Transgenic Crops

Since 1996, glyphosate-resistant (GR) crops have had a major effect on agriculture, particularly in the United States, Brazil, Argentina, and Canada (Brookes and Barfoot 2011). The introduction of GR crops in the United States helped solve a major weed-management problem that was developing at that time—the evolution of weeds resistant to the acetolactate synthase (ALS)–inhibiting and protoporphyrinogen oxidase (PPO)–inhibiting herbicides. However, the introduction of GR crops also prompted concerns about potential transfer of herbicide resistance to weed populations via crop-to-weed gene flow (Dale 1994; Warwick et al. 1999). This has been experimentally demonstrated for a small number of weed species in North America that can hybridize with a closely related crop: examples are described later in this document (see BMP 1). However, for full introgression of a herbicide-resistance gene from a crop into a weed population, the initial weed by crop hybrid must be fertile and repeated transmission of the HR gene via backcrossing to the weedy parent must take place. There are few reports of this occurring in the field. Warwick et al. (2008) described stable introgression of a HR transgene from GR canola (Brassica napus L.) to weedy birdsrape mustard (Brassica rapa L.) under field conditions, and Zapiola et al. (2008) reported persistence of a GR transgene in populations of weedy creeping bentgrass (Agrostis stolonifera L.) following accidental pollen and propagule transfer from experimental plantings of GR ornamental bentgrass (Agrostis palustris Huds.). However, none of the confirmed field reports of glyphosate resistance in North American weed species listed to date by Heap (2011b) have been traced to gene transfer from GR crops. Although there is a possibility of weeds acquiring HR genes from crops in a few cases, the accumulating evidence to date strongly suggests that a much greater risk is posed by spontaneous evolution of herbicide resistance in weed populations under selection pressure from repeated herbicide applications in HR cropping systems. The wide-scale adoption of any single herbicidal MOA contributes to the evolution of resistance to that MOA, and the unprecedented scale of glyphosate use in GR crops has clearly contributed to the number of GR weeds identified in recent years (Powles et al. 1997; Vencill et al. 2011).

In parts of North and South America, transgenic GR crops have dramatically reduced the diversity of herbicides used for weed management in those crops (Young 2006). Glyphosate-resistant crop cultivars now dominate corn (Zea mays L.), soybean [Glycine max (L.) Merr.], cotton (Gossypium hirsutum L.), and canola production in the United States (Nichols et al. 2003). As new HR crops become available, management of novel HR weeds will continue to be a challenge. Glufosinate-resistant soybean, corn, cotton, and canola are now commercialized, and, in the near future, crops resistant to the herbicides 2,4-D, dicamba, hydroxyphenylpyruvate dioxygenase (HPPD) inhibitors, and possibly to the PPO-inhibiting herbicides are expected to reach the marketplace (Green et al. 2008). Further, transgenic crops with resistance to more than one herbicide MOA (i.e., stacked traits) have also been commercialized in recent times. Strategies for managing resistant weed populations in the context of these new technologies must be developed and adopted (Duke and Powles 2008; Owen 2008; Owen and Zelaya 2005).

Although glyphosate resistance in weeds currently has a large effect on U.S. agriculture, the discussion and recommendations in this document pertain to all current and future herbicide technologies. All herbicide technologies, if used repeatedly, are at risk of losing efficacy because of herbicide resistance in weed populations. Two recent examples are the identification of goosegrass [Eleusine indica (L.) Gaertn.] populations with resistance to glufosinate (Jalaludin et al. 2010; Seng et. al. 2010) and tall waterhemp [Amaranthus tuberculatus (Moq.) Sauer] populations with resistance to the HPPD-inhibiting herbicides (Hausman et al. 2011; McMullan and Green 2011); resistance had not previously been reported to either of these two herbicide MOAs.

Consequence of Overreliance on a Single Mechanism of Action

Although a number of factors determine the frequency of resistance events in weed populations, consideration of reported incidents strongly suggests that the single most important factor leading to the evolution of herbicide resistance is overreliance on a single herbicide (or group of herbicides with the same MOA) without using other weed management options (Heap 2011b). For example, GR rigid ryegrass (Lolium rigidum Gaudin) and Italian ryegrass [Lolium perenne L. ssp. multiflorum (Lam.) Husnot] populations were identified in orchards where glyphosate had been used continually for at least 14 consecutive yr (Perez-Jones et al. 2005; Powles et al. 1997; Simarmata et al. 2005). Glyphosate-resistant horseweed [Conyza canadensis (L.) Cronq.] and common ragweed (Ambrosia artemisiifolia L.) were confirmed after continuous applications of glyphosate on GR soybean for 3 and 6 yr, respectively (Pollard et al. 2004; VanGessel 2001). Likewise, glyphosate resistance in Palmer amaranth (Amaranthus palmeri S. Wats.) and common waterhemp (Amaranthus rudis Sauer) was discovered after only 4 and 6 yr of consecutive glyphosate use (Culpeppper et al. 2006; Legleiter and Bradley 2008). This phenomenon is not unique to glyphosate. Prickly lettuce (Lactuca serriola L.), for example, evolved resistance after 5 yr of consecutive application of chlorsulfuron (Mallory-Smith et al. 1990). Other evidence includes, but is not limited to, 2,4-D resistance in wild carrot (Switzer 1957; Whitehead and Switzer 1967), common groundsel resistance to atrazine (Ryan 1970), and barnyardgrass resistance to propanil (Carey et al. 1995).

Costs of Resistance

Because herbicides are the primary means of weed management in many economically developed countries, herbicide resistance causes greater short-term cost to manage a weed population, especially with the absence of new herbicide chemistries. Implicit in the measurement of the costs of HR weeds are crop yield loss, reduced commodity prices because of weed-seed contamination, reduced land values, costs of mechanical and cultural controls, and ultimately the additional expense of alternative herbicides or cropping systems or both for managing the resistant weed. Several recent studies have described the added costs associated with the management of HR weeds. In the United States, the occurrence of GR horseweed resulted in a net increase in production cost of $28.42 ha⁻¹ in soybean (Mueller et al. 2005). The additional cost of managing GR Palmer amaranth in Georgia and Arkansas cotton production systems was estimated at $48 ha⁻¹ (Vencill et al. 2011). In Missouri, Legleiter et al. (2009) reported similar net cost increases in controlling GR common waterhemp in soybean. Similarly in Arkansas rice, the additional cost involved in controlling propanil- and quinclorac-resistant barnyardgrass was estimated at $64 ha⁻¹ (Norsworthy et al. 2007a).

Limitations to Implementation of Practices to PreventResistance

Although there is a significant cost to managing resistant weeds, growers are often hesitant to implement proactive measures to reduce the risk of resistance evolution in their fields. A key factor negatively influencing producer adoption of practices that will mitigate herbicide-resistance evolution is the expectation that there will be new herbicides available in the future (Foresman and Glasgow 2008; Llewellyn et al. 2002, 2007). With invasive plant populations, Finnoff et al. (2007) demonstrated quantitatively how managers who are cautious or financially risk averse are less likely to adopt preventive measures because prevention only reduces the risk, rather than eliminating it. This perception is perhaps true with growers weighing the value of prevention vs. control of HR weeds. For example, in a survey of more than 1,000 corn, cotton, and soybean growers in the United States, Frisvold et al. (2009) found that using multiple herbicides with different MOAs was one of the least-adopted practices for resistance management, despite this strategy being frequently advocated by weed scientists as an effective means to reduce the risk of herbicide-resistance evolution. A number of related factors may account for this. The overarching reason is that using diverse MOAs can increase current weed-control costs (Hurley et al. 2009), whereas the benefits of delaying resistance accrue in the future and are more uncertain. Some growers believe that mitigating weed resistance is beyond their control, depending more on their neighbor's behavior or natural factors (Llewellyn 2006; Wilson et al. 2008). Growers may also believe that industry will develop new chemistries, reducing the benefits of resistance management (Llewellyn et al. 2002, 2007). Finally, even when using diverse MOAs provides short-term returns comparable to current weed-management programs, growers will be less certain about the new, diversified programs.

Herbicide resistance-management strategies are typically implemented reactively when a resistant weed population has grown to problematic levels and must be controlled. Far less often are proactive management strategies implemented to mitigate the evolution of herbicide resistance. Introduction of new HR crop cultivars can provide options for managing weed populations resistant to other herbicide MOAs, but good resistance-management practices must be employed to avoid evolving resistance to the new herbicide as well.

Rationale for Best Management Practices (BMPs) and Recommendations

Herbicides constitute the key means of weed management and are the primary means for reducing soil tillage practices. Just as conservation of soil, water, and genetic diversity are fundamental for agricultural production, conservation of herbicide resources underlies the sustainability of contemporary agriculture. As such, herbicide molecules and MOAs (and other pesticides) are scarce resources upon which the public largely relies for food, feed, fiber, and energy security. Management practices that mitigate the evolution of resistance have been identified and will be invaluable in preserving these vital resources. Resistance management begins with good agronomic practices, including the implementation of integrated weed management (IWM) strategies that use diverse control tactics to reduce the frequency of herbicide applications and lessen the selection pressure they exert on weed populations. The BMPs we propose are discussed below in the context of the currently available literature documenting their effectiveness.

BMP 1: Understand the Biology of the Weeds Present

Integrating an understanding of weed biology into current weed-management systems, with emphasis on reducing the soil seedbank, is paramount for developing sustainable weed-management programs (Bhowmik 1997; Walsh and Powles 2007). By understanding weed emergence patterns, length of developmental stages, fecundity, dispersal mechanisms, and persistence of weed seed in the soil seedbank, practitioners can devise a strategy that targets the life stages most sensitive to management. Knowledge of weed biology in agricultural systems allows development of weed management programs that leverage weaknesses in the life cycle of the weed to alter the competitive relationship with the crop (Neve et al. 2003a,b). Agroecosystems typically contain diverse weed species, each with different life histories and different responses to control measures. Effective herbicide-resistance management combines a variety of chemical and nonchemical management tactics to diversify selection pressure on weed populations and minimize spread of resistance genes. Development of such strategies requires an understanding of the biology and ecology of weeds and the agroecosystems in which they occur (Mortensen et al. 2000).

In any growing season, the soil weed seedbank governs the size and species composition of the weed community (Forcella et al. 1993). In Iowa, efficacy of weed control in corn is inversely correlated with population size the previous year (Hartzler and Roth 1993). Hence, reducing current and future weed populations is not only a general goal of weed management, but also a key component in herbicide-resistance management. The greater the number of plants exposed to a herbicide, the higher the probability of selection for resistance, enriching the gene pool with an increased frequency of previously rare resistance alleles (Gressel and Levy 2006).

Long-term herbicide-resistance management, therefore, requires more than weed control aimed only at minimizing crop loss in any one season. It is also essential to reduce the number of individual weeds exposed to herbicide selection, to prevent propagation by HR weeds, and to reduce the seedbank in future years. Knowledge of the following biological characteristics of weed species allows targeted management of life stages that affect population and seedbank size, as well as more-accurate assessment of the risk of resistance evolution.

BMP 2: Use a Diversified Approach to Weed Management Focused on Reducing Weed Seed Production and the Number of Weed Seeds in the Soil Seedbank

Frequently HR weeds evolve in crop production systems where weed management strategies have not been diverse (Nichols et al. 2009; Talbert and Burgos 2007; Walsh and Powles 2007). One of the most recent weeds to affect a vast area of crops across the southern United States is GR Palmer amaranth (Nichols et al. 2009). Palmer amaranth is a genetically diverse, obligate, outcrossing weed, and the adoption of no-till or reduced-tillage production practices; minimal crop rotation, especially in cotton; and almost sole reliance on a single herbicide (i.e., glyphosate) has led to the rapid evolution of glyphosate resistance. Recent simulation modeling has shown that integration of diversified management tactics, with particular focus on reducing the soil seedbank, greatly reduces the risk of new GR Palmer amaranth populations evolving (Neve et al. 2011a,b).

For a production system to remain sustainable, the soil weed seedbank must be static or declining. An increasing seedbank is evidence of a weed that is escaping the current management regime through herbicide resistance or some other adaptation. Furthermore, the risk of resistance evolution is shown to be positively associated with the initial seedbank size (Neve et al. 2011a); therefore, keeping the soil seedbank at low levels reduces the risk of future evolution of herbicide resistance. Soil seedbank size varies considerably among fields, often because of the relative effectiveness of weed management practices. For example, overall weed seedbank densities across the Corn Belt varied from 600 to 162,000 viable seed m⁻² (Forcella et al. 1992). In Arkansas, barnyardgrass seedbank size in cotton, soybean, and rice fields varied from 0 to 121,000 viable seed m⁻², indicating the highly dynamic nature of soil seedbanks (Bagavathiannan et al. 2011b). Farming practices directly influence weed populations by affecting quantity of seed removed from and returned to the soil seedbank (Wilson 1988). The soil seedbank could be effectively reduced through preventing weed seed production and reducing the germinable fraction of weed seed in the soil through specific tillage techniques, such as stale seedbed, along with practices that encourage seed loss via herbivory.

The lasting consequences of allowing weed seed production have long been acknowledged, as stated in an old English proverb: “One year's seeding makes seven years' weeding.” Nevertheless, the idea that a limited amount of weed seed production each year is acceptable is strongly embedded in weed management strategies based on annual economic thresholds. Problems with this approach, especially the long-term effects of subthreshold weed populations, have been repeatedly demonstrated, independent of the increased risk of herbicide-resistance evolution (Cardina and Norquay 1997; Hartzler 1996; Jones and Medd 2000; Norris 1999). For example, Cardina and Norquay (1997) observed a linear increase in the soil seedbank when the annual subthreshold densities of velvetleaf plants were allowed to escape control in corn. The risks of herbicide-resistance evolution have been shown to increase with every increase in initial seedbank size (Neve et al. 2011a). Although economic thresholds have typically been calculated only to achieve an effect on crop returns on an annual basis, the long-term effects of subthreshold, noncontrolled weed populations on the soil seedbank and resistance evolution requires management considerations that aim beyond a single growing season (Gallandt 2006).

A few studies have followed an ecological approach to understanding and establishing thresholds. Jones and Medd (2000) suggested a population-based approach to weed management decisions that views weeds as more than an annual production problem. In their “economic optimum threshold” approach, seed production and subsequent additions to the soil seedbank are evaluated along with future management implications and costs. Although economic optimum thresholds are lower than annual economic thresholds (Bauer and Mortensen 1992), this approach still fails to consider the effects of weed population dynamics, such as the initial frequency of resistant alleles in the soil seedbank before herbicide use and changes in the genotypes under selection on herbicide-resistance evolution. As such, economic optimum thresholds ignore future costs associated with herbicide-resistance management. Thresholds that fail to consider the evolution of HR weeds are unsound and have long-term economic and environmental consequences. The concept of what constitutes an acceptable level of weed seed production must be abandoned in favor of a zero or near-zero threshold if the current rate of herbicide-resistance evolution is to be slowed. An increasing number of weed scientists have emphasized the need for a near-zero threshold, especially for highly competitive, prolific seed-producing weeds (Cardina and Norquay 1997; Norris 1999; Sattin et al. 1992; Swanton et al. 1999). However, with the exception of work by a small number of researchers, such as Shrestha (2004), and modeling by Neve et al. (2011a), management of the soil seedbank to delay resistance evolution has not been a major consideration in planning or assessing long-term weed management programs, which is regrettable. Modeling and field evidence indicate the long-term value of preventing weed seed production to minimize weed selection to herbicides and reduce the risk of herbicide-resistance evolution and spread (Griffith et al. 2010; Neve et al. 2011a).

The rapid spread of GR Palmer amaranth in cotton illustrates the fallacy of permitting subthreshold weed populations to produce seed. Although a density of 0.13 Palmer amaranth plants m⁻² will not cause yield loss in cotton (Keeley et al. 1987; Smith et al. 2000), those Palmer amaranth plants can produce in excess of 500,000 seed per female plant, with an annual seedbank addition of more than 16 million seed ha⁻¹ at that subthreshold level of infestation. Within 3 yr of a single GR Palmer amaranth plant escaping control in cotton, the crop field can be overwhelmed by resistance (Griffith et al. 2010). Hence, at its minimum, the goal of all weed management programs must be an aim to eliminate weed seed production from the most competitive, resistance-prone weeds in a field, including any weeds on the global list of resistance-prone weeds as identified by Heap (2011a).

Intensive use of herbicides alone may not be sufficient to prevent weed seed production and to maintain a low soil seedbank (Griffith et al. 2010; Hartzler 1996; Jones and Medd 2000; Schweizer and Zimdahl 1984). Proactively, some U.S. producers are hand-roguing fields as part of a zero seed-production approach to address the increasing occurrence of GR weeds, mainly Palmer amaranth. The expenses these producers have incurred are far less than those who allowed a few plants to escape and are later confronted with sizeable soil seedbanks of resistant weeds that must be rogued multiple times each year to achieve a harvestable crop (Culpepper et al. 2010). Methods to minimize late-season weed-seed production, such as elevated crop seeding rate and crop-topping and removing weed seed from equipment during harvest, are widely practiced in areas of Australia with major HR weed problems (Walsh and Powles 2007). Although preventing weed seed production adds to short-term management costs, it is instrumental in mitigating the evolution of herbicide resistance. Considering the long-term costs associated with future resistance management, these short-term expenses are still economically viable.

BMP 3: Plant into Weed-Free Fields and Keep Fields as Weed Free as Possible

Herbicide-resistance management plans must commence before planting because control options become limited once the crop is planted and weeds begin to emerge. At crop planting time, crop fields should be free of weeds, achieved by a range of control tools. Crops planted into established weeds are risky because even herbicide mixtures with multiple MOAs may not provide effective control. Failure of preplant measures to control emerged weeds is devastating to crop yield and even more devastating if resistance evolves.

To keep crop fields as weed-free as possible, especially in conservation-tillage systems where preplant tillage is not feasible, residual herbicides can be used before or at planting (PRE). Additional applications of residual herbicides in slow-growing or open-canopy crops, before the efficacy of the initial residual herbicide has dissipated, reduces selection pressure associated with sole reliance on POST herbicides (Neve et al. 2003b, 2011b). Crops differ in rates of growth; thus, some crops will require longer periods of residual control than will others for efficacious weed control. Once a dense crop canopy has formed, emergence of most weeds typically ceases because of changes in thermal amplitude and in the quality and quantity of light in the zone of seed germination (Egley 1999; Norsworthy 2004a). Throughout the remainder of the growing season, focus should be on removing any weeds that have escaped control measures or have emerged in safe sites after the final late-season herbicide application, including those that emerge after crop harvest, if seed production is possible before a killing frost (see BMPs 2 and 11).

BMP 4: Plant Weed-Free Crop Seed

Planting weed-free crop seed is the first step to preventing the introduction of new weeds into fields and also an important tactic in preventing the spread of herbicide resistance into new areas (Thill and Mallory-Smith 1997). Separating weed seed from crop seed can be costly and often challenging because of the similarity in size and shape of some weed and crop seed. For example, jointed goatgrass and wild oat (Avena fatua L.) seed are often difficult to separate from wheat seed, making them a common contaminant in wheat seed lots. Through an extensive seed lot survey in Utah, Dewey and Whitesides (1990) found that wild oat seed contaminated about 14% of seed samples collected, with an average of 21 seed kg⁻¹ of wheat seed. Herbicide-resistant wild oat in southeastern Idaho is thought to have immigrated from an adjacent state via contaminated crop seed (Thill and Mallory-Smith 1997). A recent survey of grower-saved crop seed in Australia revealed that about 73% of samples were contaminated with weed seed, with an average of 62 weed seed 10 kg⁻¹ of wheat, barley (Hordeum vulgare L.), or lupine (Lupinus spp.) seed (Michael et al. 2010). Among the weed seed recovered from these samples, rigid ryegrass had resistance to four different herbicides, whereas wild radish (Raphanus raphanistrum L.) and wild oat were each resistant to one herbicide (Michael et al. 2010). Herbicide-resistant rigid ryegrass seed were also found as a contaminant of Australian wheat exported to Japan, demonstrating the potential for international dispersal of resistance genes via weed seed contamination of grain shipments. In these shipments, about 4,500 resistant ryegrass seed were detected in a 20-kg wheat sample in 2006 and 2007, about 35 and 15% of which were resistant to diclofop-methyl, 5 and 6% were resistant to sethoxydim, and 56 and 60% were resistant to chlorsulfuron, respectively (Shimono et al. 2010).

BMP 5: Scout Fields Routinely

Effective scouting (i.e., moving through fields systematically to evaluate weed populations) is a vital component of any successful weed management program, including one aimed at mitigating or managing HR weeds. Timely scouting and maintaining an inventory of weed species, location, density, and size allow for subsequent assessment of the effectiveness of prescribed practices. Numerous field-sampling schemes or strategies can be used when scouting, with the strategy chosen based upon how and when the data will be used (Clay and Johnson 2002; Gold et al. 1996). Regardless of the technique used, records should document the management tools used during previous seasons along with the effectiveness of each tool on some of the most problematic weeds present. These data will be instrumental in determining the efficacy of existing weed management programs and in monitoring whether weed species shifts are occurring. A single herbicide is rarely effective against all weeds present within a field. As a result, effective scouting allows for recognizing the most efficacious herbicide for the weed spectrum present and following it with additional treatment options.

Many chemical weed-management programs aimed at reducing the risk of herbicide resistance begin with a residual herbicide, which should be chosen based on the previous year's weed spectrum from recorded inventory. In particular, weeds present at harvest are a good indicator of species that are not effectively controlled season-long and that may need further attention in the following year's crop (Reddy and Norsworthy 2010). It is imperative that in-season scouting begin soon after crop planting to determine whether preplanting measures, such as tillage or residual herbicide application, have provided the anticipated level of weed control. Further action, such as additional residual herbicide application, can be taken before the first residual herbicide has dissipated. However, timely scouting is vital in making such decisions.

The effectiveness of most POST herbicides is a function of the application rate, application timing, and size of the weed at application; in most field crop settings, a large weed is one that exceeds 10 cm in height. To ensure proper timing of POST herbicides, fields should be scouted routinely for weed sizes and spectrum. Failure to scout and apply POST herbicides in a well-timed manner could reduce the efficacy and increase the risk of herbicide resistance (see BMP 7).

Scouting allows for early identification of a lack of herbicide efficacy, which may indicate the evolution of resistance. In such instances, samples should be collected and appropriate tests conducted to determine whether herbicide failure is a result of resistance. Indicators of possible herbicide resistance include (1) failure to control a weed species normally controlled by the herbicide at the dose applied, especially if control is achieved on adjacent weeds; (2) a spreading patch of noncontrolled plants of a particular weed species; and (3) surviving plants mixed with controlled individuals of the same species. If herbicide resistance is suspected, seed production from all noncontrolled plants of the species concerned should be interrupted to prevent the perpetuation of the resistance problem, even, in extreme cases, if that means destroying the crop in the process. Timely scouting is key to making appropriate management decisions.

BMP 6: Use Multiple, Effective MOAs against the Most Troublesome Weeds and Those Prone to Herbicide Resistance

Herbicides are often the strongest selection agents for weed species present within a cropping system (Booth and Swanton 2002). Use of herbicides across vast areas containing large, genetically diverse weed populations causes intense selection pressure, often resulting in the evolution of resistance in weed communities; repeated application of effective herbicides with the same MOA is the single greatest risk factor for herbicide-resistance evolution (Beckie 2006; Powles et al. 1997). Targeting the most troublesome and resistance-prone weeds within a field by using different herbicide MOAs in annual rotations, tank mixtures, and sequential applications can delay the evolution of resistance by minimizing the selection pressure imposed on those weed populations by a particular herbicide MOA. In an annual herbicide rotation, crops in two or more subsequent cropping seasons receive different herbicide MOAs. Mixtures and sequential applications can be used as a resistance management technique by applying different herbicides to the same crop during the same growing season, simultaneously, in the case of a mixture, and at different times for a herbicide sequence.

The occurrence of resistant individuals in a weed population not previously exposed to the target herbicide is rare (Gressel and Levy 2006; Gressel and Segel 1990). However, the proportion of resistant individuals will rapidly increase with repeated use of the same herbicide MOA. For example, Beckie (2006) showed that significant levels of resistance to ALS-inhibitor herbicides can evolve in weed populations with as few as five applications.

Combining recommended rates of different herbicide MOAs, simultaneously, sequentially, or annually, greatly reduces the likelihoods of survival and reproduction of the resistant individuals. In most weed populations, individuals naturally resistant to more than one herbicide MOA will be rare. In most cases investigated to date, herbicide resistance is conferred by a single dominant or semidominant allele (Jasieniuk et al. 1996; Vila-Aiub et al. 2009). If different resistance alleles are at unlinked loci, the probability of an individual plant being resistant to multiple MOAs can be estimated as the product of the frequencies of resistance to each single MOA (Gressel and Segel 1990; Wrubel and Gressel 1994). This assumes that no single allele confers cross-resistance to multiple herbicides by mechanisms such as general detoxification or sequestration.

Modeling studies predict that strategies involving herbicide rotations or mixtures will be most effective at preventing resistance evolution in species with limited pollen movement, where recombination of resistance alleles transferred via pollen is limited, and with limited seed dispersal (Beckie et al. 2001; Diggle et al. 2003). Single-gene, target-site resistance evolves more slowly in weed populations with continuing establishment of susceptible individuals from persistent seedbanks, keeping the frequency of resistance at low levels (Powles et al. 1997). However, sequential applications in such cases may expose more than one weed cohort to different selection pressures, contributing to the longer-term evolution of genotypes resistant to multiple herbicide MOAs.

BMP 7: Apply the Labeled Herbicide Rate at Recommended Weed Sizes

In the 1990s, reduced herbicide rates were seen as an environmentally sound practice and were a part of many integrated approaches to weed management in crops (Bhowmik 1997; Buhler et al. 1992; Mickelson and Renner 1997; Steckel et al. 1990). This fit well with the concept of weed economic thresholds (i.e., weed density at which cost of control equals the financial return on the recovered crop yield) to optimize crop yield with submaximum weed control (O'Donovan 1996; Zhang et al. 2000). Weed management programs often integrate herbicide application at below-label rates with other control measures, such as tillage and enhancement of crop competitiveness, to keep weed densities below economic threshold levels as defined by crop yield loss models (Blackshaw et al. 2006; O'Donovan et al. 2007). The underlying assumption is that, although herbicide manufacturers' labeled application rates are set sufficiently high to control a range of weed species across different growth stages and site conditions, economically acceptable weed control at a given location can often be achieved with reduced herbicide rates, especially if weed pressure is low and weeds are smaller than sizes listed on the label. However, as suggested by Gressel (1995, 2002) and later established through studies (Busi and Powles 2009; Manalil et al. 2011; Neve and Powles 2005a), recurrent exposure to low herbicide doses that allow a proportion of the weed population to survive may have unintended evolutionary consequences. For example, recent evidence indicates that applying herbicides at below the labeled rate was a major contributing factor to the evolution of diclofop-resistant rigid ryegrass in an Australian wheat field (Manalil et al. 2011).

A weed population under selection from repeated low doses of a herbicide risks accumulation of minor genes, which individually confer only a slight increase in fitness but, when combined in one individual, provide agronomically significant levels of herbicide resistance. Manipulation of this type of polygenic phenotypic response in populations showing continuous trait variation is a long-established foundation of crop breeding, but few investigators have explored multigene herbicide resistance in weeds. A dominant or semidominant allele at a single locus has been found to confer herbicide resistance in most weed species investigated to date. This may only reflect that simple, major gene systems are the easiest to identify experimentally, although rapid adaptation to extreme environmental events (such as exposure to high doses of herbicide) is theoretically most likely to involve single genes with large phenotypic effects (Lande 1983; Macnair 1991). Population response to constant, but less-intense, selection pressure is more likely to involve selection in many minor genes, each with a small, individual effect, which accumulates over multiple generations to produce adapted phenotypes (Orr and Coyne 1992). This type of polygenic evolutionary-resistance response to sublethal drug or pesticide doses is well documented in bacteria (Olofsson and Cars 2007), insects (Roush and MacKenzie 1987), and fungi (Shaw 2006); evidence is emerging for a similar response in weeds, as described below.

Within-population variation for herbicide resistance has been reported in some weeds (Jasieniuk et al. 1996; Patzoldt et al. 2002), including significant survival differences among individuals at low herbicide doses, although they are uniformly susceptible to the labeled rate (Neve and Powles 2005a). Such variation may be attributed to different alleles at minor loci contributing to polygenic resistance or to different allele combinations at modifier loci affecting expression of a major resistance gene (Neve and Powles 2005a; Uyenoyama 1986). Regardless of the molecular mechanism, repeated exposure of these weed populations to reduced herbicide doses risks incremental enrichment of the gene pool with minor resistance alleles, a phenomenon termed creeping resistance by Gressel (1995).

In some weed species, individuals surviving below-label herbicide rates can initiate a rapid population response. Experimental exposure of rigid ryegrass populations to sublethal glyphosate doses resulted in a population shift to resistance in three to four cycles of selection (Busi and Powles 2009). Using low doses of the acetyl coenzyme A carboxylase (ACCase)–inhibitor diclofop-methyl in an initially susceptible rigid ryegrass population and then bulk-crossing survivors, Neve and Powles (2005a) showed that recurrent selection generated field-level resistance in only three generations. Increased levels of cross resistance to other ACCase inhibitors and to ALS-inhibiting herbicides were also reported in the selected populations, suggesting that gene pool enrichment had occurred for multiple alleles controlling nontarget site-resistance mechanisms (Neve and Powles 2005b). Enhanced herbicide detoxification as a widespread cross-resistance mechanism, as proposed by Preston (2004), is one possible outcome of repeated selection for minor resistance alleles.

Knowledge of reproductive biology as well as patterns of herbicide use can help identify weed populations that are more likely to evolve polygenic resistance through recurrent selection by low-dose applications. The risk is theoretically higher in outcrossing species that repeatedly recombine genotypes across family lineages and in populations that already possess a diverse array of minor resistance alleles or the capacity to acquire them from an external source via gene flow (Holsinger 2000; Manalil et al. 2011; Neve 2008). As such, threshold-driven weed management strategies using low-dose herbicide applications increase the risk of polygenic resistance evolution (Busi and Powles 2009; Neve and Powles 2005a,b). However, even when used at the registered rate, any volatile or slowly dissipating herbicide that repeatedly exposes a given weed cohort to a reduced dose could pose a selection risk. For example, slow-degrading, soil-applied herbicides may deliver sublethal doses to later-emerging weeds and allow for the gradual accumulation of resistance alleles, if the same residual herbicide is applied over multiple growing seasons (Maxwell and Mortimer 1994; Zhang et al. 2000). Other situations where weeds may be subjected to below-labeled rates include spraying plants larger than those recommended on the label; inadequate coverage of weeds because of size, density, or crop cover; and low rates caused by imprecise sprayer calibration or inaccurate mixing.

BMP 8: Emphasize Cultural Management Techniques that Suppress Weeds by Using Crop Competitiveness

Cultural practices can be directed to exploit crop competitiveness, thereby reducing weed emergence and growth. Some of these practices include cultivar selection (Jordan 1993; Wax and Pendleton 1968), narrow row spacing (Norsworthy and Frederick 2005; Wax and Pendleton 1968), high seeding rates (Norsworthy and Oliver 2001, 2002; Walsh and Powles 2007), adjusting planting dates (Gill and Holmes 1997; Walsh and Powles 2007; Webster et al. 2009), irrigation management (Walker and Buchanan 1982), crop rotation (Johnson and Coble 1986; Schreiber 1992; Schweizer and Zimdahl 1984), and judicial application of fertilizers with appropriate placement. By hampering weed emergence and growth, each of these practices lessens weed fecundity (Norsworthy et al. 2007b), which is critical to reducing the soil seedbank and risks of herbicide resistance (Neve et al. 2011a). Most of these practices also result in increased crop yields (Howe and Oliver 1987; Norsworthy and Frederick 2005; Norsworthy and Oliver 2001).

BMP 9: Use Mechanical and Biological Management Practices Where Appropriate

BMP 10: Prevent Field-to-field and Within-Field Movement of Weed Seed or Vegetative Propagules

The potential for field-to-field and farm-to-farm movement of weed propagules, resulting in introduction of new weed species or HR weeds into noninfested areas must be recognized. Preventative weed management incorporates practices that avoid introduction, infestation, or dispersion of weed species into areas currently free of those weeds. Natural seed dispersal in the absence of a specialized dispersal mechanism or agent is generally limited to less than 5 m from the mother plant (Verkaar et al. 1983), and often, human activities are an important means of weed seed dispersal (Burton et al. 2006; Gonzalez-Andujar et al. 2001). Hence, preventing the immigration of propagules or pollen movement from adjacent infested areas and preventing propagule dispersal from infested areas to noninfested areas should be an important consideration to all weed management programs.

Most weed populations exhibit an aggregated or patchy distribution (Wiles et al. 1992), attaining spatial stability over time (Beckie et al. 2005; Marshall and Brain 1999; Rew and Cussans 1997; Rew et al. 1996; Wilson and Brain 1991). The degree of spatial stability is more pronounced in perennial species and in large-seeded weeds, such as wild common sunflower and velvetleaf, which exhibit shattering and localized seed dispersal before crop harvest (Burton et al. 2005; Colbach et al. 2000; Dieleman and Mortensen 1999). Although short-distance, localized dispersal of any weed is still undesirable, the opportunities for long-distance dispersal events, followed by rapid colonization, even if less frequent, are of particular concern for herbicide-resistance management. In an effective herbicide-based weed management system, patches of low-density weed populations that do not form a persistent seedbank can be driven toward extinction (Humston et al. 2005). The key to successful eradication of an HR weed population is early recognition of resistance, when the infestation is still minor and manageable, with low seedbank density. However, when a few resistant plants survive within a field, their occurrence is often overlooked until the population builds up to a level that herbicide failure is recognizable; by which point, seed and pollen carrying resistance alleles may have dispersed great distances, exacerbating the problem from a single incidence for an individual producer to one that affects an entire county, state, or region.

BMP 11: Manage Weed Seed at Harvest and Postharvest to Prevent a Buildup of the Weed Seedbank

BMP 12: Prevent an Influx of Weeds into the Field by Managing Field Borders

Weeds in unmanaged habitats, including field margins, roadsides, ditchbanks, railroads, and riparian zones can serve as a corridor for the introduction and movement of new weed species (Boutin and Jobin 1998), including HR weeds. The density, diversity, and fecundity of wild and weedy plant populations are generally greatest along field edges and decrease with increasing distance into crop fields (Marshall 1989; Wilson and Aebischer 1995). Although research emphasis on field borders as a source of weed seed introduction is limited in the United States, there has been a wealth of research in this area in European countries, mainly as an opportunity for preserving farmland biodiversity (Kiss et al. 1997; Marshall 1989; Noordijk et al. 2011; Wilson and Aebischer 1995). In the context of herbicide-resistance management, allowing resistance-prone weeds to produce seed in field borders may exacerbate the risk of resistance.

Most crop producers in the United States place emphasis on maximizing crop yields by management for weed removal within fields but often neglect to manage field borders, from which some of the most invasive and noxious weeds can find ingress via seed dispersal or through gradual movement of vegetative propagules, especially in the absence of tillage. Allowing weed seed production in field borders can have long-term effects on the seedbank of crop production fields, especially when outcrossing occurs with resistant populations near a field, allowing spread of resistance through pollen and seed movement. Simulation models have shown that weeds with high dispersal potential located in or near field borders have greater potential to deleteriously affect crop yields across a field compared with more-competitive weeds that have a low potential for dispersal (Maxwell and Ghersa 1992). Several of the tactics discussed previously can be used to minimize weed entry from field borders, roadsides, and ditchbanks (see BMPs 1 and 2).

Sowing field borders to a less-weedy, preferred species that can be managed later is another tactic for reducing immigration of agriculturally important weeds from field edges. However, Noordijk et al. (2011) concluded that regular haying as a field-edge management tactic is more effective in preventing weed seed immigration than sowing wildflower mixtures along field margins. Planting and maintaining a dense grass cover in field borders may prevent the establishment of arable weed communities through shading and regular mowing. In many parts of the United States, roadsides and ditchbanks are managed by county and state entities through mowing, haying, and herbicide applications.

In the southern United States, where GR Palmer amaranth is a serious problem on many farms, weeds in ditchbanks and along roadsides cause serious concern because of potential seed entrance into adjacent crop fields (Bennett 2011). Glyphosate was traditionally used to manage field margins, but because of the widespread occurrence of resistant Palmer amaranth, other means, such as mowing, followed by repeat applications of paraquat or complete disking have been followed to prevent seed production in these populations.

Patterns of Land Ownership

The long-term economic benefits of BMPs are certain; however, many growers, out of necessity, focus on immediate economic benefits. Nationwide, approximately 38% of U.S. farmland is operated by tenants, and in parts of the Midwest and Mississippi Delta regions of the United States, more than 50% of the farms are tenant operated (USDA-NASS 2007), creating an enormous challenge for promoting BMPs with a long-term view. Farmers usually approach weed management on these rented farms differently than they do on land they own and plan to farm for many years (Soule et al. 2000). Growers on rented land often focus their attention on the economic returns of the current crop and are much less concerned with the long-term effects of management decisions on land they may not farm the following year. Also confounding the issue of rented land are restrictions imposed by the landlord on the tenant concerning crop rotations or, more specifically, concerning a particular crop to be grown annually without rotation. Educational programs that target landowners could be beneficial in altering that behavior; however, there may be economic issues involving vertical integration and investment of infrastructure of the farming operation that facilitates those management decisions. For instance, a land owner who operates a cotton-ginning facility will often rent only to tenants who will grow cotton to supply the raw materials for that business. Regardless of the motivation behind this practice, weed management strategies available to the tenant because of these stipulations can be significantly hampered in a monoculture system.

Grower Perceptions of Future Technology and Effectiveness of Current Practices

There are at least two additional factors that may hinder the adoption of practices that contribute to herbicide stewardship: (1) the expectation that the release of a new technology to solve the HR weed problems is imminent, and (2) the belief that resistance-management strategies will be futile (Webster and Sosnoskie 2010). In Australia, a survey revealed that growers who believed new herbicide chemistries would be available soon were less likely to adopt strategies designed to delay the occurrence of HR weeds, compared with those who were uncertain about those alternatives (Llewellyn et al. 2007). The basis for this belief is rooted in recent history (Webster and Sosnoskie 2010). For instance, triazine-resistant weeds were effectively managed with ALS-inhibitor herbicides following their release. Once weeds subsequently evolved resistance to ALS-inhibitors and ACCase inhibitors, GR crop technologies allowed growers to effectively manage those HR species, and many farmers believe that this trend will continue.

Although it is a common perception that new herbicides or herbicide technologies will be developed to combat HR weeds, there has not, unfortunately, been a new herbicide MOA introduced commercially since 1998. In fact, the dominance of glyphosate and GR crops resulted in less investment in research and development from agrichemical companies (Duke 2011). A significant decline in herbicide-discovery research occurred during the period from 1995 through 2010 (Gerwick 2010).

The assessment that herbicide-resistance management strategies will be unsuccessful is based on the belief that herbicide resistance in weeds occurs randomly and cannot be contained. Not only can resistance traits be moved through seed dispersal by natural elements (e.g., water, wind, and fauna) and agricultural equipment, but they can also be transferred through pollen movement (Busi et al. 2008; Hidayat et al. 2006; Murray et al. 2002; Sosnoskie et al. 2009). Transfer of herbicide-resistance genes among weed populations through pollen or seed movement has significant implications for resistance prevention and management. In a survey conducted in Australia, about 70% of the growers suspected that their neighbors were the ultimate source of HR weeds currently on their farms, through either the movement of seed or the movement of pollen (Llewellyn and Allen 2006). Given the potential for the movement of herbicide-resistance traits among weedy populations, many growers reject the assumption that investment in resistance management programs will delay the occurrence of herbicide resistance on their farm (Llewellyn and Allen 2006).

Hardin (1968) first proposed the concept of the “tragedy of the commons,” whereby a commonly held resource is depleted by those who exploit it for individual gain, even though the ultimate loss of the resource is collectively disadvantageous. This idea has been expanded and applied to agricultural settings by others (Gould 1995; Webster and Sosnoskie 2010). Hardin provides the example of a shared resource, such as a body of water, spoiled by individuals or businesses who found discharging wastes into that water was more economical for them than the costly treatment of the waste. Likewise, indiscriminate herbicide use leading to the rapid evolution of herbicide resistance in weeds may result in the loss of herbicide options for all growers, especially if resistance spreads rapidly across a region. Spoiling of the common resource of pesticide susceptibility by some growers can make others reluctant to spend time and money to implement management programs that deter the evolution of pesticide resistance (Gould 1995). Additionally, it is expected that the adoption of BMPs for resistance management will be slow because of the steep learning curve associated with proficient use of new and often complex technologies and because of the challenge of clearly demonstrating the benefit of practices involving short-term costs for longer-term prevention. The key challenge is to develop and implement a sustained education and awareness campaign to overcome the various deterrents to adoption of resistance-management practices and to establish or strengthen local networks of farmers willing to work in concert.

Recommendation One: Reduce the Weed Seedbank through Diversified Programs that Minimize Weed Seed Production

Sustainable herbicide-resistance management requires a much longer-term perspective than that of a single season. Maintaining low weed-seed numbers in the soil seedbank reduces the number of plants that will be exposed to future herbicide treatment and can substantially reduce resistance risk (Neve et al. 2011a). Thus, there are advantages to controlling weed seed production even if such management might not lead to direct economic benefits in the current season.

For herbicide-resistance management, preventing viable weed seed production is necessary. Occasionally, there will be failures, and weeds capable of producing seed may not be controlled. In situations in which initial weed management efforts are unsuccessful, producers must take appropriate action to prevent weed seed production or vegetative propagation, including postharvest weed control.

Although model data show that weed seed added to the soil seedbank increases the risk of resistant weeds, research is needed on more weed species, locales, and production scenarios to confidently incorporate vital weed-seed reduction strategies into weed resistance BMPs. Additionally, preventing seed production, including after the harvest, must be strongly emphasized by extension educators, local distributors, and others communicating these BMPs to producers and encouraging their use.

Recommendation Two: Implement a Herbicide-MOA Labeling System for All Herbicide Products, and Conduct an Awareness Campaign

Given the number of different herbicide MOAs, the unfamiliarity of MOA specification to nonspecialists, the frequent use of mixed MOA herbicide products, frequent changes in herbicide products in the market, and the complexity of herbicide chemistry and MOA biochemistry, there is a need for the introduction of clear, standard herbicide-MOA labeling on herbicide use instructions (labels) and containers. To reduce overreliance on particular herbicide MOAs and to enable diversity in herbicide use, there must be an easy and consistent identification of product MOAs. All herbicide products should be clearly identified by standard MOA designations. The inclusion of MOAs on labels is essential for basic communication and will facilitate programs to better inform producers and their advisors of the importance of herbicide diversity. This recommendation is supported by the positive experience of establishing MOA labeling in Australia and Canada, where MOA labeling has facilitated awareness and adoption of herbicide-MOA rotation as a resistance management strategy (Beckie 2006; Shaner et al. 1999). Currently, some herbicide manufacturers in the United States voluntarily provide MOA labeling on their products.

To be successful, MOA labeling must be implemented and supported by manufacturers. However, given the diversity of manufacturers, suppliers, products, and product names, some form of industry-wide coordination and compliance will be needed to ensure comprehensive implementation. In countries that have adopted the practice, the herbicide industry self-regulated to ensure MOA labeling appeared on all herbicide products and associated materials (Beckie 2006; Shaner et al. 1999). Should self-regulation not be implemented or not be effective, regulatory authorities could mandate MOA labeling. Ideally, the herbicide industry and the scientific community would work together with regulators to achieve MOA labeling. In Australia, mandatory herbicide MOA labeling was cooperatively achieved 15 yr ago. Establishing a similar requirement, whether by agreement among companies or through a regulatory agency, is essential to the implementation of herbicide-resistance management in the United States. Both the Weed Science Society of America (WSSA) and the Herbicide Resistance Action Committee have developed simple, clear MOA designations. Because only the WSSA MOA designations are approved in the United States and Canada for use on herbicide labels, the WSSA system should be used for this application.

We propose MOA labeling with a unified program of education jointly developed, supported, and espoused by the WSSA, pesticide manufacturers and their trade association, and the purveyors of HR cultivars. We also propose that herbicide MOAs for each recommended herbicide be included and easily referenced in all state recommendation guides and websites.

Recommendation Three: Communicate that Discovery of New, Highly Effective Herbicide MOAs Is Rare and that the Existing Herbicide Resource Is Exhaustible

The international herbicide-discovery industry has been spectacularly successful in producing new herbicides in the second half of the 20th century. Producers are aware of this success and, consequently, have high expectations that new herbicides capable of overcoming resistance issues will continue to enter the market (Llewellyn et al. 2002; Wilson et al. 2008). However, the reality is that herbicide discovery has slowed dramatically in recent years, as is evident in the reduction in the number of herbicide patents issued (Ruegg et al. 2007). Therefore, to encourage crop producers to achieve herbicide sustainability, current optimistic expectations about new herbicide introductions will need to be addressed. Dampening high expectations will be difficult because of the successful record of herbicide technology, the confidential nature of the development pipeline, and the unpredictability of new herbicide discovery. Achieving consensus on this matter in a highly competitive herbicide industry may also be difficult. However, focusing on important and unique herbicide examples, such as glyphosate, may be an effective way to illustrate the potential for losing valuable herbicide resources and to communicate the need to use products in a sustainable manner so as to retain their utility. Should there be a widespread reduction in the weed control spectrum of a dominant herbicide, such as glyphosate, it will be important to stress the sustainability of strategies and technologies that would replace it.

Recommendation Four: Demonstrate the Benefits and Costs of Proactive, Diversified Weed Management Systems for the Mitigation of HR Weeds

Because of their relative advantages in cost efficiency, reliability, and ease of use, herbicides will remain central to weed control strategies. Such reliance means that the case for reducing dependence on a preferred herbicide option must be strong and clearly communicated. Where the sustainability of a key herbicide is threatened, practices to sustain its use must be emphasized.

The most effective means of conserving herbicide MOAs is diversifying weed management programs by combining herbicides with different MOAs and incorporating nonchemical control tools. Diversification reduces direct selection pressure by reducing the number of applications of a single herbicide, especially the number of consecutive applications of a single MOA. Application of two MOAs in sequence tends to conserve the efficacy of both, but particularly that of the first applied because the second application may control weeds that escaped the first by reason of incipient resistance. Growers may be reluctant to replace simple weed management programs that rely on a single or very few herbicides with more diverse weed management programs because the complexity of such programs is a deterrent (Jordan et al. 2003; Liebman and Gallandt 1997; Swanton et al. 2008).

Making alternative weed management programs as simple and as cost effective as possible will be required if preemptive herbicide-resistance management is to be widely adopted. The rapid and extensive adoption of GR crops illustrates that simplicity and convenience in weed control are highly valued by modern crop producers (Dill et al. 2008; Marra et al. 2004; Piggott and Marra 2008; Shaw et al. 2009). Complexity consumes management time, attention, and labor (Kingwell 2010; Pardo et al. 2010), and those costs must be clearly addressed when promoting herbicide sustainability through adoption of diversity in weed management strategies.

A persuasive argument for adoption of diverse resistance-management practices may be that costs of control after herbicide resistance has evolved are often higher than the cost of a program for reducing the risk of resistance in the first place (Mueller et al. 2005; Orson 1999). A successful example of preemptive adoption of a resistance risk-averting practice is the campaign to encourage glyphosate sustainability in Australia using a “double knock” treatment. The double-knock treatment incorporates an application of a second herbicide MOA, other than glyphosate, shortly after the glyphosate preseeding application (Neve et al. 2003b; Weersink et al. 2005).

Recommendation Five: Foster the Development of Incentives by Government Agencies and Industry that Conserve Critical Herbicide MOAs as a Means to Encourage Adoption of Best Practices

Herbicides are the most cost-effective means of weed control in most cropping systems. Growers may accept the risk of herbicide resistance and overuse a single or a few herbicide MOAs when competitive diversified programs are not readily apparent. Diversification may occur in three ways:

We recommend the expansion of the USDA Natural Resources Conservation Service (NRCS) program, which supports producers by creating positions for technical service providers to assist in the development and implementation of sustainable herbicide-resistance management programs. USDA-NRCS is also developing plans for cost-share incentives for producers to help reduce the risks of resistance on farms, or to contend with weeds that have already evolved resistance (Robinson 2011). Support for soil-conservation farming systems and improved environmental outcomes could be a focus for conserving valuable herbicide MOAs (D'Emden et al. 2008; Givens et al. 2009; Gray et al. 1996; Marsh et al. 2006).

Recommendation Six: Promote the Application of Full, Labeled Rates at the Appropriate Weed and Crop Growth Stage

When tank mixtures are employed to control the range of weeds present in a field, each product should be used at the specified label rate appropriate for the weeds present. Repeated use of rates below those recommended on labels increases the probability of resistance (Neve and Powles 2005a,b; Norsworthy 2012). Below-labeled rates allow more weeds to survive treatment, and those weeds produce seed in the survivors that will lead to an accumulation of genes that confer resistance via multigenic mechanisms. Although producers can avoid knowingly applying a reduced herbicide rate, they must also be aware that rates can be unintentionally reduced by poor spray coverage or by applying a herbicide to weeds larger than those recommended.

It is in the herbicide manufacturers' interest to specify the manner of use of a proprietary herbicide because of probable adverse market consequences from misuse or complaints of product failure. Such stipulations are standard in herbicide labeling where specific off-label uses are not permitted, e.g., those that have a high likelihood of failure or of off-target damage. All herbicide labels give quite specific information on the application rates, weed sizes, conditions, and other factors that may affect activity of that herbicide. We recommend the crop-protection industry draft herbicide labels with mandatory minimum-use rates so the state regulatory agencies can enforce compliance. Although a pesticide label is a legal document, it may contain both mandatory and advisory statements. Minimum use rates can be made mandatory if correctly written, and that would require users to comply with those specifications to avoid increasing the risks of resistance evolution.

Recommendation Seven: Identify and Promote Individual BMPs that Fit Specific Farming Segments with the Greatest Potential Impact

We recognize that there are substantial gaps in research knowledge of herbicide-resistance BMPs. We also recognize that a BMP for a specific weed in one locale may not be the best practice in another locale or under different crop management practices. For example, tillage may offer a viable management strategy in some areas, whereas in areas susceptible to wind or water erosion, tillage would be a poor option. Likewise, herbicide use varies among locales because of differences in soils, weather, and general production practices. Weed scientists, including extension specialists, industry personnel, and producers must work cooperatively to establish and verify BMPs for problem weeds in various regions. In some cases, computer modeling can be used to predict outcomes without the need for numerous long-term field studies (Neve et al. 2011a). However, field studies are needed to collect data to parameterize models and validate model predictions. General BMPs, whether developed by model simulation or field experience, should be used as a base recommendation, allowing local weed scientists and producers to refine programs to fit individual management needs. IWM to reduce the risk of resistance combines multiple BMPs, and the path to reducing resistance risk is likely to be achieved in incremental steps from simple to more complex practices as producers gain confidence about adopting new strategies (Byerlee and Polanco 1986).

Information flow to producers occurs through private crop consultants, herbicide retailers, popular media, university research and extension systems, and other producers (Givens et al. 2009; Norsworthy et al. 2007d). Therefore, the first step in encouraging producers to adopt BMPs for herbicide sustainability in a particular state or region is to understand and access those same information pathways, using educational materials that include practices suitable for local situations and farming practices. A wide range of factors affects producer decisions regarding IWM and herbicide resistance (Shaw et al. 2009; Wilson et al. 2008, 2009). Identifying those factors at a local level allows targeted information delivery to improve decision-making (Llewellyn 2007). Many aspects of herbicide management, including managing for sustainability, require region-specific decision-making and, therefore, regional research.

Many of the BMPs offer benefits to the farming system beyond weed control, and those benefits will give producers extra incentive to adopt the practices (Llewellyn et al. 2004, 2005). Relative cost–benefit predictions for various BMPs should be presented in educational materials and programs and should be based on local farming economics. Practices that offer the greatest relative advantage in the short term are more likely to be adopted preemptively. At present, adoption of complex and costly practices is more likely with increased producer awareness of herbicide resistance, presence, or risk on their farms. However, area-wide promotion of preemptive BMPs developed to fit local production problems may encourage producers to realize the importance of these practices and increase their confidence that resistance management can be successful for their operation.

Recommendation Eight: Engage the Public and Private Sectors in the Promotion of BMPs, Including Those Concerning Appropriate Herbicide Use

The weed resistance crisis provides an excellent opportunity for partnerships between public- and private-sector consultants and scientists. The potential to reach growers and their advisors can be greatly increased when a science-based recommendation gains the support of the communication resources of a manufacturer. The double-knock concept in Australia (Walsh and Powles 2007) is one example that led to high levels of preemptive adoption of a practice.

Herbicide manufacturers should not constrain sales in any way as a part of a cooperative effort with the public sector to promote herbicide diversity in resistance management programs. However, given the realities of resistance evolution, we recommend that herbicide manufacturers unequivocally state and promote on the label and in all advertising material that the herbicide be used in a manner that embraces the principles of sustainable weed management. In this way, herbicide manufacturers are promoting a message that is consistent with public- and private-sector educational efforts.

Recommendation Nine: Direct Federal, State, and Industry Funding to Research Addressing the Substantial Knowledge Gaps in BMPs for Herbicide Resistance and Support Cooperative Extension Services as Vital Agents in Education for Resistance Management

Our current knowledge of best practices for managing herbicide resistance is incomplete. Much of the research on developing herbicide-based management strategies in cropping systems is funded by herbicide manufacturers, who are primarily interested in approaches that benefit their bottom line. Federal and state funds are, therefore, needed to meet additional research needs. Weed management is local and requires specific solutions that do not rely on one technique or one herbicide. Agriculture needs a diversity of weed management tools and strategies for every crop, including resistance management programs that use more than one herbicide or management technique in each cropping scenario. Although industry support will likely continue for some research, federal and state funding must be available and be substantial enough to support resistance-management projects for which private funding is not available or insufficient.

A vital aspect of developing and implementing successful resistance-management BMPs is information transfer to consumers of those practices. A great deal of this information is articulated through state and local extension specialists. These professionals, in addition to private crop consultants, are at the forefront of introducing producers to new information and strategies through one-on-one communication, grower and consultant meetings, cooperation in field demonstrations at research station field days, popular press news releases, and technical brochures and websites. In some states, weed resistance has compelled extraordinary cooperation among academia, extension specialists, consultants, industry personnel, and producers to develop solutions to urgent problems. However, it is vital that resistance-management research and education be supported by both private and public sectors. The need for public funding is necessary and justified given the dependence of our society on the agricultural community and its crop production.