Introduction

Genetic engineering has been used for the introduction of new traits to specific crops1,2,3,4,5, which is a good method to breed new and improved cultivars, and satisfies the increasing consumer demand for a consistent supply of high-quality crops with fewer blemishes from insects and diseases and reduced pesticide residues6,7. The steady growth of genetically modified (GM) plants in China attests to their advantages for producers8,9,10,11, and “No. 1 document” of central government in 2016 first mentioned that GM technology should be cautious introduced in agriculture, showing the determination to extend GM crops in China. However, public concerns have been raised about the environmental and food safety of GM crop planting3,12,13,14. The “No. 1 document” in 2017 did not mention the GM technology after the positive policies of past 3 years, revealing the big hurdle in applying the GM technology over the China.

One of the concerns to introduce GM crops into environment is the flow of transgene-containing pollen to the surroundings15,16,17. Gene flow itself is neutral and usually occurs between conventional crops. Gene flow is of great importance in seed production in conventional crops, and the effect of gene flow depends on the differences between the recipient and donor plants3,6,11. Pollen-mediated gene flow (PGF) makes the movement of introduced genes from GM crops to other plants possible, which usually occurs if GM pollen is delivered to the receptive stigma by insects or wind12,18,19,20,21. Thus, limiting the dispersal of GM pollen is the key to prevent cross-pollination between GM and non-GM crops5,22,23. Physical isolation measure (separation net and crop) is an effective way to constrain pollen flow with low activity of pollinators, which in turn constrains the PGF24,25,26,27. Meanwhile, physical isolation distance is also important for attenuating the gene flow risk because of the decline of PGF with increasing distance14,21,26,28,29. What has been less studied is the effectiveness of various isolation measures to constrain PGF. Unlike the developed countries, farmers in China possess relatively small cultivated area, even less than 500 m2 for per farmer, and the phenomenon that farmers select different crop varieties is very common. Thus, the coexistence of GM and non-GM crops is very likely in China, which increases the risk of PGF. Our study determining the effectiveness of various isolation measures to constrain PGF is particularly important in China, which is not only benefit for decreasing the risk of PGF, but also has potential application value in practical production and the scientific researches of GM crops.

Cotton (Gossypium hirsutum L.) is primarily grown as an irrigated annual crop in the Changjiang River Region, the Yellow River Region and the North-western Region of China. GM cotton, producing insecticidal proteins from Bacillus thuringiensis (Bt) was first introduced in China in 1997 to control lepidopteran pests30,31. Previous studies have shown that PGF in cotton could be greatly limited by low activity of pollinators, isolation distance and low speed of wind19,21,32,33. Long isolation distance cannot be applied in practical production in China because of the limited cultivated area. In the present study, we measured pollen dispersal distance and PGF in GM cotton with 90, 80, 60 and 40 holes/cm2 separation nets and Sorghum bicolor, Zea mays and Lycopersicon esculentum separation crops by pollen grain dyeing method and molecular techniques respectively to determine the effectiveness of various physical isolation measures to constrain pollen movement and PGF. Our goal was to find an effective physical isolation measure to constrain PGF.

Results

Greenhouse and field observation

The agronomic behavior of GM cotton and Shiyuan321 was similar between the greenhouse and field trials, and flowering was synchronous between two types of cottons throughout the flowering period, thus ensuring ample opportunity for pollen transfer. In greenhouses, wind velocities (most distance points: 0.5–0.6 m/s) were stable (Figure S1). In field, no pesticide was used to control pests, and Apolygus lucorum, Aphis gossypii, Apis mellifera, Pieris rapae and Harmonia axyridis were observed in both years.

Evaluation of pollen dispersal with different physical isolation measures

Pollen grain dyeing was conducted in a greenhouse during 2013 to assess the pollen flow with various physical isolation measures. As shown in Table 1, dyed pollens with various separation nets and crops were observed most at the distance of 3.2 and 4.0 m, respectively. The 90 holes/cm2 separation net and S. bicolor separation crop could minimize the pollen dispersal, whereas the constraints of pollen dispersal by 40 holes/cm2 separation net and L. esculentum separation crop were the weakest, and dyed pollens were observed at all distance points with no isolation measure (CK).

Table 1 The comparison of dyed pollen number with different physical isolation measures in greenhouse during 2013.
Full size table

Assessment of pollen-mediated gene flow in greenhouse experiment

A total of 7,200 seeds (60 seeds/sample × 4 separation nets × 7 distances × 2 yr + 60 seeds/sample × 4 separation crops × 8 distances × 2 yr) were sampled to assess PGF by DNA analyses in greenhouse experiment.

Hybridization was detected during the 2 yr of the study, and varied by the physical isolation measure and distance (Table 2). More specifically, the average PGF of all sampled distances during the 2 yr were 3.69%, 4.40%, 6.31% and 8.81% with 90, 80, 60 and 40 holes/cm2 separation nets, respectively (F3,164 = 4.534, P = 0.004), 4.69%, 8.85%, 13.13% and 19.17% with S. bicolor, Z. mays, L. esculentum separation crops and no isolation measure (CK), respectively (F3,188 = 10.766, P < 0.001). Net isolation (5.80%) was more effective to constrain PGF than crop isolation (8.89%) (t = 2.710, df = 226.845, P = 0.007). As shown in Table 2 and Fig. 1, PGF tended to decline with increasing distance, ranging from 16.67% at 0.8 and 3.2 m to 0.00% at 19.2 m in 2013 and from 18.33% at 0.8 m to 0.00% at 25.6 m in 2014 with 40 holes/cm2 separation net, from 43.33% at 4.0 m to 6.67% at 28.8 and 40.0 m in 2013 and from 53.33% at 4.0 m to 0.00% at 40.0 m in 2014 with no isolation measure (CK). Line (y = lnx) was the best-fit regression curve in Fig. 1 (90 holes/cm2 R2 = 0.8976; 80 holes/cm2 R2 = 0.8914; 60 holes/cm2 R2 = 0.7942; 40 holes/cm2 R2 = 0.9194; S. bicolor R2 = 0.9040; Z. mays R2 = 0.8633; L. esculentum R2 = 0.8477; CK R2 = 0.8222).

Table 2 Pollen-mediated gene flow frequency with different physical isolation measures in greenhouse during 2013 and 2014 (%).
Full size table
Figure 1
figure 1

Regression analysis for PGF with different separation nets (A) and crops (B) in greenhouse experiment as a function of the distance. The determination coefficient (R2) between PGF and distance was obtained using SigmaPlot 10.0 software. Regression analysis was based on the data of the 2 yr study.

Full size image

Assessment of pollen-mediated gene flow in field experiment

A total of 5,040 seeds (60 seeds/sample × 3 physical isolation measures × 7 distances × 2 directions × 2 yr) were sampled to assess PGF by DNA analyses in field experiment.

Similar to greenhouse experiment, hybridization was varied by the physical isolation measure and distance in field experiment. More specifically, the average PGF of all sampled distances during the 2 yr were 2.86%, 3.99% and 13.51% with 90 holes/cm2 separation net, S. bicolor separation crop and no isolation measure (CK), respectively (F2,249 = 51.017, P < 0.001), however PGF with net isolation and crop isolation shown no significant difference (t = 1.342, df = 166, P = 0.182)(Table 3). As shown in Table 3 and Fig. 2, PGF also declined with increasing distance, ranging from 29.17% at 4.8 m to 6.67% at 25.6 m in 2014 and from 22.50% at 4.0 m to 5.83% at 19.2 and 25.6 m in 2015 with no isolation measure (CK), whereas PGF with 90 holes/cm2 separation net and S. bicolor separation crop were relatively low among different distance points. Line (y = lnx) was the best-fit regression curve in Fig. 2 (CK R2 = 0.8560; S. bicolor R2 = 0.8748; 90 holes/cm2 R2 = 0.8832).

Table 3 Pollen-mediated gene flow frequency with different physical isolation measures in field during 2014 and 2015 (%).
Full size table
Figure 2
figure 2

Regression analysis for PGF with different physical isolation measures in field experiment as a function of the distance. The determination coefficient (R2) between PGF and distance was obtained using SigmaPlot 10.0 software. Regression analysis was based on the data of the 2 yr study.

Full size image

Discussion

Efficient pollen dispersal leading to the fertilization, is the major problem when considering the risk of transgene escape in particular for GM crops that can hybridize with non-GM lines or related wild species1,34. When the desired gene gets introgressed with other species by recombination, particularly weedy relatives and feral and volunteer crop plants, there is a potential risk for new and more invasive weeds to arise35,36,37, and domestication traits can also make weeds less invasive38. Regulation approaches of GM crops in Europe are mainly based on the process of making GM crops, whereas those of Canada are based on the GM product. The regulation of GM crops in USA focuses primarily on the characteristics of the product, and a useful oversight of the regulatory process is maintained. China implements a relative pragmatic approach to GM crop regulations, which are basically based on the products and depended on the economic interest of a given application3. Thus, preventing gene flow is important for risk management of releasing GM crops, and the physical isolation measure may be a good choice in practical production to constrain the pollen dispersal24,25,26,27. In the present study, we determined the pollen dispersal with various physical isolation measures and measured PGF between the GM cotton and non-GM cotton in greenhouse and field experiments during two years using the molecular technique, which was confirmed to be more reliable to determine PGF than biological assays because of the weaker resistance of hybrids for some assays21. Previous studies on PGF were mainly conducted in fields14,39,40,41, whereas our study was conducted in greenhouse and field, which provided a more complementary assessment.

Zhang et al.8 found that PGF from GM cotton decreased exponentially as dispersal distance increased, ranging from 10.13% at 1 m to 0.04% at 50 m for the tfd A gene, 8.16% at 1 m to 0.08% at 20 m for the Bt gene. Our early study21 found that PGF with Shiyuan321 went from 30% to <5% within 1.6 m from the test plot, and a low level from 8.33% to <1% PGF continued to occur sporadically out to a distance of 25.6 m. Van Deynze et al.33 found PGF was independent of direction from the source plot and declined exponentially with increasing distance from 7.65% at 0.3 m to <1% beyond 9 m with A. melifera. In the present study, we collected a total of 12,240 cotton seeds during the 2-yr study and then extracted DNA from seeds directly to determine PGF. Similar to other previous studies14,26,42, we found PGF with any isolation measure in greenhouse and field tended to decline significantly with increasing distance because of the decreased proportion of GM pollens over the distance.

The heights of S. bicolor, Z. mays and L. esculentum were nearly 3.2, 2.2 and 1.5 m respectively at the pollination of cottons. Cotton pollen diameter is 120–150 μm, and the hole sizes of four kind separation nets were bigger than the size of cotton pollen. Our results revealed that 90 holes/cm2 separation net and S. bicolor separation crop could constrain the pollen dispersal effectively, whereas dyed pollens were observed at all distance points with no isolation measure (CK). The constraint of pollen dispersal was benefit for reducing the environment risk of transgene escape. Similar to pollen grain dyeing results, we found that PGF in greenhouse and field was constrained by various physical isolation measures, with 90 holes/cm2 separation net and S. bicolor separation crop having lower PGF rates than other isolation measures. Meanwhile net isolation was more effective to constrain PGF than crop isolation no matter in greenhouse and field. The height of separation crop might be the key to constrain PGF. In general, the 20 m cotton buffer or 50 m bare ground is accepted to provide sufficient containment of GM cotton in small scale trials in Australia, compared to sometimes up to 800 m for foundation seed production with additional pollen buffer rows surrounding the crop in USA18. Zhang et al.8 confirmed that the farthest distance of pollen dispersal from cotton was 50 m, and recommended a 60 m buffer zone to limit pollen dispersal from small-scale field tests. However, isolation distance requirement may vary under different environmental conditions, especially wind and pollinator activity14,19,21,33. In the present study, we could still detect the gene flow occurring at 25.6 m both with 90 holes/cm2 separation net and S. bicolor separation crop in field, and the further isolation distance with S. bicolor buffer or double separation nets may be needed to constrain PGF in agricultural production and scientific researches. The application of nets may be also benefit for controlling many insect plagues, whereas the disadvantage of this method is the high expense, and the price of nets increases with the number of holes/cm2. The application of long isolation distance to constrain PGF is unrealistic in China because of the limited cultivated area, and agricultural ministries of each level should explore subsidy mechanics to promote the application of plant buffer or other physical isolation measures in practical production.

We observed that PGF occurred at a higher frequency in greenhouse than field during our 2-yr study, which might be related to the automatic wind in greenhouse. PGF or cross-fertilization in field might be more complex than those in greenhouse, which was related with changeable natural condition19,21,33,43,44. Meanwhile, the planting scale of GM crops and the genetic relationship between GM crops and adjacent crops would also influence PGF21,45. Thus, we should consider the local natural condition and agroecological environment before setting the proper isolation measure to constrain PGF. The development of transgenic technology in China has already drawn heated public concerns about the possible risk of GM food in recent years, and central government has a long way to extend consuming transgenic crops. The physical isolation measure will play a more important role in future agricultural production.

Methods

Plant materials

The inbred line Shiyuan321 (pollen receptor) and GM cotton (pollen donor) were obtained from the Cotton Research Institute, Chinese Academy of Agricultural Sciences, and were purified by consecutive two-generations of self-crosses, to ensure that each line was homozygous. The GM cotton was obtained by transferring the insect resistance gene Cry I Ac and the glyphosate resistance gene CP4 EPSPS from Agrobacterium tumefaciens into the embryonic calli of Shiyuan321 via microprojectile bombardment. The hemizygous progeny can be easily identified by PCR analyses because of the high resistance gene expression23,46,47,48.

Greenhouse layout and procedures

To evaluate the impacts of physical isolation measures on PGF, the greenhouse trials (L60 m × W8 m × H4 m) were carried out in the Shangzhuang Experimental Station of China Agricultural University (40°14′N, 116°19′E), Beijing, China during 2013 and 2014. Greenhouse provided a relatively closed environment, which was not benefit for pollen dispersal. Thus, several electric fans (2 m height) were installed every 4.5 m, turned on (from 6:00 to 18:00) to promote pollen dispersal, and wind speeds were monitored at 8:00, 12:00 and 16:00 (as 3 replications) using an anemograph (ZRQF-F303, Beijing Detector Instrument Co., Beijing, China)19. The agronomic behavior of GM cotton and Shiyuan321 was similar to a standard cultivar in the field. Plants were monitored at least twice weekly for florescence emergence and anthesis from early July through late August (24–29 °C, full-bloom stage for all plants) to ensure the flowering synchrony between two types of cottons49. Treatment rows were spaced 0.8 m apart and plants within rows were spaced 0.5 m apart.

Separation net treatment

As shown in Fig. 3, 90, 80, 60 and 40 holes/cm2 separation nets (3 m height) were installed in one greenhouse to evaluate the isolation effect. Different treatments were separated by white cloth from the ceiling to the ground. The GM cotton was planted in a 9.6 m × 8 m square in the middle of a 60 m × 8 m square area of Shiyuan321. Pollen grain dyeing was conducted during the full-bloom stage to measure the pollen dispersal according to the method of Wang et al.50 during 2013. Pollens from 15 upper flowers of the first row GM cottons were dyed with erythrosine, and dyed pollens on flowers were counted at the distance of 0.8, 1.6, 3.2, 6.4, 12.8, 19.2 and 25.6 m from GM cottons after 3 days. No significant differences in PGF were observed for flower positions on the pollen receptors32, so seeds from pollen receptors were obtained at harvest by hand from lower, middle, and upper flowers as 3 replications at the distance of 0.8, 1.6, 3.2, 6.4, 12.8, 19.2 and 25.6 m for PGF evaluation when flower mature but before excessive shattering33.

Figure 3
figure 3

Greenhouse trial design to measure the impacts of different separation nets on pollen-mediated gene flow during 2013 and 2014. The GM cotton was planted in a 9.6 m × 8 m square in the middle of a 60 m × 8 m square field of Shiyuan321. Electric fans were installed to promote pollen dispersal.

Full size image

Separation crop treatment

As shown in Fig. 4, Sorghum bicolor (Yajin1), Zea mays (Nongda80) and Lycopersicon esculentum (Zhefen702) separation crops and no isolation measure (CK) were set up in two greenhouses to evaluate the isolation effect. Different treatments were separated by white cloth from the ceiling to the ground. Treatment rows of separation crops were spaced 0.8 m apart and plants within rows were spaced 0.5 m apart. The separation crop was planted in a 4 m × 8 m square in the middle of a 46.4 m × 8 m square area of cotton. Pollen grain dyeing was similar to separation net treatment, and dyed pollens were counted at the distance of 4, 4.8, 6.4, 9.6, 16.0, 22.4, 28.8 and 40.0 m. At harvest, seeds from pollen receptors were obtained at the distance of 4, 4.8, 6.4, 9.6, 16.0, 22.4, 28.8 and 40.0 m for PGF evaluation.

Figure 4
figure 4

Greenhouse trial design to measure the impacts of different separation crops on pollen-mediated gene flow during 2013 and 2014. The separation crop was planted in a 4 m × 8 m square in the middle of a 46.4 m × 8 m square field of cotton. Electric fans were installed to promote pollen dispersal.

Full size image

Design and maintenance of experimental field plot

The field trial was also carried out in the Shangzhuang Experimental Station during 2014 and 2015. Treatment rows of both cottons and separation crops were spaced 0.8 m apart and plants within rows were spaced 0.5 m apart. As shown in Fig. 5, the 90 holes/cm2 separation net (3 m height), no isolation measure (CK) and S. bicolor separation crop were set up to evaluate the isolation effect. The GM cotton was planted in a 9.6 m × 4.5 m square in the middle of a 62.4 m × 4.5 m field. At harvest, seeds from pollen receptors were obtained from lower, middle, and upper flowers as 3 replications at the distance of 4, 4.8, 6.4, 9.6, 12.8, 19.2 and 25.6 m for PGF evaluation.

Figure 5
figure 5

Field trial design to measure the impacts of different physical isolation measures on pollen-mediated gene flow during 2014 and 2015. The 90 holes/cm2 separation net, no isolation measure (CK) and S. bicolor separation crop were set up to evaluate the isolation effect. The GM cotton was planted in a 9.6 m × 4.5 m square in the middle of a 62.4 m × 4.5 m square field.

Full size image

PGF evaluation

For each distance point, 60 seeds [20 seeds from each height of flowers (lower, middle, and upper)] were screened for hybrid confirmation by polymerase chain reaction (PCR) amplification of the fragment sequence of Cry I Ac. DNA was extracted using the cetyltrimethyl ammonium bromide (CTAB) protocol51 with the addition of 20 g/L polyvinylpyrrolidone (PVP) in the extraction buffer. A 1 μL sample of DNA was used to perform spectroscopic quantitation using a NanoDrop 2000 spectrophotometer (Thermo Fisher, USA). Then, 80 ng DNA samples were used to amplify the target gene with primers of 5′-GAAGGATTGAGCAATCTCTAC-3′ and 5′-CAATCAGCCTAGTAAGGTCGT-3′19. The PCR reaction was typically cycled 30 times at 94 °C for 60 s, 56 °C for 60 s, and 72 °C for 90 s19. Routine screening for PCR assay was performed by agarose gel electrophoresis (10 g agarose/L) with GM cotton as positive control and Shiyuan321 as negative control.

$${\rm{Gene}}\,{\rm{flow}}\,( \% )=[({\rm{number}}\,{\rm{of}}\,{\rm{seed}}\,{\rm{progeny}}\,{\rm{expressing}}\,{\rm{Cry}}\,{\rm{I}}\,{\rm{Ac}})/({\rm{number}}\,{\rm{of}}\,{\rm{tested}}\,{\rm{seeds}})]\times 100$$

Statistical analysis

All statistical analyses were conducted using the SPSS 16.0 software (IBM, Armonk, NY). All descriptive statistics were given as the mean values and standard errors of the mean (mean ± SE). The data were analyzed using Tukey HSD test and intendent t test at the P = 0.05 level of significance. Regression analyses (a potential dispersal model: y = lnx) were performed with SigmaPlot 10.0 software (Systat Software Inc., San Jose), and the determination coefficient (R2) between PGF and distance was obtained from SigmaPlot 10.0 software.