.
I tested the following hypothesize:
H1 – tillage and cover crop type will exert much more influence on arthropod abundance, richness and diversity than will the use of Bt cotton
H2 – tillage and cover crop strategy will not significantly alter the effect of cotton type on arthropod abundance, richness or diversity
All research was conducted at the Horseshoe Bend Long Term Research in Environmental Biology (LTREB) research site of the University of Georgia, Athens, GA. The 14-hectare site, surrounded by a bend in the North Oconee River, is comprised of eastern deciduous forest and a 0.8-ha agricultural plot. Agroecosystem research has been conducted at the site since 1978 when the site was divided into eight plots (28 m x 28 m) that were randomly assigned to either a conventional tillage (CT) or no-tillage (NT) management regime. The current study ran from January 2004 through October 2005. For the duration of our study, each of the eight main plots was subdivided into four subplots (14 x 14 m each) to support a fully-factorial experiment of summer cotton crop (Bt or non-Bt) crossed with winter cover crop (clover or winter wheat/rye) nested within the eight tillage plots (four conventional and four no-tillage). The experiment therefore consisted of a total of 32 subplots. Arthropod sampling was conducted each year in April (cover crop) and October (cotton crop) as described below.
During each sampling period, three sweep net samples were taken within each subplot with a net 40 cm in diameter. Each sample consisted of ten sweeps from the crop canopy taken over a five-meter transect. Three transects were run in each subplot for a total of 30 sweeps per subplot and all transects were located near the center of each subplot to avoid edge effect. Care was taken to not cross over unsampled transects. Arthropods were transferred to a kill bucket with 70 % ethanol and then placed in plastic vials for storage and identification.
During each sampling period, four pitfall traps were placed in each subplot (128 total traps). Each trap consisted of a 16-oz. plastic drink cup with a 10 cm diameter opening. A plastic specimen cup containing 30 mL of 70% ethanol was inserted into each cup. The cups were buried at ground level before placement of specimen cups to avoid collection of soil and debris. Plastic funnels were then placed in each trap. Cardboard roofing was secured approximately three inches above each trap to minimize desiccation and to keep traps from filling with rainwater. Traps remained in the field for 24 hours, after which their contents were emptied into plastic vials and held until identified.
Bt Cotton Effects
In isolation, neither Bt cotton (October sampling) nor its residue (April sampling) had any impact on total arthropod abundance, richness or diversity during either year of the study. However, there were significant effects of both cotton and cotton residue type on specific arthropod families during both years of the study. During October of 2005 the mite suborder Prostigmata (F1, 114=4.52; P=0.035) was significantly more abundant/active from pitfall traps in non-Bt cotton plots than from Bt cotton plots while the family Gryllidae was more abundant/active from Bt rather than non-Bt plots (F1,12=6.06; P=0.029). Cotton residue type also had a significant affect on the abundance of tarnished plant bugs, Lygus lineolaris, Palisot de Beauvois, and black fleahoppers Halticus bractatus, Say, (Hemiptera: Miridae) during April of 2004 (F1,12= 7.57; P=0.017).
Arthropod richness and diversity were significantly higher in No-till plots than in conventionally tilled plots in pitfall samples during October 2004 (F1,6 =11.10, P=0.015, F1,6 =13.18, P=0.011). Similarly, arthropod richness was higher in No-till plots than in conventionally tilled plots during the October of 2005 (F1,6 =8.47; P=0.027).
During both years of the study, as well as from both cotton and cover crop seasons, many individual arthropod families were significantly more abundant/active in no-till than in conventional-till plots. The taxa included mites in the suborders Prostigmata (April 2004 pitfall traps F1,6 =6.04; P=0.049, October 2005 pitfall traps F1,6 =6.24; P=0.046) and Oribatida (F1,12 =11.74; P=0.005), as well as the families Cicadellidae (F1,6 =6.45; P=0.044), Delphacidae (F1,24 =6.36; P=0.018), Miridae (October 2004: F1,6 =11.46; P=0.014, April 2005: F1,12 =34.38; P<.0001), Aphididae (F1,6 =7.24; P=0.036), Syrphidae (F1,24 =15.00; P=0.0007), Sminthuridae (October 2005: F1,12 =17.06; P=0.001, October 2005: F1,6 =6.06; P=0.049), Hypogastruridae (F1,12 =16.35; P=0.001), Linyphiidae (F1,6 =5.65; P=0.05) and Thomisidae (F1,12 =8.41; P=0.013).
There were also five taxa significantly more abundant/active in conventional-till plots than in no-till plots. These taxa were primarily adult Dipterans (Tipulidae F1,6 =7.34; P=0.035, Ceratopogonidae F1,6 =5.94; P=0.05, Chironomidae F1,12 =6.39; P=0.026), but also included thrips in the suborder Terebrantia (F1,24 =18.05; P=0.0003) and the collembolan family Isotomidae (F1,12 =15.53; P=0.002).
Cover Crop Effects
Independent effects of cover crop on arthropod communities were only detectable in the sweep-net samples of 2005. Arthropod abundance, richness and diversity were significantly higher in clover than in rye plots during April of 2005 (F1,12 =12.47, P=0.004, F1,12 =41.06, P<0.001, F1,12 =47.49, P<0.001).
Indicator Species Analysis followed by mixed model analysis of the abundance of individual taxa for April of 2005 revealed that the following taxa were significantly more abundant/active in clover than in rye subplots: Diptera (Drosophilidae F1,6 =7.16; P=0.036, Chironomidae F1,88 =26.97; P<.0001, Sciaridae F1,88 =5.07; P=0.026 and Dolichopodidae F1,24 =5.26; P=0.03), Hemiptera (Miridae F1,6 =15.53; P=0.007 and Cicadellidae F1,24 =9.42; P=0.005), Hymenoptera (Apidae: Apis mellifera, L., F1,6 =18.42; P=0.005 and Formicidae: chiefly Solenopsis invicta, Buren, F1,6 =8.31; P=0.028), Coleoptera (Nitidulidae F1,6 =10.97; P=0.016 and Tenebrionidae F1,6 =11.95; P=0.013), and Orthoptera (Acrididae F1,6 =9.35; P=0.009). Aphids however were more abundant in rye than in clover subplots (F1,6 =7.82; P=0.031).
Interactions were detected in cover crop samples from both 2004 and 2005. The impact of Bt cotton as either a live crop or as plant residue varied based on the accompanying tillage and cover crop type. Arthropod abundance was significantly lower in clover subplots with non-Bt cotton residue than in clover subplots with Bt residue during both years of the study (F1,82 =10.92, P=0.001; F1,18 =5.48; P=0.031). During April of 2004 arthropod richness and diversity were also lowest in clover subplots with non-Bt cotton residue (F1,24 =5.33, P=0.02; F1,18 =7.44, P=0.01).
Examination of the abundance of different arthropod families revealed a large number of significant interactions. However, these effects varied tremendously and rarely coincided with significant interactions at the whole community level. In cases where family level interaction effects did coincide with whole community effects (April 2004), L. lineolaris and H. bractatus (Hemiptera: Miridae) were the primary contributors.
Although the relative importance of tillage, crop and residue type varied based on year and sample type (pitfall traps or sweep net samples) tillage and cover crop appeared to have the largest impact on arthropods, both at the whole-community and individual family level. Tillage strategy had significant effects on arthropods during both years (and seasons) of the study, affecting 16 different arthropod families. Although cover crop effects were limited to one year of the study (April 2005), 12 different arthropod families were significantly affected within the sampling period. The effects of Bt cotton and its residue were, however, limited to three taxa.
Transgenic crops remain controversial and like all types of agricultural production, pose potential environmental risks. This study demonstrates that Bt cotton may have some, context specific impact on non-target arthropod communities when examined under varying cover crop and tillage practices. However, the effect of Bt was negligible relative to the role of tillage and cover crop choice in shaping the non-target arthropod community of this particular system.
Areas needing additional study
INTERACTION EFFECTS AND FUTURE WORK
The interaction reported in the present study between cotton litter and cover crop was not contingent upon tillage treatment, suggesting that the location of litter biomass above or belowground is not key to the interaction. (Flores S. 2005) demonstrated that Bt cotton litter decomposes more slowly than does the litter of non-Bt cotton. The authors, having accounted for multiple factors affecting decomposition (i.e., lignin, C:N), concluded that the presence of the Bt endotoxin was the only factor that could have influenced the rate of decomposition. Moreover, the interaction detected at the whole-community level was not consistent across individual families suggesting that the mechanisms involved may be taxon- as well as context-specific. Although we do not understand the mechanism underlying the interaction at present, it will be the focus of future work.