Anna Dutton, Swiss Federal Research Station for Agroecology and Agriculture, Reckenholzstr. 191, CH-8046 Zurich, Switzerland. E-mail: firstname.lastname@example.org
Abstract 1. Chrysoperla carnea is an important predatory insect in maize. To assess the ecological effects of Bt-maize, expressing the Cry1Ab protein, on larvae of this predator, the following factors were examined: (1) the performance of three prey herbivores (Rhopalosiphum padi, Tetranychus urticae, and Spodoptera littoralis) on transgenic Bt and non-transgenic maize plants; (2) the intake of the Cry1Ab toxin by the three herbivores; and (3) the effects on C. carnea when fed each of the prey species.
2. The intrinsic rate of natural increase (rm) was used as a measure of performance for R. padi and T. urticae. No difference in this parameter was observed between herbivores reared on Bt or non-transgenic plants. In contrast, a higher mortality rate and a delay in development were observed in S. littoralis larvae when fed Bt-maize compared with those fed the control maize plants.
3. The ingestion of Cry1Ab toxin by the different herbivores was measured using an immunological assay (ELISA). Highest amounts of Cry1Ab toxin were detected in T. urticae, followed by S. littoralis, and only trace amounts detected in R. padi.
4. Feeding C. carnea with T. urticae, which were shown to contain the Cry1Ab toxin, or with R. padi, which do not ingest the toxin, did not affect survival, development, or weight of C. carnea. In contrast, a significant increase in mortality and a delay in development were observed when predators were fed S. littoralis larvae reared on Bt-maize.
5. A combined interaction of poor prey quality and Cry1Ab toxin may account for the negative effects observed on C. carnea when fed S. littoralis. The relevance of these findings to the ecological risks of Bt-maize on C. carnea is discussed.
Since the first commercial release of genetically modified crops expressing genes from Bacillus thuringiensis (Bt), there have been concerns about their potential impact on the environment. In particular, the continuous expression of the insecticidal protein in most tissues of the plant throughout the growing season has raised concerns regarding the development of resistance in the target pest and about the possible impact of this new pest control technology on various groups of non-target organisms of ecological and economic value.
Despite the fact that Bt-toxins in commercial spray formulations are considered safe due to their selective mode of action, and are considered compatible with naturally occurring biological control agents, contrasting results on the effects of Bt-toxins on various beneficial insects have been reported (reviewed by, for example, Glare and O'Callaghan, 2000). The only beneficial insect for which direct negative effects of Bt-plants have been reported is the predatory green lacewing Chrysoperla carnea (Stephens) (Neuroptera: Chrysopidae). Higher mortality was observed when C. carnea was fed either Spodoptera littoralis (Boisduval) (Lepidoptera: Noctuidae) or Ostrinia nubilalis (Hübner) (Lepidoptera: Crambidae) larvae reared on Bt-maize expressing the cry1Ab gene, compared with C. carnea fed lepidopteran larvae of the same species but reared on untransformed maize plants (Hilbeck et al., 1998a). Similar results were obtained when the Cry1Ab toxin was incorporated into the artificial diet fed to C. carnea (Hilbeck et al., 1998b). Further studies also showed acute toxic effects of a second Bt-toxin, Cry2A, on C. carnea larvae when this toxin was incorporated into the artificial diet of its prey, S. littoralis (Hilbeck et al., 1999).
These results show that under worst-case laboratory studies, C. carnea is vulnerable to Cry1Ab expressed in maize; further research was essential to evaluate the relevance of these results for the field situation. Several factors have to be taken into account to assess the ecological effects that Bt-maize might have on a generalist predator such as C. carnea. Firstly, the range, importance, and suitability of prey available to the predator in the field need to be considered. Chrysoperla carnea larvae feed on a wide range of small, soft-bodied arthropods that include aphids and spider mites, as well as eggs and young larvae from various species including Lepidoptera (Principi & Canard, 1984; Bay et al., 1993). Secondly, it is important to establish whether the predator ingests the toxin. The ingestion of toxin can either occur when the predator feeds on contaminated prey or, in the case of phytophagous predators, via direct feeding on the transgenic plant (e.g. pollen, plant sap). Information on the mode and site of herbivore feeding and the site of toxin expression in the plant will help to predict which herbivores ingest the toxin and in what quantities. For example, different leaf-chewing Lepidoptera species consume different amounts of the Bt-toxin when feeding on Bt-expressing leaf tissue (Head et al., 2001), and phloem-sucking arthropods such as aphids do not ingest the toxin when feeding on maize varieties in which the Bt-toxin is not transported in the phloem (i.e. Event 176 and Bt11) (Head et al., 2001; Raps et al., 2001). The extent to which other cell-sucking herbivores such as spider mites or thrips, which feed on epidermal and mesophyll cells, ingest the toxin is not known. Thirdly, it is essential to evaluate the effects of Bt plants on prey quantity and quality, as both factors are known to influence predator populations (Jervis & Copland, 1996). A Bt-toxin could increase the mortality of certain prey herbivores and this may consequently affect predator populations, depending on their importance as prey and the extent of herbivore mortality. In addition, a Bt-toxin may have sub-lethal effects on prey herbivores (sick prey), which could reduce their quality as prey for predators. The insertion of new genes into the plants through genetic engineering could inadvertently change the nutritional quality of the plant itself (insertional mutagenesis or pleiotropic effects), which in turn might influence prey populations and/or prey quality.
The work reported here had three objectives. (1) To compare the performance of three different prey species of C. carnea, namely aphids, spider mites, and a non-target lepidopteran species, on either Bt or non-transgenic maize. (2) To verify and quantify the presence of Bt-toxin in the three prey herbivores. (3) To determine the effects on C. carnea larvae fed with the different prey species reared on either Bt or non-transgenic maize.
Material and methods
Plants and insects
Transgenic maize (Zea mays) plants (N4640Bt; Syngenta, Stein, Switzerland, formerly Northrup King) expressing a truncated, synthetic version of the cry1Ab gene from Bacillus thuringiensis var. kurstaki HD-1 (Bt+) and the corresponding isogenic non-transformed (Bt–) variety (N4640) were used for all experiments. The Cry1Ab toxin expression in this Bt+ hybrid is driven by the constitutive CaMV 35S promoter. All plants were planted in plastic pots (15 cm diameter) and cultivated in a greenhouse at 24 ± 4 °C, 50 ± 10% RH. Plants were fertilised (16 N : 6 P : 26 K) weekly. For all experiments, plants were used when they had reached a height of 70 cm (seven to 10 leaf stage).
Rhopalosiphum padi (L.) (Homoptera: Aphididae) and Tetranychus urticae (Koch) (Acari: Tetranychidae) were reared on Bt + or Bt – maize plants in separate greenhouses at 24 ± 4 °C, 50 ± 10% RH, and L : D 16 : 8 h photoperiod. Rhopalosiphum padi were collected by brushing them off the plant, and T. urticae were collected by tapping individual leaves with a stick and collecting the arthropods on a tray placed below. Egg masses of S. littoralis were provided by Syngenta (Stein, Switzerland), where the colony is reared on artificial diet. Newly emerged (12 h old) larvae were fed with either Bt + or Bt – maize leaves. Spodoptera littoralis larvae were kept inside an air-permeable bag (40 × 10 cm) that surrounded a maize leaf still attached to the plant. Larvae were allowed to feed on these leaves for 28 ± 4 h in a growth chamber at 27 ± 1 °C, 55 ± 5% RH, and L : D 16 : 8 h photoperiod.
Chrysoperla carnea larvae were from a permanent laboratory colony reared at the research station. Chrysoperla carnea larvae were maintained on pea aphids Acyrthosiphon pisum (Harris) (Homoptera: Aphididae), and adults were fed with a mixture of yeast, honey, and water. Rearing conditions were 22 ± 3 °C, 70 ± 5% RH, and L : D 16 : 8 h photoperiod ( Bigler, 1988 ).
Three reproductive adult females were caged on the sixth to ninth leaf of either Bt+ or Bt– maize plants in a 3.4 cm diameter clip cage in the greenhouse at 21 ± 1 °C, 70 ± 15% RH, and L : D 16 : 8 h photoperiod. After 7 h, all but three newly laid nymphs were removed from the cage. The development of the nymphs was checked daily until the start of reproduction. The number of aphids was then reduced to one per cage. The number of offspring produced by the individual aphids was counted and all nymphs produced were removed every other day. Aphid performance on Bt+ and Bt– plants was assessed by calculating the intrinsic rate of natural increase [rm = 0.74(ln FD/D), where FD = number of aphids produced over a period of time equal to that of the pre-reproductive period (D)] (Wyatt & White, 1977) of 22 aphids for each treatment. Mean comparisons were performed using Student's t-test.
Newly hatched spider mites were caged individually on the seventh or ninth leaf of either Bt+ and Bt– maize in 2 × 2 × 1.5 cm clip cages. The maize plants were then kept in a growth chamber at 27 ± 1 °C, 55 ± 5% RH, and L : D 16 : 8 h photoperiod. Intrinsic rates of natural increase were calculated as described above for a total of 35 individuals for each treatment, and means were compared using Student's t-test.
Mortality of first-instar larvae, number of days to reach second instar, and weight of second-instar larvae were measured to determine larval performance on Bt+ and Bt– plants. These parameters were selected as personal observations showed that, when given a choice, C. carnea larvae fed primarily on first-instar larvae (see also Bay et al., 1993). Groups of six newly emerged S. littoralis larvae were placed in a 3.4 cm diameter clip cage on the fourth leaf of Bt+ and Bt– plants and kept in the greenhouse at 27 ± 2 °C, 60 ± 10% RH, and L : D 16 : 8 h photoperiod. Twenty plants (replicates) per treatment were used. Observations were made daily and the number of days required for larvae to reach the second instar was recorded. Larvae were removed from cages upon reaching the second instar. Forty second-instar larvae were selected randomly and weighed individually. Per cent mortality of larvae was arcsin-square-root transformed and mean mortality was analysed using Student's t-test. Mean number of days to reach second instar and larval weight were calculated and Student's t-tests were performed to determine differences between treatments.
Cry1Ab protein levels in plants and herbivores were determined using a double sandwich ELISA kit (EnviroLogix Inc., Portland, Maine). Cry1Ab standards at concentrations 0, 0.5, 2.5, and 5 p.p.m. were used as calibrators. Spectrophotometric measurements were conducted with a microtiter plate reader (Dynatech MR 5000, Dynex Technologies, Ashford, U.K.) at 450 nm and data were analysed using the software package Biolinx 2.0 (Dynatech Laboratories Inc.) and Dynex Revelation G 3.2 (Dynex Technologies).
To quantify Bt-toxin in the three prey herbivores, protein was extracted from each herbivore reared either on Bt+ or Bt– plants. Five samples from each herbivore species reared on Bt+ and Bt– plants were analysed. Quantities (mg) of material and buffer dilutions used for the different herbivores were: 68 ± 5 mg R. padi homogenised in 0.5 ml buffer (no dilution), 24 ± 3 mg T. urticae homogenised in 0.5 ml buffer (diluted 1 : 50), and 5.7 ± 0.5 mg 2-day-old first-instar S. littoralis larvae homogenised in 0.5 ml buffer (diluted 1 : 10).
Expression of Cry1Ab in plants was verified by analysing leaf material. Fifteen randomly selected Bt+ and Bt– plants were analysed. Leaf pieces weighing 200 ± 5 mg from the fifth leaf were homogenised in 5 ml buffer and diluted 1 : 100.
All samples were centrifuged for 5 min at 13 000 r.p.m. before they were introduced into the ELISA plate.
Performance of Chrysoperla carnea
Experiments were conducted using Plexiglas® experimental cages manufactured by Isoplex AG (Regensdorf, Switzerland), similar to those used by Zwahlen et al. (2000). Cages consisted of two plates (52.6 × 13.9 × 1.9 cm), between which leaves from intact maize plants were positioned. The leaves were left attached to the plant to avoid changes in leaf quality and Bt-toxin expression. The top plate contained 3 cm diameter holes in which individual C. carnea larvae with their prey were placed. The individual holes were closed with plastic caps (1 cm deep) and fine-mesh netting. Larvae were fed twice daily ad libidum, with each of the three different herbivores reared on either Bt+ or Bt– plants. Predators that were fed with T. urticae or S. littoralis were fed exclusively with Ephestia kuehniella (Hübner) (Lepidoptera: Pyralidae) eggs 2 days after reaching the third (last) instar, because of the large prey requirements of C. carnea during this stage (Principi & Canard, 1984). Experiments were conducted in the greenhouse at 25 ± 4 °C, 60 ± 10% RH, and L : D 16 : 8 h photoperiod. Each prey treatment (R. padi, T. urticae, S. littoralis) reared on Bt+ or Bt– maize was conducted simultaneously and each experiment was replicated twice using 30 chrysopid larvae (60 larvae for each prey–plant treatment). Chrysoperla carnea larvae were examined twice daily and life-table parameters (development, mortality) were recorded. Pupal mortality was never above 2% for any one prey–plant treatment and was thus not included in the analysis. Weight of the different larval stages and adults was measured on the first day after moulting and on the day of emergence, respectively. Statistical analyses were carried out separately for each prey (Bt+ vs. Bt–). Comparisons between treatments for the stage-specific mortality data were made using a logistic regression analysis (SYSTAT, 1996). For data on development (days required to reach second and third instars, pupae, and adult emergence) and weight (second- and third-instar larvae and adults), mean comparisons between treatments were made using Student's t-test.
There was no significant difference between the intrinsic rate of increase of either R. padi or T. urticae reared on the Bt+ and Bt– plants (Table 1). In the case of S. littoralis, both the survival rate and the time required to reach the second instar were affected significantly when larvae were reared on Bt+ plants compared with larvae reared on Bt– plants (Table 2). There was some evidence for lower weight in second-instar S. littoralis larvae that were fed Bt+ leaves compared with those fed Bt– leaves, but the difference was not significant (Table 2).
Table 1. Mean intrinsic rates of natural increase (r m ± SE) of Rhopalosiphum padi and Tetranychus urticae reared on transgenic (Bt + ) and non-transgenic (Bt − ) maize.
A mean concentration of 3.4 μg Cry1Ab toxin was measured per gram fresh weight of Bt+ maize leaves (Fig. 1). In R. padi, trace concentrations of the toxin (0.02 μg g−1) were detected. Spodoptera littoralis larvae contained on average 0.72 μg g−1. The highest concentration of toxin was measured in T. urticae (2.5 μg g−1). Trace amounts of the toxin were also detected in the control samples of T. urticae (0.02 μg g−1), S. littoralis (0.01 μg g−1), and R. padi (< 0.001 μg g−1), possibly due to contamination or a cross-reaction with other proteins.
Performance of Chrysoperla carnea
Survival, development, and weight of C. carnea fed R. padi or T. urticae were not affected significantly by Bt+ plants (Fig. 2). In contrast, instar-specific survival of C. carnea fed S. littoralis reared on Bt+ plants was significantly lower (17.7%) than for prey reared on Bt– plants (55.6%) (P < 0.01). The overall survival rate of C. carnea fed S. littoralis reared on Bt– plants was much lower than when fed R. padi (93%) or T. urticae (91%) (Fig. 2a). The developmental times for first-instar (P < 0.001) and second-instar (P < 0.01) larvae and the total time to reach the adult stage (P = 0.001) were significantly longer for C. carnea fed Bt+-reared S. littoralis than for C. carnea fed Bt–-reared larvae (Fig. 2b). There were no significant effects of Bt+ plants on the development of third-instar larvae. Statistical differences in larval weight between the treatments were only observed for the first-instar C. carnea (P < 0.05), which were lighter when fed S. littoralis reared on Bt+ than when reared on S. littoralis larvae reared on Bt– (Fig. 2c).
Large differences in the concentration of Bt-toxin were detected among the three herbivores examined. In R. padi, only trace amounts of Cry1Ab were detected, whereas both T. urticae and S. littoralis were shown to ingest the toxin when feeding on transgenic plants. The data suggest that C. carnea larvae are affected by Bt-toxin when fed with S. littoralis reared on Bt+ maize, as shown by Hilbeck et al. (1998a). Surprisingly, no effects on development, mortality, or weight of C. carnea were detected when fed T. urticae, which also ingest the Bt-toxin. An interaction between prey and Bt-toxin therefore appears to account for the observed negative effects of Cry1Ab on C. carnea when S. littoralis is used as a prey herbivore.
Aphids, which are considered a major prey for C. carnea in maize fields (Coderre, 1988), do not ingest the toxin when feeding on Bt+ plants (Event 176 and Bt11), as shown here and elsewhere (Head et al., 2001; Raps et al., 2001). Given that the Bt-toxin is not present in the phloem sap in the Bt+ maize event used in this work (Raps et al., 2001), direct effects of the Bt-toxin on R. padi or on other phloem-feeding arthropods can be excluded. The intrinsic rates of natural increase (rm) observed for R. padi in the present study are similar to the findings of Lozzia et al., (1998) for the same aphid species on Bt+ maize Event 176. In contrast to the event used in this study, Event 176 contains two promoters that regulate expression of the Bt-toxin (phosphoenol pyruvate-carboxylase in combination with a pollen-specific promoter). It appears that these two commercially available events have no effect on development and fecundity of R. padi. As would be expected, no negative effects on the measured life-table parameters or on weight were observed when the predator C. carnea was fed with aphids reared on Bt+ maize.
Tetranychus urticae and other mite species such as Oligonychus pratensis (Banks) are considered sporadic but sometimes important pests in some maize-growing areas ( Pickett & Gilstrap, 1986 ), and at high infestation levels T. urticae is a major prey for C. carnea in the field. In contrast to aphids, the ELISA results demonstrate that T. urticae ingest the toxin when feeding on Bt + maize. Tetranychidae, the family to which T. urticae belongs, are known to feed on parenchyma cells ( van der Geest, 1985 ). Regardless of toxin ingestion by this arthropod, no direct negative effects of the Bt-toxin on spider mites were observed. Both development and fecundity of T. urticae were similar when reared on either Bt + or Bt – plants, as also found by Lozzia et al. (2000) for the same species on Event 176 Bt + maize . No effects were detected on the mortality, development, or weight of C. carnea fed T. urticae containing Cry1Ab.
Chrysoperla carnea larvae are known to feed on young lepidopteran larvae in maize fields, however they are considered to be a low quality prey when compared with other prey such as aphids, spider mites, or lepidopteran eggs. Chrysoperla carnea larvae develop more slowly and have a higher mortality rate when fed lepidopteran larvae than when fed other prey species, as shown here and in other studies ( Obrycki et al., 1989 ; Klingen et al., 1996 ). Choice tests have also shown that C. carnea larvae prefer aphids to lepidopteran larvae ( Meier & Hilbeck, 2001 ) and it is therefore likely that lepidopteran larvae are not a common prey for C. carnea in the field. In addition, lepidopteran larvae such as S. littoralis have the ability to escape when attacked by lacewings by dropping off the plant using a silk thread, and they are also capable of defending themselves with their mandibles when pierced by a chrysopid larva (A. Dutton, pers. obs.).
As expected for all leaf-feeding Lepidoptera, and confirmed by the ELISA test, S. littoralis consumes the toxin when feeding on Bt+ plants. Contradictory results on the effects of Cry1Ab on S. littoralis are reported in the literature. This is due to the fact that sensitivity tests have been conducted with Bt sprays, which contain a mixture of Cry proteins (i.e. sprays containing the Bt krustaki HD-1 strain contain Cry1Aa, Cry1Ab, Cry1Ac, Cry2A, and Cry2B) (Glare & O'Callaghan, 2000), with the Cry1Ab protoxin (van Frankenhuyzen, 1993; Hilbeck et al., 1999), or the activated (solubilised and trypsinised) Cry1Ab toxin (Hilbeck et al., 1999). Results from this study show that the activated Cry1Ab toxin expressed in the plant causes an increase in mortality, a delay in development, and a slight decrease in the weight of S. littoralis larvae. An increased mortality of S. littoralis was also shown by Hilbeck et al. (1999) when larvae were fed an artificial diet containing the activated pure Cry1Ab toxin, while with the protoxin this effect was not observed. These results illustrate the differences between the relatively narrow range of species-specificity of Bt-sprays (containing protoxins that need to be solubilised and tryptinised) compared with a potentially wider range of activity of Bt-toxins expressed in transgenic plants.
The fact that S. littoralis suffer when feeding on Bt+ maize suggests that these larvae present a reduced prey quality (sick prey) for C. carnea when compared with larvae fed Bt– maize. This could partly explain the negative effects observed on development and survival of C. carnea when the predator was fed with this prey species reared on Bt+ plants. A similar reduction in prey quality may also occur in other noctuid maize pests such as the corn ear worm Helicoverpa zea (Boddie) and the stalk borer Papaipema nerbis (Grunée), which develop more slowly, weigh less, and have a higher mortality rate when exposed to Bt+ maize (Pilcher et al., 1997a).
A combination of several factors might explain the negative effects observed in C. carnea when fed S. littoralis reared on Bt+ maize. Firstly, S. littoralis is not a high-quality prey; secondly, its quality is reduced further when fed Bt+ maize; and thirdly, C. carnea is susceptible to the Cry1Ab toxin (Hilbeck et al., 1998b). Whether the detrimental effects of the toxic compounds on predators are indirectly mediated solely by the quality of the sick prey, or mediated directly by the intake of the toxin by the predator is not clear from this study. Given that no negative effects were observed when C. carnea was fed T. urticae, which was shown to contain higher levels of the Bt-toxin when compared with S. littoralis, two theories can be proposed. (1) The binding of Cry1Ab to receptors in the midgut epithelium of S. littoralis causes a change in the protein structure, magnifying the toxicity in the next trophic level, as suggested by Hilbeck et al. (1999). (2) Spider mites de-activate or degrade the protein into a non-toxic form. Further studies are being conducted to investigate what happens to the toxin when it is ingested by different herbivores.
The ELISA results show that levels of Cry1Ab in T. urticae were much higher than in S. littoralis. A recent study showed similar differences in Bt-toxin measured among three species of lepidopteran larvae (Head et al., 2001). The Bt-toxin differences among species were correlated with differences in the insects' susceptibility to the protein, possibly due to the fact that susceptible herbivores fed less and thus ingested less protein. Similarly, higher amounts of the Bt-toxin were measured in T. urticae (which is not susceptible to the toxin when compared with S. littoralis), which was affected by the toxin. Although ELISA results showed quantitative differences of Bt-toxin present in the two herbivores, caution must be taken when interpreting such results. ELISA activity is based on an antibody–antigen interaction that does not necessarily indicate the level of toxin activity. Moreover, ELISA tests have the drawback of matrix effects (i.e. extraction procedures can produce products that interfere with the analysis; recovery of protein is not the same when extracted for different organisms) (Miksic, 1992), which can influence extraction of protein. This has been shown to be a problem in the extraction of Bt from Bt+ cotton tissue (Sachs et al., 1998). To verify whether the quantity and quality (i.e. activity) of the toxin differ when ingested by the different herbivores, further investigations using Western blot assays together with insect feeding assays should be performed (Fuchs et al., 1990; Head et al., 2001).
When considering the results obtained in this study, it is not surprising to find that until now no negative effects of Bt+ maize or Bt+ cotton on C. carnea have been detected in field studies (Orr & Landis, 1997; Pilcher et al., 1997b; Candolfi et al., 2000). Aphids, which are the most abundant, suitable, and preferred prey of C. carnea, do not ingest the toxin when feeding on Bt+ maize. Even feeding on herbivores such as spider mites, which ingest the toxin, does not necessarily result in lethal or sublethal effects on C. carnea. Lepidoptera larvae appear unlikely to be a common prey for C. carnea in maize, however further studies are needed to establish the proportion of different prey species in the diet of C. carnea in this particular crop ecosystem.
This project was supported by the Swiss National Science Foundation project number 5002–57820. We thank M. Walburger (FAL Zürich) for his technical assistance and H. Gu (ETH Zürich) for statistical help.