RISK OF ALFALFA TRANSGENE DISSEMINATION AND SCALE DEPENDENT EFFECTS

Paul C. St. Amanda, Daniel Z. Skinnera, and Richard N. Peadenb

aUSDA/ARS and Agronomy Department, 2004 Throckmorton Hall, Kansas State University, Manhattan KS 66506 Voice (913) 532-6168, Fax (913) 532-6167; and bUSDA/ARS, Irrigated Agricultural Research Center, Prosser WA 99350

SUMMARY

Pollen can function as a vehicle to disseminate introduced, genetically-engineered genes throughout a plant population or into a related species. The measurement of the risk of inadvertent accidental dispersion of pollen. Factors to be considered include the rate of pollen spread, the maximal dispersion distance of pollen, and the spatial dynamics of pollen movement within seed production fields; none of which are known for alfalfa (Medicago sativa L.), an insect-pollinated crop species. Using a rare, naturally-occurring molecular marker, pollen dispersal can be tracked without introducing engineered genes into the environment. In this study, a suitable marker was found, using PCR-based methods, in an intron of the alfalfa glutamine synthetase gene. This marker system was used to track pollen movement in seed-production fields; random amplified polymorphic DNA (RAPD) fragments were used to detect cross-pollination events in widely-dispersed alfalfa plants escaped from cultivation. Preliminary results indicated that leafcutter bees (Megachile spp.) used in commercial seed production show a bi-directional bias when pollinating, primarily resulting in the movement of pollen from marker plants directly toward and away from the bee domicile. Within-field pollen movement was detected only over distances of 4 m or less. Long-range dispersal of pollen from alfalfa hay production fields has been confirmed for distances up to 270 m using small trap plantings of alfalfa. Individual plants grown within urban areas at least 800 m from known alfalfa plants failed to produce seeds. An assessment of scale dependencies is lacking from most risk assessment studies. A novel field design and marker system was developed to allow an accurate investigation of scale dependencies on gene movement by comparing dispersal from research scale plots with that of commercial scale fields. Preliminary data suggest that complete containment of transgenes within alfalfa seed or hay production fields would be highly unlikely using current production practices.

Key words: Pollen dissemination, scale dependent effects, Medicago sativa L., Megachile spp., escaped plants

INTRODUCTION

Concerns have been raised over the release of genetically engineered plants into the environment, because of the risk of inadvertent dispersal of engineered genes into the cultivated crop, or into related weed species. Factors germane to that risk include the rate of pollen mediated gene spread, the maximum dispersion distance of pollen, and the spatial dynamics of pollen movement within seed production fields; none of which are known for alfalfa (Medicago sativa L.), an insect-pollinated species. The size of the area planted to the transgenic crop undoubtedly affects the dynamics of pollen dispersal; but lacking from most risk assessment studies is an investigation of scale dependencies of the data.

Manasse and Kareiva (1991) have summarized the existing body of theoretical models and real data applicable to the spread of transgenes via pollen. Essentially, the current models seek to predict spread as a function of several biological parameters. These biological parameters include such things as fitness, reproductive rate, and various life history characteristics such as a discrete breeding phase as opposed to the continuous reproduction of microbes. These mathematical models rely on certain assumptions including "idealized" populations and environments. Manasse and Kareiva (1991) realized these shortcomings, and stated "we encourage attempts to quantify the rate of spread, because they provide a common currency for comparing rates of spread among different types of transgenic organisms, ecological conditions, or release protocols. Without such a quantitative estimate of spread, risk analysis is doomed to anecdotal status". The research project reported here was directed toward contributing to that "common currency".

Handel (1983) showed that most pollen in a field of cucumbers was transported for short distances (<3 m) by honeybees (Apis spp.), but some was transported as much as 25 m. Thomson and Thomson (1989) found that bumblebees (Bombyx spp.) transport Erythronium grandiflorum pollen as much as 40 m. Ellstrand (1988) has suggested that pollen transport can occur over distances of 1000 m or more, in some systems.

Very few studies on the potential spread of actual transgenes via pollen have been conducted. Manasse and Kareiva (1991) summarized a field study carried out by the Monsanto Company with transgenic Brassica napus. It was reported that in samples of about 10,000 seeds collected 50, 100, 180, or 400 m from the transgenic source, 0.022% from 50 m, and 0.011% from 100 m from the source carried the transgene. The experimental design used was criticized by Manasse and Kareiva (1991), but does show that Brassica pollen can be dispersed to distances of at least 100 m in the field, albeit with low frequency.

Ellstrand (1988) has published a provocative paper concerning the risk of pollen dispersal. He has suggested that distances of kilometers may be required to attain a level of certainty of reproductive isolation. He also suggests using pollen sinks instead of pollen sources to measure interpopulation pollen flow. The logic of Ellstrand's (1988) proposal is that detecting rare dispersal events can much more efficiently be done by examining widely-distributed plants for non-self pollination events, rather than attempting to follow dispersal of the pollen of a specific source population. Part of the research project discussed here follows Ellstrand's suggestions, and are directed toward discovering long-distance pollen dispersal dynamics.

Alfalfa is the most important forage crop in North America. It currently is grown on about 26 million acres in the United States. Because alfalfa is easily genetically transformed, it often is considered as the ideal plant to use in large-scale production of transgene products. The alfalfa floral system is unique among cross-pollinated field crops in that the explosive tripping mechanism (Barnes et al., 1972) virtually guarantees that only the first pollinator visit will result in seed formation. Therefore, the pollen dispersal dynamics of alfalfa likely will be considerably different from other systems, and may serve as a baseline for study of other systems not subject to the one pollination per flower constraint.

Studies of transgene movement by pollen have been limited to small research scale plots. It is not known if transgene dispersal rates and maximum distances are scale dependent. Can small scale results be extrapolated to commercial scales? Are the risks scale dependent? These questions were raised, but went unanswered, at the Third International Symposium on the Biosafety Results of Field Tests of Genetically Modified Plants and Microorganisms held in Monterey, California, November, 1994.

Our goals for this study were: i. to determine the spatial dynamics of pollen movement within alfalfa seed production fields, ii. to assess the movement of pollen away from seed and hay production fields, iii. to investigate the movement of pollen between escaped alfalfa plants, and iv. to address the effect of field size on pollen movement.

MATERIALS AND METHODS

SPATIAL DYNAMICS OF POLLEN MOVEMENT WITHIN SEED PRODUCTION FIELDS

Genetic Marker and Detection. Several PCR primer pairs were designed to flank intron sequences in the published alfalfa glutamine synthetase (GS) gene (Tischer et al., 1986). Primer pairs were tested on a wide variety of alfalfa genotypes to identify naturally-occurring polymorphic DNA sequences using standard PCR techniques (Griffin and Griffin, 1994). A single plant, Varia 74 from "Varia" basic germplasm source PI 536533, was identified which carried a fragment of approximately 200 bp that was not present in any other genotypes examined. The polymorphic DNA was partially sequenced using a modified Sequenase technique (Trewick and Dearden, 1994) and primers specific to the polymorphic DNA were synthesized. These primers increased the specificity of the amplification reactions and were used routinely to generate the specific polymorphism found in Varia 74. Subsequently, a second genotype with the polymorphism was identified; it also was from within the Varia germplasm (plant Varia 57). The two individuals carrying the specific polymorphism (marker band) were crossed and the progeny were intercrossed and screened several generations to develop source plants homozygous for the marker band. The sequence of the marker band was compared to the published alfalfa GS gene using ClustalW (Thomson et al., 1994) and to GenBank using BLAST (Altschul et al., 1990) and FASTA (Pearson and Lipman, 1988) methods.

Detection of the marker band in progeny plants was performed using standard PCR techniques (Griffin and Griffin, 1994). Total DNA was extracted from entire seedlings using the NaOH method (Wang et al., 1993). In order to reduce the total number of plants examined and to increase the total number of pollination events examined, only one seed per pod was tested, since all seeds in a given pod most likely resulted from a single pollination event.

Field Design. In order to track the distance and direction of pollen movement from source plants, a circular field plot design was used. Five plants homozygous for the marker were planted in a 1 m diameter plot in the center of a rectangular 600 x 400 m alfalfa seed production field, planted to the cultivar 'Vernal,' in Prosser, WA, in the spring of 1994. A single leafcutter (Megachile spp.) bee hive was placed along the south edge of the field for pollination at the initiation of bloom. Standard seed production practices were followed. Each plot was designed to keep the radial plot angle and plot area constant (Figure 1). Plots were at regular, 2 m measured distances away from the source plants, beginning at 2 m and proceeding out to 10 m. The measured distance for each plot was the maximum distance of the plot from the source plants. The plot depth varied at each distance to create plots of uniform area (0.78 m2). Plots radiated out from the source plants at eight regular angles coinciding with the eight cardinal points of the compass (N, NE, E, etc.). Radial plot width was 11.25 degrees. Each plot was hand harvested and all pods in a plot were collected.

A similar design was used in 1995 with the following changes: ten source plants were planted in a 1 m diameter plot in the center of 2 rectangular 200 x 400 m alfalfa seed production fields in Prosser, WA, in the spring of 1995. Leafcutter bee hives were placed on the western edge of one field and on the northeastern edge of the second field. Fields were planted with the cultivar 'Vernal'. Plots were at regular, 1 m measured distances away from the source plants, beginning at 1 m and proceeding out to 10 m (Figure 2). Plot area was 0.20 m2. Plots radiated out from the source plants at sixteen regular angles coinciding with the sixteen points of the compass (N, NNE, NE, ENE, E, etc.). Radial plot width was 22.5 degrees. Each plot was hand harvested and all pods in a plot were collected.

Data Analysis. Transfer of the genetic marker from source plants to surrounding plants via pollen was measured using per plot counts of progeny containing the polymorphic DNA marker. Maximum and average distance of movement, the mean direction of movement (mean vector) relative to the beehive, and the angular variance and deviation relative to the beehive (Batschelet, 1981) were measured. Statistical evidence for non-random direction of pollen movement was tested using the Rayleigh test (Batschelet, 1981). All plots within a field were pooled for a given direction to determine the mean vector, the angular variance and deviation, and for the Rayleigh test. All data were corrected for grouping according to Batschelet (1981) and where appropriate, axial bimodal data angles were doubled (Batschelet, 1981).

POLLEN MOVEMENT FROM SEED AND HAY PRODUCTION FIELDS

Genetic Marker and Detection. The naturally-occurring polymorphic GS marker described above was used as a source plant marker. An additional marker, also a fragment of the GS gene, was present in all genotypes examined except for one plant from KS108 (Sorensen et al., 1985) germplasm. This plant, KS108-36, was vegetatively propagated and used as a pollen trap plant. Progeny from the trap plants show three distinct genotypes using the two GS markers. Progeny resulting from self-pollination lack both markers, cross-polinations between the trap plant and any other have the common GS marker, and cross-pollinations between the trap plant and the source plant have both GS markers. Presence of marker fragments were detected using PCR as described above.

Field Designs. In the spring of 1994, trap plots each having 10 plants in a one meter diameter area were planted 0, 18, 36, 54, 72, 90, 180, 270, 360, and 450 meters away from a 50 m x 50 m hay production field in Ashland, KS as a preliminary test. The area containing trap plants was searched for escaped alfalfa plants on three occasions through the growing season; none were found. Hay was harvested from this field 4 times during 1994. A small amount of flowering was noted, especially near the edge of the field. The trap plots were arranged in a straight transect, proceeding south of the hay field. Vegetation between trap plots was either wheat, soybeans, or bare ground. In the spring of 1995, trap plots each having 15 plants in a one meter diameter area were planted along road sides 0, 20, 40, 60, 80, 100, 200, 300, 400, 500, 750, and 1000 meters away from alfalfa fields (Figure 3). The area containing trap plants was searched for escaped alfalfa plants on two occasions through the growing season and the few plants found were removed. Commercial-scale production fields were used. Both types of fields, hay or seed production, were planted at two locations. One location, Prosser, WA, is an area of high pollinator activity with many leafcutter bees released especially for alfalfa pollination. The second location, Ashland, KS, normally has no alfalfa seed production, but does have a large acreage of alfalfa hay production. Hay was harvested from hay production fields 5 times at the Prosser location and 3 times in Ashland during 1995. A 2 m2 plot of approximately 300 source plants was planted at the edge of each hay and seed production field in 1995 to evaluate plot size or scale effects on marker gene movement. This planting plan, and the marker systems used, allowed the testing of different-sized source plots in the same physical space. Source plots within fields were planted adjacent to the transect of trap plots. Trap plants at all fields in both years were a minimum of 1000 m from any other alfalfa fields.

Variables Measured and Data Analysis. All seeds from trap plants were analyzed for the presence or absence of the two GS markers described above. Marker data were analyzed for minimum, maximum, and mean distance from seed and hay production fields. Models of the rate of gene spread will be based on those reviewed by Manasse and Kareiva (1991). Rates and distances of gene spread from large scale seed and hay production fields will be compared with those from the small research scale plots located within each large scale field.

POLLEN MOVEMENT BETWEEN ESCAPED PLANTS

Genetic Marker and Detection. RAPD markers were used to asses whether progeny plants were the result of self- or cross-pollination. RAPD markers have been shown to segregate according to qualitative genetic theory; however, the possibility of new markers forming from crossovers in a self pollination event, though rare, may exist. Bands found in progeny that were completely lacking in maternal parent plants indicated progeny that may have come from a cross pollination event.

Field Design. Escaped alfalfa plants were located along roadsides in and near Manhattan, KS and Prosser, WA. Individual plants were also planted in urban areas at least 800 m from known alfalfa fields. The area around each individual plant was thoroughly searched for the presence of other alfalfa plants. The distance between the individual escaped plants found and the nearest alfalfa plants was measured with either an optical rangefinder (Rangematic 1200, Ranging Co., East Bloomfield, New York) or with an automobile odometer. Distances investigated ranged from 20 m to 1,126 m. Most escaped plants were located along roadsides in cultivated regions.

Data Analysis. Minimum distances of gene flow were calculated for each maternal plant based on the shortest distance between each maternal plant and the nearest possible paternal plant.

RESULTS AND DISCUSSION

Marker System. A rare, naturally occurring, and useful molecular marker was found and is highly related to intron number six of the glutamine synthetase gene for both alfalfa and yellow lupine (Lupinus luteus). Specific primers (GS FM1 5'GGTGAAAACTCTTTTACACTTG3', and GS RM91 5'ACAAAAACAT

AGTAAATCTCTAGGG3') were designed based on the polymorphic sequence. PCR using these primers and an annealing temperature of 54°C carried out on DNA from plants homozygous for the marker produce a single 91bp marker band. Plants not carrying the marker fragment produce a band of approximately 150 bp. Progeny of marker by non-marker crosses have both bands. Examination of nearly 2000 plants indicated that this sequence is not present in a broad range of alfalfa genotypes (data not shown).

Pollen Movement within Seed Fields. Nearly 3,000 progeny from one seed production field have been evaluated thus far. Of those progeny, 3.06% carried the molecular marker. Movement of the marker within the field occurred only over very short distances. Only 0.2% of the progeny 4 m away from the marker plants carried the marker gene and no progeny 6 m away had the marker. Of those progeny carrying the marker, 97.4% occurred no more than 2 m from the source plants.

Preliminary results indicate that leafcutter bees used in commercial seed production show a bi-directional bias when pollinating alfalfa, resulting in the movement of pollen from marker plants directly toward and away from the colony hive (Figure 4). Of those progeny carrying the marker, 84.6% were located directly north or south of the marker plants in a straight transect with the bee domicile. The standard deviation about the mean vector was 28.4 degrees, indicating that bees did not deviate far from the preferred straight transect. The majority of gene movement (61.5%) was from the marker plants toward the domicile. The distribution of marked progeny around the marker plants differed significantly (P 0.0001) from a random distribution as measured by the Rayleigh test.

Pollen Movement from Alfalfa Fields. Long-range dispersal of genes from alfalfa hay production fields by pollen has been confirmed for distances up to 270 m using small trap plantings of alfalfa. Greater distances likely will be confirmed as testing continues. Modeling of the rates of gene spread will be conducted after data collection is complete. A risk assessment model describing pollen dispersal from a source population of insect-pollinated plants will be developed. The effects of field size on detectable gene flow distances will be calculated on data from the 1995 field studies.

Pollen Movement Between Escaped Plants. By examining widely dispersed, individual escaped alfalfa plants and their progeny using RAPD markers, gene movement to an alfalfa plant 230 m from the nearest neighboring plant has been confirmed. However, RAPD examination of the closest possible pollen donor indicated it was not the pollen donor. Pollen, in this case, must have traveled farther than 230 m to the escaped plant. Individual plants placed in urban areas failed to set any seed. These plants were at least 800 m from known plantings of alfalfa. The lack of self pollinations on these plants indicates a very low level of pollinator activity. Modeling of the rates of gene spread and escape will be conducted as more data are collected.

These preliminary data suggest that complete containment of transgenes within alfalfa seed or hay production fields would be highly unlikely using current production practices. The addition of non-transgenic borders surrounding fields likely will reduce the risk of transgene spread from alfalfa fields. However, our preliminary data indicate that single escaped plants growing at considerable distances from alfalfa fields can contribute to medium and long-distance gene spread away from fields and into escape populations. Since the likelihood for gene containment is small, studies of transgene fitness, hybridization, and competitive advantage should be conducted on all potential transgenes prior to release into the environment.

ACKNOWLEDGMENTS

Funding from the USDA Biotechnology Risk Assessment Program is gratefully acknowledged.

REFERENCES

Altschul, S.F., W. Gish, W. Miller, E.W. Myers, and D.J. Lipman. 1990. Basic local alignment search tool. J. Mol. Biol. 215:403-10.

Barnes, D.K., E.T. Bingham, J.D. Axtell, and W.H. Davis. 1972. The flower, sterility mechanisms, and pollination control. pp.123-142 In: Alfalfa Science and Technology. C.H. Hanson (ed.). Amer. Soc. Agron., Madison, WI.

Ellstrand, N.C. 1988. Pollen as a vehicle for the escape of engineered genes? Trends Ecol. Evol. 6:S30-S32.

Griffin, H.G. and A.M. Griffin. 1994. PCR Technology: Current Innovations. CRC Press, Boca Raton, Florida.

Handel, S.N. 1983. Contrasting gene flow patterns and genetic subdivision in adjacent populations of Cucumis sativus (Cucurbitaceae). Evolution 37:760-771.

Manasse, R. and P. Kareiva. 1991. Quantifying the spread of recombinant genes and organisms. pp.215-231 In: Assessing Ecological Risks of Biotechnology. L.R. Ginzburg (ed.). Butterworth-Heinemann, Boston.

Pearson, W. R. and D.J. Lipman. 1988. Improved tools for biological sequence comparison. Proc. Natl. Acad. Sci. USA. 85:2444-2448.

Sorensen, E. L., E. K. Horber, and D. L. Stuteville. 1985. Registration of KS108GH5 glandular-haired alfalfa germplasm with multiple pest resistance. Crop Sci. 25:1132.

Thomson, J.D., and B.A. Thomson. 1989. Dispersal of Erythronium grandiflorum pollen by bumblebees: Implications for gene flow and reproductive success. Evolution 43:657-661.

Thomson, J.D., D.G. Higgins, and T.J. Gibson. 1994. CLUSTAL W: Improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties, and weight matrix choice. NAR 22:4673-4680.

Tischer,E., S. DasSarma, and H.M. Goodman. 1986. Nucleotide sequence of an alfalfa glutamine synthetase gene. Mol. Gen. Genet. 203:221-229.

Trewick, S.A. and P. Dearden. 1994. A Rapid Protocol for DNA

Extraction and Primer Annealing for PCR Sequencing. BioTechniques. 17:842-844.

Wang, H., M. Qi, and A. J. Cutler. 1993. A simple method of preparing plant samples for PCR. NAR. 21:4153-4154.

Figure 1. Field design for the 1994 within field pollen moverment study. Marker plants were planted in a 1 m plot in the center of a 600 x 400 m field. Plots (dark grey) were 0.78 m2 in area. Plots were spaced at 2 m intervals from the marker plants.

Figure 2. Field design for the 1995 within field pollen moverment study. Marker plants were planted in a 1 m plot in the center of two 200 x 400 m fields. Plots (dark grey) were 0.2 m2 in area. Plots were spaced at 1 m intervals from the marker plants.

Figure 3. Field design for the 1995 study on pollen movement form alfalfa fields. Trap plots were 1 m in size and planted at the distances (m) shown. Seed and hay fields were of commercial size. The marker plot at the edge of each field was 2 m2 and contained approximately 300 plants.

Figure 4. Preliminary data on within field pollen movement. Contour color indicates percent of progeny carrying the marker gene. The darkest gray equals 0% and each lighter contour represents a 1% increase.