Characterization of the ecological interactions of Roundup Ready 2 Yield® soybean, MON 89788, for use in ecological risk assessment

Abstract

As part of an ecological risk assessment, Roundup Ready 2 Yield® soybean (MON 89788) was compared to a conventional control soybean variety, A3244, for disease and arthropod damage, plant response to abiotic stress and cold, effects on succeeding plant growth (allelopathic effects), plant response to a bacterial symbiont, and effects on the ability of seed to survive and volunteer in a subsequent growing season. Statistically significant differences between MON 89788 and A3244 were considered in the context of the genetic variation known to occur in soybean and were assessed for their potential impact on plant pest (weed) potential and adverse environmental impact. The results of these studies revealed no effects of the genetic modification that would result in increased pest potential or adverse environmental impact of MON 89788 compared with A3244. This paper illustrates how such characterization studies conducted in a range of environments where the crop is grown are used in an ecological risk assessment of the genetically modified (GM) crop. Furthermore, risk assessors and decision makers use this information when deciding whether to approve a GM crop for cultivation in—or grain import into—their country.

Abbreviations

DAP

days after planting

DW

dry weight

EPSPS

5-enolpyruvylshikimate-3-phosphate synthase

ERA

ecological risk assessment

GM

genetically modified

SE

standard error

Introduction

One focus of an ecological risk assessment (ERA) for a genetically modified (GM) crop is whether the GM crop has the potential to become a pest (weed) or to have other adverse environmental impacts (e.g., effects on nontarget organisms). The requirement for an ERA as a condition of product approval has been specified by a number of regulatory agencies worldwide (e.g., US Code of Federal Regulations, 2008; Canadian Food Inspection Agency, 2012; European Food Safety Authority, 2004). In the ERA, the technology developer compares the GM crop plants to plant materials produced by conventional breeding (Conner et al., 2003; Roberts et al., 2014). Several authors (Raybould, 2007; Wolt et al., 2010) have provided conceptual outlines for risk assessment of GM crops. In previous reports (Horak et al., 2007; Sammons et al., 2014), we have described the risk assessments performed for Roundup Ready® Flex¹ cotton (Gossypium spp.) and drought-tolerant corn (Zea mays L.).

In this report, we summarize studies performed to support regulatory assessments and an ERA of Roundup Ready® 2 Yield soybean, ‘MON 89788’, a second-generation glyphosate-tolerant soybean product developed by Monsanto Company. MON 89788 contains the 5-enolpyruvylshikimate-3-phosphate synthase gene derived from Agrobacterium sp. strain CP4 (cp4 epsps). This gene produces the CP4 EPSPS protein, which protects the plants from the herbicidal effects of glyphosate, the active ingredient in Roundup® agricultural herbicides. This protein has also been expressed in numerous other Roundup Ready crops (Horak et al., 2015) and to date there have been no credible reports of adverse ecological impacts associated with cultivation or import of these crops (e.g., see Center for Environmental Risk Assessment, 2011). There has been no indication that the expression of the cp4 epsps gene would result in increased plant pest potential or adverse ecological impact based on protein function in the plant, experience with soybean, or the widespread cultivation and assessment of the first-generation Roundup Ready soybean and other commercialized Roundup Ready crops that contain CP4 EPSPS. The studies reported here were conducted to confirm that these previous observations also held true for MON 89788 and to meet specific regulatory requirements for crop cultivation in—or grain import into—various countries, and for use in an ERA.

When planning and conducting the ERA for MON 89788, plausible scientific risk scenarios were considered that could suggest increased weediness or adverse environmental impact. This approach is highlighted in the Cartagena Protocol on Biodiversity in Annex III section 8(a), which states that a risk assessment should include “an identification of any novel genotypic and phenotypic characteristics associated with the living modified organism that may have adverse effects on biological diversity in the likely potential receiving environment…” (Secretariat of the Convention on Biological Diversity, 2000). While Annex III is not detailed in its guidance, and there can be different interpretations of adverse effects depending on specific national environmental policies, practical regulatory realities exist due to public concerns and fundamentally diverging views over the inherent safety of the process of genetic modification. The tension between risk assessment science and the regulatory practicalities of registering a product in countries around the world typically impels registrants to do more than is scientifically justifiable. In the case of MON 89788, one can conclude that there are no plausible risk scenarios in which the expression of the cp4 epsps gene would result in weediness or adverse ecological effects based on knowledge of—and experience with—the trait. However, in practice, providing characterization data and other related information may (a) help regulators in their assessment activities and decisions; (b) be needed to meet specific regulatory data requirements in specific countries; and/or (c) be used to answer specific scientific or general-knowledge questions anticipated about the product.

This report complements studies on the germination, plant growth and development, and pollen characteristics of MON 89788 that are reported in Horak et al. (2015) and provides additional characterization data. In all of the studies, MON 89788 was compared to A3244, a conventional soybean variety with the same genetic background (Table S1). Horak et al. (2015) concluded that “the results of these studies revealed no effects attributable to the genetic modification process or to the GM trait in the plant that would result in increased pest potential or adverse ecological impact of MON 89788 compared with A3244.”

Herein we report data on the response of MON 89788 to natural abiotic stressors, plant–arthropod and plant–disease interactions, the ability of MON 89788 seed to survive natural environmental conditions during winter and produce volunteer plants in the spring, the potential of MON 89788 to exhibit allelopathic effects on subsequent plant growth, and the interaction of MON 89788 with symbiotic bacteria (Table 1). Each of these studies was designed to assess if there were any differences indicative of a change in MON 89788 as a result of genetic modification that would result in an adverse ecological effect, including adverse effects on nontarget organisms or an increase in weediness. The results of these studies were used primarily to characterize the interactions of MON 89788 with the receiving environment. The endpoints chosen to inform the risk assessment were based on the experience of plant breeders, weed scientists, regulators, and other agricultural experts familiar with soybean and with ecological risk assessment of GM crops.

TABLE 1.

Ecological characteristics evaluated on MON 89788 and A3244 in 2004–2006

Study	Characteristic	Evaluation timinga	Evaluation description
Phenotypic Study 1, 2, 3b	Arthropod, disease, abiotic stressors	Recurring, recorded approx. every 4 wk beginning at approx. V2–V4	Qualitative assessment of specific stressors rated on a 0–9 rating scale (0 = no stressor symptoms and 9 = most severe stressor symptoms) but reported by category
Phenotypic 1 & 2c,d	Arthropod abundance	Recurring approx. every 4 wk beginning approx. 4 wk after plants reached V2 stage	Quantitative assessment of arthropods collected
Allelopathic effects	Emergence–lettuce	7 DAP: Shoot tissue evaluation; 9 DAP: Soil medium evaluation	Number of seedlings in each pot with 2 fully unfurled cotyledons
Allelopathic effects	Growth stage (leaf number)–lettuce	3 wk after planting (shoot tissue and soil medium evaluations)	Fully unfurled true leaves (i.e., not cotyledons)
Allelopathic effects	Plant height–lettuce	3 wk after planting (shoot tissue and soil medium evaluations)	Determined by measuring from the soil surface to the end of the longest leaf
Allelopathic effects	Fresh weight–lettuce	3 wk after planting (shoot tissue and soil medium evaluations)	Determined by cutting the plant at the soil surface. The shoot tissue of each plant was weighed within 1 h of harvest
Allelopathic effects	Dry weight–lettuce	3 wk after planting (shoot tissue and soil medium evaluations)	After fresh weight measurements were collected, the shoot tissue samples were oven-dried at approximately 40°C for 7 d, then weighed
Cold tolerance	Plant height	37 DAP: 20 d under either optimal temperature (29°C day/24°C night) or cold (15°C day/8°C night)	Distance from the soil line to the shoot apex
Cold tolerance	Vegetative growth stage	37 DAP: 20 d under either optimal temperature (29°C day/24°C night) or cold (15°C day/8°C night)	Developmental stage assessed by counting the uppermost fully developed trifoliate leaf node (V1 = first trifoliate leaf has unrolled leaves…V(n) = nth trifoliate leaf has unrolled leaves)
Cold tolerance	Vigor	37 DAP: 20 d under either optimal temperature (29°C day/24°C night) or cold (15°C day/8°C night)	Rated on a 0–9 scale, where 0 = dead plant, 1–3 = below average vigor, 4–6 = average vigor, and 7–9 = above average vigor
Cold tolerance	Fresh weight	37 DAP: 20 d under either optimal temperature (29°C day/24°C night) or cold (15°C day/8°C night)	Determined by cutting the plant at the soil surface and immediately weighing above-ground tissue
Cold tolerance	Dry weight	37 DAP: 20 d under either optimal temperature (29°C day/24°C night) or cold (15°C day/8°C night)	After fresh weight measurements were collected, plants were oven-dried at approximately 38°C for 9 d, then weighed
Volunteer potential	Number of emerged plants after winter sowing	6 observations per site, spring through mid-June
Symbiont study	Nodule number	4 and 6 wk after emergence	Nodules were separated from roots for counting
Symbiont study	Nodule dry weight	4 and 6 wk after emergence	Dry weight of nodules separated from roots prior to drying
Symbiont study	Root dry weight	4 and 6 wk after emergence	Dry weight of roots separated from nodules prior to drying
Symbiont study	Shoot dry weight	4 and 6 wk after emergence	Dry weight of material cut off at surface of the potting medium
Symbiont study	Shoot total nitrogen	4 and 6 wk after emergence	After drying and weighing, shoot tissue was ground and sieved. Total nitrogen was determined by combustion using a nitrogen analyzer

^aPhenotypic and Volunteer studies were performed in the field; all others were greenhouse or growth chamber studies. DAP, days after planting. Plant developmental stages are as described in Pedersen (2004).

^bMeasurements at all 27 locations used in these 3 studies (Table S2).

^cIL, MO1, and MO2 sites in Phenotypic Study 1 (Table S2).

^dIA, IN, and NE sites in Phenotypic Study 2 (Table S2).

Risk is often expressed as a function of a hazard and exposure (Roberts et al., 2014). For the ERA of a GM crop, hazard components of risk may include the potential of the crop plant itself to have an adverse effect on a nontarget organism or to become a weed of agriculture or native plant communities. For the exposure component of risk, the assessor will consider whether the crop will be cultivated, the potential area of cultivation, or, in the case of grain import into a country where the crop is not cultivated, the likelihood of environmental exposure from incidentally released grain (Roberts et al., 2014). In most cases where the crop will be cultivated, exposure is considered to be 100%, meaning that the plant will be introduced into the environment via crop planting and thus environmental exposure will occur. The data reported here and in Horak et al. (2015) were used to assess potential hazards and relevant exposure scenarios, focusing on weediness and adverse environmental effects, in order to address specific regulatory requirements and anticipated questions and ultimately to characterize the environmental risk of MON 89788.

Results

Field Observations

Phenotypic Study 1—US

Across 17 US sites (Table S2), plant responses to a total of 8 abiotic stressors, 18 diseases, and 12 arthropod categories (species or group), were evaluated. For all 212 abiotic stress and 216 disease damage observations, the results were similar for MON 89788 and the control (Tables S3–S6). Of 221 arthropod damage observations, a single qualitative difference was noted in MON 89788 plots compared to the control plots. The severity of symptoms caused by leafhopper was lower in MON 89788 plots than the control plots at the MO1 site at observation 4 (“none” vs. “slight;” Tables S7, S8). Leafhopper symptoms were not different between the MON 89788 plots and the control plots at other sites or at other observation times at the MO1 site, and the symptoms on MON 89788 fell within the range of symptoms observed in the references.

Specific pest and beneficial arthropods were collected and quantified at the IL1, MO1, and MO2 sites (Table 2 and Table S9). Overall, the arthropod abundance numbers were low due to limitations of the sampling method. No statistical differences were detected in arthropod abundance on MON 89788 compared to the control for 65 out of 68 comparisons. Abundance on MON 89788 was higher than on the control for corn earworm at IL1 collection 2 (0.7 vs. 0.0, respectively), southern corn rootworm at MO1 collection 2 (2.0 vs. 0.0, respectively), and tarnished plant bug at MO2 collection 3 (0.3 vs. 0.0, respectively). No differences were detected between MON 89788 plots and the control plots at a second collection time or site for southern corn rootworm or tarnished plant bug, respectively; additional observations were not available for corn earworm.

TABLE 2.

Phenotypic Study 1 summary of arthropod abundance data from beat-sheet samples of MON 89788, the control, and the references at 3 US sites

			Quantitative differencesb
					Arthropod abundance
Arthropod	Number of observations across sitesa	Number of observations where no differences were detected between MON 89788 and A3244	Sitec	Collection number	MON 89788 Mean (SE)d	A3244 Mean (SE)d	Reference rangee
Pests
Aphids (Aphididae)	8	8	—	—	—	—	—
Bean leaf beetle (Cerotoma trifurcata)	8	8	—	—	—	—	—
Corn earworm (Helicoverpa zea)	1	0	IL1	2	0.7 (0.3)	0.0 (nv)	0.0–0.0
Flea beetle (Chrysomelidae)	4	4	—	—	—	—	—
Garden fleahopper (Halticus bractatus)	1	1	—	—	—	—	—
Grape colaspis (Colaspis brunnea)	2	2	—	—	—	—	—
Green cloverworm (Plathypena scabra)	7	7	—	—	—	—	—
Japanese beetle (Popillia japonica)	1	1	—	—	—	—	—
Southern corn rootworm(Diabrotica undecimpunctata)	2	1	MO1	2	2.0 (nv)	0.0 (nv)	0.0–0.3
Stink bug (Pentatomidae)	5	5	—	—	—	—	—
Tarnished plant bug (Lygus spp.)	2	1	MO2	3	0.3 (0.3)	0.0 (nv)	0.0–0.0
Weevil (Curculionidae)	2	2	—	—	—	—	—
Beneficial
Asian ladybeetle (Coccinellidae)	1	1	—	—	—	—	—
Lacewing (Chrysopidae)	1	1	—	—	—	—	—
Damsel bug (Nabidae)	7	7	—	—	—	—	—
Minute pirate bug (Orius spp.)	8	8	—	—	—	—	—
Pink spotted ladybeetle (Coccinellidae)	1	1	—	—	—	—	—
7-spotted ladybeetle (Coccinellidae)	1	1	—	—	—	—	—
Spider (Araneae)	6	6	—	—	—	—	—

^aA total of 68 arthropod abundance observations were made across sites. Detailed results are provided in Table S9.

^bScores shown were statistically significant differences between MON 89788 and A3244 at p ≤ 0.05. – indicates that there were no differences detected between MON 89788 and A3244 for the arthropod indicated.

^cSite codes: IL1= Clinton Co., IL; MO1= Shelby Co., MO; MO2= Lincoln Co., MO (Table S2).

^dSE standard error, nv no variability.

^eReference range = minimum and maximum mean values among the commercial reference varieties.

Phenotypic Study 2—US

Across five US sites (Table S2), plant responses to a total of 15 arthropod categories (species or group), 17 diseases, and 9 abiotic stressors were evaluated. The results were similar for MON 89788 and the control in 47 comparisons of plant injury caused by abiotic stressors (Tables S3 and S10), 60 comparisons of plant injury caused by disease (Tables S5 and S11), or 61 comparisons of plant injury caused by arthropods (Tables S7 and S12).

Comparisons of the nontarget pest and beneficial arthropod abundance on MON 89788 and the control are presented in Table 3 and Table S13. Among 50 comparisons, no statistical differences were detected between MON 89788 and the control for 9 of 12 species evaluated. Bean leaf beetle (pest) and spider (beneficial) abundance was statistically lower at the IN site collection 3 and NE site collection 2, respectively, for MON 89788 compared to the control (4.0 vs. 23.0 and 7.0 vs. 15.7 per plot, respectively). Green cloverworm (pest) abundance was significantly higher at the IA site collection 2 for MON 89788 compared to the control (8.7 vs. 3.7 per plot). The mean abundance values for green cloverworm and spiders on MON 89788 were within their respective reference ranges. Furthermore, the statistical differences for bean leaf beetle, spider, and green cloverworm abundance were not consistent across sites or collection times.

TABLE 3.

Phenotypic Study 2 summary of arthropod abundance data from beat-sheet samples of MON 89788, A3244, and the references at 3 U.S. sites

			Quantitative differencesb
					Arthropod abundance
Arthropod	Number of observations across sitesa	Number of observations where no differences were detected between MON 89788 and A3244	Sitec	Collection number	MON 89788 Mean (SE)d	A3244 Mean (SE)d	Reference rangee
Pests
Aphids (Aphididae)	3	3	—	—	—	—	—
Bean leaf beetle (Cerotoma trifurcata)	8	7	IN	3	4.0 (2.7)	23.0 (11.1)	5.7–19.3
Green cloverworm (Plathypena scabra)	3	2	IA	2	8.7 (2.3)	3.7 (0.9)	3.0–8.7
Japanese beetle (Popillia japonica)	2	2	—	—	—	—	—
Leafhoppers (Cicadellidae)	3	3	—	—	—	—	—
Stink bug (Pentatomidae)	4	4	—	—	—	—	—
Tarnished plant bug (Lygus spp.)	1	1	—	—	—	—	—
Beneficial
Big-eyed bugs (Geocoris spp.)	2	2	—	—	—	—	—
Parasitic wasps (Hymenoptera)	2	2	—	—	—	—	—
Damsel bugs (Nabidae)	7	7	—	—	—	—	—
Minute pirate bugs (Orius spp.)	7	7	—	—	—	—	—
Spiders (Araneae)	8	7	NE	2	7.0 (1.2)	15.7 (4.1)	4.0–9.0

^aIncludes 50 observations at which both MON 89788 and A3244 abundance were ≥1 .0. Detailed results are provided in Table S13.

^bScores shown were statistically significant differences between MON 89788 and A3244 at p ≤ 0.05. – indicates that there were no differences detected between MON 89788 and A3244 for the arthropod indicated.

^cSite codes: IN = Hendricks Co., IN; IA = Jefferson Co., IA; NE = York Co., NE (Table S2).

^dSE standard error.

^eReference range = minimum and maximum mean values among the commercial reference varieties.

Phenotypic Study 3—Argentina

Across 5 sites in Argentina (Table S2), the results were similar for MON 89788 and A3244 for the 9 assessed ecological stressors including Epinotia aporema, Loxostege bifidalis, Nezara viridula, Rachiplusia nu, Cercospora kikuchii, Diaporthe complex, Septoria glycines, drought, and sunscald (Table S14).

Cold Tolerance Study

Test and control plants selected and transferred from the greenhouse to growth chambers were uniform in vigor (8.0 vs. 8.0), plant growth stage (3.0 vs. 3.0), and plant height (24.9 [SE 0.26] vs. 24.9 [SE 0.48] cm) at the time of transfer. No significant differences were detected under cold temperatures between MON 89788 and the control for plant height and biomass (plant fresh or dry weight) (Table 4). Limited variability and small sample size in the plant growth stage and vigor data precluded statistical analysis because analysis of variance assumptions could not be met. However, plant growth stage was similar between MON 89788 and the control and plants had similar vigor (Table 4). The purpose of the plants grown in the optimal chamber was to demonstrate that the experimental system was working (i.e., that soybean growth was reduced in cold temperatures. The effect of cold stress on soybean development was demonstrated by comparing the reference data from each chamber (Table 4).

TABLE 4.

Comparison of MON 89788 to A3244 after exposure to cold temperatures for 20 d

	Material/temperature
	MON 89788 / cold Mean (SE)a	A3244 / cold Mean (SE)a	Reference range: coldb	Reference range: optimalb
Plant height (cm)	31.0 (0.4)	30.7 (0.4)	20.0–47.0	77.0–161.0
Plant fresh wt (g)	20.0 (1.0)	21.3 (0.8)	13.2–29.9	54.8–116.2
Plant dry wt (g)	4.5 (0.3)	4.9 (0.2)	3.0–7.0	12.1–27.7
Vegetative stagec	4.2 (0.1)	4.0 (nv)	4.0–5.0	11.0–16.0
Vigorc,e	2.9 (0.1)	3.0 (nv)	3.0–4.0	8.0–8.0

No significant differences were detected between MON 89788 and A3244 for plant height and biomass (p ≤ 0.05).

^aSE standard error, nv no or very low variation.

^bMinimum and maximum values from 4 commercially available soybean varieties (Table S1).

^cThe analysis of variance model was not fit for soybean development stages and plant vigor ratings because the small sample size and limited variability of the observed categorical levels for these responses would not allow analysis of variance assumptions to be met.

^dVegetative stages (V stages or node stages) are defined and numbered according to the uppermost fully developed trifoliolate leaf node (unrolled leaflets at the node). V1–first-node (first trifoliolate); V2–second-node (second trifoliolate); V3–third-node (third trifoliolate);…V(n) nth-node (nth trifoliolate).

^eVigor rated on 0–9 scale where 0 = dead plant, 1–3 = below average vigor, 4–6 = average vigor, and 7–9 = above average vigor.

Volunteer Potential Study

No differences in the number of volunteer plants were detected between MON 89788 and the control at any site. No volunteer plants of MON 89788 or A3244 were observed at any time during Spring 2006 at the IL1, IN1, and MO1 sites; reference ranges for these sites were 0.0–0.3 (IL1, IN1) and 0.0–1.0 (MO1). At the NE site, a small number of volunteer plants were observed but MON 89788 was not significantly different from A3244 (mean across all dates was 4.3 [SE 2.0] for MON 89788 vs 5.3 [SE 3.3] for A3244; reference range 0.3–7.0).

Study of Allelopathic Effects

No statistically significant differences were detected in either the shoot tissue evaluation or the soil medium evaluation for plant height, fresh weight, or dry weight of lettuce plants grown in pots containing the residue of either MON 89788 or the A3244 control (Table 5). Furthermore, no gross morphological differences were observed between lettuce plants grown in pots containing the residue of MON 89788 or the control in either the shoot tissue evaluation or the soil medium evaluation (data not shown). Limited variability in the lettuce emergence and growth stage categorical data and small sample size precluded statistical analysis because analysis of variance assumptions could not be met. However, lettuce emergence and growth stage were similar for the 2 genotypes in each medium.

TABLE 5.

Comparison of MON 89788 to the control for allelopathic effects on the growth and development of lettuce

		MON 89788 Mean (SE)a	A3244 Mean (SE)a	Reference rangeb
Ground soybean shoot tissue	Emergence (%)c,d	63.9 (9.2)	75.0 (9.2)	38.9–88.9
	Growth stage (# leaves/plant)d	7.5 (0.3)	7.7 (0.3)	7.3–8.7
	Plant height (cm)	16.8 (1.0)	16.8 (0.9)	15.8–18.3
	Fresh weight (g)	5.1 (0.6)	5.1 (0.6)	4.6–7.5
	Dry weight (g)	0.24 (0.03)	0.24 (0.02)	0.2–0.4
Soil medium in which soybean plants were previously grown	Emergence (%)c,d	88.9 (8.7)	88.9 (8.7)	72.2–94.4
	Growth stage (# leaves/plant)d	8.3 (0.2)	8.2 (0.2)	7.7–8.5
	Plant height (cm)	15.7 (0.6)	15.0 (0.6)	13.8–15.0
	Fresh weight (g)	7.1 (0.7)	6.2 (0.7)	4.8–6.6
	Dry weight (g)	0.47 (0.05)	0.42 (0.05)	0.3–0.5

No statistically significant differences were detected between MON 89788 and the control at p ≤ 0.05 for lettuce plant height, fresh weight, or dry weight in either the shoot tissue or soil medium evaluation.

^aSE standard error.

^bMinimum and maximum mean values from among 6 commercially available soybean varieties.

^cPercentage of seeds planted in each pot that emerged. n = 6 seeds per pot.

^dLettuce emergence and growth stage data were not subjected to an analysis of variance because the small sample size and limited variability in the observed categorical levels for these responses would not allow analysis of variance assumptions to be met.

Symbiont Study

No significant differences were detected between MON 89788 and the control for any measured parameter, including nodule number, shoot total nitrogen, and biomass (dry weight) of nodules, shoot material, and root material at each of the 2 sampling periods (Table 6).

TABLE 6.

Comparison of symbiont interactions of MON 89788 and A3244 control

Measurement endpointa	Sampling period	MON 89788 Mean (SE)b	A3244 Mean (SE)b	Reference range
Nodule DW (mg/plant)	4 wk	0.10 (0.01)	0.10 (0.01)	0.09–0.13
	6 wk	0.35 (0.04)	0.38 (0.07)	0.31–0.42
Nodule number (per plant)	4 wk	41.25 (5.01)	51.75 (6.30)	29.63–48.75
	6 wk	147.13 (15.58)	153.25 (19.89)	94.50–103.75
Root DW (mg/plant)	4 wk	0.95 (0.16)	0.88 (0.11)	0.94–0.98
	6 wk	2.60 (0.37)	2.66 (0.26)	1.52–1.96
Shoot DW (mg/plant)	4 wk	1.22 (0.21)	1.15 (0.14)	1.59–1.99
	6 wk	4.96 (0.57)	5.43 (0.80)	4.90–7.52
Shoot total nitrogen (% DW)	4 wk	3.31 (0.16)	3.04 (0.23)	2.39–3.28
	6 wk	2.87 (0.14)	2.73 (0.15)	2.72–3.01

No significant difference (at p ≤ 0.05) was found between MON 89788 and the control for any measured parameter.

^aDW dry weight.

^bSE standard error.

DISCUSSION

Hazard Characterization and Data Transportability

pt?>The studies reported in Horak et al. (2015) and herein provide extensive plant characterization data that can be used in a hazard characterization of MON 89788, to address specific regulatory requirements, to address anticipated questions, and ultimately to characterize risk for endpoints of interest to risk assessors (US Code of Federal Regulations, 2008). Weediness and adverse ecological impacts of MON 89788 soybean are 2 potential hazards of focus in this assessment. As described in the Introduction, parts of this characterization of risk may not be in line with risk assessment theory as outlined in guidance such as the Cartagena Protocol (Secretariat of the Convention on Biological Diversity, 2000) or other publications (Raybould, 2007; Wolt et al., 2010). However, from a practical standpoint, these characterization studies provide regulators with information that is either required by government authorities or assists in decision making by addressing perceived uncertainty.

In order for MON 89788 to become a weed of agriculture or natural plant communities, several phenotypic and physiological characteristic changes would likely be required (Baker, 1974; Warwick and Stewart, 2005). Soybean is a highly domesticated crop and because of selection by plant breeders it lacks significant levels of dormancy, it has no effective dispersal mechanism, and it is selected to grow in monoculture (Warwick and Stewart, 2005). It could be hypothesized that in order for soybean to function as a weed, at a minimum it would need to possess several characteristics: (a) significant levels of seed dormancy that would allow the seed to survive in the soil until conditions are favorable for germination, e.g., not in the winter for a summer annual (Radosevich et al., 1997; Warwick and Stewart, 2005); (b) a significant seed dispersal mechanism that would move the seed to new areas for germination and subsequent plant growth (Radosevich et al., 1997); and (c) the ability to compete with native vegetation so that the plants would survive in naturally competitive plant communities (Radosevich et al., 1997). For MON 89788 there were no or few differences detected across the germination, phenotypic, and pollen studies that would increase the weediness and invasive characteristics of MON 89788, so the hazard for these particular endpoints would be negligible (Horak et al., 2015).

Based on a tiered assessment approach, the germination, emergence and plant growth, and reproduction data presented in Horak et al. (2015) did not indicate any differences that would signal the need for further assessments of weediness and invasiveness. However, in order to address specific regulatory requirements and the other needs described above, the studies reported herein were conducted. To assess the interaction of MON 89788 seed with natural environmental conditions, the volunteer potential study was conducted. Volunteer crop plants in a subsequent crop (e.g., volunteer corn in soybean) can be a significant agronomic challenge (Beckett and Stoller, 1988; Krupke et al., 2009; Marquardt et al., 2012). As reported here, the results of the volunteer potential study confirmed that there was no increased ability of MON 89788 to volunteer (germinate, emerge, and grow) in a subsequent season compared to the control and thus no increased weediness through this mechanism.

Changes in the way the plant interacts with the abiotic and biotic environment may also affect potential weediness (Radosevich et al., 1997). For example, the ability to tolerate various environmental conditions such as cold and the ability to compete interspecifically by special mechanisms such as allelopathy are both characteristics that are common in weed species (Radosevich et al., 1997). Certain changes in these characteristics (e.g., extreme cold tolerance or increased allelopathic tendencies) in MON 89788 compared to the A3244 control could be indicative of changes toward increased weediness. The absence of detected differences in plant growth under controlled cold conditions, the lack of significant difference in plant response to abiotic stressors in field studies, and the lack of detected differences in the characteristics measured in the allelopathy study between MON 89788 and the control support a conclusion that the introduction of the glyphosate tolerance trait did not increase weed potential.

These results confirm a lack of weediness of MON 89788 in a managed agricultural field and unchanged interactions with specific abiotic stressors. By extrapolation, this conclusion could be expanded to unmanaged natural plant communities. If there are few volunteers in a highly managed field with no competing vegetation, there would be little reason to believe that MON 89788 would demonstrate the ability to survive and invade native plant communities, where competition both within and among species is one of the major forces determining abundance of plant species in the plant community (Tilman, 1997).

The second potential hazard being assessed is adverse ecological impact (Canadian Food Inspection Agency, 2012; European Food Safety Authority, 2004; US Code of Federal Regulations, 2008). Although many endpoints could be considered in this category, this assessment focused on potential adverse effects to nontarget organisms including arthropods and symbionts. Assessment of impact on nontarget organisms is an important part of the ERA (Canadian Food Inspection Agency, 2012; European Food Safety Authority, 2004; Garcia-Alonso et al., 2006; US Code of Federal Regulations, 2008). Extensive safety information has been prepared for the CP4 EPSPS protein in Roundup Ready crops and was summarized by the Center for Environmental Risk Assessment (2011). The conclusion of these studies is that the CP4 EPSPS protein confers tolerance to Roundup agricultural herbicides and has no other adverse effects (Center for Environmental Risk Assessment, 2011). Although there are not thought to be direct plausible hypotheses that the CP4 EPSPS protein would affect plant ecological interactions with arthropods and symbionts in the absence of Roundup agricultural herbicides, several studies were conducted to provide information to meet specific regulatory requirements, answer anticipated questions, and inform the risk assessment. The results of the symbiont study support the conclusion that the introduction of the glyphosate tolerance trait does not alter the relationship between B. japonicum and MON 89788 compared to the conventional control. In addition, the results of qualitative and quantitative field assessments in different geographic regions (North America and South America) and years support the conclusion that the glyphosate tolerance trait did not unexpectedly alter the ecological interactions of MON 89788 compared to A3244. The results of these studies clearly supported the plant characterization of MON 89788 and indicated that there were no differences detected in the plants as revealed by symbiont interactions, disease–plant interactions, and arthropod–plant interactions that would be indicative of an adverse ecological impact. Based on the weight of evidence of these data, it can be concluded that MON 89788 did not differ from conventional soybean in characteristics that would be indicative of an adverse ecological impact. Thus, when plant characterization data and data on ecological interactions are considered for an assessment of weediness (including invasiveness) and adverse ecological interactions, the potential hazard of MON 89788 compared to the control is considered negligible. These conclusions are similar to what has been reported for other commercially available GM crops including Roundup Ready® Flex cotton and drought-tolerant corn (Horak et al., 2007; Sammons et al., 2014).

The data that are used to assess specific characteristics of soybean and soybean environmental interactions are relevant and “transportable” to other regions with similar ecological endpoints (Roberts et al., 2014). The results of the assessment of potential risks associated with MON 89788, and in particular those based on data from 2 growing seasons in the United States and one in Argentina, did not differ. Since the characteristics assessed are familiar to soybean breeders, technology developers, and risk assessors and since some endpoints may be common across regions, the data may be used to inform the risk assessment in geographies and world areas beyond where the data were collected (European Food Safety Authority, 2004; Roberts et al., 2014).

Exposure Characterization

In theory, with no plausible hypothesis of how the expression of the cp4 epsps gene could result in a potential hazard (weediness or adverse ecological impacts), an exposure assessment is not needed. In practice, however, regulators do request information on potential exposure, so these data are generated as part of the risk assessment.

In countries where cultivation of MON 89788 is expected, the risk assessment assumes a higher level of exposure of these plants to the environment (Roberts et al., 2014). In countries where MON 89788 is imported as a food or feed source and not cultivated, environmental exposure is greatly reduced. In import countries, grain imports are handled in a confined manner designed to limit environmental exposure via seed loss during transportation. Nevertheless, some incidental loss can occur along transportation routes and at seed handling/processing facilities. In these cases, environmental exposure is limited, it does not occur widely in the landscape, and it is often in environments not conducive to the long-term persistence of soybean. Thus, although exposure is not zero, it can be considered low to very low when considered in the context of the total land area of an importing region or country.

Risk Characterization

As described earlier, risk is a function of hazard and exposure (Roberts et al., 2014), so risk assessment must take both of these factors into account. Hazard assessment based upon plant characterization studies presented previously (Horak et al., 2015) and in this paper indicate that there is negligible hazard (weediness or adverse environmental impacts) associated with MON 89788 compared to the control. Because hazard is negligible, the overall risk associated with the cultivation and import of MON 89788 compared to the control is also negligible even in areas where exposure is high (i.e., in countries where the product is intended for cultivation).

Conclusions

Biological and ecological data on Roundup Ready 2 Yield soybean, MON 89788, were compared to an appropriate control, A3244. The genetic modification did not increase the pest/weed potential or result in increased adverse environmental impact of MON 89788 compared to A3244 and other conventional soybean. The results of this study support the conclusion that Roundup Ready 2 Yield soybean, MON 89788, is no more likely to pose a plant pest risk or to have adverse environmental impact than conventional soybean either when cultivated in a range of diverse geographies or when used for import for food, feed or processing.

Materials and Methods

MON 89788, Control, and Reference Materials

MON 89788 seed, plants, residue, or exudates were used in these studies (Table S1). The control materials were seed, plants, residue, or exudates of the non-glyphosate-tolerant soybean variety A3244. A3244 has a genetic background similar to the test material with the exception of the glyphosate tolerance trait; it does not contain the inserted DNA present in MON 89788. The reference materials were commercially available soybean varieties that were selected to represent a diversity of the commercially cultivated soybean and varied by study (Table S1). They provided a range of background values common to commercial soybean for the characteristics assessed.

Field Observations

Phenotypic Study 1

Trials were established at each of 17 US sites during 2005 in a randomized complete block design with 3 replications (trial and site information is reported in Table S2). Agronomic practices used to prepare and maintain each study site were characteristic of those used in each respective geographic region. Herbicides containing glyphosate were not used in this study to avoid injury to the conventional control and reference plants and to ensure all plots were managed uniformly.

Ecological interactions were assessed at all 17 sites through the qualitative evaluation of plant response to abiotic stress, disease damage, and arthropod damage data (Table 1). At each study site and observation time, each plot was evaluated for the severity of symptoms caused by abiotic stressors, diseases, and arthropods that were commonly observed at the site. The ecological stressors evaluated were not artificially induced and varied between sites. Plots were rated on a 0–9 scale (0 = none, 1 to 3 = “slight,” 4 to 6 = “moderate,” and 7 to 9 = “severe”) but the results were reported by category (none, slight, moderate, or severe). Similar qualitative assessments have been previously reported (Horak et al., 2007) and are useful in overall weight-of-evidence arguments to support the risk assessment.

At three sites (IL1, MO1, and MO2; see Table S2), arthropod abundance was quantitatively evaluated 3 times during the growing season using a beat-sheet sampling method. The beat sheet consisted of a white sheet measuring approximately 100 cm × 75 cm with a central opening to which a container lid had been glued. The attached lid had a hole in the middle to allow insects to pass through. Prior to insect collection, an empty container was attached to the lid. The beat sheet was placed flat on the ground between 2 rows and plants from both rows adjacent to the beat sheet were shaken vigorously. Dislodged insects that fell onto the beat sheet were brushed toward the center, into the container. The container was removed from the beat sheet, filled with enough alcohol to cover the insects and plant debris, and sealed with a solid lid. Two insect subsamples were collected from non-systematically selected plants in each plot, one from rows 5 and 6 and the other from rows 6 and 7. The 2 subsamples from each plot were combined into a single container, which was shipped to the University of Arkansas Department of Entomology for processing.

To focus the insect evaluation on the most abundant pest and beneficial species, the following predetermined selection criteria were employed. A list of important Midwestern pest and beneficial species was provided to the University of Arkansas. Four randomly selected samples from each collection time point at each site were examined to determine the 5 most abundant pest species and the 3 most abundant beneficial species from the list. These 8 species were then counted in each sample from each plot. Because the species counted were site- and collection-specific, they varied from site to site and from collection to collection.

Phenotypic Study 2

Trials were established at each of 5 U.S. sites during 2006 in a randomized complete block design with 3 replications (trial and site information are reported in Table S2). Agronomic practices were characteristic of those used in each respective geographic region, and no herbicides containing glyphosate were used. Qualitative data were collected on plant injury caused by arthropods, diseases, and abiotic stressors (Table 1), as described above for Phenotypic Study 1. The observed ecological stressors were not artificially induced and varied between sites.

Arthropod abundance data were collected at 3 of the 5 sites (IA, IN, and NE; see Table S2). A beat sheet was used for collection as described above for Phenotypic Study 1. Four subsamples were taken from the fifth, sixth, and seventh rows from each plot at each collection and combined into one individual sample per plot. A maximum of 5 each of the most abundant nontarget pest and beneficial arthropod taxa were determined for each collection interval at each site, except that 6 beneficial insect taxa were counted at the NE site.

Phenotypic Study 3

Trials were established in late 2004 or early 2005 at each of 5 sites in Argentina in a randomized complete block design with 3 replications (trial and site information are reported in Table S2). Agronomic practices were characteristic of those used in each respective geographic region, and no herbicides containing glyphosate were used. Qualitative data were collected on differential response to naturally occurring ecological stressors during the growing season, as described in Table 1. The observed ecological stressors were not artificially induced and varied between sites.

Cold Tolerance

Three seeds were planted per pot in each of 30 pots of MON 89788, the control, and reference materials. Plants were grown in a greenhouse with a 14-h photoperiod with a target daytime temperature of 30°C and a target night temperature of 24°C. At 7 d after planting (DAP), the pots were thinned to one seedling per pot; at 17 DAP, 20 pots of each material were selected based on uniformity in plant height, growth stage, and seedling vigor. Ten pots of each material were placed in one of 2 growth chambers and arranged in a randomized complete block design. One chamber had settings for optimal growth (29°C day, 24°C night) while the second chamber was used to expose plants to cold (suboptimal) temperatures (15°C day, 8°C night); the plants in both treatments were exposed to 14 h/d of light. Plants were grown in growth chambers for 20 d after which plant height, vegetative growth stage, vigor, fresh weight, and dry weight were determined as described in Table 1.

Volunteer Potential

Trials were established during 2005–2006 at 4 U.S. locations to assess the volunteer potential (the potential for seed to survive winter conditions and emerge and grow in the following crop growing season) of MON 89788 in major soybean-growing regions. The test, control, and reference starting seed (Table S1) were produced at the 4 sites (IL1, IN1, MO1, and NE; see Table S2). The viability of the starting seed was determined in the laboratory by conducting germination (warm) and vigor (cold) testing of each test, control, and reference starting seed lot and was acceptable for use in this trial.

The trials were established in a randomized complete block design with 3 replications. The total size of each individual plot was 5.3 m × 1.5 m. To avoid mixing of seed between adjacent plots during seed incorporation, a 4.1 m × 0.9 m confined planting area was established within each plot (0.6 m from each short edge and 0.3 m from each long edge). Each plot was hand-seeded by uniformly scattering approximately 400 seed on the soil surface within the confined plot area. Seed were then lightly incorporated with a disk in late 2005 or early 2006.

Agronomic practices used to prepare and maintain each study site were characteristic of each respective region. No irrigation was applied to the study areas and no plot management was performed after the seed were scattered in the plots unless needed. Herbicides labeled for use in conventional soybean were applied at the IL1 and IN1 sites during Spring 2006 to control weeds that could have interfered with volunteer plant emergence, observation, and counts. Volunteer plant counts commenced in Spring 2006 when environmental conditions became favorable for germination and emergence and ended by mid-June. The number of emerged soybean plants was recorded approximately every 2 wk for a total of 6 observations at each site.

Allelopathic Effects

The allelopathic effects of MON 89788, the control, and reference materials on the growth and development of lettuce (Lactuca sativa L. var. longifolia ‘Valley Heart’) were assessed by evaluating lettuce plants that were grown either in soil containing soybean shoot residue (shoot tissue evaluation) or in soil in which the soybeans were previously grown (soil medium evaluation). Lettuce is known to be sensitive to inhibitory chemicals and is frequently used in allelopathic studies (Fujii et al., 2003; Fujii et al., 2004).

Seed of MON 89788, the control, or reference materials (Table S1) were planted in an 18–33°C night/day (14-h day) greenhouse in pots arranged in a randomized complete block design with 6 replications. The pots were watered as needed and thinned to one soybean plant per pot 7 DAP. When the soybean plants reached an average growth stage of V6, the soybean plant in each pot was cut at the soil surface. The shoot tissue was chopped into approximately 2-cm segments, dried at approximately 40°C for 7 d, and then homogenized into a dry powder. The root and soil medium were separated and the roots were shaken into a bag to remove adhering soil. The root system was discarded and the remaining soil medium from each pot (including soil shaken from the roots) was transferred to a clean 15.2-cm diameter × 12.1-cm deep pot for use in the soil medium evaluation. For the shoot tissue evaluation, pots were filled with Redi-Earth soil medium that contained 3.6 kg/m³ of 14–14–14 Osmocote slow-release fertilizer. The ground shoot tissue of a single test, control, or reference soybean plant was thoroughly mixed into each pot (i.e., one shoot tissue sample per pot) at a concentration of approximately 0.6% (w/w).

For each of the allelopathy evaluations, pots (containing medium prepared as above) were arranged in a randomized complete block design in a 16–35°C night/day greenhouse, with 6 replications. Six lettuce seeds were uniformly planted in each pot at approximately 0.1 cm deep. After emergence, the pots were irrigated as needed via capillary mats. All pots from both evaluations were thinned to one lettuce plant per pot after emergence data were collected. The data collected from the lettuce plants in both evaluations are listed in Table 1.

Effect on Symbiont (Bradyrhizobium japonicum)

MON 89788, the control, and reference plants were produced from seed germinated in an environmental chamber in 2005 (Table S1). Germinated seedlings were then planted in pots containing nitrogen-free potting medium (LB2; Sun Gro Horticulture, Inc., Garland, TX) and grown in a greenhouse. Seedlings were inoculated with B. japonicum (approximately 1 × 10⁸ cells [Becker Underwood, Ames, IA] in phosphate-buffered saline) at planting, and then re-inoculated after plants emerged from the potting medium. The pots were arranged in a randomized split-block design with 8 replications. Nitrogen-free nutrient solution was added weekly after plants emerged from the potting medium. Nodule number, nodule dry weight (mg), root dry weight (mg), shoot dry weight (mg), and shoot total nitrogen (% dry weight; determined by Agvise Laboratories, Northwood, ND) were collected at each of the 2 sampling periods (4- and 6-wk; Table 1). Assessment of nodule number and mass along with plant growth and nitrogen status are commonly used to assess differences in the symbiotic association between a legume and its associated rhizobia (Israel et al., 1986).

Statistical Analysis

For each of the quantitative assessments of arthropod abundance, an analysis of variance was conducted according to a randomized complete block design using SAS² software (SAS Version 9.1.3). Because different insects were observed at different sites, only a by-site analysis was performed. Statistical significance was set at p ≤ 0.05. For Phenotypic Study 2, the study protocol specified that if the mean abundance count for a given insect, site, and collection time was <1.0 for either MON 89788 or A3244, a statistical comparison was not conducted. The minimum and maximum observed values of the references were reported but not statistically compared to the test or control.

Qualitative ecological observations (i.e., abiotic stressors, disease damage, and arthropod damage) were not statistically analyzed. Likewise, in some of the other studies described below, limited variability and small sample size or the categorical nature of the data precluded statistical analysis for some characteristics because analysis of variance assumptions could not be met. In these cases, if the values obtained were numerically the same or the categories overlapped, they were considered not different. If the values were different or the categories did not overlap, they were considered different and assessed further. Although appropriate caution should be used when assessing for differences from qualitative data, the information is useful as an indicator of a change in the plant and is considered in the overall weight of evidence for the risk assessment.

For the cold tolerance study, analysis of variance was conducted according to a randomized complete block design with 10 replications using SAS software (SAS Version 9.1.3). The level of statistical significance was predetermined to be 5% (p ≤ 0.05). Soybean developmental stages and seedling vigor ratings were not submitted for statistical analysis because the limited variability of the observed categorical levels for these responses would not allow analysis of variance assumptions to be met. No statistical comparisons were made between the test and reference substances.

For the allelopathy study, analysis of variance was conducted according to a randomized complete block design with 6 replications using SAS Version 9.1.3. The level of statistical significance was predetermined to be 5% (p ≤ 0.05). Lettuce emergence and growth stage data were not subjected to an analysis of variance because the small sample size and limited variability in the observed categorical levels for these responses would not allow analysis of variance assumptions to be met. No statistical comparisons were made between the test and reference substances. The minimum and maximum mean values (reference range) were determined from the reference means.

For the volunteer study, analysis of variance was conducted according to a randomized complete block design with 3 replications using SAS Version 9.1. The level of statistical significance was 5% (p ≤ 0.05). Low numbers of volunteer plants and low variability at 3 of the 4 sites (IL1, IN1, and MO1) precluded individual analysis of those sites and combined-site analysis because analysis of variance assumptions were not met. No statistical comparisons were made between the test and reference substances. The minimum and maximum mean values (reference range) were determined from the reference means.

For the symbiont study, an analysis of variance was conducted using a randomized split-block design with 8 replications for each test, control, and reference material at each sampling period. Data were analyzed using SAS Version 9.1, with the level of statistical significance predetermined to be 5% (p ≤ 0.05). The means of the test and control substances were compared to each other. Minimum and maximum values (reference range) were determined for the 3 reference substances. No statistical comparisons were made between test and reference substances.

Data Interpretation for Ecological Risk Assessment

Figure 1 in Horak et al. (2015) outlines the stepwise process used to determine whether a statistically significant difference between MON 89788 and A3244 for any of the characteristics assessed was potentially adverse in terms of pest potential or ecological impact. A “No” answer at any step indicated that the difference did not contribute to a biological or ecological concern for MON 89788, and subsequent steps were not considered. A detailed explanation of this decision method as applied to MON 89788 is given in Horak et al. (2015).

DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST

The authors are present or former employees of Monsanto Company, which develops and sells genetically modified crop products.

Acknowledgments

The authors acknowledge the assistance of Virginia M. Peschke of Oakside Editorial Services in the preparation of this paper and thank Aqeel Ahmad for his contributions to the Discussion. The authors also recognize the contributions of Tim Perez for statistical consultation and Tom Armstrong, S. Douglas Prosch, Jim Colyer, and Yovita Sutanto for project support.

FUNDING

This research was supported by Monsanto Company.

SUPPLEMENTAL MATERIAL

Supplemental data for this article can be accessed on the publisher's website.

Notes

1. Roundup Ready 2 Yield®, Roundup Ready®, Roundup Ready® Flex, and Roundup® are registered trademarks of Monsanto Technology, LLC, St. Louis, MO.

2. SAS® is a registered trademark of SAS Institute, Inc., Cary, NC.