Summary of 2000 Alfalfa Pollen Flow Experiment

Mark McCaslin, Sharie Fitzpatrick and Peter Reisen

Forage Genetics International


An alfalfa pollen flow study was conducted during the summer of 2000 by Forage Genetics International in association with members of the Regulatory Affairs Committee of the North American Alfalfa Improvement Conference. The study was designed to measure pollen flow potential between adjacent alfalfa seed production blocks using leafcutter bees as the predominant pollinator species and to assess the efficacy of current alfalfa seed production field isolation requirements. Pollen flow was measured using a dominant, transgenetic marker (the Roundup Ready® gene) in the pollen source plants. Non-transgenic alfalfa was planted in replicated pollen trap plots at 500, 1000 and 1500 ft from the source field edge; fallow ground surrounded these plots. A non-replicated trap plot was planted at 2000 ft; the intervening land planted to onions and winter wheat. Pollen flow was estimated as 1.39, 0.32, 0.07, 0.00% at 500, 1000, 1500 and 2000 ft, respectively. In addition to the stocked, leafcutter bees, native honeybees were also present in the field and both likely contributed to pollen flow. Findings from this study infer that alfalfa seed production plots may be significantly insulated from undesired pollen flow using spatial isolation (distance <900 ft) in combination with non-fallow cropping of the isolation zone and adherence to recommended pollinator management.


The amount of potential pollen flow between adjacent alfalfa seed production fields is a key factor in setting isolation requirements for seed stock production. Current isolation requirements are based on data collected in the 1980’s, which used pest resistance genes as pollen flow markers between seed production fields (Brown, et al. 1986). Recently, Kansas State University and USDA scientists used naturally occurring alfalfa marker genes to measure the potential for gene flow from source blocks to alfalfa plants at various distances from the source (St. Amand, et al. 2000). The experiment is useful in estimating the likely escape of transgenes to feral alfalfa plants during commercial seed production. The current study was designed to measure likely pollen flow between transgenic and non-transgenic seed production fields. This data should be useful in helping set isolation guidelines for the commercial seed production of transgenic alfalfa varieties. The importance of unintended gene flow between adjacent seed fields and to feral alfalfa is receiving renewed attention as the U. S. alfalfa seed industry prepares for commercialization of the first transgenic alfalfa varieties.

The issue of gene flow has been extensively discussed by the Regulatory Affairs Committee of the North American Alfalfa Improvement Conference. A better understanding of pollen flow will enable the Regulatory Affairs Committee of the NAAIC to contribute technical expertise in the national discussion on guidelines for commercial release of transgenic alfalfa. In 1998-1999, the Committee declared the need to begin experiments to assess the potential for pollinator-mediated movement of alfalfa genes/transgenes during commercial seed production. Based on input from the committee, this experiment was designed to measure alfalfa pollen and gene flow under conditions typical for Pacific Northwest seed production using leafcutter bees as pollinators. The Roundup Ready® (RR) trait offers unique benefits as a marker gene for the study of gene flow between seed production fields. The RR trait is controlled by a single dominant transgene, which allows for easy identification of progeny arising from the transgene-marked pollen. An efficient seedling screen for the RR phenotype was previously developed that allows rapid identification of the marker gene.


Field Plot Design The Roundup Ready® (RR) transgene (Padgette et al., 1996) was used as a marker to measure gene flow from a 1 acre RR source plot to non-RR trap plots planted at four distances from the source plot (Figure 1). In order to meet current UDSA isolation requirements for the field release and accommodate the study size to available land resources, the replicated treatments (isolation distances between source and trap blocks) were designed to radiate in a single, westerly direction at 500, 1000 and 1500 ft from the source block. The field site did not have an anticipated, prevailing wind direction during the months of pollinator activity. The land between and surrounding the source and replicated trap plots was maintained as fallow. The fallow ground simulated a worst-case interplot management where bees would have no physical or visual barriers to movement and pollen flow would be enhanced (i. e. exaggerated as compared to actual field management conditions). In contrast, the 2 acre, 2000 ft trap plot was offset from the first series by approximately 45o southwest and other crops (winter wheat and onions) were grown on the intervening land. The latter, non-fallow management being typical for alfalfa foundation seed production in the Pacific Northwest Region of the U.S.

The study was established April 7, 2000 (USDA notification #00-053-07n). Field crops surrounding the study and the location of the 2000 ft trapare shown in Figure 2. The plots were seeded as follows: All plots- 1 lb seed /A and 36 inch row spacing; Trap plots- cultivar "DK 180ML", 35 ft x 36 ft each for replicated plots and 435 ft x 200 ft (2 acres) for the 2000 ft non-replicated trap; and, Source plot- seeded with a RR experimental line "JRR99-lot ND", 435 ft x 100 ft source (1 acre). The alfalfas have similar fall dormancy reactions-FD8 and FD7, respectively. JRR99 is a non-commercial seed lot developed specifically for this study. JRR99 is Syn1 generation seed produced by the open pollination of simplex (Aaaa) RR parent plants. RR trait phenotypic expression in the Syn1 is predicted to be 75%, with genotypic array- 1 AAaa: 2 Aaaa: 1 aaaa progeny.

Source plot seedlings were sprayed at the three trifoliolate leaf stage with 4 qt/A Roundup Ultra (glyphosate herbicide) to remove the 25% null segregants. Two weeks after herbicide application, 26.6% of the source plot seedlings were counted as killed (data not shown). The surviving pollen source plants therefore, had the RR genotypes AAaa or Aaaa, estimated to occur in 33% and 67% of the RR population, respectively.

Weeds were controlled in the source plot with a single, additional application of Roundup Ultra (2 qt/A) and were controlled in the trap plots with hand tools. Plots were furrow irrigated. Leafcutter bees were introduced when the plots were at 25% bloom (June 20). The bees were removed after 4 weeks. Plants were sprayed with desiccant August 30 and seed was harvested Sept 15. Plots were harvested in the sequence: 2000’, 1500’, 1000’, 500’, and Source. The seed produced in the RR source plot was an open pollinated Syn2, "JRR99-Syn2". Each plot replicate was harvested as a 3 x 3 grid of 9 equal subplots. The 2-acre trap seed was harvested, bulked and subsampled nine times. After assay completion, all of the seed produced in this study was devitalized to prevent inadvertent transgene release.

Gene Flow Greenhouse Assay and Data Analysis One gram of threshed, scarified seed (ca. 460 seed/gram) was planted in each greenhouse 1020 flat in the FGI, Nampa, Idaho APHIS-approved greenhouses. Nine grams seed (1 g/subplot) were assayed for each of the replicated plots. Similarly, nine- 1 gram samples were assayed from the 2000 ft trap plot. Each assay flat also included one row of each the non-tolerant check (DK 180ML) and of the 2000 RR source plot produced seed (JRR99-Syn2). Seedlings were counted 6 days after planting and sprayed with Roundup Ultra (4 qts/A) at the three trifoliolate stage (17 days old). Final tolerance counts were taken 5 days after Roundup Ultra application. To validate tolerance results, tissue from all survivors at the 1500 ft plot distance (i.e. the six putative adventitious RR seedlings) was lab tested for the presence/absence of the RR protein (EPSPS-CP4) (Padgette et al., 1996).

Alfalfa is an autotetraploid (2n=4x=32) with n=2x pollen. The source plot plants were not homozygous for the RR alleles. It is important to note that by tracking only the number of trap plot progeny marked with the RR phenotype (i. e. progeny that have resulted from either AA or Ax marked pollen), the observed %RR did not account for the additional, unmarked progeny which would have arisen from the source plot pollen that does not contain the dominant transgene (i.e. null pollen - xx, 42% of total source plot pollen). The pedigree of the source plot plants can be used to estimate their 4x genotypes, the 2n gametic genotypes and the proportion of each in the source population pollen pool (Table 2). Based on this information, it is possible to calculate the percentage of pollen carrying at least one copy of the transgene. Table 2 shows the calculations performed to estimate that 58% of the pollen produced in the RR source plot carries one or more copies of the transgene. Total gene flow potential is estimated by multiplying the %RR seed observed in the trap plots by an adjustment factor of 1.72 (i. e. 1/0.58) to account for the flow of null pollen from source plots.

Data were analyzed using Proc GLM of SAS (version 6.12, SAS Institute, Cary, NC) with variances partitioned into two sources, distance (500, 1000, and 1500 ft) and strip (reps 1, 2, 3 and 4). Variation due to distance was partitioned using orthogonal contrasts to test for linear and quadratic regression responses. An analysis of residuals was also performed to confirm that the assumptions of the variance analysis were met. Based on the results of F-tests (Type 1 error rate = 0.05), SAS Proc GLM was used to generate a linear prediction model.

Results and Discussion

No differences in pollen flow were detected among the four strips (replicates) (Table 1 and Table 3). As expected, pollen flow decreased as distance from the pollen source increased. There was an 80-fold decline in pollen flow for every 500' beyond the first trap plot. Total pollen flow was estimated to be 1.39 %, 0.32 %, 0.07% and 0.00% at 500, 1000, 1500 and 2000 ft, respectively. Using data from only the replicated plots, orthogonal contrasts revealed that the decline in pollen flow was linear; the quadratic partition was not significant. The best-fit linear prediction equation was estimated as: Pollen flow = 1.916 - 0.00132 (distance ft). The standard error (se) of the intercept and regression coefficients were 0.3010 and 0.0002787, respectively. Pollen flow at 900', the isolation distance used for production of Foundation Class seed was estimated to be 0.73% (se=0.117). The 95% confidence interval of this estimate was 0 - 1.64 %.

In greenhouse flats, the mean percentage RR phenotype for the positive control (JRR99 Syn2) and negative control (DK 180ML) lines was 83% and 0%, based on 3,431 and 3,672 seedlings, respectively. RR phenotype in the greenhouse test was 100% correlated to positive laboratory tests for the RR protein (data not shown).

Summary and Conclusions

The data presented would support that the current Foundation Class seed field isolation distance of 900’ is an effective limitation to significant gene flow between adjacent alfalfa seed production fields as <1% flow would be predicted under worst-case fallow, interplot management. In this study, several factors may have (intentionally) encouraged or exaggerated gene flow above what might occur under actual alfalfa seed production practices. First, the field area planted to the RR source plants was three times greater than the collective area occupied by the small, replicated trap plots (1 ac vs. 0.35 ac). Second, the land between the source and replicated trap plots was left fallow so there was no visual or intercrop impediment to bee movement. Third, trap plots at the intermediate distances may have acted as a bridge for the bees between the source and those further away (therefore the effective isolation distance is less than 1000 or 1500 ft). No gene flow was detected in the 2000 ft trap that had been grown under actual field isolation production practices. Findings from this study infer that alfalfa seed production plots may be effectively isolated from undesired pollen flow using isolation in combination with non-fallow cropping of the isolation zone and adherence to recommended pollinator management.

Leafcutter bees are reported to have a maximum foraging radius of approximately 600 ft if there is sufficient alfalfa pollen available. Leafcutter bees were the intended pollinators in this study but routinely, native honeybees were also observed working the flowers. Honeybees are known to collect nectar and pollen at much greater distance from their hive. Honeybee flight distance estimates range from 1 to 5 miles depending upon the social cast of the bee and forage supply available to them. The present study was situated between two fields of corn (see Figure 2). Corn and corn pollen is highly attractive to honeybees. Although leafcutter bees may have been responsible for the most of the pollen movement at the shorter distances, we believe that the native honeybees may have mediated a portion of the flow, especially at the 1500’ trap distance. A blend of intended and native pollinators would be anticipated to occur in most commercial alfalfa seed production fields. Pollen will move according to the pollinator density, bloom abundance relative to pollinator density and the pollinator species present in the field.


Members of the Regulatory Affairs Committee of the North American Alfalfa Improvement Conference are acknowledged for their valuable contributions to the study design and review of the results. Thanks are expressed to Steve Bowley for statistical analysis of the data.


Brown, D. E., Grandstaff, E. L., Hanna, M. R., Hanson, A. A., Marble, V. L., Moutray, J. B. 1986. Committee Report on Alfalfa Field Isolation. Page 123, in: Report of the Thirtieth North American Alfalfa Improvement Conference. July 27-31, 1986. Reported by J. B. Moutray and J. H. Elgin.

Padgette, S., Re, D., Barry, G., Eichholtz, D., Dellanay, X., Fuchs, R., Kishore, G., Fraley, R. New Weed Control Opportunities: Development of soybean with a Roundup Readyä gene. In Herbicide Resistant Crops; Duke, S., Ed.; CRC Press: Boca Raton, FL, 1996, 53-84.

St. Amand, P. C., Skinner, D. Z. and Peaden, R.N. 2000. Risk of alfalfa transgene dissemination and scale-dependent effects. Theor. Appl. Genet. 101: 107-114.

Figure 1. 2000 Alfalfa Gene Flow Study field plot design (replicated plots only)

The above diagram illustrates the replicated experimental design. Some critical parameters and criteria were as follows:

1) Distance between source plot and trap plots (distances between closest plot edges):
  1. 500 feet (A)
  2. 1000 feet. (B)
  3. 1500 feet (C)
2) Plot size and alfalfa line description of the source and trap plots
  1. source plot — 1 acre (435 ft x 100 ft); RR alfalfa (lot JRR99-ND)
  2. trap plots — 0.03 acre (35 ft x 36 ft); non RR alfalfa — DK180 ML

distance between plots in the same rep — ca. 90 ft

3) Number of trap replications— four (each harvested in 9 sections of equal size)
4) Leafcutter pollinator density and location — four domiciles- 500’ apart, similar density to commercial practices
5) Interplot cover — fallow

Figure 2. Field plot map and crops surrounding 2000 Gene Flow Study

Table 1. Mean percent gene flow from alfalfa. Standard error (se) based on data from replicated plots at 500’, 1000’ and 1500’, only.

Trap distance


Seedlings tested

Gene Flow

















Table 2. Genotype of Roundup Ready alfalfa source plants and the proportional contribution of pollen to pool. Alfalfa is an autotetraploid (4x=32) with 2n pollen.

Genotype of Pollen


Genotype of RR Source Plant (4x)

Proportion of

Total Source Plot Pollen Pool

AAxx (0.33)

Axxx (0.67)



















Table 3. Variance analysis of gene flow from alfalfa using a Roundup Ready transgene as a marker. Test was conducted in 2000 at Nampa, Idaho.



Mean square


























Previous Page