Herbicide-resistant crops as weeds in North America

Transcription

1 Review Herbicide-resistant crops as weeds in North America Hugh J. Beckie 1, * and Micheal D. K. Owen 2 Address: 1 Agriculture and Agri-Food Canada, 107 Science Place, Saskatoon, Saskatchewan, S7N 0X2, Canada Agronomy Hall, Iowa State University, Ames, IA 50011, USA. *Correspondence: Hugh J. Beckie. beckieh@agr.gc.ca Received: 27 April 2007 Accepted: 18 June 2007 doi: /PAVSNNR The electronic version of this article is the definitive one. It is located here: g CABI Publishing 2007 (Online ISSN ) Abstract CAB Reviews: Perspectives in Agriculture, Veterinary Science, Nutrition and Natural Resources , No. 044 Growers have rapidly adopted transgenic herbicide-resistant (HR) crops, such as canola (Brassica napus L.), soybean [Glycine max (L.) Merr.], maize (Zea mays L.) and cotton (Gossypium hirsutum L.), across North America (USA and Canada) since their commercial introduction in the 1990s. With their widespread cultivation, increasing attention is focused on management of HR volunteers in crops that follow in rotation. In this review, we describe the impact and management of HR crop volunteers in different agroecosystems in North America. The relative risks of planting HR crops and subsequent potential for volunteerism of these crops are assessed. HR volunteers are common weeds and the relative weediness depends on species, genotype, seed shatter prior to harvest and disbursement of seed at harvest, management practices, and environment. Chemical control options may be more limited if the crop volunteers are HR. There are generally no marked changes in volunteer weed problems associated with these crops, except in no-tillage systems when glyphosate (GLY) is used alone to control volunteers. The increasing use of GLY in North American cropping systems, spurred by increasing area and frequency in rotation of GLY HR crops, may require increased alternative herbicide use or other novel tactics to control GLY HR crop volunteers. Keywords: Volunteer crop, Transgenic crop, Gene flow, Herbicide resistance, Agrostis stolonifera, Beta vulgaris, Brassica napus, Glycine max, Gossypium hirsutum, Medicago sativa, Oryza sativa, Triticum aestivum, Zea mays Introduction Approximately 700 million ha are planted to major crops worldwide [1]. Transgenic herbicide-resistant (HR) crops currently represent only one-tenth of this area [2]. Globally, resistance to non-selective, translocated herbicides (glyphosate, GLY; glufosinate, GLU) is the dominant trait in transgenic crops (68%, stacked traits excluded) (Table 1 [2 4]); GLY-HR crops predominate. HR soybean [Glycine max (L.) Merr.] comprises the largest area at 58.6 million ha or 57% of the area planted to transgenic crops. Other important transgenic-hr crops include maize (Zea mays L.), cotton (Gossypium hirsutum L.) and canola (Brassica napus L.). Worldwide, transgenic-hr cultivars comprise 60% of soybean, 30% of cotton, 20% of canola and 15% of maize plantings. The USA is the largest producer of transgenic-hr crops, followed by Argentina, Brazil and Canada [2]. In contrast to other countries, Canada has a unique regulatory approach that is based on substantial equivalence and trait novelty [5]. Plants with novel traits (PNTs) are regulated by the Plant Biosafety Office of the Canadian Food Inspection Agency (CFIA). A PNT is defined as a plant containing a trait not present in plants of the same species already existing as stable, cultivated populations in Canada, or is present at a level significantly outside the range of that trait in stable, cultivated populations of that plant species in Canada. PNTs may be produced by conventional breeding, mutagenesis, or more commonly, by recombinant DNA techniques [6]. Therefore, Canada regulates a broader range of HR crops, including those with herbicide resistance selected following mutagenesis

2 2 Perspectives in Agriculture, Veterinary Science, Nutrition and Natural Resources Table 1 Transgenic 1 crops grown from 2004 to 2006, listed by country, trait, and crop (adapted from James [2 4]) ha¾10 6 % ha¾10 6 % ha¾10 6 % By country: United States Argentina Brazil Canada India China Paraguay South Africa < Other 0.4 < < Total countries Total area By trait: Herbicide resistance (HR) Bt (B. thuringiensis) HR+Bt By crop: Soybean [G. max (L.) Merr.] Maize (Z. mays L.) Cotton (G. hirsutum L.) Canola (B. napus L.) IMI-HR crops are non-transgenic and therefore excluded. or introgressed following outcrossing. In the USA, the US Department of Agriculture, Animal and Plant Health Inspection Service (USDA-APHIS) regulates transgenic crops. Approvals for unconfined environmental release (deregulation) of HR crops have been granted in Canada and the USA since The first approved crops possessed input traits, mainly herbicide resistance, that enabled them to tolerate the application of selective (imidazolinones, IMIs; bromoxynil, BX) or non-selective (GLY, GLU) herbicides. The GLY-, GLU- and BX-HR crops were produced by recombinant DNA techniques through the insertion of bacterial genes (i.e. transgenic), whereas IMI resistance was attained by chemical mutagenesis (i.e. non-transgenic) [7 10]. HR field crops registered in Canada include canola (denotes Argentine canola or spring oilseed rape in the remainder of the paper), soybean, maize, wheat (Triticum aestivum L.) and lentil (Lens culinaris L.); in the USA, additional crops include cotton, rice (Oryza sativa L.), sugarbeet (Beta vulgaris L.), sunflower (Helianthus annuus L.), alfalfa (Medicago sativa L.) and creeping bentgrass (Agrostis stolonifera L.) (Table 2). In 2006, IMI-HR lentil cultivars became commercially available, but in limited seed supply. In Canada, canola occupies over 80% of the area cultivated to HR crops, followed by soybean, maize and wheat in order of decreasing area. In the US, soybean, cotton and maize comprise most of the HR crop area. Volunteer plants of one crop growing in another crop can be a significant weed problem. Volunteers are considered to be weeds because they can reduce crop yield and quality, as well as harvesting efficiency. Volunteer crops may harbour pathogens, insects and nematodes, thereby diminishing the positive effects of crop rotation on pest management [11]. Volunteer HR plants can facilitate intra- or interspecific HR gene flow in space or time. For example, the transgenic-hr traits in volunteer canola can introgress in sympatric weedy populations of bird s rape (Brassica rapa L.) in eastern Canada [12]; similarly, the IMI-HR trait in volunteer wheat can introgress into jointed goatgrass (Aegilops cylindrica Host) in the western USA [13]. There are increasingly stricter regulations to mitigate HR volunteer seed contamination of seedlots of identity-preserved crops [14]. Globally, many concerns have been raised about the release of transgenic crops. In this review, we will focus on the impact of HR crop volunteers on different agroecosystems in North America (USA and Canada only) after 10 years of growing HR cultivars. We draw upon extensive and intensive field trials, field-scale experiments, grower surveys, and post-release monitoring to assess impacts of each crop relative to their non-hr crop counterpart, whenever possible. HR canola will be discussed most extensively because it has been assessed more comprehensively than the other crops. Canola has a partially outcrossing breeding system and weedy attributes (e.g. volunteerism and seed shattering), and is cropped across a large area of the northern Great Plains of Canada and the USA. Based on our impact assessment, we make recommendations for improved stewardship to mitigate potential adverse effects of HR crop volunteers.

3 Table 2 HR field crops registered in Canada and the USA, and weediness in cropped land, non-cropped disturbed areas (including roadsides and waste ground), and natural areas (adapted from Beckie et al. [70]; Owen, unpublished data) Species HR Variety registration Regulatory approval Breeding system Alfalfa (M. sativa) GLY NA 1 Yes 2 Highly outcrossing Weediness Crop-land Disturbed areas Natural areas Yes Yes Yes Canola (B. napus) GLY Yes Yes ca. 30% Yes Yes No GLU Yes Yes outcrossing IMI Yes Yes BX Yes Yes Maize (Z. mays) GLY NA Yes Highly Yes No No GLU NA Yes outcrossing IMI NA Yes Sethoxydim NA Yes Cotton (G. hirsutum) GLY NA Yes Selfing Rarely No No GLU NA Yes BX NA Yes Creeping bentgrass GLY NA Yes 2 Highly Yes Yes Yes (A. stolonifera) outcrossing Lentil (L. culinaris) IMI Yes Yes Highly selfing Yes No No Rice (O. sativa) IMI Yes Yes Highly selfing Yes No No GLU No No Soybean (G. max) GLY Yes Yes Highly selfing Rarely No No GLU No Yes Sugarbeet (B. vulgaris) GLY NA Yes Selfing Rarely No No GLU NA Yes Sunflower (H. annuus) IMI Yes Yes Selfing Yes No No Wheat (T. aestivum) IMI Yes Yes Highly selfing Yes No No 1 NA: not required. 2 Currently restricted due to US District Court order. Hugh J. Beckie and Micheal D. K. Owen 3 Adoption of HR Crops USA Canola In 2006, ha were planted to canola in the USA [15]. North Dakota accounts for greater than 90% of US canola production, with transgenic-hr cultivars grown across most of the canola area. In 2006, GLY and GLU-HR cultivars comprised 65 and 32% of the canola area, respectively (K. Howatt, personal communication). Soybean Transgenic-HR soybean became commercially available in the USA in The cultivars utilize the CP4 gene from Agrobacterium sp. that codes for a GLY-HR form of enolpyruvylshikimate-3-phosphate synthase (EPSPS, EC ), which results in effective safety to commercialuse rates of GLY. Soybean cultivars with the GLY-HR trait represent an estimated 89% of soyabean planted in the USA in 2006 (Figure 1a). While GLU-HR soybean cultivars possessing the bar gene from Streptomyces hygroscopicus are under development, none are currently commercially available. A newly reported mechanism, N-acetylation of GLY, provides considerable resistance to GLY and is currently under development in soybean and other crops [16]. Soybean cultivars demonstrating metabolic resistance to GLY based on N-acetylation are not expected to be commercially available until 2009 at the earliest ( J. Green, personal communication). The adoption of GLY-HR soybean cultivars has been rapid and continues to increase, albeit at a slower pace given the current hectares planted to these cultivars (Figure 1a). Factors that have influenced grower adoption of GLY-HR soybean cultivars include the perception of simplicity of tactics for effective weed control, consistency of efficacy and favourable economics compared with weed control in conventional systems. A favourable environmental profile for systems based on GLY-HR soybean is also suggested to be an important consideration by growers. Notably, fewer tillage passes are required in GLY-HR soybean systems, resulting in significant reductions in soil erosion as well as a savings on petroleum fuel. Gianessi et al. [17] reported a savings of US$385 million in tillage and an estimated savings of US$ million in weed control costs attributable to the use of GLY-HR soybean cultivars. Given the current price of petroleum products, it is reasonable to suggest that the savings experienced by growers would be greater today. However, there are significant concerns about the widespread adoption of GLY-HR soybean. Notably, the perception of simplicity for weed control may not be accurate. While growers can control larger weeds, the

4 4 Perspectives in Agriculture, Veterinary Science, Nutrition and Natural Resources A Planted area (%) Maize Cotton Soybean B Planted area (%) Maize Soybean Cotton Figure 1 Adoption of transgenic HR maize, soybean and cotton in the USA (A) and percentage of planted area treated with GLY (B). Adapted from USDA [15] late applications likely result in reduced yield potential attributable to weed interference. Furthermore, growers tend to use GLY as the sole herbicide for weed control (Figure 1b) and often apply GLY at multiple times during the growing season. This high intensity of use increases the selection pressure for evolved GLY resistance and weed population shifts, thus resulting in problems for weed management [18]. Primarily found in soybean fields, more than a dozen US states have reported GLY-resistant horseweed [Conyza canadensis (L.) Cronq.] [19]. Another concern, albeit not quantifiable, is a fear that due to the widespread adoption of GLY-HR soybean cultivars, seed companies will find it economically challenging to continue breeding programmes in non-hr soybean cultivars. Thus, these cultivars may ultimately disappear from the marketplace, or at least lag GLY-HR cultivars in agronomic performance. Cotton Similarly to that for soybean, cotton resistance to GLY uses the CP4 EPSPS, and grower adoption of the GLY-HR cultivars has been rapid since their introduction in 1997 [20, 21]. In 2005, 79% of the cotton planted were transgenic-hr cultivars and adoption increased to 83% in 2006 (Figure 1a). Interestingly, cotton hectares are expected to drop by 20% in 2007, presumably as a result of the increased interested in maize grown for ethanol [22]. Cotton with resistance to BX was introduced in the USA in 1994, and GLU-HR cultivars were introduced in 2003 [23]. However grower adoption of these HR cotton cultivars has not been great. It is estimated that GLY-HR cotton cultivars have saved growers US$132 million per year [24]. Most of the savings were attributable to reduced herbicide costs and significantly fewer tillage passes. The latter also represents an important environmental benefit when compared with previous cotton production systems. The shift toward no-tillage systems for cotton production was largely attributable to the level of weed control using GLY-based systems compared with traditional production systems. Similar to soybean, however, the intense selection pressure from GLY in cotton (Figure 1b) resulted in evolved GLY resistance problems in horseweed, which is adapted to no-tillage production systems [25, 26]. Given the lack of alternative herbicides that effectively control horseweed in cotton, growers had to reinstitute significant tillage in cotton production, particularly in the Delta region of the USA. Moreover, in the Southeast cotton production region, particularly in Georgia, GLY-HR Palmer amaranth (Amaranthus palmeri S. Wats.) has become a significant problem that threatens the economic viability of cotton production [27].

5 Maize Maize cultivars with resistance to herbicides include both transgenic (GLY and GLU) and non-transgenic (sethoxydim and IMI) hybrids. IMI-HR hybrids were introduced in 1993, sethoxydim-hr hybrids in 1996, GLY-HR hybrids in 1997 and GLU-HR hybrids in 1998 [28]. Sethoxydim-HR maize hybrids were not an economic success, but did provide interesting options for grass weed control in maize [29]. IMI-HR maize hybrids are also currently available. These hybrids were selected either by pollen mutagenesis and appropriate herbicide selection of the HR trait, or through a process using cultured maize cell lines [30, 31]. The result of the different tactics to accrue IMI resistance was the selection for an altered acetolactate synthase (ALS, EC ) that demonstrated fold increased resistance to specific IMI herbicides, but limited cross-resistance to other ALS-inhibiting herbicide classes [31]. The current market share of maize with resistance to IMI herbicides is not believed to be large. Transgenic-HR maize hybrids are preferred by growers over non-transgenic HR hybrids (Figure 1a), with GLY-HR hybrids dominating. GLU-HR hybrids are available, but are estimated to be planted on only about 5% of the US maize hectares. Resistance to GLU in maize is attributable to the inclusion of the bar gene from S. hygroscopicus that results in the acylation of GLU to non-herbicidal metabolites [32]. GLY resistance in maize is the result of either the CP4 gene that codes for an altered EPSPS that does not allow binding of GLY, or N-acetylation of GLY resulting in the non-herbicidal metabolite N-acetyl GLY [33, 34]. GLY-HR maize was planted on 36% of the US hectares in 2006, which represents an 8% increase from 2005 (Figure 1a). Transgenic-HR maize accounted for 21% of the planted area, while transgenic hybrids with insect and herbicide resistance were planted on an estimated 15% of the planted area. Grower adoption of transgenic-hr maize has lagged behind that of transgenic-hr soybean and cotton. The difference in adoption rate of maize compared with that of soybean or cotton is likely attributable to a number of factors, including unavailability of the transgenic-hr trait in the higher-yielding hybrids, availability of alternative herbicides and the perception that maize is more sensitive to early-season interference by weeds resulting in significant loss of yield potential [24]. Current developments by seed companies have resolved the concern for yield potential in the transgenic-hr hybrids; however, the latter concern still resonates with many growers. Gianessi [24] reported an aggregate savings for growers of US$58 million in 2001, attributable to the adoption of weed control systems based on transgenic-hr traits. Given the increase in maize prices and the reduction in GLY cost, it is suggested that these savings may be greater now. The adoption of transgenic-hr maize hybrids appears to be increasing at an increasing rate [28, 35]. The area planted to transgenic-hr maize is expected to grow dramatically in the near future, given the emphasis on Hugh J. Beckie and Micheal D. K. Owen 5 biofuels production. Planting intentions for 2007 indicate a significant 15% increase in maize at the expense of soybean, rice and cotton, which are suggested to decline by 11, 7 and 20%, respectively. Growers feel that the perceived simplicity of transgenic-hr maize is important. Given the expected growth in maize hectares, time management will become a greater problem; the ability to spray a herbicide post-emergence (POST) and achieve effective and consistent weed control without concern for crop injury will likely drive decisions to more transgenic- HR maize. However, growers must consider the risk of allowing weeds to co-exist with the maize crop, thus potentially losing significant yield potential [36]. Rice GLU-HR rice was initially developed to manage weedy red rice (O. sativa L.) [37]. Field testing occurred from 1998 to However, marketing concerns precluded commercialization of the transgenic-hr rice cultivars. Non-transgenic (IMI)-HR rice cultivars were developed and commercialized in 2002 [38]. Grower adoption of the IMI-HR cultivars has increased steadily since their introduction. However, concerns have been expressed about the potential outcrossing of cultivated rice hybrids with red rice, and specifically the HR trait whether transgenic or non-transgenic [38 40]. Recently, the regulated transgenic-hr trait for GLU resistance (cultivar LL 601 ) was found in unregulated IMI-HR rice cultivar CLEARFIELD 1 CL131. This discovery resulted in Japan suspending importation of US long-grain rice. This problem continues to be investigated. Because of the potential for pollen-mediated gene flow and subsequent introgression, the adverse consequence was not unexpected. Wheat Wheat was planted to approximately 25 million ha in the USA in 2006 [15]. However, there are no transgenic-hr cultivars. The program to develop GLY-HR wheat cultivars was terminated in May 2004 [28]. The problems perceived by the registrant (Monsanto) included regulatory approvals, marketing approvals, stewardship programmes and grain-handling issues. Given these concerns, the registrant voluntarily deferred further efforts to introduce GLY-HR wheat into the USA and Canada. Interestingly, the technical fit of the GLY-HR wheat cultivars with current production systems was determined to be generally favourable [41]. A 10% yield advantage, primarily attributable to better weed control, was reported in the GLY-HR cultivars. However, the authors cautioned that should the technology be commercialized, appropriate stewardship practices are required to mitigate volunteer crop plants and the expected changes in the weed community [42]. Non-transgenic (IMI)-HR wheat was introduced in 2002 [23, 43]. The commercial IMI-HR winter wheat cultivars are Above and AP502 CL [44]. It is suggested that losses attributable to winter

6 6 Perspectives in Agriculture, Veterinary Science, Nutrition and Natural Resources annual grasses account for US$30 million, and that the IMI-HR wheat cultivars will be important tools to resolve this loss. However, the adoption of IMI-HR wheat is estimated to be relatively minor, and requires growers to participate in a stewardship program involving a number of stewardship tactics to protect this technology. Sunflower Resistance to IMI herbicides was introduced into domesticated sunflower through conventional breeding methods [45]. The IMI-HR gene was derived from IMI-HR weedy common sunflower. IMI-HR commercial hybrids were released to US growers in In Canada, IMI-HR sunflower was approved for unconfined environmental release [46], but is not currently grown. Anecdotal information suggests that the IMI-HR sunflower cultivars are not widely adopted by growers (R. Zollinger, personal communication). The potential for intra- and interspecific hybridization within Helianthus spp. has long been recognized [47 49]. Furthermore, the potential for the introgression of the HR trait has been demonstrated [50]. Thus, given the relatively high potential for evolved resistance to ALS-inhibiting herbicides in common sunflower [51, 52] and gene flow between cultivated and weedy sunflower, it is presumed that growers do not find the advantages of the HR cultivars to be pervasive reasons to widely adopt the trait. Currently, resistant cultivars to GLY and GLU are not commercially available. Alfalfa Alfalfa represents the third-most economically important crop in the USA, with over 9 million ha planted in 2006 [15]. Transgenic-HR alfalfa with resistance to GLY was deregulated and subsequently commercialized in 2005, becoming the fourth major HR crop grown in the USA. The deregulation occurred in the face of immediate concerns about the ability of growers to contain the trait. Given that alfalfa is an open-pollinated crop, and that pollination is accomplished by bees which can travel considerable distances with viable pollen, contamination in non-transgenic alfalfa seed was expected. In 2006, an estimated ha were planted to the GLY-HR cultivar [53]. In early 2007, a preliminary injunction order was issued indicating that USDA-APHIS had erred when the GLY-HR alfalfa was deregulated (US District Court for the Northern District of California, No. C CRB), thus halting seed sales and planting after 30 March 2007 [54, 55]. The general message from the ruling was that transgenic crops need better risk assessment [56]. However, GLY-HR alfalfa previously established could be harvested and sold. As a consequence of the court decision, grower adoption of transgenic-hr alfalfa will not be great in the foreseeable future. Creeping bentgrass Transgenic-HR creeping bentgrass with resistance to GLY was developed by Scotts Company and Monsanto (event ASR368). Field trials were established in Jefferson County, Oregon in 2003 under a permit granted by USDA APHIS [57]. Creeping bentgrass is a wind-pollinated perennial with a high potential for outcrossing, and was the first transgenic crop with synchronously-flowering wild relatives approved for cultivation in the USA [58]. Independent studies demonstrated pollen-mediated transgene flow resulting in wild plant populations expressing the GLY-HR trait [59, 60]. Consequently, further production of GLY-HR creeping bentgrass was stopped, and it was determined that USDA APHIS should have completed an environmental impact statement prior to the deregulation of the transgenic crop [56]. Sugarbeet While there has been regulatory approval for transgenic- HR sugarbeet since 1998, no cultivars have been commercially introduced [23]. It is suggested that concerns expressed by the processors about the inclusion of transgenic sugar kept transgenic-hr sugarbeet from the marketplace [24]. However, there are plans to introduce GLY-HR sugarbeet cultivars in the relatively near future. Sugarbeet is a crop that could benefit from the transgenic-hr traits. Sugarbeet receives almost 12 herbicide treatments per growing season, and growers spend an average of US$180 per ha on herbicides [24]. This expenditure represents about US$115 million on herbicides annually. Furthermore, growers typically will spend an additional US$97 million on non-herbicidal control. There are several issues for sugarbeet weed management, notably the relative poor ability of the crop to compete with weeds and sensitivity to herbicide phytotoxicity. It is suggested that the adoption of transgenic-hr sugarbeet cultivars would resolve or at least lessen the importance of these issues, and result in estimated savings in production costs of US$93 million [24]. Given the expected intensity of herbicide selection pressure that would result from the adoption of transgenic-hr sugarbeet cultivars, it is also likely that weed population shifts and evolved resistance in weed populations to these herbicides would rapidly ensue [35]. Canada Canola Transgenic-HR canola was introduced commercially in Canada in Of the approximately 5.4 million ha of canola grown in 2006 [61], about 95% (5.2 million ha) is estimated to be resistant to GLY, GLU, or IMIs (Figure 2 [62]). Over 80% of HR canola is transgenic, much higher than the global percentage (ca. 20%) [4]. Canola is primarily a western crop (Prairie provinces of Alberta, Saskatchewan and Manitoba), with cultivation in eastern Canada (Ontario, Quebec, and the Maritime provinces) accounting for only 0.6% of the nation s canola area. In 1997, the number of non-hr canola cultivars

7 Canola area (%) Year Total HR GLY-HR GLU-HR IMI-HR BX-HR Figure 2 Adoption of HR canola in Canada. Adapted from Beckie et al. [62] commercially available peaked at 46, compared with only seven HR cultivars [63]. By 2006, there was only one non-hr cultivar and 48 private-sector HR cultivars (35 GLY-HR, 10 IMI-HR and 3 GLU-HR). Not all HR canola traits have been a commercial success. Before the transgenic era, the first HR crop available to growers in Canada was triazine (TR)-HR canola in 1981 [64]. However, the TR-HR trait imposed a fitness penalty that caused lower productivity. Yields of TR-HR canola cultivars were 20 30% less than conventional canola cultivars because of reduced photosynthetic efficiency [65 67]. Additionally, atrazine did not control a number of weeds that are important economic problems in canola-growing areas. The production of TR-HR cultivars was only cost-effective in areas where competition was high from cruciferous weeds, such as wild mustard (Sinapis arvensis L.) and stinkweed/field pennycress (Thlaspi arvense L.). Because the seeds of these weeds are high in erucic acid and glucosinolates, a 5% threshold was established for their presence in canola seedlots [68]. Ethametsulfuron, introduced in 1990, controlled wild mustard and other weed species in canola, thereby reducing the utility of TR-HR cultivars. Cultivation of TR-HR canola peaked at 4% of the total canola area in 1988, and decreased to less than 1% by 1996 [64]. Breeding efforts and cultivation of TR-HR cultivars have been discontinued in Canada. The BX-HR canola trait in Canada had a similar fate as TR-HR canola, occupying less than 1% of the crop area in 2000 and 2001 (Figure 2). Although BX-HR canola did not exhibit any fitness penalty [69], the herbicides registered for use in the crop could not compete with GLY, GLU, or IMIs. BX controlled a limited number of broadleaf weeds and needed to be tank-mixed with clethodim, an acetyl- CoA carboxylase (ACCase, EC ) inhibitor, for grass weed control. Moreover, by 2000, there was widespread resistance to ACCase inhibitor herbicides in wild oat Hugh J. Beckie and Micheal D. K. Owen 7 (Avena fatua L.) and green foxtail [Setaria viridis (L.) Beauv.]. Soybean and maize Similar to canola, the adoption of HR soybean has been rapid. All registered HR cultivars are resistant to GLY (Table 2). soybean is grown primarily in eastern Canada, with over 70% of the nation s soybean area in Ontario [61]. GLY-HR soybean, first grown in Ontario in 1997, constituted about two-thirds ( ha) of the total soybean area by 2006 (Figure 3). In contrast to soybean, adoption of HR maize has been less rapid. In Canada, 59 and 35% of maize is grown in Ontario and Québec, respectively [61]. HR maize cultivars were first grown in 1998 [70]. In 2006, GLU-HR and GLY-HR maize comprised 40% of the total crop area in eastern Canada (M. Gans, personal communication). The total area of transgenic-hr maize in Canada in 2006 was about ha. Less rapid adoption of HR maize than HR soybean reflects the performance and cost of herbicide treatments in non-hr maize [71]. Wheat In 2004, CDC Imagine was the first IMI-HR spring wheat cultivar grown commercially. The cultivar was not listed in the 2004 Canadian Wheat Board variety survey [72], indicating that the planted area was exceedingly small as a possible consequence of limited seed availability. The 2006 Canadian Wheat Board variety survey [73] listed CDC Imagine at 3.3% of the area planted in the prairies to Canada Western Red Spring wheat cultivars, ranging from 1.4% in Manitoba to 4.4% in Saskatchewan. As outlined above, the rapid rate of adoption of some HR crops, such as soybean, cotton and canola, suggests a net economic benefit to growers. Adoption of HR crops is driven primarily by easier and improved weed control or higher net returns [74, 75]. Convenience in herbicide application to manage increasing farm size and concomitant time pressures is an important driver of HR crop adoption. Herbicides used in HR crops can generally be applied over a wide range of crop or weed growth stages. HR crops have facilitated the adoption of conservationtillage systems (and vice versa) by use of POST-applied herbicides (e.g. GLY, GLU, IMIs) versus pre-emergence (PRE) soil-incorporated herbicides which are commonly used in non-hr crops. Intraspecific Gene Flow and its Consequences Gene flow has occurred since plants were first domesticated as crops, but the HR traits in crops whether transgenic or non-transgenic provide simple, convenient, and precise markers to measure frequency and distance of gene flow. While the agronomic, economic and environmental consequences of gene flow are difficult to quantify, the movement or escape of HR genes has been

8 8 Perspectives in Agriculture, Veterinary Science, Nutrition and Natural Resources Soybean area (%) Year Figure 3 GLY-resistant soybean as a percentage of the total crop area in Ontario, Canada. Adapted from Beckie et al. [62] widely and sensationally reported in the popular press. These reports impact the general public s perception of biotechnology safety. Gene flow can occur via pollen if the crop species is a partial or obligate outcrosser. If HR cultivars are grown in close proximity to non-hr cultivars, HR genes may be transferred to non-hr plants and hybrid seed formed. HR crops with outcrossing potential include maize, canola, alfalfa, creeping bentgrass, rice, sunflower, and to a lesser extent, wheat (Table 2). Genes may also move via seed. Seed can remain viable in the seed bank for many years, germinate, and grow in subsequent crops as a volunteer or become mixed (termed an admixture) with other cultivars or crops in planting or harvesting equipment, and bulk storage facilities [76]. Therefore, gene flow via seed has the potential to influence agriculture temporally and generally over a larger scale than gene flow via pollen. The mechanism of gene movement is often difficult to determine because a combination of pollen and seed movement may have occurred. Intraspecific Gene Flow and Adventitious Presence (AP) There are a number of factors that influence pollenmediated gene flow. Distance from the pollen source is a critically important factor [77]. However, pollen shed density, time of pollen shed, and environmental conditions must also be considered as important factors influencing the flow of pollen-mediated genes [78]. Canola In canola, where outcrossing averages 30% and where more than one HR trait has been developed, pollenmediated gene flow can result in multiple-hr (i.e. genestacked) volunteers. Multiple-HR canola volunteers were first documented in the late 1990s [79]. In 1997 in western Canada, a field of GLY-HR canola was grown adjacent to a field of GLU-HR and IMI-HR canola. Volunteers were selected with GLY in These volunteers flowered and produced seeds that contained individuals resistant to GLY and GLU; GLY and IMI; and GLY, IMI and GLU. Two triple-hr individuals were detected, with one plant located 550 m from the GLY-HR pollen source. A subsequent study was conducted at 11 commercial field sites in Saskatchewan, Canada in 1999, where GLY-HR canola was grown adjacent to GLU-HR canola [80]. Gene flow ranged from 1.4% outcrossing at the border common to the paired fields to 0.04% at 400 m. Thus, outcrossing distance was greater than the 100 m isolation distance currently regulated for Canadian seed growers [81], and greater than the 175 m buffer zone recommended to commercial canola growers [82]. Consequently, harvested seeds of canola grown in proximity to that with a different HR trait may contain individuals with multiple herbicide resistance. The majority (80%) of growers purchase certified canola seed instead of planting seed harvested from the previous crop [83, 84]. Data from two Canadian studies have documented AP of transgenic events in certified seed. AP in non-hr canola seedlots may include individual seeds that contain HR genes, singly or stacked; in the case of HR canola cultivars, non-hr seed or another HR type would be considered as AP. AP may result as a consequence of cross-pollination, volunteer plants, seeding errors, and co-mingling during harvest, transport, storage or processing [85]. Downey and Beckie [86] found that 34 of 70 certified canola seedlots tested from 14 non-hr, open-pollinated cultivars produced in 2000 contained the gene conferring GLY resistance, and 41 seedlots (59%) contained the GLY- or GLU-HR gene. Four of 14 cultivars tested exceeded an HR frequency of 0.25%. In Canada, a maximum of 0.25% AP is permitted in certified canola seed [81]. Only two of the 14 cultivars were free of both genes. In a subsequent examination of 27 seedlots (excluding GLY-HR) by Friesen et al. [87], 78% of seedlots

9 were found to have the GLY-HR gene, 96% had either the GLY- or GLU-HR gene and 52% exceeded a frequency of 0.25% AP. Before the introduction of HR traits, there were no definitive genetic markers that allowed the precise quantification of levels of genetic purity in canola cultivar seedlots. The source of most of the AP was likely breeding nurseries during the development of the cultivars and not during the pedigreed seed multiplication process; for example, AP was found in a single variety that originated from many different seed growers. Breeders and seed companies are now monitoring seedlots for AP of the HR trait using commercially-available test strips or seed assays. Although the purity threshold is seemingly high (99.75%), the small size of canola seed and the commercial seeding rates can result in a significant number of AP seeds planted per hectare. For example, assuming that 0.25% of the seeds in a seedlot are double-hr, this would result in a potential seed bank of about 5 double-hr plants/m 2 or plants/ha. Together, AP in pedigreed canola seedlots planted and pollen-mediated gene flow can result in large, unexpected populations of single- or multiple-hr canola, and canola volunteers in subsequent years. Beckie et al. [80] found that in the year following HR canola (2000) when volunteers were mapped and characterized, gene flow as a result of pollen flow in 1999 was detected at 800 m, the limit of the study areas (Figure 4). Large variation in gene flow levels and patterns among sites was evident. The AP of double (GLY+GLU)-HR seed in GLY-HR seedlots planted at some of the sites in 1999 also contributed to the occurrence of double-hr canola volunteers in The results of the study suggest that HR gene stacking in canola volunteers in western Canada is common, and reflects pollen flow between different HR-trait canola, AP in seedlots, and/or agronomic practices employed by Canadian growers. Soybean Soybean is an autogamous species, with outcrossing ranging from 0.2 to 5% [88]. Importantly, the frequency of cross-pollination markedly declines with distance, which suggests there is a limited opportunity for intraspecific introgression of genetic traits [89]. Given that wild- and weedy-type soybean are not native to the USA or Canada, the concern for intraspecific introgression of HR traits is minimal. Cotton Cotton does not have related species that are indigenous to the USA or Canada, and thus interspecific outcrossing is not an issue [90]. Furthermore, there are no reports of outcrossing of transgenic-hr or Bacillus thuringiensis (Bt) toxin traits to non-transgenic cotton cultivars. Maize Whereas B. napus canola is a moderately outcrossing species, maize is highly outcrossing (Table 2). Hugh J. Beckie and Micheal D. K. Owen 9 GLU-resistant canola in 1999 GLY-resistant canola in Double-resistant (single plant) Double-resistant (patch) Figure 4 The occurrence of double HR canola volunteers at a site in Saskatchewan, Canada in 2000 as a result of pollen flow the previous year. Reproduced from Beckie et al. [80] by permission of the Ecological Society of America Pollen-mediated transgene flow in maize can be highly variable, as documented in studies conducted in the USA. In New York State, most pollen-mediated gene flow from GLY-HR maize to non-hr maize occurred within 5 m of the pollen source (58 81 m 2 area) [91]. Almost no pollen flow was found beyond 25 m. However, in another study, cross-pollination was measured at 200 m from the pollen source [92]. A potential problem is the contamination of seed fields, resulting in AP in seedlots. It is estimated that with a favourable wind direction and a distance of 30 m between GLY-HR and non-hr maize plots, 2% of non-hr maize may contain the GLY-HR transgene [93]. Kumar et al. [91] found that the AP of GLY-HR traits exceeded the EU threshold (0.9%) at 15 m from the pollen source in one year and at 5 m the next year. Models that predict the potential for pollen movement have been developed, but there are limitations on the accuracy of their predictions [94, 95]. The simplicity of N

10 10 Perspectives in Agriculture, Veterinary Science, Nutrition and Natural Resources current models limits the ability to predict the amount of pollen that can move further than 250 m, and to assess the impact of climatic conditions on pollen movement. Only % of viable maize pollen was estimated to move 500 m from the source plant [96]. However, the occurrence of pollen does not mean that transgenic trait movement occurs over great distances. Furthermore, the occurrence of transgenic pollen in a non-transgenic maize field does not mean pollination was successful [96]. Increasing isolation distance between transgenic maize and non-transgenic maize is a useful tool in containing gene flow [92]. Some research suggests that the isolation distance required to approach 0% probability of pollenmediated gene flow is greater than 500 m, and likely impractical for commercial maize production [97]. No Canadian data on pollen-mediated gene flow for HR maize is available. Pollen-mediated transgene (Bt) movement to non-transgenic maize was investigated in field experiments in Ontario from 2000 to 2002 [98]. Maximum outcrossing of 82% occurred in plants immediately adjacent to Bt maize, and declined exponentially to less than 1% at 28 m from the moderately-sized (27¾27 m) pollen source. The authors concluded that the current isolation distance between genotypes of 200 m, which is generally recommended for seed growers, is appropriate for Bt or HR maize. However, Ireland et al. [99] suggested that while current isolation practices could provide 99% genetic purity, considerable outcrossing could and does happen. Given the amount of pollen typically shed, even an extremely low percentage of successful pollination would result in an economically-concerning HR volunteer maize problem. If the affected maize crop was supposed to be transgene-free (e.g. organic production or specialty maize for food), the low level of outcrossing could cause the crop to be in violation of the purity contract. Given some of the logistical limitations for maize production, unintended herbicide resistance in volunteer maize is likely to become an increasing problem as the adoption of transgenic-hr hybrids increases. Rice Rice is primarily self-pollinated; outcrossing rate (HR to non-hr rice) averages 0.2%, ranging from 0 to 0.5% (as reviewed in [100]). However, the primary concern is outcrossing from HR rice to weedy red rice. Intraspecific outcrossing in rice is a major economic problem. Spontaneous crosses between cultivated rice and red rice consistently occur in and near rice fields where both plants co-exist [101]. Gene flow between IMI-HR rice cultivars and weedy red rice resulting in IMI-HR weedy red rice is one of the primary concerns of cultivating IMI-HR rice [102]. Furthermore, a regulated transgenic (GLU)-HR trait has been identified as an unwanted contaminant in commercial rice, thus resulting in significant ecological and economic problems for US rice production. Concerns also exist for other transgenic traits (i.e. Bt/CpTi) that might introgress into non-transgenic rice cultivars. While some variation was observed, frequency of transgene flow was generally quite low, ranging from 0.28% when the transgenic and non-transgenic cultivars were separated by 0.2 m to < 0.01% with a 6.2 m separation distance [103, 104]. However, given that weedy red rice co-exists within the cultivated rice fields, the presumed protection from gene flow attributable to separation distance is not a factor and the introgression of traits in the cultivated rice, whether transgenic or non-transgenic, is highly likely. While published research suggests that pollen-mediated gene flow in rice occurs at a relatively low frequency, once hybrids develop, the selection pressure from herbicide use will cause considerable change in the weedy red rice population with an introgressed HR trait within a few generations [105]. Importantly, in commercial rice fields, weedy red rice may have already hybridized with cultivated rice. Given that weedy traits tend to be dominant and that the inheritance of the IMI-HR trait is semi-dominant and does not impart any measurable fitness penalty on progeny, management of intraspecific gene flow is extremely important [38, 106]. Wheat The frequency and distance of pollen-mediated gene flow in non-hr spring wheat have been documented from a small (5¾5 m) and a medium-sized (50¾50 m) pollen source [107, 108]. Gene flow varied with cultivar, but remained below 0.5% and decreased to 0.005% at 300 m from the pollen source. Current studies are investigating pollen-mediated gene flow from IMI-HR spring wheat to non-hr spring wheat in commercial fields (Beckie et al., unpublished data). Because IMI-HR wheat is nontransgenic, AP in grain shipments is not a concern to buyers. Pollen-mediated gene flow from commercial fields of IMI-HR winter wheat to non-hr winter wheat in eastern Colorado was investigated in 2003 and 2004 [109]. Two non-hr wheat cultivars were found to have significantly higher cross-pollination rates ( %) within 5 m of IMI-HR wheat than nine other cultivars. As expected, cross-pollination declined rapidly with increasing distance from the pollen source with success rates of % at 37 m. Sunflower Sunflower is primarily self-pollinating, but outcrossing does occur via insect vectors [110]. US seed growers separate fields by km to limit outcrossing. Seed from non-imi-hr sunflower fertilized by IMI-HR pollen would be partially resistant to IMI herbicides. However, hybrid seed is rarely replanted by growers, which reduces the potential for IMI-HR gene flow via seed. The consequence of gene flow between commercial and weedy common sunflower is more important [47]. An estimated two-thirds of commercial sunflower fields occur in close proximity to weedy sunflower, and

11 phenological overlap was determined to be 52 96%. The resulting gene flow from cultivated sunflower to weedy sunflower can have significant effects on the resulting hybrid. For example, a Bt transgene that was introduced into a weedy sunflower population increased seed production and reduced insect herbivory in the hybrid progeny [111]. This effect logically increases the populations of weedy sunflower that have the introgressed transgenic trait. Alfalfa As described previously, alfalfa with the transgene conferring resistance to GLY has been the subject of a lawsuit where growers of non-transgenic alfalfa successfully brought suit against USDA APHIS, resulting in a ruling that seed sales and planting of the transgenic-hr alfalfa was deemed illegal after 30 March What is important to note is that no introgression of the transgene into non-transgenic or wild alfalfa had been identified at the time of the ruling. However, the potential for gene flow between cultivated and wild alfalfa has been identified [112]. Scientists suggest that while gene flow into the environment is a potential problem with transgenic-hr alfalfa, there are no sexually-compatible wild relatives of alfalfa except feral alfalfa [113]. Regardless, it was determined that USDA APHIS had not conducted an environment impact statement. USDA APHIS suggested that the potential for outcrossing was negligible because of temporal differences between the harvest of commercial alfalfa and alfalfa for seed [56]. Furthermore, current isolation standards for production are deemed sufficient to deter gene flow [114]. However, given that alfalfa is insect-pollinated, the isolation distance used for the production of certified alfalfa seed may not be sufficient to limit transgene flow. Studies have indicated that gene contamination declined significantly with increasing distance from the pollen source; at 4 km, gene flow was < 0.03% [115, 116]. The potential distance that gene flow occurred also depended on the specific insect pollinator. Regardless, transgenic-hr alfalfa is currently restricted because of concerns about intraspecific transgene flow. Creeping bentgrass Similar to GLY-HR alfalfa and as indicated previously, creeping bentgrass with the GLY-HR trait was also subject to restrictions brought forward by US federal court judges on 5 February 2007 [56]. The court argued that gene flow threatened the Crooked River National Grassland located in Oregon. It was countered that transgenic-hr creeping bentgrass did not pose a significant ecological threat. Nevertheless, the transgene was found in wild species well outside of the study area [60]. The presented data supported the conclusion that intraspecific transgene flow was likely attributable to wind-disseminated, pollen-mediated hybridization and crop seed dispersal. Other reports suggest that gene flow can occur as far as 21 km in sentinel creeping bentgrass plants, and 14 km in wild bentgrass plants [58]. Gene migration attributable to seed dispersal is believed to be more difficult to contain, and will occur over greater distances when compared with gene flow from pollen movement [117]. The implications of the transgenic-hr gene flow to wild bentgrass resulting in intraspecific transgenic-hr hybrids are currently unknown. However, questions about the unintended ecological consequences of transgene flow in the environment must be addressed. To date in Canada, there have been no changes in commodity market acceptance of field crops due to the presence of HR genes because non-hr and HR crops are co-mingled for domestic sale or export. The economic consequences of gene flow in canola or AP in seedlots in Canada have largely been restricted to the organic canola and honey market [118]. To date, a group of organic growers in Saskatchewan, Canada have not been successful in their legal action against the developers of GLY- and GLU-HR canola [62]. In the USA, recent court rulings against cultivation of transgenic-hr alfalfa and creeping bentgrass, and cases of a regulated HR trait discovered in commercial fields with deregulated HR crops grown for food or feed have disrupted, at least temporarily, the production and markets of these crops. Consequences of single- or stacked-hr volunteers for both adopters and non-adopters of HR crops, except alfalfa and creeping bentgrass, are described in the next section. Managing Single or Multiple HR Crop Volunteers Canola Hugh J. Beckie and Micheal D. K. Owen 11 Volunteer canola (mainly HR) was ranked 12th in relative abundance among residual (post-management) weed species when averaged across Canadian prairie field surveys conducted from 2001 to 2003; volunteer density averaged 4.3 plants/m 2 in fields where they occurred [119]. However, mean relative abundance ranking had declined from 10th position as determined from surveys conducted in the mid-1990s when canola was mainly non-hr (Figure 5) [119]. In contrast, volunteer non-hr wheat increased in rank from 18th to 8th place from the 1990s to 2000s, suggesting that the HR trait is not a major factor influencing volunteer canola abundance. Based on estimated crop yield loss, the economic impact of volunteer wheat and canola is ranked third and eighth, respectively, among weedy species [120]. Canola can produce large volunteer populations because of the large amount of seed lost before and at harvest. In a study in Saskatchewan, Canada in 1999 and 2000, average canola seed loss during harvest operations was 5.9% of crop yield (3000 viable seeds/m 2 ) as determined from measurements in 35 growers fields [121]; yield loss among growers ranged from 3.3 to 9.9% or 9 to 56 times the recommended seeding rate of canola. Canola