Running title: Interactions between transgenic plants, the diamondback moth and natural enemies

Keywords

Cotesia plutellae, Brassica napus, Bacillus thuringiensis, resistance, Plutella xylostella, transgenic crops, tritrophic interactions

Abstract

Several Brassica crops have been transformed with δ-endotoxin genes from Bacillus thuringiensis (Bt) to provide resistance to the diamondback moth (Plutella xylostella) and other lepidopteran pests. This paper describes studies of the direct and indirect effects of Bt plants on a parasitoid of P. xylostella using Cry1Ac-expressing transgenic oilseed rape (canola) as an example. The braconid wasp Cotesia plutellae is a solitary endoparasitoid of P. xylostella larvae. We present evidence that, although C. plutellae larvae that were forced to develop in Bt-treated susceptible hosts inevitably died with their hosts, behavioural factors are likely to limit the impact of this effect on field populations. Cotesia plutellae mortality in susceptible hosts was not due to a direct toxic effect of Cry1Ac, but due to premature host mortality since C. plutellae larvae developed normally in Bt-resistant hosts on Bt plants. Adult C. plutellae females were highly attracted to Bt plants damaged by Bt-resistant hosts. Any tactic aimed at suppressing pest populations risks affecting the dynamics of some natural enemies, simply due to a reduction in host or prey densities. The extent of these effects will depend on the ecological specialisation of the different natural enemy species present in the crop ecosystem. They are likely to be most pronounced for species, such as hymenopteran parasitoids, that tend towards greater host or prey specificity. However, the apparent lack of direct effects of Bt plants on survival and host-seeking ability of C. plutellae implies an environmental advantage compared with the broad-spectrum insecticides used for diamondback moth management in many areas.

Introduction

Brassica crops are grown worldwide and form an important part of the human diet, particularly in Asia (Sivapragasam et al. 1997). Pests such as the diamondback moth (Plutella xylostella (L.), Lepidoptera, Plutellidae) remain a major problem despite advances in pest control (Gujar 1999, Sivapragasam et al. 1997, Talekar 1992, Talekar & Shelton 1993). A number of Brassica crops, including cabbage (Brassica oleracea L. subsp. capitata), broccoli (B. oleracea subsp. italica) and oilseed rape (Brassica napus L.) have been transformed with lepidopteran-active δ-endotoxin genes derived from Bacillus thuringiensis Berliner (Bt) (Jin et al. 2000, Kuvshinov et al. 2001, Metz et al. 1995, Stewart et al. 1996, Zhao et al. 2000), but so far none of these Bt plants has been commercially released.

Parasitoids are very important natural enemies of Brassica pests (Billqvist & Ekbom 2001, Fournet et al. 2000, Murchie et al. 1997, Sato et al. 1999, Sivapragasam et al. 1997). They are insects that complete their larval development on a single host insect and their survival and fitness are therefore intrinsically linked to the quality and fate of their host (Quicke 1997). Parasitoids represent one of the groups of non-target insects at risk from GM crops since they may come into direct contact with the transgene product when the larva feeds on host tissues or when the adult feeds on hosts. In addition, Bt plants have the potential to affect parasitoids indirectly by reducing host availability, quality and survival.

Microbial Bt formulations have been used for some time against P. xylostella and play a major part in integrated pest management (IPM) programmes since Bt δ-endotoxins, in contrast to many synthetic insecticides, have no contact toxicity against natural enemies (Shelton et al. 1993, Sivapragasam et al. 1997, Talekar & Shelton 1993). Microbial Bt formulations applied orally or to the host, are generally considered non-toxic against parasitoids. However, some laboratory studies have reported negative effects (reviewed by Glare and O'Callaghan (2000)), but these early studies were mostly conducted with a mixture of toxin crystals and bacterial spores and it is only relatively recently that the effects of purified Bt toxins have been investigated. In addition, most wildtype Bt δ-endotoxins are protoxins, which require activation by gut proteases. In contrast, most Bt plants express already partially activated Bt toxins and this could potentially broaden their spectrum of activity.

Plutella xylostella has developed resistance to all major classes of insecticide and is considered the most damaging pest of brassicas on a worldwide scale (Ooi 1992, Talekar & Shelton 1993). Insecticide resistance in P. xylostella has led to control failures and overuse of insecticides (Gujar 1999). The susceptibility of P. xylostella to a number of Bt toxins has therefore made this pest the main target for the development of insect resistant transgenic brassicas. However, in some countries extensive use of microbial Bt formulations has already led to P. xylostella populations with resistance to Cry1A toxins (Ferré et al. 1991, Shelton et al. 1993, Tabashnik 1994, Wright et al. 1995). While Bt brassicas are considered by some a welcome new tool for the management of lepidopteran pests they are considered by others a serious threat to integrated pest management of P. xylostella. To provide factual and reliable information for regulators, growers and extension staff, further research is required into the effects of Bt brassicas on biocontrol agents, the potential role of Bt plants as a component of IPM programmes and appropriate resistance management strategies.

This paper describes the effect of transgenic oilseed rape (Brassica napus) on the solitary larval endoparasitoid Cotesia plutellae Kurdjumov (Hymenoptera, Braconidae), which is one of the major natural enemies of P. xylostella (Ooi 1992, Talekar & Shelton 1993). The interactions between Bt oilseed rape, P. xylostella and C. plutellae were studied using small-scale bioassays representing a worst-case scenario as well as behavioural and population scale choice experiments.

Transgenic plants

The Bt oilseed rape line used in this study (cv Oscar, line O52) expressed a truncated synthetic Bt cry1Ac gene under the control of the cauliflower mosaic virus 35S promoter (Stewart et al. 1996). Untransformed wildtype plants of the parent cultivar were used as controls.

Effect of Bt oilseed rape on development and survival of Cotesia plutellae in small scale bioassays

The effects of Bt oilseed rape on development and survival of C. plutellae in susceptible and Bt-resistant P. xylostella larvae were compared in no-choice Petri dish bioassays.

Susceptible P. xylostella larvae fed Bt leaves caused very little feeding damage, did not increase in size and all larvae died within five days. No parasitoids emerged from hosts fed with Bt leaves while 63% of susceptible hosts fed wildtype leaves produced viable parasitoid offspring. The remaining susceptible hosts in the wildtype treatment were not parasitised and pupated normally. When susceptible hosts were dissected within two days following exposure to parasitoid females, parasitoid eggs or I instar parasitoid larvae were found indicating that the parasitoids attacked susceptible P. xylostella larvae on Bt leaves in this no-choice situation and that eggs hatched if the host lived long enough.

Larvae of the highly Bt-resistant NO-QA P. xylostella strain (Tabashnik et al. 1997) developed normally on Bt oilseed rape and no increase in mortality compared with the control was observed (Schuler et al. 1999). The mean time from egg to parasitoid emergence from hosts ranged between seven to nine days both on Bt leaves and wildtype leaves and there was no statistically significant difference in the distribution of parasitoid emergence from hosts between plant types. The level of parasitism of resistant larvae on Bt leaves was also not decreased compared with the wildtype treatment. After emergence from hosts, C. plutellae larvae spin a cocoon in which they pupate. Successful adult parasitoid emergence from cocoons ranged between 76-100% and there was no consistent effect of the plant line. There was also no evidence for a significant effect of the Bt line on the sex ratio of the parasitoid progeny.

Windtunnel choice tests with Cotesia plutellae

Parasitoid females use volatiles released from damaged plants to locate their hosts (Turlings et al. 1990, Vinson 1991). The responses of adult female C. plutellae to Bt oilseed rape leaves were, therefore, investigated in a series of dual choice tests in a wind tunnel (Potting et al. 1999, Schuler et al. 1999) to investigate if the females distinguish between (a) Bt and wildtype oilseed rape plants and (b) plants damaged by either susceptible or Bt-resistant P. xylostella larvae. Cotesia plutellae predominantly uses plant-derived stimuli in its in-flight host searching behaviour and the presence of hosts is not essential (Potting et al. 1999). However, host-damaged leaves are more attractive to C. plutellae females than undamaged leaves while artificially-damaged leaves are as attractive as host-damaged leaves (Potting et al. 1999). Host-damaged leaves were obtained by allowing two P. xylostella larvae, either Bt-susceptible or Bt-resistant, to feed on each leaf overnight. Flights were recorded as a choice if they ended in landings on one of the leaves, within five minutes of take off. The amount of feeding damage inflicted by the hosts was measured after each bioassay.

Susceptible P. xylostella larvae caused significantly less feeding damage to the Bt leaves than to wildtype leaves and very few C. plutellae females (11%) chose to land on a Bt leaf over a wildtype leaf when these leaves were presented as a choice in the wind tunnel (Schuler et al. 1999). No significant difference was found between the feeding damage of resistant larvae on Bt compared to wildtype leaves and the parasitoid females did not distinguish between these two treatments. Similarly, parasitoids did not prefer one plant type to the other if leaves were artificially damaged to the same degree. When the parasitoids were given a choice of two Bt leaves damaged by either Bt-resistant hosts or Bt-susceptible hosts the majority (79%) of C. plutellae flew to the Bt leaves damaged by resistant hosts (Schuler et al. 1999).

Population scale studies with Cotesia plutellae

Further experiments were conducted in large cages in the laboratory to compare the level of parasitism on Bt and wildtype plants under conditions in which female parasitoids could freely choose between host populations on either plant type. The experiments were conducted with Bt-resistant P. xylostella larvae since Bt-susceptible P. xylostella larvae do not survive on Bt oilseed rape plants.

A mixture of wildtype and Bt plants were placed together in each cage as described previously (Schuler et al. 2001) and each plant was infested with P. xylostella eggs. Female C. plutellae were released into half the cages once most P. xylostella larvae had reached the III larval instar. The remaining cages served as controls. Numbers of parasitised and unparasitised hosts were assessed at the end of one parasitoid generation.

The results from one experiment showed a significantly higher level of parasitism on Bt plants but this result could not be reproduced and two further experiments showed no significant difference in parasitism between Bt and wildtype plants.

Discussion

The Bt oilseed rape plants caused 100% mortality of susceptible P. xylostella larvae and no C. plutellae larvae were able to complete their development in such hosts. However, the mortality of the parasitoid larvae was due to the premature death of the host and not due to a direct effect of Cry1Ac, since C. plutellae larvae developed normally in Bt-resistant P. xylostella feeding on Bt plants.

Parasitoid females use plant volatiles from damaged plants to guide them to hosts (Potting et al. 1999). Bt oilseed rape leaves damaged by Bt-resistant P. xylostella were highly attractive to C. plutellae females and it is possible that the ability of C. plutellae to locate and parasitise Bt-resistant hosts on transgenic crops, or crops sprayed with microbial Bt formulations, might assist with constraining the spread of genes for Bt resistance in the field. In contrast, Bt oilseed rape leaves damaged by susceptible P. xylostella were not very attractive to the parasitoid because the susceptible larvae caused only very limited feeding damage (Schuler et al. 1999). Any C. plutellae eggs deposited inside susceptible P. xylostella larvae on Bt plants would represent a waste of parasitoid resources, but the lack of attraction to this plant-host complex is likely to limit attacks of Bt-susceptible hosts in the field.

Many synthetic insecticides used for control of P. xylostella have broad spectrum contact action (Sivapragasam et al. 1997, Talekar & Shelton 1993) and cause not only parasitoid mortality indirectly through premature host mortality, but also have acute contact toxicity for adult parasitoids. In this respect, Bt plants offer an environmental advantage over broad spectrum synthetic insecticides.

In some areas P. xylostella has already developed resistance to microbial Bt sprays (Shelton et al. 2000, Tabashnik 1998, Verkerk & Wright 1997) and any commercial use of Bt brassicas has to be approached with great caution. The ability of parasitoids to detect hosts through plant-volatiles at long range, coupled with the observation that C. plutellae develops normally in Bt-resistant hosts, may contribute to resistance management. The conservation of specialist natural enemies such as C. plutellae in or near Bt crops will partly depend on the presence of weedy hosts and/ or untreated crops in the vicinity and may require active management. Further research is necessary to determine the effect of parasitoids on the development of resistant pest populations and to investigate the potential of non-Bt refuge areas for conserving parasitoids as well as Bt-susceptible pests.

The present study provides an example of how realism can be introduced into laboratory studies investigating the effect of transgenic plants on non-target organisms. Experimental methodology presented here is recommended as part of a three-tiered scheme for assessing risks of GM plants on non-target organisms (Schuler et al. 2000).

Acknowledgements

We acknowledge support for this work from the Department of the Environment, Food and Rural Affairs (DEFRA) of the UK. IACR-Rothamsted receives grant-aided support from the Biotechnology and Biological Sciences Research Council of the UK. We thank Prof. B. E. Tabashnik for providing the Bt-resistant P. xylostella strain, Prof. C. N. Stewart for supplying the oilseed rape lines, A. J. Clark for technical assistance and S. J. Clark for statistical analysis.

References