A Faunistic Approach to Assess Potential Side-Effects of Genetically Modified Bt-Corn on Non-Target Arthropods Under Field Conditions

Transcription

1 Biocontrol Science and Technology (March 2004), Vol. 14, No. 2, 129/170 A Faunistic Approach to Assess Potential Side-Effects of Genetically Modified Bt-Corn on Non-Target Arthropods Under Field Conditions M. P. CANDOLFI 1,K. BROWN 2, C. GRIMM 1, B. REBER 1 AND H. SCHMIDLI 1 1 Syngenta Crop Protection AG, CH-4002 Basel, Switzerland; 2 Ecotox Limited, Tavistock, Devon, UK (Received 31 October 2001; accepted 13 May 2003) A faunistic study investigating the potential side-effects of corn (Zea mays) genetically modified to express a truncated Cry1Ab protein derived from Bacillus thuringiensis subsp. kurstaki, on non-target arthropods was carried out under field conditions. The communities of non-target arthropods in the soil, on the leaves and flying in the crop area were monitored throughout the growing season. Water-treated, untransformed corn served as a control, and a spray application of a bacterial Bt insecticide (Delfin WG) and a synthetic insecticide (Karate Xpress) used to control the European corn borer (Ostrinia nubilalis; Lepidoptera: Pyralidae) acted as positive reference treatments. Results were analyzed using a principal response curve. Significantly lower infestations by the lepidopteran target species O. nubilalis were observed in the Bt-corn plots compared to the control. No effects of Bt-corn on the communities of soil dwelling and non-target plant dwelling arthropods were observed. A trend towards a community effect on flying arthropods was observed with lower abundance of adult Lepidoptera, flies in the families Lonchopteridae, Mycetophilidae and Syrphidae, and the hymenopteran parasitoids Ceraphronidae. Effects were weak and restricted to two sampling dates corresponding to anthesis. A short but statistically significant effect of Karate Xpress and Delfin was observed on the community of plant dwellers and a prolonged effect of Karate Xpress on the soil dwellers. Keywords: non-target arthropods, fauna, side-effects, Bt-corn, transgenic plants, Bacillus thuringiensis, Cry1Ab, pesticides, Delfin, Karate Xpress, lambda-cyhalothrin, biosafety Correspondence to: Dr. Marco Candolfi, RCC Ltd., Zelgliweg 1, CH-4452 Itingen, Switzerland. Tel.: / ; candolfi.marco@rcc.ch ISSN (print)/issn (online)/04/ DOI: / # 2004 Taylor & Francis Ltd

2 130 M. P. CANDOLFI ET AL. INTRODUCTION Insecticide formulations containing Bacillus thuringiensis (Bt) have been used in agriculture since the 1950s (Frankenhuyzen, 1993). Toxicity of these products to arthropods is predominantly due to a range of crystal proteins produced by the bacterium. Because most represent protoxins that need to be activated by insect midgut proteases to yield toxic fragments (Lereclus et al., 1993), susceptible insects must have a strongly alkaline mid-gut, proteolytic enzymes, and suitable tissue receptor-sites (Gill et al., 1992). Sprayed Bt-toxins degrade very rapidly under field conditions on foliage (Ignoffo & Garcia, 1978; Krieg, 1986; Glare & O Callaghan, 2000) and thus plant dwelling arthropods are exposed only during a very limited period of their life span. These factors contribute to the high selectivity of Btbased insecticides which are used to control lepidopteran and coleopteran pests, mosquitoes and black flies. Side-effects on non-target arthropods have only very rarely been observed under laboratory conditions. Peacock et al. (1998) found 27 of 42 lepidopteran species tested suffered increased mortality. Babrikova et al. (1982) and Babrikova and Kuzmanova (1984) reported higher mortality of the lacewing Chrysoperla carnea (Neuroptera: Chrysopidae) when adults, but not when larvae were fed with food contaminated with three different Btspray formulations under laboratory conditions. Petrova and Khrameeva (1989) reported highly deleterious effects of a spray formulation on the predatory mite Phytoseiulus persimilis (Acari: Phytoseiidae), and detrimental effects on Bracon brevicornis (Hymenoptera: Braconidae) were described by Temerak (1980) and Salama et al. (1991). However, the field use of Bt-based insecticides following good agricultural practices is generally considered safe (Flexner et al., 1986; Krieg, 1986; Melin & Cozzi, 1989; Croft, 1990, Meadows, 1993; Glare & O Callaghan, 2000). One constraint of conventional spray application of Bt-formulations is to reach pests that feed within the plant tissues. The economically most important such pest in corn is the European corn borer, Ostrinia nubilalis (Lepidoptera: Pyralidae). Eggs are laid on leaves, but larvae migrate to the relative protection of the whorls to develop, bore in to the stalk and can lead to plant breakage or lodging (Ely, 1993). In southern Europe the Mediterranean corn borer, Sesamia nonagrioides (Lepidoptera: Noctuidae), is one of the most damaging pests of corn and is also characterized by its endophytic larval behavior (González-Núñez et al., 2000). Corn has been genetically modified to express the CryIAb protein derived from B. thuringiensis, thus providing a high level of resistance to larval feeding of corn borers (Koziel et al., 1993; González-Núñez et al., 2000). Bt-corn was first commercially released in the US in 1996 (Roush, 1997) and was grown commercially on 9.9 million ha in 2002 (including stacked varieties expressing insect resistance and herbicide tolerance) in the USA, Canada, Argentina, South Africa, Spain, Honduras and Germany (James, 2002). In the Bt-corn used in this trial, the CryIAb protein is expressed in a truncated and thus partly inactivated form, under the control of the promoter phosphoenolpyrovate carboxylase in green tissues and a pollen-specific promoter in pollen, and is therefore present in significant levels in the leaves and the pollen. Highest levels related to plant fresh weight were found in the seedling stage, but on a per acre rate at anthesis (Fearing et al., 1997). Much of the selectivity data generated in the past 30 years on Bt-insecticide formulations can be applied to Bt-corn. However, potential side-effects on non-target arthropods need verification to account for the different concentration, mode and duration of exposure and the fact that CryIAb protein expressed in Bt-corn is already partially activated. A number of studies have addressed this issue in recent years. Laboratory studies conducted with Coleomegilla maculata (Coleoptera: Coccinellidae), Orius insidiosus (Heteroptera: Anthocoridae) and Chrysoperla carnea (Neuroptera: Chrysopidae) showed no detrimental effects on pre-imaginal development and survival when insect larvae were fed transgenic Bt-corn pollen (Pilcher et al., 1997). Hilbeck et al. (1998a) reported that C. carnea larvae raised with Bt-corn-fed prey / O. nubilalis or Spodoptera litoralis (Lepidoptera, Noctuidae) / showed

3 POTENTIAL EFFECTS OF Bt -CORN 131 higher pre-imaginal mortality (62%) than larvae raised on CryIAb-free prey (37%). Subsequent studies showed that C. carnea larvae fed with a liquid artificial diet mixed with the purified CryIAb toxin (concentration of 100 mg CryIAb/ml diet) exhibited a higher pre-imaginal mortality (57%) when compared to CryIAb-free diet (30%) (Hilbeck et al., 1998b). However, Lozzia et al. (1998a) found no differences in postembryonic developmental time, fecundity or survival of the aphid Rhopalosiphum padi (Homoptera: Aphididae) reared on Bt- versus untransformed corn and also reported no detrimental effects on pre-imaginal development and mortality of C. carnea fed on R. padi that had been reared on Bt-corn when compared to untransformed corn plants. A further tritrophic experiment in the laboratory also showed no effects when Orius majusculus (Heteroptera: Anthocoridae) was fed with the thrips Anaphothrips obscurus (Thysanoptera: Thripidae) reared on Bt-corn expressing CryIAb (Zwahlen et al., 2000), a result that was confirmed in the field when comparing abundances of nymphs and adults of O. insidiosus on Bt-corn with untransformed corn (Al-Deeb et al., 2001). Corn plants genetically modified to contain the event 176, encoding for a truncated CryIAb toxin, have been shown to release toxin into the soil environment by root exudates. These toxins can be persistent when bound to clay or humic acids, since they are protected against microbial degradation, and they have been shown to remain biologically active in bioasssays with tobacco hornworm, Manduca sexta (Lepidoptera: Sphingidae), and the Colorado potato beetle, Leptinotarsa decemlineata (Coleoptera: Chrysomelidae) (Saxena et al., 1999, 2002; Saxena & Stotzky, 2000). Under field conditions, Orr and Landis (1997) reported no significant negative effect of Bt-corn on the densities of O. insidiosus, coccinellids and chrysopids with a tendency towards higher densities of predators in Bt-corn fields compared to isogenic corn. Pilcher et al. (1997) reported no significant effects of Bt-corn on the abundance of species in the families Anthocoridae, Coccinellidae and Chrysopidae. Field data on additional non-target taxa were published by Lozzia and Rigamonti (1998) and Lozzia et al. (1998b). These authors reported no significant effects of Bt-corn on the abundance of carabid beetles, cicadellids, alticini beetles, parasitic hymenopterans, and syrphids. All of these studies above have compared non-target effects of Bt-corn on selected bio-indicator species with an untreated isoline. There has been little work done to compare the community-level effects of Bt-corn with the alternative of using conventional insecticides. Due to the complexity of the multitrophic interactions in the corn ecosystem, we have chosen a faunistic approach to generate additional data on potential side-effects of Bt-corn on non-target arthropods. The aim of this study was to determine whether Bt-corn affects non-target arthropods including non-target herbivores, predators, parasitoids, and other functional taxa under field conditions. We were interested in comparing potential effects of Bt-corn on non-target arthropods with the effects observed after an application of a Bt-spray insecticide formulation and a synthetic insecticide. MATERIALS AND METHODS The study design and methods described below take into account the general guidelines for field experiments described by the IOBC 1 (Hassan, 1992), the manual for field trials in plant protection (Anonymous, 1992), the Guideline to study within-season effects of insecticides on non-target terrestrial arthropods in cereals (MAFF (PSD) and HSE, 1995) and the ESCORT 2 /SETAC 3 guidance document on regulatory testing procedures for pesticides with non-target arthropods (Barrett et al., 1994). The study design and methods also comply with 1 IOBC: International Organisation for Biological and integrated Control of noxious animals and plants 2 ESCORT: European Standard Characteristics of non-target arthropod Regulatory Testing 3 SETAC: Society of Environmental Toxicology and Chemistry / Europe

4 132 M. P. CANDOLFI ET AL. the IOBC, BART 4 and EPPO 5 joint initiative principles for regulatory testing of field studies with non-target arthropods (Candolfi et al., 2000). The study was performed in compliance with Good Laboratory Practice (GLP) standards (OECD, 1997). Site Selection and Field Description The study was conducted in 1998 in Sassenay, a village near Chalon sur Saone, in the corn growing area of Burgundy in France. A graphical representation of the field including a description of the surrounding test area is presented in Figure 1. The size of the field was 15.1 ha. The soil type was characterized as a silty-clay (US classification, 45.6/49.7% clay, 37.8/ 42.1% silt and 11.2/14.1% sand) with a humus content of 4.5/5.2% and a ph of 7.1. During the 2 years preceding the trial, the experimental field was planted with corn and the following herbicides were used: tank mix of Lasso (5 L/ha, a.i. alachlor) and Quit (1 kg/ha, a.i. metsulfuron-methyl), tank mix of Banvel (0.4 L/ha, a.i. dicamba) and Quit (0.5 kg/ha, a.i. metsulfuron-methyl) and Milagro (1 L/ha, a.i. nicosulfuron). The field was also treated with the insecticide Carbolux (10 kg/ha, a.i. carbofuran) in 1996 and During the trial year, an N/P/K (20/14/15) fertilizer was applied at a rate of 850 kg/ha on 4 May After soil preparation (harrow) on 17 May 1998, corn seeds were planted 3 cm deep and spaced 13 cm within rows 75 cm apart on 18/19 May ( seeds/ha). Rows in all plots were planted in the same direction. An 18-m border strip around the perimeter was planted with untransformed control corn (see Figure 1). During 1998, the following herbicides were used in the experimental field: 21 May 1998, tank mix of Lasso (5 L/ha, a.i. alachlor) and Quit (1 kg/ha, a.i. metsulfuron-methyl), 29 May 1998, tank mix of Banvel (0.4 L/ha, a.i. dicamba) and Quit (0.5 kg/ha, a.i. metsulfuron-methyl) and 16 June 1998, Milagro (1 L/ha, a.i. nicosulfuron). No additional plant protection products were used except for the insecticide treatments included into the study design (see below). Treatments and Experimental Design Four treatments were set up in a completely randomized design. The four treatments were: 1. Bt-corn: seeds of the genetically modified Bt-176 corn hybrid Occitan Cb produced by Syngenta Seeds AG were coated with a fungicide (Thiram) and a bird repellent (Anthraquinone). Seeds were not coated with insecticides. This hybrid expresses a truncated CryIAb protein derived from B. thuringiensis subsp. kurstaki strain HD-1. CryIAb expression provides resistance to corn borers, particularly the European corn borer (O. nubilalis) and the pink stem borer (Sesamia sp.) (Koziel et al., 1993; Fearing et al., 1997). 2. Untransformed corn: the isogenic untransformed corn hybrid Occitan was used as the control. Seed treatments were the same as the Bt-corn treatment. 3. Untransformed corn sprayed with a microbial Bt-insecticide formulation: the same isogenic untransformed corn hybrid was treated with Delfin WG (Sandoz Agro AG, Basel, Switzerland) at a rate of 1.5 kg formulation/ha. Delfin contains Bt spores and crystals of the d-endotoxins CryIAa, CryIAb, CryIAc, CryIIA and CryIIB of Bacillus thuringiensis subsp. kurstaki strain HD-1 SA-11. Though Delfin is known to be effective against the European corn borer, it is not often used commercially to control this pest. The treatment was included to compare the potential effects of a Bt-spray formulation to Bt-corn, and the application rate is that recommended for use against Spodoptera spp. (Lepidoptera: Noctuidae) in corn. 4. Untransformed corn sprayed with a synthetic, broad spectrum contact insecticide: the same isogenic, untransformed corn hybrid was treated with Karate Xpress 4 BART: Beneficial Arthropod Regulatory Testing Group 5 EPPO: European and Mediterranean Plant Protection Organisation

5 POTENTIAL EFFECTS OF Bt -CORN 133 (active ingredient l-cyhalothrin), Sopra S.A. (Filiale de Zeneca S.A., Vélizy, France) applied at the recommended label rate of 0.4 kg formulation/ha. Karate Xpress is a pyrethroid insecticide currently used in the region to control the European corn borer. This treatment was included as a positive control. FIGURE 1. Plot layout and treatment allocation. Treatments 1 and 2 were replicated three times while treatments 3 and 4 were replicated two times. Plot size of each replicate varied between 1.18 and 1.69 ha. Treatment allocation and plot sizes are shown in Figure 1. Both insecticides were applied once during the growing season (14 July 1998) with a water volume of 200 L/ha. The corn plants had approximately

6 134 M. P. CANDOLFI ET AL. seven leaves and two nodes were detectable (BBCH plant growth stage 17/32 according to Weber and Bleiholder (1990) and Lancashire et al. (1991)). The control and the Bt-corn plots were treated on the same day with water (200 L/ha). Timing of the applications was chosen according to the recommendation of the regional plant protection service for treatments against European corn borer. The sprayer was equipped with an automatic computer sprayspeed controller with a boom of 36 m and Albuz green nozzles. Environmental conditions during treatment application were optimal (sunny with few clouds, temperature of 20.0/ 22.78C, a relative humidity of 25.2/31.6% and a wind speed of 0.5/1.1 m/s). Verification of CryIAb Expression in Bt-corn Certified seeds with known levels of CryIAb expression were used for the trial (Fearing et al., 1997). In order to confirm correct sowing and exclude possible contamination of control plots with transformed seeds, leaf samples were collected on 18 June 1998 (BBCH growth stage 12/14) and 11 August 1998 (BBCH growth stage 63/69) from all three replicated plots of the control (untransformed corn) and Bt-corn treatments. Twenty leaves were collected from different plants in the central part of each plot, immediately frozen with dry ice and stored at /208C until analysis. The presence/absence of the CryIAb protein in the samples was verified using a qualitative immunoassay ELISA kit (Immunovation Inc., Research Triangle Park, NC 27709, USA). The detection limit of the test is approximately 4 ng CryIAb protein per mg total protein (personal communication, Immunovation, Inc.). Six subsamples, each consisting of 10 leaf discs (approximately 8 mm diameter) cut from different leaves, were prepared from each leaf sample per plot. Each sub-sample was analyzed twice according to the method provided by the ELISA kit supplier. Corn Borer Infestation To determine European corn borer infestation, 1000 plants per plot were sampled on 18 September Plants were sampled in 10 different locations in each plot (near the pitfall traps, see below), by cutting 50 plants in five rows (10 plants per row) on both sides of the trap. The stems of the plants were cut longitudinally and the number of infested plants recorded. Sampling of Non-Target Arthropods Soil surface active arthropods were sampled using pitfall traps. Pitfall traps have been used extensively for studies on surface dwellers such as spiders, Collembola, centipedes, ants and beetles, especially Carabidae (Greenslade, 1964; Edwards et al., 1979; Clements et al., 1988). Whilst pitfall traps do not estimate actual populations (Briggs, 1961; Luff, 1975) their use is appropriate and routine in comparative studies, such as this one which also included a control and reference treatments. Pitfall traps modified from the design by Barber (1931) to facilitate emptying were used. The traps consisted of a funnel (top diameter 13.5 cm, bottom diameter 2.5 cm, height 10.5 cm), a plastic bottle (250 ml) screwed to the bottom part of the funnel and a plastic ring (24.5 cm ring diameter, 13.5 cm hole diameter) glued to the top of the funnel. The bottle was filled with 100 ml of a 50% ethylene glycol:water solution used to preserve the trapped arthropods. In order to place the traps into the soil, a hole of 16 cm diameter and 25 cm depth was dug and a plastic cylinder of the same size was positioned inside the hole. The trap was then placed inside the plastic cylinder (trap mouth level with the soil surface) and soil was carefully placed around the trap in such a way that no gaps between trap and soil occurred. The traps were covered with a roof (27/27 cm) positioned at approximately 5 cm from the soil surface to avoid flooding of the traps during rainy weather. Ten traps were set up in the central part of each plot. The traps were set up between rows and arranged in each plot as a W -shape as proposed by MAFF (PSD) and HSE guideline (1995) (Figure 1). Taxa were identified from eight sampling occasions during the growing season (Figure 2). Catches from water flooded traps were not included in statistical analyses.

7 POTENTIAL EFFECTS OF Bt -CORN 135 FIGURE 2. Trial schedule. A combination of the beating and funnel method (Wilson, 1962; Coineau, 1962) was used to sample plant dwelling arthropods. The plants were hit sharply several times with a stick and the arthropods dropping off were collected in a removal jar attached below a hole in the center of a cloth-funnel (top frame 60/40 cm, bottom hole 8 cm diameter, height 45 cm). A filter paper soaked with ethyl-acetate was positioned inside the jar to kill the arthropods. Arthropods collected were stored in a 50% ethylene glycol:water solution until identification. A total of 50 (first two samplings) or 100 (all other samplings) plants per plot and sampling time were sampled. Sampling involved beating two plants at the same time and repeating this in 50 different locations within the central part of each plot (from the center to a maximum of 20 m from the plot border). Sampling was performed seven times during the growing season (Figure 2). Yellow water traps were used to collect flying non-target arthropods in the crop. Traps (yellow containers of 32/18 cm and a height of 7 cm) were fixed to wooden poles. The trap position was adjusted during the growing season to compensate for plant growth, so that the traps were always approximately at mid-height position of the canopy. The traps were initially filled with 2 L of a 25% ethylene glycol:water solution with some drops of detergent and refilled at each sampling time. Arthropods were collected by pouring the ethylene glycol solution through a fine sieve. After collection, the arthropods were preserved in a 50% ethylene glycol:water solution until identification. Ten traps were set up between rows in the central part of each replicated plot and arranged in each plot as a W -shape. These traps were positioned at approximately 2 m distance from the pitfall traps. Sampling was performed six times during the growing season (Figure 2). Non-Target Arthropod Identification Taxa of all samples collected with pitfall traps (eight sampling intervals, 100 traps per assessment date, total of 800 samples) 6 and the beating-funnel method (seven sampling 6 Due to flooding of some pitfall traps, fewer samples were available for the 9 July 1998 sampling date (20 instead of 30 for the control, 21 instead of 30 for the Bt-corn, 16 instead of 20 for the Delfin and 15 instead of 20 for the Karate Xpress treatment).

8 136 M. P. CANDOLFI ET AL. intervals, 10 samples per assessment, total of 70 samples) were identified and analyzed. For the yellow water trap samples, taxa from five traps per plot (first to third and fifth to sixth sampling, total of 250 samples) 7 or all 10 traps (fourth sampling, total of 100) were identified and analyzed. Identification of beetles was carried out with reference to Auber (1971a, b), Lindroth (1974, 1985, 1986), Hammond (1988), Trautner and Geigenmuller (1987), Wachmann et al. (1995) and Unwin (1988). Any specimens that could not be identified to an appropriate level were identified by experts at the Natural History Museum, London, UK. Spiders were identified using Locket and Millidge (1951, 1953) Locket et al. (1974) and Roberts (1985a,b, 1987, 1995). Flies and other insects were identified using Colyer and Hammond (1951), Oldroyd (1970), Gauld and Bolton (1988), Gilbert (1993) and Chinery (1993). Aphids were identified using a field key developed by the Ministry of Agriculture, Fisheries and Food (MAFF, 1974) and with reference to Blackman and Eastop (1984). The level of identification can be seen in Tables 2/4. Data Analyses and Statistics The date of insecticide treatments (14 July 1998) is defined as day 0 on the time-scale used to analyze and describe the results. For each plot, the average abundance of the different arthropod taxa was calculated. 8 Logtransformed data of abundances (log 10 (abundance/1)) were used for statistical analyses. Graphical representation of the abundance of relevant taxa and of taxa groups against time was plotted using geometric means of abundance /1. The geometric means were calculated as follows: m((a 1 1)(a 2 1)...(a n 1)) 1=n where n is number of samples, a 1 is abundance in sample 1, a 2 is abundance in sample 2, etc. The geometric mean is considered to be more appropriate than the arithmetic mean since the log-transformed abundances are approximately normally distributed. Two main types of multivariate analyses were used to analyze the time- and treatmentdependent multivariate response of arthropods: a. Principal response curves (Van den Brink & Ter Braak, 1999): a principal response representing the whole community is estimated, which allows graphical comparison of the treated groups with the control group over time. The principal response can be explained as a virtual species which reflects the behavior of the whole community. The abundance of the principal response is a weighted sum of the abundances of the taxa. The taxa-specific weights (contributions to the PRC) are estimated such that the differences between treated groups and the control group become visible. Taxa with positive contributions to the PRC behave similarly to the principal response, i.e., if the principal response is increasing in a treated group, the taxa with positive contributions tend to increase as well. On the other hand, taxa with negative contributions tend to decrease if the principal response curve is increasing. Taxa with small contributions to the PRC are less important for showing differences between treated groups and the control group. A Monte-Carlo method (Crossvalidation/Jackknife) (Efron & Gong, 1983) was used to evaluate which treatment groups were statistically significantly different from the control for each of the sampling days. 7 Due to a very high abundance of taxa and individuals, it was decided to identify the arthropods from only 5 out of 10 traps. One trap in the Delfin treatment was lost on one sampling interval (5 th sampling). 8 To make the abundance counts comparable, the abundance of the first two beating samples / carried out on 50 plants as compared to 100 plants in later samples - were multiplied by a factor 2.

9 POTENTIAL EFFECTS OF Bt -CORN 137 b. Diversity analysis (Boyle et al., 1990): diversity indices allow a comparison of the community structures in the fields and provide information on the degree of diversity of the communities in the plots. The number of taxa and the Shannon diversity index were calculated. For the number of taxa, a separate ANOVA followed by Tukey s test (Box et al., 1978) was used to compare the effect of the different treatments for each sampling date. Let N ij be the average number of taxa in replicate plot i of treatment j for a given sampling day. Then the model used for the ANOVA can be written as: N ij /m j /o ij,werem j is the expected number of taxa for treatment j and the errors o ij are assumed to be normally distributed. The Tukey test allows to test for difference between m j and m k for all treatment pairs j and k, adjusting for multiple comparisons. It should be noted that the ANOVA is applied separately for each sampling day without adjustment for this multiple testing. The Shannon diversity index was calculated as follows: H (p 1 log(p 1 )p 2 log(p 2 )...p n log(p n )) where n is number of taxa, p 1 is proportion of individuals belonging to taxon 1, p 2 is proportion of individuals belonging to taxon 2, etc. ANOVA followed by Tukey s test (Box et al., 1978) was used to compare treatment effects on selected individual taxa and the Shannon diversity index for each sampling day. It should be noted that Tukey s test was done separately for each day and for each relevant taxon. Since a large number of tests were done, a considerable number of falsely significant results are to be expected in the univariate analyses. Corn borer infestation data were analyzed with Fisher s exact test (Zar, 1984) and treatment efficacy was calculated according to Abbott (1925): E% 1 t 100 c where E% is percent efficacy, t is average number of infested plants in the treatment group, and c is average number of infested plants in the control. For the statistical tests, the significance level was set to 5%. The software package SAS (Version 6.12 for Windows NT, SAS Institute Inc.) was used for the statistical analysis. RESULTS Verification of CryIAb Expression in Bt-corn ELISA immunoassay results confirmed the presence of CryIAb in all the Bt-corn samples analyzed at both sampling intervals. No CryIAb was found in the untransformed control plants. European Corn Borer Infestation European corn borer (O. nubilalis) infestation was very low in all treatments. Infestation rates of 1.57, 0, 0.10 and 0.85% were observed in the untransformed control, Bt-corn, untransformed corn treated with Karate Xpress and untransformed corn treated with Delfin (Table 1), respectively. A statistically significantly lower corn borer infestation was observed in all treatments when compared to the untransformed control plants. No corn borer infestation was observed in the Bt-corn plants (100% treatment efficacy). Efficacy of treatment with Karate Xpress and Delfin was 93.6 and 45.7%, respectively (Table 1).

10 138 M. P. CANDOLFI ET AL. TABLE 1. Mean percentage (9/SD) of corn plants infested with the European corn borer Treatment Percentage corn plants infested with corn borer Treatment efficacy (E) (Abbott, 1925) Control (untransformed corn) 1.579/0.50 a / Bt-corn 09/0 b 100% Untransformed corn treated with Karate Xpress 0.109/0.0 b 93.6% Untransformed corn treated with Delfin 0.859/0.07 c 45.7% abc : Different letters indicate statistically significant difference between treatments (two-tailed Fisher s exact test, P B/0.05). Soil Dwelling Arthropods A total of 76 taxa were identified from the pitfall trap samplings, including a total of individual arthropods. A list of the abundant taxa found in the pitfall traps and their density in the different treatments throughout the sampling season is presented in Table 2. The most abundant carabid beetles (Coleoptera: Carabidae) were Harpalus rufipes, Harpalus aeneus, Poecilus cupreus and Pterostichus melanarius. A very low abundance ( B/5 individuals per trap) of rove beetles (Coleoptera: Staphylinidae) was observed throughout the season in all experimental plots. The most abundant spider species were Alopecosa sp. (Araneae: Lycosidae), Oedothorax fuscus (Araneae: Linyphiidae) and Oedothorax apicatus (Araneae: Linyphiidae). The only abundant harvestmen species was Phalangium opilio (Opiliones: Phalangiidae). Collembola of the families Entomobryoidae and Sminthuridae and predatory mites (Acari) were found in very high number throughout the sampling period. Ants (Hymenoptera: Formicoidea) were also quite abundant. No statistically significant treatments effects (Tukey test, P/0.05) on the Shannon diversity indices of soil dwelling taxa (Figure 3) or on the number of soil dwelling taxa were observed throughout the season. Principal response curve (PRC) analysis for soil dwelling arthropods revealed that 69.7% of the variance observed in the system could be explained by the factor time and 10.8% by the treatment. PRC analyses showed no statistically significant treatment effects of Bt-corn or Delfin throughout the sampling period (Figure 4). Univariate analyses did not reveal statistically significant effects of the Bt-corn treatment on any individual soil dwelling taxon either. The PRC analysis showed significant treatment effects (goodness of fit R 2 /0.74, goodness of prediction Crossvalidation/Jacknife R 2 /0.62) of Karate Xpress when compared to the control treatment 14, 27 and 43 days after treatment application (Figure 4). Main contributors to the negative impact observed in the Karate Xpress treatment were millipedes (Diplopoda), the harvestman P. opilio, the carabid beetle P. melanarius, the spider O. apicatus and centipedes (Chilopoda). This indicates that the abundance of these species in the Karate Xpress treatment decreased after treatment application when compared to the control. On the other hand, some taxa (main contributor to PRC are Collembola, some carabid beetles and predatory mites) were more abundant in the Karate Xpress treatment when compared to the control (see Figure 4 for taxa with a negative PRC contribution). The population density of P. opilio is presented as a typical example where a negative response to Karate Xpress was observed (Figure 5) and the population density of Alopecosa sp. is presented as a typical example of a taxon where no effects were observed (Figure 6). Plant Dwelling Arthropods A total of 45 taxa were identified from the plant samplings. The total number of arthropods counted was A list of the abundant taxa found on corn plants and their density in the different treatments throughout the sampling season is presented in Table 3. The most

11 TABLE 2. Abundance of non-target soil dwelling arthropods (pitfall traps) in the different treatments throughout the growing season (all taxa exceeding an average of one individual per trap on at least one sampling date are given) Day (day 0/spray day) Order Family Taxon F T /26 / Acari Acari B Co BT De Ka Araneae Linyphiidae Linyphiidae juvenile B Co BT De Ka Araneae Linyphiidae Oedothorax apicatus B Co BT De Ka Araneae Linyphiidae Oedothorax fuscus B Co BT De Ka Araneae Linyphiidae Oedothorax spp. Female B Co BT De Ka POTENTIAL EFFECTS OF Bt -CORN Araneae Lycosidae Alopecosa sp. B Co BT De Ka Araneae Lycosidae Lycosidae incl. juvenile B Co BT De Ka

12 TABLE 2 (Continued) Day (day 0/spray day) Order Family Taxon F T /26 / Araneae Tetragnathidae Tetragnathidae B Co BT + ++ De + + Ka Chilopoda Chilopoda B Co BT De Ka Coleoptera Other Coleoptera I Co BT De Ka Coleoptera Carabidae Amara aeneus B Co ++ BT De + Ka + Coleoptera Carabidae Bembidion obtusum B Co + + BT De + Ka M. P. CANDOLFI ET AL. Coleoptera Carabidae Brachinus crepitans B Co BT De Ka Coleoptera Carabidae Carabid sp. incl. larvae B Co BT De Ka Coleoptera Carabidae Clivina fossor B Co BT De Ka

13 TABLE 2 (Continued) Day (day 0/spray day) Order Family Taxon F T /26 / Coleoptera Carabidae Harpalus aeneus B Co BT De Ka Coleoptera Carabidae Harpalus distinguendus B Co BT De Ka Coleoptera Carabidae Harpalus rufipes B Co BT De Ka Coleoptera Carabidae Poecilus cupreus B Co BT De Ka Coleoptera Carabidae Pterostichus B Co Melanarius BT De Ka Coleoptera Carabidae Pterostichus niger B Co ++ BT + ++ De ++ + Ka ++ POTENTIAL EFFECTS OF Bt -CORN Coleoptera Carabidae Trechus sp. B Co BT De Ka Coleoptera Elateridae Elateridae H Co BT De Ka

14 TABLE 2 (Continued) Day (day 0/spray day) Order Family Taxon F T /26 / Coleoptera Silphidae Necrophorus sp. I Co BT De ++ + Ka + + Coleoptera Staphylinidae Staphylinid larva B Co BT De Ka Coleoptera Staphylinidae Staphylinidae B Co BT De Ka Coleoptera Staphylinidae Xantholinus B Co BT De Ka Collembola Entomobryoidea I Co BT De Ka M. P. CANDOLFI ET AL. Collembola Sminthuridae Sminthuridae I Co BT De Ka Diplopoda Diplopoda B Co BT De Ka Hymenoptera Formicoidea H Co BT De Ka

15 TABLE 2 (Continued) Day (day 0/spray day) Order Family Taxon F T /26 / Isopoda Isopoda I Co BT De Ka Opiliones Phalangiinae Phalangium opilio B Co BT De Ka No symbol, 0; Individuals per plot +, 0 andb/1; ++,1toB/5; +++,5toB/10; ++++,10toB/100; +++++, 100 tob/1000. F, functional role of the taxa in the corn ecosystem; B, beneficial arthropods; H, herbivores; I, indifferent taxa; T, treatment; Co, control (untransformed corn); BT, Bt corn; De, Delfin; Ka, Karate Xpress. POTENTIAL EFFECTS OF Bt -CORN 143

16 144 M. P. CANDOLFI ET AL. FIGURE 3. Shannon diversity indices of soil dwelling taxa per trap collected with the pitfall traps throughout the sampling season: (m) untransformed corn (control); (w) Bt-corn; (') untransformed corn treated with Delfin; (j) untransformed corn treated with Karate Xpress; Day 0, spray day. No statistically significant differences between treatments were observed (Tukey test, P/0.05). abundant herbivore taxa were aphids (Rhopalosiphum maidis; Homoptera: Aphidae), leafhoppers (Zyginidia scutellaris; Homoptera: Cicadellidae; and other cicadellid species) and thrips (Thysanoptera). Predatory bugs (Orius sp.; Heteroptera: Anthocoridae), lacewings (Neuroptera: Chrysopidae) and spiders (Araneae) were the most abundant predatory taxa. No statistically significant effects (Tukey test, P/0.05) of Bt-corn and Delfin on the number of plant dwelling taxa were observed throughout the season. In the Karate Xpress treatment a statistically significantly lower number of taxa was recorded 2 days after treatment application when compared to the control (Tukey test, P B/0.05); however, recovery was observed 15 days after treatment application. No statistically significant treatment effects (Tukey test, P/0.05) on the Shannon diversity indices of plant dwelling taxa (Figure 7) were observed throughout the season (Tukey test, P/0.05). Principal response curve (PRC) analyses for plant dwelling arthropods revealed that 61.9% of the variance observed in the system could be explained by the factor time and 14.3% by treatment. PRC analyses for plant dwelling arthropods showed no significant treatment effects of Bt-corn when compared to the control throughout the sampling period (Figure 8) and univariate analyses did not reveal statistically significant effects of the Bt-corn treatment on any individual plant dwelling taxon either. PRC analyses showed significant treatment effects (goodness of fit R 2 /0.65, goodness of prediction Crossvalidation/Jacknife R 2 /0.50) of Karate Xpress and Delfin when compared to the control treatment 2 days after application (Figure 8). Main contributors to the PRC were parasitic wasps in the Eulophidae and Proctotrupoidea (Hymenoptera), the bugs Orius sp. (Heteroptera: Anthocoridae), Nabis ferus (Heteroptera: Nabidae) and Deraeocoris sp. (Heteroptera: Miridae), the leafhopper Z. scutellaris, thrips (Thysanoptera), the ladybird beetle Scymnus sp. (Coleoptera: Coccinellidae), soldier beetles (Coleoptera: Cantharidae),

17 POTENTIAL EFFECTS OF Bt -CORN 145 FIGURE 4. Principal response curve analysis for soil dwelling organinsms: zero line of the y-axis/ untransformed corn (control); (w) Bt-corn; (') untransformed corn treated with Delfin; (j) untransformed corn treated with Karate Xpress; Day 0, spray day. Statistically significant treatment effects when compared to control are circled (goodness of fit R 2 /0.74, goodness of prediction Crossvalidation/Jacknife R 2 /0.62). phorid flies (Diptera: Phoridae), and several spider taxa (Araneae: Linyphiidae, Theridiidae, Thomisidae) (see Figure 8 for single taxa contribution to PRC). Univariate analyses on individual taxa showed significant decreases of Eulophidae and Thomisidae in the Delfin treatment on day 2, and of Eulophidae, Proctotrupoidea, Cantharidae and Thomisidae on day 2 and Orius sp. on days 2 and 15 in the Karate Xpress treatments (Tukey test, P 5/0.05). The population density of Orius sp., a natural enemy of the European corn borer, is presented as a typical example where a negative response to the Karate Xpress treatment was

18 146 M. P. CANDOLFI ET AL. FIGURE 5. Population density of Phalangium opilio (Opiliones: Phalangiinae). Plotted are the geometric means of abundance /1 per trap against time: (m) untransformed corn (control); (w) Bt-corn; (') untransformed corn treated with Delfin; (j) untransformed corn treated with Karate Xpress; Day 0, spray day. No statistically significant differences between treatments and the control were observed (Tukey test, P/0.05). FIGURE 6. Population density of Alopecosa sp. (Araneae: Lycosidae). Plotted are the geometric means of abundance /1 per trap against time: (m) untransformed corn (control); (w) Bt-corn; (') untransformed corn treated with Delfin; (j) untransformed corn treated with Karate Xpress; Day 0, spray day. No statistically significant differences between treatments and the control were observed (Tukey test, P/0.05).

19 TABLE 3. Abundance of non-target plant dwelling arthropods on corn plants in the different treatments throughout the growing season (all taxa exceeding an average of five individuals per plot on at least one sampling date are given, as well as Proctotrupoidea and C. carnea larvae) Day (day 0/spray day) Order Family Taxon F T / Araneae Linyphiidae Linyphiidae Juvenile B Co BT De Ka Araneae Linyphiidae Oedothorax spp. female B Co BT De Ka Araneae Theridiidae Theridiidae sp. B Co BT De Ka Araneae Thomisidae Thomisidae B Co BT De Ka Coleoptera Cantharidae Cantharidae B Co BT De Ka ++ POTENTIAL EFFECTS OF Bt -CORN Coleoptera Elateridae Elateridae H Co BT De Ka Diptera other Nematocera H Co BT De Ka

20 TABLE 3 (Continued) Day (day 0/spray day) Order Family Taxon F T / Heteroptera Anthocoridae Orius sp. B Co BT De Ka Homoptera Aphididae Rhopalosiphum maidis H Co BT De Ka Homoptera Cicadellidae Other Cicadellidae incl. nymphs H Co BT De Ka Homoptera Cicadellidae Zyginidia scutellaris H Co BT De Ka Hymenoptera Proctotrupoidea B Co BT De Ka M. P. CANDOLFI ET AL. Hymenoptera Eulophidae Eulophidae B Co BT De Ka Neuroptera Chrysopidae Chrysoperla carnea larvae B Co BT + De + + Ka

21 TABLE 3 (Continued) Day (day 0/spray day) Order Family Taxon F T / Neuroptera Chrysopidae Other Chrysopidae B Co BT De Ka Thysanoptera Thysanoptera H Co BT De Ka No symbol, 0; Individuals per plot +, 0 andb/1; ++,1toB/5; +++,5toB/10; ++++,10toB/100; +++++, 100 tob/1000. F, functional role of the taxa in the corn ecosystem; B, beneficial arthropods; H, herbivores; I, indifferent taxa; T, treatment; Co, control (untransformed corn); BT, Bt corn; De, Delfin; Ka, Karate Xpress. POTENTIAL EFFECTS OF Bt -CORN 149

22 150 M. P. CANDOLFI ET AL. FIGURE 7. Shannon diversity indices of taxa recorded on corn plants throughout the sampling season: (m) untransformed corn (control); (w) Bt-corn; (') untransformed corn treated with Delfin; (j) untransformed corn treated with Karate Xpress; Day 0, spray day. No statistically significant differences between treatments and the control were observed (Tukey test, P/0.05). observed (Figure 9) and population densities of R. maidis and Chrysopidae are presented as typical examples of taxa where no effects were observed in any treatment (Figures 10 and 11). Arthropods Flying in the Crop Area A total of arthropods were identified in 71 taxa from the yellow-water trap samples. A list of the abundant taxa and their density in the different treatments throughout the sampling season is presented in Table 4. The most abundant taxa were found in the orders Diptera, Heteroptera, Hymenoptera, Homoptera and Thysanoptera. No statistically significant treatment effects (Tukey test, P/0.05) on the Shannon diversity indices (Figure 12) or the number of taxa of flying non-target arthropod taxa were observed throughout the season when compared to the control. Principal response curve (PRC) analyses for flying arthropods revealed that 78.5% of the variance observed in the system could be explained by the factor time and 9.0% by the treatment. A trend for an effect of the Bt-corn on the flying arthropod community was evident on days 43 and 57, although the treatment effects were not significant with P values of 0.60 on day 43 and 0.17 on day 57 (Figure 13). The taxa contributing strongest to the PRC were the dipteran families Lonchopteridae, Empididae, Phoridae, Calliphoridae, Muscidae and Dolichopodidae, the thrips (Thysanoptera), the predatory bug Orius sp. (Heteroptera: Anthocoridae), the hymenopteran families Ceraphronidae, Ichneumonidae and Eurytomidae, and Lepidoptera. Following up the community response with univariate analyses of each individual taxon, significant reductions of Exechia nigrascutellata (Diptera: Mycetophilidae) (Figure 14), Mycetophila fungarum (Diptera: Mycetophilidae) (Figure 15) and Ceraphronidae (Figure 16) were found on day 43 and Episyrphus balteatus (Diptera: Syrphidae) (Figure 17), Lonchopteridae (Figure 18) and adult Lepidoptera (Figure 19) on day 57 (Tukey test, P 5/0.05). Further taxonomic identification of Lepidoptera revealed that one of the species present was the satellite moth Eupsilia satellitia (syn. E. transversa) (Lepidoptera:

23 POTENTIAL EFFECTS OF Bt -CORN 151 FIGURE 8. Principal response curve analysis for plant dwelling organisms: zero line of the y -axis, untransformed corn (control); (w) Bt-corn; (') untransformed corn treated with Delfin; (j) untransformed corn treated with Karate Xpress; Day 0,/spray day. Statistically significant treatment effects when compared to control are circled (goodness of fit R 2 /0.65, goodness of prediction Crossvalidation/Jacknife R 2 /0.50). Noctuidae). The other noctuid species present in the samples could not be determined due to loss of wing scales during storage. Lepidoptera density was however very low in all treatments since yellow water traps are not designed to trap Lepidoptera. These data should therefore be interpreted with care. Significant increases in the Bt-corn were observed on day 57 in C. carnea and on day 72 in Cicadellidae (Homoptera) (Tukey test, P 5/0.05). The community response in the Delfin and Karate Xpress plots was different, with Delfin treatments behaving very similarly to the control and Karate Xpress treatments behaving

24 152 M. P. CANDOLFI ET AL. FIGURE 9. Population density of Orius sp. (Heteroptera: Anthocoridae). Plotted are the geometric means of abundance /1 per trap against time: (m) untransformed corn (control); (w) Bt-corn; (') untransformed corn treated with Delfin; (j) untransformed corn treated with Karate Xpress; Day 0, spray day. Statistically significant treatment effects when compared to control are circled (Tukey test, P 5/0.05). FIGURE 10. Population density of Rhopalosiphum maidis (Homoptera: Aphididae). Plotted are the geometric means of abundance /1 per trap against time: (m) untransformed corn (control); (w) Bt-corn; (') untransformed corn treated with Delfin; (j) untransformed corn treated with Karate Xpress; Day 0, spray day. Statistically significant treatment effects when compared to control are circled (Tukey test, P 5/0.05).

25 POTENTIAL EFFECTS OF Bt -CORN 153 FIGURE 11. Population density of Chrysopidae (Neuroptera). Plotted are the geometric means of abundance /1 per trap against time: (m) untransformed corn (control); (w) Bt-corn; (') untransformed corn treated with Delfin; (j) untransformed corn treated with Karate Xpress; Day 0, spray day. No statistically significant differences between treatments and the control were observed (Tukey test, P/0.05). contrary to the Bt-corn treatment, based on a reduction of Cecidomyiidae (Diptera), which was significant on day 57 (Tukey test, P 5/0.05). DISCUSSION Soil Dwelling Non-Target Arthropods No statistically significant treatment effects on the diversity indices of soil dwelling taxa were observed throughout the season. This indicates that the diversity of taxa of soil dwelling arthropods was not influenced either by the Bt-corn or by the Delfin or Karate Xpress treatments. Principal response curve analysis for soil dwelling arthropods showed no significant treatment effects of Bt-corn or Delfin when compared to the untreated control. This indicates that the behavior of the whole soil arthropod community in the Bt-corn and Delfin treatments was similar to the control. PRC analysis showed statistically significant effects of Karate Xpress 14, 27 and 43 days after treatment application when compared to the control. At subsequent sampling intervals no effects of Karate Xpress were observed. This indicates that the negative impact of Karate Xpress observed on millipedes, the harvestman P. opilio, the carabid beetle P. melanarius, the spider O. apicatus and centipedes, was only temporary and recovery occurred. These results show that the experimental methodology was appropriate to detect treatment effects on soil dwelling arthropods. Our findings confirm the results of earlier field studies where transitory negative effects of Karate on spiders and carabid beetles were reported (Dinter & Poehling, 1992; Wehling & Heimbach, 1994; Dinter, 1995; Dinter & Poehling, 1995; Wehling & Heimbach, 1995; Cole & Pilling, 1997; Van den Berg et al., 1998). CryIAb toxin can be released into the rhizosphere via root exudates of Bt-corn containing transformation event 176 (Saxena et al., 1999, 2002). Toxins may bind to clays and humic