Subject: Public Participatioin for Dicamba: New use on herbicide-tolerant cotton and soybean Public comment prepared by: Dr. David A. Mortensen, Professor, Weed and Applied Plant Ecology, Ecology Inter College Graduate Program and The Plant Sciences Department, The Pennsylvania State University, University Park, PA Summary Statement I am writing to strongly oppose Monsanto's deregulation request for dicamba resistant traits in soybean and cotton. The data and the indigenous knowledge of life--- long agricultural practicioners do not support this request. The argument is laid out in greater detail but I briefly make my argument here with supporting text below. First, deregulation will do little to address the herbicide resistance problem. Ample evidence now exists that these traits will do little to nothing to solve the herbicide resistant weed problem. In fact, in the mid--- to longer term they will exacerbate the resistance problem. The transgene facilitated herbicide treadmill (Mortensen et al. 2012) is over reliant on herbicides to achieve sustainable weed management. As recently as this past week, another paper was published underscoring the folly that is solving the glyphosate resistance problem with auxinic herbicides ( ). In addition to the deregulation request not addressing the resistance problem, we have new and promising practices that can. Cover cropping is just one such practice, one that the NRCS and USDA are championing as a multi---functional solution to pest and soil management concerns. Second, the deregulation of this herbicide use poses PROFOUND risks to broadleaf crop growers as the risk of physical and vapor drift of dicamba is large (we've spent six years studying the problem). Recent multi---year assessments of how farmers co---exist when an divergent practices are used on their farms concluded that pesticide drift should be reduced wherever possible. The deregulation request in front of the EPA greatly increases the risk of chemical trespass through herbicide drift by virtue of the fact that physical and vapor drift is extremely common with dicamba AND increasing it's use by 5---7 fold AND at times of the year when drift is highest AND when adjacent plants are most susceptible (Egan, Barlow and Mortensen 2014) will dramatically increase injury of adjacent crops and field---edge plants. Third, the deregulation of dicamba in the way proposed in the deregulation request will dramatically decrease plant diversity AND the important provisioning of natural enemies and pollinators (see Bohnenblust et al. 2016 and Kammerer et al. 2016). While data from these papers clearly documents the role that field edges play in supporting beneficial insects and how dicamba drift dramatically decreases the provisioning capacity of plants, our understanding of these dynamics is limited. It is essential that additional studies be conducted in order to know whether the dramatic effects we see in these limited studies would be commonplace if the deregulation request were approved. Fourth, it is widely researched but apparently not widely known that increasing the area treated and amount applied of herbicide increases the likelihood that herbicides appear in surface and groundwater (Barabash 1999). The fact that we would increase dicamba use by 3---7 fold at a time when we're working hard to reduce human and natural systems exposure to such compounds goes against national policy, concerns about human health and the integrity of the agroecological matrix. For these "science---based" reasons, I strongly oppose the dicamba resistant crop deregulation request in front of the EPA. Pest Problems will result and will arise in one of two ways The definition of pest is far too narrowly defined in the USDA APHIS EIS. While the EIS addresses the issue of herbicide resistant crops becoming weeds themselves this review of whether a pest problem results is far too narrow as there are other pest problems that will result from widespread adoption of the 2,4---D. First, it is abundantly clear that herbicide resistant weeds will suffer a short---term setback from early use of 2,4---D and glyphosate, however in the mid to longer term, resistance will be a growing problem. We are burying our heads in the sand if we don't believe this to be the case, one only need to read the short---sighted quotes of scientists claiming glyphosate resistance was unlikely to occur. For example in 1997 as commercialization of glyphosate resistant soybean was in it's second year, Bradshaw et al. (1997) stated "The lack of evolution of weed resistance to the herbicide glyphosate has been considered from several perspectives. Few plant species are inherently resistant to glyphosate. Furthermore, the long history of extensive use of the herbicide has resulted in no verified instances of weeds evolving resistance under field situations." Therefore "the complex manipulations that were required for the development of glyphosate---resistant crops are unlikely to be duplicated in nature to evolve glyphosate---resistant weeds." I address this issue more fully below. The second important area regarding new pests is the insect pest complex that arises broadleaf plant species are removed from the agroecological matrix. This oversight is deeply disturbing as it comes at a time when we're just beginning to understand the importance of plant diversity on our farmsteds. Here what I'm referring to is the natural enemies that are supported by a diverse flora (in the fields, field edges and other non---crop plant cover). I am perplexed that such things are overlooked in the EIS. There is mounting evidence that drift level doses reduce broadleaf plant cover, diversity and floral abundance (Egan et al. 2014b; Mortensen et al. 2012; Mortensen et al. in preparation). Here I'm not referring to "armchair" ecology but rather real effects that will matter to the farmer. Reducing floristic diversity or habitat heterogeneity results in an increase in pest pressure and a corresponding increase in insecticide use to address these pests. Our work along with that of Doug Landis (entomologist, MSU), Mary Gardiner (entomologist, OSU), John Tooker (entomologist, PSU), Felix Bianchi (entomologist, Wageningen University) and others have been quantifying the degree to which natural enemies that attack cutworm, earworm, soybean aphid (among many others) rely on field edge plants and heterogeneity in plant cover for critical life history stages. It is essential that the EIS assess these pest problems broadly in order to fully assess the impact of these Dow deregulation requests. IN FACT, recent papers out of our lab document large losses in pollination host provisioning by plants receiving drift---level doses of dicamba herbicide. In this recent work, we have documented the floral resource AND the insect visitation to plants is all be eliminated at drift---level doses of dicamba. As a leading scientist in this field I have to say it has been enormously frustrating to see this application come to EPA. We now know (we didn't have the data when the 2,4---D deregulation request was under review) that drift level doses of dicamba are enormously damaging to important insect food webs. Here we have only scratched the surface of what we need to know before such deregulation requests can be fully assessed. What we do know is drift---level doses, doses that will routinely occur in the field under the use proposed herein will knock out beneficial insect food webs, webs that include pollinators and natural enemies. What does this mean? It means that at the very same time that we're incentivizing farmer practices to preserve and enhance pollinator habitat we'd consider deregulating a crop management practice that will dramatically undermine this national effort. The same can be said for the monarch butterfly. Here again, too little research has been conducted but what has indicates that mixtures of these broadleaf herbicides (glyphosate and dicamba) are synergistic and enhance perennial weed suppression, suppression of the very plant species we're working to conserve (through practices identified in the President's Bee Foraging Summit). An increase in herbicide resistant weed pests - What follows is an excerpt from our Navigating a Critical Juncture paper (Mortensen et al. 2012), it is included because the problem of pests arising from a solely herbicide dependent form of weed control will strongly select for an increase in herbicide resistant weeds and this profoundly important point is largely or inadequately addressed in the USDA APHIS EIS. Glyphosate resistant weeds rapidly evolved in response to the intense selection pressure created by the extensive and continuous use of glyphosate in resistant crops. Anticipating the obvious criticism that the new synthetic auxin resistant cultivars will enable a similar overuse of these herbicides and a new outbreak of resistant weeds, scientists affiliated with Monsanto and Dow have argued that synthetic auxin resistant weeds will not be a problem because: i) currently very few weed species globally have evolved synthetic auxin resistance despite decades of use; ii) auxins play complex and essential roles regulating plant development, suggesting that multiple independent mutations would be necessary to confer resistance; and iii) synthetic auxin herbicides will be used in combination or rotation with glyphosate, requiring multiple resistance traits for weeds to survive (Behrens et al. 2007, Wright et al. 2010). Although these arguments have been repeated in several high---profile journals, they conspicuously leave out several important facts about current patterns in the distribution and evolution of herbicide resistant weeds. First, similar arguments were made during the release of glyphosate resistant crops. Various industry and university scientists contended that details of glyphosate's biochemical interactions with the plant enzyme EPSPS combined with the apparent lack of resistant weeds after two decades of previous glyphosate use indicated that the evolution of resistant weeds a negligible possibility (Bradshaw et al. 1997). Secondly, it is not the case that "very few" weed species have evolved resistance to the synthetic auxin herbicides. Globally, there are 28 species, with 6 resistant to dicamba specifically (Egan et al. 2011), 16 to 2,4---D, and at least two resistant to both active ingredients. And while many of these species are not thought to infest large areas or cause significant economic harm, data on the extent of resistant weeds is compiled through a passive reporting system, where area estimates are voluntarily supplied by local weed scientists once a resistant weed problem becomes apparent. Synthetic auxin resistant weeds may appear unproblematic because these species currently occur in cropping systems where other herbicide modes of action are used that can effectively mask the extent of the resistant genotypes (Walsh et al. 2007). Furthermore, the claim that dicamba resistance is unlikely to evolve due to the complex and essential functions that auxins play in plants is unsubstantiated. In many cases where resistance has evolved to synthetic auxins, the biochemical mechanism is unknown. However, in at least two cases, dicamba resistant Kochia scoparia (Preston et al. 2009) and dicamba resistant Sinapsis arvensis (Zheng and Hall 2001), resistance is conferred by a single dominant allele, indicating that resistance could develop and spread quite rapidly (Jasieniuk and Maxwell 1994). The final dimension of the industry argument is that by planting stacked resistant traits, farmers will be able to easily use two distinct herbicide modes of action and prevent the evolution of weeds simultaneously resistant to both glyphosate and dicamba or 2,4---D. The logic behind this argument is simple. Because the probability of a mutation conferring target site resistance to a single herbicide mode of action is a very small number (generally estimated as one resistant mutant per 10---5 to 10---10 individuals (Jasieniuk and Maxwell 1994), and because distinct mutations are assumed to be independent events, then the probability of multiple target site resistance to two modes of action is the product of two very small numbers, i.e. 10---10 to 10---20. For instance, if the mutation frequency for a glyphosate resistant allele in a weed population is 10---9, and the frequency for a dicamba mutant is also 10---9, then the frequency of individuals simultaneously carrying both resistant alleles would be 10---18. If the population density of this species is assumed to be around 100 seedlings per m2 of cropland (106 per ha), then it would require 1012 ha of cropland to find just one mutant individual with multiple resistance to both herbicides. For point of reference, there are only about 15 x 108 ha of cropland globally. Thus, even if the weed species was globally distributed and all of the world's crop fields were treated with both herbicides, it would appear virtually impossible to select a single weed seedling exhibiting multiple resistance. The problem with this reassuring analysis is that it contradicts recent experience. Weed species resistant to multiple herbicide modes of action are becoming more widespread and diverse (Fig. 3, Mortensen et al. 2012). There are currently 108 biotypes in 38 weed species across 12 families possessing simultaneous resistance to 2 or more modes of action, with 44% of these appearing since 2005 (Heap 2011). Common waterhemp (Amaranthus tuberculatus) simultaneously resistant to glyphosate, ALS, and PPO herbicides infests 0.5 million corn and soybean hectares in Missouri (Heap 2011). Rigid ryegrass (Lolium rigidium) populations resistant to seven distinct modes of action infest large areas of southern Australia (Heap 2011). Weeds can defy the probabilities and develop multiple resistance through a number of mechanisms. First, when a herbicide with a new mode of action is introduced into a region or cropping system where weeds resistant to an older mode of action are already widespread and problematic, the probability of selecting for multiple target site resistance is not the product of two independent, low probability mutations. In fact, the value is closer to the simple probability of finding a resistance mutation to the new mode of action within a population already extensively resistant to the old mode of action. For instance, in Tennessee, an estimated 0.8---2 million ha of soybean are infested with glyphosate resistant horseweed (Conyza canadensis) (Heap 2011). Assuming seedling densities of 100 m---2 or 106 ha---1 (Dauer et al. 2007) and a mutation frequency for synthetic auxin resistance of 10---9, this implies that next spring, there will be 800---2000 horseweed seedlings in the infested area that possess combined resistance to glyphosate and a synthetic auxin herbicide ((2x106 ha infested with glyphosate resistance) * (106 seedlings/ha) * (1 synthetic auxin resistant seedling/109 seedlings) = 2000 multiple resistant seedlings). In this example, these seedlings would be located in the very fields were farmers would most likely want to plant the new stacked glyphosate and synthetic auxin resistant soybean varieties (the fields where glyphosate resistant horseweed problems are already acute). Once glyphosate and synthetic auxin herbicides have been applied to these fields and killed the large number of susceptible genotypes, these few resistant individuals would have a strong competitive advantage and be able to spread and multiply rapidly. Secondly, several weed species have evolved cross resistance, in which a metabolic adaption allows them to degrade several different herbicide modes of action. Mutations to cytochrome P450 monoxygenase genes are a common mechanism for cross resistance (Powles and Yu 2010). Plant species typically have a large number of P450 genes (the rice genome contains 458 distinct P450 genes) involved in a variety of metabolic functions including the synthesis of plant hormones and the hydrolyzation or dealkylation of herbicides and other xenobiotics. Weeds with P450 mediated resistance are widespread and increasingly problematic. For instance, across Europe and Australia, numerous populations of Lolium rigidum and Alopecurus myosuroides occur with various combinations of P450 resistance to the ALS, ACCase, and PSII inhibitor herbicides (Powles and Yu 2010). Given the diversity and ubiquity of P450 monoxygenases in plant genomes, it is possible that in the near future a weed species could evolve a mutation that enables it to degrade glyphosate and the synthetic auxins. Historically, use of the synthetic auxins have been limited to cereals or as pre--- plant applications in broadleaf crops. The new transgenes will allow dicamba to be applied at higher rates, in new crops, in the same fields in successive years, and across dramatically expanded areas, creating intense and consistent selection pressure for the evolution of resistance. Taken together, the current number of synthetic auxin resistant species, the broad distribution of glyphosate resistant weeds, and the variety of pathways by which weeds can evolve multiple resistance suggest that the potential for synthetic auxin resistant or combined synthetic auxin/glyphosate resistant weeds in transgenic cropping systems is actually quite high. One hundred---ninety seven weed species have evolved resistance to at least one of 14 known herbicide modes of action (Heap 2011), and the discovery and development of new herbicide active ingredients has slowed dramatically over recent decades. Given that herbicides are a cornerstone of modern weed management, it seems unwise to allow the new GM herbicide resistant crops to needlessly accelerate and exacerbate resistant weed evolution. Integrated Review Needed Currently, new genetically engineered crops go through independent reviews by USDA APHIS to assess risks of gene exchange, the likelihood the transformed crop could become a pest (inadequately addressed in the current USDA APHIS Environmental Impact Statement (here forward EIS)) while "threats [that] could potentially result from proposed changes in herbicide labels are evaluated by the EPA". As the genome of the crop is manipulated in ways that directly influence pest management practices, it is essential that the review of such genetic modifications (USDA APHIS) and associated changes in pest management practices (EPA) be performed concurrently and by one integrated review panel. An integrated approach is particularly critical for herbicide resistance traits where changes in the genome result in large (5---7 fold) increases in use of the targeted herbicide. The distinction in the pest target is an important one. In the case of genetically engineered (GE) insect traits, plant incorporated protectants preclude the use of an insecticide for insect protection. GE herbicide resistance traits facilitate the use of herbicide(s) and therefore the trait is directly linked with likely increase in herbicide use. For example, in our recent BioScience paper (Mortensen et al. 2012 uploaded as a separate 'public comment') we estimate insertion of 2,4---D or dicamba resistance traits in soybean would increase auxinic herbicide use in the crop 5---7 fold. Therefore, insertion of herbicide resistance genes are inextricably linked to and positively correlated with herbicide use. This deregulation request will result in a significant increase in dicamba with that increase coming in the form of increased amount applied to a particular field and in the area treated. Of equal or possibly greater importance is the fact that dicamba will be applied over a much broader window of time, a window during which many highly sensitive broadleaf plants are leafed out and particularly vulnerable to injury. Another argument for an integrated and holistic review is found in the area of non---target effects. The deregulation requests for the Monsanto events argues dicamba resistant crops will not have significant non---target effects. Obviously, the crop cultivar itself won't adversely affect other organisms; the concern is over the herbicide use that will now be possible on the transformed dicamba resistant crops. A careful review of the peer--- reviewed environmental chemistry literature indicates that amount of herbicide use is positively correlated with the appearance of herbicides use in surface and ground water (Barbash et al., 2001). Auxinic herbicides like 2,4---D and dicamba have also been linked to a high frequency of drift injury events (see Pesticide Drift Enforcement Survey and Egan, Barlow and Mortensen, 2014a). Taken together, these spillover effects would be small in the absence of the transformed crop and associated agronomic practices. In addition to concerns about compromised environmental quality, herbicide spillover of the kind that would occur with the approval of this application will make it more difficult for fruit and vegetable farmers to coexist with grain crop farmers. Finally, large increases in dicamba would also potentially reduce the floristic diversity of field edges and the beneficial insects including pollinators and biocontrol organisms that rely on the flora (Mortensen et al., 2012 and Egan et al. 2014b). Therefore, a decision on deregulation of these Dow events must be weighed against the environmental consequences of such deregulation. Under the current review process, this is not possible. Literature cited Barbash, JE, GP Thelin, DW Kolpin, and RJ Gilliom. 1999. Distribution of Major Herbicides in Ground Water of the United States U.S. Geological Survey Water--- Resources Investigations Report 98---4245. Behrens MR, Mutlu N, Chakraborty S, Dumitru R, Jiang WZ, LaVallee BJ, Herman PL, Clemente TE, Weeks DP. 2007. Dicamba resistance: Enlarging and preserving biotechnology---based weed management strategies. Science 316: 1185---1188. Bohnenblust, EW, AD Vaudo, JF Egan, DA Mortensen, and J Tooker. 2016. Effects of the herbicide dicamba on non---target plants and pollinator visitation. Environmental Toxicology and Chemistry, Journal of Pest Science, 35:144---151. Bradshaw LD, Padgette SR, Kimball SL, Wells BH. 1997. Perspectives on glyphosate resistance. Weed Technology 11: 189---198. Dauer JT, Mortensen DA, VanGessel MJ. 2007. Temporal and spatial dynamics of long--- distance Conyza canadensis seed dispersal. Journal of Applied Ecology 44: 105--- 114. Egan, JF, BD Maxwell, DA Mortensen, MR Ryan, and RG Smith. 2011. 2, 4---D resistant crops and the potential for the evolution of 2,4---D resistant weeds. PNAS, 108(11) E38. Egan, JF and DA Mortensen. 2012. Quantifying vapor drift of dicamba herbicides applied to soybean. Environmental Toxicology and Chemistry, 31: 1---9. Egan, J.F., K.B. Barlow, and D.A. Mortensen. 2014a. A meta---analysis on the effects of 2,4---D and dicamba on soybean and cotton. Weed Science 62(1): 193---206. Egan, J.F., E. Bohnenblust , S. Goslee, D.A. Mortensen, and J. Tooker. 2014b. Herbicide drift can affect plant and arthropod communities. Agriculture, Ecosystems, and Environment 185: 77---87. Heap I. 2011. International Survey of Herbicide Resistant Weeds. (February 2011; http://www.weedscience.org/). Jasieniuk M, Maxwell BD. 1994. Population---genetics and the evolution of herbicide resistance in weeds. Phytoprotection 75: 25---35. Kammerer, MA, DJ Biddinger, EG Rajotte, and DA Mortensen. 2016. Local plant diversity across multiple habitats supports a diverse apple pollinator community. Environmental Entomology, 45:32---48. Mortensen, DA, JF Egan, BD Maxwell, MR Ryan, and RG Smith. 2012. Navigating a Critical Juncture for Sustainable Weed Management. BioScience, 62: 75---84. National Pesticide Information Retrieval System, Purdue University, West Lafayette, IN, Association of American Pesticide Control Officials, Pesticide Drift Enforcement Survey. Found at: http://aapco.ceris.purdue.edu/doc/surveys/DriftEnforce05Rpt.html Powles SB, Yu Q. 2010. Evolution in action: plants resistant to herbicides. Annual Review of Plant Biology, Vol 61 61: 317---347. Preston C, Belles DS, Westra PH, Nissen SJ, Ward SM. 2009. Inheritance of resistance to the auxinic herbicide dicamba in kochia (Kochia scoparia). Weed Science 57: 43---47. Walsh MJ, Owen MJ, Powles SB. 2007. Frequency and distribution of herbicide resistance in Raphanus raphanistrum populations randomly collected across the Western Australian wheatbelt. Weed Research 47: 542---550. Wright TR, et al. 2010. Robust crop resistance to broadleaf and grass herbicides provided by aryloxyalkanoate dioxygenase transgenes. Proceedings of the National Academy of Sciences of the United States of America 107: 20240---20245. Zheng HG, Hall JC. 2001. Understanding auxinic herbicide resistance in wild mustard: physiological, biochemical, and molecular genetic approaches. Weed Science 49: 276---281. May 31, 2016 US EPA/OSCPP/OPP 7505P Environmental Protection Agency 1200 Pennsylvania Ave. NW Washington DC 20460 RE: Docket No. EPA---HQ---OPP---2016---0187, Dicamba: proposed new use on herbicide resistant cotton and soybeans ELECTRONIC SUBMISSION via www.regulations.gov Pesticide Action Network North America (PANNA) submits the following comment regarding EPA's proposed registration of dicamba for use on genetically engineered dicamba---resistant cotton and soybeans. PANNA is a non---profit, public interest organization representing the concerns of over 100,000 supporters across the country, including farmers, farmworkers, health professionals, members of sustainable agriculture, labor, environmental and consumer groups and individuals concerned with the safety, sustainability, fairness and integrity of our food and agricultural system. Our members are deeply concerned about the serious social, economic, environmental and health harms to farmers, workers and rural communities that would accompany EPA registration of dicamba for use on genetically engineered dicamba---resistant crops. We therefore urge EPA to reject Monsanto's petition for use of dicamba on these crops. Drift damage to vulnerable crops, farmers' livelhoods and ecosystems Dicamba products on the market today are highly volatile. Dicamba has been identified by the Association of American Pesticide Control Officials as the third most commonly involved herbicide in drift occurrences.1 Volatilization leading to drift occurs more readily at higher temperatures (e.g. midseason, when dicamba could still be applied to Monsanto's dicamba---resistant varieties). Mechanical spray drift alone (e.g. when the herbicide is applied during commonly occurring wind conditions or with incorrect farm equipment) readily causes damage to vulnerable crops and adds to the threat of volatilization drift. Dicamba residues are also difficult to remove from pesticide applicators' equipment. Because miniscule, residual amounts left in a sprayer can harm crops that are subsequently sprayed with other herbicides, the likelihood that vulnerable crops treated by an applicator's dicamba---contaminated equipment will be harmed increases. 1 Association of American Pesticide Control Officials (AAPCO). "2005 AAPCO Pesticide Drift Enforcement Survey Report." 2005. On file and available at http://aapco.ceris.purdue.edu/doc/surveys/DriftEnforce05Rpt.html. Accessed May 3, 2013. OAKLAND o SACRAMENTO o MINNEAPOLIS 1611 TELEGRAPH AVE, SUITE 1200 o OAKLAND, CA 94612 o 510.788.9020 o WWW.PANNA.ORG PRINTED ON RECYCLED PAPER Dicamba is also highly toxic to broadleaf plants. Incidences of both mechanical spray and volatilization drift, as well as unintended contamination of spray equipment, are likely to rise sharply, and because of the herbicide's high toxicity, threatens growers of specialty crops and non---dicamba---resistant commodity crops with severe crop damage and yield loss. Highly sensitive crops include nearly all fruits, vegetables, seed and nut crops, such as grapes, beans, lettuce, tomatoes, soybeans, sunflower, cotton and peanuts, among others. The specialty crop industry as well as seed and vegetable oil and fiber production, would be seriously impacted. With USDA's 2015 deregulation of dicamba---resistant cotton and soybean, followed by its 2016 deregulation of dicamba---resistant corn, the window for dicamba spraying will be significantly widened, with more dicamba applications likely to occur mid---season when temperatures are warmer and volatilization occurs more readily and when vulnerable crops have leafed out and are extremely susceptible to dicamba damage. The acreage on which dicamba will be applied will also increase from current levels, as farmers begin to cultivate dicamba---resistant crops. The likelihood of dicamba drift causing crop injury and severe harm to specialty crop and organic farmers, as well as non---target species, poses a severe and unacceptable risk for thousands of American farmers. Other non---crop broadleaf plants e.g. in hedge rows, at field---edge or throughout the larger landscape, are also likely to be harmed, destroying critical habitat, food and reproductive sites for birds and other beneficial species critical to agroecosystem health (pollinators, natural enemies). Commodity growers' efforts to diversify their farms with perennials and other crops, support agriculturally critical ecosystem services, reduce wind and water erosion and diversify sources of farm income, would be undermined. Herbicide resistance and weed management U.S. farmers are facing an unprecedented crisis in the spread of herbicide---resistant weeds. USDA's approval and the subsequent widespread planting of Monsanto's Roundup Ready varieties have led directly to the current weed crisis, in which glyphosate---resistant weeds now cover over 70 million acres of farmland. With the expected surge in dicamba use that USDA and Monsanto both acknowledge will accompany cultivation of Monsanto's dicamba---resistant cotton and soybean varieties, farmers are likely to face the spread of intractable dicamba---resistant weed populations. Already several weed species are resistant to dicamba, and with resistance in the case of at least two weed species conferred by a single dominant allele, that resistance could spread swiftly. 2 Furthermore, a number of weed species have developed multiple resistance (to more than one herbicide) and/or cross---resistance (in which a metabolic adaptation in a weed species enables it to degrade several different herbicide modes of action at once). The spread of weed populations resistant to dicamba, the evolution of dicamba resistance in weed species and the emergence of volunteer dicamba---resistant corn and soybean plants all pose serious threats to the future of American farming. 2 Mortensen, David et al. 2012. Navigating a critical juncture for sustainable weed management. BioScience 62(1): 75--- 84. An EPA registration decision will spur a dramatic increase in use of an already problematic herbicide, exacerbate the weed problem by escalating the emergence and spread of resistant weeds, further trapping farmers on an out---of---control pesticide treadmill, and pushing many struggling family farmers out of business. This trajectory represents the polar opposite of the direction that American farming should be headed, namely that of ecologically---based, biodiversified, resilient farming that relies on least---toxic ecological approaches to insect and weed pest management. Economic harm from loss of inter---state and global commerce Economic harm due to crop damage and product loss caused by dicamba drift has been discussed above. Organic farmers whose crops are drifted on by dicamba face the additional possibility of losing organic certification of their crops. The absence of established tolerances for dicamba on many fruit and vegetable crops also threatens interstate commerce in these crops. This exposes specialty crop growers to risk of enforcement action by FDA, since interstate commerce is prohibited for produce lacking tolerances or exemptions. These enforcement actions could include crop confiscation and destruction, with the economic loss -- whether due to crop destruction or simply to loss of market value -- borne by the specialty crop growers themselves. Finally, conventional soybean and cotton growers may find themselves under extreme pressure to buy Monsanto's dicamba---resistant varieties, so that their own crops are not destroyed by dicamba drift. Those who have been exporting clean, non---GE soybean and cotton product to non---GE markets in Europe or Japan may find themselves unable to maintain their non---GMO production due to dicamba drift damage. The loss of these export markets will be devastating to their businesses. Health harms to farmers, workers and rural communities Adverse health effects associated with exposure to dicamba provide additional reason to reject Monsanto's proposed uses. Epidemiology studies have linked dicamba to increased rates of cancer--including non--- Hodgkin's lymphoma and multiple myeloma-- in pesticide applicators and farmers.3 Preconception exposure to dicamba has also been linked to increased risk of birth defects in farmers' male offspring, in the Ontario Farm Family Health Study.4 Dicamba has also been listed in the U.S. Toxic Release Inventory as a developmental toxin. Because dicamba has moderate persistence in the environment, is frequently detected in surface waters, and is expected to be applied more frequently throughout the growing season, the general population will also likely be more frequently exposed to dicamba than under current practice, rendering the increased risk of adverse health effects wholly unacceptable. 3 Schinasi, L and M. Leon, 2014. Non---Hodgkin lymphoma and occupational exposure to agricultural pesticide chemical groups and active ingredients: a systematic review and meta---analysis. Int J Environ Res Public Health. 2014 Apr 23;11(4):4449---527. doi: 10.3390/ijerph110404449. 4 Arbuckle, T., Z. Lin and L. Mery, 2001. An exploratory analysis of the effect of pesticide exposure on the risk of spontaneous abortion in an Ontario farm population. Environ Health Perspect. 2001 Aug; 109(8): 851-857. Conclusion In sum, we call on EPA to reject Monsanto's petition for new uses of dicamba on genetically engineered dicamba---resistant crops. EPA must protect the public against severe harms that would be exacerbated by continued and increased use of dicamba in these cropping systems. These harms include: o o o o economic harms to farmers' businesses and livelihoods caused by dicamba drift damage to vulnerable crops as well as crop loss, the cost of managing spread of intractable dicamba---resistant weeds, the emergence of dicamba---resistant soybean and cotton plants as noxious weeds themselves, restrictions on inter---state commerce, loss of organic certification for drift---damaged organic farmers and loss of access to valuable export markets; environmental harm from increased dicamba application accompanying the planting of dicamba--- resistant cotton and soybean, including reduction in farm--- and landscape---scale plant diversity that provide alternative income sources as well as protection from wind and water erosion; loss of habitat and food and reproductive resources for birds, beneficial arthropods and other species; and loss of critical ecosystem services such as pollination and natural pest control; health harm from exposure of pesticide applicators, farmers and rural communities to dicamba, including potential increased risks of cancers such as non---Hodgkin's lymphoma and multiple myeloma, birth defects and developmental toxicity; and socio---cultural harm to rural communities arising from increased conflict between neighboring farmers around issues of drift, crop damage and liability. We therefore urge EPA to prioritize the public interest and reject Monsanto's registration petition for use of dicamba on dicamba---resistant crops. Thank you for your consideration. Sincerely, Marcia Ishii---Eiteman, PhD Senior Scientist (224 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 91 of 245 Weed Technology. 2004. Volume 18:1125-1134 Education/ Exten s ion iiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiii• Determining Exposure to Auxin-Like Herbicides. I. Quantifying Injury to Cotton and Soybean 1 AUDIE S. SCIUMBATO, JAMES M. CHANDLER, SCOTT A. SENSEMAN, RODNEY W. BOVEY, and KEN L. SMITIF Abstract: Crop injury caused by drift of auxin-like herbicides has been a concem since their development. Research was conducted to desctibe a method of quantifying injury from auxin-like herbicides as a first step in detemlining crop damage. Reduced rates of 2,4-D, dicamba, and uiclopyr were applied to cotton and soybean plants in the three- to six-leaf stage in field and greenhouse studies. Injury to leaves and stems were evaluated separately, and the values were combined so that one injury estimate was obtained for each individual plant rated. Injury symptoms were typical for auxin-type herbicides and ranged fi·om slight bending of stems or petioles and wtinkled leaves to necrosis. Specific desctiptions of leaf and stem injury levels were used to describe plant injury consistently. These descriptions were very detailed for the lower injury levels, but the charactetizations became more general as the injury increased because of the prominence of factors such as necrosis. The injury evaluation method provided repeatable results for each herbicide and herbicide rate used. This injury evaluation method has many possible uses in auxin-like herbicide research and lays the foundation for forecasting the impact of early-season injury to cotton and soybean yield. Nomenclature: 2,4-D; dicamba; tiiclopyr; cotton, Gossypium hirsutum L. 'Delta Pine 50' #3 GOSHI; soybean, Glycine max (L.) 'Delta Pine 415' Men. # GLYMA. Additional index words: Epinasty, method, plant injury, rating scale. Abbreviation: DAT, days after tt·eatJnent. INTRODUCTION Auxin-like herbicides have been well received by producers since their inu·oduction in the 1940s. The first auxin-like herbicide, 2,4-D, was introduced commercially in 1945 (Peterson 1967). In 1996, 2,4-D alone was used on more than 40 crops (Bumside et al. 1996). Although agiicultural diversion programs and reduced rate recommendations have decreased herbicide use in the United States since the early 1980s (Bumside 1993), auxin-like herbicides still accounted for approximately 10% of all herbicides used (Bumside 1996). These herbicides are popular for many reasons, including their relatively low cost when compared with newer herbicides 1 Received for publication April9, 2003, and in revised form July 24, 2003. Research Associate, Professor, and Associate Professor, respectively, Texas Agricultural Experiment Station, Department of Soil and Crop Sciences, College Station, TX 77843-2474; Professor, Texas Agricultural Experiment Station, Department of Rangeland Ecology and Management, College Station, TX 77843-2126; Extension Weed Specialist, University of Arkansas-Monticello, Monticello, AR 71656. Corresponding author's E-mail: audie@tamu.edu. 3 Letters following tltis symbol are WSSA-approved computer code from Composite List of Weeds, Revised 1989. Available only on computer disk from WSSA, 810 East lOth Street, Lawrence, KS 66044-8897. 2 (Mississippi Cooperative Extension Service 1997). In addition to being inexpensive, the auxin-like herbicides pose minimal health hazards when applied properly (Bovey and Yotmg 1980; Council for Agricultural Science and Technology 1975; Fletcher and Kirkwood 1982). Although popular, auxin-like herbicides have long been scmtinized because of injury to susceptible off-target plants caused by diift (Arle 1954; Bovey and Meyer 1981; Miller et al. 1963). Drift concems have led some states to enact numerous safeguards such as the use of low-volatility f01mulations, restiictions on application times, or bruming auxin-like herbicide use (Texas Agiiculture Code 1984). Despite these eff01t s, crop injury from auxin-like herbicides still occurs. Although the auxin-like herbicides represent the oldest organic herbicide mode of action in use today, a numerical system has not been devised to describe the symptoms that occur with plant injury. The objective of this study was to devise a method of quantifying plant injury from auxin-like herbicide exposure. Such a system must be specific to these herbicides because of the unique in- 1125 ER 200 (225 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 92 of 245 SCIUMBATO ET AL.: DETERMINING AUXIN-LIKE HERBICIDE EXPOSURE Table 1. Reduced rates of 2,4-D, dicamba, and triclopyr applied to cotton and soybean plants at the four- to six-leaf stage. 2,4-D Rate Standard 4 ϫ 10Ϫ1 a 2 ϫ 10Ϫ1 a 1 ϫ 10Ϫ1 5 ϫ 10Ϫ2 1 ϫ 10Ϫ2 5 ϫ 10Ϫ3 1 ϫ 10Ϫ3 5 ϫ 10Ϫ4 1 ϫ 10Ϫ4 5 ϫ 10Ϫ5 b 1 ϫ 10Ϫ5 b a b Dicamba Concentration Use rate ␮g/mL 2,850 1,140 570 285 142 28 14 2.8 1.4 0.2 0.14 0.02 Triclopyr Concentration Use rate Concentration Use rate kg/ha ␮g/mL 0.53 2.1 ϫ 10Ϫ1 1.1 ϫ 10Ϫ1 5.3 ϫ 10Ϫ2 2.7 ϫ 10Ϫ2 5.3 ϫ 10Ϫ3 2.7 ϫ 10Ϫ3 5.3 ϫ 10Ϫ4 2.7 ϫ 10Ϫ4 5.3 ϫ 10Ϫ5 2.7 ϫ 10Ϫ5 5.3 ϫ 10Ϫ6 3,000 1,200 600 300 150 30 15 3 1.5 0.3 0.15 0.03 kg/ha ␮g/mL kg/ha 0.56 2.2 ϫ 10Ϫ1 1.1 ϫ 10Ϫ1 5.6 ϫ 10Ϫ2 2.8 ϫ 10Ϫ2 5.6 ϫ 10Ϫ3 2.8 ϫ 10Ϫ3 5.6 ϫ 10Ϫ4 2.8 ϫ 10Ϫ4 5.6 ϫ 10Ϫ5 2.8 ϫ 10Ϫ5 5.6 ϫ 10Ϫ6 6,000 2,400 1,200 600 300 60 30 6 3 0.6 0.3 0.06 4.5 2.2 1.1 5.6 1.1 5.6 1.1 5.6 1.1 5.6 1.1 1.12 ϫ 10Ϫ1 ϫ 10Ϫ1 ϫ 10Ϫ1 ϫ 10Ϫ2 ϫ 10Ϫ2 ϫ 10Ϫ3 ϫ 10Ϫ3 ϫ 10Ϫ4 ϫ 10Ϫ4 ϫ 10Ϫ5 ϫ 10Ϫ5 Applied to cotton only. Applied to soybean only. jury symptoms associated. In addition, it could be an important first step in determining crop damage severity and forecasting potential yield losses. To be useful, this scale must be clearly defined with well-described injury features along with being user friendly so that a variety of users could use it. MATERIALS AND METHODS The rating scale was constructed by observing and categorizing injury to test plants that had been exposed to reduced rates of auxin-like herbicides. The procedure was derived from both greenhouse and field experiments using cotton and soybean with four to six leaves. Auxinlike herbicides used included the dimethylamine salt of 2,4-D,4 the diglycolamine salt of dicamba,5 and the butoxyethyl ester of triclopyr.6 Visual estimates of plant injury were recorded at 1, 5, 9, and 14 d after treatment (DAT) in all experiments, although only data from the 14 DAT observations will be presented. Greenhouse Injury Evaluation Procedure. The greenhouse procedure was performed at the Norman E. Borlaug Center for Southern Crop Improvement on the campus of Texas A&M University during the fall of 1996 and spring of 1997. Each repetition consisted of six replications for each herbicide treatment for both cotton and soybean. The experimental design was completely randomized and was performed twice. ‘Delta Pine 50’7 cotton and ‘Delta Pine 415’7 soybean 4 Weedar 64௡ herbicide, Nufarm Limited, 103 Pipe Road, Laverton, North Victoria 3026, Australia. 5 Clarity௡ herbicide, BASF Corporation, Agricultural Products Group, Research Triangle Park, NC 27709. 6 Remedy௡ herbicide, Dow AgroSciences, Indianapolis, IN 46268-1189. 7 Delta and Pine Land Company, P.O. Box 157, Scott, MS 38772. seeds were planted in standard 15-cm-diam by 15-cmdeep plastic pots at populations of three plants per pot. Seeds were planted at a depth of 4 cm into growth medium composed of a 3:1 (v/v) mixture of Pro-Mix8 and Redi-Earth9 potting soils. Greenhouse conditions were 10 h of darkness at 23 C (Ϯ3 C) and 14 h of light at 29 C (Ϯ3 C). One application of N–P2O5–K2O (20:20:20) fertilizer was made to the young plants 1 wk after emergence at a rate equivalent to 23 kg/ha for N, P, and K. Nine different rates of 2,4-D, dicamba, and triclopyr were applied to the cotton and soybean plants at the fourto six-leaf stage, approximately 21 d after planting (Table 1). The herbicides were applied without surfactant in a 187 L/ha spray volume using a research track sprayer.10 The track sprayer was washed thoroughly with water before treatments of each herbicide were applied. The nine rates that were applied ranged from 4 ϫ 10Ϫ1 to 1 ϫ 10Ϫ5 times the recommended use rates of 2,4-D, dicamba, and triclopyr. The respective use rates for these herbicides were 0.53 kg ai/ha, 0.56 kg/ha, and 1.12 kg/ ha. These use rates represented concentrations of 2,850 ␮g/ml, 3,000 ␮g/ml, and 6,000 ␮g/ml for 2,4-D, dicamba, and triclopyr, respectively. Herbicide rates were similar for both plant species, but the two highest doses (4 ϫ 10Ϫ1 and 2 ϫ 10Ϫ1 times the field rates) were only used on cotton. In addition, the two lowest doses (5 ϫ 10Ϫ5 and 1 ϫ 10Ϫ5 times the field rates) were only applied to soybean. The low soybean rates were included after the highest rates of dicamba and triclopyr quickly killed the soybean test plants during preliminary trials. Potting soil, Premier Horticulture Inc., Red Hill, PA 18076. Potting soil, Scotts-Sierra Horticultural Products Company, 14111 Scottslawn Road, Marysville, OH 43041. 10 Spray chamber, De Vries Manufacturing, Hollandale, MN 56045. 8 9 1126 Volume 18, Issue 4 (October–December) 2004 ER 201 (226 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 93 of 245 WEED TECHNOLOGY Table 2. Leaf injury scale for injury evaluation on a 0-100 basis. Rating' 1- 3 4- 6 7-1 0 11- 15 16-20 21- 30 31- 40 41- 100 a Observed symptomology Slight ripples in leaf margin creating "drawstring" effect. Isolated wart-like growths may also be observed on the upper epidermis of the leaf Ripples are more pronounced and present on at least 50% of leaf margin_ Wart-like growths on leaf are common. Leaves appear stiff and brittle. Stiff areas may also display chlorosis. Necrosis begins to appear around leaf margin which is uneven and gnarled throughout its perimeterc Chlorosis is more obvious. Necrosis becomes more prominent and has affected no more than l 0% of the total new leaf area. Necrosis is the dominant factor affecting up to 20% of the young leaves. Epinasty is severe around the entire leaf margin_ Up to 30% of the leaf tissue is necrotic, with chlorosis apparent through much of the leaf perimeter. The leaf is very disfigured and chlorotic. Necrosis is evident on up to 40% of the leaf. Necrosis becomes the primary indicator of plant injury. Although epinasty is extreme throughout the leaves, chlorosis and necrosis coverage is dominant. Figure 1. Cotton leaf injury value of 5. The higher the value, the more severe the injury. The plants were retumed to an adequately ventilated greenhouse after the herbicides were applied, and injury symptoms were recorded using the rating scale. Field Injury Evaluation Procedure. The field procedme was canied out at the Texas A&M Agronomy Field Laboratory near College Station during June, July, and August of 1997. The expe1imental design was a snipsplit plot and was conducted four times during the 3-mo period. Replication in time was done to evaluate potential environmental impact on symptomology. Varieties used were Delta Pine 50 cotton and Delta Pine 415 soybean. Soil at the test site was a Belk clay (Entic Hapluderts). Four-row snips of the cotton and soybean varie- ties were planted 4 em deep on 102-cm centers for each repetition using a fom-row John Deere planter equipped with insecticide application boxes. Each experiment consisted of a fom-row strip of cotton and a four-row snip of soybean. The first replication was planted on Jtme 16, 1997, the second on June 30, 1997, the third on July 11, 1997, and the fourth on July 17, 1997. The ftmgicide metalaxyP 1 [N-(2,6-climethylphenyl)-N-(methoxyacetyl) alanine methyl ester] and the insecticide phorate12 { 0 , 0-cliethyl S-[(ethylth.io) methyl] phosphorodithioate} were applied 11 Ridomi1® fungicide, Syngenta Crop Protection, Inc. P.O. Box 18300, Greensboro, NC 27409. 12 Thimet* insecticide, BASF Corporation, Agricultural Products Group, Research Triangle Park, NC 27709. Table 3. Stem injury scale for categorical injury evaluation on a 0-100 basis. Rating' 1- 3 4-6 7- 10 11- 15 16-20 21- 30 31- 40 41- 100 a Observed symptomology Slight swelling of the stem or vascular tissues with no observed twisting or abnormal bending. Twisting and abnormal bending becoming apparent near the apical meristem, with stem bending less than 6". Leaf petioles may be slightly languid but are not horizontal. Twisting and bending in the main stem and branches range from 5 to 45°. Stem discoloration may also become apparent. Petioles have dropped no farther than horizontal. Stems or branches are bent in angles between 45 and 90°. Stem may appear red and swollen.. Petioles appear weak and have fallen beyond horizontal. Epinastic symptoms are intense with necrosis in the meristematic regions. Malformation angles range from 90 to 120°. Petioles have fallen as much as 90° beyond horizontal. Sterns or petioles are twisted at angles of 120-150°, with up to 30% dead tissue. Stem or petioles nvisted in excess of 150°. Up to 40% of the plant tissue is necrotic. Necrosis is now the primary indicator of injury to the stem, with over 40% of the tissue being necrotic. The higher the value, the more severe the injury. Figure 2. Cotton leaf injury value of 10. 1127 Volume 18, Issue 4 (October-December) 2004 ER 202 (227 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 94 of 245 SCIUMBATO ET AL.: DETERMINING AUXIN-LIKE HERBICIDE EXPOSURE Figure 3. Cotton leaf injury value of 15. Figure 5. Cotton stem injury value of 5. dming planting to protect the seeds and emerging seedlings. A preemergence application of pendimethalinl3 was also applied at 0.47 kg ai/ha to control grasses. The studies we.re furrow irrigated as needed, and tillage or hand hoeing was done to provide postemergence weed control. Plots were ananged in snips to minimize the risk of spray droplet or vapor drift from one plot to another. Using this configuration, contamination could only be spread from plot to plot by being moved lengthwise along the species sni ps. The two center rows of each fom-row snip were used to comprise the actual test plots, 13 Prow!® herbicide, BASF Corporation, Agricultural Products Group, Research Tnangle Park, NC 27709. Figure 4. Cotton leaf injury value of 20. whereas the outer two rows were used as buffer rows. In addition, fom rows separated the cotton and soybean sn·ips. The cotton and soybean rows were divided lengthwise into three. equal parts, one pa1t for each of the three herbicides to be used. Thirty 1.5-m-long test plots, 10 plots for each of the three herbicides used, were marked and separated by 1.2-m alleys in each species sni p. The plants were treated with 2,4-D, tdclopYI~ and dicamba using a C02-pressmized backpack sprayer and handheld spray boom when each replication reached the fom- to six-leaf stage. All herbicides were delivered in a 187 Llha spray solution, and the herbicide rates used in the field were the same as those used in the greenhouse injmy evaluation process (Table 1). Smfactants were not included in the spray solutions. Figure 6. Cotton stem injury value of 10. 1128 Volume 18, Issue 4 (October-December) 2004 ER 203 (228 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 95 of 245 WEED TECHNOLOGY Figure 9. Soybean leaf injury value of 5. Figure 7. Cotton stem injury value of 15. Evaluation. The scale used to quantify the injury associated with these herbicides was based on a 0 to 100 scale, with 100 representing plant death (Tables 2 and 3). Dming the evaluation process, leaves and stems were monitored separately on each plant for the effects of auxin-like herbicides. Petioles and branches were combined with the main stem and considered together as stem injury, whereas the leaves comprised leaf injury data. Injmy that was observed on leaves was documented using the scale summarized in Table 2. Symptomology and values used for stem injury evaluations are located in Table 3. Figures 1- 4 and 5-8 depict typical cotton leaf and stem injury, respectively. Soybean leaf and stem injmy is shown in Figmes 9- 12 and 13-16, respectively. The categorical ranges were used to provide general injury values from which the final evaluation would be determined. During the evaluation procedure, broad generalizations of a plant's leaves or stem appearance were Figure 8. Cotton stem injury value of 20. Figure 10. Soybean leaf injury value of 10. The herbicides were applied to the first, second, third, and fomth replications on July 7, July 24, August 4, and August 11, respectively. During the spray applications, 1.8- by 2.4-m plastic tarps were held armmd the plot being treated to reduce the lisk of spray drift particles contaminating other plots. After the herbicides we.re applied, five random test plants in each plot were chosen and rated at 1, 5, 9, and 14 DAT using the leaf and stem injury scale described later. The selected plants were marked with plastic garden stakes so that the same plants were rated each time. 1129 Volume 18, Issue 4 (October-December) 2004 ER 204 (229 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 96 of 245 SCIUMBATO ET AL.: DETERMINING AUXIN-LIKE HERBICIDE EXPOSURE Figure 11. Soybean leaf injury value of 15. Figure 13. Soybean stem injury value of 5. considered so that a basic range of injury could be detemlined. The final injury value was assigned after this approximation had been made based on the level of injury within that range. The leaf and stem injury evaluations were then averaged so that one injury value described the combined leaf and stem injmy of the plant. Mean standard enors were calculated for each treatment to detenuine variability within the evaluations. RESULTS AND DISCUSSION rating scale because leaf and stem injury from each plant was averaged together. Soybean injury appeared more consistent across leaves and stems, pruticularly in the :field. Howeve1~ severe stem injury was observed on soybean exposed to the highest triclopyr rates in the greenhouse. In addition, epinasty was easier to quantify on cotton plants because of the larger leaf size and longe1: more prominent stems. The lru·gest standru·d enor recorded for the evaluations throughout the experiment was 1.62, indicating good precision. Evaluation Scale. The majmity of the injury recorded from cotton was observed on leaves in both the field and greenhouse. This vatiability between leaf and stem injury within plants did little to affect the results of the Greenhouse. Average injury values and their standa.J·d enors ru·e shown in Tables 4-7. The overall injury evaluations of2,4-D varied from 10.3 to 1.9 for cotton (Table 4) and 3.8 to 2.4 for soybean (Table 5). The average Figure 12. Soybean leaf injury value of 20. Figure 14. Soybean stem injury value of 10. 1130 Volume 18, Issue 4 (October-December) 2004 ER 205 (230 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 97 of 245 WEED TECHNOLOGY 00~(',1 \O"Ct(',l "'d" V'l \0 (""') ..q- V'l oo...: coo 7 0 X I 0 X .-... ...-...-.. ...-.....-..,-.. r----V'I - \ 0 0 \ f""'N- N N - eeeeee ~~~ ... OOOO"Ct 0 X ~~~ ~t:":rl' ~~~ M~- -NN \0 ....... - ~ " :§" c 0 = 8 ~ ::; 0 ..c c ~ 01) c .., 01) ~ c ·~ ] > Figure 16. Soybean stem injury value of 20. 0 1131 Volume 18, Issue 4 (October-December) 2004 ER 206 (231 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 98 of 245 1132 Table 5. Greenhouse soybean injury evaluations made 14 d after treatments with 2,4-D, dicamba, or triclopyr. Herbicide rate (ϫ)a Injury category Herbicide Leaf rating 2,4-D Dicamba Triclopyr 2,4-D Dicamba Triclopyr 2,4-D Dicamba Triclopyr 1 ϫ 10Ϫ1 5 ϫ 10Ϫ2 1 ϫ 10Ϫ2 5 ϫ 10Ϫ3 1 ϫ 10Ϫ3 5 ϫ 10Ϫ4 1 ϫ 10Ϫ4 5 ϫ 10Ϫ5 1 ϫ 10Ϫ5 2.8 6.5 3.0 4.1 3.5 4.7 3.0 3.1 3.2 3.2 2.2 3.9 3.6 3.2 4.1 3.3 1.8 4.1 5.4 3.4 3.5 2.1 1.5 1.2 Average standard error average % injury Overall rating a b (0.26)b (0.51) (0.91) (0.84) (0.93) (0.86) 3.5 11.5 18.5 3.0 16.3 13.2 3.8 5.4 18.9 (0.17) (0.79) (0.98) (0.87) (0.90) (0.92) 3.4 10.9 16.0 2.9 14.3 9.4 4.4 4.7 5.2 (0.21) (0.76) (1.09) (1.39) (1.04) (1.30) 3.7 9.5 7.3 3.6 10.8 5.3 1.2 5.2 3.4 (0.21) (0.61) (0.54) (0.37) (1.24) (0.86) 2.4 8.1 4.4 3.5 8.7 3.8 3.4 4.7 4.4 (0.40) (0.99) (0.38) (1.14) (1.46) (1.34) 3.5 6.7 4.1 (0.22) (1.13) (0.12) (1.25) (1.09) (1.48) 3.4 5.0 3.8 (0.37) (0.23) (0.22) (1.04) (0.72) (1.28) 3.1 2.7 3.6 (0.62) (0.30) (0.75) (1.02) (0.61) (1.33) 3.5 2.5 4.1 (0.78) (0.29) (0.47) (0.72) (0.48) (0.51) 3.8 2.5 2.3 0.36 0.62 0.61 0.96 0.94 1.10 Rate is calculated from general use rates of 0.53 kg/ha of 2,4-D, 0.56 kg/ha of dicamba, and 1.12 kg/ha of triclopyr. Standard error of average percent injury. Table 6. Field cotton injury evaluations made 14 d after treatments with 2,4-D, dicamba, or triclopyr. Herbicide rate (ϫ)a Volume 18, Issue 4 (October–December) 2004 Injury category Herbicide Leaf rating 2,4-D Dicamba Triclopyr 2,4-D Dicamba Triclopyr 2,4-D Dicamba Triclopyr 4 ϫ 10 Ϫ1 2 ϫ 10 Ϫ1 1 ϫ 10 Ϫ1 5 ϫ 10 Ϫ2 1 ϫ 10Ϫ2 5 ϫ 10Ϫ3 1 ϫ 10Ϫ3 5 ϫ 10Ϫ4 1 ϫ 10Ϫ4 3.7 5.5 2.1 2.2 3.1 1.3 2.2 2.9 2.0 2.0 2.4 1.5 1.6 1.5 1.7 1.1 1.9 1.2 1.4 1.4 1.5 1.5 0.7 1.1 Average standard error average % injury Stem rating Overall rating a b 10.5 22.9 12.5 7.2 6.0 2.1 (0.46)b (1.10) (2.97) (1.27) (0.91) (0.38) 8.9 14.5 7.3 10.1 18.9 5.4 5.5 4.0 2.0 (0.33) (0.57) (1.66) (0.72) (0.48) (0.40) 7.8 11.5 3.7 5.4 15.9 4.8 3.9 4.4 2.0 (1.33) (0.67) (1.38) (0.77) (0.54) (0.44) 6.1 10.1 3.4 10.9 13.5 2.9 3.3 3.2 2.5 (0.78) (0.32) (0.69) (0.60) (0.48) (0.44) 7.1 8.4 2.7 9.0 9.3 2.1 2.3 2.7 1.8 (1.12) (0.79) (0.46) (0.42) (0.47) (0.44) 5.6 6.0 1.9 Rate is calculated from general use rates of 0.53 kg/ha of 2,4-D, 0.56 kg/ha of dicamba, and 1.12 kg/ha of triclopyr. Standard error of average percent injury. ER 207 (0.75) (0.97) (0.44) (0.28) (0.59) (1.40) 2.9 4.3 1.7 (0.40) (0.41) (0.40) (0.34) (0.42) (0.43) 2.1 2.7 1.8 (0.35) (0.36) (0.41) (0.37) (0.36) (0.43) 1.3 1.7 1.5 (0.30) (0.32) (0.34) (0.41) (0.33) (0.32) 1.4 1.1 1.3 0.65 0.61 0.97 0.58 0.51 0.41 SCIUMBATO ET AL.: DETERMINING AUXIN-LIKE HERBICIDE EXPOSURE Stem rating 3.4 16.3 14.9 3.6 6.6 22.2 (232 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 99 of 245 (0.43) (0.36) (0.35) (0.32) (0.30) (0.35) 1.6 1.4 2.2 1.5 (0.27) 1.9 (0.28) 1.7 (0.19) 1.2 (0.31) 1.1 (0.33) 0.7 (0.27) 1.41 1.5 1.2 2.1 1.5 2.7 1.0 1.3 1.7 1 ϫ 10Ϫ5 (0.24) (0.27) (0.15) (0.41) (0.20) (0.16) 1.5 1.2 1.3 5 ϫ 10Ϫ5 (0.25) (0.50) (0.22) (0.26) (0.25) (0.18) 1.9 2.0 1.5 1.8 1.4 1.7 1.3 1.0 0.9 1 ϫ 10Ϫ4 Rate is calculated from general use rates of 0.53 kg/ha of 2,4-D, 0.56 kg/ha of dicamba, and 1.12 kg/ha of triclopyr. Standard error of average percent injury. Field. Average evaluations and standard error values for field injury are displayed in Tables 6 and 7. The overall evaluations for 2,4-D ranged from 8.9 to 1.3 for cotton (Table 6) and 3.6 to 1.4 for soybean (Table 7). The average standard errors of 2,4-D evaluations were 0.65 and 0.58 for cotton leaf and stem injury, respectively. Those of soybean leaf and stem evaluations were 0.28 and 0.36, respectively. Injury evaluations for dicamba were again larger than those of 2,4-D, but the variability was not increased. The average standard errors for dicamba injury to cotton leaf and stem were 0.61 and 0.51, and 0.56 and 0.57 for soybean leaf and stem evaluations, respectively. The overall triclopyr injury ranged from 7.3 to 1.3 on cotton and 5.9 to 1.2 on soybean. Mean standard errors for triclopyr evaluations in the field were less than those observed in the greenhouse, although the largest mean standard error value from the field data was that of triclopyr on cotton leaf. This rating system provided consistent, repeatable results when evaluating injury on cotton and soybean, although cotton was the easier species to evaluate because of the petiole size and leaf characteristics. The system was effective in determining injury severity across both species and for all three of the auxin-like herbicides used. Injury recorded for minute levels of different herbicides were similar yet showed increased injury values with increased herbicide rates. The evaluation procedure seemed particularly effective in describing injury caused by extremely small amounts of auxin-like herbicides. The injury features outlined in this scale make it possible to quantify epinasty without the presence of necrosis or even chlorosis by determining severity of injury based on physical malformations displayed by the injured plant. This property is valuable because it is these low injury levels that are not only the most difficult to quantify but also those that are commonly encountered in cases of drift damage to cash crops. To be useful to producers, additional research featuring the use of this scale to evaluate early-season injury and its relation to plant yield is necessary. ACKNOWLEDGMENTS The authors would like to thank Curtis Jones, Greg Steele, Justin Scott, and Chris Tingle for their help in conducting this research. LITERATURE CITED b Overall rating Stem rating a (0.27) (0.54) (0.29) (0.53) (0.34) (0.27) 2.1 2.8 1.7 1.7 3.5 2.1 2.5 2.1 1.3 (0.21) (0.63) (0.33) (0.25) (0.31) (0.23) 1.5 4.3 1.3 1.6 5.4 1.9 1.4 3.1 0.7 (0.26) (0.72) (0.30) (0.34) (0.44) (0.40) 1.9 6.2 2.3 1.8 7.5 2.3 1.9 1.9 2.3 (0.26) (0.80) (0.85) (0.31) (1.24) (0.77) 1.9 8.0 4.1 2.0 7.7 4.1 1.9 8.2 4.0 (0.35)b (1.0) (0.94) (0.52) (1.72) (1.26) 3.6 9.3 5.9 2.7 9.0 4.6 4.5 9.7 1.7 Leaf rating 2,4-D Dicamba Triclopyr 2,4-D Dicamba Triclopyr 2,4-D Dicamba Triclopyr average % injury 1.9 2.7 1.9 1.8 1.4 1.1 5 ϫ 10Ϫ4 1 ϫ 10 5 ϫ 10 1 ϫ 10 Herbicide Injury category Herbicide rate (ϫ)a 1 ϫ 10Ϫ3 5 ϫ 10 Ϫ3 Ϫ2 Ϫ2 Ϫ1 Table 7. Field soybean injury evaluations made 14 d after treatments with 2,4-D, dicamba, or triclopyr. obtained with 2,4-D on cotton in both leaf and stem categories. 0.28 0.56 0.40 0.36 0.57 0.43 Average standard error WEED TECHNOLOGY Arle, H. F. 1954. The sensitivity of Acala 44 cotton to 2,4-D. West. Weed Control Conf. Proc. 4:20–25. 1133 Volume 18, Issue 4 (October–December) 2004 ER 208 (233 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 100 of 245 SCIUMBATO ET AL.: DETERMINING AUXIN-LIKE HERBICIDE EXPOSURE Bovey, R. W. and R. E. Meyer. 1981. Effects of 2,4,5-T, triclopyr and 3,6dichloropicolinic acid on crop seedling. Weed Sci. 29:256–261. Bovey, R. W. and A. L. Young. 1980. The Science of 2,4,5-T and Associated Phenoxy Herbicides. New York: J. Wiley. Chapters 1, 7, 11, 13, and 14. Burnside, O. C. 1993. Weed science—the step child. Weed Technol. 7:515– 518. Burnside, O. C. 1996. The history of 2,4-D and its impact on development of the discipline of weed science in the United States. In O. C. Burnside, ed. Biologic and Economic Assessment of Benefits from Use of Phenoxy Herbicides in the United States. Special NAPIAP Rep. 1-PA-96. Pp. 5–15. Burnside, O. C., R. W. Bovey, C. L. Elmore, E. L. Knake, C. A. Lembi, J. D. Nalewaja, M. Newton, and P. Szmedra. 1996. Biologic and economic assessment of benefits from use of phenoxy herbicides in the United States. Weed Sci. Soc. Am. Abstr. 36:36. Council for Agricultural Science and Technology. 1975. The Phenoxy Herbicides. Ames, IA: Council for Agricultural Science and Technology Rep. 39. 22 p. Fletcher, W. W. and R. C. Kirkwood. 1982. Herbicides and Plant Growth Regulators. London: Granada. Pp. 119–121. Miller, J. H., H. M. Kempen, J. A. Wilderson, and C. L. Fox. 1963. Response of Cotton to 2,4-D and Related Phenoxy Herbicides. USDA Technical Bull. 1289. Washington, DC: U.S. Government Printing Office. Mississippi Cooperative Extension Service. 1997. 1997 Weed Control Guidelines for Mississippi. Publication No. 1532. Mississippi State, MS: Mississippi State University. Peterson, G. E. 1967. The discovery and development of 2,4-D. Agric. Hist. 41:243–253. Texas Agriculture Code. 1984. St. Paul, MN: West. Chapter 75. Composite List of Weeds—1989 Available on computer disk in ASCII format The names of 2,076 weed species of current or potential importance in the United States and Canada are arranged by scientific name, by WSSA approved common name and by five-letter ‘‘Bayer code.’’ The Composite List of Weeds—1989 is available only on computer disk. Computer disk priced at US $10.00 (remittance to accompany order). (Shipping charge: $3.50 for first copy; $0.75 for each additional copy.) Ship to: Name Address City State Zip Code Credit Card Payment: Amount $ VISA # Mastercard # Expiration Signature Make checks payable to Weed Science Society of America and mail to P.O. Box 7050, 810 East 10th St., Lawrence, KS 66044-8897. Ph: (800) 627-0629 (U.S. and Canada), (785) 843-1235; Fax: (785) 843-1274; E-mail: wssa@allenpress.com. WETE/master_fillers filler07 1134 Volume 18, Issue 4 (October–December) 2004 ER 209 (234 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 101 of 245 Weed Technology. 2004. Volume 18:1135–1142 Determining Exposure to Auxin-Like Herbicides. II. Practical Application to Quantify Volatility1 AUDIE S. SCIUMBATO, JAMES M. CHANDLER, SCOTT A. SENSEMAN, RODNEY W. BOVEY, and KEN L. SMITH2 Abstract: Volatility and drift are problems commonly associated with auxin-like herbicides. Field and greenhouse studies were conducted at Texas A & M University to develop a method of quantifying volatility and subsequent off-target movement of 2,4-D, dicamba, and triclopyr. Rate–response curves were established by applying reduced rates ranging from 4 ϫ 10Ϫ1 to 1 ϫ 10Ϫ5 times the normal use rates of the herbicides to cotton and soybean and recording injury for 14 d after treatment (DAT) using a rating scale designed to quantify auxin-like herbicide injury. Injury from herbicide volatility was then produced on additional cotton and soybean plants through exposure to vapors of the dimethylamine salt of 2,4-D, diglycolamine salt of dicamba, and butoxyethyl ester of triclopyr using air chambers inside a greenhouse and volatility plots in the field. Injury resulting from this exposure was evaluated for 14 d using the same injury-evaluation scale that was used to produce the rate–response curves. Volatility-injury data were then applied to the rate–response curves so that herbicide rates corresponding with observed injury could be calculated. Using this method, herbicide volatility rates estimated from greenhouse-cotton injury were determined to be 3.0 ϫ 10Ϫ3, 1.0 ϫ 10Ϫ3, and 4.9 ϫ 10Ϫ2 times the use rates of 2,4-D, dicamba, and triclopyr, respectively. Greenhousegrown soybean developed injury consistent with 1.4 ϫ 10Ϫ2, 1.0 ϫ 10Ϫ3, and 2.5 ϫ 10Ϫ2 times the normal use rate of 2,4-D, dicamba, and triclopyr, respectively. Under field conditions, cotton developed injury symptoms that were consistent with 4.0 ϫ 10Ϫ3, 2.0 ϫ 10Ϫ3, and 1.25 ϫ 10Ϫ1 times the recommended use rates of 2,4-D, dicamba, and triclopyr, respectively. Field soybean displayed injury symptomology concordant with 1.6 ϫ 10Ϫ1, 1.0 ϫ 10Ϫ2, and 1.1 ϫ 10Ϫ1 times the normal use rates of 2,4-D, dicamba, and triclopyr, respectively. This procedure provided herbicide volatility rate estimates that were consistent with rates and injury from the rate–response injury curves. Additional research is needed to ascertain its usefulness in determining long-term effects of drift injury on crop variables such as yield. Nomenclature: 2,4-D, dicamba, triclopyr, cotton, Gossypium hirsutum L. ‘Delta Pine 50’, #3 GOSHI, soybean, Glycine max (L.) Merr. ‘Delta Pine 415’, # GLYMA. Additional index words: Injury modeling, plant injury, rate of exposure. Abbreviations: BEE, butoxyethyl ester; DAT, days after treatment; DGA, diglycolamine; DMA, dimethylamine; WAE, weeks after emergence. INTRODUCTION Volatilization, a major cause of herbicide loss, has been associated with the removal of as much as 90% of 1 Received for publication October 1, 2003, and in revised form March 9, 2004. 2 Research Associate, Professor, and Associate Professor, respectively, Texas Agricultural Experiment Station, Department of Soil and Crop Sciences, College Station, TX 77843-2474; Professor, Texas Agricultural Experiment Station, Department of Rangeland Ecology and Management, College Station, TX 77843-2126; and Extension Weed Specialist, University of Arkansas– Monticello, Monticello, AR 71656. Corresponding author’s E-mail: audie@tamu.edu. 3 Letters following this symbol are WSSA-approved computer code from Composite List of Weeds, Revised 1989. Available only on computer disk from WSSA, 810 East 10th Street, Lawrence, KS 66044-8897. an applied herbicide (Taylor and Spencer 1990). It is not uncommon for volatilization and subsequent vapor drift of auxin-like herbicides to injure susceptible crops near areas of application (Anonymous 1975; Arle 1954; Behrens and Lueschen 1979). The effect of auxin-like herbicides on crops such as cotton is both destructive and well documented (Bovey and Meyer 1981), therefore the use of most auxin-like herbicides has been restricted in areas of broadleaf crop production. These restrictions vary by location, but stipulations on time and method of application, permit requirements, or chemical formulation may apply (Texas Agriculture Code 1984). Producers must determine the extent of crop damage 1135 ER 210 (235 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 102 of 245 SCIUMBATO ET AL.: QUANTIFYING AUXINLIKE HERBICIDE VOLATILITY Table 1. Reduced rates of 2,4-D, dicamba, and triclopyr applied to cotton and soybean plants at the four- to six-leaf stage. Herbicide Rate Standard 4 ϫ 10Ϫ1a 2 ϫ 10Ϫ1a 1 ϫ 10Ϫ1 5 ϫ 10Ϫ2 1 ϫ 10Ϫ2 5 ϫ 10Ϫ3 1 ϫ 10Ϫ3 5 ϫ 10Ϫ4 1 ϫ 10Ϫ4 5 ϫ 10Ϫ5b 1 ϫ 10Ϫ5b a b 2,4-D Dicamba Triclopyr ␮g/mL kg/ha ␮g/mL kg/ha ␮g/mL 2,850 1,140 570 285 142 28 14 2.8 1.4 0.2 0.14 0.02 0.53 2.1 ϫ 10Ϫ1 1.1 ϫ 10Ϫ1 5.3 ϫ 10Ϫ2 2.7 ϫ 10Ϫ2 5.3 ϫ 10Ϫ3 2.7 ϫ 10Ϫ3 5.3 ϫ 10Ϫ4 2.7 ϫ 10Ϫ4 5.3 ϫ 10Ϫ5 2.7 ϫ 10Ϫ5 5.3 ϫ 10Ϫ6 3,000 1,200 600 300 150 30 15 3 1.5 0.3 0.15 0.03 0.56 2.2 ϫ 10Ϫ1 1.1 ϫ 10Ϫ1 5.6 ϫ 10Ϫ2 2.8 ϫ 10Ϫ2 5.6 ϫ 10Ϫ3 2.8 ϫ 10Ϫ3 5.6 ϫ 10Ϫ4 2.8 ϫ 10Ϫ4 5.6 ϫ 10Ϫ5 2.8 ϫ 10Ϫ5 5.6 ϫ 10Ϫ6 6,000 2,400 1,200 600 300 60 30 6 3 0.6 0.3 0.06 kg/ha 4.5 2.2 1.1 5.6 1.1 5.6 1.1 5.6 1.1 5.6 1.1 1.12 ϫ 10Ϫ1 ϫ 10Ϫ1 ϫ 10Ϫ1 ϫ 10Ϫ2 ϫ 10Ϫ2 ϫ 10Ϫ3 ϫ 10Ϫ3 ϫ 10Ϫ4 ϫ 10Ϫ4 ϫ 10Ϫ5 ϫ 10Ϫ5 Applied to cotton only. Applied to soybean only. on the basis of injury symptoms alone and decide whether drastic action such as replanting is necessary after crops have been injured. The uncertainty of the longterm effects of these herbicides makes decisions based on early-season injury difficult (Miller et al. 1963). A method that uses early injury symptoms to determine the amount of auxin-like herbicide that a broadleaf plant has been exposed to would be helpful in forecasting the effect of drift on crop growth and yield. One approach to quantifying herbicide exposure is through modeling procedures that rely on data from plant injury. Such a system would be well suited for practical agronomic use because it could be applied as soon as crop injury is discovered during the growing season. Upon determining drift rates, producers could forecast probable crop damage and modify management strategies immediately where exposure is found to be at unacceptable rates. The objective of this research was to describe a method for estimating auxin-like herbicide exposure rates for different herbicides and plant species. MATERIALS AND METHODS This research was carried out in a three-step process that included (1) establishment of rate–response injury curves, (2) production of equations describing those curves, and (3) insertion of volatility-injury data into model equations to estimate exposure rates. Each process was performed on cotton and soybean under both field and greenhouse conditions. All 2,4-D used was the dimethylamine (DMA) salt,4 all dicamba was the digly4 Weedar 64௡ herbicide, Rhone-Poulenc Ag Company, Research Triangle Park, NC 27709. colamine (DGA) salt,5 and all triclopyr was the butoxyethyl ester (BEE).6 Rate–Response Injury Curve Establishment. Greenhouse. Greenhouse-injury curves were produced by applying nine reduced rates of 2,4-D, dicamba, and triclopyr to cotton and soybean plants. ‘Delta Pine 50’ cotton and ‘Delta Pine 415’ soybean plants7 were grown at the seeding rate of three plants per pot in a greenhouse using standard 15-cm plastic pots. The growth medium was a 3:1 (v/v) mixture of Pro-Mix8 and Redi-Earth9 potting soils. Environmental conditions were 10 h of darkness at 23 C (Ϯ3 C) and 14 h of light at 29 C (Ϯ3 C). One fertilizer application of N:P2O5:K2O 20:20:20 was made 1 wk after emergence (WAE) at a rate equivalent to 23 kg of N, P, and K/ha, and irrigation water was provided as needed. Herbicide treatments were applied to the cotton and soybean plants when they reached the four- to six-leaf stage. Herbicide rates are listed in Table 1. The normal use rates of 0.53 kg ai/ha, 0.56 kg/ha, and 1.12 kg/ha for 2,4-D, dicamba, and triclopyr, respectively, were considered full doses when calculating the reduced rates. Each of the treatments was applied to cotton and soybean plants without surfactant in 187 L of spray solution/ha using a spray chamber.10 Each treatment was applied to six pots of each species, and each pot was considered a replication. The two highest rates, 4 ϫ 10Ϫ1 and 2 ϫ 5 Clarity௡ herbicide, BASF Corporation, Agricultural Products Group, Research Triangle Park, NC 27709. 6 Remedy௡ herbicide, Dow AgroSciences, Indianapolis, IN 46268-1189. 7 Delta and Pine Land Company, P.O. Box 157, Scott, MS 38772. 8 Potting soil, Premier Horticulture Inc., Red Hill, PA 18076. 9 Potting soil, Scotts-Sierra Horticultural Products Company, 14111 Scottslawn Road, Marysville, OH 43041. 10 Spray chamber, De Vries Manufacturing, Hollandale, MN 56045. 1136 Volume 18, Issue 4 (October–December) 2004 ER 211 (236 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 103 of 245 WEED TECHNOLOGY 10Ϫ1, were only applied to the cotton plants. Similarly, the two lowest rates, 5 ϫ 10Ϫ5 and 1 ϫ 10Ϫ5, were only applied to the soybean plants. These occlusions were based on plant responses recorded during preliminary trials. The plants were evaluated 1, 5, 9, and 14 d after treatment (DAT) using the scale outlined by Sciumbato et al. (2004) with one-leaf and one-stem evaluation being recorded for each pot. The greenhouse rate–response experiment was performed twice, and the test plants were destroyed after the 14-d evaluation periods. Field. Data for the field-injury curve were collected at the Texas A & M Agronomy Field Laboratory near College Station, TX. Delta Pine 50 cotton and Delta Pine 415 soybean were planted in four-row plots on a Belk clay (Entic Hapluderts). The fungicide metalaxyl11 [N(2,6-dimethylphenyl)-N-(methoxyacetyl) alanine methyl ester] and the insecticide phorate12 {O,O-diethyl S[(ethylthio) methyl] phosphorodithioate} were applied at planting followed by a preemergence application of pendimethalin13 at 0.47 kg/ha for weed control. The plots were furrow irrigated as needed, and postemergence weed control was performed through tillage or hand hoeing. All herbicide treatments were applied in 187 L/ha carrier volume using a CO2 backpack sprayer when the test plants reached the four- to six-leaf stage. Herbicide rates were identical to those used to produce the greenhouseinjury curves and are listed in Table 1. Plastic tarps were held around each plot receiving the herbicide application to reduce the risk of spray particles moving off target and contaminating other plots. Five test plants were then selected randomly in each plot and evaluated 1, 5, 9, and 14 DAT using the evaluation method outlined by Sciumbato et al. (2004). One-leaf and one-stem evaluation was made for each of the five plants. The field rate–response experiment was carried out four times, and the plots were destroyed after each 14-d evaluation period. Equation Estimation. Injury curve establishment was done identically for greenhouse and field data. First, rate–response data were transformed using Equation 1 and graphed in scatter plots with the natural log of the rates on the x axis. The natural log transformation was used on the x axis because differences between herbicide rates were often an order of magnitude. The arcsine 11 Ridomil௡ fungicide, Syngenta Crop Protection, Inc. P.O. Box 18300, Greensboro, NC 27409. 12 Thimet௡ insecticide, BASF Corporation, Agricultural Products Group, Research Triangle Park, NC 27709. 13 Prowl௡ herbicide, BASF Corporation, Agricultural Products Group, Research Triangle Park, NC 27709. transformation was used for injury data because most scatter plots were sigmoid after this transformation. DAT injury  ͱ   ϭ ARCSINE   leaf injury ϩ stem injury   2   100  [1] Next, parameters of Equation 2 (Seefeldt et al. 1995) were estimated using the SAS secant method (DUD method) of nonlinear regression modeling (SAS 1985) until the best possible fit of equation line to injury data was obtained for each treatment at each DAT. YϭCϩ DϪC 1 ϩ (X/I50 ) b [2] In Equation 2, D represents the response of the plants at low herbicide rates, C denotes the response of the plants at high herbicide rates, b is the slope, and I50 is the herbicide rate that caused 50% of the total plant response. Because this was a nonlinear regression procedure, the fit of the model to the data was determined using residual plot analysis. Quantitation of Volatility. Greenhouse. Herbicide volatilization and drift were produced in the greenhouse using volatility chambers (Figure 1). These chambers directed air at a constant speed first over bermudagrass (Cynodon dactylon L. Pers. CYNDA) sod that had been treated with 2,4-D, dicamba, or triclopyr, then through the canopies of cotton and soybean indicator plants placed downwind, and finally out of the greenhouse. The wind speed was maintained at 3.2 km/h and monitored using hot-wire anemometers. Delta Pine 50 cotton and Delta Pine 415 soybean to be used as indicator plants were grown in a greenhouse at the seeding rate of three plants per pot using 15-cm plastic pots. The growth medium was a 3:1 (v/v) mixture of Pro-Mix and Redi-Earth. Greenhouse conditions were 10 h of darkness at 23 C (Ϯ3 C) and 14 h of light at 29 C (Ϯ3 C). One fertilizer application of N:P2O5:K2O 20: 20:20 was made 1 WAE at a rate equivalent to 23 kg of N, P, and K/ha, and irrigation was performed as needed. The flats of bermudagrass sod used in the chambers were 42 by 23 cm and were maintained in the greenhouse for 2 mo before the experiment. The grass was cut to 7 cm before treatment. Each herbicide and water alone as a control was applied to two bermudagrass flats when the cotton and soybean indicator plants reached the four- to six-leaf stage. The herbicides were applied incarrier volumes of 187 L/ha using the same spray cham1137 Volume 18, Issue 4 (October–December) 2004 ER 212 (237 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 104 of 245 SCIUMBATO ET AL. : QUANTIFYING AUXINLIKE HERBICIDE VOLATILITY TOP VIEW and the test plants were destJ·oyed at the end of each 14d evaluation period. c ' ' ...... .' ' E SIDE VIE W 42 em I It F Figure 1. A diagram of the volatility chambers used in the greenhouse-volatility experiment. Air enters the assembly through the outermost openings of the air ducts and exits through the fan near the top. Flexible ducting was used to route exhaust outside the greenhouse. Legend: A = treated bermudagrass. B = indicator plants. C = exhaust fan. D = air intake. E = air ducts (chambers). and F = exhaust to outside. ber that was used to create the rate-response curves. Herbicide rates were 2 times the n01mal use rates, or 1.06, 1.12, and 2.24 kglha of 2,4-D, dicamba, and tJiclopyr, respectively. One hour was allowed to pass after the herbicide applications were made and before the fiats were placed in the diift chambers so that spray droplets would disperse and all herbicide movement could be attJ·ibuted to volatilization. As soon as the hour had elapsed, one bermudagrass fiat was placed at the intake end of each volatility-chamber duct In addition, one cotton and one soybean pot were inserted 40 em downwind in each chamber (Figure 1). This configuration provided a total of two cotton and two soybean replications for each herbicide treatment. The cotton and soybean were left in the airflow of the chambers for 48 h, whereupon they were removed and evaluated 1, 5, 9, and 14 dafter exposure. One-leaf and one-stem evaluation was given for each pot at each DAT. This experiment was canied out four times, Field. Herbicide volatility was assessed in the field by placing cotton and soybean indicator plants onto 15 by 15 m range plots that had been sprayed with 2,4-D, dicamba or tiiclopyr. The plots were populated ptimarily with betmudagrass and Johnsong~·ass (Sorghum halepense L. Pers. SORHA). Delta Pine 50 cotton and Delta Pine 415 soybean plants were g~·own in 15-cm plastic pots at populations of three plants per pot The growth medium was composed of a 3: 1 (v/v) mixture of the Belk clay from the field rate-response experiment location and Redi-Eatth potting soiL Indicator plants to be used in the field-volatility experiment were maintained in outdoor conditions, and water was supplied as needed. One application ofN:P2 0 5 :K2 0 20:20:20 was applied 1 WAE at the NPK rate of 23 kglha. Volatility plots were mowed to a unif01m height of 8 em when the indicator plants reached the four- to sixleaf stage. A tt·actor-mounted spray boom was used to apply 1.06, 1.12, or 2.24 kglha of 2,4-D, dicamba, or u·iclopyt~ respectively. One hour was allowed to elapse between the herbicide application and test plant placement so that spray pruticles could disperse. Five pots of each indicator species were then placed inside the treated plots. Eight cotton plants were also placed in 3.5-m buffer zones ru·ound each plot to monitor movement of herbicides outside the plots. The plants were left inside the plots for 48 h and then collected for evaluation. Plant injury was recorded 1, 5, 9, and 14 d after collection with one-leaf and one-stem injury value being assigned per pot The expetiment was cruTied out four times, and the plants were desu·oyed at the end of each 14-d evaluation petiod. Equation Application_ Plant injury values resulting from herbicide volatilization dming the field and g~·een­ house-volatility expetiments were applied to the rateresponse curves to calculate herbicide exposure rates. Injury values from the 1n(~ =~ - 1) + log(I Log(X) = b 50 ) [3] g~·eenhouse-volatility experiment were applied to the rate-response curves. Similru·ly, injury fi·om indicator plants used in the field-volatility expeiiment was applied to field rate-response curves. Volatility data were transf01med as before using Equation 1 and insett ed into Equation 3 as Y along with the g~·eenhouse 1138 Volume 18. Issue 4 (October-December) 2004 ER 213 (238 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 105 of 245 WEED TECHNOLOGY 0.25 0.45 . - - - - - - --;...-, 0.25 0.45 O.b 0.30 f--- - - -- / < - ---j 0.15 1----;;""'"'_/:::::::.__ _ _ _---l ~ ... 0.1 5 o.os • • -~ 40,05 •• ...'f .(l.05 I.• 0.30 •• c:: ~ .fj - ~ 0.30 1----===========--~ ·8 ·6 :§c-oIS -~ 0.05 ~ 0.30 tt::ii::~::i25E~ c -0.1 5 1 -- - - - - - ----l -0.25 ' - - - - - - - - - - - ' -6 .. ·2 .. 0.25 - 0.45 0.15 1 - - - - - - - - - - - j ;;;:" o.o L.,---:-----,,----,--..,---___J -8 (). IS g 0.05 ;"' :so.1s = c::" "0 ·~ ·0.05 • t c:: -0. 15 -0.25 0.0 -10 Dicamba injury on greenhouse couon ·8 ·6 -4 ·2 Dicamba injury on greenhouse soybean 0.25 OAS ·2 c :... -oos != -10 .. 2,4-0 injury on greenhouse soybean 0.25.-----------, -; . -0.25 0.0 -10 .-----------, c • ·0.15 -0.25 2 ,4-D injury on gree nhouse C<)IIOn ~0 .45 .. c:: 0.15 ·0.15 o.o '-..,.., lo--: .a---,.6,-----..,------:. .2 --:-' 0.05 0.45.-- - - - - - -""7"1 0. 15 0.30 0.30 1 -- - - - - --1----l ~ 0.05 .ii "' -0.05 0. 15 :X I 0.15 1-====~--~ -0.15 0.0 -10 .g .;; ·• 0.0' - - - - - - - - - - - ' · 10 -8 -6 -0.25 ·2 Triclopyr injury on greenhouse couon Triclopyr injury o n greenhouse soybe 0.15 = 0. 15 .., ... " .1 :?0 ;., .. = c: .. •• • .. ~ o.os "fj-o.o5 -0.15 0.0 -10 -8 ..• I '5 ~ 0.30 1 - - - - - - - ~ --=-l ~ c .., / ·10 -$ -6 0.30 1--- - - - - - - - ---l ----- 0.15 1 - - - - - =- ~ =----1 ·8 .; 0.05 -~ -0.05 Q<: -0. 15 o.oL--- - - - - - - - - - 1 -10 ~ -6 t &: d ..... .. .... .. • .. .. + ; -0.25 .. .. . • -0.25 t• - Dicamba injury on field soybean .... 0.15 -0. 15 •t -2 Dicamba injury on field colton 0.15 Q<: "' o.oL - - - - - - - - - -- ' C: ·2 0.45 . - - - - - - - - - - - - , ... ... 0.05 'f -o.os :? o IS1.:=::::::::::;;;;;-""""'----~ = • -0.25 ·• ·6 -. .. .. • .. Q<: .a .. .. • .... .. .... .. .. ....- 2,4-D injury on field soybean - 0.4 ~ OJ ... -0.25 = 'fj " .... .... 0.45 0.25 0.30 g 0.05 Ia "" -o.o5 "ii 0. 15 / 0.15 ·C.: -0.15 0.0 -10 ·2 -Q -8 ·• •t • ••• -0.25 ·1 Triclopyr injury on field conon Triclopyr injury on field soybean rate (ln) rate (In) Figure 4. Model lines and residual graphs for 2,4-D , dicamba, and triclopyr inj ury on cotton plants 14 d after treatment. The cotton plants were grown and maintained under field conditions and treated with reduced rates of the herbicides during the four- to six-leaf stage of growth. • Figure 5. Model lines and residual graphs for 2,4-D, dicamba, and triclopyr inj ury on soybean plants 14 d after treatment. The soybean plants were grown and maintained under field conditions and treated with reduced rates of the herbicides during the four- to six-leaf stage of growth. Table 2. Equation parameters used to produce rate-:response curves for 2,4-D, dicamba, and triclopyr on cotton and soybean. Equation parameter Herbicide Species 2,4-D Cotton Soybean Dicamba Cotton Soybean Triclopyr Cotton Soybean Location D• Iwb 0 bd Greenhouse Field Greenhouse Field Greenhouse Field Greenhouse Field Greenhouse Field Greenhouse Field 0.127451804 0.111256753 0.14471091 0.085808708 0.127456222 0.000000036 0.109684117 0.105134999 0.071783946 0.074766650 0.118560852 0.119530719 0.058282629 0.006530843 0.01884355 2.650930326 0.058289464 0.204063510 0.002548099 0.004593443 0.503525162 0.328585818 0.019842014 0.060349614 0.553188598 0.274637140 0.11889786 0.412926810 0.553197713 0.710261595 0.309249972 0.316954681 0.971210467 0.382724099 0.463636067 0.290065472 0.615002368 2.346593855 -27.7838911 0.560964401 0.615006164 0.252187506 Ll 13252520 0.746412742 0.388578618 0.388538078 1.788452643 1.223556518 • D, plant response at low rates. b Iso. herbicide rate causing 50% of the total plant response. < C, plant response at high rates. db, slope. 1140 Volume 18, Issue 4 (October- December) 2004 ER 215 (240 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 107 of 245 WEED TECHNOLOGY Table 3. Greenhouse herbicide exposure rates estimated for cotton and soybean 14 d after exposure. Equivalent rateb Crop Herbicide Estimationa ϫ kg/ha % Cotton 2,4-D dicamba triclopyr 2,4-D dicamba triclopyr 0.003 ad 0.001 a 0.049 a 0.014 b 0.001 c 0.025 a 0.0016 0.0006 0.0549 0.0074 0.0006 0.0280 0.1500 0.0500 2.4500 0.7000 0.0500 1.2500 Soybean Volatilizationc Table 4. Field herbicide exposure rates estimated for cotton and soybean 14 d after exposure. Equivalent rateb Crop Herbicide Estimationa ϫ kg/ha % Cotton 2,4-D dicamba triclopyr 2,4-D dicamba triclopyr 0.004 ad 0.002 a 0.125 b 0.160 a 0.010 c 0.110 b 0.0021 0.0011 0.1400 0.0848 0.0056 0.1232 0.2000 0.1000 6.2500 8.0000 0.5000 5.5000 Soybean Volatilizationc a Concentation of herbicide exposure rate expressed as a multiple of the normal use rates of 0.53 kg/ha of 2,4-D, 0.56 kg/ha of dicamba, and 1.12 kg/ ha of triclopyr. b Exposure rate expressed in kilograms per hectare. c Volatilization percentage calculated by dividing the equivalent rate by the total amount of the herbicide applied. d Values with the same letter are not significantly different from one another at the 5% level of significance. a Concentation of herbicide exposure rate expressed as a multiple of the normal use rates of 0.53 kg/ha of 2,4-D, 0.56 kg/ha of dicamba, and 1.12 kg/ ha of triclopyr. b Exposure rate expressed in kilograms per hectare. c Volatilization percentage calculated by dividing the equivalent rate by the total amount of the herbicide applied. d Values with the same letter are not significantly different from one another at the 5% level of significance. rates in the greenhouse, although this value was only significant with soybean. margins of cotton make injury to that species easier to record. That difference between species could translate to different evaluations of injury brought about by identical herbicide exposure, therefore different volatilization estimates. Volatility estimations from field data tended to be greater than those obtained from greenhouse data. This difference can be explained by the contrast between greenhouse and field conditions. High temperatures have been shown to promote herbicide volatility (Behrens and Lueschen 1979), and temperatures recorded during the field study were sometimes greater than 38 C whereas greenhouse temperatures never exceeded 29 C. The position of the test plants relative to the herbicide source may also explain some of the difference between field and greenhouse injury. Herbicide fumes may have risen from the treated surface up through the plant canopies in the field experiment, exposing the numerous stomata found on lower leaf surfaces to an upward movement of herbicide vapor. In contrast, test plants that were placed in volatility chambers were exposed to herbicide vapors moving horizontally, which may have limited the amount of herbicide retained and absorbed on plant surfaces resulting in less injury. Field. No injury was found on cotton plants placed in the buffer zones of the field-volatility plots, therefore all injury observed on test plants placed inside the plots was considered to be the result of volatility from that plot. The most obvious difference in herbicide-volatilization injury to field cotton was between the 2,4-D and dicamba salts and the triclopyr ester (Table 4). The exposure rates of the DGA salt of dicamba and the DMA salt of 2,4-D did not differ significantly with field cotton. However, both herbicides were less volatile than BEE triclopyr. Volatility injury on field soybean was different from that of field cotton (Table 4). There were significant differences among the volatility rates of all three herbicides, with the DGA salt of dicamba producing significantly lower exposure rates than the other herbicides in field soybean. However, unlike what was observed in the greenhouse, the DMA salt of 2,4-D appeared to be the most volatile of the three compounds in field soybean. This is difficult to explain because all field cotton and soybean plants received the same herbicide exposure rate during each replication. A difference in herbicide uptake because of leaf surface area is not likely. If soybean plants absorbed more of the herbicides because of surface area, all three herbicides would have increased activity proportional to that observed in cotton. This was not the case because the DGA salt of dicamba caused less injury than the DMA salt of 2,4-D. The contrast could be explained by the relative difficulty in evaluating soybean injury when compared with cotton (Sciumbato et al. 2004). The more prominent petioles and larger leaf Modeling Procedure Evaluation. The modeling procedure used in this study was effective for calculating herbicide rates that corresponded to injury from rate– response curves. Herbicide-rate estimates produced by the models were reasonable for the observed injury. However, the discrepancies between rates calculated from cotton and soybean suggest that difficulties in species evaluations can have significant effects on volatilization estimates. 1141 Volume 18, Issue 4 (October–December) 2004 ER 216 (241 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 108 of 245 SCIUMBATO ET AL.: QUANTIFYING AUXINLIKE HERBICIDE VOLATILITY The volatility chambers provided a reliable and effective method of simulating field volatility and off-target movement of herbicides. The use of the volatility chambers can easily be expanded to include other herbicides and treated surfaces. In addition, the drift chambers could be effective tools for particle-drift research after modification. Estimates of greenhouse volatility tended to be less than that from the field. The most likely explanation for this is the difference between greenhouse and field temperatures. Future research gathering yield data from plants treated with these herbicides will be necessary before it will be possible to predict the effect of drift on crop yield. ACKNOWLEDGMENTS We thank Dr. Clay Salisbury and Dr. Kevin McInnes for their assistance with this research. LITERATURE CITED Arle, H. F. 1954. The sensitivity of Acala 44 cotton to 2,4-D. West. Weed Control Conf. Proc. 14:20–25. Behrens, R. and W. E. Lueschen. 1979. Dicamba volatility. Weed Sci. 27: 486–493. Bovey, R. W. and R. E. Meyer. 1981. Effects of 2,4,5-T, triclopyr, and 3,6dichloropicolinic acid on crop seedlings. Weed Sci. 29:256–261. Council for Agricultural Science and Technology. 1975. The Phenoxy Herbicides. Ames, IA: Council for Agricultural Science and Technology Rep. 39. 22 p. Miller, J. H., H. M. Kempen, J. A. Wilderson, and C. L. Fox. 1963. Response of Cotton to 2,4-D and Related Phenoxy Herbicides. USDA Technical Bulletin 1289. [SAS] Statistical Analysis Systems. 1985. SAS User’s Guide: Statistics. 5th ed. Cary, NC: Statistical Analysis Systems Institute. P. 586. Sciumbato, A. S., J. M. Chandler, S. A. Senseman, R. W. Bovey, and K. L. Smith. 2004. Determining exposure to auxin-like herbicides. I. Quantifying injury to cotton and soybean. Weed Technol. 18:1125–1134. Seefeldt, S. S., J. E. Jensen, and E. P. Fuerst. 1995. Log-logistic analysis of herbicide dose response relationships. Weed Technol. 9:218–227. Taylor, A. W. and W. F. Spencer. 1990. Pesticides in the Soil Environment: Processes, Impacts, and Modeling. Madison, WI: Soil Science Society of America. Pp. 213–255. Texas Agriculture Code. 1984. St. Paul, MN: West. Chapter 75. 1142 Volume 18, Issue 4 (October–December) 2004 ER 217 (242 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 109 of 245 Dicamba Injury to Soybean J.D. Weidenhamer,* G. B. Triplett, Jr., and F. E. Sobotka ABSTRACT ering when 'Lee' soybean was treated with other chlorophenoxy herbicides. Wax et al. (9) noted that this cultivar is determinate and suggested that soybean yield response to dicamba at different stages of growth may depend on whether cultivars are determinate or indeterminate. Among soybean cultivars that are similarly susceptible to yield losses from dicamba, other differences in response have been observed. Yield reductions of 'Jacques 109' and 'Corsoy' soybean were approximately equal for equivalent rates of dicamba applied at the early bloom stage, but height reductions at specific rates differed markedly ( 1). The objectives of this research were (i) to determine the response of Elf and Williams soybean to dicamba over a wide range of applied rates; and (ii) to evaluate the use of dicamba injury symptoms to predict resulting losses in yield. Dicamba (3,6-dichloro-2-methoxybenzoic acid) effectively controls many dicotyledonous weeds, but nontarget species such as soybean (Glycine max (L.) Merrill) are susceptible to spray or vapor drift. Field studies were conducted on a Canfield silt loam (fineloamy, mixed, mesic Aquic Fragiudalf) soil to determine the response of 'Elf and 'Williams' soybean to dicamba over a wide range of applied rates, and to evaluate the use of dicamba injury symptoms to predict yield reductions. Soybean yield in response to increasing rates of dicamba was described by equations of the form y = Aexp(- bJC), where y = yield, A = maximum yield (rate = 0 g ha-'), b is a constant, and JC = rate of dicamba applied. Height reduction, seed number ha-•, and morphological symptoms of dicamba injury were useful in assessing yield reduction. Except for Elf soybean treated at the midbloom stage, there was no yield reduction without height reduction, regardless of foliar symptoms. Seed number ha-• decreased with increasing rates of dicamba and was closely correlated with yield. Yield reductions greater than 10% were indicated by severe morphological symptoms of injury, such as terminal bud kill, splitting of the stem, swollen petioles, and curled, malformed pods. Other foliar symptoms, such as distinctive crinkling and cupping of the terminal leaves, occurred at rates much lower than those required to cause yield reductions. MATERIALS AND METHODS Three field studies were conducted in 1980 and 1981 on a Canfield silt loam soil at the Ohio Agricultural Research and Development Center, Wooster, OH. Table I summarizes the treatment variables (cultivar, row width, dicamba formulation, growth stages at time of application, and rates) examined for each study. Fertilizer (40 kg ha- 1 P and 70 kg ha- 1 K) was applied to all sites before spring plowing, disking, and seed innoculation and planting of Elf or Williams soybean. Main plots were 3 X 27.5 m (17 rows spaced 0.18 m apart or four rows spaced 0.76 m apart), with a 1-m unplanted border between plots. Seeding rates were 13 seed m- 1 (0.18 m row spacing) and 33 to 39 seed m- 1 (0. 76 m row spacing) for Elf soybean, and 11.5 seeds m- 1 (0.18 m row spacing) and 26 to 33 seed m- 1 (0. 76 m row spacing) for Williams soybean. Following planting, 0.45 kg ha- 1 metribuzin [4-amino-6-( I, l-dimethylethyl)3-(methylthio )-1 ,2,4-triazin-5( 4H)-one] and 3.4 kg ha- 1 alachlor [2-chloro-N-(2,6-diethylphenyl)-N-(methoxymethyl)acetamide] were applied. Weeds not controlled by preemergence herbicides were removed by hand. Because dicamba causes morphological changes and yield reductions in soybean over a wide range of rates, a beltcarried C0 2 logarithmic rate sprayer was chosen to simplify experimental operations. The logarithmic sprayer was first described by Pfeiffer et al. (6) in 1955 and is particularly useful for herbicide screening studies {4, 7). Dedolph {3) has discussed questions of statistical analysis and validity of data, and described procedures for data analysis from experiments using logarithmic spraying techniques. Logarithmic applications of dicamba were made using a volume of solution necessary to deliver 12 dilutions, after which the sprayer ceased operation. The dilution rate decreased the concentration ofdicamba by 50% (one half-rate) every 1.8 to 2.2 min the 27.5 m main plots. Proper dilution was verified by applying a KCI solution, and analyzing paper strips placed at 1.8 m intervals for potassium content. The average rate of dicamba applied to each half-rate subplot 'was calculated to facilitate data analysis. These rates are shown in Table 2 and 3. All logarithmic treatments were applied in the same direction, instead of being randomized as recommended by D ICAMBA effectively controls many dicotyledonous weeds in corn (Zea mays L.) and other crops. However, its use may be limited by drift and injury to nontarget species. Soybean is especially sensitive to dicamba, and foliar symptoms can occur at rates as low as 1.0 g ha- 1 (1). Such symptoms, while highly visible, may not indicate yield loss (I) and rates considerably higher may be required before a decrease in production occurs. Data on the correlation of dicamba drift injury symptoms with actual yield reductions is needed to settle claims when drift does occur. Behrens and Lueschen (2) devised a scale (0 to 100) to evaluate dicamba drift injury to soybean at the first trifoliate leaf stage. Leaf crinkling, cupping, and malformation, as well as growth suppression and terminal bud injury, were among the symptoms observed. Yield reductions were associated with injury ratings of 60 to 70 or more. Soybean cultivar and the growth stage at time of dicamba application influence soybean response to dicamba (1,2,9). Soybean appears to be most sensitive to injury while in the flowering stage, where 9 to 11 g ha- 1 dicamba has reduced yields, compared with prebloom applications that require rates of 56 to 70 g ha- 1 to reduce yields (1,9). Contrary to studies with dicamba, Smith (8) found greater sensitivity before flowJ.D. Weidenhamer, Dep. of Chemistry, 232 Choppin Hall, Louisiana State Univ., Baton Rouge, LA 70803; G.B. Triplett, Jr., Dep. of Agronomy, Mississippi State Univ., P.O. Box 5248, Mississippi State, MS 39762; F. Sobotka, Sandoz LTD. Agro Div., 4002 Basel, Switzerland. Contribution from the Dep. of Agronomy, Ohio State Univ. Salaries and research support provided by State and Federal Funds appropriated to the Ohio Agric. Res. and Dev. Ctr., The Ohio State Univ. Journal Article no. 141-85. Received 28 June 1988. *Corresponding author. Published in Agron. J. 81:637-643 (1989). 637 ER 218 (243 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 110 of 245 638 AGRONOMY JOURNAL, VOL. 81, JULY-AUGUST 1989 Dedolph (3), because of concern that slight spray and/or vapor drift might occur between plots. If spraying direction were randomized, drift from subplots receiving the highest rates of over 100 g ha- 1 dicamba might be greater than the actual amount ( < 0.1 g ha- 1) applied to a neighboring lowrate subplot. The low rate might not produce foliar symptoms, while the low rate plus drift from the neighboring highrate subplot might show these symptoms, confounding the results. Control plots of both cultivars were present in each replication. These were divided into 12 subplots similar to the treated plots for observation and harvest. No significant variation occurred across these main nonsprayed plots for either plant height or yield. This indicates that soil and growing conditions were similar across the blocks. Precautions were taken to prevent drift. Dicamba was apJPlied under calm conditions during moderate temperatures. As a result, little drift actually occurred. As estimated by the severity of foliar symptoms on untreated plants 0.5 m from plants receiving the highest applied rates of dicamba, drift was less than 0.25 of 1% of the amount applied. Given the 0.5 m unsprayed border on each main plot, and the I m unplanted border between main plots, drift between main plots was insignificant, and within main plots, drift of this magnitude would not significantly change the applied rates. Soybean injury was assessed by measurement of height (average of three randomly selected plants per subplot), percentage stand reduction, and the presence or absence of several distinct morphological symptoms of herbicide injury; foliar aberrations, terminal bud injury, pod malformation, petiole enlargement, twisting of plant tops, splitting of the stem, canopy closure, and delayed maturity. For each symptom there was a threshold rate below which it did not occur, a small range of rates where some plants were slightly affected, and a rate above which all plants were affected. Evaluations were based on this latter rate. The center 1.4 m (eight rows spaced 0.18 m apart) or 1.5 m (two rows spaced 0. 76 m apart) of each subplot was machine harvested. Seed weight was determined for 100 randomly selected seed per sample. The number of seed produced per hectare was calculated from yield and 100 seed weight. All yields and 100 seed weights were adjusted to 135 g kg- 1 moisture content. Logarithmic Rate Study The initial study was conducted to correlate yield reductions from dicamba injury with morphological symptoms, over a wide range of applied rates. Elf (a semi-drawf determinate cultivar) and Williams soybean (an indeterminate cultivar) were treated at the late prebloom (41 DAP, days after planting) and midbloom (78 DAP) stages of growth. Rates were chosen to produce the complete range of dicamba injury to soybean, from complete kill to no effect. The experiment was arranged in a randomized complete-block design with three replications. The ex]Perimental site was planted 7 May 1980. Observations were made at 1.8 m intervals along the main plots. Effects of prebloom treatments were evaluated approximately 5 and 10 wk after application. Effects of midbloom treatments were observed approximately I 0 wk after application. The times chosen for evaluation allowed for full development of injury symptoms. Significant recovery from injury occurred following prebloom applications, and this was assessed in the I 0 wk evaluation. The subplots receiving the two highest rates of dicamba were hand harvested because of reduced plant height. Discrete Rate Study A discrete rate study was conducted to verifY the rewlts obtained in the logarithmic rate study. In addition, row width was incorporated as a treatment variable along with cultivar and stage of growth at application. Main treatments consisted of all combinations of cultivars, growth stage at application, and row widths. They were arranged as a randomized complete block in three replications. Subplots received six rates of dicamba including a control. Rates ranged above and below the minimum rate required for yield reduction in 1980, with the median rate being the highest rate that did not reduce yields in 1980. Because of differences in 1980 for cultivars and times of application, four sets of rates were chosen. These are listed in Table 1. Dicamba was applied to the subplots in strips 4.0 m long, leaving 0.6 m borders. Discrete rates were applied by filling both the concentrate chamber and the diluent tank o:f the sprayer with the same herbicide solution. Late prebloom treatments were applied 52 DAP, immediately before Howering. Midbloom treatments were applied 70 DAP, when plants were blooming and the first pod was up to 20 mm long. The harvested subplot lenj~th was 3.4 m. Prebloom treatments were evaluated 1, 2, and 6 wk after application, and midbloom treatments 2 and 6 to 7 wk after application. The I and 2 wk evaluations were included to better quantifY the development of injury symptoms with time. A later (10 wk) evaluation ofprebloom treatments was unnecessary because droughty conditions reduced growth and recovery from injury did not occur as in 1980. Formulation Study This study was conducted to determine whether then~ was any difference in the effects of dicamba dimethylamine (DMA) and Na salts on soybean at a specific stage of growth. Table 1. Summary of treatment variables and herbicide rates examined. Row width (m) Dicamba formulation Time of application Type of application Rates, g active ingredient ha- 1 Evaluation of injury Elf, Williams 0.18 DMAt Pre- and midbloom Logarithmic 0.028-115 5WATt (Pre-, midbloom) IOWAT (Pre bloom) Discrete rate study (1981) Elf, Williams 0.18, 0.76 DMA Pre· and midbloom Discrete 0, 0.32, 5.0, II, 40, 81 (Elf prebloom) 0, 0.32, 5.0, 22, 71, 110 (Elf midbloom) 0, 0.64, 10, 20, 40, 81 (Williams prebloom) 0, 0.32, 2.5, 9.5, 20, 81 (Williams midbloom) I, 2, 6WAT (Pre bloom) 2, 6-7 WAT (Midbloom) Formulation study (1981) Williams 0.18 DMA, Na§ Midbloom Logarithmic 0.056-460 2, 6-7 WAT Experiment Cullivars Logarithmic rate study ( 1980) t DMA = Dimethylamine salt. t WAT = Weeks after treatment. § Na = Sodium salt. ER 219 (244 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 111 of 245 639 WEIDENHAMER ET AL.: DICAMBA INJURY TO SOYBEAN The initial rate was increased to 460 g ha-• because complete kill had not been obtained in the Logarithmic Rate Study at 115 g ha-•. The experimental design was a randomized complete block with three replications. Three main treat- ments consisted of each herbicide plus an untreated control. Williams soybean was planted on 22 May and flowering began 14 to 16 July 1981. Treatments were applied on 24 July 63 DAP, during the midbloom growth stage. Detailed ob- Table 2. Effect of dicamba treatment at the prebloom and midbloom stages of growth on yield and morphology of two soybean cultivars (Elf and Williams) planted in 0.18 m rows (Logarithmic study, 1980). Average dicamba rate Yieldt (g ha-•) (Mg ha-•) Height (%of control) Plant morphology Terminal leaf morphology* CUP CRK LS< LM< LS> LM> TT DST TBK SR POD Elf soybean 0.04 0.08 0.16 0.32 0.63 1.3 2.5 5.0 10 20 40 80 Control§ R'~ Pre Mid Pre Mid 4.07 3.76 Ill 107 3.42 3.01 107 104 3.32 3.64 102 101 3.22 3.70 100 102 3.38 3.66 98 109 3.31 2.93 95 110 3.29 3.22 so 109 3.45 3.84 71 106 3.62 3.81 63 102 2.61 3.53 61 94 2.61 3.35 47 88 2.20 2.19 38 82 3.37 0.56m0.82m 0.55 0.23 0.86 0.92 + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + Williams soybean 80 Control§ Pre Mid Pre 3.29 3.90 92 3.23 3.25 95 3.39 3.73 97 3.69 3.23 92 3.76 3.12 83 3.35 3.88 84 3.48 3.23 78 3.33 3.76 66 3.42 3.26 66 3.42 2.68 63 2.84 2.89 50 2.14 2.76 38 3.59 0.57m R 2* 0.46 0.04 0.08 0.16 0.32 0.63 1.3 2.5 5.0 10 20 40 0.39 Mid 98 100 106 104 102 100 96 92 92 79 74 68 !.Om + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + + 0.80 0.82 • All R 2 values are significant at the 0.05 level. t Due to loss of yield data from one replicate of both Elf and Williams controls, statistically valid comparisons to control cannot be made. * + =symptom present,- =symptom absent, CRK =crinkling, and CUP= cuppimg of terminal leaves; LS< or LS> =size of terminal leaves reduced 40%; LM< =minor leaf margin damage (<25% of margin); LM> =severe leaf margin damage; DST =gross distortion ofleafvenation patterns; TT = twisted tops; TBK = terminal bud kill; POD = curled, malformed pods; SR = stand reduced > 10%. When the terminal bud was killed, any foliar symptoms (CRK, CUP, etc.) occurred on developing lateral shoots. Treatments were evaluated S wk after treatment. § Control plots received no dicamba application. Table 3. Effect of dicamba DMA treatment at the midbloom stage of growth on yield and morphology of Williams soybeans planted in 0.18 m rows (Formulation study, 1981). Average dicamba rate (g ha-') 0.08 0.2 0.4 0.8 1.6 3.5 7.4 16 33 70 149 316 Control* R'* Terminal leaf morphologyt Yield Height (% of control) (% of control) 98 94 87 89 101 86 86 77 51 23 2 0 2.49 Mg ha-• 0.93 Maturity delay 96 94 94 84 85 75 70 64 54 42 0.83 m 0.90 CRK CUP LS< LS> LM< LM> DST TT TBS TBK TBK+ + + + + + + + 95 95 Plant morphology + + + + + + + + + + + + + + + + + + + + + + + + ss PET POD CAN SR -' + + + + + + + + + + + + + + + + + + + + + + + + + • All R2 values are significant at the 0.05 level. t TBS = terminal bud stunted, TBK + = more than terminal bud killed, SS = split stems, PET = abnormal enlargement of petioles, CAN = loss of canopy closure. For other symptoms, see Table 2. Data for the dicamba Na treatment showed no differences from the dicamba DMA treatment and are not presented. * Control plots received no dicamba application. Treatments were evaluated 7 wk after treatment. ER 220 (245 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 112 of 245 640 AGRONOMY JOURNAL, VOL. 81, JULY-AUGUST 1989 servations of soybean injury were made at approximately 2 and 7 wk after spraying. Statistical Analysis All data were subjected to standard analysis of variance and regression procedures. The Gauss-Newton method of iteration was used for nonlinear regression. All comparisons made are significant at the 5% probability level, unless noted otherwise. RESULTS AND DISCUSSION Yield Soybean yield response to increasing rates of dicamba is described by equations of the form y = Aexp (- bx), where y = yield, A = maximum yield (rate = 0 g ha- 1), b is a constant, and x = rate of dicamba applied (Fig. 1). Equations were fitted by the modified Gauss-Newton method of iteration because normal probability plots of residuals indicated that the appropriate model is y = Aexp(- bx) + ~. rather than y = Aexp(- bx) X ~. where~ represents sampling error. Were the latter model correct, the data could be analyzed more simply by taking the natural logarithm of both sides of the equation, which gives the equation for a straight line. Data from other studies also suggest an exponential decrease in soybean yield with increasing rates of dicamba (1,9). However, data in those studies were evaluated not by determining yield-response curves, which is more appropriate when treatments vary levels of a quantitative factor (5), but by multiple comparison procedures. For the variables studied, differences between dicamba treatments in their effect on yield were minor (see Table 2). Differences in yield reductions caused by prebloom and midbloom applications were less than previous reports (1,9). This may have been due to later application of the prebloom treatments in our studies. No difference in yield reductions was seen between the dicamba DMA and Na formulations (F = 0.21, Pr > F = NS). Row width had a small but significant effect on yield reductions caused by dicamba treatment. At rates 120 • _I above 1 g ha- 1, yields of soybean in 0.18 m rows were as much as 10% lower than yields of soybean in 0. 76 m rows (Because oflength, most data from the discrete rate study are not presented, but are available fi·om the senior author on request). At the time of spraying, soybean planted in 0.18 m rows had formed a complete canopy, while those planted in 0.76 m rows had not. Thus a higher percentage of the dicamba spray was probably intercepted by foliage in 0.18 m rows than in 0. 76 m rows. At the rates used, dicamba that did not contact the foliage was considered to have a negligible effect. Yield reductions were not as great in 1980 as in 1981 for equivalent rates of dicamba. In 1980, at least 15 g ha- 1 dicamba were required for a 10% reduction from maximum observed yields (Table 2). In 1981 ratt:s as low as 1.3 g ha- 1 dicamba caused a 10% yield reduction (Table 3). This difference is attributed to droughty conditions in 1981 (240 mm less rainfall during June through August) when soybean were less able to recover from dicamba injury. Auch and Arnold (1) also observed that dry conditions following dicamba applications resulted in greater yield reductions. Components of Yield Soybean yield is a function of plant population, the number of seed produced per plant and seed weight. Treatment with dicamba affected each of these components of yield. In both 1980 and 1981 stand reductions occurred at rates of 40 to 80 g ha- 1 dicamba and above for prebloom applications, and above 150 g ha- 1 for midbloom applications (Table 2 and 3). In 1980 stand reductions of approximately 30% were associated with yields reduced 20 to 40% of maximum observed yh~lds. Under droughty conditions in 1981, stand reductions were associated with yields already greatly reduced by other factors. Seed number was closely correlated with yield in both 1980 and 1981. The number of seed produced per hectare (a function of plant population and. the number of seed produced per plant) decreased as much as 95 to 1OOo/o with increasing rates of dicamba. Figure 2 summarizes the regression analysis of mean values of seed number ha- 1 and yield for all treatments in the 100 y= 1.06X -0.83 1- z 0 80 u.. 60 _I • • • ~ a: 80 OICAMBA DlMETHYLAMINE SALT '1 0 9~.4 I •18,81 ~ 0- r2=0.99 -' 0 (,) 0 .. 100 0 a: I- 2 0 u r 2 • 0.93 60 u.. 0 40 ~ w ;::: 40 MEAN VALUES FOR ALL TREATMENTS 0- 20 -' 20 UJ ): 0 0 100 DICAMBA 200 RATE, o--~~-~-~~.~~~~~---- o 300 g ha"' 20 40 -I NO. SEED ha, Fig. 1. Yield-rate response curve for midbloom applications of dicamba dimethylamine salt in the formulation study. 60 80 100 'Yo OF CONTROL Fig. 2. Relationship of yield and number seed ha-• for all treatments in the discrete rate study. ER 221 (246 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 113 of 245 641 WEIDENHAMER ET AL.: DICAMBA INJURY TO SOYBEAN discrete rate study. The slope was approximately 1.0. Similar results were observed in the logarithmic rate study (y ""' 0.9l2x + 9.84, R 2 = 0.93) for all treatments except Elf soybean treated prebloom, which did not show any decrease in seed number at the highest applied rates. Thus, the primary component of soybean yield affected by dicamba applied shortly before or during flowering appears to be seed number. At higher rates, dicamba affected seed weight (Fig. 3a and 3b). Midbloom applications of dicamba above 30 g ha- 1 increased seed weight approximately 2 g 100 seed- 1 in 1980 (Fig. 3a). This increase in seed size is probably due to the reduction in the number of seed produced. Prebloom treatment with dicamba reduced Elf soybean seed size (Fig. 2b). This was the primary component of yield affected for this treatment. Seed number was not reduced because the plants produced many lateral shoots and set the same number of seed as untreated plants. Height Low rates ofdicamba did not reduce soybean height. Height was reduced as much as 62% by higher rates of dicamba (Table 2 and 3). Quadratic equations described the relationship of height and log rate of dicamba applied (Fig. 4). For midbloom treatments, the effect of dicamba on height was similar in both 1980 and 1981. Because little growth occurred following midbloom treatments, there was little change in this relationship from 2 wk after treatment until harvest. With prebloom treatments, however, significant changes in the magnitude of height reductiens at a given dicamba rate did occur with time as the plants grew. Also, height reductions from prebloom applications were greater in 1981 than in 1980, when higher rainfall levels facilitated recovery from dicamba injury (Fig. 4). The correlation of height with yield was typically high for individual treatments (Fig. 5). Use of height reductions to predict reductions in yield was expected to be a valuable tool in evaluating actual drift injury where the amount of drift in unknown. However, the quantitative relationship of yield and height varied widely between years for both cultivars and times of application (Fig. 4b). Height reduction are, therefore, only a qualitative indicator of dicamba injury. Plant Morphology Injury increased with dicamba rate, and at higher rates(> 15 g ha- 1 in 1980 and 1.3 g ha- 1 in 1981) dicamba greatly affected the growth and development ofsoybean plants (Table 2 and 3). Symptoms of severe injury included stand reductions, death of the terminal bud, curled malformed pods, split stems, swollen petioles, and twisting of plant tops. Foliar aberrations included distinctive crinkling and cupping of the terminal leaves, leaf margin injury, and size reduction. These symptoms, which are understandably worrisome to growers, occurred at rates as lowas0.2gha- 1 in J980and0.06g ha-' in 1981, much lower than rates required for yield reduction (Table 2 and 3). Foliar injury developed sooner and continued to develop on new growth longer at higher rates of dicamba (data not shown). In the formulation study, symptoms appeared 1 to 2 d after treatment at rates above 100 g ha·', and after 13 d, foliar symptoms were present at all rates. Soybean response to dicamba vaned depending on growth stage at the time of application. Foliar symptoms were most pronounced and occurred at lower 21 21 20 20 19 01 01 18 1- 1- (.!) (.!) 18 :I: :I: i£j ~ 0 UJ UJ i£j ~ 0 UJ UJ ILSD.o~ 1/) ~ 1/) 16 2 15 14 MID BLOOM: o ELF e WILLIAMS PREBLOOM : o ELF • WILLIAMS 13 12 12~~~~~~~--~~--~~--- - 1.00 I LSD.o5 0 14 13 17 0.00 LOG ( DICAMBA 1.00 RATE, g 2.00 ho 1 ) -1.00 0.00 LOG ( DICAMBA Fig. 3a. Relationship of 100-seed weight and log rate dicamba applied for midbloom treatments in the logarithmic rate study. 1.00 RATE, g 2.00 ho'> Fig. Jb. Relationship of I 00-seed weight and log rate dicamba applied for prebloom treatments in the logarithmic rate study. ER 222 (247 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 114 of 245 642 AGRONOMY JOURNAL, VOL. 81, JULY-AUGUST 1989 Table 4. Tentative criteria for evaluation of drift injury; symptoms associated with yield reductions of less than and greater than 10'% of control. Symptoms that may be present with yield reductions < IO%t Cultivar-Time of treatment CRK CUP LS< LM< DST TT + + + + + + + + + + + + + + + + + Williams-Prebloom Williams-Midbloom Elf-Prebloom Elf-Midbloom + + + TBK + NHR§ + + + Symptoms that indicate yield reductions > IO%:j: Williams-Prebloom TBS TBK TT ss PET POD CAN + + + + + + +§ + + + + + + + + + Williams-Midbloom + Elf-Prebloom Elf-Midbloom + + + + MAT SR SON + + >15%(0.18m) > 5%(0.76m) + + + > 5%(0.1.8m) > 10%(0.76m) + + HT > 10%(0.76m) + t NHR = yields not reduced if there was no height reduction. For definitions of other symptoms, see footnotes of Tables 2 and 3. :j: HT = height reduction 5 to 10 wk after treatment, and SON = reduction in seed number ha·• > 10%. § 0.18 m rows only. 110 . . .. . . \~· :'~ • 100 ~-.. ...J 0 0::: I- : 90 80 0 70 \ (!) 60 iLi 50 0 ~ ·.·. \ Q \ ·.\ 40 !:!:! ·. >- 20 \ ·•··... ··.. -1.5 ~ 60 _J ··.. 40 • • (.) IL. ··.\. I 100 ~ 80 0 \ ~ II ...J • 0 0::: ' r , ........ ··,···... ··... IL. ~ --~-~ ... z 0 (.) 120 0 • • •• 50 -0.5 +0.5 +1.5 +2.5 1 LOG ( OICAMBA RATE, g ho ) Fig. 4. Relationship of height 7 wk after treatment and log rate dicamba applied for the midbloom dicamba DMA treatment in the formulation study (data "points and_, y = -4.67r -6.1Sx +91.4, R 2 =0.90). Height of untreated controls was 0.8 m. Dashed lines indicate height-log rate relationships for Williams soybean in 0.18 m rows treated prebloom in 1980 (10 wk after treatment, --)and 1981 (6 wk after treatment, · • • ·). rates on growing plants treated at the prebloom stage. The fewest symptoms were observed on Elf soybean treated midbloom (Table 2). Elf is a determinate cultivar that ceases vegetative growth after flowering. Following midbloom treatment with dicamba, there was little or no vegetative growth evident on severely injured plants, such as lateral branch formation, and foliar symptoms were largely absent. With prebloom treatments, severe injury and death of the terminal bud occurred at high rates, but these plants formed numerous lateral shoots. These lateral shoots typically showed symptoms of dicamba injury. The effect of row width on plant morphology was similar to that noted for yield. Morphological aber- 60 70 80 90 100 HEIGHT, 'Yo OF CONTROL 110 Fig. 5. Relationship of yield and height 7 wk after treatment for midbloom dicamba DMA treatments in the formulation study (data points and--· y = -0.0406x2 +8.4Sx -342, r=0.84). Yield and height of untreated controls were 2490 kg ha·• and 0.8 m respectively. Dashed lines indicate yield-height relationships for Williams soybean in 0.18 m rows treated prebloom in 1980 (5 wk after treatment,----) and 1981 (6 wk after treatment,·· · ·). rations and height reductions were greater in 0.18 m than in 0.76 m rows (data not shown). Field Evaluation of Drift Injury The data demonstrate that, for a given treatment in a given year, height reduction and plant morphology are good predictors of yield reductions from dicamba injury. Results of all three studies were compar{:d to develop tentative criteria for the evaluation of drift injury (Table 4). Most foliar symptoms (i.e., crinlding and cupping of the terminal leaves, leaf margin injury and size reduction, and distorted venation patterns) were not indicative of reductions in yield. With the exception ofElfsoybean treated at the midbloom stage, there was no yield reduction without height reduction, ER 223 (248 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 115 of 245 UNGAR ET AL.: FLORAL BUD DEVELOPMENT IN COTTON regardless of any foliar symptoms present. Since Elf is a determinate cultivar, growth in height is essentially complete at the midbloom stage. For all treatments, severe injury symptoms (i.e., terminal bud kill, splitting of the stem, swelling of the petioles, and curled, malformed pods) were associated with substantial reductions in yield. We feel this approach is useful for evaluating dicamba drift injury to soybeans. Further experiments should be undertaken to validate it under a wider range of environmental conditions. ACKNOWLEDGMENTS Statistician Bert Bishop of the Ohio Agricultural Research and Development Center provided helpful suggestions and advice on data analysis. 643 REFERENCES I. Auch, D.E., and W.E. Arnold. 1978. Dicamba use and injury on soybeans (Glycine max) in South Dakota. Weed Sci. 26:471-475. 2. Behrens, R., and W.E. Lueschen. 1979. Dicamba volatility. Weed Sci. 27:486-493. 3. Dedolph, R.R. 1960. A suggested method of handling data obtained with an exponential (variable dosage) sprayer. Proc. Am. Soc. Hortic. Sci. 75:789-798. 4. Friesen, G. 1958. The use of variable dosage sprayers in weed control research. Can. J. Plant Sci. 38:300-306. 5. Petersen, R.G. 1977. Use and misuse of multiple comparison procedures. Agron. J. 69:205-208. 6. Pfeiffer, R., R.T. Brunskill, and G.S. Hartley. 1955. A variable dosage sprayer for agricultural experiments. Nature 176:472-473. 7. Selleck, G.W. 1958. Note on the place of a logarithmic sprayer in testing herbicides for weed control. Can. J. Plant Sci. 38(2):270273. 8. Smith, R.J., Jr. 1965. Effect of chlorophenoxy herbicides on soybeans. Weeds 13:168-169. 9. Wax, L.M., L.A. Knuth, and F.W. Slife. 1969. Response of soybeans to 2,4-D, dicamba and picloram. Weed Sci. 17:388-393. ER 224 (249 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 116 of 245 Soybean Foliage Residues of Dicamba and 2,4-D and Correlation to Application Rates and Yield Shane M. Andersen, S. A. Clay,* L. J. Wrage, and D. Matthees Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved. ABSTRACT 1979; Weidenhamer et al., 1989; Kelley et al., 2002). Small amounts of PGR herbicides left in spray tanks after treating labeled crops also can result in soybean injury. Injury symptoms include leaf cupping, stunting, death of the apical bud, and malformations of the stem (Fribourg and Johnson, 1955; Auch, 1977; Behrens and Lueschen, 1979; Al-Khatib and Peterson, 1999). In addition, yield loss due to PGR exposure can be substantial under some conditions. Low detection levels are needed to document PGR herbicide contamination due to the low concentrations that can cause soybean injury. In addition, sampling for residue frequently occurs long after herbicide exposure. Concentrations may be low due to volatilization losses from the plant leaf, dilution due to plant growth, and/or degradation of the herbicide within the growing plant. Extraction of PGR herbicides from plant tissue requires acidification along with alkaline hydrolysis to remove free, bound, and conjugated forms of the herbicide (Yip and Ney, 1966; Chow et al., 1971). Detection and quantification of PGR residue in tissue extract has been problematic due to poor sensitivity and background interference when using gas chromatography (GC) and electron capture (ECD) techniques (Marquardt and Luce, 1961; Yip, 1962; Lorah and Hemphill, 1974), with typical detection levels ranging from 0.05 to 2 ␮g gϪ1. Detection limits have been lowered by about 10-fold (to 0.005 ␮g gϪ1) when using a GC coupled with mass spectrometry (MS) and selective ion monitoring (SIM) because much of the background interference is eliminated and confirmation ions of each herbicide are monitored. Documenting soybean injury and yield loss from PGR herbicides typically involves describing plant symptoms and their extent in the field, analyzing vegetative material for residue, and quantifying yield losses in areas of suspected exposure. The advancements in detection and quantification may allow for detection of PGR residues long after exposure to very low levels of these herbicides. However, the relationship between the amount recovered and plant yield is tenuous. The objectives of this study were to quantify the amount of dicamba and 2,4-D in soybean foliage 0, 6, 12, 24, and 48 DAT at the three-leaf (V3) stage of growth and determine if these concentrations were correlated to initial application rate, grain yield, or both. Plant growth regulator (PGR) herbicides dicamba (3,6-dichloro2-methyloxybenzoic acid) and 2,4-D [(2,4-dichlorophenoxy)acetic acid] can severely injure soybean [Glycine max (L.) Merr.] by drift or tank contamination and reduce yield. Often in regulatory disputes, tissue is analyzed for PGR residue. However, relationships between grain yield reduction and foliar residue concentrations at various times after exposure are not well documented. This 2-yr study quantified the amount of dicamba and 2,4-D in soybean foliage 0, 6, 12, 24, and 48 d after treatment (DAT) when treated with 1 to 20% of 0.56 kg a.e. haϪ1 [labeled rate for corn (Zea mays L.)] at the three-leaf (V3) stage of growth and determined if these concentrations were correlated to initial application rate or grain yield. Herbicide concentrations were determined using gas chromatography/mass spectrometry techniques with selective ion monitoring. Visual symptoms were slight (Ͻ10%) to severe (90%) and included leaf cupping, epinasty and, in some cases, death of the apical bud. Grain yields from dicamba-treated plants were reduced from 14 to 93% compared with untreated plant yield, whereas only 2,4-D at the highest rate reduced yield. In both years, foliar residue concentrations were correlated with initial application rates and yield reduction up to 24 DAT for dicamba and 12 DAT for 2,4-D, with all treatments having residue amounts similar to untreated plants after these intervals. The data suggest that plant samples should be collected as soon as possible after suspected PGR exposure for accurate detection and quantification of PGR residue. P lant growth regulator herbicides dicamba and 2,4-D are widely used for broadleaf weed control in corn, sorghum [Sorghum bicolor (L.) Moench], small grains, and pasture. These two herbicides consistently rank among the top 25 herbicides in annual usage in the United States (USEPA Office of Pestic. Progr., 2002). For example, dicamba was among the five most applied herbicides to corn in the USA during 2001, with 15% of all corn treated with an average of 0.17 kg a.e. haϪ1 dicamba (USDA National Agricultural Statistics Service, 2002). The active ingredient 2,4-D was applied to about 8% of corn in 2000 at a rate of 0.42 kg a.e. haϪ1 whereas in spring and winter wheat (Triticum spp.), 2,4-D was applied to 45 and 13% of these crops in 2000, respectively. Soybean is often placed in a rotation with corn and wheat and is highly sensitive to PGR herbicides. The close proximity of soybean fields to areas treated with PGR herbicides increases the risk for soybean exposure by off-target movement from field applications due to particle drift or volatilization (Behrens and Lueschen, S.M. Andersen, S.A. Clay, and L.J. Wrage, Plant Sci. Dep., and D. Matthees, Dep. of Chem. and Biochem., South Dakota State Univ., Brookings, SD 57007. Received 4 September 2003. *Corresponding author (sharon_clay@sdstate.edu). Abbreviations: COC, crop oil concentrate; DAT, days after treatment; ECD, electron capture detector; GC, gas chromatography; GDD, growing degree days; MS, mass spectrometry; PGR, plant growth regulator; SIM, selective ion monitoring; VCRR, visual crop response rating. Published in Agron. J. 96:750–760 (2004).  American Society of Agronomy 677 S. Segoe Rd., Madison, WI 53711 USA 750 ER 225 (250 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 117 of 245 ANDERSEN ET AL.: DICAMBA AND 2,4-D RESIDUE AFFECT SOYBEAN MATERIALS AND METHODS Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved. Site Description Field experiments were conducted at the Southeast Research Station (SE farm) near Beresford, SD, in 2001 and at the Brookings Agronomy Farm, Brookings, SD, in 2002. Soil at the SE farm was an Egan silty clay, 0 to 2% slope (finesilty, mixed, superactive, mesic Udic Haplustoll), with a sand, silt, and clay content of 180, 420, and 400 g kgϪ1, respectively, and a pH of 6.6. Soil at Brookings was a Vienna clay loam, 2 to 6% slope (fine-loamy, mixed, superactive, frigid Calcic Hapludoll), with a sand, silt, and clay content of 420, 280, and 300 g kgϪ1, respectively, and a pH of 6.7. Daily temperature and precipitation data for 1 May through 30 September of 2001 and 2002 were obtained from the South Dakota Cooperative Extension Service weather website and were used to calculate monthly averages. In addition, daily temperatures were used to calculate growing degree days (GDD): GDD ϭ ⌺{[Max. Daily Temp. (ЊC) ϩ Min. Daily Temp. (ЊC)]/2} Ϫ Base Temp. (ЊC) A base temperature of 10ЊC and a ceiling temperature of 30ЊC were used in the calculations. The monthly and seasonal averages were compared with the 30-yr averages (1961–1990) obtained from the NRCS National Water and Climate Center website. Plot Preparation and Maintenance The seedbed was tilled to a depth of about 10 cm with two passes of a field cultivator. Prairie Brand1 (‘PB1901RR’) (Maturity Group 1.9) soybean was planted at the SE farm on 29 May 2001. Asgrow (‘AG1301RR’) (Maturity Group 1.3) soybean was planted at Brookings on 21 May 2002. Seeding rate was 419 900 seeds ha–1, and planting depth was about 2.5 cm. Plots were maintained weed-free using a combination of herbicides, cultivation, and hand weeding. At both locations, sulfentrazone {N-2,4-dichloro-5[4-(dihydro-3-methyl-5-oxo-1H-1,2, 4-triazol-1-yl)phenyl]methanesulfonamide} (281 g a.i. haϪ1) plus cloransulam-methyl {3-chloro-2-[[(5-ethoxy-7-fluoro[1,2,4]triazolo[1,5-c]pyrimidin-2yl)sulfonyl]amino]benzoic acid, methyl ester} (36 g a.i. haϪ1) were applied 2 d after planting to control broadleaf weeds. At Brookings, S-metolachlor [2-chloro-N(2-ethyl-6-methylphenyl)-N-(2-methoxy-1-methylethyl)acetamide] (2.1 kg a.i. haϪ1) also was applied for grass control. All maintenance herbicides were applied with an air-pressurized tractor-mounted sprayer equipped with flat-fan nozzles spaced 51 cm apart that delivered 187 L haϪ1 at 290 kPa at 4.5 km hϪ1. Plots at the SE farm were cultivated to a depth of 5 cm 30 d after planting. Hand weeding was used throughout the season as needed. Treatments and Experimental Design Treatments were arranged in a randomized complete block design with four replications and included an untreated control and herbicide applications that ranged from 1 to 20% of a labeled rate for corn. The rates of the diglycolamine salt of dicamba were 0.0056, 0.0112, and 0.056 kg a.e. ha–1, which corresponded to 1, 2, and 10% of a 0.56 kg a.e. haϪ1 label rate, respectively. The highest rate of dicamba was included in two treatments, one with a 1% volume per volume (v/v) rate of crop oil concentrate (COC) and the other without 1 The use of trade names is for the convenience of the reader only and does not imply endorsement by South Dakota State University. 751 COC. The rates of the dimethylamine salt of 2,4-D were 0.0112, 0.056, and 0.112 kg a.e. ha–1, which corresponded to 2, 10, and 20% of a 0.56 kg a.e. haϪ1 corn rate, respectively. Herbicides were applied to soybean at V3 growth stage (Ritchie et al., 1997) (3 July 2001 and 25 June 2002) with an air-pressurized bicycle-type sprayer equipped with six flat-fan nozzles spaced 51 cm apart and 46 cm above the crop. Delivery rate was 187 L haϪ1 at 290 kPa at 4.5 km hϪ1. Plots were 3 m wide (four 76-cm soybean rows) by 12 m long. Four soybean rows were left untreated between each plot as a buffer to limit herbicide drift among treatments. A 6-m untreated buffer also was left between replications. Plant Evaluation and Sampling Plant height, phenological development (Ritchie et al., 1997), and visual crop response rating (VCRR) (Behrens and Lueschen, 1979) (Table 1) were recorded just before herbicide application, 4 h after treatment, and 6, 12, 24, and 48 DAT. Plant height from the soil surface to the top node of the main stem was determined on six random plants from the outer two rows of each plot. These plants were clipped at soil level and placed in sealed polyethylene bags on ice. Samples were separated according to application rate and herbicide, transferred to several freezers to prevent possible contamination among treatments, and stored at –20ЊC until analyzed for herbicide residue. At phenological maturity of the untreated control, maturity index [days earlier (–) or later (ϩ) than the untreated control] and lodging score (based on average erectness of the main stem of 12 plants within each treatment) were estimated. The center two rows of each plot were combined for grain yield using a plot combine when the seeds were at 15% moisture or less. Yields were calculated on a 13% moisture content basis and expressed as kilograms per hectare. Herbicide Residue Extraction and Detection A 25-g subsample (fresh weight) of each soybean sample was cut into 1-cm lengths, blended at high speed for 5 min with 200 mL of 0.1 M NaOH, and filtered. The volume of filtrate was measured, transferred to a separatory funnel that contained 25 g of NaCl, acidified to pH 1 using 3 M H2SO4, and partitioned with CH2Cl2. After centrifugation, the aqueous layer was poured off and discarded. The organic layer was filtered through phase-separation filter paper and condensed to about 1 mL using a rotary evaporator. This solution was methylated with diazomethane, transferred to a Florisil column (Alltech Assoc., Deerfield, IL) prewet with 3% acetone in hexane, eluted with 40 mL of the mobile phase, and evaporated under N2 to about 5 mL, and the volume was increased to 10 mL with hexane. Herbicide residues were quantified using a GC/MS (Model 5890 GC plus a 5971 series massselective detector, Hewlett-Packard, Wilmington, DE) using the SIM data acquisition mode monitoring at m/z 234 and 236. Injection volumes were 2 ␮L. Average recovery of each herbicide from fortified plant tissue was 88%. This method determined bound as well as free forms of both herbicides, with a detection limit of 0.001 ␮g chemical residue per gram fresh plant material. Residue values were calculated on a microgram chemical residue per gram dry plant material basis by drying the remaining plant sample at 65ЊC for 48 h, weighing, and correcting fresh weight for plant water content. Data Analyses All plant, grain, and residue data were analyzed using Statistical Analysis Systems software (SAS Inst., 1990). Significant ER 226 (251 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 118 of 245 752 AGRONOMY JOURNAL, VOL. 96, MAY–JUNE 2004 Table 1. Injury levels and corresponding symptoms used to rate plant growth regulator herbicide injury to soybean as developed and reported by Behrens and Lueschen (1979). Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved. Injury level 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Symptoms No effect, plants normal Slight crinkle of leaflets of terminal leaf Cupping of terminal leaflets, slight crinkle of leaflets of second leaf, growth rate normal Leaflets of two terminal leaves cupped, expansion of terminal leaf suppressed slightly Malformation and growth suppression of two terminal leaves, terminal leaf size less than one-half that of control No expansion of terminal leaf, second leaf size one-half or less that of control Slight terminal growth, vigorous, malformed axillary shoot growth developing Terminal bud dead, substantial, strongly malformed, axillary shoot growth Limited axillary shoot growth, leaves present at time of treatment chlorotic with slight necrosis Plants dying, leaves mostly necrotic Plants dead differences among treatment means were determined using analysis of variance (ANOVA) with the GLM procedure at a significance level of P Ͼ 0.05 unless otherwise noted. Treatment differences are reported as a least significant difference (LSD) (Steel et al., 1997). Regression analysis was used to quantify the relationship between initial application rates and residue concentrations at each sampling date. A correlation procedure (SAS Inst., 1990) was used to determine the Pearson correlation coefficients between residue concentrations at each sampling date and yield. RESULTS A third study site was located at the Northeast Research Farm (Watertown, SD) in 2002, and results for injury and yield loss are similar to data reported herein (Andersen, 2003). Due to budget constraints, residue analyses were not done for this site. Climate Data In general, the 2001 growing season at the SE farm had average temperatures with total GDD accumulation about 4% above the 30-yr average of 1467 (data not shown). Rainfall over the growing season was 18% below the 30-yr average (42.2 cm). Rainfall in July (just after application) was 33% above and in August was 56% below the 30-yr average. The 2002 growing season at Brookings was warm, with total GDD accumulation about 16% above the 30-yr average of 1220. The season was very dry (23% below the 30-yr average of 40.3 cm). Rainfall in July (just after application) was 93% less than average whereas in August, rainfall was 56% above average. Plant Injury Soybean injury was similar each year, and VCRR ranged from 5% (slight leaflet malformations) to 90% (necrosis of all leaves), depending on treatment and sampling date (Tables 1 and 2). Plant injury was less severe with 2,4-D than dicamba (Table 2). Wax et al. (1969) also reported greater soybean injury with dicamba than 2,4-D at equal exposure rates. The addition of COC to dicamba generally resulted in higher injury ratings, especially 6 and 12 DAT. At 6 DAT, injury symptoms of dicamba-treated plants included shoot and petiole epinasty, cupping, and marginal chlorosis of terminal leaflets and were similar to those reported by others (Wax et al., 1969; Auch and Arnold, 1978; Behrens and Lueschen, 1979). At 12, 24, and 48 DAT, injury was slightly greater than the injury observed 6 DAT (Table 2). However, apical meristems died, and with this loss of apical dominance, lower leaf axillary buds were released, resulting in significant amounts of lateral branching. This response also has been reported in other dicamba studies (Wax et al., 1969; Weidenhamer et al., 1989). Trifoliates of the lateral branches were cupped and distorted with an oblong Table 2. Visual crop response rating (VCRR) of soybean treated at the V3 stage of soybean growth with several sublethal rates of dicamba and 2,4-D. VCRR SE farm (2001) Brookings (2002) Days after treatment Treatment Rate 6 12 24 0 30 35 80 85 5 20 35 3 0 40 45 85 90 5 25 35 5 0 40 50 85 90 5 10 30 4 haϪ1 Check Dicamba‡ Dicamba Dicamba Dicamba ϩ COC§ 2,4-D¶ 2,4-D 2,4-D kg a.e. 0 0.0056 0.0112 0.056 0.056 ϩ 1% (v/v) 0.0112 0.056 0.112 LSD(0.05) † Plant injury rated according to Behrens and Lueschen (1979). ‡ Dicamba applied as the diglycolamine salt formulation. § COC, crop oil concentrate. ¶ 2,4-D applied as the dimethylamine salt formulation. ER 227 48 6 injury rating, %† 0 0 40 40 55 50 90 80 90 90 5 5 10 20 30 30 5 3 12 24 48 0 45 50 85 90 5 20 30 4 0 45 50 90 90 5 10 25 4 0 45 55 90 90 5 10 30 4 (252 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 119 of 245 753 ANDERSEN ET AL.: DICAMBA AND 2,4-D RESIDUE AFFECT SOYBEAN Table 3. Plant biomass (g plantϪ1) of soybean treated at the V3 stage of soybean growth with several sublethal rates of dicamba and 2,4-D. Plant biomass SE farm (2001) Brookings (2002) Days after treatment Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved. Treatment Rate 0 6 12 24 haϪ1 Check Dicamba† Dicamba Dicamba Dicamba ϩ COC‡ 2,4-D§ 2,4-D 2,4-D kg a.e. 0 0.0056 0.0112 0.056 0.056 ϩ 1% (v/v) 0.0112 0.056 0.112 LSD(0.05) 1.32 1.35 1.44 1.33 1.30 1.63 1.49 1.44 NS 2.49 2.04 2.39 1.94 1.69 2.44 2.10 1.89 0.53 4.31 3.46 3.25 2.68 2.74 4.22 3.81 3.01 0.69 9.92 5.92 5.46 4.60 3.63 9.04 6.86 6.78 1.47 48 g 22.25 13.98 12.76 8.52 7.62 17.87 17.13 15.60 4.56 0 6 12 24 48 1.11 1.33 1.33 1.22 1.19 1.28 1.13 1.14 NS 2.31 1.97 2.02 1.56 1.56 1.79 1.54 1.30 0.50 3.44 3.12 2.64 1.66 1.48 2.69 2.56 2.51 0.45 5.02 5.41 4.01 2.56 2.53 5.68 4.08 4.94 0.81 11.91 8.10 7.62 8.84 9.43 11.55 7.71 7.87 3.67 plantϪ1 † Dicamba applied as the diglycolamine salt formulation. ‡ COC, crop oil concentrate. § 2,4-D applied as the dimethylamine salt formulation. shape and chlorotic tips, similar to the symptoms of the terminal growth at 6 DAT. At higher rates, the point of petiole attachment to the main stem was weak and easily broken. Plants treated with higher rates of dicamba had small, curled, and malformed pods 48 DAT. Soybean biomass (Table 3) and plant height (Andersen, 2003) were reduced by dicamba compared with the check at 12, 24, and 48 DAT. Reductions in both parameters generally were greater as the dicamba rate increased, with maximum reductions of 70 and 66% for height and biomass, respectively. Visual injury symptoms caused by 2,4-D were similar to those reported by others (Slife, 1956; Rojas-Garciduenas and Kommedahl, 1958; Smith, 1965; Kelley et al., 2002). Unlike dicamba symptoms that were not noted until 6 DAT, 2,4-D injury was noticeable as leaf epinasty 4 h after treatment. At 6 DAT, shoot and petiole epinasty near the apical meristem had increased to greater than a 90-degree angle from the main stem. New trifoliate growth appeared strapped with parallel venation. At 12 DAT, main stems and petioles had upright growth. About 20% of plants treated with the lowest 2,4-D rate had growth from the unifoliate axil, indicating apical meristem release, but branching was less than that of plants treated with an identical rate of dicamba. A bend in the lower portion of the main stem developed in plants treated with the two highest 2,4-D rates and was attributed to epinasty that occurred immediately after treatment. The bend persisted throughout the remainder of the growing season. Callusing and cracking of the lower 12 cm of the main stem also were observed at the highest 2,4-D rate. Maximum height reduction from 2,4-D occurred at 6 and 12 DAT whereas maximum dicamba height reduction occurred at 24 and 48 DAT (Andersen, 2003). Plants treated with 2,4-D were better able to recover from early-season stunting than those treated with dicamba. Maximum biomass reduction (32%) occurred at 24 DAT with the 0.112 kg a.e. ha–1 2,4-D treatment at the SE farm (Table 3). All treatments of dicamba and 2,4-D delayed vegetative plant development and maturity (data not shown), with dicamba-treated plants being more affected. Soybean planted at SE farm was from Maturity Group 1.9 whereas Group 1.3 was planted at Brookings. However, maturity delay was similar at both locations. The high rate of dicamba ϩ COC virtually stopped vegetative development of soybean plants up to 24 DAT. Treated plants, except those treated with 0.056 kg a.e. haϪ1 dicamba, reached the reproductive stages at similar times as the untreated plants although maturity was delayed about 7 d in dicamba treatments and about 1 d in 2,4-D treatments. Auch and Arnold (1978) noted a maturity delay of 12 d when 0.056 kg a.e. ha–1 dicamba was applied to soybean in the early-bloom stage. Wax et al. (1969) applied simulated drift-type rates of 2,4-D and dicamba at prebloom (V3) and also reported that dicamba delayed maturity more than 2,4-D. Plant lodging was not prevalent with any treatment (Andersen, 2003). Soybean Grain Yield All rates of dicamba reduced yield (Table 4), with reductions ranging from 14 to 93%. Maximum reductions occurred with the 0.056 kg a.e. ha–1 ϩ COC treatment. Yield and initial dicamba rate were highly negatively correlated [r ϭ –0.98 (2001) and –0.94 (2002); p ϭ 0.01]. Only the 0.112 kg a.e. ha–1 rate of 2,4-D reduced yield. Weidenhamer et al. (1989) stated that soybean generally is able to tolerate considerable early-season foliar injury without reducing yield. Al-Khatib and Peterson (1999) noted that visual injury ratings of dicambatreated soybean were always greater than yield loss. The loss of apical dominance and release of axillary buds that produce new branches and, eventually, flowers and seed pods (Moore, 1979) can compensate for a portion of the expected yield loss when soybean is exposed to PGR herbicides. Plant Growth Regulator Herbicide Residue Before application and throughout the sampling period each season, plants from the untreated check had low concentrations of PGR residue present in foliage (Table 5). Other studies have reported positive PGR ER 228 (253 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 120 of 245 754 AGRONOMY JOURNAL, VOL. 96, MAY–JUNE 2004 Table 4. Soybean yield after treatment at the V3 stage of soybean growth with sublethal rates of dicamba and 2,4-D during 2001 and 2002. SE farm (2001) Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved. Treatment Check Dicamba† Dicamba Dicamba Dicamba ϩ COC‡ 2,4-D§ 2,4-D 2,4-D Brookings (2002) Rate Grain yield Yield reduction Grain yield Yield reduction kg a.e. haϪ1 0 0.0056 0.0112 0.056 0.056 ϩ 1% (v/v) 0.0112 0.056 0.112 LSD(0.05) kg haϪ1 3097 2663 2670 884 605 3109 2874 2114 313 % 0 14.0 13.8 71.5 80.5 0 7.2 31.7 kg haϪ1 2567 1708 1505 426 175 2510 2381 1933 283 % 0 33.5 41.4 83.4 93.2 2.2 7.2 24.7 † Dicamba applied as the diglycolamine salt formulation. ‡ COC, crop oil concentrate. § 2,4-D applied as the dimethylamine salt formulation. residue detection on control samples (Cessna, 1980; Hemphill and Montgomery, 1981; Smith, 1984). In addition, 10 random soybean samples taken in 2000 from fields that had no visual PGR injury symptoms had 2,4-D detections ranging from 0.009 to 0.036 ␮g gϪ1 and three had positive dicamba detections (unpublished data, 2000). The herbicide source is unknown but was not due to laboratory contamination because samples having no PGR herbicide detections were obtained. Contamination before application may have been due to drift or volatilized chemical deposition from areas outside of the treatment sites (Behrens and Lueschen, 1979). Herbicide residue concentrations of both dicamba and 2,4-D were greater throughout most of 2002 compared with concentrations measured in 2001 (Table 5). Herbicide treatments in both years were applied at V3; however, plant biomass was greater (2–32%) in 2001 than in 2002 (Table 3) due to the dry conditions in 2002. If the amount of residue per whole plant is compared, residue levels for the 2 yr are similar. Foliar residue concentration of PGRs dropped quickly over time (Table 5). It is unclear if the decrease was due to metabolism, dilution as the plant grew, or both. In an extensive literature search, the rate of dicamba metabolism in soybean was not found. Dicamba metabolism in tartary buckwheat [Fagopyrum tataricum (L.) Gaertn.] was slow, with only 10% of the herbicide detoxified 20 DAT whereas 50% of the dicamba was detoxified 1 DAT in wheat (Chang and Vanden Born, 1971). Metabolism of 2,4-D in tomato (Lycopersicon spp.) was very slow, with characteristic injury symptoms to transplanted buds still evident 60 DAT when grafted onto treated plants (Muzik and Whitworth, 1960). These data from other sensitive plants would suggest that the decrease in soybean was due to dilution of the herbicide during plant growth and not metabolism. The decrease in foliar residues of both dicamba and 2,4-D agrees with results reported by Auch and Arnold (1978) although they reported no detection of dicamba residue after 7 d, most likely due to the less sensitive detection limit of the GC/ECD system. Correlations of application rate to residue level and residue level to yield reduction were similar each year. Only 2001 data are presented here although data for 2002 parameters are reported in Andersen (2003). Dicamba foliar concentrations were correlated with application rate up to 24 DAT, whereas 2,4-D residue amounts were correlated with application rate up to 12 DAT (Fig. 1 and 2). The addition of COC to the high Table 5. Foliar residue concentrations of soybean treated at the V3 stage of growth with several sublethal rates of dicamba and 2,4-D at the SE farm (2001) and Brookings (2002). Foliar herbicide residue SE farm (2001) Brookings (2002) Days after treatment Treatment Rate 0 6 12 kg a.e. haϪ1 Check Dicamba† Dicamba Dicamba Dicamba ϩ COC‡ Check 2,4-D§ 2,4-D 2,4-D 0.0056 0.0112 0.056 0.056 ϩ 1% (v/v) LSD(0.05) 0.142 2.951 4.831 21.401 25.463 5.01 0.249 0.583 0.969 7.207 9.709 1.43 0.153 0.187 0.301 2.230 4.358 1.14 0.0112 0.056 0.112 LSD(0.05) 0.592 5.438 16.069 39.519 4.10 0.620 0.628 1.455 4.923 1.66 0.304 0.444 0.405 1.703 0.48 † Dicamba applied as the diglycolamine salt formulation. ‡ COC, crop oil concentrate. § 2,4-D applied as the dimethylamine salt formulation. ER 229 24 48 0 6 ␮g dicamba/g dry plant material 0.041 0.047 0.068 0.093 0.034 0.030 2.882 0.646 0.057 0.017 5.259 1.099 0.120 0.031 26.08 18.59 0.173 0.033 30.20 15.64 0.09 0.03 7.94 4.56 ␮g 2,4-D/g dry plant material 0.134 0.150 0.211 0.550 0.076 0.073 4.806 1.010 0.100 0.082 23.78 4.761 0.159 0.084 59.86 18.507 0.08 0.11 18.46 4.12 12 24 48 0.049 0.393 0.733 7.342 12.17 5.21 0.091 0.129 0.216 2.536 4.337 0.76 0.075 0.030 0.020 0.053 0.062 0.04 0.325 0.592 1.437 3.446 0.92 0.514 0.261 0.223 0.424 0.23 0.393 0.080 0.069 0.071 0.16 (254 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 121 of 245 755 ANDERSEN ET AL.: DICAMBA AND 2,4-D RESIDUE AFFECT SOYBEAN 35 30 ~ 25 r = 0.96 (p= O.Ol) ODAT I 20 • 15 10 Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved. 5 o ~----~----~----~----~------~--~(~A) 12 ~--------------------------------------------- 101 8 ~ 6 J r = 0.95 (p = O.OI) 6DAT § 0 4 2 o ~~~~----~----~----~------~--~(~B) 7 ~------------------------------------- 6 s r = 0.85 (p = O.OI) 12DAT 0 G 4 3 2 o~~~~QO____~------~----~------~----o~(C~) I o .3 5 r-=:---------------------~ 0.30 0.25 r - 0.70 (p = O.Ol) 24DAT • • • 0.1 5 0.1 0 ~------1-·-------------------!·!l 0.05 i • • • 0.20 o.oo -t-----,------,-------,,------------,------.--~·~(D~) Dicamba rate applied (kg ae/ha) Fig. 1. Correlation of dicamba residue with original dicamba application rate at (A) application [0 d after treatment (DAT)] and (B) 6, (C) 12, (D) 24, and (E) 48 DAT. Data collected at the SE farm, Beresford, SD, in 2001. ER 230 (255 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 122 of 245 756 AGRONOMY JOURNAL, VOL. 96, MAY–JUNE 2004 50 r = 0.98 (p = O.Ol) 40 • 0 OAT 30 20 Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved. 10 (A) 0 10 r = 0.86 (p = O.Ol) 8 60AT 0 6 4 2 ,.-.., ~ -c (B) 0 ~ ~ E ~ Q. 3.0 r=0.81 (p = O.Ol) 2.5 c: 2.0 -- 1.5 "0 !OJ) 120AT 0 0 !OJ) :t ._., c .s -.."' '; ,_ c 1.0 @ G le • • • • 60 • • • r = 0.2 1 (p = NS) 30 (E) 48DAT 0 0.00 0.05 0. 10 0.15 0.20 0.25 0.30 0.35 2,4-D concentJ·ation (J..Lg/g dry plant material) Fig. 4. Correlation of yield with 2,4-D residue in plant at (A) 0, (B) 6, (C) 12, (D) 24, and (E) 48 d after treatment (DAT). Data collected at the SE farm, Beresford, SD, in 2001. ER 233 (258 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 125 of 245 Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved. ANDERSEN ET AL.: DICAMBA AND 2,4-D RESIDUE AFFECT SOYBEAN rate of dicamba increased residue amounts at 0 DAT and resulted in higher residue levels up to 24 DAT (Table 5). Crop oil concentrate generally enhances the absorption of herbicides by increasing movement through leaf cuticles and may reduce herbicide losses due to volatilization and photodecomposition (Jansen et al., 1961). Volatilization losses of dicamba can be considerable (Behrens and Lueschen, 1979) although the diglycolamine salt formulation is less volatile than the amine formulation. Yield loss was correlated with dicamba foliar concentrations from 0 to 24 DAT (r Ն –0.69; p ϭ 0.01) (Fig. 3). Although yield reductions were not reported to be statistically significant for most 2,4-D treatments, the measured reductions were correlated to foliar 2,4-D concentrations up to 12 DAT (r Ն –0.83; p ϭ 0.01) (Fig. 4). 759 possible after suspected exposure to accurately assess the amount of chemical exposure. If possible, healthy soybean plants from the area also should be collected at the same time for residue analysis for comparison. Differences among soybean residue PGR concentrations from the lowest exposure rates and untreated material were indistinguishable 6 DAT; however, plants treated with low dicamba rates suffered yield loss whereas plants treated with 2,4-D did not. At higher exposure rates, differences between treated and background residue levels could be distinguished up to 24 DAT. Collecting samples after 24 DAT may be of little or no value since the analysis is costly and there is no correlation between residue levels and yield. ACKNOWLEDGMENTS CONCLUSIONS Residue analysis of soybean plants exhibiting PGR herbicide symptomology is requested frequently in regulatory investigations as a means of substantiating the cause of injury and possible source. Yet there is little information on the relationship between the amount of residue detected in soybean tissue at various sampling dates and soybean injury or yield reduction. This lack of information is most likely due to quantification techniques that lacked sensitivity to very low levels of PGR in plant tissues. However, now care must be taken to accurately interpret very low residue amounts. For example, before application, plants from the untreated control had some PGR residue present, and at 24 and 48 DAT, residue concentrations were similar among all treatments. In this study, PGR exposure occurred at V3 stage of soybean growth, when a large percentage of PGR herbicides are applied to labeled crops. Results suggest that to document PGR injury, suspected exposure areas should be scouted, noting the plant symptomology and severity of the injury. The most severe dicamba injury was noted between 12 to 48 DAT whereas the most severe 2,4-D injury occurred 6 DAT. Other studies have reported more severe soybean injury and yield reduction when PGR herbicides are applied at V5 or later growth stages (Slife, 1956; Wax et al., 1969; Auch and Arnold, 1978); however, the possibility of PGR exposure in this study area would be greatly reduced due to the earlier timing of these applications. Environmental conditions also affected plant injury and ultimately grain yield. Both parameters were more severe during 2002, a very dry year, than during 2001 when water was not limiting. Auch and Arnold (1978) also reported increased injury and yield reduction due to PGR exposure during dry years. All leaflets and petioles of the plant were taken for residue analysis in this study. This system was clear-cut and less ambiguous than collecting the top one-third of growth or only affected leaflets (a method that could not be used for asymptomatic plants). Different sampling methods may lead to different data interpretation. However, when all leaflets and petioles were collected for residue analysis, samples had to be collected as soon as This research was partially funded by USDA, South Dakota Agricultural Experiment Station, and South Dakota Department of Agriculture. South Dakota Agricultural Experiment Station Manuscript no. 3386. REFERENCES Al-Khatib, K., and D. Peterson. 1999. Soybean (Glycine max) response to simulated drift from selected sulfonylurea herbicides, dicamba, glyphosate, and glufosinate. Weed Technol. 13:264–270. Andersen, S.M. 2003. Analysis of soybean [Glycine max (L.) Merr.] response to simulated drift rates of PGR herbicides using a foliar residue test. M.S. thesis. South Dakota State Univ., Brookings. Auch, D.E. 1977. Dicamba use in relationship to drift injury on soybeans in southeastern South Dakota. M.S. thesis. South Dakota State Univ., Brookings. Auch, D.E., and W.E. Arnold. 1978. Dicamba use and injury on soybean (Glycine max) in South Dakota. Weed Sci. 26:471–475. Behrens, R., and W.E. Lueschen. 1979. Dicamba volatility. Weed Sci. 27:486–493. Cessna, A.J. 1980. Simultaneous extraction and detection of residues of (2,4-dichlorophenoxy) acetic acid and bromoxynil from wheat. J. Agric. Food Chem. 28:1229–1232. Chang, F.Y., and W.H. Vanden Born. 1971. Dicamba uptake, translocation, metabolism, and selectivity. Weed Sci. 19:113–117. Chow, C., M.L. Montgomery, and T.C. Yu. 1971. Methodology and analysis for residues of MCP and 2,4,5-T in wheat. Bull. Environ. Contam. Toxicol. 6:576–580. Fribourg, H.A., and I.J. Johnson. 1955. Response of soybean strains to 2,4-D and 2,4,5-T. Agron. J. 47:171–174. Hemphill, D.D., Jr., and M.L. Montgomery. 1981. Response of vegetable crops to sublethal application of 2,4-D. Weed Sci. 29:632–635. Jansen, L.L., W.A. Genter, and W.C. Shaw. 1961. Effects of surfactants on the herbicidal activity of several herbicides in aqueous spray systems. Weeds 9:381–405. Kelley, K.B., L.M. Wax, A.G. Hager, and D.E. Riechers. 2002. Soybean response to simulated drift and spray tank contamination of plant growth regulator herbicides. In Abstr., Annu. Meet., North Cent. Weed Sci. Soc., 57, St. Louis, MO [CD-ROM]. 2–5 Dec. 2002. North Cent. Weed Sci. Soc.,Champaign, IL. Lorah, E.J., and D.D. Hemphill. 1974. Direct chromatography of some N-methylcarbamate pesticides. J. Assoc. Off. Anal. Chem. 57:570– 575. Marquardt, R.P., and E.N. Luce. 1961. A new basic procedure for determining phenoxy acid herbicides in agricultural products. J. Agric. Food Chem. 9:266–270. Moore, T.C. 1979. Biochemistry and physiology of plant hormones. Springer-Verlag, New York. Muzik, T., and J.W. Whitworth. 1960. A biological method for determining persistence of chemicals in plant tissues. Plant Physiol. 35(suppl.):xxvi (abstr.). Ritchie, S.W., J.J. Hanway, H.E. Thompson, and G.O. Benson. 1997. How a soybean plant develops. Spec. Rep. 53. Rev. ed. Iowa State Univ. Coop. Ext. Serv., Ames. ER 234 (259 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 126 of 245 Reproduced from Agronomy Journal. Published by American Society of Agronomy. All copyrights reserved. 760 AGRONOMY JOURNAL, VOL. 96, MAY–JUNE 2004 Rojas-Garciduenas, M., and T. Kommedahl. 1958. The effects of 2,4diclorophenoxyacetic acid on radicle development and stem anatomy of soybeans. Weeds 6:49–51. SAS Institute. 1990. SAS user’s guide: Statistics. 4th ed. SAS Inst., Cary, NC. Slife, F.W. 1956. The effect of 2,4-D and several other herbicides on weeds and soybeans when applied as post-emergence sprays. Weeds 4:61–68. Smith, A.E. 1984. Gas chromatographic method for analysis of 2,4-D in wheat: Interlaboratory study. J. Assoc. Off. Anal. Chem. 67:794– 798. Smith, R.J., Jr. 1965. Effect of chlorophenoxy herbicides on soybeans. Weeds 13:168–169. Steel, G.D., J.H. Torrie, and D.A. Dickey. 1997. Principles and proce- dures of statistics: A biometrical approach. 3rd ed. McGraw-Hill, New York. USDA National Agricultural Statistics Service. Crop production annual summary, 2002. USDA Natl. Agric. Stat. Serv., Washington, DC. USEPA Office of Pesticide Programs. Pesticide industry sales and usage, 2002. USEPA Office of Pestic. Progr., Washington, DC. Wax, L.M., L.A. Knuth, and F.W. Slife. 1969. Response of soybeans to 2,4-D, dicamba, and picloram. Weed Sci. 17:388–393. Weidenhamer, J.D., G.B. Triplett, Jr., and F.E. Sobotka. 1989. Dicamba injury to soybean. Agron. J. 81:637–643. Yip, G. 1962. Determination of 2,4-D and other chlorinated phenoxy alkyl acids. J. Assoc. Off. Anal. Chem. 45:367–376. Yip, G., and R.E. Ney, Jr. 1966. Analysis of 2,4-D residues in milk and forage grasses. Weeds 14:167–170. ER 235 (260 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 127 of 245 Crop Protection 28 (2009) 539–542 Contents lists available at ScienceDirect Crop Protection journal homepage: www.elsevier.com/locate/cropro Soybean response to simulated dicamba/diflufenzopyr drift followed by postemergence herbicides Lynette R. Brown a, *, Darren E. Robinson a, Robert E. Nurse b, Clarence J. Swanton c, Peter H. Sikkema a a Department of Plant Agriculture, University of Guelph, Ridgetown Campus, 120 Main Street East, Ridgetown, Ontario N0P 2C0, Canada Agriculture and Agri-Food Canada, Greenhouse and Processing Crops Centre, 2585 County Road 20, Harrow, Ontario N0R 1G0, Canada c Department of Plant Agriculture, University of Guelph, Guelph, Ontario N1G 2W1, Canada b a r t i c l e i n f o a b s t r a c t Article history: Received 31 January 2008 Received in revised form 9 February 2009 Accepted 18 February 2009 Five field experiments were conducted in 2007 to determine the effect of simulated dicamba/diflu fenzopyr drift followed by postemergence applications of chlorimuron ethyl, imazethapyr or bentazon on soybean (Glycine max Merr.) crop injury, dry weight, height and yield. In the absence of a post emergence herbicide, as the dose of simulated dicamba/diflufenzopyr increased there was an increase in soybean injury and a decrease in dry weight, height and yield. The application of registered post emergence herbicides following simulated dicamba/diflufenzopyr drift resulted in a synergistic increase in crop injury in some environments. There was no synergistic response in respect to dry weight and height when simulated drift was followed by postemergence herbicides. A synergistic yield response was observed with yield being decreased 4 7% more than expected due to simulated dicamba/diflufenzopyr drift followed by the application of chlorimuron ethyl. No synergistic yield response was observed for dicamba/diflufenzopyr drift followed by either imazethapyr or bentazon. Ó 2009 Elsevier Ltd. All rights reserved. Keywords: Bentazon Chlorimuron-ethyl Dicamba/diflufenzopyr Imazethapyr Synergism 1. Introduction Maize (Zea mays) and soybean (Glycine max Merr.) are frequently planted in fields that are adjacent to one another. When these two crops are grown in close proximity, there is the potential for soybean injury due to herbicide drift from an adjacent maize field. Previous research conducted by Maybank et al. (1978) and Wolf et al. (1992) has determined that herbicide drift from unshielded sprayers can range from 1 to 16% depending on nozzle type, spray additives, boom height and wind velocity. Possible synergistic responses in respect to crop injury and yield from herbicide drift followed by the application of a registered post emergence herbicide have been postulated. The interaction of herbicides when applied either simulta neously or sequentially, may result in responses that are not predictable based on their response when applied alone. The interaction of herbicides in combination is synergistic if the actual effect is greater than the sum of the effects from the two herbicides applied individually (Gressel, 1990; Lich et al., 1997). These herbi cide combinations can cause a synergistic response that increases crop damage. An example of this was documented by Simpson and * Corresponding author. Tel.: þ1 519 674 1645; fax: þ1 519 674 1600. E-mail address: lbrown@ridgetownc.uoguelph.ca (L.R. Brown). Stoller (1996) who reported that individual applications of thifensulfuron (4.4 g a.i. haÀ1) and imazethapyr (70 g a.i. haÀ1) caused 0 and 28% injury, respectively in soybean, but the combi nation of both herbicides caused 50% injury. Dicamba/diflufenzopyr is a postemergence herbicide registered for broadleaf weed control in maize in Canada. Diflufenzopyr is an auxin transport inhibitor and dicamba causes irregular accumula tion of indoleacetic acid (Vencill, 2002) and stimulates ethylene production. Soybean injury symptoms due to dicamba/diflufenzo pyr drift appear as cupping and puckering of the leaves, twisted stems, shortened internodes, and a triangular shaped canopy. Chlorimuron ethyl and imazethapyr are registered for post emergence broadleaf weed control in soybean. They are aceto lactate synthase (ALS) inhibitors (Vencill, 2002) and are widely used due to their low mammalian toxicity, broad spectrum weed control, and flexibility of use on a wide variety of crops. Both chlorimuron ethyl and imazethapyr have the potential to cause some initial soybean injuries. Bentazon is a photosystem II inhibitor (Vencill, 2002) that provides annual broadleaf weed control in soybean. The objective of this research was to determine if simulated dicamba/diflufenzopyr drift followed by postemergence applica tions of chlorimuron ethyl, imazethapyr or bentazon has a syner gistic effect on soybean crop injury, dry weight, height and yield. 0261-2194/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.cropro.2009.02.004 ER 236 (261 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 128 of 245 540 L.R. Brown et al. / Crop Protection 28 (2009) 539–542 2. Materials and methods Five field experiments were established in 2007 at the Univer sity of Guelph, Elora, Ontario, at the Agriculture and Agri Food Canada Research Centre, Harrow, Ontario and at the University of Guelph Ridgetown Campus, Ridgetown, Ontario. At Elora and Ridgetown, trial areas were moldboard plowed in the fall and worked twice with a cultivator with rolling basket harrows in the spring to prepare the seedbed. At Harrow the seedbed was prepared by cultivation in the spring. At each location, the experiments were established as a randomized complete block design with four replications. Glyphosate resistant soybean were planted in 60 76 cm rows at a seeding rate of 400,000 480,000 seeds haÀ1 at Elora (DK27 02) on May 24, 2007, Harrow (DK31 52) on June 11, 2007, and Ridge town (DK30 07) on May 23 and May 24, 2007, into plots that were 2 by 7 m, 1.8 by 8 m, and 2 by 8 m, respectively. The soil at Elora was a silt loam with 31% sand, 50% silt, 19% clay, 4.2% organic matter, and a pH of 7.4. The soil at Harrow was sandy loam with 83% sand, 5% silt, 12% clay, 2.6% organic matter, and a pH of 6.0. The soil at two of the Ridgetown locations was a sandy clay loam with 52% sand, 26% silt, 21% clay, 5.3% organic matter, and a pH of 6.8. The soil at the third Ridgetown location was a sandy loam with 54% sand, 27% silt, 19% clay, 5.6% organic matter, and a pH of 6.4. The two herbicide treatments (simulated drift followed by the registered post emergence herbicide) at the Ridgetown sites were applied on June 15 and 18, June 22 and 25 and June 25 and 28. At Elora, the treat ments were made on June 23 and 25 and at Harrow, the treatments were applied on July 3 and 7. Plots were maintained weed free with s metolachlor/benoxacor plus glyphosate applied preemergence, glyphosate applied postemergence and hand hoeing as required. The sodium salt of dicamba in combination with diflufenzopyr (5:2 ratio) was applied to soybean at the two to three trifoliate stages at 0, 2, 10, 20 and 40 g a.i. haÀ1, representing 0, 1, 5, 10, and 20% of the recommended labeled dose, respectively, to simulate herbicide drift. Chlorimuron ethyl (9 g a.i. haÀ1), ammonium salt of imazethapyr (100 g a.i. haÀ1) or sodium salt of bentazon (1080 g a.i. haÀ1) was applied 2 4 days after the simulated dicamba/ diflufenzopyr drift application. Chlorimuron ethyl, imazethapyr and bentazon treatments included urea ammonium nitrate (UAN) at 2.0% v/v. A nonionic surfactant was added at 0.10% v/v to the chlorimuron ethyl treatments and at 0.25% v/v to the imazethapyr treatments. Herbicides were applied with a CO2 pressurized backpack sprayer equipped with 120 02 ultra low drift nozzles calibrated to deliver 200 L haÀ1 at 207 kPa at the Elora and Ridge town locations, and using flat fan 11004XR (Teejet Spraying Systems Co. Wheaton, IL) nozzles at Harrow. Crop injury was rated 7, 14, 28, and 56 days after application (DAA) with 0% indicating no crop injury and a rating of 100% indicating complete plant death. Average soybean height was determined by measuring the height of 10 plants from the soil surface to the top trifoliate leaf 28 DAA. Soybean dry weight was determined at 42 DAA by destructively harvesting 10 plants per plot at ground level and placing them into a paper bag. The plants were then dried at 60 C to a constant moisture, and the weight was recorded. Soybean grain yield was determined by harvesting the middle two rows of each plot with a small plot combine. Yields were adjusted to 13.0% moisture. All data were subjected to analysis of variance, and analyzed using the PROC MIXED procedure of SAS (Ver. 9.1, SAS Inst., Cary NC). To meet the assumptions of variance analyses, means of injury ratings 7 DAA (Ridgetown and Harrow) were transformed using log and square root transformations. Injury ratings 14 DAA (Ridge town) and 56 DAA were transformed using an arcsine square root transformation. Dry weight was transformed using a square root transformation. Means were back transformed to the original scale for presentation of results. Injury at 28 DAA, height and yield data met the assumptions of normality, therefore no transformations were necessary. The random effect of location and its interaction with herbicide treatments was significant for several of the vari ables analyzed. As a result, data for some parameters were reported by location. Means were separated using Fisher’s protected LSD at P < 0.05. Colby (1967) Equation (1) was used to determine the expected combination means by using the observed means for dicamba/diflufenzopyr (A) alone and the postemergence herbicide (B) alone. expected AþB A  B=100 (1) Yield, dry weight and height were calculated as a percent of the untreated check and Colby’s modified Equation (2) for percent of control values was used to determine the expected combination means. expected A  B=100 (2) Following the calculation of the expected means, observed versus expected means were compared at the 0.05 level of significance using a paired t test in order to determine synergistic or antago nistic responses. 3. Results and discussion 3.1. Crop injury Generally, as dicamba/diflufenzopyr dose increased, soybean foliar injury increased at 7 (data not shown), 14, 28, and 56 DAA (Tables 1 3). Dicamba/diflufenzopyr injury included cupping and crinkling of newly emerged leaves and twisting of the fully expanded leaves. Similar dicamba injury symptoms have been reported by Andersen et al. (2004), Kelley et al. (2005) and Wei denhamer et al. (1989). Chlorimuron ethyl, imazethapyr and ben tazon applied postemergence caused little injury to soybean. Combinations of the simulated dicamba/diflufenzopyr drift fol lowed by the postemergence herbicides resulted in synergistic responses in some environments. When dicamba/diflufenzopyr drift was simulated without the postemergence herbicide, there was an increase in soybean injury with increasing dose at all three locations 7 DAA (data not shown). Generally, there was no synergistic response from simulated dicamba/diflufenzopyr drift followed by the postemergence herbicides (chlorimuron ethyl, imazethapyr or bentazon) at the Harrow and Elora locations. In contrast, synergistic responses were observed at Ridgetown with all three postemergence herbicides. The application of dicamba/diflufenzopyr at 2, 10, 20 and 40 g a.i. haÀ1 followed by chlorimuron ethyl or imazethapyr resulted in a synergistic response. In addition, the simulated drift of dicamba/ diflufenzopyr at 10 and 20 g a.i. haÀ1 followed by bentazon appli cation showed a synergistic response. For example, the observed crop injury from the dicamba/diflufenzopyr drift at 2, 10, 20 and 40 g a.i. haÀ1 followed by chlorimuron ethyl was 6, 31, 27 and 11% greater than the expected values indicating a synergistic response. Similar increases in crop injury were observed with the application of imazethapyr and bentazon. At 14 DAA, there was an increase in crop injury in soybean with increasing doses of dicamba/diflufenzopyr (Table 1). Kelley et al. (2005) also reported observing more injury with increasing doses of dicamba/diflufenzopyr. When dicamba/diflufenzopyr was applied at 0.2 g a.i. haÀ1 plus 0.08 g a.i. haÀ1 and 2.0 g a.i. haÀ1 plus 0.8 g a.i. haÀ1, soybean injury two weeks after application was 22 and 42%, respectively (Kelley et al., 2005). Generally, there was no ER 237 (262 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 129 of 245 LR. Brown et al. [erJ Hotection 28 539?542 541 Table 1 Percent soybean injury 14 DAA with simulated dicamba/di?ufenzopyr drift alone or followed by the application of a postemergence herbicide at Elora. Harrow and the three sites at Ridgetown in 2007.?1 Table 3 Percent soybean injury 56 DAA with simulated dicamba/diflufenzopyr drift alone or followed by the application of a postemergence herbicide at Elora. Harrow and Ridgetpwn in 2007.?1 Dicambaldi?ufenzowr Injury (14 drift (dose) Dicamba/diflufenmwr Injury (56 drift (dose) (8 1L ha Dicamba/ Drift fb Drift fb Drift fb (8 81 ha Dicamba/ Drift fb Drift fb Drift fb difldenzowr drlon'muron-ethyl imazethapyr bentazon diflufenzopyr chlorimuron-ethyl imazethawr bentazon Elora Untreated check Untreated check 0a 0a 0a 0a 0 0a 03(0)c 0a(0) 0a(0) 0 0a 53(5)c 10 b(10) 03(0) 2 11 13b(11) 17b(11)+ 16 b(11)+ 2 11 16 b(15) 13 b(19) 13 (11) 10 23 25 (2324c (20) 13 (11) 20 28 34d (28H 33 (28H 32 20 18 24 (22) 26 (26) 21 (18) 40 42 47 45 (42) 47 40 28 36 d(31) 28 (35) 27 (282'1 Abbreviations: DAA. days after application: fb, followed by. Untreated check 03 0a 0 a 03 Means have been back?transforrned to original scale. Means followed same letter in each column are not signi?cantly different according to Fisher's 2 16 16 (16) 16 (16) 16 mined? LSD test (P 0?05)? . 30 20 N30) 29 25d (35) Expected responses based on Colby?s equation (E A 8/100), for 20 34 39 (34) 34 (34) 28 (38) combinations are shown in parmtheses following each observed response. Signi?- 40 55 58 (55) 46 (55) 53 (58) cant differences based on a paired t?test between observed and expected values are SE 44 4? 38 3 6 shown bya sign to indicate synergism I Untreated chedt (0) drlorimuron ethyl had 17. 20. 15 and 6% more injury. respectively 2 ?5 33 33 '3 27 than what was expected indicating a synergistic response. Simi (51? 63 (51 61 (51? larly. synergistic responses were also observed drcamba/ 70 76e(7o)+ 72e(70) 71 (70) defenzopyr at 2. and aj. imazethapyr se 33 3.6 3.4 33 and bentazon. Abbreviations: DAA days after application: lb, followed by. Ridgetown means havebear back-transformed to original scale. Means followed by the same letter in each column are not signi?cantly different according to Fisher's protected LSD test (P 0.05). Btpected responses. based on Colby's equation (E A A 8/100). for combinations are shown in parentheses following each observed response. Signi?- cant differences based on a paired r?test between observed and expected values are shown by a sign to indicate synergism and a sign to indicate antagonism synergistic response from simulated dicamba/diflufenzopyr drift followed by the three postemergence herbicides at the Harrow and Elora locations. At Harrow. there was an antagonistic response with the simulated drift (2 a.i. ha?) followed by bentazon. However. this was an isolated response that was not observed in any of the other data. At Ridgetown. there were synergistic responses with all three postemergence herbicides. The application of dicamba/ diflufenzopyr at 2. 10. 20 and 40 ai. ha'1 followed by Table 2 Percent soybean injury 28 DAA with simulated dicamba/diflufenzopyr drift alone or followed by the application of a postemergence herbicide at Elora. Harrow and Ridgetown in 2007.2 Dicambaldi?ufenzopyr Injury (28 21:25:25? Dicamba/ Drift fb Drift fb Drift fb chlorimuron-ethyl imazethawr bentazon Untreated chedt 011(0)c 0a(0) 1a(1) 2 18b 21 b(18) 28b(18)+ 24b(18(43) 40 60e 64d(60) 63d(60) 62e(61) SE 2.6 2.8 2.7 26 Abbreviations: DAA days after application; fb, followed by. Means followed by the same letter in each column are not signi?cantly different according to Fisher?s protected [50 test 0.05). Expected responses. based on Colby?s equation (E A A 8/100). for combinations are shown in parentheses following each observed response. Signi?- cant differences based on a paired t-test between observed and expected values are shown by a sign to indicate synergism. Soybean injury at 28 DAA increased with inqeasing doses of dicamba/di?ufenzopyr (Table 2). The application of dicamba/ diflufenzopyr at 10 and 20 at ha?1 followed by chlorimuron ethyl and imazethapyr resulted in 6 11% more injury than expected indicating a synergistic response. Similarly. the application of dicamba/di?ufenzopyr at 2. 10 and 20 a.i.ha'1 followed by irna zethapyr resulted in a synergistic response. There was no syner gistic response from simulated dicamba/diflufenzopyr drift followed by bentazon. At 56 DAA. there was an increase in injury in soybean with increasing doses of dicamba/diflufenzopyr (Table 3). The applica tion of dimmba/diflufenzopyr at 2 and 10 a.i.ha? followed by either imazethapyr or bentazon resulted in a synergistic response. In addition. the simulated drift of dicamba/di?ufenzopyr at 20 a.i.ha?1 followed by either chlorimuron ethyl. imazethapyr or bentazon. resulted in injury that was 6. 5 and 4% greater than the expected values. respectively. Similarly. synergistic responses were also observed with the applimtion of dicamba/diflufenzopyr at 40 ai. ha?1 followed by either chlorimuron ethyl or bentazon. 32. Crop height Height decreased with increasing doses of dicamba/diflufenzo (data not shown). Height decreased by 13. 29. 37 and 47% when dicamba/diflufenzopyr was applied at 2. 10. 20 and 40 a.i. ha?. respectively. Similarly. Kelley et al. (2005) also reported a 12 24% height reduction, despite the development of lateral branches on the more severely injured plants. There were no synergistic responses from simulated dicamba/di?ufenzopyr drift followed by any postemergence herbicides. 3.3. Crop dry weight Percent dry weight re?ected the amount of injury observed. Dry weight decreased by 19. 26. 36 and 50%. respectively. with increasing doses of dicamba/diflufenzopyr (data not shown). There were no synergistic responses from simulated dicamba/diflu fenzopyr drift followed by the postemergence herbicides. ER 238 (263 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 130 of 245 542 LR. Brown et a1. Crop Protection 28 (2009) 539-542 Table 4 Percent soybean yield in comparison to an untreated check when dicamba/diflu? fenzopyr drift wm simulated alone or followed by the application of a post- emergence herbicide at all three locations in 2007.? Dicamba/di?ufmzopyr Yield" (24) gf??sf; Dicamba] on'rt fb Drift lb Drilt lb di?ufenzowr chlorimuron-ethyl imazethawr bentazon Untreated check 100 (102)c 95 ab (95) 98 a (98) 2 94 ab 98 a (96) 94 abc (90) 96 a (92) 10 943b 90 ab (97) 86 bc(90) 90ab (92) 20 87 81 82 83 (85) 40 69c 62 c(70)+ 63 (65) 64c (67) SE 1.7 1.8 1.6 1.6 Abbreviations: DAA. days after application; fb. followed by. Means followed by the same letter in each column are not signi?cantly dilferent according to Fisher?s protected LSD test (P 0.05). Expected raponses, based on Colby's equation (E A 8/100). for combina- tions are shown in parentheses following each observed response. Significant differences based on a paired t-test between observed and expected values are shown by a sign to indicate synergism 3.4. Crop yield Soybean yield decreased with increasing doses of dicamba/ di?ufenzopyr (Table 4). Synergistic responses were observed with the simulated drift of dicamba/diflufenzopvyr at 10. 20 and 40g a.i. ha? followed by the d'rlorimuron ethyl application whid'r resulted in a 7 8% greater yield reduction than expected. The application of dicamba/diflufenzopyr at 2. 10. 20 and 40 a.i. ha'1 followed by imazethapyr or bentazon did not result in a synergistic yield response. Similarly. dicamba/di?ufenzopyr (2 a.i.ha") fol lowed by dilorimuron ethyl also did not result in a synergistic response in yield. This study con?rms that injury in soybean is accentuated when dicamba/diflufenzopyr drift is followed by some registered post emergence herbicides. A synergistic response for crop injury 7. 14. 28 and 56 DAA and soybean yield was documented when soybean was stressed due to herbicide drilt followed by the postemergence herbicide. Weed management practitioners and cop consultants must be aware that herbicides with an acceptable margin of safety can cause injury if the crop was stressed due to a previous herbicide drift event. Acknowledgements The authors would like to acknowledge the University of Guelph and Harrow Research Station weeds labs for their expertise and tedinical assistance in these studies. Funding for this project was provided in part by the Ontario Soybean Growers and the Canada Ontario Researd'r and Development Program. References Andersen. S.M., Clay. S.A., Wrage, Lj.. Matthees, D.. 2004. Soybean foliage residues of dicamba and and correlation to application rates and yield. Agron.]. 96, 750-760. Colby. SR. 1967. Calculating synergistic and antagonistic responses of herbicirh combinations. Weeds 15. 20?22. Gresel. 1., 1990. Synergizing herbicides. Rev. Weed Sci. 5. 49?82. Kelley. Wax. LM.. Hager. A.G.. Riechas. D.E.. 2005. Soybean response in plant growth regulator herbicides is affected by other postemergence herbicides Weed Sci. 53. 101?112. Lich, Renner, KA., Penner. D.. 1997. Interaction of with post- emergence soybean (Glycine max) herbicides Weed Sci. 45. 12?21. Maybank. Yoshida. K.. Grover, R.. 1978. Spray drift from agricultural pesticich applications. J. Air Pollut. Control Assoc. 28. 1009-1014. Simpson. D.M.. Stoller. E.W., 1996. Physiological mechanisms in the synergism between thifensulfuron and imazethapyr in sulforrylurea-tolerant soybean (Glycine max). Weed Sci. 44, 209-214. Vendll, W.K.. 2002. Herbicide Handbook. eighth ed. Weed Science Society of America. 493 pp. Weidenhamer, Triplet! jr., G.B., Sobotka, RE. 1989. Dicamba injury to soybean. Agron. 81. 637?643. Wolf. T.M., Grover, R., Wallace, K, Shewchuk. S.R.. Maybank. 1., 1992. Effect of protective shields on drift and deposition characteristics of ?eld Sprayers. Can. Plant Sci. 73.1261?1273. ER 239 (264 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 131 of 245 Uti izing Geospat al echnology to Assess Off target Dicamba Injury and Yield Loss in Missouri Soybean F elds Shea a ell* D Kent Shannon E c Ose and Mandy Bish and Kevin B ad ey Un ve sty o Missou i Columba IN RODUC ION RESUL S RESUL S n 20 6 he mao ty o the cot on c ea e in the m d South we e p an ed w th d camba- o e ant D ) va e i s and a im ted numbe o D oybean va e i s we e a so pan ed Du ng the 20 6 g owi g eason he e we e no app o ed die mba he b c de o mu at ons o pos - me gence app i at on o hese c ops yet a subset o g owe s made i egal app ca io s an w y Ove 40 000 ac es o non-D soyb an we e n u ed by o ta get movement o d camba caus ng u known amoun s o yi ld lo s ev ous esea ch as s own soy ean y e d oss is depe dent on g owth tage nd exposu e dosa e Howe e in eld s t in s p a tt one s n ve know he peci c dose o ic mba hat co tac ed the non D oybean n e po a ed v sual nju y eva ua ons ( gu es 2A a d 2C) we e a s gned and g o ped by in u y nte si y n o nc men s o 20°/o he ota a ea o ea h n u y ange was ca cu a ed ( ab es and ) he est mat d ve a e yie d o each nju y a ge w s mu t p ed by ts ep ese ta i e a ea hen added to de e m ne to al y e d e t m ton ( ab es and ) n e po a ed y e d m ps ( gu es 28 a d 20) we e c assi ed by ompa ng 20 6 y e ds o he c mb ned ave age y e d o he h e ie d a ea ep e ent ng e ch y e d p ev ous oybean o at ons ange was cal ul ted ( ables and 2) A e age y e d o ea h n u y ange ac o s all el s col ec ve y was compu ed o u he nal ze i amount o yie d I ss an e a soc ated w th v sua I n u y ( ab e 3) gu %Y B A % 0 d OB EC IVES o de e m ne i ate seas n v sual val at ons o d camba n u y c n p ed ct y e d oss on a eld sea e le el • MA ERIALS AND ME HODS - e d bounda es we e mapp d to SMS Mobi e Ag eade so wa e o am le g d c e t on ( g Bounda y) y % Samp e oca ons we e sta I shed us ng a ce te g id o mat Sampl ng G ds) at 5 me e spac ngs ( g - - ~- y % 9 9 9 9 0 9 90 9 0 99 0 0 0 9 - ~ y # y d B % -------- - - -- - - -- - - h - - - - - - - - - v ~ 'J ~ ~ ~ s ~ ~ . ~ 0 0 0 0 00 0 0 00 0 0 0 0 0 0 0 0 0 0 9 9 9 9 9 9 9 00 0 9 9 0 9 999 9 om n g neal v sual nju y nd y e d oss was moe p eva ent nd eve e asp oxim ty o he sou e nc eased u he ana ys s w II be onduc ed o deem ne i yie d lo s p t e ns wi h n a e d an eve as bet e ind ca o s o expe ted y e d lo s aused by o - a et mo eme to d camba Re ul s om t is s udy w I he p a me s and p o es iona s n he ag i ul u e ind st y be te i ual ze and un e s and he e ec s hat o ta get movement o d camba ( i u e 3) has on oybean y e d s B 0 9 9 s 9 0 0 0 ER 240 0 On ve a e in 20 6 es ima ed y e d I sses ased on ineason n u y a ngs we e wi h n 2 5°/o o he ac ual y e d he s udy w I be epea ed in 20 7 c o s a I ou ie ds 00 9 B 9 CONCLUSIONS & FU URE WORK B 9 9 In 20 6 y eld es ima io s us ng th s p oce u e ang d -5 2°/o to 8°/o 0 he act al 20 6 y eld ( abe 4) y d 0 Ad i i nal y a elep one u vey s eing cond cted o ob a n p odu e s inp ton ie d lo ses due to d camba n u y %Y y % 0% 0 99 Ha ves ed yi ld da a was o gan zed in o ones de in ng how a eas o elds y e ded in 20 6 compa d o he hi to i al ave age he mean ie d o e ch one was us d o al ula e to al y e d ( ab es and 2) gu % % 00 0 e ds we e g ouped n o anges ep esen i g ow i es eci c y e d va ues compa ed o the e d ave age then n e po a ed o ca cu ate a a ( ig nte po a ed e d Map) 00 0 0 0 0 0 9% % % % 0 S t -spe i ic y e d n o mat on o ea h samp ed oca on was ie d Map) ob a ned th ough comb ne y e d mon o s ( g 0 0 0 0 0 % % 9 # o de e m ne i is al i ju y a in s c n se ve as in ic to s o yi ld at ha ve t va ues o e t m te ie d we e de e m ned o each n u y ange Va ues sed o es ima e y e d pes nt he ve age y eld p oduc d w hin ach a ge when a I v sual nju y at ngs and a company ng ha v sted y e ds we e compa ed ( ab es and 2) n u y a ngs we e g ou ed o p ed ct y e d oss anges based on he MEANS p ocedu e and om a ed to he ctua y elds 9% % 9 e im na y da a i die tes yi ld oss d d not occu un il at le st 0°/o vi ual in u y was bse ed and >25°/o y eld lo s occu ed when at east 40°/o n u y was p sent ( ab e 3) 00 0 0 0 0 0 0 # 0 0 %Y 0 0 0 0 00 So bean ie d and vi ual nju y a ngs t each p e ete m ned s mp e oca on we e compa ed in SAS us ng he MEANS p ocedu e at the 0 05 le el o s gni can e % B 0 V sual so bean n u y a i gs we e e o ded once oybean eached the R6 R7 s age o g ow h Hand he d G S un ts we e u ed o n vi ate tot e p ede e mi ed g id oca ions so at ngs cou d be geo e e enced ( g Vi ual I ju y R t ngs) n u y eva ua ions wee g ouped y se e ty and i te po ated n GIS so a a a socat d wth e ch ange c uld be c lc Ia ed ( g In e ola ed In u y Ra i gs) y % R su ts om a meta ana ys s nd i ate owe ng soyb an exposed o OOOX a e o die mba equal o/o yi ld I ss nd exposu e to OOX ate equ Is a 9°/o y eld oss d n 20 6 4 s pa a e on-D soybean e ds that we e n u ed by o ta get movement o d camba we e isu I y a ed u ing he Beh ens and ue chen sea e ( 979) 0% 00 - - 0 9 0 s (265 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 132 of 245 Weed Science 2014 62:193 206 A Meta-Analysis on the Effects of 2,4-D and Dicamba Drift on Soybean and Cotton J. Franklin Egan, Kathryn M. Barlow, and David A. Mortensen* Commercial introduction of cultivars of soybean and cotton genetically modified with resistance to the synthetic auxin herbicides dicamba and 2,4-D will allow these compounds to be used with greater flexibility but may expose susceptible soybean and cotton cultivars to nontarget herbicide drift. From past experience, it is well known that soybean and cotton are both highly sensitive to low-dose exposures of dicamba and 2,4-D. In this study, a meta-analysis approach was used to synthesize data from over seven decades of simulated drift experiments in which investigators treated soybean and cotton with low doses of dicamba and 2,4-D and measured the resulting yields. These data were used to produce global dose–response curves for each crop and herbicide, with crop yield plotted against herbicide dose. The meta-analysis showed that soybean is more susceptible to dicamba in the flowering stage and relatively tolerant to 2,4-D at all growth stages. Conversely, cotton is tolerant to dicamba but extremely sensitive to 2,4-D, especially in the vegetative and preflowering squaring stages. Both crops are highly variable in their responses to synthetic auxin herbicide exposure, with soil moisture and air temperature at the time of exposure identified as key factors. Visual injury symptoms, especially during vegetative stages, are not predictive of final yield loss. Global dose–response curves generated by this meta-analysis can inform guidelines for herbicide applications and provide producers and agricultural professionals with a benchmark of the mean and range of crop yield loss that can be expected from drift or other nontarget exposures to 2,4-D or dicamba. Nomenclature: 2,4-D (2,4-dichlorophenoxyacetic acid); dicamba (3,6-dichloro-2-methoxy benzoic acid); glyphosate; soybean, Glycine max (L.) Merr.; cotton, Gossypium hirsutum L. Key words: Dose–response curves, Glycine max, Gossypium hirsutum, herbicide drift, herbicideresistant crops, meta-analysis. Biotechnology companies are currently developing cultivars of corn, soybean, and cotton engineered with transgenic resistance to the syntheticauxin herbicides 2,4-D and dicamba (Behrens et al. 2007; Waltz 2010; Wright et al. 2010). Dow AgroSciences is currently developing corn (Zea mays L.), soybean, and cotton cultivars resistant to 2,4-D, and the Monsanto Company in collaboration with BASF is developing cultivars of soybean and cotton resistant to dicamba. Dicamba and 2,4-D have been widely used for decades for selective weed control of broadleaf plants in grass and cereal crops (Monaco et al. 2002). However, the new resistant cultivars will enable these compounds to be applied in new crops, at new times during the growing season (including more POST applications), and over greatly expanded areas, potentially leading to DOI: 10.1614/WS D 13 00025.1 * Research Associate, Pasture Systems Watershed Management Research Unit, USDA Agricultural Research Service, Building 3702, Curtain Road, University Park, PA 16802; Research Technician and Professor, Department of Plant Science, The Pennsylvania State University, 116 ASI Building, University Park, PA 16802. Corresponding author’s E mail: Franklin.Egan@ars.usda.gov increased problems with nontarget drift onto susceptible crops, including non-transgenic soybean and cotton (Mortensen et al. 2012). Soon after the commercialization of 2,4-D in the 1940s and dicamba in the 1960s, recurrent problems of nontarget exposures to susceptible crops began to occur (Staten 1946; Wax et al. 1969). Continuing to the present, the Association of American Pesticide Control Officers (AAPCO) consistently ranks 2,4-D and dicamba at or near the top of herbicide active ingredients implicated in crop injury complaints (AAPCO 2005), and several states and municipalities have special restrictions on the use of these compounds to help prevent crop injury problems (Louisiana Department of Agriculture and Forestry 2011; Texas Department of Agriculture 2012). This high frequency of crop injury complaints relative to other herbicides is likely due to several factors specific to 2,4-D and dicamba. First, synthetic auxin herbicides can cause distinctive injury symptoms on many broadleaf crops, including twisting or epinasty of stems and cupping of leaves, such that even slight injury can be readily recognized by growers and land owners. Egan et al.: 2,4-D and dicamba drift on soybean and cotton ER 241 N 193 (266 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 133 of 245 Secondly, several broadleaf crops, including soybean and cotton, are very sensitive to these compounds, creating the potential for noticeable injury and potential yield loss following very low-dose exposures. Finally, several commercially available 2,4-D and dicamba products include moderately volatile herbicide formulations that can travel away from treated fields as vapor drift (Behrens and Lueschen 1979; Egan and Mortensen 2012; Grover et al. 1972). In regions where synthetic auxin–resistant cultivars of cotton and soybean will be widely adopted, the use of 2,4-D and dicamba is likely to increase substantially over the next 5 to 10 years (Mortensen et al. 2012). These trends could increase the risk of injury and yield loss to susceptible crops, including non-transgenic soybean and cotton through a variety of mechanisms. First, as with all herbicides, 2,4-D and dicamba can move as particle drift from ground or aerial application equipment, especially when herbicides are applied under windy conditions or when spray equipment not designed to reduce particle drift is used (Wang and Rautman 2008). Secondly, as previously mentioned, if volatile formulations are used under high temperature conditions, 2,4-D and dicamba can move from treated fields onto susceptible fields. Third, 2,4-D and dicamba residues are known to be difficult to clean from equipment, and small amounts of these compounds could be inadvertently applied to susceptible crops if the same equipment was recently used to treat dicamba or 2,4-D resistant or tolerant crops (Boerboom 2004). Finally, in regions where 2,4-D or dicamba are used frequently and over large areas, such as the wheat (Triticum aestivum L.) cropping systems of the Canadian prairie provinces, herbicide residues can accumulate in the atmosphere and return to fields as precipitation at concentrations high enough to cause injury to susceptible crops (Hill et al. 2002; Tuduri et al. 2006). Anticipating potential problems, Dow AgroSciences, BASF, and Monsanto Company have been working with growers, agricultural service providers, and university extension to develop stewardship practices for these technologies. These practices will include the development of extremely low volatility formulations of 2,4-D and dicamba, adjuvants and herbicide premixes that reduce particle drift, and advanced spray nozzle designs that limit fine spray droplets (Dow AgroSciences 2011a, 2011b; Thomas et al. 2012). However, due to the combination of exposure routes, it remains 194 N Weed Science 62, January–March 2014 likely that nontarget drift to susceptible crops will be a significant concern for growers, especially during the early phase of commercialization of these technologies. Because soybean and cotton are among the crop species most susceptible to these compounds, the risk of crop injury and potential yield loss will perhaps be greatest to soybean and cotton growers who choose not to use resistant cultivars in regions where the transgenic cultivars and associated herbicide programs are widely adopted by neighboring farmers. Importantly, because 2,4-D resistant crops will be susceptible to dicamba (and vice versa), crop injury risk could extend to growers that choose 2,4-D resistant cultivars in regions where dicamba resistant cultivars are more popular (and vice versa). In order to prepare for crop injury incidents and potential yield loss, growers and agricultural professionals may find it helpful to be equipped with a detailed understanding of the likely responses of cotton and soybean to low-dose exposures of 2,4-D and dicamba. Fortunately, the dose–response patterns of crop injury and yield loss in cotton and soybean to 2,4D and dicamba have already been extensively researched. Beginning in the 1950s with cotton, weed scientists in the United States and internationally began conducting simulated drift bioassay experiments to determine the herbicide doses that are likely to cause noticeable injury symptoms and the doses that are likely to cause significant yield loss. In this paper, a meta-analysis approach was used to review and synthesize the results from many of these previously published simulated drift experiments. After conducting an exhaustive search of the literature, an extensive dataset on the response of cotton and soybean to simulated drift was used to answer four key interrelated questions. First, what is the dose–response pattern of herbicide exposure dose (in g ha21) and crop yield? Second, how is the dose–response pattern affected by crop phenology at the time of exposure? Third, what other environmental and agronomic factors may influence crop dose–response? Finally, how do visual injury symptoms in soybean and cotton from 2,4-D and dicamba exposure correlate with yield loss? Material and Methods Literature Search. Literature searches were performed using the CAB Direct database in October and November of 2011 (Centre for Agriculture and Biosciences International 2012). CAB Direct spe- ER 242 (267 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 134 of 245 cializes in agricultural research and indexes many materials that would be missed by more general literature search engines such as Web of Science or Google Scholar, including conference proceedings and research reports from state agricultural experiment stations. The search was defined using the terms (cotton or soy*) and (2,4-D or dicamba) and (drift or injury or sensitiv* or toleran*). Several pilot searches using broader search terms such as ‘‘yield’’ were also conducted, but it was found that the more specific search terms captured all of the relevant material. Studies were classified as relevant based on the following criteria: 1. The study must have employed a replicated field experiment in which cotton or soybean was exposed to at least one dose of dicamba or 2,4-D and a water or untreated control. Studies in which dicamba or 2,4-D was applied in mixtures with other herbicidal compounds were not included. Herbicide treatment doses must have been applied as a foliar spray and presented in grams per hectare or in units that could be converted into grams per hectare doses. 2. Because 2,4-D or dicamba applications could affect crop performance by influencing weed– crop competition in addition to having direct phytotoxic effects, studies were selected that eliminated background weed communities with appropriate herbicides or cultivation. In some instances, studies selected as relevant did not specifically describe background weed control practices, but based on their methodologies and stated objectives, it could be safely inferred that weeds were effectively managed. 3. Studies must have collected and reported data on grain yield for soybean and seed cotton or lint yield for cotton. Data on visual injury ratings from studies that reported yield were also included. CAB Direct indexes international publications and proceedings, but it reports all abstracts in English. For studies not published in English, the abstract, tables, and figures were reviewed to determine if the paper was likely to fit the criteria defined above. For these likely papers, international colleagues at The Pennsylvania State University were recruited to assist with interpretation. One study originally published in Portuguese (Constantin et al. 2007) was fully translated. For each study that was selected as relevant, the bibliography was also reviewed and a backward search was then performed using the same criteria. Data Coding. For each relevant study, information was gathered from tables and figures, and the available data on the dose of dicamba or 2,4-D exposure in each treatment and the resulting yield and visual injury (most commonly reported on a 0 to 100 scale) was coded. Yield response and injury data were coded as the mean value for a given dose as presented in each study’s tables or figures. All yield data were normalized as the proportion of the respective control dose. Data presented in figures were extracted using the software program EnGauge (M. Mitchell 2007, Engauge Digitizing Software). The crop growth stage at the time of exposure was coded by grouping the phenology described by the authors into either vegetative, flowering, or pod formation stages for soybean and into vegetative, preflower squaring (flower buds are first forming), early flowering, or mature flowering/boll formation stages for cotton. Several studies did not clearly describe the crop’s growth stage, but instead listed the crop’s height or number of leaves or nodes at the time of herbicide treatment. In these cases, the crop’s phenology was inferred based on other studies in the dataset that reported height, leaf number, and phenology from similar locations and using similar cultivars or using information presented in Barker et al. (1985), Oosterhuis and Jernstedt (1999), and Pedersen (2004). Many studies quantified yield response under several unique experimental conditions, for instance crop response may have been measured to both dicamba and 2,4-D or to different 2,4-D doses at multiple phenological timings of herbicide exposure. Within a study, a unique set of experimental conditions was classified as a unique ‘‘sequence.’’ For instance Kelley et al. (2005) present data from an experiment in which soybean were exposed to dicamba at the V3, V7, and R2 growth stages, providing three unique dose–response sequences. They also report on a separate experiment in which soybean were exposed to dicamba at the V3 and V7 stages in two different years, with each year reported separately. In total, this reference therefore contributed seven different dicamba dose–response sequences to the dataset. Because each sequence contained either one or two different doses of dicamba exposure, Kelley et al. (2005) contributed 10 data points to this meta-analysis for dicamba and soybean. Statistical Analysis. The dose–response patterns of dicamba or 2,4-D simulated drift on soybean and cotton yield were analyzed by fitting log-logistic Egan et al.: 2,4-D and dicamba drift on soybean and cotton ER 243 N 195 (268 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 135 of 245 curves to the data using the nls package in R (R Core Team 2011; Ritz and Streibig 2008). Because the untreated control yields were defined as 1.0, and because label rates of dicamba and 2,4-D are obviously fatal to both cotton and soybean, a twoparameter log-logistic function with the upper asymptote set to 1 and the lower asymptote set to zero was used (Equation 1). Yieldi ~1=ð1z expfb ½logðDosei Þ logðe ފgÞ À Á zei , ei *N 0,s2 ½1Š where e is the dose that causes a 50% yield loss and b is the slope of the curve at the e-parameter dose (Ritz and Streibig 2005). Because all yields were normalized as a proportion of the control, the untreated control yields were removed from the data set before fitting the models to avoid heteroscedasticity and nonnormality of residuals. To quantify the influence of crop phenology at the time of exposure, separate models were fitted for each combination of crop growth stage and herbicide active ingredient (dicamba or 2,4-D). For cotton, a fraction of studies reported yield in terms of both raw seed cotton yield and ginned lint yield. To test for any potential interaction of herbicide dose and lint yield as a percentage of seed cotton yield (ginning percentage), the correlation between normalized seed cotton and lint yield was assessed for this subset of studies. Linear regression results indicated a near perfect correlation (0.999) with a slope nearly equal to one (0.993), implying no influence of either herbicide on ginning percentage. Studies that reported seed cotton or lint cotton yields were therefore pooled into the same dose–response curves. In this meta-analysis, the effect size or response statistic is the ratio of treatment yield to control yield. Meta-analysis requires estimating the withinstudy variation in effect size as a measure of the precision with which the authors of a given study were able to estimate an effect or response in their experiments (Cooper et al. 2009). Within-study variation statistics can then be used to weight more precise studies more heavily than less precise studies in the meta-analysis. For response ratio data, Hedges et al. (1999) suggested that the variance of response ratios provides a good estimate of withinstudy variation. Hedges et al. (1999) further suggested that the natural logarithm of response ratios and its associated variance are better effect size statistics because they weight changes in control and treatment response equally and tend to be more 196 N Weed Science 62, January–March 2014 normally distributed than the untransformed response ratio statistic. But, for log-logistic models, log-transforming the y-axis and using log response ratios would lose the biological meaning of the model’s asymptotes, because log(1) equals zero and log(0) equals negative infinity. Therefore the more interpretable untransformed response ratio statistic and its associated variance were used in this analysis, as defined in Equation 2 and in Appendix A of Hedges et al. (1999). Variance of Response Ratio~ Á À ÂÀ Áà ðXT =XC Þ2 SE2T nT XT2 z SE2C nC XC2 ½2Š where, XT is the mean from a treatment group, SET is the standard error of that mean, nT is the sample size of the treatment group, and XC, SEC, and nC are the analogous quantities for the control group. For several studies in this meta-analysis, the values for the standard error of the mean that were needed to calculate Equation 2 were either reported directly or could be back-calculated using summary statistics reported by the authors, such as the Fisher’s Least Significant Difference (Kuehl 2000; Zar 1998). Using reported values or back-calculations, Equation 2 could be calculated for four studies and 35 sequences for soybean and eight studies and 57 sequences for cotton. Unfortunately, in many cases authors did not report sufficient information to calculate standard errors of the means but instead only reported sample size or the number of replications. Rather than simply exclude studies that did not report variance statistics from this meta-analysis, the subset of studies for which Equation 2 could be calculated was used to exploit a correlation between within-study variance and sample size. The pattern evident in this correlation was then applied to the entire dataset using the following bootstrap procedure. For the subset of studies for which Equation 2 could be calculated, the variance of the response ratio statistics were binned into three replicate size classes (three, four or five, and more than six replicates). Figure 1 indicates a clear pattern of decreasing variation with increasing sample size, as is expected from basic statistical theory (Crawley 2005). Next, the subset for which Equation 2 could be calculated was separated by crop and the variance statistics were then randomly sampled with replacement from each replicate size class. These randomly sampled values were then assigned to data points with corresponding sampling sizes in the full dataset. A two-parameter log-logistic model was then fit using the randomly ER 244 (269 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 136 of 245 Soybean 0003 0.002 0.001 .Q "§ 0.010 0.005 0.000 3 4 5 6 Replicates Figure 1. Relationship between the number of replications and the variance of the response ratio of yields in ueated to untreated conuol plots. Data are compiled from previously published experiments that measured the effect of simulated dicamba and 2,4 D drift on yields of soybean and cotton. assigned variance stacistics as weights (Equation 3; Ritz and Streibig 2008). Yieldi = 1/ ( 1 + exp { b x [log( Dosei) +ei j..Jwi, ei"' N(O,CJ2 ) log(e)]}) [3] where wi is the variance of response racio randomly assigned based on sample size. For each combinacion of crop growth stage and herbicide, this procedure was repeated 100 times, and results are reported as rhe median from this bootstrapped distribution of fitted models. R code for this procedure is available from the authors upon request. Synthetic auxin herbicides are widely reported to cause stimulatory effects to crops at low doses, also known as hormesis. To test for me possibility of hormesis, the log-logistic model was compared to the Cedergreen hormesis model (Cedergreen et al. 2005) using Akaike's Information Criterion (Burnham and Anderson 2002) and t-tests of me Cedergreen f parameter (Cedergreen et al. 2 005). For all combinations of crop growth stage and herbicide, the log-logistic model showed a substantially berrer fir to the data, indicating no evidence fo r a hormesis effect (data nor presenred). To assess potential yield loss from herbicide drift, the predicted yield loss for each crop growth srage and herbicide combination was calculated ar doses of 0.56, 5.6, and 56 g ha- 1. Assuming a field application rate of 560 g ha-1, these doses roughly correspond to a vapor drift exposure in an adjacent field (Egan and Mortensen 20 12; Grover er al. 1972), a particle drift exposure in an adjacent field (Browner al. 2004; Carlsen et al. 2006; de Jong er al. 2008; United States Environmental Protectio n Agency 2006; Wang and Rautman 2008), or a serious application error, respeccively. For each bootstrapped model fir, 95% confidence intervals were calculated around iliese yield loss estimates using the delta method in the R package car (Fox and Weiseberg 2010). The median intervals from rhe bootstrapped disrriburion of 100 models are reported. For me subset of studies mat reported data on born yield and visual injury for d1e same treatments, simple linear models of the relacionship between yield and injury rating were calculated. Multiple linear (including a quadratic term for injury) and logistic regression models fo r the relationship between yield and injury were also calculated. For soybean, logistic models provided a better fir, bur for cotton, linear, multiple linear, and logistic models all produced similar fits. However, for both soybean and cotton, all models led ro a similar interpretation of the results, ilierefore results will only be presenred for rhe simple linear models. Injury ratings 12 to 16 d after treatment (DAT) were used because this was the most commonly reported injury rating interval. For one cotton 2,4-D study (Goodman et al. 1955), injury 3 DAT was used, since this was the closest reported interval. Mitigating Environmental and Agronomic Factors. Many authors conducted experiments over multiple years and sires or crossed herbicide dose wirh another porencially important factor such as herbicide formulation or crop genetics. Because data on these potentially important factors was not collected or reported consistently across the family of studies in the dataset, their significance could not be statistically assessed. Ins read each author's explanation and discussion of potentially important factors that could influence dose- response patterns was carefully examined, and these findings are presented here as a narrative review. Egan et al.: 2,4-D and dicamba drift on soybean and cotton ER 245 197 (270 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 137 of 245 Table 1. Summary of the number of studies, dose response sequences, and unique mean data points excluding controls (n). collected from a literature search of studies that measured the yield response of soybean and cotton to simulated dicamba and 2,4 D drift at different crop growth stages. Crop, herbicide Growth stage Soybean, dicamba Vegetative Soybean, 2,4 D Cotton, dicamba Cotton, 2,4 D Studies Sequences 6 20 n 61 Flowering 4 22 80 Pod Vegetative 2 9 9 25 26 81 Flowering 5 13 35 Vegetative 6 11 42 Squaring 4 6 18 Flowering 5 7 19 Vegetative 15 44 117 Squaring 11 28 76 Flowering 12 34 75 Boll 10 33 64 Citations Al Khatib and Peterson 1999; Andersen et al. 2004; Auch and Arnold 1978; Johnson et al. 2012; Kelley et al. 2005; Wax et al. 1969 Auch and Arnold 1978; Kelley et al. 2005; Wax et al. 1969; Weidenhamer et al. 1989 Auch and Arnold 1978; Weidenhamer et al. 1989 Andersen et al. 2004; Johnson et al. 2012; Kelley et al. 2005; Merotto et al. 1999; Robinson et al. 2013; Slife 1956; Smith 1965; Wax et al. 1969; Wiese and Martin 1963 Kelley et al. 2005; Slife 1956; Smith 1965; Wax et al. 1969; Wiese and Martin 1963 Everitt and Keeling 2009; Johnson et al. 2012; Lanini 1999; Marple et al. 2007, 2008; Smith and Wiese 1972 Everitt and Keeling 2009; Hamilton and Arle 1979; Marple et al. 2008; Smith 1972 Everitt and Keeling 2009; Hamilton and Arle 1979; Lanini 1999; Marple et al. 2008; Smith 1972 Behrens et al. 1955; Carns and Goodman 1956; Charles et al. 2007; Epps 1953; Everitt and Keeling 2009; Goodman et al. 1955; Goodman 1953; Johnson et al. 2012; Lanini 2000; Marple et al. 2007, 2008; Miller et al. 1963; Smith 1972; Watson 1955; Wiese and Martin 1963 Arle 1954; Banks and Schroeder 2002; Behrens et al. 1955; Carns and Goodman 1956; Charles et al. 2007; Everitt and Keeling 2009; Goodman et al. 1955; Marple et al. 2008; Miller et al. 1963; Smith 1972; Wiese and Martin 1963 Arle 1954; Behrens et al. 1955; Carns and Goodman 1956; Charles et al. 2007; Constantin et al. 2007; Everitt and Keeling 2009; Goodman et al. 1955; Lanini 1999; Marple et al. 2008; Miller et al. 1963; Smith 1972; Watson et al. 1955 Arle 1954; Behrens et al. 1955; Carns and Goodman 1956; Charles et al. 2007; Constantin et al. 2007; Epps 1953; Goodman et al. 1955; Kittock and Arle 1977; Kittock et al. 1973; Miller et al. 1963 Results and Discussion Search Results. The literature searches retrieved 512 unique studies, of which 23 were classified as relevant. Backward searches produced an additional five relevant studies. Two recent papers (Johnson et al. 2012; Robinson et al. 2013) that were published after the conclusion of the literature review were added. In total, the dataset includes 30 studies and a total number of 252 sequences. The number of studies, sequences, and unique dose-response data points (excluding control points) for each crop growth stage and herbicide combination is summarized in Table 1. Dose–Response Curves. Soybean. Soybean was far more sensitive to dicamba than to 2,4-D and was more sensitive to both herbicides in the flowering growth stage than in other stages (Tables 2 and 3, Figure 2). During the flowering stage, the dose– response curves indicate a mean yield loss of ,1.0% 198 N Weed Science 62, January–March 2014 from dicamba vapor drift exposures (0.56 g ha21) and 8.7% from dicamba particle drift exposures (5.6 g ha21). Yield losses were basically zero for 0.56 g ha21 exposures during vegetative and pod formation stages, and slight (3.7%) for 5.6 g ha21 exposures during vegetative stages. For serious misapplication exposures (56.1 g ha21), all soybean growth stages showed drastic yield losses of 48% or greater. The dose–response curves suggest that soybean has surprisingly high tolerance to 2,4-D. During both vegetative and flowering stages, soybean showed essentially no yield loss to vapor or particle drift level exposures and only slight yield losses (1.5 to 3.0%) to even serious misapplication exposures. There were no data available for pod formation stage exposures to 2,4-D. These data suggest that yield loss from synthetic auxin drift to soybean is more likely to be an issue when soybean is exposed to dicamba during the flowering stage. Because soybean may be planted ER 246 (271 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 138 of 245 Table 2. Summary of log logistic dose response models for the effects of dicamba and 2,4 D exposure on yields of soybean and cotton at different crop growth stages. Values reflect the median parameter estimates across 100 bootstrapped model fits.a Crop, herbicide Soybean, dicamba Soybean, 2,4 D Cotton, dicamba Cotton, 2,4 D Growth stage eb bc r2d Vegetative Flowering Pod Vegetative Flowering Vegetative Squaring Flowering Vegetative Squaring Flowering Boll 58 60 51 651 461 6730 109 92 61 15 72 328 1.40 0.99 3.41 1.42 2.00 0.46 1.46 1.15 0.33 0.70 0.63 0.66 0.60 0.58 0.84 0.47 0.63 0.22 0.63 0.60 0.28 0.48 0.44 0.38 a Yield 5 1/(1 + exp{b 3 [log(Dose) log(e)]}), with yield normalized as proportion of untreated control. b The e parameter is the herbicide dose causing a 50% loss in yield (in units of g ha 1). c The b parameter describes the slope of the curve at the e parameter dose. d The r2 statistic is the squared Pearson correlation of predicted and observed values for each curve. over a long period (6 wk or longer, particularly in the southern United States), such a scenario is more likely to occur if POST applications of dicamba herbicides become more common in soybean production areas. The new resistant traits will make later POST applications of dicamba in soybean and corn a weed control option that may be very attractive to growers where glyphosate-resistant and tolerant weeds are a serious problem. Thus, it will remain important to use appropriate application techniques and stewardship practices when using dicamba near susceptible soybean and other crops. Cotton. Cotton was far more sensitive to 2,4-D than dicamba (Tables 2 and 3, Figure 2), and for 2,4-D, cotton showed the most sensitivity relative to all of the other three crop–herbicide combinations. Cotton was most sensitive to dicamba during early flowering, with slight losses (1.3%) predicted from vapor drift exposures and slightly more substantial (3.9%) losses predicted from particle drift exposures. During vegetative and squaring stages, basically no yield loss is predicted from vapor drift exposures of dicamba, but more substantial yield losses are possible from particle drift exposures. Serious misapplication doses (56.1 g ha21) indicated yield losses of 10% or more from all growth stages (no data were available for dicamba exposures in the boll stage). As has been widely appreciated nearly since the discovery of 2,4-D, cotton is extremely sensitive to this herbicide, especially during vegetative and preflowering squaring stages. During vegetative stages, average yield losses of more than 19% are predicted just from vapor drift exposures, and 32% and 49% yield losses are possible from particle drift or misapplication exposures. During preflowering squaring stages, cotton showed less sensitivity to vapor drift exposures (9% yield loss), but greater sensitivity to particle drift (33% loss) and misapplication (71% loss) doses. Cotton sensitivity declines somewhat as plants mature and begin Table 3. Predicted yield of soybean or cotton exposed to three doses of dicamba or 2,4 D at different crop growth stages. Yield is presented as the proportion of untreated or control yield, and doses represent probable exposures to vapor drift, particle drift, or herbicide misapplication onto a sensitive crop adjacent to a field treated at 560 g ha 1 with either herbicide. Predictions are derived from log logistic dose responsea curves fit to data from previously published simulated drift experiments. Values reflect the median estimates across 100 bootstrapped model fits with 95% confidence intervals displayed in parentheses. Yield Crop, herbicide Soybean, dicamba Soybean, 2,4 D Cotton, dicamba Cotton, 2,4 D a Growth stage 0.56 g ha Vegetative Flowering Pod Vegetative Flowering Vegetative Squaring Flowering Vegetative Squaring Flowering Boll 0.998 0.990 1.000 1.000 1.000 0.989 1.000 0.997 0.805 0.912 0.956 0.985 Yield 5 1/(1 + exp{b 3 [log(Dose) 1 vapor drift (0.995, (0.979, (1.000, (1.000, (1.000, (0.966, (0.998, (0.989, (0.712, (0.844, (0.906, (0.963, 1.002) 1.002) 1.000) 1.000) 1.000) 1.015) 1.001) 1.005) 0.900) 0.978) 1.001) 1.005) 5.6 g ha 0.963 0.913 0.999 0.999 1.000 0.969 0.986 0.961 0.680 0.670 0.835 0.937 1 particle drift (0.920, (0.873, (0.998, (0.997, (0.999, (0.923, (0.959, (0.903, (0.601, (0.577, (0.747, (0.890, 1.006) 0.953) 1.001) 1.001) 1.000) 1.008) 1.012) 1.017) 0.756) 0.763) 0.914) 0.983) 56 g ha 0.511 0.515 0.414 0.970 0.985 0.904 0.727 0.642 0.509 0.293 0.545 0.761 1 misapplication (0.414, (0.455, (0.278, (0.945, (0.965, (0.858, (0.624, (0.520, (0.412, (0.223, (0.456, (0.704, 0.607) 0.576) 0.550) 0.996) 1.006) 0.955) 0.823) 0.751) 0.605) 0.361) 0.626) 0.817) log(e)]}). Egan et al.: 2,4-D and dicamba drift on soybean and cotton ER 247 N 199 (272 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 139 of 245 Soybean and Dicamba Soybean and 2,4-D 12 A 1.2 0 o 0 A • 10 10 08 0.8 0.6- --(),- o• ·+· .! \ . .o•' ' ' ' pod go, Oa ........ e c 8 t '·' 0.0 -(),- 04 0 o vegetative flowering \ 6 02 -- 0.6 0 vegetative flowering 1:J. 0 . ....... 0.2 \ 0 00 ' c: 0 O.o1 t 0.1 10 0.1 100 100 10 1000 0 a. e .e: "0 Cotton and Dicamba 12 - 1.2 Q) >= Cotton and 2,4-D a 0 t A 1.0 0.8 + 06 0 4- • ~0 .... 0 ...,._ vegetative -tr squaring · + · flowering A '-t + 10 ~ ·\ 08 . + • ·., ~ ·.,: ..... A 06 ...,._ vegetative --6- squaring ·+ · flowering -*· boll '.\ 0.4 + ',\ A 02 + • ._, '.\ 0.2 ·~ 00 00 0,01 0,1 10 O.G1 100 0,1 10 100 Dose (g ha· 1) Figure 2. Yield response of soybean and cotton co dicamba or 2,4 D exposure across different crop growth stages. Data are compiled from previously published simulated drift experiments. Dose response curve lines reflect the log logistic models defined in Table 2. developing bolls, but small yield losses are still possible from vapor drift (1.5%) or particle drift exposures (6.3%) during boll stages. Interestingly, cotton was also far more variable in irs response to 2,4-D as compared wim soybean's response ro either herbicide (Table 3, Figure 2). This variability may reflect inherent variation in me uptake and biochemical response of cotton tO 2,4-D, or it may reflect me fact mat this dataset contains a greater number of studies, locations, and environmental conditions for corron and 2,4-D as compared with the other herbicide-crop combinations in this metaanalysis (Table 1) . These data suggest that yield losses from 2,4-D drift tO corron may be a substantial problem i f new resistant crops make postemergence 2,4-D applications common when susceptible cotton is growing nearby. Such a scenario is probable, because in 200 • Weed Science 62, January-March 2014 much of the southern United States corron is planted before soybean. The observed variability in response indicates it will be very difficult to anticipate yield loss following crop injury, bur that yield losses could potentially be high. It will be critically important to use low volatility formulations, state-of-the-art application equipment, and perform applications w1der appropriate environmental conditions. Dicamba may be a safer option than 2,4-D if susceptible cotton is nearby, and it may be more appropriate to avoid synthetic auxins all together and integrate alternative weed management practices instead. Mitigating Environmental and Agronomic Factors. The studies in this dataset reflect broad heterogeneity with regard to many factors that are well known to influence crop response to herbicides, ER 248 (273 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 140 of 245 Table 4. Summary of environmental and agronomic factors found to influence soybean and cotton sensitivity to yield loss and injury from simulated dicamba or 2,4 D drift. Crop Soybean Cotton Herbicide Effect on sensitivity Factor Dicamba Crop oil adjuvants in spray solution Dicamba, 2,4 D Dry conditions around exposure Increased Increased Dicamba Higher temperatures around exposure Dicamba, 2,4 D Crop cultivar Increased Variable 2,4 D Formulation (ester vs. amine) Dicamba, 2,4 D ‘‘Thickening agent’’ (Norbak) added to spray solution Dicamba Narrower row spacing Dicamba Formulation (DMA vs. Na) 2,4 D Favorable fall weather facilitates recovery Increased No effect 2,4 D Higher carrier volume in simulated drift studies Dicamba, 2,4 D Dry conditions around exposure Decreased Increased Dicamba, 2,4 D Moist conditions around exposure Increased 2,4 D 2,4 D Higher temperatures around exposure Formulation (ester vs. amine) Increased Increased 2,4 D Soil quality facilitates recovery Decreased including meteorological and edaphic conditions at the time of spraying, crop cultivar and genetics, herbicide formulation, and herbicide carrier volume. Because these factors were generally not balanced in this dataset in a way that permitted a rigorous statistical analysis, the dose–response curves reflect the mean or expected yield loss across this broad heterogeneity. The often substantial variability around these mean curves (Table 3, Figure 2) reflects the combined contributions of these mitigating factors. Nevertheless, many authors explored crop herbicide response over multiple site years or over different experimental conditions and offered some explanations for the variation they observed in crop response. These factors are summarized in Table 4, and a few consistent themes emerge. First, environmental conditions before, during, and following herbicide exposure play a very important role determining crop sensitivity to herbicide drift. Soil moisture level and air temperature were identified by several authors as key factors. For soybean, dry conditions were consistently associated with increased dicamba and 2,4-D sensitivity relative to conditions with less water stress. For cotton, the effect of soil moisture was more nuanced. For vegetative and squaring stage Increased No effect Decreased Citations Andersen et al. 2004 Andersen et al. 2004; Auch and Arnold 1978; Kelley et al. 2005; Weidenha mer et al. 1989 Al Khatib and Peterson 1999 Auch and Arnold 1978; Wax et al 1969; Weidenhamer et al. 1989 Smith 1965; Weise and Martin 1963 Wax et al. 1969 Weidenhamer et al. 1989 Weidenhamer et al. 1989 Arle 1954; Behrens 1955; Carns and Goodman 1956; Miller et al. 1963 Banks and Schroeder 2002 Behrens 1955; Carns and Goodman 1956; Marple et al. 2007; Marple et al. 2008 Carns and Goodman 1956; Goodman 1953; Marple et al. 2007 Kittock and Arle 1953 Marple et al. 2007; Wiese and Martin 1963 Miller et al. 1963 exposures, several authors noted that sufficient soil moisture and humid conditions led to plants that were actively growing and therefore absorbed and translocated more herbicide, leading to greater sensitivity. However, for late flowering or boll stage exposures, dry conditions were found to affect the floral abscission and boll development process negatively, such that sensitivity to 2,4-D was increased. For vegetative growth stages, Marple et al. (2007) found that dry conditions increased sensitivity, especially with ester formulations of 2,4D. Marple et al. (2007) suggested this occurred because esters are less polar molecules relative to amine formulations and therefore may be more likely to cross the waxy cuticle of cotton leaves under dry conditions. Several authors also concluded that higher air temperatures increased herbicide uptake and resulted in greater injury and yield loss in both cotton and soybean. As an indeterminate species, cotton produces squares and flowers continuously until arrested by low night temperatures (,5 C) or by ‘‘cut-out,’’ a physiological end of a flowering cycle that depends on latitude, night temperatures, cultivar, and fruit load (Bednarz and Nichols 2005). Consequently, cotton plants injured by 2,4-D or dicamba early in development can often resume flowering and Egan et al.: 2,4-D and dicamba drift on soybean and cotton ER 249 N 201 (274 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 141 of 245 compensate with later fruit set. However, depending on latitude, climate, and weather patterns during a particular season, injured plants may hit the end of the growing season before they are able to fully recover. Several authors (Arle 1954; Behrens 1955; Carns and Goodman 1956; Miller et al. 1963) documented that in seasons with delayed frosts and extended growing seasons, yield losses from 2,4-D exposures during vegetative, preflowering squaring, and flowering stages were substantially reduced from losses observed during shorter growing seasons. Cultivars and crop genetics are also likely key factors, but the effect of crop cultivar on sensitivity was only well explored for soybean. Weidenhamer et al. (1989) found that for flowering stage exposures to dicamba, a determinate soybean cultivar that ceases vegetative growth at the onset of flowering was less sensitive than an indeterminate cultivar. Wax et al. (1969) also commented that indeterminate cultivars were likely to be more sensitive to dicamba during flowering stages than determinate cultivars, but that determinate cultivars may be more sensitive during vegetative stages. Auch and Arnold (1978) observed variation for dicamba sensitivity across cultivars, but did not find that any cultivars were especially tolerant. Several authors working with cotton discussed the possibility of selecting cotton cultivars with increased tolerance to 2,4-D (Charles et al. 2007; Marple et al. 2008), but none compared different cultivars statistically. Particularly with 2,4-D, the specific active ingredient and formulation of the herbicide was also identified as an important factor. Both cotton and soybean were consistently shown to be more sensitive to esters vs. amine simulated drift of 2,4-D. As part of their resistant crop cultivar technology packages, Dow AgroSciences and Monsanto/BASF are both promoting new low volatility formulations (Dow AgroSciences 2011a; Thomas et al. 2012). Dow is promoting Cholex-D, a quaternary choline salt of 2,4-D, and BASF (the primary manufacturer of dicamba and business partner of Monsanto) has developed EnGenia, an aminopropyl methylamine salt of dicamba. However, there are currently no published data on how susceptible crops will respond to these formulations. Only one study in this dataset systematically compared dicamba formulations (Weidenhamer et al. 1989) and found no difference in soybean susceptibility to dimethlylamine vs. sodium salt dicamba treatments. Depending on the nature of the adjuvant, incorporating 202 N Weed Science 62, January–March 2014 adjuvants were observed to either increase or have no effect on the sensitivity of soybean to simulated dicamba or 2,4-D drift (Table 4). Herbicide carrier volume was addressed in one study as an important factor influencing crop sensitivity (Banks and Schroeder 2002) and has also been highlighted by authors conducting simulated drift studies on other crop–herbicide combinations (Ellis et al. 2002; Roider et al. 2008). In simulated drift studies, experimenters typically hold the carrier volume constant while reducing the grams per hectare dose, thus effectively also reducing the grams per liter herbicide concentration of the treatment. When carrier volume is reduced across a dose gradient, such that grams per liter concentrations are kept constant, the crop response is often more severe. Most of the studies in this dataset used field application rate carrier volumes (,187 L ha21). Thus, the dose–response curves presented here may in fact underestimate the real yield losses that can occur from particle or vapor drift exposures. Considering the range of factors affecting crop sensitivity to herbicide drift, it is important to consider that these dose–response curves are not meant to predict yield loss in any specific field event. Instead, global dose–response curves can provide a statistically valid estimate of the mean and variation of potential yield loss and also highlight the important differences between crops, herbicides, and growth stages at the time of exposure. Visual Injury and Yield Loss. From the subset of studies that measured both yield loss and visual injury symptoms ,14 DAT, linear models with visual injury symptoms generally overestimated yield loss (Figure 3). For both cotton and soybean, data was mainly available only for vegetative growth stages, so it remains unclear how well the patterns documented in Figure 3 translate across growth stages, or to other circumstances beyond this sample of studies. For soybean, injury seemed to correlate fairly closely with yield loss for both dicamba (r2 0.62) and 2,4-D (r2 0.61). However, for both herbicides the slope was somewhat above unity, indicating that a linear model for injury will overestimate yield loss and that plants exposed in vegetative stages can generally grow out of low to moderate injury symptoms. For cotton and dicamba, injury appeared to greatly overestimate yield loss, indicating that plants may sometimes express severe synthetic auxin injury symptoms but will generally grow out of the injury without suffering ER 250 (275 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 142 of 245 Soybean and Dicamba Soybean and 2,4-D 12 1.2 o vegetative A flowering A 1.0 ..... "' 08 ' ' 0 ' ' 0.6 I .. ' -eg 0 • ' ' ~ '' ' 0.6 '' ' ' o• '' ' ' --o• 0 00 02 00 0 12 (ii . + OA "0 >= ' .... 0 10 ''"-'-· .....,_' + .... 08 o\c o •o 0 0.8 0.0 10 ' ' A 0.4 + . A 02 A . ~ 0 '' ' + '' '' 02 0.4 0.6 0 '' ' '' 06 ' "o' 0.8 1.0 0 + A o : o 10 - 0 00 • . A "< 0.6 vegetative + + 04 '' ' ' •'' + o ' o'- AA 02 '' ' 08 0 0 • .... A ' ' .... • 0 0 0 08 A squaring + flowering 00 00 ' A 12 .. '+' 04 0.2 ~ 0 '' Cotton and 2,4-D . . • . .. •• o.,.. .. '' ' ' ,. A 06 . 0 '' ' () 0 Cotton and Dicamba c. e .eo 06 0.4 '' 02 , .. bO c c ' 04 0.2 :e0 0 ~ 08 • 04 - 10 0 0 ' ' '' ' ~· 0 00 02 0.4 06 oo '' 0.0 10 O...o 08 ' 10 Injury Figure 3. Correlations between yields of soybean and cotton and visual ratings of dicamba or 2,4 0 14 d after rreatment. Data are compiled from previously published experimentS evaluating the response of soybean or cotton to simulated dicarnba and 2,4 0 drift. Solid lines reflect the fit of linear regression models and dashed lines reflea an idealized 1:1 exact correlation. Symbols reflect different crop growth stages, but regressions for each crop and herbicide were calculated with all growth stages combined. Regression equations for each panel are as follows: soybean and dicamba, Yield= 1.08 0.81 X Injury (r2 = 0.62); soybean and 2,4 0, Yield= 1.02 0.64 X Injury (i = 0.61); cotton and dicarnba, Yield= 0.92 0.17 X Injury (r2 = 0.01); cotton and 2,4 0 , Yield= 0.95 0.72 X Injury (? = 0.32). substantial yield loss. For cotton and 2,4-D the slope of the trend line was close to unity, indicating broad agreement between injury sympmms and yidd loss. However, there was large variation around the trend line indicating that in specific circumstances, injury can only serve as a rough predictor of final yield loss. U tility of Meta-Analysis. Meta-analysis has long been a standard research mol in the biomedical sciences (Cooper et al. 2 009) bur has been sddom applied in weed science research. For instance, a recent Web of Science search on the topic " metaanalysis" in the journals Weed Science and Weed Technology retrieved only three publications (Rinella and Sheley 2005; Schutte et al. 2010; Wagner et al. 2007). Statistical approaches for the meta-analysis of dose- response curves continue to be developed (Ba.gnardi et al. 2004; Paul et al . 2006; Ritz and Streibig 2008; Thompson and Higgins 2002). A5 has been demonstrated in this paper, these approaches can readily be adapted to the symhesis of data on dose- response patterns of crops and weeds to herbicides. As weed science continues to address changes in weed comrnunmes across cropping systems, the evolution of new resistant weed species, and d1e commercialization of new herbicide-resistance traits, carefully synthesized information describing the sensitivity of crops m herbicide active ingredients will continue to be Egan et al.: 2,4-D and dicamba drift on soybean and cotton ER 251 203 (276 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 143 of 245 useful. During the time work was progressing on this meta-analysis, other research groups interested in synthetic auxin drift from herbicide-resistant cropping systems published results from new field experiments assessing the response of cotton and soybean to simulated dicamba and 2,4-D drift (Johnson et al. 2012; Robinson et al. 2013). While these studies added several valuable data points to this meta-analysis, the multiple site-year experiments described in these papers were no doubt very costly and time-consuming to conduct, and on their own, they provide an understanding of cotton and soybean sensitivity to herbicide drift that is limited to their site and experimental conditions. When an opportunity arises to use existing chemistries in new contexts, meta-analysis approaches can provide a supplement to new experiments and can be a powerful and costeffective approach towards understanding the dose– response patterns of weeds and crops. Acknowledgments The authors thank Bill Curran for helpful insights on this project and manuscript. They would also like to thank Duy Vu and Wen-Yu Hua at the PSU Statistical Consulting Center for helpful guidance and suggestions on data analysis methods. Kathy Fescemeyer at the PSU Life Sciences Library provided assistance developing literature search and evaluation methods. Gustavo Camargo and Alexander Savelyev provided assistance evaluating articles in Portuguese and Russian. The authors also thank Sarah Goslee, Jason Hill, and Eric Nord for help programming in R. This work was supported through an EPA STAR (FP917131012) fellowship awarded to J. Franklin Egan. Mention of trade names or commercial products in this publication is solely for the purpose of providing specific information and does not imply recommendation or endorsement by the U.S. Department of Agriculture. USDA is an equal opportunity provider and employer. Literature Cited [AAPCO] Association of American Pesticide Control Officials (2005) Pesticide Drift Enforcement Survey Report. Milford, DE: AAPCO. http://aapco.ceris.purdue.edu/doc/surveys/Drift Enforce05Rpt.htm. Accessed November 26, 2012. Al Khatib K, Peterson D (1999) Soybean (Glycine max) response to simulated drift from selected sulfonylurea herbicides, dicamba, glyphosate, and glufosinate. Weed Technol 13:264 270 Andersen SM, Clay SA, Wrage LJ, Matthees D (2004) Soybean foliage residues of dicamba and 2,4 D and correlation to application rates and yield. Agron J 96:750 760 Arle HF (1954) The sensitivity of Acala 44 cotton to 2,4 D. Pages 20 25 in Proceedings of the 14th Western Weed Control Conference. Tuscon, AZ: Western Weed Science Society 204 N Weed Science 62, January–March 2014 Auch DE, Arnold WE (1978) Dicamba use and injury on soybeans (Glycine max) in South Dakota. Weed Sci 26:471 475 Bagnardi V, Zambon A, Quatto P, Corrao G (2004) Flexible meta regression functions for modeling aggregate dose response data, with an application to alcohol and mortality. Am J Epidemiol 159:1077 1086 Banks PA, Schroeder J (2002) Carrier volume affects herbicide activity in simulated spray drift studies. Weed Technol 16:833 837 Barker GL, Hatfield JL, Wanjura DF (1985) Cotton phenology parameters as affected by wind. Field Crops Res 12:33 47 Bednarz CW, Nichols RL (2005) Phenological and morpholog ical components of cotton crop maturity. Crop Sci 45: 1497 1503 Behrens R, Hall WC, Fisher CE (1955) Field responses of cotton to four phenoxy type herbicides. Pages 72 75 in Proceedings of the 8th Southern Weed Control Conference. St. Petersburg, FL: Southern Weed Science Society Behrens MR, Mutlu N, Chakraborty S, Dumitru R, Jiang WZ, LaVallee BJ, Herman PL, Clemente TE, Weeks DP (2007) Dicamba resistance: Enlarging and preserving biotechnology based weed management strategies. Science 316:1185 1188 Behrens R, Lueschen WE (1979) Dicamba volatility. Weed Sci 27:486 493 Boerboom C (2004) Field case studies of dicamba movement to soybeans. Wisconsin Fertilizer, Aglime, and Pest Management Conference, Madison, WI. 5 p Brown RB, Carter MH, Stephenson GR (2004) Buffer zone and windbreak effects on spray drift deposition in a simulated wetland. Pest Manag Sci 60:1085 1090 Burnham KP, Anderson DM (2002) Model Selection and Multimodel Inference. New York: Springer Verlag. 488 p Carlsen SCK, Spliid NH, Svensmark B (2006) Drift of 10 herbicides after tractor spray application. 2. Primary drift (droplet drift). Chemosphere 64:778 786 Carns H, Goodman VH (1956) Responses of cotton to 2,4 D Bulletin of the Mississippi Agricultural Experiment Station. 15 p Cedergreen N, Ritz C, Streibig JC (2005) Improved empirical models describing hormesis. Environ Toxicol Chem 24: 3166 3172 Centre for Agriculture and Biosciences International (2012) CAB Direct Database. http://www.cabdirect.org. Accessed August November 2011. Charles GW, Constable GA, Llewellyn DJ, Hickman MA (2007) Tolerance of cotton expressing a 2,4 D detoxification gene to 2,4 D applied in the field. Aust. J. Agric. Res 58:780 787 Constantin J, Oliveira Ju´nior RS de, Fagliari JR, Pagliari PH, Arantes JGZ de, Cavalieri SD, Framesqui VP, Gonc¸alves DA (2007) Effect of sub lethal dosages of 2,4 D on cotton yield and crop susceptibility as a function of its development stage. Engenharia Agrı´cola 27:24 29 Cooper H, Hedges LV, Valentine JC (2009) The Handbook of Research Synthesis and Meta Analysis. 2nd ed. New York: Russel Sage Foundation. 615 p Crawley MJ (2005) Statistics: An Introduction Using R. West Sussex, UK: John Wiley and Sons. Pages 33 50 de Jong FMW, de Snoo GR, van de Zande JC (2008) Estimated nationwide effects of pesticide spray drift on terrestrial habitats in the Netherlands. J Environ Manage 86:721 730 ER 252 (277 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 144 of 245 Dow AgroSciences (2011a) Research Demonstrating Reduction in Potential Off Target Movement for 2,4 D Choline and Colex DTM Technology as Part of the EnlistTM Weed Control System. M Krieger, editor. Supplemental Information for Petition for Determination of Nonregulated Status for Herbicide Tolerant DAS 68416 4 Soybean. 13 p Dow AgroSciences (2011b) Stewardship of Herbicide Tolerant Trait Technology for DAS 68416 4 Soybean. M Krieger, editor. Supplemental Information for Petition for Determina tion of Non regulated Status for Herbicide Tolerant DAS 68416 4 Soybean. 12 p Egan JF, Mortensen DA (2012) Quantifying vapor drift of dicamba herbicides applied to soybean. Environ Toxicol Chem 31:1023 1031 Ellis JM, Griffin JL, Jones CA (2002) Effect of carrier volume on corn (Zea mays) and soybean (Glycine max) response to simulated drift of glyphosate and glufosinate. Weed Technol 16:587 592 Epps EA (1953) Growth regulators. Effect of 2,4 D on growth and yield of cotton. J Agric Food Chem 1:1009 1010 Everitt JD, Keeling JW (2009) Cotton growth and yield response to simulated 2,4 D and dicamba drift. Weed Technol 23:503 506 Fox J, Weiseberg S (2010) An R Companion to Applied Regression. 2nd ed. Thousand Oaks, CA: Sage Publications. 249 p Goodman VH (1953) The yield and progeny seedling responses of cotton to treatment with 2,4 D, 2,4,5 T, and MCPA. Pages 47 56 in Proceedings of the 6th Southern Weed Control Conference. Knoxville, TN: Southern Weed Science Society Goodman VH, Ennis WB, Palmer RD (1955) Cotton responses to 2,4 D, 2,4,5 T, MCPA and related growth regulators. Pages 76 81 in Proceedings of the 8th Southern Weed Control Conference. St. Petersburg, FL: Southern Weed Science Society Grover R, Yoshida K, Maybank J (1972) Droplet and vapor drift from butyl ester and dimethylamine salt of 2,4 D. Weed Sci 20:320 324 Hamilton KC, Arle HF (1979) Response of cotton (Gossypium hirsutum) to dicamba. Weed Sci 27:604 607 Hedges LV, Gurevitch J, Curtis PS (1999) The meta analysis of response ratios in experimental ecology. Ecology 80:1150 1156 Hill BD, Harker KN, Hasselback P, Moyer JR, Inaba DJ, Byers SD (2002) Phenoxy herbicides in Alberta rainfall: Potential effects on sensitive crops. Can J Plant Sci 82:481 484 Johnson VA, Fisher LR, Jordan DL, Edmisten KE, Stewart AM, York AC (2012) Cotton, peanut, and soybean response to sublethal rates of dicamba, glufosinate, and 2,4 D. Weed Technol 26:195 206 Kelley KB, Wax LM, Hager AG, Riechers DE (2005) Soybean response to plant growth regulator herbicides is affected by other postemergence herbicides. Weed Sci 53:101 112 Kittock DL, Arle HF (1977) Termination of late season cotton fruiting with plant growth regulators. Crop Sci 17:320 324 Kittock DL, Mauney JR, Arle HF, Bariola LA (1973) Termination of late season cotton fruiting with growth regulators as an insect control technique. J Environ Qual 2:405 408 Kuehl RO (2000) Design of experiments: statistical principles of research design and analysis. 2nd edition. Pacific Grove, CA: Brooks and Cole. 666 p Lanini WT (2000) Simulated drift of herbicides on grapes, tomatoes, cotton, and sunflower. Pages 107 110 in Proceed ings of the Annual Meeting of the California Weed Science Society. Fremont, CA: California Weed Science Society Louisiana Department of Agriculture and Forestry. 2,4 D Herbicide Regulations. Louisiana Department of Agriculture and Forestry Pesticide Rules and Regulations, Section 143. 24 p Marple ME, Al Khatib K, Peterson DE (2008) Cotton injury and yield as affected by simulated drift of 2,4 D and dicamba. Weed Technol 22:609 614 Marple ME, Al Khatib K, Shoup D, Peterson DE, Claassen M (2007) Cotton response to simulated drift of seven hormonal type herbicides. Weed Technol 21:987 992 Merotto A Jr, Vidal RA, Fleck NG (1999) Soybean tolerance to synthetic auxin and potential of mixtures with protox inhibiting herbicides. Pages 319 324 in Brighton Crop Protection Conference: Weeds. Brighton, UK: British Crop Protection Council Miller JH, Kempen HM, Wilkerson JA, Foy CL (1963) Response of cotton to 2,4 D and related phenoxy herbicides. United States Department of Agriculture Technical Bulletin 1289. 28 p Monaco TJ, Weller SC, Ashton FM (2002) Weed Science: Principles and Practices. New York: John Wiley and Sons. Pp 291 310 Mortensen DA, Egan JF, Maxwell BD, Ryan MR, Smith RG (2012) Navigating a critical juncture for sustainable weed management. BioScience 62:75 85 Oosterhuis DM, Jernstedt J (1999) Morphology and anatomy of the cotton plant. Pages 175 206 in Smith CW and Cothren JT, eds. Cotton: Origin, History, Technology, and Produc tion. New York: Wiley and Sons Paul PA, Lipps PE, Madden LV (2006) Meta analysis of regression coefficients for the relationship between fusarium head blight and deoxynivalenol content of wheat. Phytopa thology 96:951 961 Pedersen P (2004) Soybean Growth and Development. Ames, IA: Iowa State University Extension. 28 p R Core Team (2011) R: A Language and Environment for Statistical Computing. Vienna, Austria: R Foundation for Statistical Computing, ISBN: 3 900051 07 0. Available on line at http://www.R project.org/ Rinella MJ, Sheley RL (2005) A model for predicting invasive weed and grass dynamics. II. Accuracy evaluation. Weed Sci 53:605 614 Ritz C, Streibig JC (2005) Bioassay analysis using R. J Stat Softw 12:1 22 Ritz C, Streibig JC (2008) Nonlinear Regression with R. New York: Springer Science and Business Media. 148 p Robinson AP, Davis VM, Simpson DM, Johnson WG (2013) Response of soybean yield components to 2,4 D. Weed Sci 61:68 76 Roider CA, Griffin JL, Harrison SA, Jones CA (2008) Carrier volume affects wheat response to simulated glyphosate drift. Weed Technol 22:453 458 Schutte BJ, Hager AG, Davis AS (2010) Respray requests on custom applied, glyphosate resistant soybeans in Illinois: how many and why. Weed Technol 24:590 598 Slife FW (1956) The effect of 2,4 D and several other herbicides on weeds and soybeans when applied as post emergence sprays. Weeds 4:61 68 Egan et al.: 2,4-D and dicamba drift on soybean and cotton ER 253 N 205 (278 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 145 of 245 Smith DT, Wiese AF (1972) Cotton Response to Low Rates of 2,4 D and Other Herbicides Texas Agricultural Experiment Station Bulletin No. B 1120. 8 p Smith RJ (1965) Effects of chlorophenoxy herbicides on soybeans. Weeds 13:168 169 Staten G (1946) Contamination of cotton fields by 2,4 D or hormone type weed sprays. Agron J 38:536 544 Texas Department of Agriculture (2012) Regulated Herbicides and Regulated Herbicide Counties. http://www.texasagriculture. gov/RegulatoryPrograms/Pesticides/ RegulatedHerbicides/Regulated HerbicidesCounties.aspx. Accessed November 20, 2012 Thomas WE, Staal M, Bowe S, Bozeman L (2012) Engenia herbicide: maximizing on target spray deposition. Page 83 in ASA, CSSA, SSSA International Meetings. Cincinnati, OH: ASA, CSSA, SSSA Thompson SG, Higgins JPT (2002) How should meta regression analyses be undertaken and interpreted? Stat Med 21:1559 1573 Tuduri L, Harner T, Blanchard P, Li YF, Poissant L, Waite DT, Murphy C, Belzer W (2006) A review of currently used pesticides (CUPS) in Canadian air and precipitation. Part 2: Regional information and perspectives. Atmos Environ 40: 1579 1589 [US EPA] United States Environmental Protection Agency (2006) Reregistration Eligibility Decision for Dicamba and Associated Salts. Washington, DC: US EPA. 165 p Wagner NC, Maxwell BD, Taper ML, Rew LJ (2007) Developing an empirical yield prediction model based on wheat and wild oat (Avena fatua) density, nitrogen and 206 N Weed Science 62, January–March 2014 herbicide rate, and growing season precipitation. Weed Sci 55:652 664 Waltz E (2010) Glyphosate resistance threatens Roundup hegemony. Nat Biotechnol 28:537 538 Wang M, Rautman D (2008) A simple probabilistic estimation of spray drift factors determining spray drift and development of a model. Environ Toxicol Chem 27:2617 2626 Watson AJ (1955) The response of cotton to low rates of 2,4 D, 2,4,5 T, MCP and Silvex. Pages 82 86 in Proceedings of the 8th Southern Weed Control Conference. St. Petersburg, FL: Southern Weed Science Society Wax LM, Knuth LA, Slife FW (1969) Response of soybeans to 2,4 D, dicamba and picloram. Weed Sci 17:388 393 Weidenhamer JD, Triplett GB, Sobotka FE (1989) Dicamba injury to soybean. Agron J 81:637 643 Wiese AF, Martin AG (1963) Toxicity of benzoic acid herbicide to cotton and soybeans. Weeds 11:7 10 Wright TR, Shan GM, Walsh TA, Lira JM, Cui C, Song P, Zhuang MB, Arnold NL, Lin GF, Yau K, Russell SM, Cicchillo RM, Peterson MA, Simpson DM, Zhou N, Ponsamuel J, Zhang ZY (2010) Robust crop resistance to broadleaf and grass herbicides provided by aryloxyalkanoate dioxygenase transgenes. Proc Natl Acad Sci USA 107: 20240 20245 Zar JH (2009) Biostatistical Analysis. 5th ed. Upper Saddle River, NJ: Pearson. 960 p Received February 19, 2013, and approved July 23, 2013. ER 254 (279 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 146 of 245 Weed Science, 53:101–112. 2005 Soybean response to plant growth regulator herbicides is affected by other postemergence herbicides Kevin B. Kelley Department of Crop Sciences, University of Illinois, Urbana, IL 61801 Loyd M. Wax U.S. Department of Agriculture–Agricultural Research Service, Department of Crop Sciences, University of Illinois, Urbana, IL 61801 Aaron G. Hager Department of Crop Sciences, University of Illinois, Urbana, IL 61801 Dean E. Riechers Corresponding author. Department of Crop Sciences, University of Illinois, Urbana, IL 61801; riechers@uiuc.edu Two studies investigated off-target exposure of soybean to plant growth regulator (PGR) herbicides and determined if simultaneous exposure to PGR herbicides and labeled soybean herbicides increase PGR injury. The PGR herbicides, 2,4-D, clopyralid, and dicamba, as well as dicamba plus the auxin transport inhibitor diflufenzopyr, were applied to glyphosate-resistant soybean at the V3, V7, and R2 soybean growth stages. Two rates were chosen from previous and preliminary research to approximate threshold rates that would cause a yield reduction so as to distinguish differences in sensitivity between growth stages. All four PGR herbicides caused significant soybean injury, height reduction, and yield loss at one or more application rates and growth stages. Relative to other PGR herbicides, dicamba reduced soybean yield at the lowest rate (a potential rate from residues remaining in improperly cleaned application equipment), followed by clopyralid, with 2,4-D requiring the highest rate to reduce soybean yield (a potential rate from a high level of spray drift). Dicamba and dicamba plus diflufenzopyr were applied at equal fractions of labeled use rates for corn to compare them directly at equivalent levels of off-target movement. Dicamba plus diflufenzopyr caused less injury and yield loss than dicamba applied alone. In a second study, the highest labeled soybean use rates of glyphosate, imazethapyr, imazamox, and fomesafen were applied alone and in combination with the highest rate of dicamba used in the first study (1% of a labeled use rate for corn) at the V3 and V7 stages. Dicamba demonstrated synergistic interactions with imazamox, imazethapyr, and fomesafen (but not with glyphosate) to further reduce yield under some circumstances, especially when applied at the V7 stage. Several treatments that included dicamba reduced soybean seed weight when applied at either the V3 or V7 stage and reduced the number of seeds per pod at the V7 stage. Nomenclature: clopyralid; 2,4-D; dicamba; diflufenzopyr; fomesafen; glyphosate; imazamox; imazethapyr; corn, Zea mays L.; soybean, Glycine max (L.) Merr. ‘Pioneer 94B01RR’. Key words: Auxinic herbicides, crop injury, herbicide interaction, spray drift, spray tank contamination, synergy. Plant growth regulator (PGR) herbicides have been widely used in monocotyledonous crops for many years and effectively control a broad spectrum of dicotyledonous weeds. Compared with herbicides with other modes of action, weed resistance to PGR herbicides has been slow to develop (Sterling and Hall 1997), which also increases their appeal. However, soybean is frequently grown in close proximity and often in rotation with monocot crops and is very sensitive to PGR herbicides (Al-Khatib and Peterson 1999; Wax et al. 1969). Reports of soybean injury with symptoms resembling off-target exposure to PGR herbicides have been widespread and recurring (Boerboom 2004; Hager and Nordby 2004), although the cause of injury is not often readily identifiable. PGR herbicide injury to soybean can result in yield loss, but abnormal foliar symptoms and other developmental abnormalities can occur at rates lower than those required to reduce yield (Auch and Arnold 1978; Wax et al. 1969; Weidenhamer et al. 1989). The PGR herbicides most commonly used in close proximity to soybean fields include 2,4-D, clopyralid, and dicamba. Also, the auxin transport inhibitor diflufenzopyr is used in combination with dicamba and synergizes its activity on dicot weeds (Grossman et al. 2002), although there is no information available on the effect that the addition of diflufenzopyr to dicamba has on potential soybean injury. Soybean differ in sensitivity between dicamba and 2,4-D. When directly applied at the V3 soybean growth stage, 5.6 g haϪ1 of dicamba (1% of a labeled use rate for corn) reduced soybean yield 14 to 34%, whereas 112 g haϪ1 of 2,4-D (20% of a labeled use rate for corn) was required to cause a similar reduction (25 to 32%) (Andersen et al. 2004). In addition, off-target movement of dicamba has been reported to result in more soybean injury than 2,4-D. In 1974 in Minnesota, postemergence (POST) use of dicamba and 2,4-D in corn resulted in 68 reports of dicamba injury to soybean and 7 reports of 2,4-D injury to soybean, although 2,4-D was applied to over three times as many hectares of corn as was dicamba (Behrens and Lueschen 1979). Clopyralid also has been shown to cause soybean injury (Bovey and Meyer 1981). A 50% soybean yield reduction was caused by nearly equal rates of clopyralid and dicamba (Smith and Geronimo 1977), although the rates were not reported. Soybean sensitivity to PGR herbicides varies at different growth stages. Dicamba caused greater yield reductions when exposure occurred at a late vegetative or early reproductive stage, relative to an early vegetative stage (Auch and Arnold 1978; Wax et al. 1969). Reports of soybean sensiKelley et al.: Off-target growth regulator herbicide injury ER 255 • 101 (280 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 147 of 245 tivity to 2,4-D are somewhat conflicting, however, with 2,4D causing the greatest yield response when applied at early vegetative stages (Smith 1965), whereas others report little difference in sensitivity between growth stages (Wax et al. 1969), and yet others report that soybean is more sensitive to 2,4-D as it grows taller (Slife 1956). Little has been reported about soybean sensitivity to clopyralid at different growth stages. PGR herbicides can unintentionally come in contact with soybean and cause injury through several routes of exposure. Spray particles or volatile active ingredients can drift from neighboring fields. Spray particles can drift in air currents with injury often showing a pattern that follows wind direction (Bode 1987), and many herbicide labels have statements regarding wind speed and drift. Risk of vapor drift depends on the volatility of the herbicide formulation used and can be influenced by environmental factors. Short-chain esters of 2,4-D are very volatile, whereas volatility is lower with long-chain esters and is almost eliminated by amine salts of 2,4-D (Que Hee and Sutherland 1974). Dicamba can volatilize as the free acid and injure soybean even when applied as the dimethylamine salt formulation (Behrens and Lueschen 1979). However, dicamba volatility is reduced by lower temperatures and higher relative humidity. PGR herbicide residues remaining in application equipment after previous applications to a corn crop can also be dislodged when the spray equipment is used in soybean. Labels of products containing dicamba provide information describing how to clean equipment to remove these residues. However, even after following recommended cleaning procedures, dicamba residues can remain in application equipment and be detected in a subsequent spray solution at levels as high as 0.63% of a field use rate in corn (Boerboom 2004). Previous research has described the effects of PGR herbicides on soybean growth and yield when these herbicides are applied alone. However, it is not currently known if there is an interaction between PGR herbicides and herbicides labeled for POST use in soybean that may increase injury. Data from the National Agricultural Statistics Service (NASS 2002) indicate there has been an increase in the use of POST herbicides in soybean with a concomitant decrease in the use of soil-applied herbicides. The increase in POST herbicide use in soybean increases the potential for herbicides labeled for use in soybean to be present when off-target soybean exposure to PGR herbicides occurs. Dicamba and clopyralid interacted with diclofop to increase yield loss in sunflower (Helianthus annuus L.) and lentils (Lens culinaris L.), respectively (Derksen 1989). PGR herbicides could also potentially interact with soybean herbicides to increase soybean injury. An interaction is possible if PGR herbicide residues are not cleaned from application equipment or if a PGR herbicide drifts from neighboring fields at or near the time of a herbicide application to soybean. The increased dependence on POST herbicides in soybean increases the necessity to understand how herbicides labeled for use in soybean affect soybean exposed to PGR herbicides. In this study, PGR herbicides commonly used near soybean fields were applied directly to soybean at reduced rates at different growth stages to determine the effect of offtarget PGR herbicide exposure on growth, development, and yield. Soybean herbicides with different modes of action 102 • Weed Science 53, January–February 2005 were included for comparison and to obtain tissue samples for lab analysis (Kelley et al. 2004). In addition, dicamba and several soybean herbicides were applied alone and in combination at two vegetative growth stages to determine whether the presence of POST herbicides labeled for use in soybean would increase the injury caused by dicamba. Dicamba was chosen because of its widespread use in corn and the high number of soybean injury reports attributed to dicamba. Materials and Methods Two soybean field experiments were conducted at the Crop Sciences Research and Education Center in Urbana, IL. Fields were planted to corn in previous years and had been chisel plowed each fall after corn harvest. In the spring, fields were tilled with a field cultivator. Glyphosate-resistant soybean variety ‘Pioneer 94B01RR’ was planted in 0.76-m rows at a rate of 400,000 seeds haϪ1 in 2001 and 2002 and 420,000 seeds haϪ1 in 2003. Plots were kept weed free with a preemergence application of 2.14 kg haϪ1 metolachlor, 44 g haϪ1 chlorimuron-ethyl, and 0.27 kg haϪ1 metribuzin. All treatments were applied with a CO2-pressurized backpack sprayer equipped with a 2.3-m-wide handheld boom and five 8003 flat-fan nozzles1 spaced 46 cm apart that delivered 187 L haϪ1 at 221 kPa. The spray boom, narrower than the plot width (3.0 m), was centered over each four-row plot so that the two outside rows were not completely within the spray pattern and acted as a buffer to reduce movement between adjacent plots. Applications were made under mostly calm conditions (wind speed was 4 m sϪ1 or less) to further reduce drift. PGR Herbicide Study To evaluate the effects of current PGR herbicides on soybean development, reduced rates of PGR herbicides were applied in 2001, 2002, and 2003. The soil was a Flanagan silt loam (fine, smectitic, mesic Aquic Argiudolls) in 2001 and 2003 and a Catlin silt loam (fine-silty, mixed, superactive, mesic Oxyaquic Argiudolls) in 2002. The soil organic matter was 4.8, 4.0, and 4.8%, and the soil pH was 6.6, 6.5, and 6.6, respectively. Soybean was planted on May 30, 2001, June 1, 2002, and May 21, 2003. Treatments included the diglycolamine salt of dicamba, the sodium salt of dicamba plus the sodium salt of diflufenzopyr, the monoethanolamine salt of clopyralid, the isooctylester formulation of 2,4-D, imazethapyr as a free acid, the isopropylamine salt of glyphosate, and the sodium salt of fomesafen, each applied at the soybean growth stages and rates presented in Table 1. Imazethapyr, fomesafen, and glyphosate are three of the most commonly used POST herbicides labeled for use in soybean and were included in the experiment so that PGR herbicide injury could be compared with the effects of herbicides labeled for use in soybean. The rates chosen for the PGR herbicides were based on preliminary research (data not shown) to bracket the threshold rate that would cause a yield reduction so as to distinguish any differences in soybean sensitivity to these herbicides at the different growth stages. Less dicamba was included with diflufenzopyr than dicamba applied alone, although these are equal fractions of corn field use rates because diflufenzopyr allows for less dicamba to provide similar weed control ER 256 (281 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 148 of 245 TABLE 1. Soybean injury caused by reduced rates of PGR herbicides applied at the V3, V7, and R2 stages of soybean growth combined across 2001, 2002, and 2003.a,b Late soybean injury Early soybean injuryc 2 WAT Herbicide Dicamba Dicamba ϩ diflufenzopyr 2,4-D Clopyralid Imazethapyr Glyphosate Fomesafen Untreated control Rate g ae/ha 0.56 5.6 0.2 ϩ 0.08 2.0 ϩ 0.8 56 180 2.1 6.6 71 840 330 V3 V7 R2 6 WAT 6–7 WAT 4–5 WAT V3 V7 R2 16 c 29 ab 9d 21 bc 3e 30 ab 17 c 33 a 0e 0e 0e 0e 23 cd 36 b 12 e 28 c 4f 20 d 26 cd 61 a 0g 0g 0g 0g 26 b 38 a 18 c 28 b 7d 25 bc 28 b 45 a 0f 0f 2e 0f % 37 d 50 b 22 e 42 cd 8f 49 bc 41 cd 65 a 4g 1h 8f 0h 31 e 41 c 17 f 38 cd 22 f 52 b 32 de 64 a 6g 1h 8g 0h 25 e 41 b 18 f 34 cd 19 f 37 bc 29 de 47 a 1i 5h 12 g 0i a Abbreviations: PGR, plant growth regulator; WAT, weeks after treatment. b Means within a column (treatments applied at the same growth stage) followed by the same letter(s) are not significantly different according to Fisher’s Protected LSD (0.05). c Visual injury ratings on a scale of 0 to 100% with 0% ϭ no injury and 100% ϭ complete death. (Grossman et al. 2002). This allows a direct comparison of the effect that the addition of diflufenzopyr to dicamba has on the potential for soybean injury caused by off-target exposure. The fractions of a field use rate in corn represented in this study are 0.1 and 1% for dicamba or dicamba plus diflufenzopyr, 10 and 32% for 2,4-D, and 1 and 3.2% for clopyralid. The higher rates of 2,4-D and clopyralid were not included in 2001 but were added in 2002 and 2003. Because application equipment cleaned using recommended procedures may contain dicamba residues as high as 0.63% of a field use rate (Boerboom 2004), equipment that was not properly cleaned could contain PGR herbicide levels similar to the rates applied in this study of dicamba, dicamba plus diflufenzopyr, or even possibly clopyralid. Also, if a PGR herbicide is applied adjacent to a soybean field at a high spray pressure and with high wind speeds, it is feasible for a PGR herbicide to drift onto soybean at rates as high as the rates of 2,4-D applied in this study. All PGR herbicides were applied with 0.25% (v/v) of a nonionic surfactant.2 Glyphosate was applied with ammonium sulfate at 1.9 kg haϪ1. Methylated seed oil (MSO)3 and 28% urea ammonium nitrogen (UAN) were each included with imazethapyr at 1.25% (v/v) and with fomesafen at 1.0 and 2.5% (v/v), respectively. Soybean growth stages for PGR herbicide applications were chosen to include a vegetative stage when many herbicides are commonly applied to corn (soybean V3 stage), a growth stage when later rescue treatments for weed escapes in corn are often applied (soybean V7 stage), and a reproductive stage when drift from other sources, such as noncrop and pasture areas, may occur. The experiment was established as a randomized complete block design with three replications and a factorial arrangement of treatments. Herbicide treatments and growth stages were separate factors. Plots measured 3.0 m wide by 9.1 m in length. All herbicide treatments were applied to soybean in the V3 stage 30 to 37 d after planting (DAP), the V7 stage 43 to 51 DAP, and the R2 stage 61 to 66 DAP. At the V3 application, soybean were 9 cm tall in 2001, 16 cm tall in 2002, and 22 cm tall in 2003. At the V7 application, soybean were 31 cm tall in 2001, 38 cm tall in 2002, and 44 cm tall in 2003. At the R2 application, soybean were 65 cm tall in 2001, 70 cm tall in 2002, and 72 cm tall in 2003. Soybean injury and height were recorded 2 wk after treatment (WAT) and again 4 to 7 WAT, depending on the time of application. Visual soybean injury ratings were made on a scale of 0 to 100%, where 0 equals no crop injury and 100 equals complete crop death. Final height was measured when plants reached full height before leaf senescence. Delayed maturity was measured by recording the day on which 95% of the soybean pods in each plot reached a mature color and then comparing that with the day when the untreated control plots matured. Yield was measured by machine harvesting the center two rows from each plot and adjusting the moisture to 13%. Soybean Herbicide Interaction Study In 2002 and 2003, four herbicides labeled for use in soybean and a reduced rate of dicamba were applied alone and in combination to evaluate an interaction of soybean herbicides and injury caused by dicamba. The soil was a Flanagan silt loam in 2002, and a Drummer silty clay loam (fine-silty, mixed, superactive, mesic Typic Endoaquolls) in 2003. The soil organic matter was 4.8 and 5.4% and the soil pH was 6.3 and 6.6, respectively. Soybean was planted on June 3, 2002, and May 21, 2003. The isopropylamine salt of glyphosate, imazethapyr as a free acid, the ammonium salt of imazamox, and the sodium salt of fomesafen were applied with and without the diglycolamine salt of dicamba, as well as dicamba applied alone, at the growth stages and rates listed in Table 3. The adjuvants and rates included with each herbicide were the same as in the PGR herbicide study. Imazamox was applied with Kelley et al.: Off-target growth regulator herbicide injury ER 257 • 103 (282 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 149 of 245 MSO and 28% UAN, both at 1.25% (v/v). Soybean growth stages for soybean herbicide interaction applications were chosen to include an early vegetative stage (V3) when soybean herbicides are commonly applied and a late vegetative stage (V7) before flowering when rescue treatments for weed escapes are often applied. Drift of a PGR herbicide from outside the soybean field could injure soybean at any stage. However, soybean exposure to a PGR herbicide can occur in the presence of a herbicide labeled for use in soybean only when herbicides are applied to soybean (most commonly during vegetative growth stages). The rate of dicamba chosen to be sufficient to cause a yield reduction but not plant death is equivalent to the highest rate used in the PGR herbicide study and represents a potential rate from improperly cleaned application equipment (Boerboom 2004). The rates of the soybean herbicides were the maximum labeled rates at the time of application. With the highest labeled use rates for soybean herbicides and a rate of dicamba expected to cause a yield reduction, these treatments represent a worst-case scenario to determine whether there is potential for dicamba to interact with soybean herbicides and cause a greater yield loss in their presence than if soybean were exposed to dicamba alone. The experimental design and number of replications were the same as the PGR herbicide study. Plot size was 3.0 m wide by 11.6 m long in 2002, and 3.0 m wide by 9.1 m long in 2003. Treatments were applied to soybean in the V3 stage 30 to 37 DAP and the V7 stage 45 to 51 DAP. At the V3 application, soybean were 20 cm tall in 2002 and 22 cm tall in 2003. At the V7 application, soybean were 40 cm tall in 2002 and 50 cm tall in 2003. Soybean injury and height were recorded 2 and 6 WAT. Final soybean height, delayed maturity, and grain yield were measured in the same fashion as the PGR herbicide study. Before harvest, 10 plants in a row from the center of each plot were collected and used for yield component analysis. All data were analyzed with a mixed linear model with Statistical Analysis Systems (SAS 1999). In the PGR herbicide study, data from the 3 yr were combined and years were treated as random effects. In the soybean herbicide interaction study, each year was analyzed separately assuming that 2 yr are not a sufficient random sample to represent the larger population (Carmer et al. 1989). Visual injury data were transformed by arcsine square root before statistical analysis to stabilize variances. Untransformed data are presented with statistical interpretation based on transformed data. Visual injury data for applications at each growth stage were analyzed separately because of the data being collected at different times and under different conditions. Within each factor (herbicide treatment and growth stage), means were separated using Fisher’s Protected LSD at the 0.05 level of significance. Synergistic and antagonistic responses between dicamba and soybean herbicides were determined using the method described by Colby (1967) to calculate expected response of herbicide tank mixtures. Expected response values were calculated by expressing values as a percent of the untreated control, and taking the product of values for each herbicide applied alone included in the combination and dividing by 100. Synergistic or antagonistic responses were determined by significant differences between the expected and observed responses using Fisher’s protected LSD at the 0.05 level of 104 • Weed Science 53, January–February 2005 significance. When expected and observed responses are not significantly different, interactions between herbicides in a combination are considered additive. Results and Discussion PGR Herbicide Study By 2 wk after all applications (V3, V7, and R2), soybean had significant foliar injury in response to all PGR herbicides, with more injury as rates increased (Table 1). Dicamba and dicamba plus diflufenzopyr resulted in new trifoliolate leaves that were cupped and crinkled, with the higher rates resulting in smaller leaves and reduced overall growth compared with the lower rates (Figures 1A and 1B). Symptoms caused by 2,4-D included epinasty of leaves and stems and swollen, cracked stems. Clopyralid injury resembled dicamba injury, but there were more thin, strapped leaves with parallel venation and less cupping injury (Figures 1C and 1D). Similar symptoms have been described previously (AlKhatib and Peterson 1999; Andersen et al. 2004; Auch and Arnold 1978; Wax et al. 1969; Weidenhamer et al. 1989). Fomesafen caused temporary necrosis of leaf tissue but had no effect on subsequent growth, whereas imazethapyr temporarily stunted plant growth. Glyphosate caused no visible plant injury, except that the youngest leaves temporarily exhibited chlorosis after the R2 application. The terminal growing point was killed by the higher rate of dicamba or clopyralid at all application timings, by the higher rate of dicamba plus diflufenzopyr at V3 and V7, and by the lower rate of clopyralid at the V7 application. Two WAT at all growth stages, soybean plants treated with the higher rates of PGR herbicides were 10 to 50% shorter than the untreated control (data not shown). Soybean treated with the higher rates of 2,4-D or clopyralid at all growth stages showed little to no increase in height during the 2 wk after treatment. By 4 to 7 WAT, soybean had recovered from injury caused by fomesafen, imazethapyr, and glyphosate, and injury caused by most PGR herbicides had decreased (Table 1). Soybean treated with the lower rate of dicamba at V3 and the lower rates of dicamba plus diflufenzopyr and 2,4-D at both V3 and V7 showed signs of recovery (emerging trifoliolate leaves lacked injury symptoms). Injury symptoms from both rates of clopyralid and the higher rates of the other PGR herbicides remained more persistent, with the most severe injury from the high rate of clopyralid applied at V7. All PGR herbicides resulted in a significant reduction in final soybean height, except for the lower rate of dicamba applied at R2 and the lower rate of dicamba plus diflufenzopyr at V7 and R2. Treatments that resulted in the death of the terminal growing point (as mentioned previously) stimulated development of lateral branches for subsequent growth, yet resulted in a 16 to 42% reduction in final height (Table 2). Although the higher rate of 2,4-D did not kill the terminal growing point, it resulted in soybean with severe stem epinasty and an 18 to 25% reduction in final height. The greatest height reductions resulted from the higher rates of all four PGR herbicide treatments at V7, with the higher rate of clopyralid reducing height the most. Several PGR herbicide treatments caused significant delays in soybean maturity (Table 2). Except for the R2 ap- ER 258 (283 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 150 of 245 FtGURE I. Foliar leaf abnormalities caused by plant growth regulator (PGR) herbicides in soybean. (A) and (B) created wich dicamba at 5.6 g ha- 1, (C) and (D) created wich clopyralid at 2.1 g ha- 1• (D) Clopyralid injured aifoliolate on the left and untreated rrifoliolate on me right. plication at the lower rates, all dicamba and clopyralid treatments resulted in delayed maturity, whereas only the V3 and V7 applications of the higher rate of dicamba plus diflufenzopyr caused a delay. Maturity was delayed by 2,4-D at both rates applied at R2 and the higher rate applied at V7. Injury from higher rates of dicamba, 2,4-D, and clopyralid applied during flowering development (R2) caused the greatest delay in maturity. Notwithstanding significant injury and reduced height, many PGR herbicide treatments did not result in yield reductions (Table 2). Dicamba plus diflufenzopyr reduced yield by 8% when the higher rate was applied at V3, whereas dicamba reduced yield from 6 to 12% after application of the higher rate at all growth stages and the lower rate at V3. Yield was reduced by 15 to 25% from the higher rates of2,4-D applied at all growth stages, and clopyralid reduced yield by 9 to 48% from the higher rate applied at all stages and the lower rate applied at V3. The higher rate of clopyralid applied at V7 resulted in the lowest yield (Table 2). lmazethapyr applied at V7 also reduced yield by 7%. The growth stages at which soybean were most sensitive to height or yield reductions (or both) varied among the herbicides (Table 2). The highest rate of clopyralid applied at V7 reduced height and yield more than the same rate applied at the other growth stages. T he highest rate of dicamba applied at V7 also reduced height more than the same rate applied at the other growth stages, but dicamba did not have a similar effect on yield. Previous research (Auch and Arnold 1978; Wax et al. 1969) showed that dicamba caused greater injury and yield reduction when apKelley et al.: Off-target growth regulator herbicide injury ER 259 • 105 (284 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 151 of 245 106 • Weed Science 53, January–February 2005 TABLE 2. Soybean yield, rate of maturity, and height in response to application of reduced rates of PGR herbicides applied at the V3, V7 and R2 stages of soybean growth combined across 2001, 2002, and 2003.a,b Soybean yieldc Herbicide Dicamba Dicamba ϩ diflufenzopyr 2,4-D Clopyralid Imazethapyr Glyphosate Fomesafen Untreated control LSD (0.05)f LSD (0.05)g Rate g ae/ha 0.56 5.6 0.2 ϩ 0.08 2.0 ϩ 0.8 56 180 2.1 6.6 71 840 330 V3 V7 Maturity delayd R2 V3 Ϫ1 2,820 bc 2,830 bc 2,970 ab 2,790 bc 2,850 ab 2,270 d 2,740 bc 2,580 c 2,890 ab 3,040 a 2,920 ab 3,020 a kg ha 3,120 a 2,660 de 3,080 ab 3,000 abc 2,890 bc 2,520 e 2,940 abc 1,560 f 2,820 cd 2,900 abc 2,960 abc 3,020 ab 270 220 3,270 a 2,800 de 3,040 bc 3,150 ab 2,970 bcd 2,570 e 3,180 ab 2,670 e 2,980 bcd 3,110 abc 2,930 cd 3,020 bc 4 bc 7a 1d 5 ab 2 cd 1d 6 ab 6 ab 1d 1d 2 cd 0d V7 d 3 abc 4 ab 0d 4 ab 2 bcd 3 abc 5a 2 bcd 2 bcd 1 cd 2 bcd 0d 3 3 a Abbreviation: PGR, plant growth regulator. b Means within a column (treatments applied at the same growth stage) followed by the same letter(s) are not significantly different c Measured by harvesting the center two rows from each plot and adjusting moisture to 13%. d Measured by recording the day when 95% of the pods reached a mature color and comparing with the untreated control. e Measured when plants reached full height before leaf senescence. Final height of untreated plants was 102 cm. f Between growth stages for the same herbicide treatment (only the higher rates of 2,4-D and clopyralid—applied only in 2002 and g Between growth stages for the same herbicide treatment (only those treatments applied in all years). ER 260 Heighte R2 0d 9a 1 cd 1 cd 4b 8a 0d 8a 1 cd 3 cd 0d 0d V3 88 c 77 e 95 b 84 cd 95 b 82 de 86 cd 71 f 98 ab 98 ab 98 ab 100 a V7 % untreated control 87 c 65 f 98 ab 76 e 88 c 75 e 81 d 58 g 96 b 99 ab 98 ab 100 a 6 4 according to Fisher’s Protected LSD (0.05). 2003). R2 97 a 81 c 97 a 88 b 87 b 79 c 92 b 72 d 99 a 99 a 98 a 100 a (285 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 152 of 245 TABLE 3. Soybean injury caused by combinations of dicamba and herbicides labeled for use in soybean applied at the V3 and V7 stages of soybean growth.a Growth stage Herbicide V3 Glyphosate Imazethapyr Imazamox Fomesafen Dicamba Glyphosate ϩ dicamba Imazethapyr ϩ dicamba Imazamox ϩ dicamba Fomesafen ϩ dicamba Untreated control V7 Glyphosate Imazethapyr Imazamox Fomesafen Dicamba Glyphosate ϩ dicamba Imazethapyr ϩ dicamba Imazamox ϩ dicamba Fomesafen ϩ dicamba Untreated control Rate Early-season injuryb Late-season injury 2 WATc 6 WAT 2002 g ae/ha 1,270 71 44 330 5.6 1,270 ϩ 5.6 71 ϩ 5.6 44 ϩ 5.6 330 ϩ 5.6 2003 2002 2003 % 1,270 71 44 330 5.6 1,270 ϩ 5.6 71 ϩ 5.6 44 ϩ 5.6 330 ϩ 5.6 0d 2d 3 cd 5c 42 b 47 b 50 ab*d 53 ab* 60 a* 0d 0e 3d 5d 5d 32 c 33 bc 42 ab* 43 a* 48 a* 0e 0c 0c 1c 0c 27 b 30 b 29 b 32 ab 37 a* 0c 0d 0d 0d 0d 23 c 25 bc 28 ab* 30 a* 30 a* 0d 0c 0c 2c 5b 27 a 30 a 37 a* 33 a* 37 a* 0c 2e 5d 5d 7d 28 c 35 bc* 40 ab* 40 ab* 48 a* 0e 0e 0e 0e 0e 30 d 33 c 40 a* 42 a* 38 b* 0e 0c 0c 0c 0c 37 b 43 b* 50 a* 52 a* 57 a* 0c a Means of treatments applied in the same year at the same growth stage followed by the same letter(s) are not significantly different according to Fisher’s Protected LSD (0.05). b Visual injury ratings on a scale of 0 to 100% with 0% ϭ no injury and 100% ϭ complete death. c Abbreviation: WAT, weeks after treatment. d * Indicates significant synergistic interaction at the 0.05 level. plied near the R2 or V7 stage, relative to the V3 stage. The higher rate of 2,4-D resulted in the lowest yield from the V3 application, significantly lower than the R2 application at P Ͻ 0.05, and the V7 application at P Ͻ 0.1. Smith (1965) also reported lower yields after 2,4-D was applied to soybean at an early vegetative stage compared with at a reproductive stage. Many of the applications of dicamba plus diflufenzopyr resulted in reduced crop injury (Table 1) and greater height and yield (Table 2) compared with an equal fraction of a field use rate of dicamba in corn. The addition of diflufenzopyr to dicamba, which allows a reduction in the amount of dicamba necessary to achieve adequate weed control, may reduce injury caused by off-target exposure to dicamba-containing products. Soybean Herbicide Interaction Study By 2 wk after the V3 and V7 applications, treatments that included dicamba caused a considerable amount of injury (Table 3), including death of the terminal growing point and leaf cupping symptoms (Figures 1A and 1B). When applied alone, fomesafen caused temporary leaf necrosis but had no effect on subsequent growth, imazethapyr and imazamox temporarily stunted plant growth, and glyphosate caused no significant plant injury. Imazethapyr, imazamox, and fomesafen all demonstrated synergistic interactions with dicamba, increasing soybean injury at 2 wk after both application timings in both years, and glyphosate had a similar interaction with dicamba after the V7 application in 2003 (Table 3). By 6 wk after both application timings, dicamba-treated soybean were still showing foliar leaf cupping symptoms and were reduced in height (Table 3). When applied at V3, there were synergistic interactions between the following soybean herbicides and dicamba to increase soybean injury 6 WAT: fomesafen in both years and imazethapyr and imazamox in 2003. When applied at V7, there were synergistic interactions between the following herbicides and dicamba to increase injury 6 WAT: imazethapyr, imazamox, and fomesafen in both years, and glyphosate in 2003. Dicamba-treated plants (alone or with another herbicide) failed to achieve canopy closure and all leaves that emerged after application exhibited cupping injury symptoms, with leaves that were smaller than leaves from plants not treated with dicamba (data not shown). All treatments that included dicamba caused a significant reduction in final soybean height, whereas herbicides labeled for use in soybean did not reduce final height in the absence of dicamba (Table 4). Dicamba applied alone at V3 reduced final soybean height by 21 to 22%, and when applied at V7, dicamba applied alone reduced height by 25 to 28%. When applied at the V7 application both years, there were synergistic interactions between dicamba and imazamox or fomesafen to further reduce final soybean height, whereas similar interactions occurred with dicamba plus imazethapyr in 2002 and dicamba plus glyphosate in 2003. Dicamba treatments had a significant effect on the rate of soybean maturity, but the effect varied between 2002 and 2003 (Table 4). Most treatments containing dicamba reKelley et al.: Off-target growth regulator herbicide injury ER 261 • 107 (286 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 153 of 245 108 • Weed Science 53, January–February 2005 TABLE 4. Soybean yield, rate of maturity, and height in response to combinations of dicamba and herbicides labeled for use in soybean applied at the V3 and V7 stages of soybean growth.a Soybean yieldb Growth stage Herbicide V3 Glyphosate Imazethapyr Imazamox Fomesafen Dicamba Glyphosate ϩ dicamba Imazethapyr ϩ dicamba Imazamox ϩ dicamba Fomesafen ϩ dicamba Untreated V7 Glyphosate Imazethapyr Imazamox Fomesafen Dicamba Glyphosate ϩ dicamba Imazethapyr ϩ dicamba Imazamox ϩ dicamba Fomesafen ϩ dicamba Untreated LSD (0.05)f Rate g ae/ha 1,270 71 44 330 5.6 1,270 ϩ 5.6 71 ϩ 5.6 44 ϩ 5.6 330 ϩ 5.6 1,270 71 44 330 5.6 1,270 ϩ 5.6 71 ϩ 5.6 44 ϩ 5.6 330 ϩ 5.6 2002 3,280 a 3,320 a 2,870 bcd 3,190 ab 2,690 cde 2,730 cd 2,950 abc 2,540 de 2,340 e 3,160 ab Delayed maturityc 2003 kg haϪ1 3,370 abc 3,130 bcd 3,390 ab 3,470 ab 2,720 ef 3,010 cde 2,440 f 2,930 de 2,440 f 3,490 a 3,200 a 3,160 ab 3,010 abc 3,280 a 2,790 c 2,580 d 2,060 e* 1,970 e* 2,070 e* 3,160 ab 390 3,270 a 3,340 a 3,400 a 3,330 a 2,500 b 2,300 bc 2,200 bc 2,110 bc 2,060 c* 3,490 a 410 2002 days delayed 0b 0b 0b 0b 0b 0b 0b 1b 5a 6a 4a 7a 3a 6a 4a 7a 2 ab 7a 0b 0b 1a 1a 0a 1a Ϫ4 b Ϫ4 b Ϫ4 b Ϫ4 b Ϫ4 b 0a 3 a Means of treatments applied in the same year at the same growth stage followed by the same letter(s) are not significantly different according to b Measured by harvesting the center two rows from each plot and adjusting moisture to 13%. c Measured by recording the day when 95% of the pods reached a mature color and comparing with the untreated control. NS, not significant. d Measured when plants reached full height before leaf senescence. Final heights of untreated plants were 81 cm in 2002 and 111 cm in 2003. e * Indicates significant synergistic interaction at the 0.05 level. f Between growth stages for the same herbicide treatment. ER 262 2003 0b 0b 0b 1b 5a 6a 5a 6a 6a 0b NS Final heightd 2002 2003 % untreated control 104 a 98 a 104 a 95 a 96 a 94 a 103 a 94 a 78 b 79 b 80 b 79 b 76 b 73 b 73 b 77 b 71 b 74 b 100 a 100 a 99 a 102 a 102 a 100 a 75 b 73 bc 64 cd* 63 d* 63 d* 100 a 8 Fisher’s Protected LSD (0.05). 98 a 99 a 94 a 96 a 72 b 63 c*e 65 bc 61 c* 60 c* 100 a 8 (287 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 154 of 245 TABLE 5. Rainfall before and after treatments of the soybean herbicide interaction study.a 2002 V3 2003 V7 V3 V7 64 6 106 0 151 0 52 5 mm MBTb 1 1 WAT 2 WAT 3 WAT 59 0 0.3 35 1 35 24 0 a Illinois Climate Network Data, Illinois State Water Survey for Champaign, IL. b Abbreviations: MBT, month before treatment, WAT, week after treatment. sulted in delayed maturity. However, applications at V7 in 2002 resulted in earlier maturity. The late-season injury (Table 3) and final height data (Table 4) indicate that dicamba applied at V7 was potentially more damaging to soybean than at V3. Precipitation received before and after the V7 application in 2002 was less favorable than in 2003, as illustrated by the rainfall data in Table 5. These plants in 2002 had received only 1 mm of rainfall during the entire month before treatment, resulting in drought-stressed plants that were further stressed by dicamba. Soybean received rainfall after herbicide treatment (Table 5), but it is likely that the addition of dicamba injury and drought stress at application caused enough damage to result in premature senescence (Table 4), although this did not appreciably alter the response of soybean height or yield to dicamba injury between the 2 yr. Notably, the addition of another herbicide to dicamba did not affect maturity. Yield results also show a significant impact of the presence of herbicides labeled for use in soybean on dicamba injury. Less favorable rainfall in 2002 resulted in lower yields, with the untreated control yielding 3,160 kg haϪ1 in 2002 compared with 3,490 kg haϪ1 in 2003. Soybean yield after applications containing dicamba ranged from 7 to 38% less than the untreated control in 2002 and 14 to 41% less in 2003 (Table 4). Herbicide treatments applied at V7 that resulted in significantly lower yield (P Ͻ 0.05) than the same treatment applied at V3 included imazamox plus dicamba in both years, imazethapyr plus dicamba in 2002, and glyphosate plus dicamba in 2003. After the V7 application in 2002, there were synergistic interactions between dicamba and imazethapyr, imazamox, or fomesafen to further decrease yield (Table 4), and when applied at V7 in 2003, fomesafen had a similar interaction with dicamba. If the significance level is set at P Ͻ 0.1, then imazamox applied at V7 in 2002 and fomesafen applied at V3 in 2002 also demonstrated synergistic interactions with dicamba to further decrease yield. Fomesafen plus dicamba resulted in the highest soybean injury rating (Table 3) and the greatest height reduction (Table 4) of all V3 applications in 2002. These results demonstrate that dicamba can cause a greater yield loss in the presence of a herbicide labeled for use in soybean than if there is no other herbicide present, and that among the soybean herbicides included in this study, fomesafen exacerbated yield losses caused by dicamba more than other herbicides. To determine which growth process was affected to reduce yield, plant samples were collected before harvest for yield component analysis (Table 6). Seeds per pod were significantly reduced by all applications at V7 that included dicamba in both years. The stress from dicamba may have affected seed development during flowering, which began shortly after the V7 stage. Other treatments that reduced seeds per pod included dicamba applied alone at V3 in 2002, glyphosate plus dicamba at V3 in 2003, and imazethapyr or imazamox applied alone at V7 in 2002. Pods per plant were not significantly affected by dicamba applications (Table 6). Pods per node were reduced in response to dicamba, but nodes per plant were increased (data not shown). Although plants were shorter, they were able to produce sufficient nodes on lateral branches from which pods could develop to offset any reduction in pod set during flowering. The degree of seed and pod development vs. floral abortion is influenced by auxin (Cho et al. 2002). Also, exogenous auxin enhances the growth of different tissues (roots, buds, stems), but only at specific concentrations, with higher concentrations inhibiting growth (Gardner et al. 1985). Therefore, it would be anticipated that PGR herbicides, which overstimulate auxin receptors (Sterling and Hall 1997), would inhibit floral development at a sublethal dose if applied near flowering. Seed weight was significantly reduced by several treatments that included dicamba at both the V3 and V7 stages (Table 6). This appears unusual because seed fill does not begin until late in development, several weeks after the V3 stage. However, decreased seed weight may be due to diminished photosynthetic capacity caused by reduced leaf area, given that dicamba prevented canopy closure and resulted in smaller, malformed leaves (data not shown). All the herbicides labeled for use in soybean that exacerbated yield losses caused by dicamba are not phytotoxic to soybean due to rapid metabolism of the herbicide (Skipsey et al. 1997; Tecle et al. 1993). However, glyphosate, which did not significantly increase dicamba injury, is not phytotoxic due to an insensitive target site in soybean (Padgette et al. 1995). It may therefore be possible that dicamba injury prevented soybean from metabolizing these herbicides at a sufficient rate to prevent phytotoxicity. Because the presence of herbicides labeled for use in soybean may affect the level of soybean injury and yield loss caused by dicamba, there is added significance in identifying the route of exposure to a PGR herbicide in a reported case of injury. However, with some reports of soybean symptoms resembling PGR herbicide injury, there is not a readily determined source of PGR herbicide exposure. It could be possible for other sources of stress, such as herbicides with a different mode of action, aphid feeding, or infection by certain soybean viruses, to cause symptoms that are mistaken for PGR herbicide injury (Proost et al. 2004). This makes it difficult to accurately assess the cause of soybean injury, especially because no diagnostic tools are available to conclusively verify that a PGR herbicide is the cause of injury. Another study performed in conjunction with this one explores the development of a diagnostic assay for PGR herbicide injury in soybean based on the expression of auxinresponsive genes (Kelley et al. 2004). The results of this study reveal differences in the way that soybean responds to PGR herbicides and may influence decisions on their use. Clopyralid caused much greater yield losses at 6.6 g haϪ1 when applied at a late-vegetative stage Kelley et al.: Off-target growth regulator herbicide injury ER 263 • 109 (288 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 155 of 245 110 • Weed Science 53, January–February 2005 TABLE 6. Components of soybean yield in response to combinations of dicamba and herbicides labeled for use in soybean applied at the V3 and V7 stages of soybean growth.a Seeds per pod Growth stage Herbicide Rate 2002 Pods per plant 2003 2002 2003 2.38 ab 2.42 a 2.27 ab 2.29 ab 2.20 b 2.29 ab 2.26 ab 2.29 ab 2.27 ab 2.41 a 2.47 ab 2.52 ab 2.51 ab 2.53 a 2.48 ab 2.35 b 2.51 ab 2.43 ab 2.44 ab 2.53 a 28 a 26 a 28 a 26 a 23 a 21 a 22 a 25 a 25 a 25 a 28 a 31 a 37 a 30 a 30 a 32 a 25 a 28 a 30 a 35 a g per 100 seeds 17.03 a 13.84 ab 16.27 ab 13.66 ab 16.79 a 13.75 ab 16.64 a 13.28 ab 14.54 cd 12.69 b 13.96 d 13.08 ab 14.88 bcd 12.28 b 15.75 abc 13.05 ab 14.29 cd 14.15 a 16.52 a 14.34 a 2.31 ab 2.23 bc 1.97 d 2.43 a 2.15 bcd 2.10 cd 2.05 cd 2.00 d 2.00 d 2.41 a 0.20 2.48 a 2.43 a 2.42 a 2.52 a 1.90 b 2.00 b 1.86 b 1.86 b 1.87 b 2.53 a 0.19 31 a 29 a 28 a 25 a 26 a 26 a 26 a 26 a 26 a 25 a NSc 34 a 30 a 31 a 28 a 33 a 30 a 29 a 28 a 26 a 35 a NS 16.47 abc 16.54 ab 17.12 a 16.17 abc 14.76 c 15.54 bc 15.23 c 15.42 bc 15.37 bc 16.52 ab 1.47 Ϫ1 V3 Glyphosate Imazethapyr Imazamox Fomesafen Dicamba Glyphosate ϩ dicamba Imazethapyr ϩ dicamba Imazamox ϩ dicamba Fomesafen ϩ dicamba Untreated control V7 Glyphosate Imazethapyr Imazamox Fomesafen Dicamba Glyphosate ϩ dicamba Imazethapyr ϩ dicamba Imazamox ϩ dicamba Fomesafen ϩ dicamba Untreated control LSD (0.05)b g ae ha 1,270 71 44 330 5.6 1,270 ϩ 5.6 71 ϩ 5.6 44 ϩ 5.6 330 ϩ 5.6 1,270 71 44 330 5.6 1,270 ϩ 5.6 71 ϩ 5.6 44 ϩ 5.6 330 ϩ 5.6 Seed weight 2002 2003 13.56 ab 13.88 ab 14.44 a 14.12 ab 12.66 b 12.83 b 13.41 ab 13.41 ab 13.18 ab 14.34 a NS a Measured from a sample of 10 plants taken from the center of each plot before harvest. Means of treatments applied in the same year at the same growth stage followed by the same letter(s) are not significantly different according to Fisher’s Protected LSD (0.05). Treatments that include more than one herbicide were tested for synergistic interactions using the method described by Colby (1967), but no treatments were significant at the 0.05 level. b Between growth stages for the same herbicide treatment. c Abbreviation: NS, nonsignificant. ER 264 (289 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 156 of 245 approaching flowering than at an early-vegetative stage or during flowering, whereas dicamba and 2,4-D showed less of a difference among growth stages. Dicamba caused yield losses at the lowest rate, with 2,4-D requiring the highest rate to reduce yield and clopyralid causing yield losses at a rate in between dicamba and 2,4-D. The addition of diflufenzopyr to dicamba, which allows for less dicamba to be applied to maintain adequate weed control, resulted in less of a yield effect than dicamba applied alone at an equal fraction of a field use rate in corn. This indicates that the use of diflufenzopyr may reduce the risk for unintended soybean injury due to dicamba. Results show that soybean responds differently to the various PGR herbicides examined in our study, and an understanding of these differences will allow growers to select a PGR herbicide based on an assessment of their weed management needs and the potential for soybean injury due to off-target movement. Previous research on the effects of PGR herbicides in soybean has not addressed the impact of the presence of herbicides labeled for use in soybean. However, our results clearly show that the presence of a POST soybean herbicide can significantly exacerbate yield losses caused by off-target dicamba exposure. Dicamba can interact with a soybean herbicide when dicamba herbicide residues are present in application equipment used for soybean. The rate used in this study would not likely be present in application equipment that was cleaned properly, which emphasizes the need to clean application equipment thoroughly after use of a PGR herbicide. Dicamba may also interact in the plant with soybean herbicides when dicamba drifts onto soybean from a neighboring corn field at or near the time of a POST application to soybean, although this type of interaction was not evaluated in this study and may have different consequences than those reported here. Results showed a difference between herbicides that are selective in soybean due to metabolism (imidazolinones and fomesafen) vs. an insensitive target site (glyphosate), which indicates that a reduction in the ability of soybean to metabolize either herbicide may play a role in the interaction between the soybean herbicide and dicamba. Regardless of the mechanism, it is clear that under certain circumstances, the presence of some soybean herbicides can aggravate injury and yield losses caused by dicamba. It would also be of interest to determine if dicamba injury to soybean is affected by other POST herbicides in soybean where selectivity is due to engineered metabolism (e.g., glufosinate-resistant soybean). Sources of Materials 1 TeeJet standard flat spray tips, Spraying Systems Co., P.O. Box 7900, Wheaton, IL 60189-7900. 2 Activator-90, nonionic surfactant, a mixture of alkylphenyl hydroxypolyoxyethylene and fatty acids, Loveland Industries Inc., P.O. Box 1289, Greeley, CO 80632-1289. 3 MSO, methylated seed oil and emulsifying surfactants 100%, Loveland Industries Inc., P.O. Box 1289, Greeley, CO 806321289. Acknowledgments This research was supported by the Illinois Soybean Program Operating Board and Dow AgroSciences. We thank Doug Maxwell and Ryan Hasty for the planting, maintaining, and harvesting of field research plots and Joshua Strom and Anna Ferguson for assisting in the collection of yield component data. Literature Cited Al-Khatib, K. and D. Peterson. 1999. Soybean (Glycine max) response to simulated drift from selected sulfonylurea herbicides, dicamba, glyphosate, and glufosinate. Weed Technol. 13:264–270. Andersen, S. M., S. A. Clay, L. J. Wrage, and D. Matthees. 2004. Soybean foliage residues of dicamba and 2,4-D and correlation to application rates and yield. Agron. J. 96:750–760. Auch, D. E. and W. E. Arnold. 1978. Dicamba use and injury on soybeans (Glycine max) in South Dakota. Weed Sci. 26:471–475. Behrens, R. and W. E. Lueschen. 1979. Dicamba volatility. Weed Sci. 27: 486–493. Bode, L. E. 1987. Spray application technology. Pages 85–110 in C. G. McWhorter and M. R. Gebhardt, eds. Methods of Applying Herbicides. Weed Science Society of America Monograph 4. Champaign, IL: Weed Science Society of America. Boerboom, C. 2004. Field case studies of dicamba movement to soybeans. www.soils.wisc.edu/extension/FAPM/fertaglime04.htm. Bovey, R. W. and R. E. Meyer. 1981. Effects of 2,4,5-T, triclopyr, and 3,6dicloropicolinic acid on crop seedlings. Weed Sci. 29:256–261. Carmer, S. G., W. E. Nyquist, and W. M. Walker. 1989. Least significant differences for combined analyses of experiments with two- or threefactor treatment designs. Agron. J. 81:665–672. Cho, Y., S. K. Suh, H. K. Park, and A. Wood. 2002. Impact of 2,4-DP and BAP upon pod set and seed yield in soybean treated at reproductive stages. Plant Growth Regul. 36:215–221. Colby, S. R. 1967. Calculating synergistic and antagonistic responses of herbicide combinations. Weeds 15:20–22. Derksen, D. A. 1989. Dicamba, chlorsulfuron, and clopyralid as sprayer contaminants on sunflower (Helianthus annuus), mustard (Brassica juncea), and lentil (Lens culinaris), respectively. Weed Sci. 37:616–621. Gardner, F. P., R. B. Pearce, and R. L. Mitchell. 1985. Plant growth regulation. Pages 156–186 in F. P. Gardner, R. B. Pearce, and R. L. Mitchell, eds. Physiology of Crop Plants. 1st ed. Ames, IA: The Iowa State University Press. Grossman, K., G. Caspar, J. Kwiatkowski, and S. J. Bowe. 2002. On the mechanism of selectivity of the corn herbicide BAS 662H: a combination of the novel auxin transport inhibitor diflufenzopyr and the auxin herbicide dicamba. Pest Manage. Sci. 58:1002–1014. Hager, A. and D. Nordby. 2004. Soybean leaf cupping. Pest Manage. Crop Develop Bull. 15:133–135. [www.ipm.uiuc.edu/bulletin/index.php] Kelley, K. B., K. N. Lambert, A. G. Hager, and D. E. Riechers. 2004. Quantitative expression analysis of GH3, a gene induced by plant growth regulator herbicides in soybean. J. Agric. Food Chem. 52:474– 478. [NASS] National Agricultural Statistics Service. 2002. Agricultural Chemical Use Database. www.pestmanagement.info/nass. Padgette, S. R., K. H. Kolacz, X. Delannay, et al. 1995. Development, identification, and characterization of a glyphosate-tolerant soybean line. Crop Sci. 35:1451–1461. Proost, R., C. Boerboom, and R. Schmidt. 2004. Dicamba Injury to Soybean. Integrated Pest and Crop Management, University of Wisconsin-Madison. http://ipcm.wisc.edu/pubs/pest/Dicamba2004.htm. Que Hee, S. S. and R. G. Sutherland. 1974. Volatilization of various esters and salts of 2,4-D. Weed Sci. 22:313–318. [SAS] Statistical Analysis Systems. 1999. SAS/STAT User’s Guide. Version 7-1. Cary, NC: Statistical Analysis Systems Institute. Skipsey, M., C. J. Andrews, J. K. Townson, I. Jepson, and R. Edwards. 1997. Substrate and thiol specificity of a stress-inducible glutathione transferase from soybean. FEBS Lett. 409:370–374. Slife, F. W. 1956. The effect of 2,4-D and several other herbicides on weeds and soybeans when applied as post-emergence sprays. Weeds 4:61–68. Smith, L. L. and J. Geronimo. 1977. Response of seven crops to six hormone-like herbicides. Weed Sci. Abstr. p. 32 Smith, R. J. 1965. Effects of chlorophenoxy herbicides on soybeans. Weeds 13:168–169. Sterling, T. M. and J. C. Hall. 1997. Mechanism of action of natural auxins and the auxinic herbicides. Pages 111–141 in R. M. Roe, J. D. Burton, and R. J. Kuhr, eds. Herbicide Activity: Toxicology, Biochemistry, and Molecular Biology. Burke, VA: IOS. Tecle, B., A. Dacunha, and D. L. Shaner. 1993. Differential routes of Kelley et al.: Off-target growth regulator herbicide injury ER 265 • 111 (290 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 157 of 245 metabolism of imidazolinones: basis for soybean (Glycine max) selectivity. Pestic. Biochem. Physiol. 46:120–130. Wax, L. M., L. A. Knuth, and F. W. Slife. 1969. Response of soybean to 2,4-D, dicamba, and picloram. Weed Sci. 17:388–393. 112 • Weed Science 53, January–February 2005 Weidenhamer, J. D., G. B. Triplett, and F. E. Sobotka. 1989. Dicamba injury to soybean. Agron. J. 81:637–643. Received April 5, 2004, and approved August 20, 2004. ER 266 (291 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 158 of 245 Weed Technology 2014 28:454 464 Influence of Application Timings and Sublethal Rates of Synthetic Auxin Herbicides on Soybean Craig B. Solomon and Kevin W. Bradley* Synthetic auxin herbicides have long been utilized for the selective control of broadleaf weeds in a variety of crop and noncrop environments. Recently, two agrochemical companies have begun to develop soybean with resistance to 2,4-D and dicamba which might lead to an increase in the application of these herbicides in soybean production areas in the near future. Additionally, little research has been published pertaining to the effects of a newly-discovered synthetic auxin herbicide, aminocyclopyrachlor, on soybean phytotoxicity. Two field trials were conducted in 2011 and 2012 to evaluate the effects of sublethal rates of 2,4-D amine, aminocyclopyrachlor, aminopyralid, clopyralid, dicamba, fluroxypyr, picloram, and triclopyr on visible estimates of soybean injury, height reduction, maturity, yield, and yield components. Each of these herbicides was applied to soybean at the V3 and R2 stages of growth at 0.028, 0.28, 2.8, and 28 g ae ha 1. Greater height reductions occurred with all herbicides, except 2,4-D amine and triclopyr when applied at the V3 compared to the R2 stage of growth. Greater soybean yield loss occurred with all herbicides except 2,4-D amine when applied at the R2 compared to the V3 stage of growth. The only herbicide applied that resulted in no yield loss at either stage was 2,4-D amine. When applied at 28 g ae ha 1 at the V3 stage of growth, the general order of herbicide-induced yield reductions to soybean from greatest to least was aminopyralid . aminocyclopyrachlor clopyralid picloram . fluroxypyr . triclopyr . dicamba . 2,4-D amine. At the R2 stage of growth, the general order of herbicide-induced yield reductions from greatest to least was aminopyralid . aminocyclopyrachlor picloram . clopyralid . dicamba . fluroxypyr triclopyr . 2,4-D amine. Yield reductions appeared to be more correlated with seeds per pod than to pods per plant and seed weight. An 18- to 26-d delay in soybean maturity also occurred with R2 applications of all synthetic auxin herbicides at 28 g ae ha 1 except 2,4-D. Results from this research indicate that there are vast differences in the relative phytotoxicity of these synthetic auxin herbicides to soybean, and that the timing of the synthetic auxin herbicide exposure will have a significant impact on the severity of soybean height and/or yield reductions. Nomenclature: Aminocyclopyrachlor; aminopyralid; clopyralid; dicamba; fluroxypyr; picloram; triclopyr; 2,4-D; soybean, Glycine max (L.) Merr. Key words: Growth regulator herbicides, herbicide-resistant crops, off-target spray, spray drift, tank contamination. Los herbicidas auxinas sint´eticas han sido utilizados por un largo tiempo para el control selectivo de malezas de hoja ancha en una variedad de situaciones con y sin cultivos. Recientemente, dos compan˜ ´ıas de agroqu´ımicos iniciaron el desarrollo de soya con resistencia a 2,4 D y dicamba, lo que podr´ıa llevar a un incremento en la aplicacio´ n de estos herbicidas en zonas productoras de soya en un futuro cercano. Adicionalmente, pocas investigaciones han sido publicadas en relacio´ n a los efectos de aminocyclopyrachlor, un herbicida auxina sint´etica recientemente descubierto, sobre la fitotoxicidad en soya. Se realizaron dos experimentos de campo en 2011 y 2012 para evaluar los efectos de dosis subletales de 2,4 D amine, aminocyclopyrachlor, aminopyralid, clopyralid, dicamba, fluroxypyr, picloram, y triclopyr sobre los estimados visuales de dan˜ o en soya, la reduccio´ n en la altura, la madurez, el rendimiento, y los componentes de rendimiento. Cada uno de estos herbicidas fue aplicado a soya en los estadios de desarrollo V3 y R2 a 0.028, 0.28, 2.8, y 28 g ae haÀ1. Las mayores reducciones en altura ocurrieron con todos los herbicidas, excepto 2,4 D amine y triclopyr cuando se aplico´ en el estadio de desarrollo V3 en comparacio´ n con R2. Las mayores pe´rdidas en el rendimiento de la soya ocurrieron con todos los herbicidas excepto 2,4 D amine cuando se aplic´o en el estadio R2 en comparaci´on con V3. El u´ nico herbicida aplicado que no resulto´ en pe´rdidas de rendimiento en ninguno de los estadios de desarrollo fue 2,4 D amine. Cuando se aplico´ a 28 g ae haÀ1 en el estadio V3, el orden general de mayor a menor, de reducciones en el rendimiento de la soya inducidas por el herbicida fue: aminopyralid . aminocyclopyrachlor ¼ clopyralid ¼ picloram . fluroxypyr . triclopyr . dicamba . 2,4 DOI: 10.1614/WT D 13 00145.1 * Graduate Research Assistant and Associate Professor, Division of Plant Sciences, University of Missouri, Columbia, MO 65201. Corresponding author’s E mail: BradleyKe@missouri.edu 454 Weed Technology 28, July–September 2014 ER 267 (292 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 159 of 245 D amine. En el estadio de desarrollo R2, el orden general, de mayor a menor, de reducciones en el rendimiento de la soya inducidas por el herbicida fue: aminopyralid . aminocyclopyrachlor ¼ picloram . clopyralid . dicamba . fluroxypyr ¼ triclopyr . 2,4 D amine. Las reducciones en el rendimiento parecieron estar ma´s correlacionadas con el nu´ mero de semillas por vaina que el nu´ mero de vainas por planta o el peso de la semilla. Un retraso de 18 a 26 d en la madurez de la soya tambi´en ocurri´o con aplicaciones en R2 de todos los herbicidas auxinas sint´eticas a 28 g ae haÀ1 excepto 2,4 D. Los resultados de esta investigacio´ n indican que existen amplias diferencias en la fitotoxicidad relativa de esos herbicidas auxinas sinte´ticas en soya, y que el momento de exposicio´ n a estos herbicidas tendra´ un impacto significativo en la severidad de las reducciones en altura y/o rendimiento de la soya. As of 2012, 93% of soybean hectares planted in the United States were genetically engineered, herbicide-resistant varieties (USDA 2012). Due to the increase in the occurrence of glyphosate-, protoporphyrinogen oxidase- (PPO) and acetolactate synthase/acetohydroxyacid synthase- (ALS/ AHAS) resistant weed populations, several new herbicide-resistant crop offerings are expected to be introduced onto the marketplace in the near future. Among these are soybean that have been genetically modified to withstand applications of either 2,4-D (Wright et al. 2010) or dicamba (Behrens et al. 2007). Although 2,4-D was first introduced in 1945 (Troyer 2001) and dicamba in 1967 (CCME 1999), weeds with resistance to these herbicides have been relatively slow to evolve. To date, only 30 weed species in the world have been characterized with resistance to at least one of the members of the synthetic auxin herbicide family (Heap 2013). Specifically, there have been 18 species characterized with resistance to 2,4-D, and six with resistance to dicamba (Heap 2013). In these instances, resistance to synthetic auxin herbicides was associated with continuous applications of a single active ingredient over many years (Cranston et al. 2001; Heap and Morrison 1992; Holt and LeBaron 1990). Common symptoms of off-target movement of synthetic auxin herbicides include leaf cupping, stem and leaf epinasty, and cracked and swollen stems, as well as chlorosis and necrosis (Al-Khatib and Peterson 1999; Andersen et al. 2004; Auch and Arnold 1978; Kelley et al. 2005; Sciumbato et al. 2004; Wax et al. 1969). Kelley et al. (2005) described that dicamba applications to soybean resulted in new trifoliate leaves being cupped and crinkled, with higher rates resulting in smaller leaves and reduced overall growth compared to lower rates. Symptoms associated with 2,4-D include leaf and stem epinasty, leaf elongation (often known as ‘‘strapping’’), as well as swollen and cracked stems (Kelley et al. 2005; Wax et al. 1969). Clopyralid injury has been described as similar to dicamba, but with more thin, elongated leaves with parallel venation and less leaf cupping (Kelley et al. 2005). Due to the diversity of cropping systems in the United States, it is not uncommon for crops that are tolerant of synthetic auxin herbicides to be grown in close proximity to crops that are more susceptible to these herbicides, and often in rotation with one another (Wax et al. 1969). Thus, off-target movement can become a major concern due to the widespread use of 2,4-D, dicamba, picloram, triclopyr, and clopyralid in controlling emerged broadleaf weeds in corn (Zea mays L.), sorghum (Sorghum bicolor L. Moench), small grains, fallow land, turfgrasses, pastures, and rangelands. Injury to susceptible plants from off-target movement of synthetic auxins has been well documented in many crops, including cotton (Gossypium hirsutum L.) (Everitt and Keeling 2009; Johnson et al. 2012; Marple et al. 2007), alfalfa (Medicago sativa L.) (AlKhatib et al. 1992), common sunflower (Helianthus annuus L.) (Derksen 1989; Lanini 2000), peanut (Arachis hypogaea L.) (Johnson et al. 2012), wine grape (Vitis vinifera L.) (Al-Khatib et al. 1993), and many other crops (Derksen 1989; Hemphill and Montgomery 1981; Lanini 2000). As a result, certain states have laws that dictate which synthetic auxin herbicides may be applied, the chemical formulation, and at what time of year the herbicide may be applied (ASPB 2012; Texas Agriculture Code 1984). Soybean are especially at risk of injury from offtarget movement of synthetic auxin herbicides due to their similar geographic vicinity and rotation with monocot crops (Wax et al. 1969). Al-Khatib and Peterson (1999) evaluated the response of soybean to reduced rates of dicamba and other herbicides when applied at the V2 to V3 stage of growth. In their research, they found that 187 g ae ha 1 of dicamba (33% of the labeled use rate in corn) resulted in yield reductions of 92 and 80%, Solomon and Bradley: Sublethal rates of synthetic auxin herbicides on soybean ER 268 455 (293 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 160 of 245 respectively. In the same study, 56 g ae ha 1 of dicamba (10% of the labeled use rate in corn) resulted in yields 45% lower than the control (AlKhatib and Peterson 1999). Andersen et al. (2004) found that when 5.6 g ae ha 1 of dicamba (1% of the labeled use rate in corn) was applied to soybean at the V3 stage of growth, yield reductions of 14 to 34% occurred. The same study reported that it took applications of 112 g ae ha 1 of 2,4-D (20% of the labeled use rate in corn) to provide similar yield reductions (Andersen et al. 2004). In a similar study, Kelley et al. (2005) observed that applications of 5.6 g ae ha 1 dicamba to V3 soybean resulted in yield reductions of 6%, whereas applications of 2,4-D at 180 g ae ha 1 resulted in a 25% yield reduction. Dicamba applications of 0.56 and 5.6 g ae ha 1 to soybean in the R2 stage of growth resulted in yield reductions of 0 and 7%, respectively, and 2 and 15% for 56 and 180 g ae ha 1 of 2,4-D, respectively (Kelley et al. 2005). In the same study, clopyralid was applied at 2.1 and 6.6 g ae ha 1 to both V3 and R2 soybean, respectively, resulting in yield reductions of 9 and 15%, respectively, for the V3 applications, and 0 and 12%, respectively, for the R2 applications (Kelley et al. 2005). With the exception of 5.6 g ae ha 1 dicamba, all treatments resulted in lower yields when applied at the V3 compared to the R2 stage of growth (Kelley et al. 2005). This is in contrast to previous research, which reported greater injury and yield reductions when dicamba was applied at later soybean growth stages (Auch and Arnold 1978; Slife 1956; Wax et al. 1969). Wax et al. (1969) determined that approximately 16.7 g ae ha 1 of dicamba applied to soybean at the prebloom and bloom growth stages resulted in yield reductions of 11 and 49%, respectively, with 2,4-D applications at these stages resulting in no yield losses. In the same study, 8.75 g ae ha 1 of picloram resulted in soybean yield reductions of 18 and 98% when applied at the prebloom and bloom stages, respectively (Wax et al. 1969). Delayed maturity of soybean following exposure to synthetic auxin herbicides has also been documented in a number of previous experiments (Auch and Arnold 1978; Kelley et al. 2005; Wax et al. 1969). Wax et al. (1969) observed greater maturity delay when dicamba and picloram were applied during the reproductive stages compared to earlier vegetative stages. When picloram was applied at 456 8.75 g ae ha 1 to soybean in the prebloom and bloom growth stages, soybean maturity was delayed 2 and 27 d, respectively (Wax et al. 1969). Dicamba applied at 16.7 g ae ha 1 to soybean in the prebloom and bloom growth stages resulted in delays in maturity of 4 and 14 d, respectively (Wax et al. 1969). Auch and Arnold (1978) also observed a delay in soybean maturity from foliar applications of dicamba throughout the reproductive growth stages. When comparing early-bloom, midbloom, early-pod, and late-pod dicamba applications, most rates and applications resulted in additional delays in maturity as soybean further developed (Auch and Arnold 1978). A variety of research has been conducted to determine the effects of synthetic auxin herbicides on soybean phytotoxicity and yield loss. However, few of these studies have provided results pertaining to aminocyclopyrachlor and aminopyralid, which are two of the newest synthetic auxin herbicides introduced onto the marketplace. Some authors have evaluated the response of soybean to different rates of synthetic auxin herbicides and the rates selected were based on fractions of the recommended use rate of these herbicides in other cropping systems (Andersen et al. 2004; Sciumbato et al. 2004; Weidenhamer et al. 1989), whereas other authors (Everitt and Keeling 2009; Marple et al. 2007; Thompson et al. 2007) have conducted this research with equivalent rates of the synthetic auxin herbicides to determine the relative response of all synthetic auxin herbicides to each other. The objective of this research was to determine the relative effects of sublethal rates of 2,4-D amine, aminocyclopyrachlor, aminopyralid, clopyralid, dicamba, fluroxypyr, picloram, and triclopyr on visible soybean injury, height reduction, yield, and yield components when applied to plants in the V3 and R2 stages of growth. Materials and Methods General Trial Information. Duplicate field trials were conducted during 2011 and 2012 in Boone County, Missouri at the University of Missouri Bradford Research Center (38.90898N, 92.208W). The soil was a Mexico silt loam (fine, smectic, mesic Aeric Vertic Epiaqualfs) with 2.3% organic matter and pH of 6.0 in 2011 and a pH of 6.3 and organic matter content of 2.4% in 2012. On June 6, 2011 Weed Technology 28, July–September 2014 ER 269 (294 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 161 of 245 Table 1. Sources of materials used in the experiment. Common namea Trade name 2,4 D amine Dicamba Clopyralid Picloram Triclopyr Aminopyralid Aminocyclopyrachlor Fluroxypyr a b Weedar 64 Clarity Transline Tordon 22K Remedy Ultra Milestone MAT28 Starane Formulationb 456 g L 480 g L 360 g L 240 g L 480 g L 240 g L 0.50 g g 180 g L Manufacturer 1 EC EC 1 EC 1 EC 1 EC 1 EC 1 SG 1 EC Nufarm, Inc., Burr Ridge, IL (www.nufarm.com/US) BASF Crop Research Triangle Park, NC (www.agro.basf.com) Dow Agrosciences, Indianapolis, IN (www.dowagro.com) Dow Agrosciences Dow Agrosciences Dow Agrosciences DuPont Corporation, Wilmington, DE (www.dupont.com) Dow Agrosciences 1 InterLockt at 0.208% v/v was added to each herbicide solution. Abbreviations: EC, emulsifiable concentrate; SG, soluble granule. and May 22, 2012, Asgrow 3803 glyphosateresistant soybean were planted into a conventionally-tilled seedbed in rows spaced 76 cm apart at a rate of 432,000 seeds ha 1. All treatments were arranged in a randomized complete block (RCB) design with six replications. Individual plots were 2 by 8 m in size. In both years, the entire trial was maintained weed-free with a PRE application of sulfentrazone plus cloransulam plus pendimethalin (139 þ 18 þ 780 g ae ha 1) followed by POST applications of glyphosate (1,121 g ae ha 1). Treatments included the eight synthetic auxin herbicides listed in Table 1. Each of these herbicides was applied at the V3 and R2 stages of soybean growth at 0.028, 0.28, 2.8, and 28 g ae or ai ha 1. In 2011, V3 and R2 applications were made on July 1 and August 3, respectively, whereas in 2012, V3 and R2 applications were made on June 18 and July Table 2. Monthly rainfall (mm) and average monthly temperatures (C) from April through October in 2011 and 2012 in comparison to the 30 yr average in Boone County, Missouri. Rainfall Month 2011 2012 Temperature 30 yr averagea 2011 mm April May June July August September October Total 72 130 77 59 61 46 26 471 171 25 39 18 5 46 68 372 2012 30 yr averagea C 121 127 94 101 75 78 99 695 13.6 16.5 24.0 27.6 24.6 17.4 13.8 13.9 21.0 24.1 28.5 24.7 18.6 11.7 13.6 18.9 23.8 25.7 24.8 20.4 14.0 a 30 yr averages (1981 2010) obtained from National Climatic Data Center (2011). 13, respectively. All treatments were applied with a CO2-pressurized backpack sprayer equipped with 80025 air induction nozzles that delivered coarse to extremely coarse droplets at 140 L ha 1 and 117 kPa. In an effort to minimize spray drift and/or contamination between plots: (1) drift shields were established on three sides of the spray boom during treatment; (2) all treatments included a drift reduction agent (InterLockt, 0.2% v/v; Winfield Solutions LLC, P.O. Box 64589, St. Paul, MN 55164); and (3) each herbicide was applied using a specific boom that had never been used before and was designated for that active ingredient only. Monthly rainfall totals and average monthly temperatures for each year are presented in Table 2. Treatment Evaluation and Data Collection. Visible herbicide injury and soybean height were evaluated at 2 and 4 wk after treatment (WAT). Visible estimates of injury were evaluated on a scale from 0 to 100%, where 0 equals no injury and 100 was equivalent to complete crop death. Soybean height was evaluated by measuring six random soybean plants per plot (three from each row) from the soil surface to the top of the central stem. Delayed maturity was measured by recording the day on which 95% of the soybean pods in each plot reached a mature color and then comparing that with the day when the nontreated control plots reached maturity. Before harvest, a sample of six random soybean plants from the center of each plot were collected and used for yield component analysis. Each sample was evaluated by counting the number of seeds per pod and pods per plant to determine an average value for each respective treatment. Soybean were harvested from the center two rows of each plot with a small plot combine, and seed yields were adjusted to 13% moisture Solomon and Bradley: Sublethal rates of synthetic auxin herbicides on soybean ER 270 457 (295 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 162 of 245 content. A 100-count seed subsample was collected from each plot to determine seed weight. Statistical Analysis. All data were checked for normality to meet basic assumptions prior to statistical analysis. Visible estimates of injury, soybean height, yield component analyses, and soybean yield were subjected to ANOVA using the PROC MIXED procedure in SAS (SAS 9.2, SASt Institute Inc.) and tested for appropriate interactions. Year–location combinations were considered an environment sampled at random, as suggested by Carmer et al. (1989) and Blouin et al. (2011). Herbicide, herbicide rate, and application timing were considered fixed effects in the model, whereas environment, replications, subsamples, and interactions within environment were considered random effects. Analyses were performed on the means and least squares means and detected using Fisher’s protected LSD at a ¼ 0.05. Results and Discussion Visible Estimates of Injury. At 2 WAT, injury symptoms were dependent on herbicide and rate, regardless of growth stage (Table 3). In general, injury intensity increased with increasing herbicide rates. No significant injury was noted following any application of 2,4-D amine. Soybean injury was greatest in response to aminopyralid, aminocyclopyrachlor, picloram, clopyralid, and dicamba, and least with triclopyr and 2,4-D amine (Table 3). By 2 WAT, 28 g ha 1 aminocyclopyrachlor and picloram applied at the V3 stage of growth resulted in terminal clusters of undeveloped buds, moderate epinasty, and chlorosis, with noticeable cupping of leaves. Applications of aminopyralid and clopyralid at the same rate resulted in more necrotic buds and bleached tissues, but less cupping than many of the other synthetic auxin herbicides. Although there were varying degrees of symptomology observed, by 2 WAT of the V3 application timing, 28 g ha 1 aminopyralid, aminocyclopyrachlor, picloram, clopyralid, and fluroxypyr resulted in 56 to 73% visible soybean injury, which was the highest observed in these trials (Table 3). Dicamba and triclopyr at 28 g ha 1 resulted in intermediate levels of soybean injury at 44 and 29%, respectively, with soybean exhibiting fewer necrotic buds and overall leaf cupping in response to these herbicides. Although leaf cupping is more characteristic of dicamba 458 exposure to soybean, at 28 g ha 1 leaves that developed following herbicide treatment did not expand further than bud clusters; thus, visible leaf cupping was minimal. Similar symptoms have been described previously (Al-Khatib and Peterson 1999; Andersen et al. 2004; Auch and Arnold 1978; Kelley et al. 2005; Wax et al. 1969; Weidenhamer et al. 1989). When applied at the V3 stage of growth, 28 g ha 1 2,4-D amine resulted in only 3% soybean injury, which was the lowest level of injury observed in these experiments. There were no leaf or stem epinastic symptoms observed following treatment with triclopyr or 2,4-D amine at any rate. Applications of aminopyralid, picloram, clopyralid, aminocyclopyrachlor, and dicamba at 2.8 and 0.28 g ha 1 to soybean in the V3 stage of growth caused noticeable leaf cupping and leaf mottling/ puckering, as well as chlorotic, undeveloped bud clusters 2 WAT. Due to fewer necrotic buds and stems, visible injury values were overall lower compared to the 28 g ha 1 rate of these same herbicides. In response to V3 applications of 0.028 g ha 1 aminopyralid and dicamba, soybean exhibited a moderate degree of leaf cupping and chlorosis of leaf edges, with dicamba displaying more cupped bud clusters than the other synthetic auxin herbicides. No significant soybean injury was noted 2 WAT of the V3 applications of 0.028 g ha 1 aminocyclopyrachlor and 0.028, 0.28, and 2.8 g ha 1 2,4-D, triclopyr, and fluroxypyr (Table 3). Aminopyralid, clopyralid, picloram, and aminocyclopyrachlor applied at 28 g ha 1 to R2 soybean resulted in the greatest injury (30 to 39%) 2 WAT (Table 3). These treatments resulted in terminal bud death, loss of apical dominance/expansion, and severe stem chlorosis and epinasty. Soybean stems had splits, callouses, and angles of 45 to 120 degrees. These symptoms predominantly occurred on newer plant tissues, and therefore visible injury ratings were overall much lower than V3 applications. Equivalent applications of dicamba and triclopyr to R2 soybean resulted in similar bud necrosis/death, but less epinasty and chlorosis. Overall injury was 15 and 18% in response to 28 g ha 1 triclopyr and dicamba, respectively (Table 3). R2 applications of 0.028, 0.28, and 2.8 g ha 1 dicamba all resulted in similar levels of leaf cupping/ mottling. At the same timing, 0.028, 0.28, and 2.8 g ae ha 1 of aminopyralid and clopyralid resulted in terminal leaf cupping/chlorosis and bud abortions, Weed Technology 28, July–September 2014 ER 271 (296 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 163 of 245 Table 3. Soybean injury, rate of maturity, and height in response to eight synthetic auxin herbicides applied at the V3 and R2 stages of soybean growth combined across 2011 and 2012. Injurya 2 WAT Herbicide Rate g ae ha 2,4 D amine Aminocyclopyrachlor Aminopyralid Clopyralid Dicamba Fluroxypyr Picloram Triclopyr Nontreated LSD (0.05)d 0.028 0.28 2.8 28 0.028 0.28 2.8 28 0.028 0.28 2.8 28 0.028 0.28 2.8 28 0.028 0.28 2.8 28 0.028 0.28 2.8 28 0.028 0.28 2.8 28 0.028 0.28 2.8 28 V3 4 WAT R2 1 Soybean height V3 2 WAT R2 V3 %cd 2 1 1 3 5 11 32* 70* 31* 41* 48* 73* 7 11 41* 60* 21 28 32* 44* 1 1 4 56* 10 11 30* 69* 1 3 2 29 1 18 0 0 0 0 3 9 13 33 12 11 14 39 10 12 14 30 15 17 14 18 0 1 1 15 5 7 10 32 0 1 0 15 0 9 R2 4 WAT V3 R2 % of nontreated controlcd 1 1 0 0 2 4 11 63* 7 14 43* 65* 1 2 7 68* 10 9 9 12 0 0 1 36* 2 2 5 66* 0 1 0 7 0 5 0 1 0 0 3 8 14 29 9 11 13 34 7* 8* 14* 21 17* 16* 15* 14 1 2 2 8 4 6 12* 25 0 1 1 10 0 3 96 102 99 94 103 95 78 52 87 84 74 44 93 92 83 52 89 85 79 80* 102 101 93 58 98 98 85 52 97 98 98* 71 100 6 102 100 101 95 100 97 85* 68* 91 91* 80* 59* 102* 96 86 56 94 93* 86* 74 102 99 97 74* 98 96 85 64* 99 98 92 76 100 4 103 101 101 99 104 99 83* 47 92* 88 66 26 97 95 83 35 94 90 75 74* 101 101 96 59 99 99 90 46 100 98 99 78* 100 6 Maturity delayb V3 R2 No. dayscd 103 100 101 98 101 99 76 59* 86 84 71 53* 101 93 80 57* 89 85 77 62 102 100 99 72* 101 98 84* 56* 101 100 96 62 100 4 0 0 0 0 0 0 4 8 1 1 3 21 0 0 2 8 0 3 3 5 0 0 0 4 0 0 1 8 0 0 0 0 0 1 0 0 0 0 0 0 10* 23* 1 1 16* 23* 0 0 1 26* 0 0 1 24* 0 0 0 18* 0 0 10* 26* 0 0 0 18* 0 1 a Injury ratings on a scale of 0 (no injury) to 100% (complete kill). Measured by recording the day when 95% of the soybean pods in each plot reached maturity compared to the nontreated control. c Values followed by an asterisk indicate a significantly higher level of visible injury, soybean height reduction, and maturity delay between the V3 and R2 applications of a given active ingredient and rate, LSD (0.05). d LSD (0.05) within a column between herbicide treatments applied at the same soybean growth stage. b with 0.28 and 2.8 g ha 1 of aminopyralid displaying unexpanded/undeveloped bud clusters and stem epinasty. Aminocyclopyrachlor at 2.8 g ha 1 exhibited chlorotic terminal leaf cupping and mottling, as well as undeveloped bud clusters similar to aminopyralid. The 0.028, 0.28, and 2.8 g ha 1 rates of picloram applied at R2 resulted in slight cupping of the newest trifoliates. This differential response to the eight synthetic auxin herbicides was not surprising because plants absorb, translocate, and metabolize herbicides at different rates. By 4 WAT, all soybean exposed to synthetic auxin herbicides at the V3 growth stage, except for Solomon and Bradley: Sublethal rates of synthetic auxin herbicides on soybean ER 272 459 (297 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 164 of 245 28 g ha 1 clopyralid, picloram, aminocyclopyrachlor, and 2.8 and 28 g ha 1 of aminopyralid, had recovered from 2 wk prior (Table 3). Conversely, soybean treated with synthetic auxin herbicides at the R2 stage of growth did not recover as well, and in many instances exhibited similar levels of injury as 2 WAT. Soybean Height. Previous research has correlated soybean yield loss with reductions in plant height following an application of dicamba (Weidenhamer et al. 1989). In this research, reductions in plant height were generally correlated with, but less severe than, visible injury estimates. Greater height reductions occurred with all herbicides except for 2,4-D amine and triclopyr when applied at the V3 compared to the R2 stage of growth (Table 3). Auch and Arnold (1978) observed that the greatest soybean height reductions from dicamba applications were made at the early-bloom stage, as compared to applications made at vegetative growth stages or from midbloom through late-pod. At 2 WAT, soybean height was not reduced following V3 or R2 applications of 2,4-D and triclopyr at 0.028, 0.28, and 2.8 g ha 1, and for aminocyclopyrachlor, fluroxypyr, and picloram at 0.028 and 0.28 g ha 1, but was reduced for all rates of aminopyralid, clopyralid, and dicamba (Table 3). At 2 WAT when herbicides were applied at 28 g ae ha 1, soybean height expressed as a percent of the nontreated was equal for V3 and R2 applications of 2,4-D (94 and 95% of the nontreated) and clopyralid (52 and 56%), but height reduction for 28 g ha 1 was greater for R2 compared to V3 applications for aminocyclopyrachlor (52 and 68%), aminopyralid (44 and 59%), dicamba (80 and 74%), fluroxypyr (58 and 74%), picloram (52 and 64%), and triclopyr (71 and 76%). At 4 WAT soybean height compared with the nontreated control was reduced with V3 and R2 applications of aminocyclopyrachlor, aminopyralid, clopyralid, dicamba, and picloram at 2.8 and 28 g ha 1 and with fluroxypyr and triclopyr at 28 g ha 1. Soybean Maturity. The specific herbicide, herbicide rate, and timing of herbicide application had significant effects on the delay in soybean maturity (Table 3). In general, applications made to soybean in the R2 stage of growth resulted in greater delays in soybean maturity compared to V3 herbicide applications. Wax et al. (1969) also observed greater 460 maturity delays following dicamba and picloram applications to soybean in the reproductive stages of growth compared to the prebloom stages of growth. Applications of aminocyclopyrachlor, clopyralid, dicamba, and picloram at 28 g ha 1 delayed maturity 5 to 8 d when applied at the V3 stage of growth and 23 to 26 d when applied at the R2 stage of growth (Table 3). V3 and R2 applications of 28 g ha 1 aminopyralid delayed maturity 21 and 23 d, respectively. Applications of aminocyclopyrachlor, aminopyralid, dicamba, and picloram at 2.8 g ha 1 delayed soybean maturity 1 to 4 d when applied at the V3 stage of growth and 1 to 16 d when applied at the R2 stage of growth. Soybean maturity was not delayed for 2,4-D regardless of application timing or for triclopyr at all rates at V3. Wax et al. (1969) also reported that dicamba delayed soybean maturity more than 2,4-D. Triclopyr applied at R2 delayed maturity 18 d for only the 28 g ha 1 rate. Soybean Yield. In general, herbicide treatments and rates resulting in less than 10% injury 2 WAT did not reduce yield (Tables 3 and 4). Except for either application timing of 2,4-D amine and V3 applications of dicamba, all herbicides resulted in greater soybean yield loss with increasing herbicide rates (Table 4). Additionally, greater soybean yield loss occurred with applications made to R2 compared to V3 soybean, except for 2,4-D amine, which did not reduce soybean yield compared to the nontreated control at either application timing. This result is consistent with previous research; Slife (1956) and Wax et al. (1969) reported less yield reduction from early compared to later 2,4-D treatments, and Robinson et al. (2013) reported soybean yield losses of 5% with V2 or R2 applications of 2,4-D at rates up to 116 g ha 1. Soybean yield after R2 applications of dicamba ranged from 2 to 67% less than the nontreated control, but V3 applications of dicamba did not result in any soybean yield loss. This result is in agreement with previous research, where 9 to 11 g ha 1 dicamba reduced yields in the flowering stage, compared with prebloom applications that required rates of 56 to 70 g ha 1 to reduce yields (Auch and Arnold 1978; Wax et al. 1969). In relation to the significant injury following early-season dicamba applications, Behrens and Leuschen (1979) determined yield reductions following dicamba drift injury to soybean at the first trifoliate stage were associated with injury ratings of 60 to 70 or more. Weed Technology 28, July–September 2014 ER 273 (298 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 165 of 245 Table 4. Soybean yield and yield components in response to eight synthetic auxin herbicides applied at the V3 and R2 stages of soybean growth combined across 2011 and 2012. Herbicide Rate g ae ha 2,4 D amine Aminocyclopyrachlor Aminopyralid Clopyralid Dicamba Fluroxypyr Picloram Triclopyr Nontreated LSD (0.05)b 0.028 0.28 2.8 28 0.028 0.28 2.8 28 0.028 0.28 2.8 28 0.028 0.28 2.8 28 0.028 0.28 2.8 28 0.028 0.28 2.8 28 0.028 0.28 2.8 28 0.028 0.28 2.8 28 1 Soybean yieldab Seeds per podab Pods per plantab Seed weightab V3 V3 V3 V3 R2 kg ha 4,345 4,306 4,462 4,306 4,513 4,440 4,222* 1,927* 4,141 4,086 3,329* 423* 4,369 4,015 3,944 1,838* 4,147 4,260 4,178* 4,128* 4,463 4,447 4,289 3,079* 4,464 4,401 4,088* 2,070* 4,446 4,360 4,543 3,832* 4327 267 R2 1 R2 No. 4,340 4,395 4,354 4,373 4,466 4,594 3,823 435 4,016 3,898 2,752 135 4,640* 4,073 3,795 622 4,222 4,052 3,730 1,427 4,671 4,425 4,530 2,306 4,511 4,242 3,653 480 4,464 4,550 4,513 2,468 4,327 234 2.22 2.27 2.26 2.23 2.28 2.20 2.27* 2.23 2.27 2.26* 2.10* 0.76* 2.25 2.19 2.24 2.28* 2.17 2.17 2.16* 2.20* 2.29 2.23 2.28 2.30* 2.27 2.22 2.28 2.29* 2.13 2.25 2.35 2.31* 2.27 0.12 2.33 2.22 2.20 2.20 2.24 2.18 2.02 0.19* 2.17 2.07 1.93 0.01 2.20 2.15 2.00* 0.08 2.06 2.07 2.00 0.64 2.17 2.22 2.30 1.07 2.27 2.18 2.15 0.12 2.20 2.23 2.33 1.07 2.27 0.14 R2 g 100 seeds 45 45 49 51 46 46 48* 45* 45 49* 44 16* 44 47 48* 49* 45 50 45 50* 50 45 49* 50* 47 45 44 53* 51 50 47 49* 48 8 55* 53* 48 45 48 45 37 7 40 40 41 1 48 46 40 9 42 43 39 13 46 48 40 15 44 44 42 10 53 49 45 11 48 6 16.77 16.68 16.63 16.88 16.72 16.40 16.24 16.42 16.37 16.25 16.24 16.61 16.52 16.08 16.14 16.33 16.23 16.35 16.44 16.35 16.47 16.60 16.80 16.45 16.79 16.53 16.39 16.34 16.78 16.67 16.87 16.45 16.70 0.37 1 16.62 16.83 16.66 17.25 17.11 17.18 19.37* 17.16* 17.99* 17.54* 18.79* 15.87 17.50 17.27* 18.01* 17.87* 18.11* 18.35* 17.73* 18.99* 17.02 16.99 17.35 18.98* 17.11 17.10 18.38* 16.67 16.67 17.07 17.69* 20.41* 16.70 0.89 a Values followed by an asterisk indicate a significantly higher level of soybean yield, seeds per pod, pods per plant, and seed weight between the V3 and R2 applications of a given active ingredient and rate, LSD (0.05). b LSD (0.05) within a column between herbicide treatments applied at the same soybean growth stage. Other authors (Auch and Arnold 1978; Slife 1956; Wax et al. 1969) have also noted greater yield reductions following dicamba applications to soybean in the reproductive rather than vegetative stages of growth. Conversely, Kelley et al. (2005) reported equivalent or greater yield reductions from V3 applications of dicamba, 2,4-D, and clopyralid, compared to R2 applications of these same herbicides. Regardless of growth stage, yields were significantly reduced following 0.28, 2.8, and 28 g ha 1 clopyralid and 2.8 and 28 g ha 1 picloram. Only 2.8 and 28 g ha 1 aminopyralid applied to V3 soybean reduced yield, while all aminopyralid rates applied to R2 soybean resulted in yields 7 to 97% less than the nontreated control. Similarly, only 28 g ha 1 aminocyclopyrachlor applied to V3 soybean reduced yield, while the 2.8 and 28 g ha 1 rates applied at the R2 stage reduced yield 12 and 90%, respectively. Lastly, only 28 g ha 1 of triclopyr and fluroxypyr applied at either growth stage resulted in yields less than the nontreated control. When Solomon and Bradley: Sublethal rates of synthetic auxin herbicides on soybean ER 274 461 (299 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 166 of 245 applied at 28 g ha 1 at the V3 stage of growth, the general order of herbicide-induced yield reductions to soybean from greatest to least was aminopyralid . aminocyclopyrachlor ¼ clopyralid ¼ picloram . fluroxypyr . triclopyr . dicamba 2,4-D amine. At the R2 stage of growth, the general order of herbicide-induced yield reductions from greatest to least was aminopyralid . aminocyclopyrachlor ¼ picloram . clopyralid . dicamba . fluroxypyr ¼ triclopyr . 2,4D amine. Interestingly, certain synthetic auxin treatments resulted in yields higher than the nontreated control (Table 4). When applied at the R2 stage of growth, 0.028 g ha 1 clopyralid and fluroxypyr resulted in yields 313 and 344 kg ha 1 greater than the nontreated control. This response can be explained by a phenomenon known as herbicide hormesis (Southman and Ehrlich 1943), or the Arndt-Schultz law (Thimann 1956), which states that every toxicant is a stimulant at low levels (Schabenberger et al. 1999). Several other authors have reported stimulatory effects on field crops from low concentrations of 2,4-D and other synthetic auxin herbicides (Miller et al. 1962; Taylor 1946; Wiedman and Appleby 1972). Soybean Yield Components. Generally, all synthetic auxin herbicides other than 2,4-D amine reduced soybean seeds per pod in response to increasing herbicide rates. All rates of 2,4-D amine resulted in seeds per pod equivalent to the nontreated control. In general, R2 applications of synthetic auxin herbicides influenced seeds per pod more than V3 applications, but the response varied by herbicide and rate (Table 4). Kelley et al. (2005) found that 5.6 g ha 1 dicamba reduced seeds per pod more when applied to soybean at V7 compared to V3 in 1 of 2 yr. Dicamba was the only herbicide where all rates applied to R2 soybean resulted in fewer seeds per pod than the nontreated control (Table 4). Following V3 applications, all herbicides except triclopyr and aminopyralid resulted in similar numbers of seeds per pod, regardless of herbicide rate. When compared to the nontreated control, 2.8 and 28 g ha 1 aminopyralid and 0.028 g ae ha 1 triclopyr were the only herbicides applied at the V3 timing that reduced soybean seeds per pod. Overall, seeds per pod were most affected by aminopyralid and least by 2,4-D amine; therefore, 462 the number of soybean seeds per pod were strongly correlated with the soybean yield losses observed. Following V3 applications, the number of pods per plant was only reduced in response to the highest rate of aminopyralid; all other synthetic auxin herbicides and rates resulted in a similar number of pods per plant as the nontreated control (Table 4). Kelley et al. (2005) reported that soybean treated at the V3 and V7 stages with 5.6 g ha 1 dicamba resulted in a similar number of pods per plant as the nontreated control. In contrast, following R2 applications, the number of pods per plant was highly influenced by herbicide rate. All synthetic auxin herbicides applied at the R2 stage of soybean growth resulted in significant differences in pods per plant in response to rate, with higher rates reducing pods per plant more than lower rates (Table 4). The lowest rate of 2,4-D applied to R2 soybean was the only treatment that resulted in more pods per plant than the nontreated control. All rates of aminopyralid, 2.8 and 28 g ha 1 dicamba, clopyralid, aminocyclopyrachlor, and fluroxypyr, and 28 g ha 1picloram and triclopyr applied to R2 soybean reduced pods per plant in comparison to the nontreated control. As with seeds per pod, the differences in pods per plant was greatest with aminopyralid and least with 2,4-D. Soybean seed weight was variable, with no consistent trend in response to either application timing. When applied at the V3 growth stage, there were no treatments that resulted in soybean seed weight greater than the nontreated control, whereas the same treatments applied to the R2 growth stage resulted in no seed weights less than the nontreated control (Table 4). Applications of 2,4-D at either soybean growth stage resulted in similar soybean seed weight as the nontreated control. Robinson et al. (2013) observed similar seed weight as the nontreated control with doses 560 g ha 1 2,4-D. Only 0.028 g ha 1 dicamba, 2.8 g ha 1 aminocyclopyrachlor, and 0.28 and 2.8 g ha 1 clopyralid and aminopyralid applied to V3 soybean resulted in seed weight less than the nontreated control. Wax et al. (1969) reported . 1 g reductions in seed weight per 100 seeds following prebloom applications of 1 to 33 g ha 1 dicamba. Following R2 applications, all rates of dicamba, and several rates of all other synthetic auxin herbicides other than 2,4-D resulted in seed weight greater than the nontreated control (Table 4). Weidenhamer et al. (1989) also observed Weed Technology 28, July–September 2014 ER 275 (300 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 167 of 245 increases in seed weight following later applications of dicamba, whereas earlier dicamba applications reduced seed weight. Wax et al. (1969) also reported greater soybean seed weight from latecompared to early-season treatments of dicamba and picloram, noting that the increased seed size did not counteract the reduction in seed number and thus resulted in lower yields. The increase in seed weight was likely due to the reduction in the number of seeds produced. The results from this research indicate that the risk to soybean from herbicide drift and/or tank contamination is dependent on herbicide, herbicide rate, and maturity of soybean following exposure. Overall, soybean are more likely to recover from misapplications of synthetic auxin herbicides made earlier, rather than later in the growing season. In this research, soybean exposed to synthetic auxin herbicides in early vegetative stages were able to maintain seed and pod set more efficiently than equivalent exposure to these herbicides at reproductive stages. In general, herbicide-induced injury increased with increasing herbicide rate, with aminopyralid, clopyralid, aminocyclopyrachlor, and dicamba resulting in more phytotoxicity to soybean than 2,4-D amine, triclopyr, and fluroxypyr. In this study, yield reductions were correlated with seeds per pod and pods per plant more so than seed weight. Literature Cited Al Khatib K, Peterson D (1999) Soybean (Glycine max) response to simulated drift from selected sulfonylurea herbicides, dicamba, glyphosate, and glufosinate. Weed Technol 13:264 270 Al Khatib K, Parker R, Fuerst EP (1992) Alfalfa response to simulated herbicide spray drift. Weed Technol 6:956 960 Al Khatib K, Parker R, Fuerst EP (1993) Wine grape response to simulated herbicide drift. Weed Technol 7:97 102 Andersen SM, Clay SA, Wrage LJ, Matthees D (2004) Soybean foliage residues of dicamba and 2,4 D and correlation to application rates and yield. Agron J 96:750 760 [ASPB] Arkansas State Plant Board (2012) Class F (2,4 D) Restricted Pesticide List. Little Rock, AR: Arkansas Agriculture Department. 3 p Auch DE, Arnold WE (1978) Dicamba use and injury on soybenas (Glycine max) in South Dakota. Weed Sci 26:471 475 Behrens MR, Mutlu N, Chakraborty S, Dumitru R, Jiang WZ, LaVallee BJ, Herman PL, Clemente TE, Weeks DP (2007) Dicamba resistance: enlarging and preserving biotechnology based weed management strategies. Science 316:1185 1188 Behrens R., Lueschen WE (1979) Dicamba volatility. Weed Sci 27:486 492 Blouin DC, Webster EP, Bond JA (2011) On the analysis of combined experiments. Weed Technol 25:165 169 [CCME] Canadian Council of Ministers of the Environment (1999) Canadian water quality guidelines for the protection of aquatic life: dicamba. Pages 1 3 in Canadian environmental quality guidelines, 1999.Winnipeg, Manitoba, Canada: Ca nadian Council of Ministers of the Environment Carmer SG, Nyquist WE, Walker WM (1989) Least significant differences for combined analysis of experiments with two or three factor treatment designs Agron J 81:665 672 Cranston HJ, Kern AJ, Hackett JL, Miller EK, Maxwell BD, Dyer WE (2001) Dicamba resistance in kochia. Weed Sci 49:164 170 Derksen DA (1989) Dicamba, chlorsulfuron, and clopyralid as sprayer contaminants on sunflower, mustard, and lentil, respectively. Weed Sci 37:616 621 Everitt JD, Keeling JW (2009) Cotton growth and yield response to simulated 2,4 D and dicamba drift. Weed Technol 23:503 506 Heap I (2013) International Survey of Herbicide Resistant Weeds. http://www.weedscience.org. Accessed: April 29, 2013 Heap I, Morrison IN (1992) Resistance to auxin type herbicides in wild mustard (Sinapis arvensis L.) populations in western Canada. Weed Sci Soc Am Ann Meeting Abstr 32:164 [Abstract] Hemphill DD, Montgomery ML (1981) Response of vegetable crops to sublethal application of 2,4 D. Weed Sci 29:632 635 Holt JS, LeBaron HM (1990) Significance and distribution of herbicide resistance. Weed Technol 4:141 149 Johnson VA, Fisher LR, Jordan DL, Edmisten KE, Stewart AM, York AC (2012) Cotton, peanut, and soybean response to sublethal rates of dicamba, glufosinate, and 2,4 D. Weed Technol 26:195 206 Kelley KB, Wax LM, Hager AG, Riechers DE (2005) Soybean response to plant growth regulator herbicides is affected by other postemergence herbicides. Weed Sci 53:101 112 Lanini WT (2000) Simulated drift of herbicides on grapes, tomatoes, cotton, and sunflower. Proc Calif Weed Conf 52:107 110 Marple ME, Al Khatib K, Shoup D, Peterson DE, Claassen M (2007) Cotton response to simulated drift of seven hormonal type herbicides. Weed Technol 21:987 992 Miller MD, Mikkelsen DS, Huffaker RC (1962) Effects of stimulatory and inhibitory levels of 2,4 D and iron on growth and yield of field beans. Crop Sci 2:114 116 National Climatic Data Center. http://www.ncdc.noaa.gov/. Accessesed April 29, 2013 Robinson AP, Davis VM, Simpson DM, Johnson WG (2013) Response of soybean yield components to 2,4 D. Weed Sci 61:68 76 Schabenberger O, Tharp BE, Kells JJ, Penner D (1999) Statistical tests for hormesis and effective dosages in herbicide dose response. Agron J 91:713 721 Sciumbato AS, Chandler JM, Senseman SA, Bovey RW, Smith KL (2004) Determining exposure to auxin like herbicides. I. Quantifying injury to cotton and soybean. Weed Technol 18:1125 1134 Solomon and Bradley: Sublethal rates of synthetic auxin herbicides on soybean ER 276 463 (301 of 886) Case: 17-70196, 02/09/2018, ID: 10759012, DktEntry: 71-2, Page 168 of 245 Slife FW (1956) The effect of 2,4 D and several other herbicides on weeds and soybeans when applied as post emergence sprays. Weeds 4:61 68 Southman CM, Ehrlich J (1943) Effects of extract of western red cedar heartwood on certain wood decaying fungi in culture. Phytopathology 33:517 524 Taylor DL (1946) Observations on the growth of certain plants in nutrient solutions containing synthetic growth regulating substances. I. Some effects of 2,4 D acid. Bot Gaz 107:597 611 Texas Agriculture Code (1984) Chapter 75. Pages 213 255. St. Paul, MN: West Thimann KV (1956) Promotion and inhibition: twin themes of physiology. Am Nat 40:145 162 Thompson MA, Steckel LE, Ellis AT, Mueller TC (2007) Soybean tolerance to early preplant applications of 2,4 D ester, 2,4 D amine, and dicamba. Weed Technol 21:882 885 Troyer JR (2001) In the beginning: the multiple discovery of the first hormone herbicides. Weed Sci 49:290 297 464 [USDA] United States Department of Agriculture, Economic Research Service (2012) 2013 http://ers.usda.gov/data products/adoption of genetically engineered crops in the us. aspx. Accessed April 29 Wax LM, Knuth LA, Slife FW (1969) Response of soybeans to 2,4 D, dicamba, and picloram. Weed Sci 17:388 393 Weidenhamer JD, Triplett GB, Sobotka FE (1989) Dicamba injury to soybean. Agron J 81:637 643 Wiedman SJ, Appleby AP (1972) Plant growth stimulation by sublethal concentrations of herbicides. Weed Res 12:65 74 Wright TR, Shan G, Walsh TA, Lira JM, Cui C, Song P, Zhuang M, Arnold NL, Lin G, Yau K, Russell SM, Cicchillo RM, Peterson MA, Simpson DM, Zhou N, Ponsamuel J, Zhang Z (2010) Robust crop resistance to broadleaf and grass herbicides provided by aryloxyalkanoate dioxygenase trans genes. Proc Natl Acad Sci U S A 107:20240 20245 Received September 23, 2013, and approved February 21, 2014. Weed Technology 28, July–September 2014 ER 277