Two transgenic maize varieties, DKc3421Bt (event MON810) and DKc5143Bt (event MON88017), both from Monsanto Company (St. Louis, MO, USA), and their corresponding non-transformed near isolines, DKc3421 and DKc5143, were used for the experiments. DKc3421Bt plants express a truncated, synthetic version of the cry1Ab gene from B. thuringiensis ssp. kurstaki HD-1, targeting Lepidoptera, while DKc5143Bt plants express a synthetically modified cry3Bb1 gene from wild-type B. thuringiensis ssp. kumamotoensis EG4691, targeting Coleoptera (CERA 2010). Bt expression levels in the plants used for the experiments were measured using double-antibody sandwich enzyme-linked immunosorbent assays (DAS-ELISA) from Agdia (Elkhard Indiana, USA) (see below for details).
The four maize varieties were grown simultaneously under the same environmental conditions of 25 ± 1°C, 70 ± 5% RH, and a 16-h photoperiod. Different growth chambers were used for transgenic and non-transgenic maize plants to prevent spider mites from moving between treatments. Three seeds were planted in one plastic pot (12 l) filled with humus-rich soil (Ökohum-Staudenerde, Obi-Ter, Märwil, Switzerland). Plants were fertilized weekly with 400–800 ml of a 0.2% aqueous solution of Vegesan standard (80 g N, 70 g P2O5, and 80 g K2O per liter, Hauert HBG Dünger AG, Grossaffoltern, Switzerland) and watered as required.
Maize plants were used for the spider mite rearing once they had reached the 4–5 leaf stage and were removed when they reached anthesis.
Mixed stages of two arthropod species and eggs of a third species were used as prey for A. bipunctata. The two-spotted spider mite, T. urticae, was reared on Bt maize or the corresponding control plants in the growth chambers where the plants were raised. A colony of the pea aphid, Acyrthosiphon pisum (Harris) (Hemiptera: Aphididae), was kept as a continuous culture on broad bean plants (Vicia faba L.) in the glasshouse at 22 ± 3°C, 70 ± 5% RH, and a 16-h photoperiod. UV-irradiated eggs of the lepidopteran E. kuehniella were supplied by Biotop (Valbonne, France) and stored at 4°C. Mixed stages of A. pisum and eggs of E. kuehniella, both of which are high quality food for A. bipunctata larvae (Blackman 1967; De Clercq et al. 2005), were used to determine the food quality of mixed stages of T. urticae for A. bipunctata larvae.
Eggs of A. bipunctata were purchased from Andermatt Biocontrol (Grossdietwil, Switzerland). Upon arrival, egg masses were placed in Petri dishes (9 cm diameter) and kept at 25 ± 1°C, 70 ± 5% RH, and a 16-h photoperiod until larvae emerged. Once larvae had eaten their egg shell and started searching for food (≈12 h old), they were individually transferred to the experimental arenas and offered a single egg of E. kuehniella to enhance larval survival. After about 8 h, larvae were switched to their respective prey treatments.
Stock solutions of purified Cry1Ab in 25 mM carbonate/bicarbonate buffer (pH 10.6) and Cry3Bb1 in 10 mM sodium carbonate/bicarbonate buffer (pH 10.0) were provided by Monsanto Company. The Cry1Ab protein used for the bioassays was the purified trypsin-resistant core of Cry1Ab from recombinant B. thuringiensis strain SIC1837. The Cry3Bb1 protein used was the purified Cry3Bb1.11098(Q349R) produced by recombinant E. coli containing the pMON72735 expression plasmid. Both proteins were purified using SDS–Page/Densitometry. The Cry3Bb1 protein is equivalent to protein expressed in Bt maize (events MON863 and MON88017) in terms of biochemical or toxical characteristics (US EPA 2007). The purity-corrected Cry1Ab and Cry3Bb1 protein concentration was 2.1 and 6.3 mg ml−1, respectively. Bioactivity of both Cry proteins was confirmed by Monsanto Company in sensitive insect bioassays (unpublished information, provided with the protein certificate of analysis). Cry1Ab was tested on larvae of Helicoverpa zea (Boddie) (Lepidoptera: Noctuidae) and a 50% effect (EC50) on larval weight was reported for 0.012 μg trypsin resistant core. Cry3Bb1 was tested on larvae of Leptinotarsa decemlineata (Say) (Coleoptera: Chrysomelidae) and a concentration causing 50% mortality (LC50) of 0.4 μg protein was reported. In addition, the bioactivity of the Cry3Bb1 protein was confirmed in a L. decemlineata bioassay in our laboratory (Meissle and Romeis 2009a).
Lyophilized snowdrop lectin (Galanthus nivalis agglutinin, GNA) was obtained from Els van Damme (Ghent University, Belgium). A detailed description of the isolation from snowdrop bulbs is given by Van Damme et al. (1987).
The inorganic toxin potassium arsenate was purchased from Sigma–Aldrich (Buchs, Switzerland).
Experimental arenas consisted of small plastic Petri dishes (5 cm diameter, 1.3 cm height) covered with lids containing a mesh window (0.5 cm diameter with 0.2 mm openings) for ventilation. All experiments were conducted in a climate chamber at 25 ± 1°C, 70 ± 5% RH, and a 16-h photoperiod.
Consumption of E. kuehniella eggs by A. bipunctata
Visual observations were conducted to determine how young A. bipunctata larvae feed on E. kuehniella eggs. First and second instars of A. bipunctata were offered undamaged eggs and were observed with a stereomicroscope. Furthermore, the number of E. kuehniella eggs consumed during the first two instars was determined. Neonate A. bipunctata were fed daily ad libitum with undamaged E. kuehniella eggs, and the depleted egg shells were counted daily until A. bipunctata reached the third instar. Before eggs were provided to the predator, they were examined with a stereomicroscope, and damaged eggs were removed. The assay was performed using eight larvae.
Assessment of prey quality
A bioassay was conducted to investigate whether A. bipunctata larvae are able to develop by feeding exclusively on T. urticae. Neonates of A. bipunctata were fed ad libitum with one of the following food sources: E. kuehniella eggs or mixed stages (immatures and adults) of A. pisum or T. urticae. Mixed stages of A. pisum were brushed from bean leaves and added to the test arenas. For T. urticae, an approximately 9-cm2 maize (Dkc5143) leaf disc infested with spider mites was placed upside-down in each Petri dish. All food sources were replaced daily. The experiment was performed with 30 A. bipunctata larvae per treatment. Assessment of variables is described in the next section.
Tritrophic feeding study
This bioassay assessed the prey-mediated effects of Bt protein-expressing maize on A. bipunctata. Neonates were fed ad libitum with T. urticae reared on Bt-transgenic plants or on the respective non-transformed near isolines for several generations as described above. The bioassay was conducted in three subsequent runs with 15 replications each, resulting in a total of 45 A. bipunctata larvae per treatment. For every run, the A. bipunctata larvae were obtained from a different shipment and were fed with T. urticae reared on different plants.
In both the prey-quality and tritrophic feeding assays, A. bipunctata larvae were observed twice daily, in the morning and in the evening, until they reached the third instar. Then, larvae were frozen at −20°C, dried at 50°C for 24 h, and subsequently weighed on a microbalance (Mettler Toledo MX5, division d = 1 μg; tolerance ± 2 μg). Larval development time (L1–L3), dry weight of third-instars, and mortality were measured. We focused on the early larval instars for three main reasons: (1) early instars are generally regarded as being more sensitive to Bt Cry proteins than older instars or adults (Glare and O’Callaghan 2000), (2) T. urticae are not an optimum food for A. bipunctata larvae, and extended feeding on T. urticae would have caused unacceptably high control mortality levels, and (3) Schmidt et al. (2009) only reported Cry protein effects on the first instar in their bioassays.
Verification of the transfer of Cry proteins through the food chain
This experiment was conducted to verify that A. bipunctata larvae were exposed to Cry protein when feeding on Bt maize-fed spider mites and to quantify the protein levels. Samples of maize leaves and T. urticae were collected from Bt plants. Spider mites were collected in a tray kept below a maize leaf by shaking the leaf with a stick. Maize leaf material and T. urticae obtained from the 7th leaf from one plant were separately put into 1.5-ml micro-reaction tubes. For each of the three runs, two samples of leaves and T. urticae (7–10 mg fresh weight) were collected from different maize plants, resulting in a total of six samples. Leaf material and T. urticae were similarly obtained from control plants.
Neonate A. bipunctata were fed Bt maize-reared T. urticae for either 2 or 5 days, equivalent to first and second larval instars, respectively. As a control, larvae were fed T. urticae raised on non-transgenic maize plants. Five A. bipunctata were pooled into a 1.5-ml micro-reaction tube as one sample; for each of the three runs, two samples of both larval instars were analyzed.
All plant and insect samples were frozen and stored at −20°C for less than 11 weeks for Cry1Ab or Cry3Bb1 measurements. Bt protein levels in maize leaves, T. urticae, and A. bipunctata were measured using DAS-ELISA from Agdia following the protocol described in detail in Meissle and Romeis (2009b). Standard curves were made using solutions of the purified Cry1Ab or Cry3Bb1 proteins that were provided by Monsanto Company and for which purity was known. Protein concentrations in μg g−1 fresh weight (FW) were calculated from the standard curves using regression analysis. For the clear separation of positive readings from controls, the limit of detection (LOD) of the test was determined based on the standard deviation of the OD values of buffer-only controls multiplied by three (ICH 2005). Subsequently, the detection limit of each sample was calculated from the dilution, sample weight and amount of added buffer. Measurements of all Bt samples revealed ODs above the respective LOD.
Direct feeding study
A direct feeding study was conducted to evaluate the effect of purified Cry1Ab and Cry3Bb1 proteins on the pre-imaginal development of A. bipunctata and the weight of the emerging adult beetles. The bioassay aimed to cover some of the limitations of the tritrophic experiment that used T. urticae as a carrier for the Cry proteins, namely (1) T. urticae is a suboptimum food for A. bipunctata, (2) the tritrophic experiment had to be restricted to the first two intars of A. bipunctata, and (3) we could not include a positive control to show that the experimental set-up was able to detect adverse effects on the life-history parameters that were measured.
Bt protein concentrations were approximately 10 times higher than those measured in spider mites that had fed on Bt maize. Snowdrop lectin (GNA) and potassium arsenate were used as positive control treatments. That GNA can have a deleterious effect on larvae of A. bipunctata at high concentrations has been previously demonstrated (Hogervorst et al. 2006), and potassium arsenate is an inorganic compound that is highly toxic to insects and that is often used as a positive control in toxicological studies including those with ladybird beetles (Duan et al. 2002, 2006, 2008). The aim of the positive controls was to show that the ladybird larvae actually ingested the sucrose solution containing the test compounds and that the experimental set-up was able to detect adverse effects on the measured life-history parameters (Rose 2007).
The experiment was initiated with neonate A. bipunctata. Each larva received two droplets of a 2 M sucrose solution that was prepared with deionized water (L1 and L2: 0.5 μl, L3 and L4: 1 μl) or a sucrose solution containing 45 μg ml−1 Cry1Ab, 200 μg ml−1 Cry3Bb1, 10,000 μg ml−1 GNA, or 300 μg ml−1 potassium arsenate on the first day of each instar. After 24 h, larvae were transferred to clean Petri dishes and subsequently fed ad libitum with E. kuehniella eggs to continue development. After emergence, adults were frozen at −20°C and later dried at 50°C for 24 h. Subsequently, the dried adults were weighed on the microbalance and their sex was determined by dissection. Larval and pupal development time, adult dry weight, and mortality were recorded. The experiment was conducted with 34–41 A. bipunctata larvae per treatment.
In the assay in which A. bipunctata was fed with three different prey species, larval development time (L1–L3) and L3 dry weight were analysed using non-parametric statistics because data were not normally distributed (Kolmogorov–Smirnov test). Three pairwise comparisons were conducted using the Mann–Whitney U test, adjusted for ties, and significance levels were adjusted using the sequential Bonferroni procedure (Holm 1979). In the experiment that determined the impact of Bt-expressing maize on A. bipunctata, data for larval development time (L1–L3) and L3 dry weight were compared between the Bt maize and the respective non-Bt maize treatments using Student’s t-test.
In the direct feeding assay, larval and pupal development time and adult dry weight were analyzed between the control treatment and the respective insecticidal compounds in pairwise comparisons using Student’s t-test. Significance levels were adjusted using the sequential Bonferroni procedure.
In all assays, mortality data were compared using Chi-square tests. Again, significance levels were adjusted using the sequential Bonferroni procedure to correct for multiple pairwise comparisons.
For all tests, the overall α-level was set at 5%. Statistical analyses were conducted using the software package Statistica (Version 7.1, StatSoft Inc., Tulsa, OK, USA).
Statistical power analyses were performed before conducting the tritrophic feeding study using PASS (Version 2005, NCCS, Kaysville, UT, USA). Data (mean and standard deviation) recorded for the A. bipunctata larvae fed with T. urticae in the prey-quality bioassay were used for the calculations. With 45 replications, a power of 80%, and α = 0.05, the detectable differences between control and Bt treatment for larval development time (L1–L3), L3 dry weight, and larval mortality were 12, 8, and 29%, respectively. Calculations were based on two-sided t-tests (larval development time and weight) or Chi-square tests (mortality). The sensitivity of the experiment was judged to be sufficient because 80% power at α = 0.05 to detect a 50% effect is often stated as acceptable for risk assessment research (Rose 2007; EFSA 2009). Power analyses were not conducted for the direct feeding bioassay because the positive control treatments that were included provide some indication about the size of the detectable effect in that bioassay.