measuring gene flow in the cultivation of transgenic cotton (gossypium hirsutum l.)

Gene Flow in Transgenic Cotton 11

MOLECULAR BIOTECHNOLOGY Volume 31, 2005

RESEARCH

11

Molecular Biotechnology © 2005 Humana Press Inc. All rights of any nature whatsoever reserved. 1073–6085/2005/31:1/011–020/$30.00

*Author to whom all correspondence and reprint requests should be addressed. 1The Institute of Environmental and Human Health, andDepartment of Environmental Toxicology, Texas Tech University, Lubbock, TX 79409-1163, USA; E-mail: [email protected]; 2KeyLaboratory of Cotton Genetic Improvement of the Ministry of Agriculture, Cotton Research Institute, Chinese Academy of AgriculturalSciences, Anyang, Henna 455112, China;3Institute of Heze Agricultural Science, Heze, Shandong 273000, China; 4Henan Institute of Sci-ence and Technology, Xinxiang, Hehan 453003, China.

Abstract

Measuring Gene Flow in the Cultivation of Transgenic Cotton(Gossypium hirsutum L.)

Bao-Hong Zhang,1,2,4,* Xiao-Ping Pan,1,4 Teng-Long Guo,3 Qing-Lian Wang,4

and Todd A. Anderson1

Transgenic Bt cotton NewCott 33B and transgenic tfd A cotton TFD were chosen to evaluate pollendispersal frequency and distance of transgenic cotton (Gossypium hirsutum L.) in the Huanghe Valley Cot-ton-producing Zone, China. The objective was to evaluate the efficacy of biosafety procedures used toreduce pollen movement. A field test plot of transgenic cotton (6 × 6 m) was planted in the middle of anontransgenic field measuring 210 × 210 m. The results indicated that the pollen of Bt cotton or tfd Acotton could be dispersed into the environment. Out-crossing was highest within the central test plot whereprogeny from nontransgenic plants, immediately adjacent to transgenic plants, had resistant plant progenyat frequencies up to 10.48%. Dispersal frequency decreased significantly and exponentially as dispersaldistance increased. The flow frequency and distance of tfd A and Bt genes were similar, but the pollen-mediated gene flow of tfd A cotton was higher and further to the transgenic block than that of Bt cotton (χ2

= 11.712, 1 degree of freedom, p < 0.001). For the tfd A gene, out-crossing ranged from 10.13% at 1 m to0.04% at 50 m from the transgenic plants. For the Bt gene, out-crossing ranged from 8.16% at 1 m to 0.08%at 20 m from the transgenic plants. These data were fit to a power curve model: y = 10.1321x–1.4133 with acorrelation coefficient of 0.999, and y = 8.0031x–1.483 with a correlation coefficient of 0.998, respectively. Inthis experiment, the farthest distance of pollen dispersal from transgenic cotton was 50 m. These resultsindicate that a 60-m buffer zone would serve to limit dispersal of transgenic pollen from small-scale fieldtests.

Index Entries: Transgenic cotton; pollen dispersal; Bt gene; tfd A gene; biosafety; transgene.

1. IntroductionGene flow is one of the primary safety issues

associated with the release of transgenic plantsinto the environment (1–5). Unrestricted geneflow can, through introgression with compatiblesecondary recipients, give rise to new weedy andferal genotypes that may impinge on the long-term stability of natural ecosystems. Thus, sincethe initial release of transgenic plants, manyecologists and other scientists have focused con-siderable attention on this issue (6–12).

Cotton is one of the most important fiber andeconomic crops, and was among the first geneti-cally modified crops to be commercialized (13–17). The two major kinds of transgenic cottonplanted commercially around the world are insect-resistant and herbicide-resistant cotton. Insect-resis-tant Bt cotton was welcomed by cotton farmersbecause of its high efficiency in controlling cottonpests. To date, transgenic Bt cotton has beenplanted commercially in the United States, China,India, Australia, Mexico, Argentina, South Africa,

11_20_Zhang_MB04_0094 8/8/05, 1:26 PM11

12 Zhang et al.


and Spain. In the United States, since 2000, morethan 75–80% of the total cotton area was plantedwith transgenic varieties. In China, at least sixtransgenic cotton varieties are planted (15). Othercountries, such as Colombia, Bolivia, El Salva-dor, Greece, Israel, Paraguay, Thailand, and Zim-babwe, are in the process of testing Bt cotton(13,14,17,18). In addition, transgenic herbicide-and disease-resistant cotton has also been plantedin the United States and China (13,17,19).

Although great progress has been made throughgenetically modified cotton, the environmentalrisks and biosafety of transgenic cotton haverarely been reported. American and Australianscientists have studied in small trials transgeniccotton pollen dispersal using transgenic cottoncarrying a marker gene (20,21). Both groups re-ported that pollen transmission rates were in-versely related to the distance from transgenicplants to nontransgenic plants. However, transgenicgene flow is environment-dependent, and may bespecies-specific with respect to genotype, loca-tion, and season (22). Thus, it is important tostudy gene flow in different environments, includ-ing different countries. Although it is unlikely thatdifferent transgenes significantly affect the mor-phology of the transgenic, which directly relatesto gene flow, transgenes may affect the agronomicpractices for transgenic crops (such as applicationof pesticides), especially for insect-resistanttransgenic plants such as Bt cotton. This in turncould affect pollinator numbers and behavior andsubsequently pollen transmission. Thus, it is impor-tant to study different types of transgenic plants,especially commercial transgenic crops, underdifferent environmental conditions.

We have obtained transgenic Bt pest-resistantcotton and tfd A herbicide-resistant cotton byAgrobacterium-mediated transformation. Severalgreenhouse experiments and field trials indicatedthat Bt cotton was highly resistant to pests. Cot-ton farmers adopting Bt cotton save 70–80% la-bor and pesticides compared with plantingnontransgenic cotton. The herbicide 2,4-dichlo-rophenoxyacetic acid (2,4-D) is one type of plantgrowth regulator that can effectively kill dicoty-ledon weeds. It is widely applied to monocotyle-

dons (such as wheat, corn) to kill dicotyledonweeds in fields. As a dicotyledon, cotton is highlysensitive to 2,4-D. Low concentrations of 2,4-D,even spray drift from other fields, can lead todamage and abnormal growth. Thus, scientistsand plant breeders have tried to build up thecotton’s resistance to 2,4-D by using tfd A gene.We have obtained transgenic 2,4-D-resistant cot-ton, and field experiments indicated that thetransgenic tfd A cotton could grow well if 800 mg/L of 2,4-D was sprayed. The tfd A cotton exhib-ited 50- to 100-fold more tolerance to 2,4-D com-pared with nontransformed plants (23,24). Inaddition, these transgenic Bt and tfd A cotton havebeen planted in China for 10 yr.

The objective of the research reported here wasto estimate the frequency and distance of pollendispersal from two types of transgenic cotton (in-sect-resistant Bt cotton and herbicide-resistant tfdA cotton) under Chinese environmental condi-tions, and to evaluate biosafety procedures usedto reduce pollen movement.

2. Materials and Methods2.1. Plant Materials

Commercial transgenic insect-resistant cottoncultivar NewCott 33B and transgenic herbicide-resistant cotton line TFD were selected for thisexperiment. NewCott 33B carries a Bt gene andexpresses cry IA insecticidal protein for resis-tance to Heliothis virescenes and Helicoverpa zea(Lipidoptera: Noctuidae). The transgenic lineTFD carries the tfd A gene and expresses 2,4-Dmonooxygenase for resistance to the herbicide2,4-D. NewCott 33B was imported by the JidaCompany and authorized for planting in 1997, andpassed the biosafety evaluation by the Chinesegovernment in 1998. The transgenic cotton lineTFD was obtained from the Cotton Research In-stitute in Shanxi Province, and then was purifiedby self-crosssing at the Cotton Research Institute,Chinese Academy of Agricultural Sciences. Thenontransgenic cultivar CCRI 19 was obtainedfrom the Cotton Research Institute, ChineseAcademy of Agricultural Sciences. All materialswere purified by consecutive two-generation self-crosses, to ensure that each line was homozygousfor the transgenes.

11_20_Zhang_MB04_0094 8/8/05, 1:26 PM12



2.2. Experimental Design for Field Trials

The experiments were conducted in 1997 and1998 at the Huanghe Valley Cotton-producingZone, China. Bt cotton and TFD cotton were sownduring the same season, but in different fieldsseparated by 2 km. In 1997, transgenic andnontransgenic cotton were sown by conventionalmethods at a rate of approx 15 seed/m of row on a1.0-m row spacing. Final plant density was 5 to 7plants/m2. Transgenic cotton plants and non-transgenic cotton plants were sown in the samefield. Transgenic cotton plants were sown in a 6-× 6-m square in the middle of a 210- × 210-msquare field according to the scheme presented inFig. 1. The area surrounding this central plot wasplanted with the nontransgenic cotton varietyCCRI 19. Standard cultivation practices and in-sect control measures were used in an attempt to

Fig. 1. Field trial design to measure gene flow from transgenic cotton in China during 1997 and 1998. Thetransgenic cotton test plots were planted in a 6- × 6-m square in the middle of a 210- × 210-m square field ofnontransgenic cotton plants. Bars indicate sampling points at 1, 2, 5, 10, 20, 50, 60, and 100 m.

optimize yields. At harvest, 50 cotton bolls werechosen randomly from east, south, west, and northof the transgenic cotton at distances of 1, 2, 5, 10,20, 50, 60, and 100 m.

In 1998, seeds obtained from the trial withtransgenic herbicide-resistant tfd A cotton in 1997were sown on a 30-cm row spacing and 10-cm plantspacing. At the four- to five-leaf stage, plants weresprayed with 2–5 mL of 200 mg/L 2,4-D perplant. Two weeks after spraying, the growth ofcotton plants was investigated, and the damage tothe cotton leaves was determined. Plants whoseleaves were twisted, damaged, or absent wereidentified as nonresistant plants. Plants whoseleaves grew normally were identified as resistantplants. The number of resistant seedlings and non-resistant seedlings per sampling point was re-corded.

11_20_Zhang_MB04_0094 8/8/05, 1:26 PM13

14 Zhang et al.


In 1998, seeds obtained from the trial withtransgenic insect-resistant Bt cotton in 1997 weresown on a 100-cm row spacing and 15- to 20-cmplant spacing. At the 10-leaf stage, the insect resis-tance of each plant was estimated using a field testand lab bioassay methods described by Zhang etal. (25). The number of resistant seedlings persampling point were recorded. For the field test,three to five bollworms were placed on the shootof all tested plants. Bollworm mortality and dam-age to cotton plants was measured. Plants withoutbollworm damage were identified as pest-resis-tant plants. Plants with bollworm damage wereidentified as susceptible plants. The lab bioassaywas used to further test undamaged plants to con-firm that they were really pest-resistant plants.Fresh, top opened leaves were put into Petridishes containing two young bollworm larvae.After 2 d of exposure, the results were recorded.Leaves without damage were identified as resis-tant plants. If damage was present, the plant wasidentified as susceptible.

2.3. Verification of Resistant Seedlingsby Molecular Analyses

DNA and protein were extracted from resistantplants screened in the field and lab tests. Poly-merase chain reaction (PCR) and Western blot-ting were used to confirm the reliability of ourassay and to identify resistant plants carrying theBt gene and tfd A gene. Our previous molecularanalyses (25,26) also showed that there was a highcorrelation between the presence of the resistancegene and the response of cotton plants to 2,4-D orbollworms; all resistant seedlings identified byfield screening contained the Bt or tfd A gene.Thus, in this experiment, plant resistance to 2,4-Dor bollworm was considered confirmatory for aplant containing the tfd A gene and the Bt gene,respectively.

2.4. Statistical AnalysesAll obtained data were processed using stan-

dard statistical analysis software (SigmaPlot, Ver-sion 8.0, and SigmaStat, Version 2.03, SPSS,Chicago, IL). The pollen dispersal models werefit using the same software.

3. Results and Analysis3.1. Pollen Dispersal of Transgenic tfd ACotton

After 2 wk of spraying with 2–5 mL of 200 mg/L 2,4-D per plant, two categories of cotton seed-lings were identified: resistant seedlings and seed-lings susceptible to the herbicide 2,4-D. Resistantand susceptible plants were distinguished easilyby observing their phenotype. The 2,4-D-resistantseedlings grew as well as seedlings that had neverbeen sprayed with 2,4-D. Non-2,4-D-resistantseedlings produced distinct chemical damage,such as twisted leaves. Additional experiments in-dicated that transgenic tfd A cotton plants can growand develop normally ever after 14 d of exposureto 2–5 mL of 200 mg/L 2,4-D, but nontransgenicplants never grow well and present abnormalsymptoms (26). The conventional cultivar CCRI19 produced offspring containing the resistantgene to 2,4-D after being planted near the transgenictfd A cotton resistant to 2,4-D. This indicated thatnontransgenic cotton plants were pollinated natu-rally by transgenic cotton plants, and the foreigntfd A gene was out-crossed into surrounding cot-ton plants through pollen dispersal.

Distance was the key factor that affected tfd Agene flow by pollen dispersal. Table 1 showsclearly that the proportion of seeds containing thetfd A gene decreases sharply with increasing dis-tance away from transgenic plants. Most cottonplants resistant to 2,4-D were found 1 m awayfrom the tfd A cotton plants, about 9.8–10.48%,with an average of 10.13%. When the distancefrom transgenic plants increased to 2 m, 3.29% ofthe plants exhibited resistance to the herbicide2,4-D. When the distance was increased to 20 m,only 0.26% of the plants showed 2,4-D resistance.When the distance was increased to 50 m, only 1cotton plant of 2680 examined showed resistanceto 2,4-D. No resistant plants were observed fromthe plant population harvested from locationsmore than 50 m away from transgenic cottonplants. This indicated that the tfd A gene flow bypollen dispersal decreased significantly withdistance (χ2 = 997.353, 6 degrees of freedom, p <0.0001) (Table 1). In this experiment, the farthest

11_20_Zhang_MB04_0094 8/8/05, 1:26 PM14



distance of tfd A gene flow was 50 m; beyond thisdistance, resistant plants produced by pollen dis-persal could not be found. The data were fit to apower curve model: y = 10.1321x–1.4133 with acorrelation coefficient of 0.999.

Direction slightly affected the distance and fre-quency of tfd A gene flow by pollen dispersal (χ2

= 12.733, 3 degrees of freedom, p = 0.005) (Table1). When the distance from transgenic cottonplants was 1 m, there were no obvious differencesamong the frequency of gene flow in the four car-dinal directions; frequencies of gene flow were10.07% (east), 9.87% (south), 10.11% (west), and10.48% (north). At 2 m from transgenic plants, thelowest frequency of gene flow was 1.88% (south),and the highest frequency was 6.17% (north). Thedispersal frequency to the north was 3.28 timesgreater than the frequency to the south. The re-sults indicated that gene flow by pollen dispersalin these trials was easier to the north and west thanto the east or south.

3.2. Pollen Dispersal of TransgenicBt Cotton

Without the control of cotton bollworms, non-transgenic cotton plants were seriously damaged.The primary phenotype showed the top leaves andapical buds badly damaged and several holes inthe plant leaves. In contrast, the transgenic Bt cot-ton plants grew and developed normally. To avoid

experimental errors caused from possible unbal-anced dispersal of bollworms or false-positive re-sults, both healthy and slightly damaged plantswere tested in the field and laboratory accordingto our previous reports (25,26). The experimentalresults showed that resistant plants could be foundin the offspring populations of nontransgenic cul-tivar CCRI 19 planted around transgenic Bt cot-ton (Table 2). This indicated that the exogenousBt gene was out-crossed to nearby cotton by pol-len dispersal.

The pattern, frequency, and distance of Bt geneflow were similar that of the tfd A gene, althoughthe pollen-mediated gene flow of tfd A cotton wasmuch higher, and much further into the transgenicblock than that of Bt cotton. The frequency ofgene flow decreased quickly with increasing dis-tance (χ2 = 606.916, 5 degrees of freedom, p <0.0001). Resistant plants (5.24–10.05%) werefound in the CCRI 19 offspring populations ob-tained from locations 1 m away from transgenicinsect-resistant Bt cotton. When the distance was20 m, the frequency of resistant plants decreasedto 0.08%. At 50 m, no resistant plants were ob-served in the offspring populations. The data werefit by a power curve model, y = 8.0031x–1.483 witha correlation coefficient of 0.998. Although thefrequency of the Bt gene flow by pollen dispersalwas slightly higher to the north and lowest to thesouth, statistical analysis of data from the Bt cot-

Table 1Frequency and Distance of Transgenic Pollen Movement From Transgenic tfd A Cotton Into Surrounding

Nontransgenic Cotton

Distance from tfdEast South West North Total

A cotton (m) Total Ra R% Total R R% Total R R% Total R R% Total R R%

1 685 69 10.07 689 68 9.87 643 65 10.11 668 70 10.48 2685 272 10.132 678 16 2.36 691 13 1.88 692 33 4.77 681 42 6.17 2742 104 3.795 700 2 0.29 647 8 1.24 691 10 1.45 672 10 1.49 2710 30 1.11

10 653 0 0.00 663 0 0.00 632 1 0.16 678 5 0.74 2626 6 0.2320 676 1 0.15 682 0 0.00 667 3 0.45 651 3 0.46 2676 7 0.2650 666 0 0.00 654 0 0.00 682 0 0.00 678 1 0.15 2680 1 0.0460 681 0 0.00 279 0 0.00 600 0 0.00 703 0 0.00 2263 0 0.00

100 643 0 0.00 669 0 0.00 682 0 0.00 668 0 0.00 2662 0 0.00

aR = resistant.

11_20_Zhang_MB04_0094 8/8/05, 1:26 PM15

16 Zhang et al.


ton trial indicated that the number of insect-resis-tant progeny recovered in the nontransgenic cot-ton field was independent of direction. This resultwas in contrast to results of transgenic tfd A cot-ton (χ2 = 11.712, 1 degree of freedom, p = 0.001).

3.3. Molecular Confirmation of ResistantSeedlings

The ability of the resistant screening assay toidentify transgenic plants reliably was confirmedby performing PCR and Western blotting. Resis-tant plants screened in the field and laboratorywere randomly chosen for molecular analysis.Molecular analysis showed a high correlation be-tween the presence of the resistance gene and theresponse of cotton plants to 2,4-D or bollworm.All resistant seedlings screened in the field con-tained the Bt or tfd A gene.

4. Discussion4.1. Sampling and Error

In this type of experiment on pollen-mediatedgene flow, transgenic plants are usually plantedin the middle of the field surrounded by non-transgenic plants. This means that the total numberof plants exponentially increases with distancefrom transgenic plants. Thus, it is important tochoose an efficient sampling method to avoidsampling errors. Some experiments have shownthat it is a valid method to sample at different dis-

tances along a line in each of the four cardinaldirections (11,20,21). However, random activityof insects or other environmental conditions maycause sampling errors with increasing distance.To avoid this sampling error, we increased thesample number. However, caution is still war-ranted when evaluating pollen-mediated geneflow at long distances.

4.2. Pollen Dispersal and Gene FlowPollen dispersal is a major issue when consid-

ering the escape of transgenes from geneticallymodified plants, especially for crops that can hybrid-ize with wild relatives growing nearby (5,27–29).Cotton has more than 40 species to which it canbe crossed, and some hybridization has beenapplied to cotton breeding (19,30). Althoughmany of these hybrids can be realized in thegreenhouse, it is more difficult to produce themin nature. However, more important for this situa-tion is the presence of potential pollen recipients(e.g., Gossypium araboreum and feral or volun-teer G. hirsutem L.) near transgenic cotton fields.

For most of the important crops in the world,gene flow by pollen dispersal between cultivarsand between wild and weedy relatives has alwaystaken place (5). Pollen-mediated gene flow pro-duces a source of variation for plant breeding aswell as new weeds, such as weed rye in the UnitedStates (31) and weed beets in Europe (32). How-ever, this gene flow has not attracted much atten-

Table 2Frequency and Distance of Transgenic Pollen Movement From Transgenic Bt Cotton

Into Surrounding Nontransgenic Cotton

Distance fromEast South West North Total

Bt-cotton (m) Total Ra R% Total R R% Total R R% Total R R% Total R R%

1 587 59 10.05 611 32 5.24 602 48 7.97 578 50 8.65 2378 194 8.162 602 15 2.49 608 18 2.96 609 18 2.96 589 21 3.57 2408 72 2.995 613 1 0.16 605 2 0.33 583 3 0.51 583 5 0.86 2384 11 0.46

10 599 0 0.00 597 1 0.17 611 3 0.49 605 5 0.83 2412 9 0.3720 588 1 0.17 601 0 0.00 607 1 0.16 598 0 0.00 2394 2 0.0850 600 0 0.00 598 0 0.00 617 0 0.00 602 0 0.00 2417 0 0.0060 577 0 0.00 475 0 0.00 502 0 0.00 533 0 0.00 2087 0 0.00

100 591 0 0.00 593 0 0.00 574 0 0.00 608 0 0.00 2366 0 0.00

aR = resistant.

11_20_Zhang_MB04_0094 8/8/05, 1:26 PM16



tion before the commercialization of transgenicplants. Genetic engineering raises additional con-cerns because the genes inserted into transgenicplants usually are desired genes (such as pest re-sistance). Weeds could get increased fitness andbecome superweeds if desired genes flow into wildrelatives or weeds growing near the transgenicplants (33). Enhanced understanding of this pro-cess, and more important, of the impact of geneflow and the factors controlling pollen dispersalwill help us reduce the potential risk and promotethe adoption of transgenic plants.

Gene flow by pollen dispersal is a complex pro-cess dependent on many factors (5,10,34,35), in-cluding environmental conditions, pollinatorbehavior, crop variety, and field practices (suchas pesticide spraying and plant density). Mostlikely, cotton pollen is carried by insects and in-sect prevalence strongly influences out-crossingrates (36). The most important insects are thosebelonging to the order Hymenoptera. Of the vari-ous hymenopteran species that act as cotton polli-nators, bees are the most significant (21,37). Thebees most frequently mentioned are bumble bees(Bombus spp.) and honey bees (Apis spp., such asApis dorsata, A. indica, A. mellifera, and A.florea). Other insects such as wasps, flies, and hibis-cus beetles can also serve as pollinators. The fre-quency of out-crossing depends on how many andhow often the pollinators visit the different flow-ers: the higher the pollinator population and activ-ity, the higher the ratio of out-crossing.

Insect pollinator species and their populationdensities vary geographically and seasonally (20).The determination of out-crossing rates and hencethe potential for pollen-mediated gene flow fromtransgenic plants needs to be assessed based onregion, season, and plant species. Umbeck et al.(21) examined the degree of pollen dispersal froma US field test of transgenic cotton (136 × 30 m)surround by 25 m of nontransgenic cotton. Theyobserved that the resistant marker gene waspresent in the immediate adjacent row (1 m) ofnontransgenic cotton (5.7%), and decreased tobelow 1% at 7 m. The farthest distance they foundresistant plants was 25 m (21). Because such asmall field was used, it is impossible to say that

pollen-mediated gene flow was limited to 25 m.In Australia, Llewellyn and Fitt (20) observed asimilar pattern of pollen-mediated gene flow bymeasuring the frequency of the dominant selectablemarker, neomycin phosphotransferase (Npt II), inthe progeny of buffer nontransgenic cotton plants.The frequency of transgene dispersal was muchlower (e.g., in 1992/1993 only 0.36% in the firstadjacent row), even in samples collected close tothe transgenic plants and it decreased rapidly tobelow 0.17% at 4 m (20). Llewellyn and Fitt be-lieved that the difference in the magnitude of thepollen-mediated gene flow between the trials inthe United States and Australia was probably areflection of differences in pollinator species (20).In the United States, the bumble bee, honey bee,and Melissodes bee are considered most impor-tant as pollinators of cotton (37–39). However, inAustralia, Llewellyn and Fitt (20) did not observeany pollinators including bees except for smallnumbers of wasps and flies in the field duringboth transgenic trails. Honeybees were assumedto be the main vector for cross-pollination of cot-ton in Australia (40). Among cotton pollinators,bumble bees and honey bees are more likely tocause cross-pollination when feeding on nectarfrom partially opened flowers, and visiting a largenumber of flowers before returning to the hive(20,41). Wasps and flies are unlikely to cause ahigh frequency of cross-pollination. The absenceof efficient pollinators, especially bumble bees,may be the reason for the low frequency of pol-len-mediated gene flow of transgenic cotton in theAustralia trials (20).

Our study was conducted on a large commer-cial cotton field, and used large sample sizes andfield size to detect pollen-mediated gene flow oftwo desired genes (insect-resistant Bt gene andherbicide-resistant tfd A gene). In other studies,relatively small sources have been used to onlystudy a selectable marker gene. Although thetransgenes did not significantly affect either theviability of the pollen or the activity of the polli-nator, the characteristics controlled by thesetransgenes affect agronomic practices (such asapplication of pesticides) that affect pollinatoractivity and affect insect-mediated pollen dis-

11_20_Zhang_MB04_0094 8/8/05, 1:26 PM17

18 Zhang et al.


persal. In our trial, the frequency of pollen-medi-ated gene flow decreased rapidly with distancefrom the transgenic cotton plot. However, the fre-quency of pollen-mediated transgene flow wasmuch higher, and the dispersal distance of cottonpollen was much further than that in trials in theUnited States and in Australia. The highest fre-quency of pollen-mediated gene flow was 10.13%for Bt cotton and 8.16% for tfd A cotton, bothhigher than in trials in the United States (5.7%)(21) and Australia (1.76%) (20). Pooling all theresults at equivalent distances from the transgeniccotton plants, total out-crossing was 9.20% in thefirst meter, decreasing to 0.80% by 5 m. Althoughresistant plants were found at 10, 20, and 50 m, thefrequency was very low (at 50 m from transgenicplants, only one resistant plant was observed in5097 seedlings tested). The decrease in frequencyof pollen-mediated gene flow was significant (forBt cotton, χ2 = 606.916, 5 degrees of freedom, p <0.0001; for tfd A cotton, χ2 = 997.353, 6 degreesof freedom, p < 0.0001), and was fit to a nonlinermodel: y = 8.731 / x – 0.338 with R2 = 0.991, p <0.001.

In China, pollen dispersal was further andgreater from transgenic plants than that in theUnited States or Australia. One of the reasonsmay be the different environmental conditions,especially the type, number, and behavior of thepollinators. In China, the primary pollinators arebumble bees and honey bees; secondary pollina-tors are diurnal hawk months and butterflies (42).Near the field where our trial took place there wasa bee farm; lots of honey bees visited the cottonfield. In addition, other pollinators, includingbumble bees, butterflies, ladybugs, wasps, andflies, were found in the cotton field during the tri-als (data not shown).

In Australia, 20-m buffer zones would serve tolimit dispersal of transgenic pollen from small-scale field tests (20). In the United States, out-crossing from transgenic cotton to nontransgeniccotton decreased from 5 to <1% at 7 m from thetest plot. A low level of pollen dispersal (<1%)continued to occur sporadically in the remainingborder rows to a distance of 25 m (21). In our ex-periment, the results revealed that transgenic

plants were identified at 50 m. This suggests thatenvironmental conditions affect the dispersal oftransgenes through pollen. At the same time, ourresults do not support the recommendation of pre-vious reports that a 20-m buffer zone is sufficientto prevent gene flow (20,21). To prevent geneflow from transgenic cotton to nontransgenic cot-ton, we should enlarge buffer zones from 20 m(20,21) to at least 60 m.

Modified traits affect pollen-mediated geneflow. In this trial, although a similar pattern wasobserved in tfd A cotton and Bt cotton, the pollen-mediated gene flow of tfd A cotton was muchhigher and much further into the nontransgenicblock than that of the Bt cotton (χ2 = 11.712, 1degree of freedom, p < 0.001). The reason for thisdifference is unclear. However, one importantreason may the result of environmental condi-tions. Because these two trials were conducted indifferent fields, there may have been a different pol-linator density or other environmental conditions.

Weather conditions may affect pollen dispersaland gene flow. For wind-pollinated species, windis an important factor controlling gene flow. Cornis a wind-pollinated species. Cross pollination fre-quency between transgenic and nontransgenic cornis 1% at a distance of 30 m from the transgenicdonor area (43). This is much higher than insect-pollinated species. Cotton is an insect-pollinatedspecies. It is not easy to disperse cotton pollen bywind because cotton pollen is large and sticky(20,44). However, we still found slight effects ongene flow as a result of direction for tfd A cotton.The frequency of transgene tfd A flow by pollendispersal was slightly higher in the north, andlowest in the south (χ2 = 12.733, 3 degrees of free-dom, p = 0.005). Field weather records showed thatwind coming from the south and southeast wasprevalent during the cotton growing season, espe-cially during the blooming season. The greatermovement of pollen toward the north and westpresumably related to the movement of pollina-tors and was consistent with the prevailing winddirection at the trial sites. Llewellyn and Fitt (20)also observed a directional effect of gene flow.They hypothesized that it was due to pollinatormovement. Australia is located at the southern

11_20_Zhang_MB04_0094 8/8/05, 1:26 PM18



hemisphere; the wind coming from the north isprevalent during the cotton growing season. Wedo not think it is a coincidence that a similar pat-tern was reached in the two different environmen-tal conditions at individual field trials. Out-crossingof cotton may be caused primarily by insect, sec-ondly by wind.

4.3. Risk ManagementBiosafety is key to risk assessment of transgenic

plants (1,45). Gene flow by pollen dispersal is acritical factor to be considered in estimating po-tential environmental risks. When the desiredgene invades other species, particularly weedyrelatives and feral and volunteer crop plants, thereis a potential for new and more invasive weeds toarise (46–48). Thus, preventing gene flow is nec-essary for risk management of releasing transgenicplants into the environment.

For cotton, gene flow appears to be an infre-quent event in our study and other studies (20,21).It can be adequately controlled for nearly all prac-tical purposes with existing agricultural practices,such as using buffer zones, crop rotation, and con-trol of escapes (49). In most cases, there is no needto demand extra purity. If we need extra purityfor cotton production, newly developed molecu-lar technology can be used to control gene escapefrom transgenic cotton (50,51).

AcknowledgmentsThis work was partially supported by grants

from NSF of China, the Minister of Science andTechnology of China; and the Department of Sci-ence and Technology, and the Department ofEducation of Henan of China. We also sincerelythank Dr. Stephen B. Cox of Texas Tech Univer-sity for statistical analyses of data, and Dr. DavidBecker of the Bayer Crop Science and Dr. DwightRomanovicz of the University of Texas at Austinand the University Writing Center for critical reviewof the manuscript.

References1. Mikkelsen, T. R., Andersen, B., and Jorgensen, R. B.

(2001) The risk of crop transgene spread. Nature 380,31.

2. Ghosh, S. K. (2001) GM crops: rationally irresistible.Curr. Sci. 81, 655–660.

3. Dale, P. J., Clarke, B., and Fontes, E. M. G. (2002)Potential for the environmental impact of transgeniccrops. Nat. Biotechnol. 20, 567–574.

4. Snow, A. A. (2002) Transgenic crops—why geneflow matters. Nature Biotech. 20, 542.

5. Messeguer, J. (2003) Gene flow assessment in trans-genic plants. Plant Cell Tiss. Org. Cul. 73, 201–212.

6. Chevre, A. M., Eber, F., Baranger, A., and Renard,M. (1997) Gene flow from transgenic crops. Nature389, 924.

7. Rognli, O. A., Nilsson, N. O., and Nurminiemi, M.(2000) Effects of distance and pollen competition ongene flow in the wind-pollinated grass Festuca pratensisHuds. Heredity 85, 550–560.

8. Gueritaine, G., Sester, M., Eber, F., Chevre, A. M.,and Darmency, H. (2002) Fitness of backcross six ofhybrids between transgenic oilseed rape (Brassicanapus) and wild radish (Raphanus raphanistrum).Mol. Ecol. 11, 1419–1426.

9. Ilardi, V. and Barba, M. (2002) Assessment of func-tional transgene flow in tomato fields. Mol. Breed. 8,311–315.

10. Rieger, M. A., Lamond, M., Preston, C., Powles, S.B., and Roush, R. T. (2002) Pollen-mediated move-ment of herbicide resistance between commercialcanola fields. Science 296, 2386–2388.

11. Beckie, H. J., Warwick, S. I., Nair, H., and Seguin-Swartz, G. S. (2003) Gene flow in commercial fieldsof herbicide-resistant canola (Brassica napus). Ecol.Appl. 13, 1276–1294.

12. Burke, J. M. and Rieseberg L. H. (2003) Fitness ef-fects of transgenic disease resistance in sunflowers.Science 300, 1250.

13. Wilkins, T. A., Rajasekaran, K., and Anderson, D. M.(2000) Cotton biotechnology. Critical Rev. Plant Sci.19, 511–550.

14. Zhang, B. H. and Feng, R. (2000) Insect-Resistanceof Cotton and Resistant-Pest Cotton. Chinese Agri.Sci. Tech. Press, Beijing.

15. Zhang, B. H. ,Liu, F., Yao, C. B., and Wang, K. B.(2000) Recent progress in cotton biotechnology andgenetic engineering in China. Curr. Sci. 79, 37–44.

16. Zhang, B. H., Wang, Q. L., Wang, K. B., Zhou, D.Y., and Liu, F. (2004) Bt cotton in India. Curr. Sci.86, 758–760.

17. Zhang, B. H. and Wang, Q. L. (2003) Transgenic cot-ton. In: Plant Genetic Engineering, vol. 3 Improve-ment of Commercial Plants (Singh, R. P. and JaiwalP. K., eds.), pp. 1–64. Sci-Tech Publisher Company,Houston, TX.

18. Zhang, B. H., and Zhang, L. Z. (1998) Pest-Resis-tance Cotton and its Cultivation. Shandong Sci. Tech.Press, Jinan, China.

19. Zhang, B. H. (1997) Cotton Biotechnology and its Ap-plication. Chinese Agricultural Press, Beijing, China.

11_20_Zhang_MB04_0094 8/8/05, 1:26 PM19

20 Zhang et al.


20. Llewellyn, D. and Fitt, G. (1996) Pollen dispersalfrom two trials of transgenic cotton in the Namoj val-ley, Australia. Mol. Breed. 2, 157–166.

21. Umbeck, P. F., Barton, K. A., Nordheim, E. V.,Mccarty, J. C., Parrott, W. L., and Jenkins, J. N.(1991) Degree of pollen dispersal by insects from afield-test of genetically engineered cotton. J. Econ.Entom. 84, 1943–1950.

22. Elliott, L. J., Mason, D. C., Wilkinson, M. J., et al.(2004) The role of satellite image-processing for na-tional-scale estimates of gene flow from geneticallymodified crops: rapeseed in the UK as a model. J.Appl. Ecol. 41, 1174–1184.

23. Bayley, C., Trolinder, N. L., Ray, C., Morgan, M.,and Quisenberry, J. E. (1992) Engineering 2,4-D re-sistant into cotton. Theor. Appl. Genet. 83, 645–649.

24. Chen, Z. X., Fan, Y. L., and Liewellyn, D. J. (1994)2,4-D resistant transgenic cotton plants produced byAgrobacterium-mediated gene transfer. Sci. Agri. Sin.27, 31–37.

25. Zhang, B. H., Guo, T. L., and Wang, Q. L. (2000)Inheritance and segregation of exogenous genes intransgenic cotton. J. Genet. 79, 71–75.

26. Zhang, B. H., Wang, H. M., Liu, F., Li, Y. H., andLiu, Z. D. (2001) In vitro assay for 2,4-D resistancein transgenic cotton. In Vitro Cell. Devel. Biol.-Plan37, 300–304.

27. Dale, P. J. (1992) Spread of engineered genes to wildrelatives. Plant Physiol. 100, 13–15.

28. Lavigne, C., Klein, E. K., Vallee, P., Pierre, J.,Godelle, B., and Renard, M. (1998) A pollen-dis-persal experiment with transgenic oilseed rape. Esti-mation of the average pollen dispersal of an individualplant within a field. Theor. Appl. Genet. 96, 886–896.

29. Stewart, C. N., Halfhill, M. D., and Warwick, S. I.(2003) Transgene introgression from geneticallymodified crops to their wild relatives. Nature Rev.Genet, 4, 806–817.

30. Fryxell, P. A., Craven, L. A., and Stewart, J. M.(1992) A revision of Gossypium sect Grandicalyx(Malvaceae), including the description of 6 new spe-cies. Sys. Bot. 17, 91–114.

31. Sun, M. and Corke, H. (1992) Population genetics ofcolonizing success of weedy rye in northern Califor-nia. Theor. App. Genet. 83, 321–329.

32. Boudry, P., Morchen, M., Saumitoulaprade, P., Vernet,P., and van Dijk, H. (1993) The origin and evolution ofweed beets—consequences for the breeding and re-lease of herbicide-resistant transgenic sugar-beets.Theor. Appl. Genet. 87, 471–478.

33. Pan, X. P., Zhang, B. H., and Cobb, G. P. (2005)Transgenic plants: environmental benefits and risks.Physiol. Mol. Biol. Plants 11, 13–32.

34. Barton, J. E. and Dracup, M. (2000) Geneticallymodified crops and the environment. Agron. J. 92,797–803.

35. Ellstrand, N. C., Prentice, H. C., and Hancock, J. F.(1999) Gene flow and introgression from domesti-cated plants into their wild relatives. Annu. Rev. Ecol.Syst. 30, 539–563.

36. Elfawal, M. A., Bishr, M. A., and Hassoub, E. K. (1976)Natural cross pollination in Egyptian cotton (Gossypiumbarbadense L.). J. Agric. Sci. 86, 205–209.

37. McGregor, S. E. (1959) Cotton-flower visitation andpollen distribution by honey bees. Science 129, 97,98.

38. Loper, G. M. and Davis, D. D. (1985) Disparity ofcotton pollen dispersal by honey bees visiting uplandand Pima pollen parents. Crop Sci. 25, 585–589.

39. Stephens, S. G. and Finkner, M. D. (1953) Naturalcrossing in cotton. Econ. Bot. 7, 257–269.

40. Mungomery, V. E. and Glassop, A. J. (1969) Naturalcross-pollination of cotton in central Queensland,Queensland. J. Agric. Sci. 26, 69–74.

41. Free, J. B. (1993) Insect Population of Crops, 2nd ed.Academic Press, London.

42. Tian, J. P., Liu, K. M., and Hu, G. W. (2004) Pollina-tion ecology and pollination system of Impatiensreptans (Balsaminaceae) endemic to China. Ann. Bot.93, 167–175.

43. Jemison, J. M. and Vayda, M. (2000) Pollen trans-port from genetically modified corn. In: AgronomyAbstracts. ASA, Madison, WI, p. 145.

44. Thies, S. A. (1953) Agents concerned with natural cross-ing of cotton in Oklahoma. Agron. J. 45, 481–484.

45. Snow, A. A. and Palma, P. M. (1997) Commercial-ization of transgenic plants: potential ecological risks.Bioscience 47, 86–96.

46. Desplanque, B., Hautekeete, N., and Van Dijk, H.(2002) Transgenic weed beets: possible, probable,avoidable? J. Appl. Ecol. 39, 561–571.

47. Adam, D. (2003) Transgenic crop trial’s gene flowturns weeds into wimps. Nature 421, 462.

48. Haygood, R., Ives, A. R., and Andow, D. A. (2003)Consequences of recurrent gene flow from crops towild relatives. Proc. Royal Soc. London Ser. B-Biol.Sci. 270, 1879–1886.

49. Ritala, A., Nuutila, A. M., Aikasalo, R., Kauppinen,V., and Tammisola, J. (2002) Measuring gene flow inthe cultivation of transgenic barley. Crop Sci. 42,278–285.

50. Daniell, H. (2002) Molecular strategies for gene con-tainment in transgenic crops. Nat. Biotechnol. 20,581–586.

51. Kuvshinov, V., Koivu, K., Kanerva, A., and Pehu, E.(2001) Molecular control of transgene escape fromgenetically modified plants. Plant Sci. 160, 517–522.

11_20_Zhang_MB04_0094 8/8/05, 1:26 PM20

measuring gene flow in the cultivation of transgenic cotton (gossypium hirsutum l.)

Documents