Feral rice from introgression of weedy rice genes into transgenic herbicide-resistant hybrid-rice progeny

Feral rice from introgression of weedy rice genes into transgenic herbicide-resistant hybrid-rice... Abstract Pollen-mediated transgenic flow of herbicide resistance occurs bidirectionally between transgenic cultivated rice and weedy rice. The potential risk of weedy traits introgressing into hybrid rice has been underestimated and is poorly understood. In this study, two glufosinate-resistant transgenic rice varieties, hybrid rice (F1), and their succeeding generations (F2–F4) were planted for 3 years in field plots free of weedy rice adjacent to experimental weedy-rice fields. Weedy-rice-like (feral) plants that were both glufosinate-resistant and had red-pericarp seed were initially found only among the F3 generations of the two glufosinate-resistant transgenic hybrid cultivars. The composite fitness (an index based on eight productivity and weediness traits) of the feral progeny was significantly higher than that of the glufosinate-resistant transgenic hybrid (the original female parent of the feral progeny) under monoculture common garden conditions. The hybrid rice progeny segregated into individuals of variable height and extended flowering. The hybrid rice F2 generations had higher outcrossing rates by pollen reception (0.96–1.65%) than their progenitors (0.07–0.98%). The results show that herbicide-resistant weedy rice can rapidly arise by pollen-mediated gene flow from weedy to transgenic hybrid rice, and their segregating pollen-receptive progeny pose a greater agro-ecological risk than transgenic varieties. The safety assessment and management regulations for transgenic hybrid rice should take into account the risk of bidirectional gene flow. Ecological risk, gene flow, herbicide-resistance, hybrid rice, transgenic rice, weedy rice Introduction Herbicide resistance has been the dominant trait of transgenic crops. The global area of transgenic crops had reached 185.1 million ha by 2016, of which over 80% were herbicide-resistant, including those with stacked traits (ISAAA, 2016). Most herbicide-resistant transgenic crops are transformed to withstand otherwise non-selective herbicides. These chemicals have a wide spectrum of weed control and thus kill the vast majority of plants with the exception of the resistant transgenic crop. The evolution of herbicide-resistant weeds caused by intense selection pressure from recurrent use of a single herbicide in transgenic crops has become widespread (Beckie and Hall, 2014; Busconi et al., 2014; Bonny, 2016). In addition, when weedy relatives infest transgenic crops, transgene flow can transfer the resistant trait to them and thus aggravate the weed problem (Lu and Yang, 2009). Therefore, the safe and sustainable use of transgenic technology depends on the implementation of proper prevention and mitigation of gene flow based on a thorough agro-ecological risk assessment of transgene escape (Gressel and Valverde, 2009; Rong et al., 2012). Weedy rice (Oryza sativa), conspecific to cultivated rice (also O. sativa), is one of the most serious weeds in paddy fields worldwide (Valverde, 2005; Delouche et al., 2007). It is frequently closely related to its companion cultivated rice, as it has been detected in morphological and genetic studies (Ishikawa et al., 2005; Akasaka et al., 2009; Prathepha, 2009; Gross et al., 2010; Li et al., 2017), suggesting a possible origin of weedy rice as a de-domesticated (feral) form. Although rice is predominantly self-pollinated, bidirectional gene flow between the cultivated and weedy forms occurs, although at a relatively low rate of usually less than 1% (Messeguer et al., 2004; Gealy, 2005; Rong et al., 2007; Shivrain et al., 2007). However, this rate is of practical relevance, as observed with the commercial cultivation of imidazolinone (IMI)-herbicide-resistant (ClearfieldTM) rice varieties that have been responsible for the quick introgression of the IMI-resistance genes into weedy rice populations wherever this rice is grown (Zhang et al., 2006; Gressel and Valverde, 2009; Valverde, 2013; Sudianto et al., 2013). Moreover, despite some cross-incompatibility between the two major subspecies of rice (O. sativa ssp. indica and O. sativa ssp. japonica), introgression of genes for enhanced traits does occur in the field, as observed for the transfer of IMI resistance from tropical japonica rice to indica weedy rice in the USA and elsewhere (Craig et al., 2014). Most research on gene flow has focused on the transfer of transgenes from transgenic rice to weedy rice (the ‘forward’ direction) (Gealy, 2005; Song et al., 2009), because of the concerns about the escape and persistence of such transgenes in weedy rice populations through seed-shattering and dormancy. Under properly managed commercial production, the density of cultivated rice greatly surpasses that of weedy rice even at high infestations (Valverde, 2005). However, ‘reverse’ gene flow (from weedy to cultivated rice), will also result in herbicide-resistant weedy individuals (Serrat et al., 2013). Therefore, gene flow between transgenic and weedy rice in any direction may lead to similar results (Olsen et al., 2007). In addition, it is known that the crop itself has the potential to back-mutate into weedy types, and thus many types of weedy rice are closely related to the varieties they associate with (Ammann et al., 2000; Dale et al., 2002; Zhang et al., 2012; Sun et al., 2013). Hence, transgenes are likely to escape from transgenic rice by the cultivated form reverting to a weed type. A detailed study based on cytoplasmic maternal inheritance has recently demonstrated that some weedy rice accessions derive from cultivated hybrid rice as an initial female parent (Zhang et al., 2015). This suggests that the herbicide-resistant transgenic rice itself also has the potential of being pollinated by weedy rice, resulting in herbicide-resistant weedy progeny. The risk of a herbicide-resistance transgene escape depends on the actual gene flow frequency between the transgenic and weedy rice or the reversion to weediness of a transgenic cultivar, and on the ability of the new herbicide-resistant progeny to persist. Gene flow from transgenic herbicide-resistant rice to weedy rice usually results in progeny at least equally as fit as their weedy rice parents (Cao et al., 2009; Song et al., 2011). Similarly, when cultivated rice serves as the female parent in a cross with weedy rice, the progeny is also as fit as the respective weedy rice parent (Shivrain et al., 2009a). However, the risk of reverse (weed to crop) gene flow has been seriously underestimated and even ignored completely because it was thought that the non-shattering hybrid seed would remain on the spike and be removed together with the crop grain. Hence, limited research has addressed this issue (Serrat et al., 2013). However, a considerable amount of rice seed can fall onto the soil during harvest (Gealy, 2005; Valverde, 2005), increasing the impact of reverse gene flow. This needs to be recognised and integrated into the assessment systems of potential agro-ecological risk of transgenic hybrid rice. Differences in the intrinsic genetic variability and growth characteristics of weedy rice biotypes also play a role in the outcome of bi-directional gene flow (Zuo et al., 2011). Particular attention should be paid to hybrid rice because of its heterozygous nature and its segregating progeny, easily observed by the variability in plant height, flowering time, and even pollen fertility (Yuan, 2002), all of which affect their outcrossing rate. These special characteristics, as well as protruding anthers (Gressel, 2002), may confer transgenic hybrid rice with a higher gene-flow risk and hence justify the need to impose stricter ecological safety evaluation and regulations. In this study, we established a 3-year experiment using transgenic rice varieties and hybrids to explore the likelihood and the frequency of these rice cultivars turning into herbicide-resistant weeds through receiving pollen from weedy rice. We hypothesise that, as previously proposed (Gressel, 2002), the comparatively higher outcrossing rate of hybrid rice to that of autogamous varieties may be a key factor for the introgression of weedy traits into the progeny of hybrid rice interacting with the weedy form in commercial fields. An additional gene flow experiment was conducted to indentify descriptors for important weediness traits. Our findings suggest that there is a need to revise the transgenic rice safety assessments and management regulations that were originally established based on the premise of unidirectional gene flow, and to consider the relevance of gene flow from weeds to transgenic crops. Materials and methods Plant material Glufosinate-resistant transgenic rice varieties and hybrids together with their progeny when appropriate were used in the experiments. The transgenic varieties were Bar68-1, 86B, and T1C-19, all homozygous for the bar transgene (Tang et al., 2006; Cui et al., 2013; Sun et al., 2015). The transgenic hybrids were Xiang125s/Bar68-1 (abbreviated as X125B) (Sun et al., 2015) and II you 86B (IIU86B) (Cui et al., 2013). X125B is a two-line hybrid whose male parent is Bar68-1. IIU86B is a three-line hybrid whose male and female parents are 86B and a wild abortive (WA) male-sterile line, respectively. Both the hybrid rice cultivars are hemizygous, containing one copy each of the bar gene. The conventional hybrid rice cultivars used in this experiment included the two-line hybrid rice 6 Liang you 9368 (6LY) (China Rice Data Center, http://www.ricedata.cn/variety/) and its F2 generation (F2-6LY), and the three-line hybrid rice II you 98 (IIY) (China Rice Data Center) and its F2 generation (F2-IIY). The F2 generations are the offspring of the rice hybrids that by definition are F1s. Nine weedy rice accessions were also included and were collected in Liaoning (WR1, WR5), Jiangsu (WR3, WR4), Anhui (WR2), Hunan (WR6), Guangdong (WR7, WR9), and Guangxi (WR8) provinces and selected from the weedy rice germplasm bank of the Weed Research Laboratory at Nanjing Agricultural University (Supplementary Table S3 at JXB online). Experimental field conditions Experimental fields were located at the Jiangpu Transgenic Rice Experiment Base of the Weed Research Laboratory at Nanjing Agricultural University (32°0′34″N, 118°36′51″E). The area was fenced and surrounded by poplar trees (Supplementary Fig. S1). A ‘pollen inflow experiment’ was established in a field section devoid of weedy rice where only clean agricultural implements were allowed. The nine weedy rice accessions were grown in paddy fields (two fields totaling 3000 m2 and one field of 1000 m2 in 2011 and 2012, respectively) situated east and north-east of the pollen inflow experiment field that represent the sole source of weedy rice pollen (Supplementary Fig. S1). A separate gene flow experiment was established in an isolated field nearby. Pollen inflow experiment The experiment consisted of six plots (Supplementary Fig. S2A), four of which were 70 m2 each (A1, A2, B1, and B2) together with two larger ones of 210 m2 each (A3 and B3). Different generations of the transgenic rice cultivars 86B, Bar68-1, IIU86B, and X125B were grown in the small plots between 2011 and 2013 (Supplementary Fig. S2B). The larger plots were used to increase the populations of segregating progeny of the IIU86B and X125B transgenic hybrids. All plots were inspected for the presence of weedy-rice-like (feral) plants with red-pericarp seed at maturity. The plots were established by direct water-seeding of broadcast pre-germinated seed at 10 g m−2 (dry seed weight) and sprayed each year with 450 g active ingredient (a.i.) ha−1 glufosinate ammonium (Basta, Bayer AG) at 40 d after seeding. Flowering time and plant height In each of the six plots in each of the 3 years 200 plants were randomly selected to evaluate flowering time (recorded as the period from first to last plant to begin flowering) and plant height at maturity. The flowering time obtained by this descriptor only reflects the period during which individuals began flowering and not the length of flowering, and hence was used only to compare flowering synchrony among the populations. Measurement of emergence rate of crop volunteers After harvest, seed obtained from the rice crop in each plot was broadcast onto the soil surface (20 g m−2) and allowed to overwinter without ploughing. In the following year, irrigation was provided at the beginning of May to stimulate germination. Emerged crop volunteers (from the broadcast seed and the naturally shed seed) were counted in the entire plot at the beginning of June and then removed from the field. Emergence of volunteers was expressed as density per plot. Screening and identification of feral plants All rice individuals in every plot were carefully inspected at maturity in each of the 3 years. Feral plants and their panicles with mature seed possessing red pericarps were collected and stored at low temperature (5–15 °C) until further use. Young leaves of these feral plants were collected for total DNA extraction. The molecular marker RID14 (forward: 5′-TCCAGGCACC ACACAGAGA-3′; reverse: 5′-GGCACTGAAATCACCTTGG-3′) based on the red-pericarp Rc gene (Yang et al., 2009) was used to genetically confirm the red pericarp phenotype, and to determine whether plants were homozygous or heterozygous for the Rc gene. The marker orWA352 (forward: 5′-GTTGATGGGTATGGATAGAG-3′; reverse: 5′-CGCAGGGCCTCGGTATATCTA-3′) based on the wild abortive cytoplasm male-sterility WA352 gene (Luo et al., 2013) was used to determine whether the female parent of feral plants was hybrid rice. The marker bar (forward: 5′-GCACCATCGTCAACCACTAC-3′; reverse: 5′-GCCAGAAACCCACGTCAT-3′) based on the glufosinate-resistance bar gene was used to detect the presence of the bar gene in the feral individuals. The PCR products of the marker RID14 and those of orWA352 and bar were separated by 6% and 1.5% polyacrylamide gel electrophoresis, respectively. Evaluation of weediness traits in feral individuals The seed produced by each feral individual was considered as a feral group. Seed of feral plants collected in 2012 from the F3 generations of transgenic two- and three-line hybrid rice in the A3 (33 feral groups) and B3 (27 feral groups) plots, and those of X125B, IIU86B (female parents of feral plants), and the nine weedy rice accessions (WR1–WR9, potential male parent of feral plants) were sown in a nursery. When the plants reached the six-leaf growth stage they were transplanted into a common garden at a distance of 25 cm within rows and 30 cm between rows. Glufosinate ammonium (450 g a.i. ha−1) was applied 15 d after transplanting to verify the expression of glufosinate resistance in the feral plants and to control weeds. Uncontrolled weeds and those emerging in plots planted with weedy rice accessions, where the herbicide was not applied, were eliminated manually. Eight traits related to weediness were measured at the reproductive stage: plant height, tiller number per plant, estimated flag leaf area (calculated as the product of leaf length × width), panicle length, spikelet number per panicle, seed-set rate, seed-shattering rate, and 100-seed weight. Ten plants of each feral group, X125B, IIU86B, and the nine weedy rice accessions were randomly selected for the measurements. The relative fitness of feral groups compared to that of their initial female crop parent and potential weedy male parents (the nine weedy rice accessions) was calculated according to Song et al., (2004) and Song et al., (2011). Briefly, the mean value for each trait measured in the female parent was assigned a value of ‘1.0’. The ratio with the mean value of the trait in each of the corresponding feral groups was then calculated. Similarly, the relative fitness of each feral group in relation to each of its potential (weedy rice) male parents was calculated, this time assigning the value of ‘1.0’ to the mean value for the trait measured in the corresponding weedy accession. For each individual comparison, a value above 1.0 indicated enhanced performance of the feral group. Finally, a composite fitness value was assigned to each feral group (in relation to each of its female or potential male parents) as the mean value of the eight relative fitness values. Gene flow-experiment Seedlings of transgenic varieties and conventional hybrids at the six-leaf growth stage were transplanted at a distance of 25 cm in rows separated by 30 cm (Supplementary Fig. S3). The transgenic rice varieties 86B, Bar68-1, and T1C-19 served as pollen donors while hybrid rice 6LY and IIY and their offspring (F2 generations designated as F2-6LY and F2-IIY, respectively) were used as pollen receptors (12 gene-flow combinations). Each combination was repeated four times; there were 60 plants of each pollen receptor in each replicate. The length of flowering (the duration for which flowers were open) of each pollen donor and receptor was recorded to calculate the overlapping flowering time for each gene-flow combination. The height of 10 randomly selected plants of each type was measured at flowering. The male sterility of plants in the F2-6LY and F2-IIY populations in each gene-flow combination was confirmed by the standard pollen I2-KI staining method. A single panicle from each male-sterile plant was selected, and five panicles were collected in each combination at maturity for detection of glufosinate-resistant seed. Seeds of pollen receptors in each combination were harvested at maturity, and at least 10% of them were used to verify glufosinate resistance (Song et al., 2011) for calculation of the rate of gene flow. Statistical analysis ANOVA was used to determine the significance of differences in plant height and outcrossing rates among pollen donors and receptors in the gene-flow experiment, as well as the significance of differences in volunteer density among the types used in the ‘pollen inflow experiment’ in each year. Independent t-tests were used for all comparisons involving pairs of means, including for plant height and seed-shattering rate comparisons between the feral progeny and their initial female parent, the outcrossing-rate comparisons between the two hybrid rice generations in the gene-flow experiment, and the relative fitness comparisons between the feral progeny and their original female parent and original potential male parent. All statistical analyses were carried out using PASW Statistics version 18.0 for Windows (IBM Inc., New York, USA). Results Herbicide-resistant feral plants are found in the F3 progeny of field-grown transgenic hybrid rice Weedy-rice-like (feral) plants with red pericarps were identified by inspecting every rice plant in the experimental paddy field. None was found associated with the transgenic varieties or with the transgenic hybrids (F1) themselves or the progeny (F2) of the hybrids in the first year of the experiment (2011). Distinctive feral plants appeared in some plots in the following two years (Fig. 1A). In the second year (2012), 33 and 27 feral plants were found exclusively among the hybrid rice generations F3-IIU86B and F3-X125B growing in Plots A3 and B3, respectively (Fig. 1B). In the third year, in addition to 20 and nine feral plants found in Plots A3 and B3 planted with F4-IIU86B and F4-X125B, respectively, four and one feral plants were also found in Plots A2 and B2, which were planted with F3-IIU86B and F3-X125B, respectively (Fig. 1C). Interestingly, no feral plants were ever found throughout the 3-year experiment in Plots A1 or B1, which were respectively sown with saved seed from the cultivars 86B and Bar68-1 harvested the previous season. Thus, feral plants with red pericarps began appearing only in the F3 generations of transgenic hybrid rice. Fig. 1. View largeDownload slide Occurrence and frequency of weedy-rice-like (feral) plants and their identification. (A) Feral plants with red pericarps. (B, C), Experimental plots in which feral plants were found (highlighted in red) in 2012 (B) and in 2013 (C) together with the number of feral individuals found in each plot (underlined); the identity of the plant material grown in each plot is indicated. (D) Marker RID14 amplifies a 154-bp fragment from weedy rice homozygous for the Rc gene (lane 1) and a 140-bp fragment from cultivated rice homozygous for the rc gene (lane 2). All the feral plants (lanes 3–12) were heterozygous (Rc/rc) for the red pericarp gene except for one individual of F4-IIU86B that was homozygous (lane not shown). M represents the DL500 DNA Marker. (E) The bar marker only amplifies a 429-bp fragment from rice containing the glufosinate-resistance bar gene (lane 1) that is absent in weedy rice (lane2). All feral plants contained the bar gene (lanes 3–12). M represents the DL500 DNA Marker. (F, G) The marker orWR352 only amplifies a 1432-bp fragment from rice containing the wild abortive (WA) cytoplasmic male-sterility gene WA352 (lane 1) that is absent in a reference male-fertile restorer line, Zhenshan 97B (lane 2). The feral plants found in Plots A2 and A3 contained the WA352 gene (F, lanes 3–12) but it was absent in all feral plants found in Plots B2 and B3 (G, lanes 3–12). M represents the DL2000 DNA Marker. Fig. 1. View largeDownload slide Occurrence and frequency of weedy-rice-like (feral) plants and their identification. (A) Feral plants with red pericarps. (B, C), Experimental plots in which feral plants were found (highlighted in red) in 2012 (B) and in 2013 (C) together with the number of feral individuals found in each plot (underlined); the identity of the plant material grown in each plot is indicated. (D) Marker RID14 amplifies a 154-bp fragment from weedy rice homozygous for the Rc gene (lane 1) and a 140-bp fragment from cultivated rice homozygous for the rc gene (lane 2). All the feral plants (lanes 3–12) were heterozygous (Rc/rc) for the red pericarp gene except for one individual of F4-IIU86B that was homozygous (lane not shown). M represents the DL500 DNA Marker. (E) The bar marker only amplifies a 429-bp fragment from rice containing the glufosinate-resistance bar gene (lane 1) that is absent in weedy rice (lane2). All feral plants contained the bar gene (lanes 3–12). M represents the DL500 DNA Marker. (F, G) The marker orWR352 only amplifies a 1432-bp fragment from rice containing the wild abortive (WA) cytoplasmic male-sterility gene WA352 (lane 1) that is absent in a reference male-fertile restorer line, Zhenshan 97B (lane 2). The feral plants found in Plots A2 and A3 contained the WA352 gene (F, lanes 3–12) but it was absent in all feral plants found in Plots B2 and B3 (G, lanes 3–12). M represents the DL2000 DNA Marker. The red pericarp genotype was confirmed by detecting the Rc gene through a molecular marker test. All feral plants were heterozygous for the Rc gene (Fig. 1D), except for one feral individual of F4-IIU86B that was homozygous. Additionally, a separate molecular marker test confirmed that all feral plants had the glufosinate-resistance bar gene (Fig. 1E). All feral plants found in Plots A2 and A3 (that had been initially planted with hybrid rice bred using a WA male-sterile line) contained the WA cytoplasmic male-sterility gene WA352, but it was absent in all feral plants found in Plots B2 and B3 (Fig. 1F) that were initially planted with the two-line hybrid rice X125B (in whose production no WA male-sterile line was involved). Thus, it is highly likely that these feral plants were directly derived from transgenic hybrid rice within the same plot. Fitness traits engender weediness in feral progeny Although a red pericarp is a common characteristic among weedy rice accessions, the most agronomically relevant traits of weedy rice are its competitive ability and shattering of differentially dormant seed. The calculated composite fitness of each progeny group based on eight characteristics (vegetative: plant height, tiller number, estimated flag leaf area; and reproductive: panicle length, spikelet number, seed set rate, shattering rate, 100-seed weight) measured under common garden conditions indicated that most feral groups had higher relative fitness than their original female parents (transgenic hybrid rice). Many of the feral groups even had higher relative fitness than their weedy rice parents, i.e. the nine weedy rice accessions (Fig. 2). Feral plants consistently had greater relative fitness for plant height (i.e. they were taller) than their crop female parent or any of the potential weedy rice male parents (Supplementary Fig. S4), except those designated as WR7 and WR8 (Supplementary Tables S1, S2). Likewise, they also had greater relative fitness for leaf area except for that of the potential WR4 and WR6 male parents. All feral plants bore fewer tillers than their known female or their putative male parents (Supplementary Tables S1, S2). In general, feral progeny were vegetatively superior to their female crop progenitor and equivalent or superior to their potential weedy rice parents. In relation to reproductive traits, feral progeny developed equivalent or longer panicles, equivalent or more spikelets, but set less seed than their parents (crop or potential weedy). The feral progeny produced lighter seed than their female parents, but equivalent or heavier than that of their weedy rice parents (Supplementary Tables S1, S2). The seed of the feral progeny shattered substantially more than their female parents (Supplementary Fig. S4), but less than their weedy rice parents (Supplementary Tables S1, S2). Fig. 2. View largeDownload slide Composite fitness of weedy-rice-like (feral) progeny groups compared with their corresponding parents. (A) Composite fitness of 33 feral progeny groups compared with the original female parent (IIU86B) and original potential male parents (nine weedy rice accessions, WR1–WR9), obtained from plants grown in Plot-A3 in 2012. Most progeny had higher fitness than their original female parent or each of their original potential male parents. (B) Composite fitness of 27 feral progeny groups compared with the original female parent (X125B) and original potential male parents (nine weedy rice accessions, WR1–WR9), obtained from plants grown in Plot-B3 in 2012. All progeny also had higher fitness than their corresponding original female parent and most of the original potential male parents. In (A, B) the bars represent numbers of feral groups (left y-axis) and each dot represents the relative fitness of one feral group (right y-axis). The data are divided into relative fitness values greater than 1 (red) and smaller than 1 (black). Fig. 2. View largeDownload slide Composite fitness of weedy-rice-like (feral) progeny groups compared with their corresponding parents. (A) Composite fitness of 33 feral progeny groups compared with the original female parent (IIU86B) and original potential male parents (nine weedy rice accessions, WR1–WR9), obtained from plants grown in Plot-A3 in 2012. Most progeny had higher fitness than their original female parent or each of their original potential male parents. (B) Composite fitness of 27 feral progeny groups compared with the original female parent (X125B) and original potential male parents (nine weedy rice accessions, WR1–WR9), obtained from plants grown in Plot-B3 in 2012. All progeny also had higher fitness than their corresponding original female parent and most of the original potential male parents. In (A, B) the bars represent numbers of feral groups (left y-axis) and each dot represents the relative fitness of one feral group (right y-axis). The data are divided into relative fitness values greater than 1 (red) and smaller than 1 (black). The high relative (both vegetative and reproductive) and composite fitness, weediness traits, and glufosinate-resistance of the feral plants suggest that they could be potentially harmful and difficult to control in paddy fields, particularly if glufosinate-resistant transgenic rice is grown sequentially. Hybrid rice segregating progeny have extended flowering time and height diversity Plant height and flowering duration were recorded for every generation in the experiments, including for the transgenic rice varieties 86B and Bar68-1 and the transgenic hybrids X125B and IIU86B and their F2–F4 generation progeny, which included the feral plants whose characteristics have been described in the previous section. The transgenic cultivars had uniform and narrow flowering duration (8–14 d) and plant height (height difference between 15 to 21 cm). The segregating progeny (F2–F4 generations) of transgenic hybrid IIU86B began flowering earlier than their progenitor, and that of X125B initiated flowering at the same time or earlier than their X125B female parent. All progeny had a substantially extended initiation of flowering of to up to 36 d. Their range of values for plant height was also widened, with the difference between the shortest and tallest plant being 86 cm (Fig. 3). Fig. 3. View largeDownload slide The progeny of transgenic hybrid rice were highly variable in flowering time and plant height because of character segregation. (A) The segregating progeny of hybrid rice included early maturing individuals that were already ripening their seed while others were just flowering. The images were taken on the same day. (B, C) The transformed rice materials 86B, Bar68-1, IIU86B, and X125B had similar flowering time durations and ranges of plant height (indicated by the bars). The segregating progeny (F2–F4) of the transgenic hybrids IIU86B and X125B had extended flowering periods and diverse plant heights. The line graphs inside the bars represent the distribution of individuals within the respective ranges of flowering time and height. Fig. 3. View largeDownload slide The progeny of transgenic hybrid rice were highly variable in flowering time and plant height because of character segregation. (A) The segregating progeny of hybrid rice included early maturing individuals that were already ripening their seed while others were just flowering. The images were taken on the same day. (B, C) The transformed rice materials 86B, Bar68-1, IIU86B, and X125B had similar flowering time durations and ranges of plant height (indicated by the bars). The segregating progeny (F2–F4) of the transgenic hybrids IIU86B and X125B had extended flowering periods and diverse plant heights. The line graphs inside the bars represent the distribution of individuals within the respective ranges of flowering time and height. Segregating progeny of hybrid rice have higher outcrossing rates than hybrid rice The outcrossing rates of two non-transgenic commercial hybrid rice cultivars (6LY and IIY) and their respective offspring (F2 generation, F2-6LY and F2-IIY), serving as pollen receptors, were compared in a separate gene-flow experiment. Every pollen receptor was planted together with three varieties of transgenic rice (Bar68-1, TIC-19, and 86B) as pollen donors. In all combinations, the outcrossing rates of both F2 generations were significantly higher (0.96–1.65%) than those of their respective parental hybrid rice (0.07–0.98%) (Fig. 4A, B). Correspondingly, both F2 generations had longer overlapping times of flowering with each pollen donor than the two hybrid rice cultivars, which resulted from their longer flowering periods (Fig. 4D). Fig. 4. View largeDownload slide Gene flow between hybrid rice cultivars (F1, F2 generations) as pollen receptors and transgenic rice varieties as pollen donors. (A, B) The outcrossing rate (top) and the overlapping of flowering time (bottom) of pollen donors and pollen receptors. In (A) the black and red bars represent the hybrid rice cultivar 6LY and F2-6LY (progeny of 6LY) as pollen receptors, respectively, with the transgenic rice varieties Bar68-1, T1C-19, and 86B as pollen donors. In (B) the black and red bars represent the hybrid rice cultivar IIY and F2-IIY (progeny of IIY) as pollen receptors, respectively, with the same transgenic rice varieties Bar68-1, T1C-19, and 86B as pollen donors. (C) Outcrossing rate (by pollen reception) of male-sterile plants found with F2-6LY (black) and F2-IIY (red). (D) Flowering stage (top) and plant height (bottom) of pollen donors (black) and pollen receptors (red). Data are means (±SD). Different letters indicate significant differences at P<0.05. Fig. 4. View largeDownload slide Gene flow between hybrid rice cultivars (F1, F2 generations) as pollen receptors and transgenic rice varieties as pollen donors. (A, B) The outcrossing rate (top) and the overlapping of flowering time (bottom) of pollen donors and pollen receptors. In (A) the black and red bars represent the hybrid rice cultivar 6LY and F2-6LY (progeny of 6LY) as pollen receptors, respectively, with the transgenic rice varieties Bar68-1, T1C-19, and 86B as pollen donors. In (B) the black and red bars represent the hybrid rice cultivar IIY and F2-IIY (progeny of IIY) as pollen receptors, respectively, with the same transgenic rice varieties Bar68-1, T1C-19, and 86B as pollen donors. (C) Outcrossing rate (by pollen reception) of male-sterile plants found with F2-6LY (black) and F2-IIY (red). (D) Flowering stage (top) and plant height (bottom) of pollen donors (black) and pollen receptors (red). Data are means (±SD). Different letters indicate significant differences at P<0.05. The outcrossing rate from transgenic rice pollen donors to male-sterile plants among two F2 generations (F2-6LY and F2-IIY) was also determined. Overall, the obligate cross-pollinated male-sterile plants had relatively higher outcrossing rates (8.8–22.4%) (Fig. 4C) compared with the two F2 generations (0.96–1.65%) (Fig. 4A, B). It is notably that the transgenic cultivar Bar68-1 was significantly shorter than the pollen receptors (Fig. 4D), probably making it an unsuccessful donor. In addition, Bar68-1 was at a further disadvantage as a pollen donor because of its earlier flowering and shorter flowering period that results, which resulted in a limited overlapping time of flowering with the pollen receptors (Fig. 4D). Consequently, the combinations in which 6LY and IIY were the pollen receptors had low outcrossing rates. However, the combinations in which the slightly shorter (than their respective female progenitors) F2-6LY and F2-IIY were the pollen receptors had relatively high outcrossing rates (Fig. 4A, B). In addition, the male-sterile plants among these two combinations had the highest outcrossing rates through cross-pollination (Fig. 4C). In a field situation, the presence of male-sterile plants among segregating progeny of hybrid rice flowering earlier and being surrounded by typically early-flowering weedy rice plants would be highly conducive for ‘reverse’ gene flow (i.e. from weed to crop volunteer). Under these conditions, weedy rice, and not the crop, would be the most likely source of pollen. Progeny of transgenic hybrid rice can potentially persist as volunteers Farmers do not save seed fromtheir harvested hybrid rice for subsequent sowings. Thus, the possibility of progeny of hybrid rice becoming established depends on volunteers. According to our quantitative observations from 2012 to 2014, all the rice materials that was planted produced volunteers in the following season (0.26–9.62 plants m−2) (Fig. 5). The density of volunteers emerging in a particular year depended on the lines planted. For example, in 2012 the F2 and F3 of the two-line transgenic hybrid rice X125B had more volunteers (9.62 and 5.58 plants m−2, respectively; P≤0.05) than the F2 and F3 of the three-line transgenic hybrid rice IIU86B (1.64 and 1.22 plants m−2, respectively). In the following two years, progeny (F3–F5) of X125B also consistently had more volunteers than the respective progeny of IIU86B. In general, the number of volunteers produced by the transformed varieties was much lower than most hybrid progeny (Fig. 5). Therefore, the ability to produce volunteers was influenced by the genetic background of the rice. The volunteer density also varied with cropping season (year) for each particular rice line. These varying densities, particularly of the transgenic rice varieties 86B (0.4, 0.26, and 4.46 plants m−2 in 2012, 2013, and 2014, respectively) and Bar68-1 (1.58, 0.36, and 5.2 plants m−2), strongly suggest that the occurrence of volunteers is influenced by environmental conditions. Fig. 5. View largeDownload slide The density of volunteer plants of different rice material cultivated in the pollen-inflow experiment from 2012 to 2014. Bar68-1 and 86B are transgenic varieties, IIU86B and X125B are transgenic hybrid rice; F2-, F3-, and F4- denote the first, second, and third offspring of the respective hybrid (F1). Data are means (±SD). Different letters indicate significant differences at P<0.05. Fig. 5. View largeDownload slide The density of volunteer plants of different rice material cultivated in the pollen-inflow experiment from 2012 to 2014. Bar68-1 and 86B are transgenic varieties, IIU86B and X125B are transgenic hybrid rice; F2-, F3-, and F4- denote the first, second, and third offspring of the respective hybrid (F1). Data are means (±SD). Different letters indicate significant differences at P<0.05. Discussion Herbicide-resistant weedy rice can originate through reverse gene flow Several generations of the transgenic rice varieties 86B and Bar68-1 and of the transgenic hybrid rice IIU86B and X125B were grown over three years in an experimental field that was under glufosinate selection pressure and free of weedy rice, but in close proximity to weedy rice growing in adjacent fields (Supplementary Figs S1A, S2). Weedy-rice-like (feral) individuals with red pericarps were initially found in plots (A3 and B3) as part of the F3 generation of two transgenic hybrid rice (F1) cultivars grown in 2012 (Fig. 1). Similarly, feral individuals also appeared in 2013 in plots (A2 and B2) that were planted with the F3 generations of the two transgenic hybrid rice cultivars. Feral plants were never found in plots (A1 and B1) planted with saved seed of two transgenic rice varieties even though these plots were placed side-by-side with the ones planted with hybrid rice. The feral individuals were glufosinate-resistant and carried the bar gene. In addition, the plasmagene WA352 contained in IIU86B was only detected in feral plants collected in plots (A2 and A3) that were planted with the F3 or F4 generations of IIU86B. Therefore, feral plants most likely directly derived from transgenic hybrid rice within each plot. In the majority of cases, the red pericarp character in rice is under the control of a dominant Rc gene, while the white pericarp character is under the control of the recessive rc gene that carries a 14-bp deletion in the seventh exon of the Rc gene (Sweeney et al., 2006; Furukawa et al., 2007). All transgenic rice cultivars (varieties and hybrids) used in this study possessed white pericarps and were homozygous for the rc gene. The feral individuals indentified by their red pericarps were heterozygous for the Rc gene, except for a single individual of the F4-IIU86B generation found in one of the plots that was homozygous for this trait. This individual was probably an offspring of a feral plant from the previous year. There are two known potential possibilities for the presence of the Rc gene in a population that originally lacks the gene: gene recombination and gene flow. It is unlikely that the rc gene would mutate into the Rc gene with such high frequency and only in a way that happened in specific generations of the transgenic hybrid rice. Although not being directly a back-mutation, red pericarps can arise in otherwise white-pericarp cultivated material by a second deletion in the Rc gene that restores the red color as a dominant trait (Brooks et al., 2008; Lee et al., 2009), but this does not seem plausible in our case because all feral plants carried the Rc gene (without the deletion). A possible explanation, for which we are not aware of any previous direct evidence, is that two white-pericarp rice varieties could produce a red-pericarp hybrid progeny through gene recombination; however, there are claims that indica×japonica hybrids may segregate individuals with red pericarps (Xiong et al., 2012). The possibility of gene recombination would also fail to explain why there were no red-pericarp individuals among the F2 generation of the transgenic hybrid rice. Therefore, the only plausible explanation is that the dominant allele in feral plants heterozygous for the Rc gene comes from gene flow, and that the pollen donor (with red pericarps) was the weedy rice growing in the adjacent fields. Traits conferring competitive ability and weediness to feral rice and their progeny will determine the success of the progeny as newly derived weedy rice biotypes. Progeny of hybrids between cultivated rice and weedy rice have always been found to be at least as fit as their parents (Cao et al., 2009; Shivrain et al., 2009a; Song et al., 2011). Our results were consistent with those of previous studies in that most feral groups had higher composite fitness indexes than their initial female parents (transgenic hybrid rice) (Fig. 2). The progeny of feral plants were taller and exhibited more seed shattering, which might confer a greater competitive advantage and a better opportunity to persist as weedy forms compared to their initial transgenic hybrid rice female parents (Supplementary Tables S1, S2). In addition, feral progeny could survive under glufosinate selection pressure together with glufosinate-resistant crop volunteers. Depending on their weedy attributes, they could persist and subsequently evolve into herbicide-resistant weedy rice biotypes. Clearly reverse gene flow from weedy rice to transgenic hybrid rice will increase the likelihood of the rapid evolution of herbicide-resistant weedy rice, especially under the selection pressure of the herbicide. This is a major agro-ecological risk that needs to be considered and mitigated otherwise it poses a threat to the sustainability of transgenic herbicide-resistant rice technology. Transgenic hybrid rice is more prone to transgene escape than transgenic varieties The fact that feral plants were found beginning with the F3 hybrid rice generation has important implications (Fig.1B). If such plants arose from reverse gene flow, then the F2 progeny and not the F1 (the transgenic hybrid rice itself) must had been the pollen receptor. Both generations were juxtaposed within the experimental plot arrangement, which later also included the F3 generations. This would suggest that the F2 generation had higher crossing ability by pollen receptivity than the cultivated hybrid rice. Furthermore, no feral individuals were among subsequent plantings of saved seed of the two transgenic varieties, also emphasising the apparent increased outcrossing rate of the F2 hybrid generation. In the gene-flow experiment, we indeed determined that the F2 generation of hybrid rice had a substantially higher outcrossing rate due to pollen reception than the cultivated hybrid itself. The outcrossing rate is determined by both biotic and abiotic factors (Gealy, 2005; Rong et al., 2007; Yuan et al., 2007; Shivrain et al., 2009b; Zuo et al., 2011). Abiotic factors include temperature, wind speed and direction, and air humidity, all of which should have been uniform across the plots given the scale of our experiment. Therefore, the increased outcrossing rate observed in the F2 offspring must have been associated with differential biological characteristics of this segregating progeny. First, individuals in the progeny of the hybrid rice had unsynchronised flowering times because of character segregation (Fig. 4). Thus, the F2 generation of hybrid rice had a flowering duration that was almost doubled compared to its hybrid rice parent in the gene-flow experiment, and it was further extended in the F3 and F4 generations (Fig. 3). The implication of this extension of the flowering period in comparison to both hybrid rice and conventional varieties is that the pollen-receptor function of the hybrid rice progeny would have overlapped for longer with the different pollen donors. The longer overlapping times in flowering were most likely the determining factor in the higher outcrossing rates of the hybrid rice progeny. In addition, transgenic hybrid rice could also be a pollen donor, in which case the extended flowering period would increase the risk of transgene escape. Quantifying this outcrossing route was not the objective of this current research, but it is something that needs to be considered, particularly in relation to dealing with the risk of movement and establishment (introgression) of herbicide-resistance transgenes to weedy rice, especially in the absence of mitigating measures (Gressel and Valverde, 2009). Second, among the segregating progeny of hybrid rice there were male-sterile individuals that had obligate cross-pollination, except for hybrid rice in whose production a gametophyte abortion-type sterile line had been used. In the gene-flow experiment, the male-sterile plants had a significantly higher outcrossing (by pollen reception) rate (8.8–22.4%) than the overall rate of the hybrid rice F2 generation (0.96–1.65%) (Fig. 4A, B). Consequently, the male-sterile individuals would be expected to have greatly contributed to the relatively high outcrossing rate of the hybrid rice F2 generation. Moreover, the greater range of plant heights of the hybrid rice progeny should also favor bidirectional outcrossing, except, of course, for male-sterile individuals. Therefore, the differential outcrossing rates between weedy rice and hybrid rice (both as F1 and F2 generations) together with conventional rice varieties should be addressed in more detail in future gene-flow studies. It is worth noting that even if the presence of male-sterile plants is beneficial for reverse gene flow, their success will depend on the ability of the offspring to predominantly self-pollinate and set seed. Fertility restoration is easy for male-sterile plants among two-line hybrid rice progeny, because almost any other rice material will possess dominant fertility genes that restore the photo-thermo-sensitive male-sterility gene. For male-sterile plants among three-line hybrid rice progeny, however, pollen with the required specific restoring gene is needed for fertility restoration. As expected, the overall average rate of seed set of feral groups whose initial female parent was X125B (two-line hybrid rice) was significantly higher than for progeny derived from IIU86B (three-line hybrid rice) as the initial female parent (Supplementary Fig. S5). In theory, the grain of transgenic varieties could be saved as seed for future planting but hybrid rice farmers, who understand the problems associated with segregation, purchase fresh commercial seed every cropping season. Thus, the risk of herbicide-resistant transgenes establishing in the field may actually be less with hybrid rice than with varietal material. In the pollen-inflow field experiment, each generation of transgenic hybrid rice was able to persist as volunteers in the following season (Fig. 5). The persistence of volunteers may be the first step towards plants becoming feral (Marambe, 2005). The occurrence of volunteers was influenced by the rice genetic background and environmental conditions. According to a recent study, hybrid rice would be expected to persist more as volunteers than conventional rice because of the higher overwintering capacity of its seed (Singh et al., 2016). If possible, transgenic hybrid rice material with lower ability to survive as volunteers should be preferentially selected for commercial use, especially in regions with warmer climates and under continuous rice cultivation that are more conducive for the establishment and persistence of crop volunteers. A paddy field devoid of weedy rice was used for the pollen-inflow experiment, and weedy rice was only present in adjacent fields. Even if the glufosinate is used to eliminate the weedy rice population in a paddy field planted with glufosinate-resistant rice, some weedy rice plants are likely to escape treatment (Busconi et al., 2014). Bidirectional gene flow will be more likely to occur under these circumstances because both the transgenic and weedy rice will be in close proximity. It should be noted that our experimental design with exceedingly high densities of transgenic hybrid rice progeny in experimental plots and weedy rice in the adjacent fields was highly conducive for obtaining feral plants. It is possible that the frequency of feral plants under a different design, or with different wind patterns at flowering, could vary, as has been observed in other gene-flow studies (Rong et al., 2012; Sun et al., 2015). In our case the wind patterns at flowering (Supplementary Fig. S1) clearly allowed reverse gene flow to occur, as clearly demonstrated by the phenotypic and genetic characterisation of the feral plants. It is undeniable that because of the increased outcrossing rate of its segregating progeny, transgenic hybrid rice should have a higher risk of transgene escape than transgenic rice varieties. This agro-ecological risk must not be neglected, particularly if herbicide-resistant hybrid rice that is either transgenic but without transgenic mitigation (Gressel and Valverde, 2009) or is conventionally bred becomes commercially successful and is planted over large areas, thus increasing the likelihood of greater areas being infested by herbicide-resistant weedy rice. Based on the results obtained in this study, it is evident that a more stringent management or stewardship program should be implemented for transgenic hybrid rice cultivation compared to transgenic varieties because of the higher agro-ecological risk associated with the higher outcrossing rate of the hybrid progeny. Special attention should be given to the management of weedy rice in adjacent fields to prevent reverse gene flow. Weedy rice should be properly controlled in fields and neighboring areas where transgenic hybrid rice seed is being produced because of the risk of a proportion of the male-sterile line being pollinated by weedy individuals. Weedy rice control with the herbicide to which the transgenic rice is resistant should be complemented with other practices that are properly integrated into the production system; for example, stale seedbed preparation, when feasible, to decrease the initial density of weedy and volunteer rice that can become established within the crop, particularly in direct-seeded fields at low latitude regions with suitable conditions for continuous rice production systems. It is advisable to initially plant transgenic herbicide-resistant varieties instead of hybrids in fields heavily infested with weedy rice, since varietal material poses a lower risk for gene flow than hybrid rice. In any case, it is best to have a proper rotation system in which varieties or hybrids with transgenic or conventionally bred resistance to herbicides with different modes of action are alternated. The rotation scheme can also be based on the deployment of hybrid rice that incorporates transgenic mitigation of gene flow (Gressel and Valverde, 2009). Finally, safety assessments and management regulations for transgenic hybrid rice should be further improved to take into account the risk of reverse gene flow. Supplementary data Supplementary data are available at JXB online. Table S1. Calculated relative fitness of the progeny of feral plants derived from the F3 generation of the transgenic herbicide-resistant three-line hybrid rice IIU86B (F1) compared to their original female parent and original potential male parents. Table S2. Calculated relative fitness of the progeny of feral plants derived from the F3 generation of the transgenic herbicide-resistant three-line hybrid rice X125B (F1) compared to their original female parent and original potential male parents. Table S3. Source of weedy rice materials used in this study. Fig. S1. Experimental location and surrounding environment. Fig. S2. Field experiment layout and cropping schedule in the pollen-inflow experiment. Fig. S3. Diagram of an individual experimental plot showing the planting layout in the gene-flow experiment. Fig. S4. Seed-shattering rate and plant height of feral progeny and their original female parent. Fig. S5. Mean seed-setting rate of feral groups. Acknowledgements This study was financially supported by the China Transgenic Organism Research and Commercialization Project (2016ZX08011-001), the Doctoral Program of Higher Education of the Ministry of Education (20130097130006), the 111 Project, and the National Natural Science Foundation of China (30800604). Rice materials were kindly provided by the Institute of Subtropical Agriculture of the Chinese Academy of Sciences (Bar68-1 and Xiang125s/Bar68-1), the South China Botanical Garden of the Chinese Academy of Sciences (86B and II you 86B), Huazhong Agricultural University (T1C-19), and Nanjing Shenzhou Seeds Industry Co., Ltd. (6 Liang you 9368 and II You 98). The authors have no conflicts of interest to declare. References Akasaka M , Ushiki J , Iwata H , Ishikawa R , Ishii T . 2009 . Genetic relationships and diversity of weedy rice (Oryza sativa L.) and cultivated rice varieties in Okayama Prefecture, Japan . Breeding Science 59 , 401 – 409 . Google Scholar CrossRef Search ADS Ammann K , Jacot Y , Mazyad PRA . 2000 . Weediness in the light of new transgenic crops and their potential hybrids . Journal of Plant Diseases and Protection 107 , 19 – 29 . Beckie HJ , Hall LM . 2014 . Genetically-modified herbicide-resistant (GMHR) crops a two-edged sword? An Americas perspective on development and effect on weed management . Crop Protection 66 , 40 – 45 . Google Scholar CrossRef Search ADS Bonny S . 2016 . Genetically modified herbicide-tolerant crops, weeds, and herbicides: overview and impact . Environmental Management 57 , 31 – 48 . Google Scholar CrossRef Search ADS PubMed Brooks SA , Yan W , Jackson AK , Deren CW . 2008 . A natural mutation in rc reverts white-rice-pericarp to red and results in a new, dominant, wild-type allele: Rc-g . Theoretical and Applied Genetics 117 , 575 – 580 . Google Scholar CrossRef Search ADS PubMed Busconi M , Baldi G , Lorenzoni C , Fogher C . 2014 . Gene flow from transgenic rice to red rice (Oryza sativa L.) in the field . Plant Biology 16 , 22 – 27 . Google Scholar CrossRef Search ADS PubMed Cao QJ , Xia H , Yang X , Lu BR . 2009 . Performance of hybrids between weedy rice and insect-resistant transgenic rice under field experiments: implication for environmental biosafety assessment . Journal of Integrative Plant Biology 51 , 1138 – 1148 . Google Scholar CrossRef Search ADS PubMed Craig SM , Reagon M , Resnick LE , Caicedo AL . 2014 . Allele distributions at hybrid incompatibility loci facilitate the potential for gene flow between cultivated and weedy rice in the US . PLoS ONE 9 , e86647 . Google Scholar CrossRef Search ADS PubMed Cui RR , Dai WM , Qiang S , Song XL . 2013 . Gene flow from transgenic glufosinate-resistant rice II you 86B and its restorer line 86B to weedy rice . Jiangsu Journal of Agriculture Sciences 29 , 708 – 714 (in Chinese). Dale PJ , Clarke B , Fontes EM . 2002 . Potential for the environmental impact of transgenic crops . Nature Biotechnology 20 , 567 – 574 . Google Scholar CrossRef Search ADS PubMed Delouche JC , Labrada R , Rosell C . 2007 . Weedy rices – origin, biology, ecology and control . Rome, Italy : Food and Agriculture Organization of the United Nations . Furukawa T , Maekawa M , Oki T , Suda I , Iida S , Shimada H , Takamure I , Kadowaki K . 2007 . The Rc and Rd genes are involved in proanthocyanidin synthesis in rice pericarp . The Plant Journal 49 , 91 – 102 . Google Scholar CrossRef Search ADS PubMed Gealy DR . 2005 . Gene movement between rice (Oryza sativa) and weedy rice (Oryza sativa): a U.S. temperate rice perspective . In: Gressel J , ed. Crop ferality and volunteerism . Boca Raton, FL : CRC Press , 323 – 354 . Google Scholar CrossRef Search ADS Gressel J . 2002 . Molecular biology of weed control , vol. 1 . Boca Raton, FL : CRC Press . Gressel J , Valverde BE . 2009 . A strategy to provide long-term control of weedy rice while mitigating herbicide resistance transgene flow, and its potential use for other crops with related weeds . Pest Management Science 65 , 723 – 731 . Google Scholar CrossRef Search ADS PubMed Gross BL , Reagon M , Hsu SC , Caicedo AL , Jia Y , Olsen KM . 2010 . Seeing red: the origin of grain pigmentation in US weedy rice . Molecular Ecology 19 , 3380 – 3393 . Google Scholar CrossRef Search ADS PubMed ISAAA . 2016 . Global status of commercialized biotech/GM Crops: 2016. ISAAA Brief No. 52 . Ithaca, NY : ISAAA . Ishikawa R , Toki N , Imai K , Sato YI , Yamagishi H , Shimamoto Y , Ueno K , Morishima H , Sato T . 2005 . Origin of weedy rice grown in Bhutan and the force of genetic diversity . Genetic Resources and Crop Evolution 52 , 395 – 403 . Google Scholar CrossRef Search ADS Lee D , Lupotto E , Powell W . 2009 . G-string slippage turns white rice red . Genome 52 , 490 – 493 . Google Scholar CrossRef Search ADS PubMed Li LF , Li YL , Jia Y , Caicedo AL , Olsen KM . 2017 . Signatures of adaptation in the weedy rice genome . Nature Genetics 49 , 811 – 814 . Google Scholar CrossRef Search ADS PubMed Lu BR , Yang C . 2009 . Gene flow from genetically modified rice to its wild relatives: assessing potential ecological consequences . Biotechnology Advances 27 , 1083 – 1091 . Google Scholar CrossRef Search ADS PubMed Luo D , Xu H , Liu Z , et al. 2013 . A detrimental mitochondrial–nuclear interaction causes cytoplasmic male sterility in rice . Nature Genetics 45 , 573 – 577 . Google Scholar CrossRef Search ADS PubMed Marambe B . 2005 . Properties of rice growing in abandoned paddies in Sri Lanka . In: Gressel J , ed. Crop ferality and volunteerism . Boca Raton, FL : CRC Press , 295 – 303 . Google Scholar CrossRef Search ADS Messeguer J , Marfa V , Catala MM , Guiderdoni E , Mele E . 2004 . A field study of pollen-mediated gene flow from Mediterranean GM rice to conventional rice and the red rice weed . Molecular Breeding 13 , 103 – 112 . Google Scholar CrossRef Search ADS Olsen KM , Caicedo AL , Jia Y . 2007 . Evolutionary genomics of weedy rice in the USA . Journal of Integrative Plant Biology 49 , 811 – 816 . Google Scholar CrossRef Search ADS Prathepha P . 2009 . Seed morphological traits and genotypic diversity of weedy rice (Oryza sativa f. spontanea) populations found in the Thai Hom Mali rice fields of north-eastern Thailand . Weed Biology and Management 9 , 1 – 9 . Google Scholar CrossRef Search ADS Rong J , Lu BR , Song Z , Su J , Snow AA , Zhang X , Sun S , Chen R , Wang F . 2007 . Dramatic reduction of crop-to-crop gene flow within a short distance from transgenic rice fields . New Phytologist 173 , 346 – 353 . Google Scholar CrossRef Search ADS PubMed Rong J , Wang F , Song Z , Su J , Chen R , Lu BR . 2012 . Scale effect on rice pollen-mediated gene flow: implications in assessing transgene flow from genetically engineered plants . Annals of Applied Biology 161 , 3 – 11 . Google Scholar CrossRef Search ADS Serrat X , Esteban R , Peñas G , Català MM , Melé E , Messeguer J . 2013 . Direct and reverse pollen-mediated gene flow between GM rice and red rice weed . AoB Plants 5 , plt050 . Google Scholar CrossRef Search ADS Shivrain VK , Burgos NR , Anders MM , Rajguru SN , Moore J , Sales MA . 2007 . Gene flow between Clearfield™ rice and red rice . Crop Protection 26 , 349 – 356 . Google Scholar CrossRef Search ADS Shivrain VK , Burgos NR , Gealy DR , Sales MA , Smith KL . 2009a . Gene flow from weedy red rice (Oryza sativa L.) to cultivated rice and fitness of hybrids . Pest Management Science 65 , 1124 – 1129 . Google Scholar CrossRef Search ADS Shivrain VK , Burgos NR , Sales MA , Mauromoustakos A , Gealy DR , Smith KL , Blank HL , Jia M . 2009b . Factors affecting the outcrossing rate between Clearfield™ rice and red rice (Oryza sativa) . Weed Science 57 , 394 – 403 . Google Scholar CrossRef Search ADS Singh V , Burgos N , Singh S , Abugho S , Earnest L , Gbur E , Scott R . 2016 . Herbicide and winter flood treatments for controlling volunteer rice off-season . Crop Protection 79 , 87 – 96 . Google Scholar CrossRef Search ADS Song X , Liu L , Wang Z , Qiang S . 2009 . Potential gene flow from transgenic rice (Oryza sativa L.) to different weedy rice (Oryza sativa f. spontanea) accessions based on reproductive compatibility . Pest Management Science 65 , 862 – 869 . Google Scholar CrossRef Search ADS PubMed Song X , Wang Z , Qiang S . 2011 . Agronomic performance of F1, F2 and F3 hybrids between weedy rice and transgenic glufosinate-resistant rice . Pest Management Science 67 , 921 – 931 . Google Scholar CrossRef Search ADS PubMed Song ZP , Lu BR , Wang B , Chen JK . 2004 . Fitness estimation through performance comparison of F1 hybrids with their parental species Oryza rufipogon and O. sativa . Annals of Botany 93 , 311 – 316 . Google Scholar CrossRef Search ADS PubMed Sudianto E , Beng-Kah S , Ting-Xiang N , Saldain NE , Scott RC , Burgos NR . 2013 . Clearfield® rice: its development, success, and key challenges on a global perspective . Crop Protection 49 , 40 – 51 . Google Scholar CrossRef Search ADS Sun G , Dai W , Cui R , Qiang S , Song X . 2015 . Gene flow from glufosinate-resistant transgenic hybrid rice Xiang 125S/Bar68-1 to weedy rice and cultivated rice under different experimental designs . Euphytica 204 , 211 – 227 . Google Scholar CrossRef Search ADS Sun J , Qian Q , Ma DR , Xu ZJ , Liu D , Du HB , Chen WF . 2013 . Introgression and selection shaping the genome and adaptive loci of weedy rice in northern China . New Phytologist 197 , 290 – 299 . Google Scholar CrossRef Search ADS PubMed Sweeney MT , Thomson MJ , Pfeil BE , McCouch S . 2006 . Caught red-handed: Rc encodes a basic helix-loop-helix protein conditioning red pericarp in rice . The Plant Cell 18 , 283 – 294 . Google Scholar CrossRef Search ADS PubMed Tang W , Chen H , Xu C , Li X , Lin Y , Zhang Q . 2006 . Development of insect-resistant transgenic indica rice with a synthetic cry1C* gene . Molecular Breeding 18 , 1 – 10 . Google Scholar CrossRef Search ADS Valverde BE . 2005 . The damage by weedy rice: can feral rice remain undetected ? In: Gressel J , ed. Crop ferality and volunteerism . Boca Raton, FL : CRC Press , 279 – 294 . Google Scholar CrossRef Search ADS Valverde BE . 2013 . Is herbicide resistant rice the ultimate solution for controlling weedy rice? Experiences from the Americas . Korean Journal of Weed Science 33 , 11 – 23 . Xiong HB , Xu HY , Xu Q , et al. 2012 . Origin and evolution of weedy rice revealed by inter-subspecific and inter-varietal hybridizations in rice . Molecular Plant Breeding 10 , 131 – 139 . Yang J , Wang J , Cao Q , Chen ZD , Tang LH , Zhong WG . 2009 . Development of a functional marker to red pericarp gene in weedy rice (Oryza sativa L. f. spontaneous) . Molecular Plant Breeding 7 , 721 – 726 . Yuan LP . 2002 . Science of hybrid rice . Beijing, China : China Agriculture Press . Yuan QH , Shi L , Wang F , Cao B , Qian Q , Lei XM , Liao YL , Liu WG , Cheng L , Jia SR . 2007 . Investigation of rice transgene flow in compass sectors by using male sterile line as a pollen detector . Theoretical and Applied Genetics 115 , 549 – 560 . Google Scholar CrossRef Search ADS PubMed Zhang JX , Lu ZM , Dai WM , Song XL , Peng YF , Valverde BE , Qiang S . 2015 . Cytoplasmic-genetic male sterility gene provides direct evidence for some hybrid rice recently evolving into weedy rice . Scientific Reports 5 , 10591 . Google Scholar CrossRef Search ADS PubMed Zhang L , Dai W , Wu C , Song X , Qiang S . 2012 . Genetic diversity and origin of Japonica- and Indica-like rice biotypes of weedy rice in the Guangdong and Liaoning provinces of China . Genetic Resources and Crop Evolution 59 , 399 – 410 . Google Scholar CrossRef Search ADS Zhang W , Linscombe SD , Webster E , Tan S , Oard J . 2006 . Risk assessment of the transfer of imazethapyr herbicide tolerance from Clearfield rice to red rice (Oryza sativa) . Euphytica 152 , 75 – 86 . Google Scholar CrossRef Search ADS Zuo J , Zhang L , Song X , Dai W , Qiang S . 2011 . Innate factors causing differences in gene flow frequency from transgenic rice to different weedy rice biotypes . Pest Management Science 67 , 677 – 690 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model) http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Experimental Botany Oxford University Press

Feral rice from introgression of weedy rice genes into transgenic herbicide-resistant hybrid-rice progeny

Journal of Experimental Botany , Volume Advance Article (16) – Jun 5, 2018

Loading next page...
 
/lp/ou_press/feral-rice-from-introgression-of-weedy-rice-genes-into-transgenic-a8BynGb90O
Publisher
Oxford University Press
Copyright
© The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. All rights reserved. For permissions, please email: journals.permissions@oup.com
ISSN
0022-0957
eISSN
1460-2431
D.O.I.
10.1093/jxb/ery210
Publisher site
See Article on Publisher Site

Abstract

Abstract Pollen-mediated transgenic flow of herbicide resistance occurs bidirectionally between transgenic cultivated rice and weedy rice. The potential risk of weedy traits introgressing into hybrid rice has been underestimated and is poorly understood. In this study, two glufosinate-resistant transgenic rice varieties, hybrid rice (F1), and their succeeding generations (F2–F4) were planted for 3 years in field plots free of weedy rice adjacent to experimental weedy-rice fields. Weedy-rice-like (feral) plants that were both glufosinate-resistant and had red-pericarp seed were initially found only among the F3 generations of the two glufosinate-resistant transgenic hybrid cultivars. The composite fitness (an index based on eight productivity and weediness traits) of the feral progeny was significantly higher than that of the glufosinate-resistant transgenic hybrid (the original female parent of the feral progeny) under monoculture common garden conditions. The hybrid rice progeny segregated into individuals of variable height and extended flowering. The hybrid rice F2 generations had higher outcrossing rates by pollen reception (0.96–1.65%) than their progenitors (0.07–0.98%). The results show that herbicide-resistant weedy rice can rapidly arise by pollen-mediated gene flow from weedy to transgenic hybrid rice, and their segregating pollen-receptive progeny pose a greater agro-ecological risk than transgenic varieties. The safety assessment and management regulations for transgenic hybrid rice should take into account the risk of bidirectional gene flow. Ecological risk, gene flow, herbicide-resistance, hybrid rice, transgenic rice, weedy rice Introduction Herbicide resistance has been the dominant trait of transgenic crops. The global area of transgenic crops had reached 185.1 million ha by 2016, of which over 80% were herbicide-resistant, including those with stacked traits (ISAAA, 2016). Most herbicide-resistant transgenic crops are transformed to withstand otherwise non-selective herbicides. These chemicals have a wide spectrum of weed control and thus kill the vast majority of plants with the exception of the resistant transgenic crop. The evolution of herbicide-resistant weeds caused by intense selection pressure from recurrent use of a single herbicide in transgenic crops has become widespread (Beckie and Hall, 2014; Busconi et al., 2014; Bonny, 2016). In addition, when weedy relatives infest transgenic crops, transgene flow can transfer the resistant trait to them and thus aggravate the weed problem (Lu and Yang, 2009). Therefore, the safe and sustainable use of transgenic technology depends on the implementation of proper prevention and mitigation of gene flow based on a thorough agro-ecological risk assessment of transgene escape (Gressel and Valverde, 2009; Rong et al., 2012). Weedy rice (Oryza sativa), conspecific to cultivated rice (also O. sativa), is one of the most serious weeds in paddy fields worldwide (Valverde, 2005; Delouche et al., 2007). It is frequently closely related to its companion cultivated rice, as it has been detected in morphological and genetic studies (Ishikawa et al., 2005; Akasaka et al., 2009; Prathepha, 2009; Gross et al., 2010; Li et al., 2017), suggesting a possible origin of weedy rice as a de-domesticated (feral) form. Although rice is predominantly self-pollinated, bidirectional gene flow between the cultivated and weedy forms occurs, although at a relatively low rate of usually less than 1% (Messeguer et al., 2004; Gealy, 2005; Rong et al., 2007; Shivrain et al., 2007). However, this rate is of practical relevance, as observed with the commercial cultivation of imidazolinone (IMI)-herbicide-resistant (ClearfieldTM) rice varieties that have been responsible for the quick introgression of the IMI-resistance genes into weedy rice populations wherever this rice is grown (Zhang et al., 2006; Gressel and Valverde, 2009; Valverde, 2013; Sudianto et al., 2013). Moreover, despite some cross-incompatibility between the two major subspecies of rice (O. sativa ssp. indica and O. sativa ssp. japonica), introgression of genes for enhanced traits does occur in the field, as observed for the transfer of IMI resistance from tropical japonica rice to indica weedy rice in the USA and elsewhere (Craig et al., 2014). Most research on gene flow has focused on the transfer of transgenes from transgenic rice to weedy rice (the ‘forward’ direction) (Gealy, 2005; Song et al., 2009), because of the concerns about the escape and persistence of such transgenes in weedy rice populations through seed-shattering and dormancy. Under properly managed commercial production, the density of cultivated rice greatly surpasses that of weedy rice even at high infestations (Valverde, 2005). However, ‘reverse’ gene flow (from weedy to cultivated rice), will also result in herbicide-resistant weedy individuals (Serrat et al., 2013). Therefore, gene flow between transgenic and weedy rice in any direction may lead to similar results (Olsen et al., 2007). In addition, it is known that the crop itself has the potential to back-mutate into weedy types, and thus many types of weedy rice are closely related to the varieties they associate with (Ammann et al., 2000; Dale et al., 2002; Zhang et al., 2012; Sun et al., 2013). Hence, transgenes are likely to escape from transgenic rice by the cultivated form reverting to a weed type. A detailed study based on cytoplasmic maternal inheritance has recently demonstrated that some weedy rice accessions derive from cultivated hybrid rice as an initial female parent (Zhang et al., 2015). This suggests that the herbicide-resistant transgenic rice itself also has the potential of being pollinated by weedy rice, resulting in herbicide-resistant weedy progeny. The risk of a herbicide-resistance transgene escape depends on the actual gene flow frequency between the transgenic and weedy rice or the reversion to weediness of a transgenic cultivar, and on the ability of the new herbicide-resistant progeny to persist. Gene flow from transgenic herbicide-resistant rice to weedy rice usually results in progeny at least equally as fit as their weedy rice parents (Cao et al., 2009; Song et al., 2011). Similarly, when cultivated rice serves as the female parent in a cross with weedy rice, the progeny is also as fit as the respective weedy rice parent (Shivrain et al., 2009a). However, the risk of reverse (weed to crop) gene flow has been seriously underestimated and even ignored completely because it was thought that the non-shattering hybrid seed would remain on the spike and be removed together with the crop grain. Hence, limited research has addressed this issue (Serrat et al., 2013). However, a considerable amount of rice seed can fall onto the soil during harvest (Gealy, 2005; Valverde, 2005), increasing the impact of reverse gene flow. This needs to be recognised and integrated into the assessment systems of potential agro-ecological risk of transgenic hybrid rice. Differences in the intrinsic genetic variability and growth characteristics of weedy rice biotypes also play a role in the outcome of bi-directional gene flow (Zuo et al., 2011). Particular attention should be paid to hybrid rice because of its heterozygous nature and its segregating progeny, easily observed by the variability in plant height, flowering time, and even pollen fertility (Yuan, 2002), all of which affect their outcrossing rate. These special characteristics, as well as protruding anthers (Gressel, 2002), may confer transgenic hybrid rice with a higher gene-flow risk and hence justify the need to impose stricter ecological safety evaluation and regulations. In this study, we established a 3-year experiment using transgenic rice varieties and hybrids to explore the likelihood and the frequency of these rice cultivars turning into herbicide-resistant weeds through receiving pollen from weedy rice. We hypothesise that, as previously proposed (Gressel, 2002), the comparatively higher outcrossing rate of hybrid rice to that of autogamous varieties may be a key factor for the introgression of weedy traits into the progeny of hybrid rice interacting with the weedy form in commercial fields. An additional gene flow experiment was conducted to indentify descriptors for important weediness traits. Our findings suggest that there is a need to revise the transgenic rice safety assessments and management regulations that were originally established based on the premise of unidirectional gene flow, and to consider the relevance of gene flow from weeds to transgenic crops. Materials and methods Plant material Glufosinate-resistant transgenic rice varieties and hybrids together with their progeny when appropriate were used in the experiments. The transgenic varieties were Bar68-1, 86B, and T1C-19, all homozygous for the bar transgene (Tang et al., 2006; Cui et al., 2013; Sun et al., 2015). The transgenic hybrids were Xiang125s/Bar68-1 (abbreviated as X125B) (Sun et al., 2015) and II you 86B (IIU86B) (Cui et al., 2013). X125B is a two-line hybrid whose male parent is Bar68-1. IIU86B is a three-line hybrid whose male and female parents are 86B and a wild abortive (WA) male-sterile line, respectively. Both the hybrid rice cultivars are hemizygous, containing one copy each of the bar gene. The conventional hybrid rice cultivars used in this experiment included the two-line hybrid rice 6 Liang you 9368 (6LY) (China Rice Data Center, http://www.ricedata.cn/variety/) and its F2 generation (F2-6LY), and the three-line hybrid rice II you 98 (IIY) (China Rice Data Center) and its F2 generation (F2-IIY). The F2 generations are the offspring of the rice hybrids that by definition are F1s. Nine weedy rice accessions were also included and were collected in Liaoning (WR1, WR5), Jiangsu (WR3, WR4), Anhui (WR2), Hunan (WR6), Guangdong (WR7, WR9), and Guangxi (WR8) provinces and selected from the weedy rice germplasm bank of the Weed Research Laboratory at Nanjing Agricultural University (Supplementary Table S3 at JXB online). Experimental field conditions Experimental fields were located at the Jiangpu Transgenic Rice Experiment Base of the Weed Research Laboratory at Nanjing Agricultural University (32°0′34″N, 118°36′51″E). The area was fenced and surrounded by poplar trees (Supplementary Fig. S1). A ‘pollen inflow experiment’ was established in a field section devoid of weedy rice where only clean agricultural implements were allowed. The nine weedy rice accessions were grown in paddy fields (two fields totaling 3000 m2 and one field of 1000 m2 in 2011 and 2012, respectively) situated east and north-east of the pollen inflow experiment field that represent the sole source of weedy rice pollen (Supplementary Fig. S1). A separate gene flow experiment was established in an isolated field nearby. Pollen inflow experiment The experiment consisted of six plots (Supplementary Fig. S2A), four of which were 70 m2 each (A1, A2, B1, and B2) together with two larger ones of 210 m2 each (A3 and B3). Different generations of the transgenic rice cultivars 86B, Bar68-1, IIU86B, and X125B were grown in the small plots between 2011 and 2013 (Supplementary Fig. S2B). The larger plots were used to increase the populations of segregating progeny of the IIU86B and X125B transgenic hybrids. All plots were inspected for the presence of weedy-rice-like (feral) plants with red-pericarp seed at maturity. The plots were established by direct water-seeding of broadcast pre-germinated seed at 10 g m−2 (dry seed weight) and sprayed each year with 450 g active ingredient (a.i.) ha−1 glufosinate ammonium (Basta, Bayer AG) at 40 d after seeding. Flowering time and plant height In each of the six plots in each of the 3 years 200 plants were randomly selected to evaluate flowering time (recorded as the period from first to last plant to begin flowering) and plant height at maturity. The flowering time obtained by this descriptor only reflects the period during which individuals began flowering and not the length of flowering, and hence was used only to compare flowering synchrony among the populations. Measurement of emergence rate of crop volunteers After harvest, seed obtained from the rice crop in each plot was broadcast onto the soil surface (20 g m−2) and allowed to overwinter without ploughing. In the following year, irrigation was provided at the beginning of May to stimulate germination. Emerged crop volunteers (from the broadcast seed and the naturally shed seed) were counted in the entire plot at the beginning of June and then removed from the field. Emergence of volunteers was expressed as density per plot. Screening and identification of feral plants All rice individuals in every plot were carefully inspected at maturity in each of the 3 years. Feral plants and their panicles with mature seed possessing red pericarps were collected and stored at low temperature (5–15 °C) until further use. Young leaves of these feral plants were collected for total DNA extraction. The molecular marker RID14 (forward: 5′-TCCAGGCACC ACACAGAGA-3′; reverse: 5′-GGCACTGAAATCACCTTGG-3′) based on the red-pericarp Rc gene (Yang et al., 2009) was used to genetically confirm the red pericarp phenotype, and to determine whether plants were homozygous or heterozygous for the Rc gene. The marker orWA352 (forward: 5′-GTTGATGGGTATGGATAGAG-3′; reverse: 5′-CGCAGGGCCTCGGTATATCTA-3′) based on the wild abortive cytoplasm male-sterility WA352 gene (Luo et al., 2013) was used to determine whether the female parent of feral plants was hybrid rice. The marker bar (forward: 5′-GCACCATCGTCAACCACTAC-3′; reverse: 5′-GCCAGAAACCCACGTCAT-3′) based on the glufosinate-resistance bar gene was used to detect the presence of the bar gene in the feral individuals. The PCR products of the marker RID14 and those of orWA352 and bar were separated by 6% and 1.5% polyacrylamide gel electrophoresis, respectively. Evaluation of weediness traits in feral individuals The seed produced by each feral individual was considered as a feral group. Seed of feral plants collected in 2012 from the F3 generations of transgenic two- and three-line hybrid rice in the A3 (33 feral groups) and B3 (27 feral groups) plots, and those of X125B, IIU86B (female parents of feral plants), and the nine weedy rice accessions (WR1–WR9, potential male parent of feral plants) were sown in a nursery. When the plants reached the six-leaf growth stage they were transplanted into a common garden at a distance of 25 cm within rows and 30 cm between rows. Glufosinate ammonium (450 g a.i. ha−1) was applied 15 d after transplanting to verify the expression of glufosinate resistance in the feral plants and to control weeds. Uncontrolled weeds and those emerging in plots planted with weedy rice accessions, where the herbicide was not applied, were eliminated manually. Eight traits related to weediness were measured at the reproductive stage: plant height, tiller number per plant, estimated flag leaf area (calculated as the product of leaf length × width), panicle length, spikelet number per panicle, seed-set rate, seed-shattering rate, and 100-seed weight. Ten plants of each feral group, X125B, IIU86B, and the nine weedy rice accessions were randomly selected for the measurements. The relative fitness of feral groups compared to that of their initial female crop parent and potential weedy male parents (the nine weedy rice accessions) was calculated according to Song et al., (2004) and Song et al., (2011). Briefly, the mean value for each trait measured in the female parent was assigned a value of ‘1.0’. The ratio with the mean value of the trait in each of the corresponding feral groups was then calculated. Similarly, the relative fitness of each feral group in relation to each of its potential (weedy rice) male parents was calculated, this time assigning the value of ‘1.0’ to the mean value for the trait measured in the corresponding weedy accession. For each individual comparison, a value above 1.0 indicated enhanced performance of the feral group. Finally, a composite fitness value was assigned to each feral group (in relation to each of its female or potential male parents) as the mean value of the eight relative fitness values. Gene flow-experiment Seedlings of transgenic varieties and conventional hybrids at the six-leaf growth stage were transplanted at a distance of 25 cm in rows separated by 30 cm (Supplementary Fig. S3). The transgenic rice varieties 86B, Bar68-1, and T1C-19 served as pollen donors while hybrid rice 6LY and IIY and their offspring (F2 generations designated as F2-6LY and F2-IIY, respectively) were used as pollen receptors (12 gene-flow combinations). Each combination was repeated four times; there were 60 plants of each pollen receptor in each replicate. The length of flowering (the duration for which flowers were open) of each pollen donor and receptor was recorded to calculate the overlapping flowering time for each gene-flow combination. The height of 10 randomly selected plants of each type was measured at flowering. The male sterility of plants in the F2-6LY and F2-IIY populations in each gene-flow combination was confirmed by the standard pollen I2-KI staining method. A single panicle from each male-sterile plant was selected, and five panicles were collected in each combination at maturity for detection of glufosinate-resistant seed. Seeds of pollen receptors in each combination were harvested at maturity, and at least 10% of them were used to verify glufosinate resistance (Song et al., 2011) for calculation of the rate of gene flow. Statistical analysis ANOVA was used to determine the significance of differences in plant height and outcrossing rates among pollen donors and receptors in the gene-flow experiment, as well as the significance of differences in volunteer density among the types used in the ‘pollen inflow experiment’ in each year. Independent t-tests were used for all comparisons involving pairs of means, including for plant height and seed-shattering rate comparisons between the feral progeny and their initial female parent, the outcrossing-rate comparisons between the two hybrid rice generations in the gene-flow experiment, and the relative fitness comparisons between the feral progeny and their original female parent and original potential male parent. All statistical analyses were carried out using PASW Statistics version 18.0 for Windows (IBM Inc., New York, USA). Results Herbicide-resistant feral plants are found in the F3 progeny of field-grown transgenic hybrid rice Weedy-rice-like (feral) plants with red pericarps were identified by inspecting every rice plant in the experimental paddy field. None was found associated with the transgenic varieties or with the transgenic hybrids (F1) themselves or the progeny (F2) of the hybrids in the first year of the experiment (2011). Distinctive feral plants appeared in some plots in the following two years (Fig. 1A). In the second year (2012), 33 and 27 feral plants were found exclusively among the hybrid rice generations F3-IIU86B and F3-X125B growing in Plots A3 and B3, respectively (Fig. 1B). In the third year, in addition to 20 and nine feral plants found in Plots A3 and B3 planted with F4-IIU86B and F4-X125B, respectively, four and one feral plants were also found in Plots A2 and B2, which were planted with F3-IIU86B and F3-X125B, respectively (Fig. 1C). Interestingly, no feral plants were ever found throughout the 3-year experiment in Plots A1 or B1, which were respectively sown with saved seed from the cultivars 86B and Bar68-1 harvested the previous season. Thus, feral plants with red pericarps began appearing only in the F3 generations of transgenic hybrid rice. Fig. 1. View largeDownload slide Occurrence and frequency of weedy-rice-like (feral) plants and their identification. (A) Feral plants with red pericarps. (B, C), Experimental plots in which feral plants were found (highlighted in red) in 2012 (B) and in 2013 (C) together with the number of feral individuals found in each plot (underlined); the identity of the plant material grown in each plot is indicated. (D) Marker RID14 amplifies a 154-bp fragment from weedy rice homozygous for the Rc gene (lane 1) and a 140-bp fragment from cultivated rice homozygous for the rc gene (lane 2). All the feral plants (lanes 3–12) were heterozygous (Rc/rc) for the red pericarp gene except for one individual of F4-IIU86B that was homozygous (lane not shown). M represents the DL500 DNA Marker. (E) The bar marker only amplifies a 429-bp fragment from rice containing the glufosinate-resistance bar gene (lane 1) that is absent in weedy rice (lane2). All feral plants contained the bar gene (lanes 3–12). M represents the DL500 DNA Marker. (F, G) The marker orWR352 only amplifies a 1432-bp fragment from rice containing the wild abortive (WA) cytoplasmic male-sterility gene WA352 (lane 1) that is absent in a reference male-fertile restorer line, Zhenshan 97B (lane 2). The feral plants found in Plots A2 and A3 contained the WA352 gene (F, lanes 3–12) but it was absent in all feral plants found in Plots B2 and B3 (G, lanes 3–12). M represents the DL2000 DNA Marker. Fig. 1. View largeDownload slide Occurrence and frequency of weedy-rice-like (feral) plants and their identification. (A) Feral plants with red pericarps. (B, C), Experimental plots in which feral plants were found (highlighted in red) in 2012 (B) and in 2013 (C) together with the number of feral individuals found in each plot (underlined); the identity of the plant material grown in each plot is indicated. (D) Marker RID14 amplifies a 154-bp fragment from weedy rice homozygous for the Rc gene (lane 1) and a 140-bp fragment from cultivated rice homozygous for the rc gene (lane 2). All the feral plants (lanes 3–12) were heterozygous (Rc/rc) for the red pericarp gene except for one individual of F4-IIU86B that was homozygous (lane not shown). M represents the DL500 DNA Marker. (E) The bar marker only amplifies a 429-bp fragment from rice containing the glufosinate-resistance bar gene (lane 1) that is absent in weedy rice (lane2). All feral plants contained the bar gene (lanes 3–12). M represents the DL500 DNA Marker. (F, G) The marker orWR352 only amplifies a 1432-bp fragment from rice containing the wild abortive (WA) cytoplasmic male-sterility gene WA352 (lane 1) that is absent in a reference male-fertile restorer line, Zhenshan 97B (lane 2). The feral plants found in Plots A2 and A3 contained the WA352 gene (F, lanes 3–12) but it was absent in all feral plants found in Plots B2 and B3 (G, lanes 3–12). M represents the DL2000 DNA Marker. The red pericarp genotype was confirmed by detecting the Rc gene through a molecular marker test. All feral plants were heterozygous for the Rc gene (Fig. 1D), except for one feral individual of F4-IIU86B that was homozygous. Additionally, a separate molecular marker test confirmed that all feral plants had the glufosinate-resistance bar gene (Fig. 1E). All feral plants found in Plots A2 and A3 (that had been initially planted with hybrid rice bred using a WA male-sterile line) contained the WA cytoplasmic male-sterility gene WA352, but it was absent in all feral plants found in Plots B2 and B3 (Fig. 1F) that were initially planted with the two-line hybrid rice X125B (in whose production no WA male-sterile line was involved). Thus, it is highly likely that these feral plants were directly derived from transgenic hybrid rice within the same plot. Fitness traits engender weediness in feral progeny Although a red pericarp is a common characteristic among weedy rice accessions, the most agronomically relevant traits of weedy rice are its competitive ability and shattering of differentially dormant seed. The calculated composite fitness of each progeny group based on eight characteristics (vegetative: plant height, tiller number, estimated flag leaf area; and reproductive: panicle length, spikelet number, seed set rate, shattering rate, 100-seed weight) measured under common garden conditions indicated that most feral groups had higher relative fitness than their original female parents (transgenic hybrid rice). Many of the feral groups even had higher relative fitness than their weedy rice parents, i.e. the nine weedy rice accessions (Fig. 2). Feral plants consistently had greater relative fitness for plant height (i.e. they were taller) than their crop female parent or any of the potential weedy rice male parents (Supplementary Fig. S4), except those designated as WR7 and WR8 (Supplementary Tables S1, S2). Likewise, they also had greater relative fitness for leaf area except for that of the potential WR4 and WR6 male parents. All feral plants bore fewer tillers than their known female or their putative male parents (Supplementary Tables S1, S2). In general, feral progeny were vegetatively superior to their female crop progenitor and equivalent or superior to their potential weedy rice parents. In relation to reproductive traits, feral progeny developed equivalent or longer panicles, equivalent or more spikelets, but set less seed than their parents (crop or potential weedy). The feral progeny produced lighter seed than their female parents, but equivalent or heavier than that of their weedy rice parents (Supplementary Tables S1, S2). The seed of the feral progeny shattered substantially more than their female parents (Supplementary Fig. S4), but less than their weedy rice parents (Supplementary Tables S1, S2). Fig. 2. View largeDownload slide Composite fitness of weedy-rice-like (feral) progeny groups compared with their corresponding parents. (A) Composite fitness of 33 feral progeny groups compared with the original female parent (IIU86B) and original potential male parents (nine weedy rice accessions, WR1–WR9), obtained from plants grown in Plot-A3 in 2012. Most progeny had higher fitness than their original female parent or each of their original potential male parents. (B) Composite fitness of 27 feral progeny groups compared with the original female parent (X125B) and original potential male parents (nine weedy rice accessions, WR1–WR9), obtained from plants grown in Plot-B3 in 2012. All progeny also had higher fitness than their corresponding original female parent and most of the original potential male parents. In (A, B) the bars represent numbers of feral groups (left y-axis) and each dot represents the relative fitness of one feral group (right y-axis). The data are divided into relative fitness values greater than 1 (red) and smaller than 1 (black). Fig. 2. View largeDownload slide Composite fitness of weedy-rice-like (feral) progeny groups compared with their corresponding parents. (A) Composite fitness of 33 feral progeny groups compared with the original female parent (IIU86B) and original potential male parents (nine weedy rice accessions, WR1–WR9), obtained from plants grown in Plot-A3 in 2012. Most progeny had higher fitness than their original female parent or each of their original potential male parents. (B) Composite fitness of 27 feral progeny groups compared with the original female parent (X125B) and original potential male parents (nine weedy rice accessions, WR1–WR9), obtained from plants grown in Plot-B3 in 2012. All progeny also had higher fitness than their corresponding original female parent and most of the original potential male parents. In (A, B) the bars represent numbers of feral groups (left y-axis) and each dot represents the relative fitness of one feral group (right y-axis). The data are divided into relative fitness values greater than 1 (red) and smaller than 1 (black). The high relative (both vegetative and reproductive) and composite fitness, weediness traits, and glufosinate-resistance of the feral plants suggest that they could be potentially harmful and difficult to control in paddy fields, particularly if glufosinate-resistant transgenic rice is grown sequentially. Hybrid rice segregating progeny have extended flowering time and height diversity Plant height and flowering duration were recorded for every generation in the experiments, including for the transgenic rice varieties 86B and Bar68-1 and the transgenic hybrids X125B and IIU86B and their F2–F4 generation progeny, which included the feral plants whose characteristics have been described in the previous section. The transgenic cultivars had uniform and narrow flowering duration (8–14 d) and plant height (height difference between 15 to 21 cm). The segregating progeny (F2–F4 generations) of transgenic hybrid IIU86B began flowering earlier than their progenitor, and that of X125B initiated flowering at the same time or earlier than their X125B female parent. All progeny had a substantially extended initiation of flowering of to up to 36 d. Their range of values for plant height was also widened, with the difference between the shortest and tallest plant being 86 cm (Fig. 3). Fig. 3. View largeDownload slide The progeny of transgenic hybrid rice were highly variable in flowering time and plant height because of character segregation. (A) The segregating progeny of hybrid rice included early maturing individuals that were already ripening their seed while others were just flowering. The images were taken on the same day. (B, C) The transformed rice materials 86B, Bar68-1, IIU86B, and X125B had similar flowering time durations and ranges of plant height (indicated by the bars). The segregating progeny (F2–F4) of the transgenic hybrids IIU86B and X125B had extended flowering periods and diverse plant heights. The line graphs inside the bars represent the distribution of individuals within the respective ranges of flowering time and height. Fig. 3. View largeDownload slide The progeny of transgenic hybrid rice were highly variable in flowering time and plant height because of character segregation. (A) The segregating progeny of hybrid rice included early maturing individuals that were already ripening their seed while others were just flowering. The images were taken on the same day. (B, C) The transformed rice materials 86B, Bar68-1, IIU86B, and X125B had similar flowering time durations and ranges of plant height (indicated by the bars). The segregating progeny (F2–F4) of the transgenic hybrids IIU86B and X125B had extended flowering periods and diverse plant heights. The line graphs inside the bars represent the distribution of individuals within the respective ranges of flowering time and height. Segregating progeny of hybrid rice have higher outcrossing rates than hybrid rice The outcrossing rates of two non-transgenic commercial hybrid rice cultivars (6LY and IIY) and their respective offspring (F2 generation, F2-6LY and F2-IIY), serving as pollen receptors, were compared in a separate gene-flow experiment. Every pollen receptor was planted together with three varieties of transgenic rice (Bar68-1, TIC-19, and 86B) as pollen donors. In all combinations, the outcrossing rates of both F2 generations were significantly higher (0.96–1.65%) than those of their respective parental hybrid rice (0.07–0.98%) (Fig. 4A, B). Correspondingly, both F2 generations had longer overlapping times of flowering with each pollen donor than the two hybrid rice cultivars, which resulted from their longer flowering periods (Fig. 4D). Fig. 4. View largeDownload slide Gene flow between hybrid rice cultivars (F1, F2 generations) as pollen receptors and transgenic rice varieties as pollen donors. (A, B) The outcrossing rate (top) and the overlapping of flowering time (bottom) of pollen donors and pollen receptors. In (A) the black and red bars represent the hybrid rice cultivar 6LY and F2-6LY (progeny of 6LY) as pollen receptors, respectively, with the transgenic rice varieties Bar68-1, T1C-19, and 86B as pollen donors. In (B) the black and red bars represent the hybrid rice cultivar IIY and F2-IIY (progeny of IIY) as pollen receptors, respectively, with the same transgenic rice varieties Bar68-1, T1C-19, and 86B as pollen donors. (C) Outcrossing rate (by pollen reception) of male-sterile plants found with F2-6LY (black) and F2-IIY (red). (D) Flowering stage (top) and plant height (bottom) of pollen donors (black) and pollen receptors (red). Data are means (±SD). Different letters indicate significant differences at P<0.05. Fig. 4. View largeDownload slide Gene flow between hybrid rice cultivars (F1, F2 generations) as pollen receptors and transgenic rice varieties as pollen donors. (A, B) The outcrossing rate (top) and the overlapping of flowering time (bottom) of pollen donors and pollen receptors. In (A) the black and red bars represent the hybrid rice cultivar 6LY and F2-6LY (progeny of 6LY) as pollen receptors, respectively, with the transgenic rice varieties Bar68-1, T1C-19, and 86B as pollen donors. In (B) the black and red bars represent the hybrid rice cultivar IIY and F2-IIY (progeny of IIY) as pollen receptors, respectively, with the same transgenic rice varieties Bar68-1, T1C-19, and 86B as pollen donors. (C) Outcrossing rate (by pollen reception) of male-sterile plants found with F2-6LY (black) and F2-IIY (red). (D) Flowering stage (top) and plant height (bottom) of pollen donors (black) and pollen receptors (red). Data are means (±SD). Different letters indicate significant differences at P<0.05. The outcrossing rate from transgenic rice pollen donors to male-sterile plants among two F2 generations (F2-6LY and F2-IIY) was also determined. Overall, the obligate cross-pollinated male-sterile plants had relatively higher outcrossing rates (8.8–22.4%) (Fig. 4C) compared with the two F2 generations (0.96–1.65%) (Fig. 4A, B). It is notably that the transgenic cultivar Bar68-1 was significantly shorter than the pollen receptors (Fig. 4D), probably making it an unsuccessful donor. In addition, Bar68-1 was at a further disadvantage as a pollen donor because of its earlier flowering and shorter flowering period that results, which resulted in a limited overlapping time of flowering with the pollen receptors (Fig. 4D). Consequently, the combinations in which 6LY and IIY were the pollen receptors had low outcrossing rates. However, the combinations in which the slightly shorter (than their respective female progenitors) F2-6LY and F2-IIY were the pollen receptors had relatively high outcrossing rates (Fig. 4A, B). In addition, the male-sterile plants among these two combinations had the highest outcrossing rates through cross-pollination (Fig. 4C). In a field situation, the presence of male-sterile plants among segregating progeny of hybrid rice flowering earlier and being surrounded by typically early-flowering weedy rice plants would be highly conducive for ‘reverse’ gene flow (i.e. from weed to crop volunteer). Under these conditions, weedy rice, and not the crop, would be the most likely source of pollen. Progeny of transgenic hybrid rice can potentially persist as volunteers Farmers do not save seed fromtheir harvested hybrid rice for subsequent sowings. Thus, the possibility of progeny of hybrid rice becoming established depends on volunteers. According to our quantitative observations from 2012 to 2014, all the rice materials that was planted produced volunteers in the following season (0.26–9.62 plants m−2) (Fig. 5). The density of volunteers emerging in a particular year depended on the lines planted. For example, in 2012 the F2 and F3 of the two-line transgenic hybrid rice X125B had more volunteers (9.62 and 5.58 plants m−2, respectively; P≤0.05) than the F2 and F3 of the three-line transgenic hybrid rice IIU86B (1.64 and 1.22 plants m−2, respectively). In the following two years, progeny (F3–F5) of X125B also consistently had more volunteers than the respective progeny of IIU86B. In general, the number of volunteers produced by the transformed varieties was much lower than most hybrid progeny (Fig. 5). Therefore, the ability to produce volunteers was influenced by the genetic background of the rice. The volunteer density also varied with cropping season (year) for each particular rice line. These varying densities, particularly of the transgenic rice varieties 86B (0.4, 0.26, and 4.46 plants m−2 in 2012, 2013, and 2014, respectively) and Bar68-1 (1.58, 0.36, and 5.2 plants m−2), strongly suggest that the occurrence of volunteers is influenced by environmental conditions. Fig. 5. View largeDownload slide The density of volunteer plants of different rice material cultivated in the pollen-inflow experiment from 2012 to 2014. Bar68-1 and 86B are transgenic varieties, IIU86B and X125B are transgenic hybrid rice; F2-, F3-, and F4- denote the first, second, and third offspring of the respective hybrid (F1). Data are means (±SD). Different letters indicate significant differences at P<0.05. Fig. 5. View largeDownload slide The density of volunteer plants of different rice material cultivated in the pollen-inflow experiment from 2012 to 2014. Bar68-1 and 86B are transgenic varieties, IIU86B and X125B are transgenic hybrid rice; F2-, F3-, and F4- denote the first, second, and third offspring of the respective hybrid (F1). Data are means (±SD). Different letters indicate significant differences at P<0.05. Discussion Herbicide-resistant weedy rice can originate through reverse gene flow Several generations of the transgenic rice varieties 86B and Bar68-1 and of the transgenic hybrid rice IIU86B and X125B were grown over three years in an experimental field that was under glufosinate selection pressure and free of weedy rice, but in close proximity to weedy rice growing in adjacent fields (Supplementary Figs S1A, S2). Weedy-rice-like (feral) individuals with red pericarps were initially found in plots (A3 and B3) as part of the F3 generation of two transgenic hybrid rice (F1) cultivars grown in 2012 (Fig. 1). Similarly, feral individuals also appeared in 2013 in plots (A2 and B2) that were planted with the F3 generations of the two transgenic hybrid rice cultivars. Feral plants were never found in plots (A1 and B1) planted with saved seed of two transgenic rice varieties even though these plots were placed side-by-side with the ones planted with hybrid rice. The feral individuals were glufosinate-resistant and carried the bar gene. In addition, the plasmagene WA352 contained in IIU86B was only detected in feral plants collected in plots (A2 and A3) that were planted with the F3 or F4 generations of IIU86B. Therefore, feral plants most likely directly derived from transgenic hybrid rice within each plot. In the majority of cases, the red pericarp character in rice is under the control of a dominant Rc gene, while the white pericarp character is under the control of the recessive rc gene that carries a 14-bp deletion in the seventh exon of the Rc gene (Sweeney et al., 2006; Furukawa et al., 2007). All transgenic rice cultivars (varieties and hybrids) used in this study possessed white pericarps and were homozygous for the rc gene. The feral individuals indentified by their red pericarps were heterozygous for the Rc gene, except for a single individual of the F4-IIU86B generation found in one of the plots that was homozygous for this trait. This individual was probably an offspring of a feral plant from the previous year. There are two known potential possibilities for the presence of the Rc gene in a population that originally lacks the gene: gene recombination and gene flow. It is unlikely that the rc gene would mutate into the Rc gene with such high frequency and only in a way that happened in specific generations of the transgenic hybrid rice. Although not being directly a back-mutation, red pericarps can arise in otherwise white-pericarp cultivated material by a second deletion in the Rc gene that restores the red color as a dominant trait (Brooks et al., 2008; Lee et al., 2009), but this does not seem plausible in our case because all feral plants carried the Rc gene (without the deletion). A possible explanation, for which we are not aware of any previous direct evidence, is that two white-pericarp rice varieties could produce a red-pericarp hybrid progeny through gene recombination; however, there are claims that indica×japonica hybrids may segregate individuals with red pericarps (Xiong et al., 2012). The possibility of gene recombination would also fail to explain why there were no red-pericarp individuals among the F2 generation of the transgenic hybrid rice. Therefore, the only plausible explanation is that the dominant allele in feral plants heterozygous for the Rc gene comes from gene flow, and that the pollen donor (with red pericarps) was the weedy rice growing in the adjacent fields. Traits conferring competitive ability and weediness to feral rice and their progeny will determine the success of the progeny as newly derived weedy rice biotypes. Progeny of hybrids between cultivated rice and weedy rice have always been found to be at least as fit as their parents (Cao et al., 2009; Shivrain et al., 2009a; Song et al., 2011). Our results were consistent with those of previous studies in that most feral groups had higher composite fitness indexes than their initial female parents (transgenic hybrid rice) (Fig. 2). The progeny of feral plants were taller and exhibited more seed shattering, which might confer a greater competitive advantage and a better opportunity to persist as weedy forms compared to their initial transgenic hybrid rice female parents (Supplementary Tables S1, S2). In addition, feral progeny could survive under glufosinate selection pressure together with glufosinate-resistant crop volunteers. Depending on their weedy attributes, they could persist and subsequently evolve into herbicide-resistant weedy rice biotypes. Clearly reverse gene flow from weedy rice to transgenic hybrid rice will increase the likelihood of the rapid evolution of herbicide-resistant weedy rice, especially under the selection pressure of the herbicide. This is a major agro-ecological risk that needs to be considered and mitigated otherwise it poses a threat to the sustainability of transgenic herbicide-resistant rice technology. Transgenic hybrid rice is more prone to transgene escape than transgenic varieties The fact that feral plants were found beginning with the F3 hybrid rice generation has important implications (Fig.1B). If such plants arose from reverse gene flow, then the F2 progeny and not the F1 (the transgenic hybrid rice itself) must had been the pollen receptor. Both generations were juxtaposed within the experimental plot arrangement, which later also included the F3 generations. This would suggest that the F2 generation had higher crossing ability by pollen receptivity than the cultivated hybrid rice. Furthermore, no feral individuals were among subsequent plantings of saved seed of the two transgenic varieties, also emphasising the apparent increased outcrossing rate of the F2 hybrid generation. In the gene-flow experiment, we indeed determined that the F2 generation of hybrid rice had a substantially higher outcrossing rate due to pollen reception than the cultivated hybrid itself. The outcrossing rate is determined by both biotic and abiotic factors (Gealy, 2005; Rong et al., 2007; Yuan et al., 2007; Shivrain et al., 2009b; Zuo et al., 2011). Abiotic factors include temperature, wind speed and direction, and air humidity, all of which should have been uniform across the plots given the scale of our experiment. Therefore, the increased outcrossing rate observed in the F2 offspring must have been associated with differential biological characteristics of this segregating progeny. First, individuals in the progeny of the hybrid rice had unsynchronised flowering times because of character segregation (Fig. 4). Thus, the F2 generation of hybrid rice had a flowering duration that was almost doubled compared to its hybrid rice parent in the gene-flow experiment, and it was further extended in the F3 and F4 generations (Fig. 3). The implication of this extension of the flowering period in comparison to both hybrid rice and conventional varieties is that the pollen-receptor function of the hybrid rice progeny would have overlapped for longer with the different pollen donors. The longer overlapping times in flowering were most likely the determining factor in the higher outcrossing rates of the hybrid rice progeny. In addition, transgenic hybrid rice could also be a pollen donor, in which case the extended flowering period would increase the risk of transgene escape. Quantifying this outcrossing route was not the objective of this current research, but it is something that needs to be considered, particularly in relation to dealing with the risk of movement and establishment (introgression) of herbicide-resistance transgenes to weedy rice, especially in the absence of mitigating measures (Gressel and Valverde, 2009). Second, among the segregating progeny of hybrid rice there were male-sterile individuals that had obligate cross-pollination, except for hybrid rice in whose production a gametophyte abortion-type sterile line had been used. In the gene-flow experiment, the male-sterile plants had a significantly higher outcrossing (by pollen reception) rate (8.8–22.4%) than the overall rate of the hybrid rice F2 generation (0.96–1.65%) (Fig. 4A, B). Consequently, the male-sterile individuals would be expected to have greatly contributed to the relatively high outcrossing rate of the hybrid rice F2 generation. Moreover, the greater range of plant heights of the hybrid rice progeny should also favor bidirectional outcrossing, except, of course, for male-sterile individuals. Therefore, the differential outcrossing rates between weedy rice and hybrid rice (both as F1 and F2 generations) together with conventional rice varieties should be addressed in more detail in future gene-flow studies. It is worth noting that even if the presence of male-sterile plants is beneficial for reverse gene flow, their success will depend on the ability of the offspring to predominantly self-pollinate and set seed. Fertility restoration is easy for male-sterile plants among two-line hybrid rice progeny, because almost any other rice material will possess dominant fertility genes that restore the photo-thermo-sensitive male-sterility gene. For male-sterile plants among three-line hybrid rice progeny, however, pollen with the required specific restoring gene is needed for fertility restoration. As expected, the overall average rate of seed set of feral groups whose initial female parent was X125B (two-line hybrid rice) was significantly higher than for progeny derived from IIU86B (three-line hybrid rice) as the initial female parent (Supplementary Fig. S5). In theory, the grain of transgenic varieties could be saved as seed for future planting but hybrid rice farmers, who understand the problems associated with segregation, purchase fresh commercial seed every cropping season. Thus, the risk of herbicide-resistant transgenes establishing in the field may actually be less with hybrid rice than with varietal material. In the pollen-inflow field experiment, each generation of transgenic hybrid rice was able to persist as volunteers in the following season (Fig. 5). The persistence of volunteers may be the first step towards plants becoming feral (Marambe, 2005). The occurrence of volunteers was influenced by the rice genetic background and environmental conditions. According to a recent study, hybrid rice would be expected to persist more as volunteers than conventional rice because of the higher overwintering capacity of its seed (Singh et al., 2016). If possible, transgenic hybrid rice material with lower ability to survive as volunteers should be preferentially selected for commercial use, especially in regions with warmer climates and under continuous rice cultivation that are more conducive for the establishment and persistence of crop volunteers. A paddy field devoid of weedy rice was used for the pollen-inflow experiment, and weedy rice was only present in adjacent fields. Even if the glufosinate is used to eliminate the weedy rice population in a paddy field planted with glufosinate-resistant rice, some weedy rice plants are likely to escape treatment (Busconi et al., 2014). Bidirectional gene flow will be more likely to occur under these circumstances because both the transgenic and weedy rice will be in close proximity. It should be noted that our experimental design with exceedingly high densities of transgenic hybrid rice progeny in experimental plots and weedy rice in the adjacent fields was highly conducive for obtaining feral plants. It is possible that the frequency of feral plants under a different design, or with different wind patterns at flowering, could vary, as has been observed in other gene-flow studies (Rong et al., 2012; Sun et al., 2015). In our case the wind patterns at flowering (Supplementary Fig. S1) clearly allowed reverse gene flow to occur, as clearly demonstrated by the phenotypic and genetic characterisation of the feral plants. It is undeniable that because of the increased outcrossing rate of its segregating progeny, transgenic hybrid rice should have a higher risk of transgene escape than transgenic rice varieties. This agro-ecological risk must not be neglected, particularly if herbicide-resistant hybrid rice that is either transgenic but without transgenic mitigation (Gressel and Valverde, 2009) or is conventionally bred becomes commercially successful and is planted over large areas, thus increasing the likelihood of greater areas being infested by herbicide-resistant weedy rice. Based on the results obtained in this study, it is evident that a more stringent management or stewardship program should be implemented for transgenic hybrid rice cultivation compared to transgenic varieties because of the higher agro-ecological risk associated with the higher outcrossing rate of the hybrid progeny. Special attention should be given to the management of weedy rice in adjacent fields to prevent reverse gene flow. Weedy rice should be properly controlled in fields and neighboring areas where transgenic hybrid rice seed is being produced because of the risk of a proportion of the male-sterile line being pollinated by weedy individuals. Weedy rice control with the herbicide to which the transgenic rice is resistant should be complemented with other practices that are properly integrated into the production system; for example, stale seedbed preparation, when feasible, to decrease the initial density of weedy and volunteer rice that can become established within the crop, particularly in direct-seeded fields at low latitude regions with suitable conditions for continuous rice production systems. It is advisable to initially plant transgenic herbicide-resistant varieties instead of hybrids in fields heavily infested with weedy rice, since varietal material poses a lower risk for gene flow than hybrid rice. In any case, it is best to have a proper rotation system in which varieties or hybrids with transgenic or conventionally bred resistance to herbicides with different modes of action are alternated. The rotation scheme can also be based on the deployment of hybrid rice that incorporates transgenic mitigation of gene flow (Gressel and Valverde, 2009). Finally, safety assessments and management regulations for transgenic hybrid rice should be further improved to take into account the risk of reverse gene flow. Supplementary data Supplementary data are available at JXB online. Table S1. Calculated relative fitness of the progeny of feral plants derived from the F3 generation of the transgenic herbicide-resistant three-line hybrid rice IIU86B (F1) compared to their original female parent and original potential male parents. Table S2. Calculated relative fitness of the progeny of feral plants derived from the F3 generation of the transgenic herbicide-resistant three-line hybrid rice X125B (F1) compared to their original female parent and original potential male parents. Table S3. Source of weedy rice materials used in this study. Fig. S1. Experimental location and surrounding environment. Fig. S2. Field experiment layout and cropping schedule in the pollen-inflow experiment. Fig. S3. Diagram of an individual experimental plot showing the planting layout in the gene-flow experiment. Fig. S4. Seed-shattering rate and plant height of feral progeny and their original female parent. Fig. S5. Mean seed-setting rate of feral groups. Acknowledgements This study was financially supported by the China Transgenic Organism Research and Commercialization Project (2016ZX08011-001), the Doctoral Program of Higher Education of the Ministry of Education (20130097130006), the 111 Project, and the National Natural Science Foundation of China (30800604). Rice materials were kindly provided by the Institute of Subtropical Agriculture of the Chinese Academy of Sciences (Bar68-1 and Xiang125s/Bar68-1), the South China Botanical Garden of the Chinese Academy of Sciences (86B and II you 86B), Huazhong Agricultural University (T1C-19), and Nanjing Shenzhou Seeds Industry Co., Ltd. (6 Liang you 9368 and II You 98). The authors have no conflicts of interest to declare. References Akasaka M , Ushiki J , Iwata H , Ishikawa R , Ishii T . 2009 . Genetic relationships and diversity of weedy rice (Oryza sativa L.) and cultivated rice varieties in Okayama Prefecture, Japan . Breeding Science 59 , 401 – 409 . Google Scholar CrossRef Search ADS Ammann K , Jacot Y , Mazyad PRA . 2000 . Weediness in the light of new transgenic crops and their potential hybrids . Journal of Plant Diseases and Protection 107 , 19 – 29 . Beckie HJ , Hall LM . 2014 . Genetically-modified herbicide-resistant (GMHR) crops a two-edged sword? An Americas perspective on development and effect on weed management . Crop Protection 66 , 40 – 45 . Google Scholar CrossRef Search ADS Bonny S . 2016 . Genetically modified herbicide-tolerant crops, weeds, and herbicides: overview and impact . Environmental Management 57 , 31 – 48 . Google Scholar CrossRef Search ADS PubMed Brooks SA , Yan W , Jackson AK , Deren CW . 2008 . A natural mutation in rc reverts white-rice-pericarp to red and results in a new, dominant, wild-type allele: Rc-g . Theoretical and Applied Genetics 117 , 575 – 580 . Google Scholar CrossRef Search ADS PubMed Busconi M , Baldi G , Lorenzoni C , Fogher C . 2014 . Gene flow from transgenic rice to red rice (Oryza sativa L.) in the field . Plant Biology 16 , 22 – 27 . Google Scholar CrossRef Search ADS PubMed Cao QJ , Xia H , Yang X , Lu BR . 2009 . Performance of hybrids between weedy rice and insect-resistant transgenic rice under field experiments: implication for environmental biosafety assessment . Journal of Integrative Plant Biology 51 , 1138 – 1148 . Google Scholar CrossRef Search ADS PubMed Craig SM , Reagon M , Resnick LE , Caicedo AL . 2014 . Allele distributions at hybrid incompatibility loci facilitate the potential for gene flow between cultivated and weedy rice in the US . PLoS ONE 9 , e86647 . Google Scholar CrossRef Search ADS PubMed Cui RR , Dai WM , Qiang S , Song XL . 2013 . Gene flow from transgenic glufosinate-resistant rice II you 86B and its restorer line 86B to weedy rice . Jiangsu Journal of Agriculture Sciences 29 , 708 – 714 (in Chinese). Dale PJ , Clarke B , Fontes EM . 2002 . Potential for the environmental impact of transgenic crops . Nature Biotechnology 20 , 567 – 574 . Google Scholar CrossRef Search ADS PubMed Delouche JC , Labrada R , Rosell C . 2007 . Weedy rices – origin, biology, ecology and control . Rome, Italy : Food and Agriculture Organization of the United Nations . Furukawa T , Maekawa M , Oki T , Suda I , Iida S , Shimada H , Takamure I , Kadowaki K . 2007 . The Rc and Rd genes are involved in proanthocyanidin synthesis in rice pericarp . The Plant Journal 49 , 91 – 102 . Google Scholar CrossRef Search ADS PubMed Gealy DR . 2005 . Gene movement between rice (Oryza sativa) and weedy rice (Oryza sativa): a U.S. temperate rice perspective . In: Gressel J , ed. Crop ferality and volunteerism . Boca Raton, FL : CRC Press , 323 – 354 . Google Scholar CrossRef Search ADS Gressel J . 2002 . Molecular biology of weed control , vol. 1 . Boca Raton, FL : CRC Press . Gressel J , Valverde BE . 2009 . A strategy to provide long-term control of weedy rice while mitigating herbicide resistance transgene flow, and its potential use for other crops with related weeds . Pest Management Science 65 , 723 – 731 . Google Scholar CrossRef Search ADS PubMed Gross BL , Reagon M , Hsu SC , Caicedo AL , Jia Y , Olsen KM . 2010 . Seeing red: the origin of grain pigmentation in US weedy rice . Molecular Ecology 19 , 3380 – 3393 . Google Scholar CrossRef Search ADS PubMed ISAAA . 2016 . Global status of commercialized biotech/GM Crops: 2016. ISAAA Brief No. 52 . Ithaca, NY : ISAAA . Ishikawa R , Toki N , Imai K , Sato YI , Yamagishi H , Shimamoto Y , Ueno K , Morishima H , Sato T . 2005 . Origin of weedy rice grown in Bhutan and the force of genetic diversity . Genetic Resources and Crop Evolution 52 , 395 – 403 . Google Scholar CrossRef Search ADS Lee D , Lupotto E , Powell W . 2009 . G-string slippage turns white rice red . Genome 52 , 490 – 493 . Google Scholar CrossRef Search ADS PubMed Li LF , Li YL , Jia Y , Caicedo AL , Olsen KM . 2017 . Signatures of adaptation in the weedy rice genome . Nature Genetics 49 , 811 – 814 . Google Scholar CrossRef Search ADS PubMed Lu BR , Yang C . 2009 . Gene flow from genetically modified rice to its wild relatives: assessing potential ecological consequences . Biotechnology Advances 27 , 1083 – 1091 . Google Scholar CrossRef Search ADS PubMed Luo D , Xu H , Liu Z , et al. 2013 . A detrimental mitochondrial–nuclear interaction causes cytoplasmic male sterility in rice . Nature Genetics 45 , 573 – 577 . Google Scholar CrossRef Search ADS PubMed Marambe B . 2005 . Properties of rice growing in abandoned paddies in Sri Lanka . In: Gressel J , ed. Crop ferality and volunteerism . Boca Raton, FL : CRC Press , 295 – 303 . Google Scholar CrossRef Search ADS Messeguer J , Marfa V , Catala MM , Guiderdoni E , Mele E . 2004 . A field study of pollen-mediated gene flow from Mediterranean GM rice to conventional rice and the red rice weed . Molecular Breeding 13 , 103 – 112 . Google Scholar CrossRef Search ADS Olsen KM , Caicedo AL , Jia Y . 2007 . Evolutionary genomics of weedy rice in the USA . Journal of Integrative Plant Biology 49 , 811 – 816 . Google Scholar CrossRef Search ADS Prathepha P . 2009 . Seed morphological traits and genotypic diversity of weedy rice (Oryza sativa f. spontanea) populations found in the Thai Hom Mali rice fields of north-eastern Thailand . Weed Biology and Management 9 , 1 – 9 . Google Scholar CrossRef Search ADS Rong J , Lu BR , Song Z , Su J , Snow AA , Zhang X , Sun S , Chen R , Wang F . 2007 . Dramatic reduction of crop-to-crop gene flow within a short distance from transgenic rice fields . New Phytologist 173 , 346 – 353 . Google Scholar CrossRef Search ADS PubMed Rong J , Wang F , Song Z , Su J , Chen R , Lu BR . 2012 . Scale effect on rice pollen-mediated gene flow: implications in assessing transgene flow from genetically engineered plants . Annals of Applied Biology 161 , 3 – 11 . Google Scholar CrossRef Search ADS Serrat X , Esteban R , Peñas G , Català MM , Melé E , Messeguer J . 2013 . Direct and reverse pollen-mediated gene flow between GM rice and red rice weed . AoB Plants 5 , plt050 . Google Scholar CrossRef Search ADS Shivrain VK , Burgos NR , Anders MM , Rajguru SN , Moore J , Sales MA . 2007 . Gene flow between Clearfield™ rice and red rice . Crop Protection 26 , 349 – 356 . Google Scholar CrossRef Search ADS Shivrain VK , Burgos NR , Gealy DR , Sales MA , Smith KL . 2009a . Gene flow from weedy red rice (Oryza sativa L.) to cultivated rice and fitness of hybrids . Pest Management Science 65 , 1124 – 1129 . Google Scholar CrossRef Search ADS Shivrain VK , Burgos NR , Sales MA , Mauromoustakos A , Gealy DR , Smith KL , Blank HL , Jia M . 2009b . Factors affecting the outcrossing rate between Clearfield™ rice and red rice (Oryza sativa) . Weed Science 57 , 394 – 403 . Google Scholar CrossRef Search ADS Singh V , Burgos N , Singh S , Abugho S , Earnest L , Gbur E , Scott R . 2016 . Herbicide and winter flood treatments for controlling volunteer rice off-season . Crop Protection 79 , 87 – 96 . Google Scholar CrossRef Search ADS Song X , Liu L , Wang Z , Qiang S . 2009 . Potential gene flow from transgenic rice (Oryza sativa L.) to different weedy rice (Oryza sativa f. spontanea) accessions based on reproductive compatibility . Pest Management Science 65 , 862 – 869 . Google Scholar CrossRef Search ADS PubMed Song X , Wang Z , Qiang S . 2011 . Agronomic performance of F1, F2 and F3 hybrids between weedy rice and transgenic glufosinate-resistant rice . Pest Management Science 67 , 921 – 931 . Google Scholar CrossRef Search ADS PubMed Song ZP , Lu BR , Wang B , Chen JK . 2004 . Fitness estimation through performance comparison of F1 hybrids with their parental species Oryza rufipogon and O. sativa . Annals of Botany 93 , 311 – 316 . Google Scholar CrossRef Search ADS PubMed Sudianto E , Beng-Kah S , Ting-Xiang N , Saldain NE , Scott RC , Burgos NR . 2013 . Clearfield® rice: its development, success, and key challenges on a global perspective . Crop Protection 49 , 40 – 51 . Google Scholar CrossRef Search ADS Sun G , Dai W , Cui R , Qiang S , Song X . 2015 . Gene flow from glufosinate-resistant transgenic hybrid rice Xiang 125S/Bar68-1 to weedy rice and cultivated rice under different experimental designs . Euphytica 204 , 211 – 227 . Google Scholar CrossRef Search ADS Sun J , Qian Q , Ma DR , Xu ZJ , Liu D , Du HB , Chen WF . 2013 . Introgression and selection shaping the genome and adaptive loci of weedy rice in northern China . New Phytologist 197 , 290 – 299 . Google Scholar CrossRef Search ADS PubMed Sweeney MT , Thomson MJ , Pfeil BE , McCouch S . 2006 . Caught red-handed: Rc encodes a basic helix-loop-helix protein conditioning red pericarp in rice . The Plant Cell 18 , 283 – 294 . Google Scholar CrossRef Search ADS PubMed Tang W , Chen H , Xu C , Li X , Lin Y , Zhang Q . 2006 . Development of insect-resistant transgenic indica rice with a synthetic cry1C* gene . Molecular Breeding 18 , 1 – 10 . Google Scholar CrossRef Search ADS Valverde BE . 2005 . The damage by weedy rice: can feral rice remain undetected ? In: Gressel J , ed. Crop ferality and volunteerism . Boca Raton, FL : CRC Press , 279 – 294 . Google Scholar CrossRef Search ADS Valverde BE . 2013 . Is herbicide resistant rice the ultimate solution for controlling weedy rice? Experiences from the Americas . Korean Journal of Weed Science 33 , 11 – 23 . Xiong HB , Xu HY , Xu Q , et al. 2012 . Origin and evolution of weedy rice revealed by inter-subspecific and inter-varietal hybridizations in rice . Molecular Plant Breeding 10 , 131 – 139 . Yang J , Wang J , Cao Q , Chen ZD , Tang LH , Zhong WG . 2009 . Development of a functional marker to red pericarp gene in weedy rice (Oryza sativa L. f. spontaneous) . Molecular Plant Breeding 7 , 721 – 726 . Yuan LP . 2002 . Science of hybrid rice . Beijing, China : China Agriculture Press . Yuan QH , Shi L , Wang F , Cao B , Qian Q , Lei XM , Liao YL , Liu WG , Cheng L , Jia SR . 2007 . Investigation of rice transgene flow in compass sectors by using male sterile line as a pollen detector . Theoretical and Applied Genetics 115 , 549 – 560 . Google Scholar CrossRef Search ADS PubMed Zhang JX , Lu ZM , Dai WM , Song XL , Peng YF , Valverde BE , Qiang S . 2015 . Cytoplasmic-genetic male sterility gene provides direct evidence for some hybrid rice recently evolving into weedy rice . Scientific Reports 5 , 10591 . Google Scholar CrossRef Search ADS PubMed Zhang L , Dai W , Wu C , Song X , Qiang S . 2012 . Genetic diversity and origin of Japonica- and Indica-like rice biotypes of weedy rice in the Guangdong and Liaoning provinces of China . Genetic Resources and Crop Evolution 59 , 399 – 410 . Google Scholar CrossRef Search ADS Zhang W , Linscombe SD , Webster E , Tan S , Oard J . 2006 . Risk assessment of the transfer of imazethapyr herbicide tolerance from Clearfield rice to red rice (Oryza sativa) . Euphytica 152 , 75 – 86 . Google Scholar CrossRef Search ADS Zuo J , Zhang L , Song X , Dai W , Qiang S . 2011 . Innate factors causing differences in gene flow frequency from transgenic rice to different weedy rice biotypes . Pest Management Science 67 , 677 – 690 . Google Scholar CrossRef Search ADS PubMed © The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Experimental Biology. All rights reserved. For permissions, please email: journals.permissions@oup.com This article is published and distributed under the terms of the Oxford University Press, Standard Journals Publication Model (https://academic.oup.com/journals/pages/open_access/funder_policies/chorus/standard_publication_model)

Journal

Journal of Experimental BotanyOxford University Press

Published: Jun 5, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off