1 Introduction

Genetically modified herbicide-resistant (GMHR) sugar beet (Beta vulgaris L.) was deregulated in 2006 and introduced to the US and Canadian market in 2007. It was rapidly adopted by farmers and now accounts for nearly 95 percent of the acreage of all sugar beets produced in the US. The GMHR sugar beet is resistant to the herbicide active ingredient glyphosate. An application for cultivation of the variety H7-1 was made in Europe in 2008 and is currently in the approval process under Directive 2001/18/EC (European Commission, 2001*). This regulatory framework specifies a systematic environmental risk assessment (ERA) and mandatory post-market environmental monitoring (PMEM) after approval. This is designed to handle uncertainties about potential adverse environmental effects still remaining after the ERA, which is primarily based on short-term and field-scale releases of the GM crops.

Various public and scientific concerns have been raised both in the US and in the EU about potential adverse agro-environmental effects of widespread GMHR sugar beet cultivation, leading to its provisional ban in some US states (Burkett, 2010). The concerns include cross-pollination with both conventional non-GM and organic Beta vulgaris varieties cultivated for sugar or for seed production (Lange et al., 1999*; OECD, 2006*), thereby reducing or even destroying their market value. Hybridisation with and gene flow to related wild species within the Beta section may occur (Bartsch et al., 2003*; Sukopp et al., 2005*), forming normally vigorous HR hybrids that may spread in the environment. The modified herbicide application regime with glyphosate includes multiple changes in cultivation practice (Graef, 2009*) that may adversely impact the agro-environment and cause negative weed flora changes (Heard et al., 2003b*,a*; Roy et al., 2003), direct toxic environmental effects (Benachour and Séralini, 2009*; Brausch and Smith, 2007*; Relyea, 2005a*,b*,c*), or the development of weed resistance through selection pressure (Heap, 2009; Zelaya et al., 2007). The concerns cover a wide range of environmental, agronomic, economic, and social aspects which are often partly interlinked, but our review focuses primarily on potential environmental and agronomic aspects of GMHR sugar beet cultivation.

Currently, the only GM crops approved for cultivation purposes in the EU are the Bt-Maize MON 810, which has been cultivated primarily in Spain, and the Amflora potato, which was approved in early March 2010. Despite their regulation under Directive 2001/18/EC (European Commission, 2001*) and under the Supplementing Guidance Notes (European Commission, 2002*), both the general approaches to ERA and the PMEM of GM crops are under continuing scientific debate. The quality and quantity of data provided on the ERA process presently does not satisfy scientific and technical standards (Dolezel et al., 2009*). Moreover, the monitoring plans submitted and the monitoring reports presented by the applicants (BVL, 2007) still lack a profound science-based design. More general guidance on the ERA was developed by EFSA (2008*). PMEM guidance was presented by ACRE (2004*), more sharply delimiting the different monitoring intensity categories of general surveillance and case-specific monitoring.

Various additional scientific aspects have been developed and outlined in the last few years. These include a more targeted ERA involving well-defined hypotheses, precisely defined and prioritized hazards and quantifying elements of exposure (Andow and Hilbeck, 2004; Johnson et al., 2007; Wilkinson et al., 2003), the selection of test species for the risk assessment for non-target organisms (Hilbeck et al., 2008a*), the selection of indicator organisms for the PMEM (Hilbeck et al., 2008b*), systematic approaches for landscape-scale or ecoregion-based PMEM (Graef et al., 2005a*,b*), a science-based systematic step-by-step approach in PMEM (Züghart et al., 2008*), the enhanced quality of statistics required for the ERA (Lövei and Arpaia, 2005*; Perry et al., 2009*), and a proposal for the definition of environmental damage (Bartz et al., 2009*). Nonetheless, open questions and shortcomings in the present ERA and monitoring practice remain. Key issues to be further tackled are

• improvements in guidance and standardization of risk assessment methodology, e.g., guidance on selecting representative locations for the assessment of agronomic and environmental behaviour of a particular GM crop, on the details of field trial designs, and on the risk assessment of long-term and cumulative effects (Dolezel et al., 2009*),
• normative indications and thresholds for ecological hazards and damages associated with GM crops (Bartz et al., 2009*; Breckling et al., 2009*; Regal, 1994*),
• the necessary field test and PMEM design required to yield scientifically sound data, (De Jong, 2010; Graef et al., 2005b*; Lövei and Arpaia, 2005*; Perry et al., 2009*; Züghart et al., 2008*)
• the ERA and monitoring data management and technical implementation using structured databases (Reuter et al., 2010a),
• the methodology of the ERA and monitoring of GM crops with multiple stacked transgenic events within one crop (De Schrijver et al., 2007),
• the methodology of data upscaling and interpretation as more field testing and monitoring data become available with more widely spread GM crop cultivation (ACRE, 2004; Breckling et al., 2009*; Squire et al., 2009*),
• the monitoring data coordination and harmonisation at national and/or EU levels (Finck et al., 2006*; Graef et al., 2008*),
• the applicability and use of existing national and/or EU environmental monitoring programmes and data infrastructure schemes for genetically modified organism (GMO) monitoring (Graef et al., 2005b*; EU Monitoring Working Group, 2010).

The uncertainty connected to these key issues is also reflected in the often contradictory comments of EU member state experts during GM crop approval processes. Research is underway to tackle some of these shortcomings, for instance in national research programmes or as part of the Framework Programme on research by the European Commission.

So far, environmental risk-related data on GM plants is mainly concentrated on the lower levels of spatial extension such as molecular detection, laboratory trials, and short-term greenhouse or field studies to assess effects on the population level. Experiments on larger landscape scales are sparse, the most prominent being the Farm Scale Evaluations in the UK (Firbank et al., 2003*). The usual ERA practice is to analyse and assess the greenhouse and field-scale results and extrapolate them to the European scale of (future) crop cultivation, potentially entailing large inference errors. Extrapolating ecological effects of GM crops from field scale to larger landscape scales, however, requires an up-scaling approach based on reliable data on various scales of GM exposition. This has been demonstrated in a special issue of the journal “Ecological Indicators” with GM oilseed rape (Brassica napus L.) (Breckling et al., 2009*; Middelhoff et al., 2010; Reuter et al., 2010b). Many of those results are generally valid for other GM crops such as for GMHR sugar beet: it has certain biological features in common with oilseed rape, for instance wind-pollination, hybridisation with wild relatives, and persistent seeds in the soil.

It is important mentioning that not only GM crop but also the nonGM crop cultivation may entail various environmental effects that can occur on different spatial and temporal scales. Furthermore, the ecological importance of environmental effects is difficult to determine and may vary depending on the type of effect. Little focus has been placed on ERA and PMEM of sugar beet. This literature review is designed to identify likely adverse effects of GMHR sugar beet cultivation at the various spatio-temporal scales relevant for the ERA and the PMEM. The key question is whether experimental greenhouse- or field-scale-based data on specific potential adverse effects of GMHR sugar beet cultivation are scientifically-based and sufficient to be upscaled to larger areas such as landscapes. We thus identify the results and shortcomings of small- (field-) scale findings and indicate whether they enable inferring the outcomes at larger scales and on required PMEM.

In the context of this paper the term ‘field organism’ is defined as all organisms living in or visiting the field and its margins, such as plants, epigeic and endogeic invertebrates, birds, mammals and amphibians. The term ‘herbicide resistance’ is applied according to the WSSA (Weed Science Society of America) definition as the inherited ability of a plant to survive and reproduce following exposure to a dose of herbicide normally lethal to the wild type. In plants, resistance may be naturally occurring or induced by techniques such as genetic engineering or selection of variants produced by tissue culture or mutagenesis. The term ‘agro-environment’ is used here as the cultivated area along with neighbouring fields and biotopes. The different terms for scale levels used in this review are a) ‘laboratory scale’, b) ‘greenhouse scale’, c) ‘field scale’, which applies to pre-commercial experimental and large-scale field trials (European Commission, 2002*) that are limited in number, their extension, and the duration of observations, and d) ‘landscape scale’. The latter is the commercial GM plant cultivation scale and may range from smaller ecoregions, for instance at the 1:50,000 scale (Graef et al., 2005b*), to larger climatic regions or ecoregions (Bailey, 2002; EFSA, 2008) or even (bio)geographical regions at the national or European scale (Eiden et al., 2000; Metzger et al., 2005).