| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Colorado State Univ., Dep. of Soil and Crop Sciences, 1170 Campus Delivery, Fort Collins, CO 80523
b Dep. of Bioagricultural Sciences and Pest Management, 1177 Campus Delivery, Fort Collins, CO 80523
c USDA/ARS Central Great Plains Research Station, Akron, CO 80720, present address, Room 326 Dorman Hall, Stone Blvd., Mississippi State, MS 39762
d USDA/ARS Water Management Research, 2150 Centre Avenue, Fort Collins, CO 80526
e Dep. of Statistics, 1877 Campus Delivery, Fort Collins, CO 80523
* Corresponding author e-mail: Patrick.Byrne{at}colostate.edu.
 |
ABSTRACT |
|---|
 TOPNOTES ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Abbreviations: ALS, acetolactate synthase • AIC, Aikake information criterion • GLMM, generalized linear mixed model • GWM, general wheat model • IR, imidazolinone-resistant • IS, imidazolinone-susceptible • PCR, polymerase chain reaction • PMGF, pollen-mediated gene flow • RHC, relative heading class
a Colorado State Univ., Dep. of Soil and Crop Sciences, 1170 Campus Delivery, Fort Collins, CO 80523
b Dep. of Bioagricultural Sciences and Pest Management, 1177 Campus Delivery, Fort Collins, CO 80523
c USDA/ARS Central Great Plains Research Station, Akron, CO 80720, present address, Room 326 Dorman Hall, Stone Blvd., Mississippi State, MS 39762
d USDA/ARS Water Management Research, 2150 Centre Avenue, Fort Collins, CO 80526
e Dep. of Statistics, 1877 Campus Delivery, Fort Collins, CO 80523
* Corresponding author e-mail: Patrick.Byrne{at}colostate.edu.
The potential introduction of wheat (Triticum aestivum L.) cultivars with transgenic traits has generated increased interest in pollen-mediated gene flow (PMGF). The objectives of this study were to estimate wheat PMGF between commercial fields across multiple years and locations, and to compare estimates from large fields to those from smaller experimental plots. The study was conducted in a total of 56 commercial field locations in eastern Colorado in 2003, 2004, and 2005. We measured PMGF by tracking the movement of an imidazolinone herbicide resistance gene from resistant to susceptible cultivars, sampled at distances of 0.23 to 61 m. At least one sample from all 56 fields and from all 18 evaluated cultivars had detectable PMGF. The highest observed PMGF was 5.3% at 0.23 m. The farthest distance at which PMGF was detected was 61 m and the highest PMGF at that distance was 0.25%. Higher levels and greater distances of PMGF were detected in commercial fields than in experimental plots. Based on estimates from a generalized linear mixed model with a random location effect, the distance required to ensure 95% confidence that 95% of locations have PMGF less than 0.9% is 41.1 m for cultivars heading earlier than the pollen source and 0.7 m for cultivars heading later than the pollen source. These confidence limits should represent the highest levels of PMGF expected to occur in winter wheat in the west-central Great Plains and will be useful for wheat biotechnology regulation.
Abbreviations: ALS, acetolactate synthase • AIC, Aikake information criterion • GLMM, generalized linear mixed model • GWM, general wheat model • IR, imidazolinone-resistant • IS, imidazolinone-susceptible • PCR, polymerase chain reaction • PMGF, pollen-mediated gene flow • RHC, relative heading class
|
INTRODUCTION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Wheat seed production guidelines address gene flow between fields. In Colorado, isolation distances of 3 m are required for certified seed production (Anonymous, 2003). Several authors have suggested that to ensure adequate purity in wheat seed production, a larger isolation distance is needed, such as 30 m (Matus-Cadiz et al., 2004) or 45 m (Hanson et al., 2005). Gene flow in wheat has been detected at distances as great as 2.75 km, though at the very low level of 0.01% (Matus-Cadiz et al., 2007).
Pollen movement is the first step required for gene flow to occur. Synchrony in flowering between fields is necessary, because viable pollen must successfully enter florets, germinate on stigmas, and fertilize ovules (Waines and Hegde, 2003). Anthesis in wheat can last up to 10 d (De Vries, 1973) and stigmas can be receptive for a period of four to 13 days (De Vries, 1971). Cross-pollination in wheat can be affected by both genetic and environmental factors. Cultivars are known to vary in receptivity to gene flow under controlled conditions (Griffin, 1987; Hucl, 1996; Hucl and Matus-Cadiz, 2001; Lawrie et al., 2006; Martin, 1990). The optimal temperature for pollen production is 21°C (Porter and Gawith, 1999), but freezing or near-freezing temperatures at the boot, heading, and anthesis stages can kill anthers, resulting in male sterility (Shroyer et al., 1995) that leaves the surviving stigmas more receptive to foreign pollen. High temperatures or drought can increase cross-pollination frequency (Briggs et al., 1999; Waines and Hegde, 2003) by increasing glume opening, which is associated with higher outcrossing (De Vries, 1971; Hucl, 1996; Waines and Hegde, 2003).
Pollen source size is known to influence outcrossing in recipient wheat (De Vries, 1974). Several studies in wheat have used single row pollen sources to measure gene flow via pollen over close distance (Griffin, 1987; Hucl, 1996; Martin, 1990). Other studies have used pollen source blocks ranging in size from 5 m2 (Hucl and Matus-Cadiz, 2001) to 50 m2 (Matus-Cadiz et al., 2004) to 1640 m2 (Hanson et al., 2005). Matus-Cadiz et al. (2004) suggested that further research at the commercial field scale was necessary, because small pollen sources may underestimate gene flow in wheat.
Realistic measurements of gene flow require adequate sampling over space and time to reveal the range of variation that may occur (Rieger et al., 2002). The release in 2001 of a non-transgenic herbicide-resistant winter wheat cultivar in Colorado presented a unique opportunity to monitor gene flow at the commercial field scale. This cultivar, Above (Haley et al., 2003), has single-gene imidazolinone herbicide resistance due to a mutant allele at the acetolactate synthase (ALS) locus on the D genome (Anderson et al., 2004). Above and several subsequently released cultivars were developed through induced mutagenesis followed by conventional plant breeding (Newhouse et al., 1992; Tan et al., 2005). Above was first planted by commercial grain producers in fall of 2002 and the first summer during which PMGF could have occurred on a large scale was 2003. In this system, gene flow is identified by the presence of heterozygous imidazolinone-resistant (IR) plants in the progeny of imidazolinone-susceptible (IS) plants.
Ensuring coexistence of transgenic and non-transgenic wheat, including appropriate biotechnology regulations for transgenic wheat field trials and appropriate seed production standards, will require knowledge of commercial field-scale gene flow variation. A comparison between gene flow in commercial fields and in experiments with small pollen sources would also aid in evaluating previous wheat gene flow studies. This study utilized the novel IR trait to (i) estimate gene flow in wheat between commercial fields across years and locations and (ii) compare gene flow estimates from commercial field sampling to those obtained from small experimental plots.
|
MATERIALS AND METHODS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Commercial field samples were collected by hand harvesting single rows of the IS cultivar along a transect perpendicular to the field border at distances from 0.23 to 61 m from the IR border. Each sample was a composite of at least 20 subsamples of all wheat heads in a 1 m length of row, spaced at least 3 m apart, in a single row at each distance from the IR pollen source. For the strip trials, a single cultivar was represented by one to three composite samples at different distances from the IR cultivar. Samples were grouped by distance and threshed individually with a stationary thresher, starting with samples from the farthest distances and proceeding to the closest distances to minimize potential contamination between samples.
A total of 455 samples were collected from 56 locations in eastern Colorado (Table 1 ). The samples included 18 commercial wheat cultivars, representing six relative heading classes (RHC) for eastern Colorado (Table 2 ). Relative heading classes range from 1 to 8, with 1 being earliest and adjacent classes differing by approximately 1.5 d (S. Haley, personal communication, 2006). At all locations IS wheat was sampled next to Above, with the exception of one location where IS wheat was sampled next to Bond CL. Above is classified as an RHC 3 cultivar, and Bond CL is considered to be in RHC 5. All locations had some degree of heading date overlap between IS and IR cultivars sufficient to provide flowering synchrony. Sampling sites included 25 cultivar strip trials (177 samples) and 36 large commercial fields (278 samples). Some locations included both commercial field transects and cultivar strip trials. Commercial field locations ranged in size from 32 to 130 ha for pollen source and recipient fields, and border length varied from 200 to 800 m. Cultivar trial pollen source and recipient plots ranged from 6- to 24-m wide with a 200- to 400-m long border.
|
|
The number of surviving plants was determined in late May. A heterozygous phenotype, indicative of gene flow, was identified by mild foliar chlorosis, stunting, increased tillering, and twisted spikes. This phenotype was consistent with greenhouse observations of imidazolinone-treated heterozygous plants (Pozniak and Hucl, 2004). We estimated the total number of emerged plants in each plot by counting subsamples. The subsample plant population was used to estimate the average number of plants per meter of row, and that number was multiplied by the number of meters of row in a plot to estimate the total number of emerged plants.
Survivor Verification
Resistance to ALS-inhibiting herbicides in plants can occur via spontaneous mutation (Saari et al., 1994). To verify that plants scored as heterozygous carried one copy of the ALS mutant allele from Above and one copy of the wild-type ALS allele from the susceptible parent, we conducted polymerase chain reaction (PCR) analysis on a subsample of plants. Proprietary PCR-based protocols and primers specific to the D genome ALS mutant allele from Above and the wild-type D genome ALS allele were provided by the BASF Corp. (Research Triangle Park, NC). Reactions were performed on a PTC-100 thermal cycler (Bio-Rad Laboratories, Hercules, CA). Amplification products were separated on 2% agarose gels, stained with ethidium bromide, and visualized under ultraviolet light.
Data Analysis
Percent PMGF was calculated as (number of verified survivors/total number of plants) x 100. Analysis of variance for PMGF was performed in SAS Proc GLM (SAS, 2004) on verified PMGF results from samples taken within 6 m of the pollen source. Means separation was conducted using Fisher's Least Significant Difference (LSD) at 
= 0.05.
Experimental Plots
In a separate experiment, a Nelder wheel design (Nelder, 1962) was used to measure gene flow from smaller pollen sources. Plots were planted in fall 2003 and 2004 and sampled the following summer. Alternating strips of the IS cultivars Prairie Red and Halt were planted around a central 10-by-10 m block of Above. Dates of 50% heading for the three cultivars were recorded. Wind speed and direction data were obtained from a permanent USDA weather station 400 m from the site. Samples were collected by hand harvesting all wheat heads within a 1 m2 area at 1, 3, 7, 11, 15, 23, 31, and 39 m from the edge of the pollen source along eight transects in north, northeast, east, southeast, south, southwest, west, and northwest directions. In 2004, two additional transects were sampled on a north by northwest and a west by northwest orientation. Farther distances were sampled where the size of the recipient wheat plot permitted, up to 70 m from the pollen source. These samples were processed separately from commercial field samples and evaluated using a greenhouse screening method.
Greenhouse Screening Method
Samples from the 2004 Nelder wheel were screened by planting 12 rows of 15 seeds each in rectangular flats with commercial potting media (Sunshine Mix #3, SunGro Horticulture, Bellevue, WA). Each sample was planted in two flats in each of two replications. Flats were placed in a greenhouse under natural light supplemented with 400 W sodium halide lamps to provide a 14 h daylength and watered as required. Plants were sprayed at the two- to three-leaf stage with imazamox at 35 g ha–1, 0.25% (v/v) non-ionic surfactant, and 1.0% (v/v) urea ammonium nitrate in a research track sprayer (DeVries Manufacturing Corp., Hollandale, MN) calibrated to apply 187 L ha–1 at 206 kPa. Two days after treatment, plants were clipped to approximately 1 cm above the newest emerging leaf. Plants that re-grew and displayed an injured phenotype were identified as heterozygous survivors. The number of survivors divided by the number of emerged plants in each sample was used to calculate PMGF.
Samples from the 2005 Nelder wheel experiment were screened with a more efficient method of soaking seeds in imazamox, described by Gaines et al. (2007). A foliar application of imazamox as described for the 2004 samples was applied 10 to 14 d after emergence to eliminate any susceptible plants that escaped imazamox treatment during seed soaking. The number of survivors divided by the expected number of emerged plants in each sample was used to calculate PMGF. The methods used in 2004 and 2005 were compared by correlating PMGF estimates from a set of commercial field samples taken in 2003.
Empirical Model
As a starting equation for this empirical modeling approach, we used the following "General Wheat Model" (GWM) proposed by Gustafson et al. (2005):
 | [1] |
 | [2] |
 | [3] |
Equation [3] was modified for additional analysis in Proc Nlmixed. For a given location, Hi (number of heterozygous plants) is binomially distributed with ni trials and a probability given by
 | [4] |
. Three lines were computed based on the model parameter estimates. The first line is a model of the median response, which represents a typical location where l = 0,
 | [5] |
The second line represents an estimate of the response for a location at the upper 95th percentile of locations, 10
, where
 | [6] |
The third line represents an upper 95% confidence limit on 10
and requires a standard error of calculated using established methods for variances of linear combinations of parameter estimates. An interpretation of this line is that we have 95% confidence that 95% of locations in a given RHC have gene flow less than this bound at a given distance.
|
RESULTS |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The highest observed gene flow in a sample was 5.3% in ‘Jagger’ at 0.23 m, while no gene flow was detectable in 125 (27%) of the samples at distances ranging from 0.3 to 61 m from the pollen source. Gene flow was detected in at least one sample from every location and from each of 18 cultivars tested. Across all three years, 33% of samples were taken within 6 m of the pollen source and 67% of total observed gene flow occurred within that distance; 77% of samples were taken within 30 m and 92% of observed gene flow occurred within that distance. The farthest distance at which gene flow was detected was 61 m, our furthest sampling point, and the maximum outcrossing at that distance was 0.25% in a Prairie Red sample.
Analysis of variance for gene flow in samples taken within 6 m of the pollen source indicated that cultivar, year, and a cultivar-by-year interaction were all significant (P < 0.01, data not shown). Cultivars with mean PMGF > 1% in at least one year were Jagger, Prairie Red, and TAM 107 (Table 3 ), all of which are early heading cultivars (RHC 1 or 2, Table 2). Ankor (RHC 5) also had relatively high cross-pollination rates in two of the three years.
|
|
Prediction Estimates
The final equation for the GLMM was:
 | [7] |
Parameter estimates were produced in Proc Nlmixed and used to construct estimates of the lines for a median location (Eq. [5]), a location at the upper 95th percentile (Eq. [6]), and an upper 95% confidence limit for the upper 95th percentile of locations for each relative heading class z (Fig. 1A
-1F). Model predictions were converted from probabilities to percentages for graphing purposes. Data from commercial fields and cultivar strip trials were included in the GLMM. Parameters and standard errors in parentheses estimated from the GLMM were a = –6.75 (0.02), b = 0.41 (0.01), d = 2.54 (0.11), and = 1.14 (0.12), all significant with P < 0.0001. Median location estimates (Eq. [5]) and upper 95% confidence limits on the upper 95th percentile of locations by RHC (z) are shown in Fig. 2A and 2B
.
|
|
|
DISCUSSION |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
At three of 56 locations (5%), all samples exceeded the estimated 95% confidence limit for the upper 95th percentile of locations, which is acceptable within the prediction limit. These included one location each of Jagger (Fig. 1B), ‘Ankor’ (Fig. 1D), and ‘Platte’ (Fig. 1E). In addition, one sample of Ankor out of 10 total samples at one cultivar strip trial location in 2003 exceeded the prediction limit (Fig. 1D). Weather stations near the three locations that exceeded model predictions recorded near-freezing or freezing temperatures at some point during boot and heading stages (Colorado Agricultural Meteorological Network, available at http://ccc.atmos.colostate.edu/
coagmet; accessed 15 June 2006; verified 12 Sept. 2007). This may have been cold enough to cause low levels of male sterility in the recipient fields by killing developing anthers (Shroyer et al., 1995), thus increasing the rate of outcrossing.
The RHC parameter explains significant variation in these data, indicating that cultivars that headed earlier than the pollen source (RHC 1 and 2) had higher levels of gene flow (Fig. 2A and 2B). The 95% confidence limit estimated for RHC 1 is high in part due to the influence of gene flow observations at greater distances (Fig. 1A). Whether the higher PMGF estimates in classes 1 and 2 are due to timing of heading relative to the pollen source or are confounded with other genetic attributes of Prairie Red and TAM 107 (both RHC 1) and Jagger (RHC 2) is difficult to determine because these cultivars are the only ones sampled in their respective classes. Based on the observation that Prairie Red and Jagger had wider floret opening at anthesis (Gaines, 2006), genetic variation in floret opening and stigma receptivity to foreign pollen may account for part of the cultivar variation.
Less variation in PMGF was observed in experimental plots than in commercial fields. Gene flow was not detected as far from the pollen source, lower average PMGF was found, and a higher percentage of samples had no detectable gene flow. This was probably due to the small area of the pollen source. The association of earlier relative heading with higher gene flow was consistent with the commercial field results. Prairie Red is earlier heading than Halt and had higher mean and maximum gene flow; however, this relationship would have been difficult to infer based solely on experimental plot data because only two cultivars were included in the design.
Modeling wheat PMGF with a GLMM allows for two important improvements over the GWM proposed by Gustafson et al. (2005). First, it allows the inclusion of a random location effect which combines such site-specific variables as wind speed and direction, temperature, humidity, and topography. Second, a GLMM allows PMGF to be modeled explicitly as a multiple of a binomial random variable, for which values of zero are allowed. This approach allows the binomial variance to be modeled separately from the location variance. Advantages of Proc Nlmixed over other considered alternatives include the ability to explicitly estimate c (the transformation used for distance), to compare models using AIC to evaluate fit, and to provide an approximate covariance matrix of all model parameters including (the standard deviation of the random location effect). Because many locations were sampled, an estimate of location to location variation enables a prediction of gene flow at a typical location and an upper estimate of gene flow. Allowing for the random effect in predictions of upper bounds is very important, as the upper 95% confidence limit line for the upper 95th percentile line is approximately 10 times the median location estimate (Fig. 2A and 2B).
Median and upper 95% confidence limit gene flow estimates in this study are higher under early heading date classes than the GWM proposed by Gustafson et al. (2005). Part of this difference may be due to the sampling of many more cultivars and locations in this study than the studies used to develop the GWM (Griffin, 1987; Hucl, 1996; Hucl and Matus-Cadiz, 2001; Martin, 1990; Matus-Cadiz et al., 2004). Four of the five data sets used to develop the GWM measured gene flow in spring wheat cultivars, as opposed to winter wheat cultivars in our study. Winter wheat in Colorado is vulnerable to environmental stresses such as freezing temperatures during floral development and drought stress, two factors known to increase outcrossing. The pollen source in this study, Above, tends to head earlier than most varieties. Recipient cultivars that were earlier heading than Above tended to have higher gene flow, while later ones had less. This result is consistent with the report of Matus-Cadiz et al. (2004) that noted overall gene flow was higher if the pollen source headed later than the recipient plot. This trend may indicate that bordering fields are more receptive to gene flow if environmental conditions cause some male sterility in recipient plants. If under these conditions a recipient cultivar is earlier than the neighboring pollen source, pollen will be available for male sterile florets when the source cultivar sheds pollen.
In summary, we have shown that significant gene flow occurs in wheat at the commercial field-scale, with higher levels and greater distances of PMGF than in smaller experimental plots. Factors associated with elevated gene flow included a pollen source that sheds pollen somewhat later than the receiving fields and weather conditions that may have caused some level of male sterility in the recipient plants. Flowering traits specific to certain cultivars may also result in higher levels of PMGF. A GLMM was used to fit a median estimate of PMGF based on relative heading date and an upper estimate to account for unpredictable environmental variables, such as wind speed and direction. Based on variation due to these and other factors, our confidence limits should represent the highest levels of gene flow expected to occur in wheat in the west-central Great Plains. Our results will be useful to biotechnology regulatory agencies, seed production organizations, wheat growers, and others seeking to minimize gene flow in wheat.
|
ACKNOWLEDGMENTS |
|---|
|
NOTES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Received for publication July 23, 2007.
|
REFERENCES |
|---|
|
TOPNOTESABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. J. Willenborg, E. C. Luschei, A. L. Brule-Babel, and R. C. Van Acker Crop Genotype and Plant Population Density Impact Flowering Phenology and Synchrony between Cropped and Volunteer Spring Wheat Agron. J., November 1, 2009; 101(6): 1311 - 1321. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. J. Willenborg, E. C. Luschei, A. L. Brule-Babel, and R. C. Van Acker Flowering Phenology and Synchrony between Volunteer and Cropped Spring Wheat: Implications for Pollen-Mediated Gene Flow Crop Sci., May 11, 2009; 49(3): 1029 - 1039. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. J. Willenborg, A. L. Brule-Babel, L. F. Friesen, and R. C. Van Acker Response of Heterozygous and Homozygous Imidazolinone-Resistant Spring Wheat Genotypes to Imazamox Crop Sci., November 24, 2008; 48(6): 2107 - 2114. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| The SCI Journals | Agronomy Journal | Vadose Zone Journal | |||
| Journal of Natural Resources and Life Sciences Education |
Soil Science Society of America Journal | ||||
| Journal of Plant Registrations | Journal of Environmental Quality |
The Plant Genome | |||
6 m from the pollen source, by cultivar and year.
