Increased mortality is predicted of Inachis io larvae caused by Bt-maize pollen in European farmland
Highlights
► We modelled the phenology of Bt-maize pollen and Inachis io larvae. ► Second generation larvae coincided with the presence of Bt-maize pollen. ► The dose–response model of larvae on Bt-maize pollen predicted increased mortality. ► Our model differs in design from earlier models; it has more biological detail. ► Our model differs in predictions from earlier models; we found considerable effects.
Introduction
Transgenic crops are grown over large areas in the America (James, 2011). While in Europe imports prevail and only few transgenic crops have been authorised for cultivation, currently only transgenic Bt-maize (event MON810) is grown commercially and mostly in Spain (CERA, 2010). Seven European countries have a moratorium on growing transgenic crops under field conditions, and in several cases the reason is concerns over possible negative environmental impacts. The environmental effects of transgenic crops vary in direction and magnitude (Andow et al., 2006, Catacora-Vargas, 2011) and include effects on ecosystem services (Arpaia, 2010, Lövei, 2001).
Bt-maize, expressing Cry toxins harmful to Lepidoptera or Coleoptera (Koziel et al., 1993, Van Rie, 2000), is cultivated on a large scale (e.g., in the USA; see USDA, 2011). As most promotors are constitutive, the Bt protein is expressed in all plant tissues, including pollen, in which toxin concentration is highly variable (Szekacs et al., 2010). Herbivores ingesting such pollen may suffer toxic effects. For example, wind dispersed pollen from Bt-maize deposited on the leaves of milkweed (Asclepias sp.) can cause mortality to Monarch butterfly (Danaus plexippus) larvae (Losey et al., 1999). Lepidoptera-specific Bt toxins affect most Lepidoptera but species differ greatly in their susceptibility (van Frankenhuyzen, 2009, Peacock et al., 1998). The effects of Bt toxins include increased mortality, reduced growth rate, prolonged development time, and reduced adult body mass and size (Lang and Otto, 2010).
Since maize pollen is deposited in the vicinity of maize fields (Hofmann et al., 2010, Pleasants et al., 2001;), lepidopteran larvae occurring in field margins or habitats nearby Bt-maize fields are possibly at risk (Losey et al., 2003). In Europe, several butterfly and moth species occurring in arable landscapes could be affected (Felke and Langenbruch, 2005, Lang et al., 2004, Schmitz et al., 2003). Of all Austrian butterfly species, 70% are found in agricultural landscapes, of which only eight species would not risk exposure to Bt-maize pollen; the larvae of most species would temporally overlap with the maize pollen shedding period (Traxler et al., 2005).
The population-wide effect of Bt-maize on a species will depend on the species-specific susceptibility to Bt toxins, the actual exposure to the toxin and the ecology of the species. Yet, only few risk assessments studies take these factors into account. For instance, Dively et al. (2004) detected adverse effects of Bt-maize (events MON810 and Bt11) on the Monarch butterfly, but concluded that the natural population would suffer no harm from Bt-maize cultivation due to a low exposure of the larvae (cf. Sears et al., 2001). A similar conclusion was reached for the pale grass blue butterfly (Pseudozizeeria maha) in Japan (Wolt et al., 2005). In a more detailed analysis, Peterson et al. (2006) modelled the phenology of both maize pollen shedding and larval feeding and combined this with a GIS-based analysis of the co-occurrence of Bt-maize fields and larval habitats. For the species in question (Lycaeides melissa samuelis), they concluded that in most places and for most years, maize pollen shedding would occur after the majority of the butterfly larva population had finished eating. However, while most larval habitats where at a safe distance from Bt-maize fields, there were a few regions where they were interspersed and monitoring of the butterfly population was recommended.
In order to estimate the risk involved with Bt-maize in Europe, Perry et al. (2010) modelled the effects of MON810 Bt-maize cultivation on two butterfly species (Inachis io, Vanessa atalanta) and one moth (Plutella xylostella). In contrast to the model of Peterson et al. (2006), their model did not explicitly include the mechanisms of phenology to simulate temporal overlap of maize pollen and larvae, nor was the calculation of spatial overlap presented in detail. The use of generic parameters, subsuming several mechanisms and informally estimated by the authors, were later debated (Lang et al., 2011, Perry et al., 2011); a more recent version of the model maintains this design (Perry et al., 2012).
In this paper, we present a new model BtButTox which includes the mechanisms of temporal coincidence between pollen shedding and larval feeding. We focused on the nettle (Urtica dioica) feeding butterfly species, Peacock butterfly (I. io) for which empirically based parameter estimates could be obtained from literature, both on the dose–response curve of larvae exposed to Bt-maize pollen and the phenology. Moreover, we chose specific scenarios in regions of Germany where historical data exists, from which the expected period of maize pollen shedding could be derived. The model was implemented in a transparent object-oriented design as open source code. The model predicted that in Northern Germany, where I. io is mostly univoltine, the cultivation of Bt-maize would pose a negligible risk, because pollen shedding is predicted later than larval feeding. But in Southern Germany, where I. io is bivoltine, the second generation of larvae coincides with the peak of maize pollen deposition and consequently is at risk.
Section snippets
Model design
The model follows the object-oriented modelling paradigm for model design (Silvert, 1993). Object-oriented software is ideally composed of well-defined, loosely coupled objects, in our case model components, with a minimal interface to other objects (Martin, 2006). The BtButTox model presented here consists of three model components simulating (1) the pollen density on nettle leaves, (2) the dose–response curve of I. io larvae to pollen density and (3) the phenology of I. io taking the imposed
Results
The I. io phenology model reproduced well the expected periods when adults are usually observed flying (Fig. 5 top), both in scenario 1 (univoltine) and scenario 2 (bivoltine). In 1996 some pupae in scenario 2 did not eclose as adults before the onset of winter. It is not known whether this actually happened, but the phenomenon had no impact on the analysis, because it was the number of larvae reaching pupation that was used to express final population size.
In scenario 1 the single generation
Discussion
We detected a risk to I. io living in an agricultural landscape where Bt-maize is grown. However, the risk depended on I. io phenology which depends on the climate. Only in Southern and Central Europe (scenario 2), where the species is bivoltine, would the period of larval feeding coincide with that of maize pollen shedding (Fig. 5).
The peak average density of pollen on nettle leaves (Npeak, Fig. 2) was important for the resulting reduction in I. io population survival (Fig. 6). The lower
Conclusions
Our model indicated that in European farmland, the exposure of butterfly larvae to Bt-maize toxin constitutes a realistic risk. Specifically I. io is at risk in Central and Southern Europe where it is bivoltine. This suggests that a more comprehensive assessment is warranted of the risk implied to butterflies when and where Bt-maize is grown. We contend that such an assessment is best carried out using empirical data, which invites scientific review and integration of knowledge, rather than on
Acknowledgements
We thank Tobias Vogt and Dr. Klaus-Peter Wittich from the German National Meteorological Service (DWD) for the provision of temperature data used in this model and, two anonymous reviewers for helpful comments. This work was supported by the German Federal Agency for Nature Conservation. Authorship in this paper was determined by ranking.
References (67)
- et al.
An aerobiological framework for assessing cross-pollination in maize
Agricultural and Forest Meteorology
(2003) - et al.
Bacillus thuringiensis Cry3Aa toxin increases the susceptibility of Crioceris quatuordecimpunctata to Beauveria bassiana infection
Journal of Invertebrate Pathology
(2012) - et al.
The effects of pollen consumption of transgenic Bt maize on the common swallowtail, Papilio machaon L. (Lepidoptera, Papilionidae)
Basic and Applied Ecology
(2006) - et al.
Winter predation on two species of hibernating butterflies: monitoring rodent attacks with infrared cameras
Animal Behaviour
(2011) Object-oriented ecosystem modeling
Ecological Modelling
(1993)Bacillus thuringiensis and its use in transgenic insect control technologies
International Journal of Medical Microbiology
(2000)- Abkin, M.H., Wolf, C., 1976. Distributed Delay Routines. Class Document No. 8. Dept Agric Econom, Michigan State Univ.,...
- et al.
Bt transgenic crops, natural enemies and implications for environmental risk assessment
Nature Biotechnology
(2006) Genetically modified plants and non-target organisms: analysing the functioning of the agro-ecosystem
Collection of Biosafety Reviews (ICGEB Biosafety Unit)
(2010)- et al.
Thermal ecology of gregarious and solitary nettle-feeding nymphalid butterfly larvae
Oecologia (Berlin)
(2000)