Environmental Pollution 142 (2006) 212e216 www.elsevier.com/locate/envpol

Consequences for Protaphorura armata (Collembola: ) following exposure to genetically modified Bacillus thuringiensis (Bt) maize and non-Bt maize

Lars-Henrik Heckmann a,1, Bryan S. Griffiths b, Sandra Caul b, Jacqueline Thompson b, Marianne Pusztai-Carey c, William J. Moar d, Mathias N. Andersen e, Paul Henning Krogh a,*

a National Environmental Research Institute, Department of Terrestrial Ecology, Vejlsøvej 25, PO Box 314, DK-8600 Silkeborg, Denmark b Scottish Crop Research Institute, Department of Soil Plant Dynamics, Invergowrie, Dundee DD2 5DA, UK c Case Western Reserve University, Cleveland, OH 44106, USA d Auburn University, Department of Entomology and Plant Pathology, Auburn, AL 36849, USA e Danish Institute of Agricultural Sciences, Research Centre Foulum, PO Box 50, DK-8830 Tjele, Denmark Received 2 June 2005; received in revised form 5 October 2005; accepted 13 October 2005

Protaphorura armata performed equally well when reared on two Bt and three non-Bt maize varieties.

Abstract

Studies on the effect of genetically modified Bacillus thuringiensis (Bt) crops on true soil dwelling non-target are scarce. The objective of this study was to assess the influence of a 4-week exposure to two Bt maize varieties (Cry1Ab) Cascade and MEB307 on the collembolan Protaphorura armata. For comparison three non-Bt maize varieties, Rivaldo (isogenic to Cascade), Monumental (isogenic to MEB307) and DK242, and two control diets based on baker’s yeast (uncontaminated and contaminated with Bt toxin Cry1Ab) were also tested. Due to a lower C:N ratio, individuals reared on yeast performed significantly better in all of the measured endpoints than those reared on maize. P. armata performed equally well when reared on two Bt and three non-Bt maize varieties. Although there were no negative effects of Bt maize in this experiment, we recommend future studies on Bt crops to focus on species interactions in long-term, multi-species experiments. Ó 2005 Elsevier Ltd. All rights reserved.

Keywords: GMO; Cry1Ab; Collembola; Life history traits; Population

1. Introduction dwelling (i.e. euedaphic) arthropods that may be directly affected by the traits of many GM crops (Yu et al., 1997). While the use of genetically modified (GM) crops is Bacillus thuringiensis (Bt) insecticidal Cry toxin (e.g. Cry1Ab) currently at debate in Europe, several countries (e.g. USA, is one of the traits that have been genetically engineered into Argentina, China and Australia) are growing GM crops com- crops. Cry toxins are active in insects when ingested orally as mercially on a large scale. Numerous studies concerning the their mode of action is in the midgut. Genetically modified Bt impact of GM crops on terrestrial organisms have been per- crops express the Cry toxin constitutively throughout the en- formed (e.g. Losey et al., 1999; Saxena and Stotzky, 2001a). tire plant (Groot and Dicke, 2002). In the insect midgut Cry But only few have investigated the impact on true soil toxins synthesised by B. thuringiensis, need to be proteolyti- cally cleaved to form the active d-endotoxin before binding to insect midgut receptors (Schnepf et al., 1998). Binding of * Corresponding author. Tel.: þ45 89 20 15 88; fax: þ45 89 20 14 13. E-mail address: [email protected] (P.H. Krogh). the d-endotoxin to the receptors is a requirement for insecti- 1 Present address: The University of Reading, School of Biological Sciences, cidal activity, although binding alone is not sufficient for Environmental Biology, PO Box 228, Reading RG6 6AJ, UK. toxicity. When binding is followed by pore formation in the

0269-7491/$ - see front matter Ó 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.envpol.2005.10.008 L.-H. Heckmann et al. / Environmental Pollution 142 (2006) 212e216 213 midgut membrane, the midgut cells are disrupted and the in- maize varieties (Table 1); dry baker’s yeast contaminated with Bt toxin sect dies. The toxicity of different Cry toxins is specific, e.g. (Cry1Ab); and uncontaminated dry baker’s yeast (Table 1). Each replicate con- Cry1A mainly targets lepidopterans, whereas Cry3A targets sisted of a Petri dish (5.4 cm diam., 1.2 cm height, with a plaster of Paris and activated charcoal (8:1 weight:weight) covering the bottom) with 10 adult Chrysomelidae (Coleoptera). Thus, Cry toxins are toxic to P. armata (w4e10 weeks old). Synchronisation of P. armata is very difficult, some species whilst being more or less harmless to others and thus the body area of the individuals in each Petri dish was measured using (Schnepf et al., 1998). Digital Image Processing (DIP). The experimental duration was four weeks Below ground, the rhizosphere is a hot spot with respect to during which P. armata were kept in a climate room with a constant temper- abundance, species diversity and biological interactions of the ature of 20 1 C and a light:dark regime of 12:12 h. The relative humidity was w100%, which was adjusted weekly by adding demineralised water. soil biota (Coleman et al., 1988). The presence of Cry1Ab toxin P. armata were starved for 2 d prior to the experiment, after which they has been observed in the rhizosphere as root exudate from Bt were fed one of the above-mentioned diets once a week (w4 mg DW per rep- maize (Saxena et al., 1999). No apparent effects have been licate). Somatic growth was measured as body surface area weekly using DIP identified in earthworms (Vercesi et al., in press), nematodes, (Krogh et al., 1998). During DIP, the individuals were anaesthetised with CO2. protozoa, bacteria and fungi (Saxena and Stotzky, 2001a) since The total number of juveniles produced was counted manually after six weeks, two weeks after removal of the surviving adults, thus estimating the total re- the Cry toxins primarily target insects. Some studies have been production during the four weeks as mean egg development takes w14 d at performed on the effects of Bt crops on non-target arthropods 19 C(Bengtsson et al., 1985). Survival and population growth rate (pgr) (NTA), but they mainly focused on aboveground insects (e.g. was estimated after four weeks. Jasinski et al., 2003). So, as concluded by Groot and Dicke The maize root material was obtained from frozen plants harvested after (2002), studies on the effect of Bt crops on euedaphic NTA field trials performed in Denmark in 2001 (varieties Rivaldo (non-Bt) and Cascade (Bt)) and 2002 (DK242 (non-Bt), Monumental (non-Bt) and MEB307 are scarce. (Bt)) (Griffiths et al., in press). The roots were washed, dried and lyophilised Collembolans comprise a large group of NTA, and play over night at 40 C under a vacuum of w30 mTorr. Subsequently, the root a key role in the functioning of the soil environment (Larink, tissue was finely ground in a mortar and kept dry, dark and cold (w5 C) until 1997). Thus, our main objective was to examine the impact of use. Treatments with baker’s yeast were processed according to the method of root tissue from Bt (Cry1Ab) and non-Bt maize varieties Sims and Martin (1997). One gram of dry baker’s yeast was suspended in a Petri dish in 5 ml of distilled water containing 500 mg of purified Cry1Ab toxin. offered as food to the collembolan Protaphorura armata The suspension was quickly frozen in a 80 C freezer, and subsequently (Onychiuridae). Moreover, two diets based on dry baker’s lyophilised as described above. Solubilized, trypsinized (i.e. proteolytically ac- yeast were included as a series of controls to evaluate the over- tivated) and HPLC purified Cry1Ab1 toxin was applied. Protein activity of all performance of P. armata on maize root tissue. Cry1Ab1 was verified at Rothamsted Research (Harpenden, UK) using a bio- assay mortality test with Plutella xylostella (Lepidoptera: Yponomeutidae). Quantification of Cry1Ab in yeast, Bt and non-Bt root tissue was performed 2. Materials and methods with QuantiPlate Kit for Cry1Ab/Cry1Ac (Envirologix Inc., Portland, ME, USA). Carbon:Nitrogen (C:N) ratios were obtained according to the method 2.1. Test species described in Scrimgeour and Robinson (2003). About 1 mg of the samples were weighed into 6 4 mm tin cups for analysis by continuous flow Dumas P. armata Tullberg (formerly Onychiurus armatus) is a euedaphic NTA combustion using a Roboprep CN sample converter (Europa Scientific, Crewe, and a taxonomic representative of the Collembola. It is a known pest on the UK). Nitrogen and carbon were selectively detected as N2 and CO2 using roots of sugar beet (Joosse and Koelman, 1979), and as such is one of the a Tracermass mass spectrometer (Europa Scientific, Crewe, UK). Samples few Collembola known to be herbivorous. P. armata is a slow-growing, were quantified relative to a leucine/citric acid mixture of known C and N long-lived (maximum age w2 y) that reproduces parthenogenetically content. Reference gravimetric samples of wheat flour were used for quality (Hopkin, 1997). The individuals used in the experiment were taken directly control. The experimental part was performed at the National Environmental from the stock culture at the National Environmental Research Institute, Research Institute, Denmark, while the Bt and C:N analyses were completed Silkeborg, Denmark. at the Scottish Crop Research Institute, UK.

2.2. Experimental design 2.3. Data analysis and statistical methods

The experimental set-up consisted of 7 diet treatments (5 replicates each): All data were log-transformed. One-way ANOVAwas performed in Enter- 5 treatments with dried ground root tissue, 2 Bt maize varieties and 3 non-Bt prise GuideÒ Version 1 (SAS Institute Inc, Cary, NC, USA), to test for

Table 1 Different diets fed to Protaphorura armata Diet Cry1Ab (mg/g)* C:N ratio Diet group Remarks Cascade 1.37 0.34 32.5 0.50a Bt maize Transformation event MON810 MEB307 1.01 0.23 31.1 1.48a Bt maize Transformation event MON810 Rivaldo N.d. 40.2 2.49b Non-Bt maize Isogenic to Cascade Monumental N.d. 19.1 1.01c Non-Bt maize Isogenic to MEB307 DK242 N.d. 19.8 0.34c Non-Bt maize Dry baker’s yeast N.d. 3.90 0.07d Yeast Dry baker’s 183 35.2 4.52 0.23d Yeast Nominal concentration 500 mg/g yeast contaminated with Cry1Ab N.d. signifies ‘‘None detected’’, different superscript letters signify a significant difference (ANOVA, P < 0.05) between carbon:nitrogen (C:N) ratios (mean standard error, n ¼ 3). * mg/g dry maize roots or yeast (mean standard error, n ¼ 4). 214 L.-H. Heckmann et al. / Environmental Pollution 142 (2006) 212e216

significant differences between somatic growth, reproduction, survival and population growth rate of the different treatments. Tukey’s honestly significant difference (HSD) was applied for post hoc comparisons. Equality of variance was tested using Levene’s test. For all tests, a significant level of 5% was applied. Population growth rate was estimated as the natural rate of increase, r by the following equation: ÿ Á r ¼ loge Nf =No =DT ð1Þ

where Nf is the final number of individuals, No is the initial number of individ- uals and DT is the change in time (number of days the experiment was run). Positive values of r indicate a growing population, r ¼ 0 indicates a stable population, and negative r values indicate a population in decline and headed toward extinction.

3. Results

3.1. Quantification of Cry1Ab and C:N ratio

In the yeast treatment, the actual concentration of Bt toxin Cry1Ab was approximately 2.7-fold lower than the nominal concentration of 500 mg/g (Table 1). The root tissue of Bt maize varieties Cascade and MEB307 contained the same level of Bt toxin, whereas none of the non-Bt maize varieties Rivaldo, Monumental and DK242 contained any Bt toxin (Table 1). The carbon:nitrogen (C:N) ratio differed significantly (P < 0.05) between treatments, with the C:N ratio of the yeast treatments being 5e10-fold lower than the maize treatments (Table 1).

3.2. Effects on life history traits and population growth rate

Throughout the experiment, somatic growth, measured as body surface area, was significantly higher (P < 0.05) in P. armata reared on yeast diets than in individuals reared on any of the maize diets (Fig. 1A). With the exception of Bt maize variety MEB307 (P ¼ 0.245), reproduction was only significantly higher (P < 0.05) in P. armata exposed to yeast contaminated with Cry1Ab compared with those exposed to maize diets

Fig. 1. Consequences for Protaphorura armata following a 4-week exposure to Bacillus thuringiensis (Bt) maize, non-Bt maize and yeast (mean standard error, n ¼ 5). Note different y-axes. Significant differences (ANOVA, P < 0.05) between individual maize and yeast treatments are denoted by dif- ferent letters; a (Y þ Bt) and b (Y and Y þ Bt, respectively). (A) Somatic growth measured as body surface area following 0 d (white bars), 7 d (down- ward hatched bars), 14 d (spotted bars), 21 d (black bars) and 28 d (upward hatched bars). (B) Reproduction shown as total number of juveniles. (C) Survival in percent. (D) Population growth rate (pgr) estimated as the natural rate of increase (r) per day. A group-wise comparison between yeast and maize revealed a significantly higher (ANOVA, P < 0.05) reproduction, survival and pgr in individuals reared on yeast. Abbreviations: Y ¼ Dry baker’s yeast; Y þ Bt ¼ Dry baker’s yeast contaminated with purified Cry1Ab (500 mg/g); Mon ¼ Monumental (non-Bt maize isogenic to MEB307); Riv ¼ Rivaldo (non-Bt maize isogenic to Cascade); 242 ¼ DK242 (non-Bt maize); Cas Btþ¼Cascade (Bt maize isogenic to Rivaldo); MEB Btþ¼MEB307 (Bt maize isogenic to Monumental). L.-H. Heckmann et al. / Environmental Pollution 142 (2006) 212e216 215

(Fig. 1B). Comparing the two treatment groups revealed a sig- pollen (Losey et al., 1999), and reduced juvenile survival of nificantly higher (P < 0.05) reproduction in the group fed the tobacco hornworm Manduca sexta exposed to root exudate yeast compared with that in the group fed maize. of N4046Bt (Saxena et al., 1999). But considering that Survival did not differ significantly (P > 0.05) between Cry1Ab targets lepidopterans these results are to be expected. diets (Fig. 1C). But, a group-wise comparison of the yeast For comparison, we found no significant (P < 0.05) differ- and maize diets revealed a significantly higher (P < 0.05) sur- ences in the concentration of Cry1Ab between Cascade, vival in the P. armata reared on yeast. MEB307 and N4046Bt in a pot experiment (unpublished re- P. armata had a positive pgr in all yeast diets, whereas in- sults). However, altered expression of secondary plant metab- dividuals reared on maize were in decline, except for Bt maize olites following the transformation event may influence the variety MEB307 where pgr was stable (Fig. 1D). Again only overall impact of a GM crop (e.g. Saxena and Stotzky, 2001b). P. armata reared on yeast contaminated with Cry1Ab had The actual concentration of purified Cry1Ab was more than a significantly higher (P < 0.05) pgr than those reared on the two orders of magnitude higher in the yeast treatment than in maize treatments, except for Bt maize variety MEB307 the concentrations in the two Bt maize varieties, Cascade and (P ¼ 0.215). Comparing the treatment groups revealed a signif- MEB307. However, P. armata was unaffected when fed yeast icantly reduced (P < 0.05) pgr in maize treatments compared contaminated with purified Bt Cry1Ab, as were the collembo- with the pgr in yeast. lans F. candida and Xenylla grisea after exposure to Cry1Ab, Cry1Ac, Cry2A and Cry3A (Sims and Martin, 1997). The pu- 4. Discussion rified toxin had been proteolytically activated by trypsin. Thus, the d-endotoxin had little or no affinity to the midgut receptors The main objective of this study was to determine whether of these collembolan species. Even if receptor binding did take P. armata would perform differently on a diet of Bt maize place, no distinct pore formation and harmful disruption of compared to non-Bt maize. No significant differences were ob- the midgut membrane occurred. P. armata (Onychiuroidea), served between P. armata reared on Bt and non-Bt maize. In F. candida (Isotomoidea) and X. grisea (Hypogastruroidea) fact, although not significant, P. armata reared on Bt variety belong to different superfamilies, suggesting that collembolans MEB307 had the highest somatic growth, reproduction, sur- in general seem to be very tolerant to Bt Cry toxins. vival and pgr of all the maize treatments, while P. armata reared on the non-Bt variety DK242 had the poorest perfor- 5. Conclusions mance. Overall, P. armata performed significantly better on contaminated and uncontaminated yeast compared with maize The major reason why P. armata was unaffected by root tissue. This corresponds to a recent study, where the Cry1Ab seems to be due to low target site interaction. It seems collembolan Folsomia candida performed far better on yeast, unlikely, from the results of our study, that P. armata would than it did on root tissue from transgenic wheat and its isogenic perform differently in a Bt maize field compared with an iso- varieties (Romeis et al., 2003). At the end of the 4-week expo- genic non-Bt maize field. However, in the field, Cry1Ab is still sure, there was a consistent decrease in body surface area in all detectable in plant residues more than a year after harvest treatments indicating that P. armata was becoming generally (Zwahlen et al., 2003). We emphasize that future studies stressed by the experimental conditions. should be long-term, multi-species experiments that may re- Since body surface area and reproduction were affected in veal the ecological impact of Bt crops on species interactions. both the Bt and non-Bt maize treatments, the reduced repro- duction was presumably not caused directly by Cry1Ab. It was more likely an artefact of a reduced somatic growth that Acknowledgements resulted in a lowered fecundity, and thus in a reduced pgr. The general poor performance on maize reflects the high This http://www.ECOGEN.dk research was supported by C:N ratio in both Bt and non-Bt maize, thus having a lower the EU RTD Quality of Life Programme contract no. QLRT- food quality compared with yeast. Terrestrial isopods Oniscus 2001-01666. The Scottish Crop Research Institute receives asellus and Porcellio scaber also had a higher performance on grant-in-aid from the Scottish Executive Environment and plant material with a low C:N compared with plant material Rural Affairs Department. We thank Zdenek Gavor and Elin with high C:N ratios (Zimmer and Topp, 2000). Jørgensen, National Environmental Research Institute, for A recent study on the effect of plant tissue from Bt maize technical assistance, and Dr Helen L. Hooper and two anony- variety N4046Bt (Cry1Ab) came to a similar finding as our re- mous reviewers for valuable comments on the manuscript. sults on P. armata, showing that P. scaber perform equally when reared on plant tissue from Bt compared with non-Bt References maize (Wandeler et al., 2002). Additionally, Yu et al. (1997) report that F. candida and Oppia nitens (Acari) were unaffected Bengtsson, G., Gunnarsson, T., Rundgren, S., 1985. Influence of metals on when reared on plant tissue from Bt cotton (Cry1Ab and reproduction, mortality and population growth in Onychiurus armatus (Collembola). Journal of Applied Ecology 22, 967e978. Cry1Ac) and Bt potato (Cry3A). However, researchers also Coleman, D.C., Crossley, D.A., Beare, M.H., Hendrix, P.F., 1988. Interactions report on reduced juvenile survival of the monarch butterfly of organisms at root/soil and litter/soil interfaces in terrestrial ecosystems. Danaus plexippus fed milkweed contaminated with N4046Bt Agriculture Ecosystems & Environment 24, 117e134. 216 L.-H. Heckmann et al. / Environmental Pollution 142 (2006) 212e216

Griffiths, B.S., Caul, S., Thompson, J., Birch, A.N.E., Scrimgeour, C., earthworms, nematodes, protozoa, bacteria, and fungi in soil. Soil Biology Andersen, M.N., Cortet, J., Messe´an, A., Sausse, C., Lacroix, B., Krogh, & Biochemistry 33, 1225e1230. P.H. A comparison of soil microbial community structure, protozoa and Saxena, D., Stotzky, G., 2001b. Bt corn has a higher lignin content than non-Bt nematodes in field plots of conventional and genetically modified maize corn. American Journal of Botany 88, 1704e1706. expressing the Bacillus thuringiensis CryIAb toxin. Plant and Soil, in press. Schnepf, E., Crickmore, N., Van Rie, J., Lereclus, D., Baum, J., Feitelson, J., Groot, A.T., Dicke, M., 2002. Insect-resistant transgenic plants in a multi- Zeigler, D.R., Dean, D.H., 1998. Bacillus thuringiensis and its pesticidal crys- trophic context. Plant Journal 31, 387e406. tal proteins. Microbiology and Molecular Biology Reviews 62, 775e806. Hopkin, S., 1997. Biology of the (Insecta: Collembola). Oxford Scrimgeour, C.M., Robinson, D., 2003. Stable isotope analysis and appli- University Press, New York, NY, USA. cations. In: Smith, K.A., Cresser, M.S. (Eds.), Soil and Environmental Jasinski, J.R., Eisley, J.B., Young, C.E., Kovach, J., Willson, H., 2003. Select Analysis: Modern Instrumental Techniques, third ed. Marcel Dekker nontarget abundance in transgenic and nontransgenic field crops Inc., New York, NY, USA, pp. 381e431. in Ohio. Environmental Entomology 32, 407e413. Sims, S.R., Martin, J.W., 1997. Effect of the Bacillus thuringiensis insecticidal Joosse, E.N.G., Koelman, T., 1979. Evidence for the presence of aggregation proteins CryIA(b), CryIA(c), CryIIA, and CryIIIA on Folsomia candida pheromones in Onychiurus armatus (Collembola), a pest insect in sugar- and Xenylla grisea (Insecta: Collembola). Pedobiologia 41, 412e416. beet. Entomologia Experimentalis et Applicata 26, 197e201. Vercesi, M.L., Holmstrup, M., Krogh, P.H. Effects of Bacillus thuringiensis Krogh, P.H., Johansen, K., Holmstrup, M., 1998. Automatic counting (Bt) corn residues and Bt-corn plants on life-history traits in the earthworm of collembolans for laboratory experiments. Applied Soil Ecology 7, Aporrectodea caliginosa. Applied Soil Ecology, in press, doi:10.1016/ 201e205. j.apsoil.2005.07.002. Larink, O., 1997. Springtails and mites: important knots in the food web of Wandeler, H., Bahylova, J., Nebtwig, W., 2002. Consumption of two Bt and six soils. In: Benckiser, G. (Ed.), Fauna in Soil Ecosystems. Recycling Pro- non-Bt corn varieties by the woodlouse Porcellio scaber. Basic Applied cesses, Nutrient Fluxes, and Agricultural Production. Marcel Dekker Ecology 3, 357e365. Inc., New York, NY, USA, pp. 225e264. Yu, L., Berry, R.E., Croft, B.A., 1997. Effects of Bacillus thuringiensis toxins Losey, J.E., Rayor, L.S., Carter, M.E., 1999. Transgenic pollen harms monarch in transgenic cotton and potato on Folsomia candida (Collembola: larvae. Nature 399, 214. Isotomidae) and Oppia nitens (Acari: Orbatidae). Journal of Economic Romeis, J., Battini, M., Bigler, F., 2003. Transgenic wheat with enhanced Entomology 90, 113e118. fungal resistance causes no effects on Folsomia candida (Collembola: Zimmer, M., Topp, W., 2000. Species-specific utilization of food sources by Isotomidae). Pedobiologia 47, 141e147. sympatric woodlice (Isopoda: Oniscidea). Journal of Ecology 69, Saxena, D., Flores, S., Stotzky, G., 1999. Transgenic plants e insecticidal 1071e1082. toxin in root exudates from Bt maize. Nature 402, 480. Zwahlen, C., Hilbeck, A., Gugerli, P., Nentwig, W., 2003. Degradation of the Saxena, D., Stotzky, G., 2001a. Bacillus thuringiensis (Bt) toxin released from Cry1Ab protein within transgenic Bacillus thuringiensis corn tissue in the root exudates and biomass of Bt maize has no apparent effect on field. Molecular Ecology 12, 765e775.