J. Brown, D.C. Thill, A.P. Brown, C. Mallory-Smith, T.A. Brammer and H.S. Nair

Department of Plant, Soil, and Entomological Sciences, University of Idaho, Moscow, ID 83844-2339.


Brassica species are particularly receptive to gene transformation techniques. There now exist genetically engineered canola genotypes with resistance to glyphosate, sulfonylurea or glufosinate herbicides. The main concerns of introducing herbicide resistant cultivars into commercial agriculture are: i. these crops will become difficult to handle volunteer weeds in following crops and ii. introgression of the engineered gene into related weed species. A survey of northern Idaho and eastern Washington showed that a number of weeds closely related to cultivated canola, wild mustard (B. kabel (DC) L.C. Wheeler), tumble mustard (Sisymbrium altissi,u, L.), birdsrape mustard (B. rapa L.), flixweed (Descurainia sophia L.), black mustard (B. nigra L.) and field pennycress (Thiaspi arvense L.) all bloom simultaneously with canola crops. In 1993 and 1994, a field study to investigate pollen movement and cross-pollination between herbicide resistant canola and non-resistant canola was carried out at two locations in Northern Idaho. In these studies, glufosinate resistant canola lines were planted in the center of a 62 m, square border, of non-resistant canola. The non-resistant border was made up with a mixture of three canola cultivars (`Westar', `Legend' and `Helios') each with a different flowering time in order to ensure synchrony of flowering with the resistant center and non-resistant border. Seed was collected from the susceptible border every 1.5 m along 16 rays spaced 22.5° apart. Each ray was 26 m long. Each sample was threshed separately and the seedling progeny screened for herbicide resistance (any resistant plants being the result of hybridization with resistant plants from the center plot). A second field study was carried out at two locations to determine the frequency of natural pollination that may occur when herbicide (glufosinate) resistant canola plants are grown in close proximity to local weeds. Each of three weed species (wild mustard, black mustard and birdsrape mustard) were examined in plots with a 1:1 mixture of weed and transgenic herbicide resistant canola. At harvest, weed plants and canola plants were separated by hand and the seed from each threshed separately. All weed seeds collected were screened for herbicide resistance (resulting from interspecific crossing). In a greenhouse study, canola was used as both male and female parents in crosses to the same three related weed species. After pollination, pollen grain germination, pollen tube development, fertilization, embryo and endosperm development were monitored on each of the 64 possible cross combinations (including selfs) over time. The potential of backcrossing, after initial hybrids are created, was also investigated in the greenhouse. Findings from these studies will be used in conjunction with field studies to develop simulation models of what may happen in nature. From the preliminary results it is difficult to make strong conclusions. However some indications already observed are: i. canola seed can be readily transported throughout a region and therefore there is a risk that these crops will become volunteer weeds; ii. canola pollen can move more than 26 m and movement is affected by wind direction; iii. canola and some related weeds can combine to produce hybrid plants under glasshouse and possibly field conditions; iv. herbicide resistance is expressed in the hybrid; and v. bridge crosses could play a major role in the movement of herbicide resistant genes into the natural weed population.

Key words: Brassica species, transgenic herbicide resistance, gene transfer


Over the past decade, plant genetic engineering techniques have been developed where specific characters (genes) can be introduced into a plant in a relatively straightforward manner, provided the genes coding for the character have been identified. Brassica species are particularly receptive to gene transformation techniques and several reports have been presented where canola (Brassica napus L. and B. campestris L.) has been genetically transformed for specific genes (DeBlock et al., 1989; Moloney et al., 1989; Thomzik and Hain, 1990).

Effective weed control is a major problem in canola production. Although no information is available regarding the situation within the USA, yield losses due to weeds in Canadian canola were about 10% and have been estimated to cost Canadian farmers over $306 million each year (Chandler et al., 1984). Brassicaceae weeds also can affect canola oil and meal quality (Thomas, 1984). It is no surprise therefore, that a major objective of molecular biologists has been to produce transgenic canola lines with herbicide resistance. There now exist canola genotypes with transgenic resistance genes for glyphosate, sulfonylurea and glufosinate herbicides.

A major concern of introducing transgenic herbicide resistant crops into agriculture is the spread of the engineered gene(s), particularly by pollen, to related weed species (Keeler, 1989; Williamson, 1991). New weeds could potentially be produced from backcrossing to a hybrid and their invasion into the natural ecosystem would cause change (Keeler, 1989). Indeed, some researchers believe that producing genetically engineered herbicide resistant crop species will ultimately lead to an increase in herbicide use (Hoffman, 1990; Ellstrand and Hoffman, 1990; Williamson, 1991). In addition to hybridization with weed species, herbicide resistant plants could flourish as volunteer weeds on neighboring farms (Botterman and Leemans, 1988; Williamson, 1991).

Preliminary surveys of grass and broadleaf weeds that infect canola fields in the Inland Northwest and Montana show that the same weeds that commonly infest small grain cereal fields also infest canola crops (Johnston, 1992; Brennan and Thill, 1993). These include several Brassicaceae genera including wild mustard (B. kaber (DC.) L.C. Wheeler), black mustard (B. nigra (L.) W.J.D. Koch), birdsrape mustard (B. rapa L.), shepherdspurse (Capsella bursa-pastoris (L). Medik.) tumble mustard (Sisymbrium altissimim L.) and field pennycress (Thlaspi arvense L.). Preliminary studies under controlled conditions (Brown and Brown, 1995) have suggested a strong possibility of hybridization between some of these weed species and transgenic canola.

One of the major pathways in plant evolution is through hybridization between related species (Williamson, 1991). Many crop-weed comparisons show that plants can evolve into invasive genotypes based on a few gene polymorphisms (Keeler, 1989; Hoffman, 1990). Herbicide resistance in many cases can be achieved by the transfer of a single gene (Schulz et al., 1990). Gene expression varies with genetic background, due to epistasis, linkage and pleiotropy. Therefore, it can be difficult to predict how the genetically engineered gene(s) will be expressed in a related weed species (Colwell et al., 1985; Tiedje et al., 1989).

There is substantial recent literature on other intergeneric crosses within the Brassicaceae family, and an even larger number on hybridization between different Brassica species. B. napus has been combined with several species of Diplotaxis and Eruca by sexual crosses and embryo rescue (Ringdahl et al., 1987; Batra et al., 1990), as well with various other Brassica species (Jourdan et al., 1989; Sjodin and Glimelius, 1989; Glimelius et al., 1990). Fusions of B. napus with both diploid Brassicas (e.g. B. oleracea, B. nigra) and with amphidiploid ones (e.g. B. juncea, B. carinata) have yielded viable plants.

Other intergeneric Brassica hybrids recently produced include various combinations of B. juncea, B. campestris, Diplotaxis, Eruca, Raphanus, Moricandia, and Trachystoma (e.g. Toriyama et al., 1987a; 1987b; Agnihotri et al., 1990a; 1990b; Sikdar et al., 1990; Takahata and Takeda, 1990; Kirti et al., 1992). These hybrids generally showed some degree of fertility in crosses and a range of chromosome numbers in the progeny. Preferential loss of chromosomes from one parent was occasionally observed (Fahleson et al. 1988).

Most interspecific crosses do not produce mature seed due to failure of endosperm development, resulting in wrinkled or empty seeds (Nishiyama et al., 1991). However, normally incompatible interspecific hybridization can spontaneously produce a few seeds which usually yield true F1 plants as a result of unexpected ploidy changes (Nashiyama et al., 1991). This phenomenon has been documented between B. napus and B. campestris, and between B. napus and B. nigra.

The aim of this project is to determine the feasibility and frequency of gene flow between transgenic canola (B. napus) and related weed species. Experiments carried out under controlled conditions will provide data that can be used to model possible genetic transfer that may occur under field conditions. Similarly, competitive fitness of plants from hybrid combinations will add invaluable information concerning the potential risks to the environment should genetic transfer occur between cultivated and weedy Brassica's.


Field Pollen Movement. A field pollen movement study was carried out in the spring of 1993. In this experiment a plot (10 m2) of glufosinate resistant canola was surrounded by an 8 m border of glufosinate susceptible canola (`Helios'). At harvest the susceptible border was sampled in four directions (north, east, south and west) every 1.5 m from the herbicide resistant center. Four samples, each of 200 seeds, were sown from each sample position. When seedlings reached a 1-2 leaf stage they were sprayed with the recommended field rate of glufosinate. Surviving plants were sprayed a second time to avoid errors caused by escapes. Plants surviving the second spray were counted and assumed to have resulted from cross pollination with plants from the herbicide resistant center.

In the following year (1994) pollen movement was examined in greater detail in an experiment grown at two locations in Northern Idaho (Moscow and Genesee). At both locations a plot of transgenic glufosinate resistant canola was surrounded by a 30 m wide border of herbicide susceptible canola. The susceptible canola border was planted using a mixture of seed from three canola cultivars (`Legend', `Westar' and `Helios') each with a different flowering time to ensure synchronous flowering with resistant genotypes in the center. At maturity, seeds were collected from the non-transgenic canola border every 1.5 m along sixteen 26 m long rays spaced 22.5° apart. The seed from each sample position was grown in the greenhouse to the 1-2 leaf stage, and sprayed with glufosinate at the 0.42 kg ai/ha. Sprayed plants were evaluated 7 days after treatment. To ensure there were no escapes, all surviving plants were sprayed again with the herbicide dose previously mentioned and survivors counted.

In both years of study, herbicide resistant and susceptible cultivars were included in the herbicide spray design as controls.

Greenhouse Interspecific Hybridization. Interspecific and intergeneric hybridization between transgenic herbicide resistant canola and three related weed species (wild mustard, birdsrape mustard and black mustard) was examined in greenhouse experiments. Two genotypes of transgenic canola and one susceptible canola (`Cyclone') were crossed in all possible combinations, including selfs and reciprocals in a full diallele crossing design with the three weed species.

Mercury lights were used to supplement natural lighting and provide a 16 hour day-length. Greenhouse temperature was not effectively controlled. However, daytime temperatures were around 21 ± 5°C.

Two days after pollination, 10 siliques were taken at random to examine pollen germination on the stigma and pollen tube development down the style towards the ovary. The method used was similar to that of Martin (1959). Pollinated styles were fixed in absolute alcohol:glacial acetic acid (3:1) 48 hours after pollination and stored at 4°C until examined. The fixative was replaced with 1N sodium hydroxide (NaOH) and the styles were left at room temperature. Hydrolysis was completed with a change of 1N NaOH and a further half hour at 60°C. After the hydrolyzed styles were rinsed with water and stained with a methyl blue solution [0.2% methyl blue + 2.0% K3PO4 (w/v)] they were observed for fluorescence using wavelength of light in the range 350-400 nm. The styles were examined for pollen germination and pollen tube growth in the style and ovary.

After pollination, developing siliques were harvested at 2 day intervals between 4 and 28 days after pollination and fixed in absolute alcohol:glacial acetic acid (3:1) and stored at 4°C until examination. The ovary was opened and any developing ovules were removed. Individual ovules were opened and a few drops of HCl-carmine (Snow, 1963) placed inside. After a few minutes, excess stain was rinsed away with 70% ethanol. The ethanol was removed using a piece of filter paper and replaced with Rattenburys' Fluid (45% acetic acid:glycerine, 10:1). The endosperm and embryo were teased from the ovular tissue, which was then discarded. The prepared material was examined using a Jenval transmitted light research microscope. The developmental stage of the embryo and endosperm and any abnormalities present were noted. A full data set was not available from this study at the time of producing this report and only a sub-set of observations from one silique per sample are presented.

Field Interspecific Hybridization. Transgenic, glufosinate-resistant canola was planted in a 1:1 mixture with three weed species (wild mustard, birdsrape mustard and black mustard) in 1 x 5 m plots arranged as a randomized complete block design with four replications. Seeding rate of each species was adjusted prior to planting to account for any differences in germination. At harvest weed plants were threshed separately from the canola mixtures. Weed seed collected from the canola:weed mixtures were grown in the greenhouse to the 1-2 leaf stage, and sprayed with glufosinate at 0.42 kg ai/ha. Sprayed plants were evaluated 7 days after treatment. To ensure there were no escapes, all surviving plants were sprayed again with the herbicide dose previously mentioned and survivors counted. Herbicide resistant and susceptible cultivars were included as mentioned previously.

Biological Fitness Study. A number of fertile interspecific plants were produced from hand crossing canola and birdsrape mustard. Most of the hybrid plants were sterile. However, two F1 families were found to set self seed. These two hybrid lines were backcrossed to each of the parent species. This resulted in eight families (canola, birdsrape mustard, F1(A), F1(A) x canola, F1(A) x birdsrape mustard, F1(B), F1(B) x canola and F1(B) x birdsrape mustard) which were grown in a randomized block design in the greenhouse with two replicates each of five single plants per family. Data were collected from individual plants for pre-harvest characters and seed yield.


Field Survey of Weed Populations. Twenty fields were surveyed in Latah, Nez Perce and Lewis counties in Idaho and Whitman county in Washington. Wild mustard (B. kabel (DC) L.C. Wheeler), tumble mustard (Sisymbrium altissi,u, L.), birdsrape mustard (B. rapa L.), flixweed (Descurainia sophiaL.), and field pennycress (Thiaspi arvense L.) all bloomed simultaneously with canola (Brassica napus L.). The weeds occurred in the same field with canola, or in ditches next to canola fields. Among the weed populations observed wild mustard and birdsrape mustard occurred with greatest frequency. A number of canola plants were also found by roadsides and ditches which must have volunteered from farmers fields.

Pollen Movement and Cross Pollination Study. The percentage cross pollination between herbicide resistant plants and the non-resistant border decreased with greater distance from the herbicide resistant pollen source. When grown adjacent to each other, there was 6.3% cross pollination between plants. However, cross pollination was greater than 1:200 seeds when plants were separated by 7.5 m (Figure 1a). The percentage cross pollination changed according to direction from the pollen source (Figure 1b). Greatest cross pollination occurred down wind of the pollen source.

The weather conditions in 1994 were drastically different from those in the previous year. 1994 was a hotter and dryer season which greatly reduced the time plants remained in flower, pollen load per flower and the number of insects observed in the experiment. With the exception of the adjacent samples, this resulted in fewer cross pollinations compared to the wetter and cooler conditions of the previous season. The percentage cross pollination according to distance from pollen source is shown in Figure 2. This figure shows that hybridization decreases rapidly with increasing distance from the transgenic canola. A very high frequency of cross pollination was found in the sample adjacent to the transgenic resistant center. This high frequency should be treated with caution as there is some suggestion that a proportion of these samples was taken from the transgenic center by error. Beyond 5 m from the pollen source the frequency of cross pollination is only about one fertilization in 1000, although there was no suggestion of the frequency reducing below that frequency over greater distances than was examined in the study. Indeed, a small number of transgenic resistant hybrids were found even at the furthest distance from the pollen source (i.e. up to 26 m).

Hybridization frequency and pollen movement also varied by site and by wind direction. Site differences included terrain (side of hill vs. on top of hill), and climate (wind direction, speed). The average wind direction was from the southwest at both locations and the average wind speed was 2.5 m/s at Genesee and 1.8 m/s at Moscow.

Data collected on the frequency of pollination related to distance of pollen movement were analyzed using a maximum likelihood approach to assess the risk of gene escape. A general model was as follows:


where c is the frequency of cross pollination when two plants are adjacent and R(x) is the probability that pollen grains travel at least x distance away from the source plant.

The best fit of possible forms of R(x) was the Weilbull model. This model states that pollen is carried from the source and is deposited at a variable rate, which is distance dependent. Therefore:

R(x)= exp(-axb);

where x is the mean distance traveled by the pollen, and a, b and c are parameters estimated from the data collected.

Table 1 shows the estimated parameters from the Weilbull model at each location (Genesee and Moscow) estimated from up wind and down wind directions. One of the most important parameters is c, which indicated a changing rate of hybridization between upwind and downwind directions, especially at the Genesee site. This suggests that wind direction was also one of the factors effecting the rate of pollen movement; and therefore, the hybridization between the transgenic and non-transgenic canola. On average, pollen movement (Table 1) was always less than 1 m from the source plant. However, as is shown in Figure 1 and Figure 2, cross pollination still occurs at low frequency at far greater distances.

Greenhouse Interspecific Hybridization. Pollen grain germination and pollen tube development was observed in all of the 12 hybrid species combinations (including reciprocals) studied. Some hybrid combinations, for example canola x black mustard, showed mostly an incompatible pollen germination reaction, where many short twisted pollen tubes did not penetrate the stigma. In these cases very few pollen tubes were observed around the ovary. Other hybrid combinations, for example wild mustard x canola, showed considerably more compatibility with a much larger presence of pollen tubes in the ovary. Pollen tubes were observed penetrating the micropyle in all crosses.

Fertilization occurred in 10 of the possible 12 hybrid combinations, as evidenced by the onset of embryo and endosperm development. A great deal of variation existed in the stage of development observed 16 days after pollination (Table 2). Some cross combinations aborted at the early globular stage, very soon after fertilization. However, excluding selfed crosses, 30% of cross combinations studied developed to the heart stage or a later stage of embryo growth. Hybridizations where black mustard was used as the maternal parent consistently resulted in early embryo abortion. However, as a pollen parent in hybrid combinations, embryo development of black mustard hybrids advanced to later stages of development.

Overall, the three weed species examined combined to a relatively high degree in hybrid crosses with other weed species, while weed x canola hybrids showed the least embryo development compared to the development of selfed species.

The different species examined in this study showed different rates of embryo development (Table 3). The rate of embryo development in canola was faster than any of the weed species, after self pollination. Canola developed globular embryos after 8 days and heart shaped embryos after 12 days. Globular embryos were not observed in any of the three weed species within the first 10 days after fertilization. Heart shaped embryos were not observed in birdsrape mustard that had been self pollinated until 14 days after pollination and after 16 and 18 days, respectively, for wild mustard and black mustard.

The general trend of embryo developmental rates, where embryos were observed, tended to follow the rate of the slower developing parent, irrespective of which is used as the maternal parent.

The normal pattern of endosperm development in Brassica is to progress from a coenocytic state in the early stages, to a cellular endosperm corresponding to the heart stage of embryo development. The primary cause of early embryo abortion was failure of the endosperm to develop. Very small amounts of degenerating endosperm were observed. Aborting endosperm is characterized by the presence of large nuclei and nucleoli. In this study, dumb-bell nuclei, anaphase bridges, lagging chromosomes, various widths of the metaphase plate and split spindles were observed in dividing hybrid endosperm.

A proportion of pollinations were allowed to mature and set seed. Mature seed was obtained from the interspecific weed hybridizations of birdsrape mustard x wild mustard and wild mustard x birdsrape mustard. In addition mature seed was also obtained from crosses between transgenic canola and birdsrape mustard.

Biological Fitness of Hybrid Combinations. As most successful hybrid cross combinations have been between canola and birdsrape mustard, this cross combination was examined with respect to the biological fitness of the hybrids and their progeny. Most hybrid plants produced were sterile (probably due to chromosome abnormalities or a lack of homologous pairing at meiosis). All hybrid plants produced; however, expressed herbicide resistance, although to varying degrees. This observation suggests a gene interaction effect.

Fertile hybrid plants were readily produced by treating with colchicine. However, two F1 hybrid plants produced self seed without colchicine treatment. These two hybrid lines were backcrossed to both parents and grown in a replicated greenhouse study to examine the biological fitness of the hybrids and backcrosses compared to the parental species. F1 hybrid plants appeared to show lower seedling vigor compared to either parent species. However, vigor was restored to be equal or exceed the parent species after backcrossing (Table 4). Similarly, F1 hybrid plants were shorter with fewer leaves, less leaf area and lower yield than the parent species. However, backcrossing to either parental species resulted in increased plant height, leaf number and area and also greater yield.

Field Hybridization. Seed collected from weed plants in resistant canola:weed species mixtures were all screened for herbicide resistance. No species hybrid seedlings (i.e. herbicide resistant seedlings) were found in black mustard or wild mustard. However, a few (approximately 1:1000 seedlings examined) hybrid plants were obtained between canola and birdsrape mustard. These hybrid plants are presently being investigated in more detail in cytological studies.


Pollen Movement. Canola pollen can move, by insect or wind, further than 26 m. Wind direction and speed are factors in pollen movement with greatest frequency of pollination being down wind. In the 1994 study, on average, pollen moved less than 1 m from parent source. High temperatures and drought conditions in the 1994 season may have had a restrictive effect on cross pollination due to reduced flowering time, fewer insects and lower pollen load. However, a low frequency of cross pollination was observed up to 26 m from the source, the greatest distance considered in this study. Herbicide resistant canola fields grown adjacent to non-resistant canola crops will therefore have high potential of cross pollination between these fields.

Glasshouse Interspecific Hybridization. Mature hybrid seed was obtained from the crosses, birdsrape mustard x wild mustard and its' reciprocal and the interspecific cross canola x birdsrape mustard. The latter canola x weed hybrid was also found to occur under field conditions.

In all hybrid combinations examined, pollen germinated and pollen tubes penetrated the style. Only two species cross combinations did not show any pollen tubes in the ovary after hybridization. It should, however, be noted that one of these was the self species cross wild mustard, which later produced mature seeds. Therefore, this suggests that the sample size presented in this report is too small to detect some necessary differences or occurrences. It also should be noted that under field conditions, there is the potential of millions of hybridizations that could occur rather than the relatively few examined here.

A number of hybrid embryos developed to the heart stage which appears to be a crucial stage of embryo development (Brown, 1985). If the embryo develops to this stage, there is a potential of continuing development to maturity.

Spontaneously doubled fertile F1 hybrids showed lower vigor and seed yield compared to either parent species. However, it was noted that the backcrossed plants, obtained by crossing the F1 to either canola or birdsrape mustard, were as vigorous and high yielding as their parent species.

Many hybrid cross combinations in Brassica result in sterile pollen due to cytological defects and lack of chromosome pairing in meiosis. Sterile plants have a greater tendency to be cross pollinated by other plants, or species, than fertile ones and a second cycle of hybridization may occur in gene flow situations. The possibility of bridge crossing is just beginning to be explored as part of this project.

In order to explain potential bridge crosses, consider the hybrid combination birdsrape mustard x wild mustard (which produced mature seed in this study, in either direction). Birdsrape mustard has 10 paired chromosomes (2n = 20) and wild mustard has 9 paired chromosomes (2n = 18). A hybrid between the species would most likely be an allotetraploid with 19 paired chromosomes (2n = 38) and hence will have the same chromosome number as canola. Having the same chromosome number, especially with a common genome (the A genome from B. rapa) may be an important factor in hybrid formation. A hybrid between these two weed species could therefore act as a bridging species with canola and could further add to the risk of gene flow of transgenic herbicide resistance into weed species.

In addition, bridging and backcrossing also offers the prospect of combining multiple herbicide tolerance within the same weed species. Further studies need to be carried out to determine the possibility of gene flow with sterile single hybrid combinations. For example, F1 hybrids need to be inter-crossed to determine the possibility of bridge crossing (i.e. [canola x weed A] x weed B, or [weed A x weed B] x canola). In addition, the degree of fertility in hybrid combinations and the potential of backcrossing needs further investigation.

Overall, it is difficult to make strong conclusions as the data sets are only partially complete. However some indications already observed are: i. canola seed can be readily transported throughout a region and therefore there is a risk that these crops will become volunteer weeds; ii. canola pollen can move at least, or more than, 26 m and movement is affected by wind direction; iii. canola and some related weeds could combine to produce hybrid plants under glasshouse and field condition; iv. herbicide resistance is expressed in species hybrids; v. biological fitness of hybrid plants is less than parent species although fitness is increased by backcrossing; vi. bridge crosses could play a major role in the movement of herbicide resistant genes into natural weed populations.


The authors would like to thank Plant Genetic Systems N.V. Ghent, Belgium for providing the transgenic herbicide genotypes used in this research.


Agnihotri A, Gupta B, Lakshmikumaran MS, Shivanna KR, Prakash S, Jagannathan V (1990a) Production of Eruca-Brassica hybrids by embryo rescue. Plant Breeding 104:281-289.

Agnihotri A, Shivanna KR, Raina SN, Lakshmikumaran M, Prakash S, Jagannathan V (1990b) Production of Brassica napus x Raphanobrassica hybrids by embryo rescue: an attempt to introduce shattering resistance into B. napus. Plant Breeding 104:281-289.

Batra V, Prakash S, Shivanna KR (1990) Intergeneric hybridization between Diplotaxis siifolia, a wild species and crop Brassicas. Theor. Appl. Genet. 80:537-541.

Botterman J, Leemans J (1988) Engineering of herbicide resistance in plants. Biotechnol Genet. Engineer Rev. 6:321-340.

Brennan, JS, Thill DC (1993) Weed management in canola. In Proc. Idaho Weed Control Association. pp 44-51.

Brown AP (1985) Pollen, embryo and endosperm development following cross pollination within and between the crop species Brassica campestris, B. oleracea, B. napus and Raphanus sativus. Ph.D. Thesis, University of Edinburgh, Scotland, UK.

Brown J, Brown AP (1995) Potential gene transfer between Brassica napus and related weed species. (submitted to press).

Chandler JM, Hamill AS, Thomas AG (1984) Crop losses due to weeds in Canada and the United States. Weed Science Society of America, Champaign, IL 61820.

Colwell JK, Norse EA, Pimentel D, Sharples FE, Simberloff D (1985) Genetic engineering in agriculture. Science 229:111-112.

DeBlock M, DeBrouwer D, Tenning P (1989) Transformation of Brassica napus and Brassica oleraceausing Agrobacterium temefaciens and the expression of the bar and neo genes in the transgenic plants. Plant Physiol. 91:694-701.

Ellstrand NC, Hoffman CA (1990) Hybridization as an avenue of escape for engineered genes. BioScience Vol. 40 No. 6:438-442.

Fahleson J, Rahlen L, Glimelius, K (1988) Analysis of plants regenerated from protoplast fusions between Brassica napus and Eruca sativa. Theor. Appl. Genet. 76:507-512.

Glimelius K, Fahleson J, Landren M, Sjodin C, Sundberg E (1990) Improvements of the Brassica crops by transfer of genes from alien species via somatic hybridization. In Progress in Plant Cellular and Molecular Biology (Nijkamp HJJ, Van Der Plas LHW, Van Aartrijk J, eds.). Kluwer Acad Press. pp.299-304.

Heyn, FW (1977) Analysis of unreduced gametes in the Brassicaceae by crosses between species and ploidy levels. Z. Pflanz. 78:13-30.

Hoffman, CA (1990) Ecological risks of genetic engineering of crop species. BioScience 40:434-437.

Johnson P (1992) Montana Canola Weed Survey. In Proc. Western Canola Development Meeting. pp 37-43.

Jourdan PS, Earle ED, Mutschler MA (1989) Synthesis of male sterile, triazine-resistant Brassica napusby somatic hybridization between cytoplasmic male sterile B. oleracea and atrazine-resistant B. campestris. Theor. Appl. Genet. 78:445-455.

Kareiva P, Manasse R, Morris W (1990) Using models to integrate data from field trials and estimate risks of gene escape and gene spread. In Biological Monitoring of Genetically Engineered Plants and Microbes. Agricultural Research Institute, Bethesda. pp 31-42.

Keeler KH (1989) Can genetically engineered crops become weeds? BioTechnology I 1134-1139.

Kirti PB, Narasimhulu SB, Prakash S, Chopra VL (1992) Production and characterization of intergeneric somatic hybrids of Trachystoma ballii and Brassica juncea. Plant Cell Rep. 11:90-92.

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

Martin FW (1959) Staining and observing pollen tubes in the style by means of fluorescence. Stain Tech.34:125-128.

Moloney MM, Walker JM, Sharma KK (1989) High efficiency transformation of Brassica napus using Agrobacterium vectors. Plant Cell Rep.. 8:238-242.

Nishiyama I, Sarashima M, Matsuzawa Y (1991) Critical discussion of abortive interspecific crosses in Brassica. Plant Breeding 107:288-302.

Ringdahl EA, McVetty PBE, Sernyk AL (1987) Intergeneric hybridization of Diplotaxis spp. with Brassica napus: a source of new CMS systems? Can. J. Plant Sci. 67:239-243.

Schulz A, Wengenmayer F, Goodman HM (1990) Genetic engineering of herbicide resistance in high plants. Crti. Rev. Plant Sci. 9:1-15.

Sikdar SR, Chatterjee, Das S, Sen SK (1990) "Erussica", the intergeneric fertile somatic hybrid developed through protoplast fusion between Eruca sativa Lam. and Brassica juncea L. Czern. Theor. Appl. Genet. 79:561-567.

Sjodin C, Glimelius K (1989) Brassica naponigra, a somatic hybrid resistant to Phoma lingam. Theor. Appl. Genet. 77:651-656.

Snow, R (1963) Alcoholic hydrochloric acid-carmine as a stain for chromosomes in squash preparations. Stain Tech. 38:9-13.

Takahata Y, Takeda T (1990) Intergeneric (intersubtribe) hybridization between Moricandia arvensis and Brassica A and B genome species by ovary culture. Theor. Appl. Genet. 80:38-42.

Tiedje JM, Colwell RK, Grossman YL. Hodson RE, Lenski RE, Mack RN, Regal PJ (1989) The planned introduction of genetically engineered organisms : ecological considerations and recommendations. Ecology 70:298-315.

Thomas P (1984) Canola Growers Manual. Canola Council of Canada.

Thomzik JE, Hain R (1990) Transgenic Brassica napus plants obtained by cocultivation of protoplasts with Agrobacterium tumefaciens. Plant Cell Rep. 9:233-236.

Toriyama K, Hinata K, Kameya T (1987a) Production of somatic hybrid plants, "Brassicomoricandia" through protoplast fusion between Moricandia arvensis and Brassica oleracea. Plant Sci. 48: 123-128.

Toriyama K, Kameyta T, Hinata K (1987b) Selection of a universal hybridizer in Sinapis turgida Del. and regeneration of plantlets from somatic hybrids with Brassica species. Planta 170:308-313.

Williamson M (1991) Assessment of the hazards from genetically engineered plants : The work of the advisory committee on genetic manipulation intentional introduction sub-committee. In Herbicide resistance in weeds and crops (J.C. Casely, G.W. Cussans and R.K. Atkin Eds.) pp 375-386.

Figure 1. Percentage cross pollination observed in the 1993 field study according to distance from pollen source (a) and percentage pollination according to direction from pollen source (b).

Figure 2. Log of the percentage cross pollination according to distance from Genesee in 1994.