Scientific Narrative: Gene Flow¹ via Pollination and Crops Derived through Biotechnology

Table of Contents

1. Introduction
2. Background: Assessing Possible Impacts of Gene Flow
3. Background: Documenting Gene Flow
4. Gene Flow via Pollination - Maize
5. Gene Flow via Pollination - Soybeans
6. Gene Flow via Pollination - Oilseed Brassica Species
7. Molecular Strategies to Prevent Gene Flow from Biotech Crops
8. Endnotes

Farmers have grown crops derived through biotechnology since the mid-1990s. Examples of such crops include:

• corn and cotton with transgenes that protect the crop from insect pests, and
• corn, soybeans and canola with transgenes protecting the crop from a specific herbicide.

Farmers have rapidly adopted crops derived through biotechnology because they offer increased yield, cost savings, enhanced efficiency, convenience and improved environmental stewardship.

Despite the benefits of biotechnology, there have been concerns expressed by some organizations and individuals. Among these concerns is the potential of transgenes to move, via pollen, from biotech crops to other plants, and, in doing so, alter ecological relationships in unmanaged ecosystems or crop management practices in agricultural ecosystems. Gene flow could happen with any crop and related plants, whether it is biotech or non-biotech derived. This paper focuses on the potential impacts of transgenes moving from biotech crops to other related plants, wild or cultivated. It will address the following questions:

• What is the likelihood that a transgene will move from the biotech crop to an uncultivated, wild plant?
• What would be the impact, if any, of such gene flow on native populations of wild plants?
• What is the likelihood a transgene will move from the biotech crop to non-biotech, cultivated forms of the crop?
• What would be the impact, if any, of gene flow on neighboring, non-biotech crops?

Answers to these questions vary with the biotech crop, nature of the trait encoded by the transgene, the plant receiving the transgene, and a variety of environmental factors. Therefore, to thoroughly address the possibility of gene flow and its potential impacts, one must evaluate each situation systematically, on a crop-by-crop basis, using information on variables such as:

• presence or absence of relatives, either wild or cultivated, of the biotech crop;
• pollination system of the biotech crop and recipient plant;
• reproductive compatibility of the biotech crop and recipient plant;
• physical proximity of biotech crop to wild or cultivated relatives;
• population density of recipient plant in the vicinity of the biotech crop;
• possible effect on fitness² of the wild plant of the trait encoded by the transgene; and
• inherent weediness of wild relative or invasiveness of the cultivated variety³.

Any discussion of the effects of gene flow from biotech crops to other plants must clearly distinguish between the possible impacts on wild, uncultivated relatives and possible impacts on non-biotech, cultivated crops. These two potential paths differ from one another in virtually all key parameters, such as:

• the probability gene flow will occur;
• possible means by which it can occur; and
• nature of its potential impacts.

In addition, when focusing on the impacts of gene flow on cultivated crops, it may be necessary to differentiate improved, commercial varieties from cultivated landraces maintained by local farmers who save seed from one year to the next. The probability, mechanisms and significance of gene flow in these two classes of cultivated crops can differ.

Discussions should also recognize that gene flow can be multi-directional, which means that it can be from:

• biotech to non-biotech varieties and landraces, and vice versa;
• wild relatives to biotech varieties, and vice versa; and
• wild relatives to non-biotech varieties and landraces, and vice versa.

Finally, gene flow from crops to wild relatives is equally likely for biotech or non-biotech varieties (Hancock et al., 1996). There are a number of well-documented instances of gene flow from non-biotech varieties to wild relatives (reviewed by Barrett, 1983). Ellstrand et al. (1999) presents evidence for hybridization between non-biotech crops and wild relatives for 13 of the world's most important crops.

Return to Table of Contents

Pollen transfer from biotech crops to other plants does not, in and of itself, lead to fertilization; fertilization does not necessarily result in gene flow; and gene flow from biotech crops to other plants may or may not have significant ecological or agricultural impacts.

The ecological or agricultural significance of pollen transfer from a biotech crop to either a wild relative or a non-biotech cultivar is determined by the extent to which the following events occur.

• Pollen from the biotech crop fertilizes the recipient plant (cross-pollination).
• The fertilization produces fertile seeds that germinate into hybrid plants that express the gene and pass it to their offspring (hybridization).
• The transgenes are retained in a population (introgression).
• Plants with the transgene are more or less able to survive in their environment (fitness).

The first and second factors determine the probability of gene flow from the biotech crop to other plants, while the third and fourth affect the potential ecological or agricultural significance of the gene flow^4. A number of factors reduce the joint probability of gene flow both occurring and having adverse effects. These factors significantly decrease the potential impact of gene flow from biotech crops to other plants⁵.

2.1   Probability of Cross-Pollination

Gene flow depends, first and foremost, on successful cross-pollination, which we define as the transfer of pollen from one plant to a recipient plant and subsequent fertilization of the recipient plant's eggs. For cross-pollination between two plants to occur, viable pollen from one plant must reach the second plant prior to fertilization of the recipient's egg(s) by other pollen grains. Therefore, the amount of cross-pollination between any two plants depends upon many of the following factors.

• Reproductive compatibility, which depends on the degree of relatedness. (For example, corn will only cross-pollinate with other related maize plants.)
• Synchronicity between pollen shedding by one plant and egg maturation of the recipient plant.
• Pollen competition, which is affected by relative amount of pollen shed by one plant compared to other possible pollen sources.
• Physical proximity of pollen donor to recipient plant.
• Longevity of viable pollen.
• Distance pollen travels.
• Pollination mechanics (wind vs. insect pollinated; out crossing vs. self pollinating).
• Floral morphology.

For gene flow from biotech crops to other plants to occur, viable pollen containing the transgene and the capability of fertilizing the other plant, must reach the recipient plant prior to fertilization of its egg(s) by other, non-biotech pollen grains. The recipient plant might be a commercially improved, but non-biotech, variety, local cultivar or landrace, or a non-cultivated (wild) relative of the biotech crop. Cross-pollination between a biotech crop and a different plant can also occur in the opposite direction - the non-biotech variety, landrace or wild relative serves as the pollen donor and fertilizes the biotech crop.

Because non-biotech, cultivated varieties are planted in blocks and can self-pollinate, pollen from within the field is more likely to fertilize the crop than pollen from a biotech crop at some distance from the non-biotech variety. However, the adventitious presence⁶ of biotech seed with non-biotech seed may result in some biotech plants within a field of a non-biotech variety.

Seed companies spend considerable resources to ensure the highest possible level of seed quality and purity. Even so, virtually all seed supplies contain the adventitious presence of small amounts of genetic material or whole seeds from another seed line. Existing national and international regulatory bodies have taken adventitious presence into account for years, establishing thresholds and policies for purity levels in commercially improved seed supplies.

Adventitious presence can sometimes be traced to the seeds farmers purchase. During commercial seed production, pollen mixing or inadvertent mixing of parental seed lines during planting, harvesting or transporting can lead to the adventitious presence of other seeds or genetic material. On the farm, seeds also can be mixed if planters and other equipment are not thoroughly cleaned when farmers change varieties during planting.

Often the potential for cross-pollination is asymmetrical. The probability that cross-pollination between two plants is successful will vary according to which plant serves as the pollen donor and pollen recipient. Therefore, a demonstration that cross-pollination can successfully occur when the wild relative is the pollen donor does not necessarily mean the reverse is true. To accurately estimate the potential for gene flow from cultivated crops to their wild relatives, pollen from the crop must be used to fertilize the wild relative.

2.2   Successful Hybridization - Crop to Wild Relative

Cross-pollination is necessary, but not sufficient, for successful hybridization leading to gene flow. Once pollen fertilizes the egg of the recipient plant, post-fertilization barriers to successful hybridization may impede subsequent development. Gene flow can occur only if those barriers to hybridization are overcome, and the resulting hybrid offspring are viable and fertile.

Because successful hybridization is possible only between plants that are closely related to each other, gene flow from crops to wild plants is possible only if the crop is grown in the vicinity of its wild relatives. Therefore, it is occurs, it is most likely to occur in "center of diversity" regions in which relatives of the crop are common. Each crop has its own distinct and limited center(s) of diversity. For some crops, but not others, this region is also the center of origin for the crop. For example, centers of diversity for maize are located in Mexico, Central and South America, and the center of origin for maize is in Mexico (Piperno and Flannery, 2001; Smith, 2001; Matsuoka et al., 2002). However, according to Harlan (1992), the Near East, specifically along the Jordan rift, near the Mediterranean, is the center or origin of barley, but its centers of diversity are Ethiopia, Tibet, China and Japan. Therefore, gene flow from corn to wild relatives is germane only to corn growing in Mexico, Central America, and regions in South America; while gene flow from barley to wild relatives is relevant for barley grown in the Near East, Ethiopia, Tibet, China and Japan.

Another variable that should be evaluated when assessing the probability of gene flow from crops to wild relatives is the direction in which the hybridization takes place. Asymmetrical reproductive compatibility often leads to directionality constraints to hybridization. Most of the currently available data on reproductive compatibility between crops and wild relatives were collected by breeders attempting to improve cultivated crops with genes from wild relatives (pollen donor). To estimate gene flow from crops to wild relatives, the success of hybridization must be assessed with crops as the pollen donors.

Finally, according to Raybould and Gray (1993), it is highly unlikely that transgenes will change the rate at which crops hybridize with wild relatives, or the range of species with which they are sexually compatible. Therefore, the probability of successful hybridization leading to gene flow is as likely for biotech as non-biotech crop varieties.

2.3   Introgression

Successful hybridization of a biotech crop with a wild relative or non-biotech variety does not necessarily mean that the resulting gene flow will have a significant ecological or agricultural impact. To have a significant ecological or agricultural impact, the gene must first become established, or "introgressed" in a sizeable portion of the recipient plant population.

2.3a   Wild Relatives

For gene flow from wild relatives to crops and vice versa, introgression is first and foremost a function of the reproductive compatibility of the first generation (F1) hybrid (crop X wild plant) with the parental strains. If there are significant genome differences between the hybrid offspring and a parent, then meiotic abnormalities and recombination barriers will retard or prevent introgression, even if the F1 hybrids are viable and fertile when crossed with each other. For a new gene to become established in a wild plant population, it must be transferred repeatedly through a recurring series of hybridizations between the F1 and parental generations.

2.3b   Crop Varieties

In the case of gene flow among crop cultivars, there are two primary ways for transgenes to become incorporated into non-biotech cultivars: pollen and seeds. With some crops, cross-pollination can occur if fields of non-biotech and biotech crops are in proximity to each other. Some of the variables affecting the probability of this occurring are discussed above in the general description of cross-pollination, as well as below for each crop. However, even in those instances where fields of biotech crops are not growing in proximity to non-biotech varieties or land races, biotech material can become incorporated into non-biotech, commercial varieties and land races through a number of routes involving seeds.

2.3c   Seed Banks and Volunteer Plants

Seeds left in the field by harvesting equipment can germinate the next year. If these seeds were from a biotech variety, then the volunteer plants will produce pollen containing the transgene. Typically, about half of the pollen shed by a transgenic hybrid carries the transgenic trait. Assuming a non-biotech variety of the same crop is planted in the next growing season, the pollen from the previous year could fertilize the non-biotech variety. If the number of biotech volunteers is small and the farmer purchases new seed every year, then the degree of introgression of transgenes into a non-biotech variety will be negligible, and will, therefore, not pose a significant crop management problem.

2.3d   Saving and Sharing Seeds

Although most farmers in industrialized nations purchase new seed for hybrid crops every year, many in developing countries selectively save seed from one growing season for planting during subsequent seasons. Studies have shown that these farmers may purchase seeds from commercially improved maize varieties not only for producing that variety and saving its seeds, but also to intentionally incorporate the improved germplasm into local cultivars and landraces (Bellon and Brush, 1994; Louette and Smale, 1996; Louette et al., 1997). If farmers save and share seeds derived from either the commercial variety or the hybrid (commercially improved variety x landrace) with other local farmers, then introgression of the new genetic material can occur. With the exception of soybeans in the U.S. and Canada, seed saving of non-hybrid crop seed is common in all countries, regardless of the level of industrialization. Also, some farmers in developing countries purchase grain intended as livestock feed and use it for seed.

2.3e   Impacts of Introgression

Can hybridization and introgression alone have negative impacts? It depends on the recipient plant and the trait.

In the case of wild relatives, hybridization and introgression alone do not necessarily lead to negative environmental effects. Usually, significant ecological impacts from the presence of a new gene in a wild plant population will occur only if the new trait affects the plant's fitness (discussed below). However, under a certain, but unlikely, set of conditions, hybridization and introgression alone could have negative effects on wild plant populations, even in the absence of selective pressure. Mathematical models (Ellstrand and Elam, 1993) have demonstrated that if

• hybridization and introgression of crop genes into wild relatives occur at a high rate, and
• the wild relative is a locally rare species,

significant genetic erosion⁷ of the wild plant population can occur after several generations of backcrossing the hybrid (wild plant x crop) to the crop.

In terms of the impacts of introgression of new genes into landraces, small farmers in developing countries have intentionally incorporated genetic material from commercially improved crop cultivars into the locally distinct cultivars and landraces they have developed over thousands of years of selectively saving seeds. Such intentional introgression of new genes from improved varieties reflects the openness of traditional farmers to genetically shaping cultivars to the demands of the local environment in order to better meet their needs.

In spite of these benefits, some scientists and activists have expressed concern that such introgression of genetic material from either conventionally improved varieties or biotech crops will lead to the loss of genetic diversity in the landraces and local cultivars used by farmers in developing countries (Iltis, 1974; Serratos et al., 1997). Research has indicated that this not the case. Genetic erosion of local cultivars is a complex function of the amount of acreage occupied by commercial varieties (either conventional or biotech) versus the area planted to local cultivars; the diversity within and between the commercial varieties and local cultivars; and the extent to which local varieties have been abandoned or substituted. A number of studies have demonstrated that traditional farmers that grow commercial varieties do not necessarily abandon their local varieties or landraces (Brush et al., 1992; Bellon and Brush, 1994; Bellon and Taylor, 1993). In fact, it appeared that farmers might be transforming commercial varieties into local ones rather than the reverse. As long as the function of the new genetic material is complementary to that of the local germplasm, genetic diversity is likely to increase, not decrease.

2.4    Effects on Fitness of the Hybrid

The potential impact of gene flow does not depend solely on cross-pollination, successful hybridization and introgression. A transgene that does not offer an evolutionary advantage may not persist in a population of wild plants; or, if it does become established through introgression, the presence of the new gene/trait in the wild plant population may not lead to significant ecological or agricultural impacts. For example, a marker gene is unlikely to have a significant impact on either a wild plant population or crop management practices, even if the rates of gene flow were high enough for introgression to occur. On the other hand, there is evidence that gene flow from some crops (non-biotech varieties) to a weedy relative has had an impact on crop management practices in certain areas. For example, weed species that are very closely related to pearl millet (Pennisetum americanum) and sorghum (Sorghum bicolor) have become more difficult for farmers in some regions to control (Ellstrand and Hoffman, 1990).

To alter ecological relationships in wild plant populations, transgenes must either increase or decrease the fitness of the wild plant by enhancing or reducing its reproductive success. For example, if a population of wild plants were controlled by a lepidopteran insect, then acquisition of the Bt gene would give the (crop x wild plant) hybrids a selective advantage over related plants without the Bt gene. On the other hand, additional genes transferred from the crop to the wild plant, such as genes that limit seed dispersal, would be disadvantageous to the hybrids. This disadvantage could offset any advantage provided by the Bt gene. If the disadvantage were great enough, acquisition of the trait that limits seed dispersal could lead to decreased fitness (outbreeding depression).

Just as the probability of gene flow varies with the crop, environment and recipient plant, the potential impacts of gene flow vary not only with the nature of the transgene, but also with the recipient plant. Some of the ecological factors that determine the potential impact of gene flow are the recipient plant's potential for aggressive weediness, or, if the recipient is a crop, its invasiveness in agricultural systems; the initial population size of the recipient plant; and the factors that naturally limit or can be used to control an increase in the number of the recipient plant hybrids.

An often-discussed negative impact of gene flow from improved varieties to either wild plant populations or non-biotech crops is the potential to convert crops into "supercrops" or wild plants into "superweeds" when they acquire the improved germplasm.

Concern about the creation of supercrops is restricted to the handful of crops exhibiting weed-like traits that allow them to become invasive in agricultural systems or persistent in natural environments. Examples of such crops include canola, rice, alfalfa and sorghum. A ten-year study of four biotech crops - oilseed rape, sugar beets, potatoes and maize - demonstrated they are no more likely than their non-biotech counterparts to persist or invade habitats (Crawley et al., 2001). Researchers planted non-biotech and biotech varieties next to each other and within four years all plots of maize, beet and rape had died out. Only one plot of potatoes lasted the full decade and all survivors in that plot were non-biotech.

The concept of converting a wild plant into a superweed and a crop into a supercrop is most often discussed for herbicide resistant crops. Some believe acquisition of herbicide resistance transgenes will improve fitness of wild plants or make non-biotech crops more difficult to control. In a few instances, herbicide tolerance traits in crops either derived through biotechnology or through conventional breeding, have moved to other varieties and made control of volunteers more difficult, but manageable, in some crop rotation systems. Experience with conventionally bred crops shows that if crops with invasive tendencies acquire resistance to certain herbicides either by cross-pollination or the natural forces of evolution, these crops can be controlled with other herbicides on the market.

On the other hand, if a wild relative acquires herbicide resistance through gene flow, it may never be subjected to the selective pressure that would provide an adaptive advantage, i.e., the application of that herbicide. Therefore, the concept of increased fitness would not apply. As described above, acquisition of a herbicide resistance gene through gene flow makes a plant resistant to that herbicide, but it remains susceptible to many others.

Return to Table of Contents

Even though naturally occurring crop/wild relative hybrids are easily found for some crops, determining whether crop genes have introgressed into wild relatives can be quite difficult. As Doebley (1990) pointed out, if a crop and its wild relative in a particular geographical region both possess a particular morphological trait, it may seem reasonable to conclude that the trait was transferred by cross-pollination from the crop to the wild relative, or vice versa. Another explanation, however, might be that these populations, growing under the same environmental constraints, may simply share the trait as a result of either convergent evolution or joint inheritance of the trait from their common ancestor. The latter possibility is especially relevant because crops and their progenitors are separated by relatively short evolutionary periods of 10,000 years or less and are particularly likely to possess similar genotypes.

Ellstrand et al. (1999) believed that the most appropriate way to test whether hybridization of a crop and wild relative is likely to occur under field conditions is to create experimental stands of the crop associated with the wild relative under conditions that simulate those which occur in an agronomic setting. They suggested that gene flow from crops to wild relatives could then be measured by testing the progeny of the wild plant for crop-specific genetically based markers, thus answering the question: Does hybridization occur under field conditions? Additional questions to be answered to assess impacts include:

• Will the hybrids reproduce in the wild?
• Will the transferred genes enhance weediness of the wild plant?

The economic and environmental impact of that enhanced weediness, if it occurs, will depend on the degree to which the new trait increases fitness and the environmental conditions.

Return to Table of Contents

4.1   Background - Maize Pollination System

Maize is a cross-pollinated, wind-pollinated crop with pollen dispersal driven by a variety of physical factors, such as wind speed, wind direction, wind turbulence, air density and air velocity (Di-Giovanni and Kevan, 1991; Di-Giovanni et al., 1995; Jones and Brooks, 1950). Pollen size and density also affect the distance over which corn pollen disperses. Maize pollen is much larger than pollen from most wind-pollinated plants, such as ragweed pollen (100 microns vs. 20 microns in diameter). Due to its large size, maize pollen settles at a rate that is approximately 10 times faster than pollen from other wind-pollinated plants (Di-Giovanni et al., 1995). Raynor et al. (1972) showed maize pollen is not transported as far by the wind as smaller pollen grain; does not disperse as widely horizontally or vertically; and settles to earth more quickly, much of it within the source itself. At 60 meters from the source, in the downwind direction, maize pollen concentrations average about 1% of those at 1 meter. Each hybrid maize plant is capable of producing over 10 million pollen grains.

It is important to emphasize that off-site pollen movement does not necessarily result in gene flow. To pollinate another maize plant, viable pollen must arrive when silks are receptive and be the first to travel down a particular silk. When pollen within a field is plentiful, this latter requirement is difficult to achieve.

4.2   Assessing the Potential for Gene Flow from Maize

4.2a   The Likelihood of Gene Flow through Cross-pollination

Early field experiments by Bateman (1947a, 1947b) showed that wind speed, direction, and flowering date had large effects on the degree of cross-pollination in maize. He also confirmed that there is a rapid fall in pollination with increasing distance. In addition to the fieldwork, he developed mathematical formulas to depict what happens in the field with a variety of crops. He reached the conclusion that the rate of cross-pollination decreases per unit increase of isolation distance. Many other studies later confirmed that distance between pollen source plants and pollen receptor plants is an important factor that influences cross-pollination.

Three recent studies examined the effect of isolation distance vs. rate of cross-pollination. One was a literature review that focused on research in non-commercial settings, while the other two were relevant for commercial settings. In the literature review, Ingram (2000) came to the conclusion that separation distances of biotech maize to non-biotech maize of 200 meters (660 ft.) and 300 meters (990 ft.) are recommended to maintain cross-pollination for grain production below 1% and 0.5%, respectively. Again, most of those studies reviewed by Ingram have not been carried out on a commercial scale.

A recent study commissioned by the British Department for Environment, Food and Rural Affairs (DEFRA; Henry et al., 2003; Report EPG 1/5/138) was performed in a setting that represented the potential for gene flow under realistic farming practices. A total of 55 maize farm scale evaluation sites were used and samples from conventional maize fields were collected from various distances away from the junction with biotech maize. The results indicate that cross-pollination can be predicted to be less than 0.9% and 0.3% at a separation distance of 25 meters (80 ft.) and 80 meters (265 ft.), respectively.

Another recent report from Spain, where maize developed through biotechnology (Bt maize) and conventional maize have been grown for several years, demonstrates that coexistence is possible if certain conditions are applied. The study was conducted by the Spanish Institute for Agriculture & Food Research and Technology (IRTA). The findings are based on the results of field trials where Bt maize was grown in a 0.25 hectares patch within a 7.5-hectare conventional maize field. Conditions were chosen that guaranteed a maximum level of cross-pollination. The average presence of the Bt gene in conventional maize samples was 0.9% at distances of 2 to 10 meters (7 to 33 feet) from the Bt maize. The average rate of cross-pollination throughout the conventional maize field was less than 0.2%.

Another important factor that determines the probability of cross-pollination is pollen viability, which is affected by temperature and humidity. Maize pollen has a high water concentration and is generally considered desiccation intolerant compared to other pollen because it loses water rapidly and its viability decreases sharply as it dehydrates (Hoekstra, 1986; Buitinik et al., 1996; Luna et al., 2001). High temperature also decreases pollen viability (Roy et al., 1995). ). Baltazar and Schoper (2002a) demonstrated that as temperature increases and relative humidity decreases pollen grains have to face more arid atmospheric conditions, which decreases the likelihood of cross-pollination.

Field experiments to measure pollen viability, conducted in two growing seasons in Nayarit, Mexico, demonstrated 100% loss of viability after 2 hours of atmospheric exposure, irrespective of the temperature and relative humidity. On average, 80% of the pollen was not viable after one hour, with a range of 96% under conditions of high temperature and low relative humidity compared to 58% under the more favorable conditions of high relative humidity and lower temperatures (Luna et al., 2001).

The limited viability of pollen under high temperature and low humidity condition, may explain the low rate of cross-pollination measured by Luna et al. (2001). Small plots of maize (12.8m2; 4 rows, 4m long) were grown at various distances (100, 150, 200, 300 and 400m) from the north, south, east and west margins of a large (4000m2) planting of a maize variety that served as pollen source and contained genetic markers for either a leaf or seed trait. To ensure maximum yields were obtained (and therefore maximum opportunities for gene flow had occurred), maize plants in the twenty small plots and single, large pollen source were grown under optimal cultural practices, and researchers selected varieties that exhibited synchronicity in pollen shedding and silk emergence. They found that the maximum distance over which any cross-pollination occurred was 200m, and this was a single kernel event. A single kernel pollination was also observed at both 150m and 100m. Therefore, a total of 3 pollen grains from the 4000m2 pollen source fertilized a single egg of one maize plant growing at 100, 150 and 200m (Luna et.al., 2001).

Castillo and Goodman (1997) reviewed two experiments conducted in Mexico that examined gene flow between adjacent plots with simultaneous flowering. In both experiments, a maize variety expressing a marker trait serving as a pollinator was surrounded with another variety. The degree of gene flow was determined from plants in border rows and from those that were 12-15 meters away from the pollinator marker plants. Whereas 8 to 60% of the kernels from border row plants were cross-pollinated, the rate dropped dramatically to 0.16 to 3% cross-pollinated kernels from plants 12 to 15 m away.

4.2b   Assessing Gene Flow between Maize and Its Wild Relatives

As a result of the geographic distribution of teosinte (Zea spp.), the wild relative of maize, gene flow between the different subspecies and hybrid maize and improved open pollinated varieties is only possible in Mexico and Central America. Doebley (1990) determined that all teosintes (except the tetraploid, Z. perennis) can produce fertile hybrids when crossed with maize. Doebley also determined that it is common to observe hybrids and backcrossed individuals between maize and teosinte under field conditions, which suggests free and constant gene exchange between them (Blancas, 2002; Sanchez and Corral, 1997; Wilkes, 1967, 1972). However, it is not clear yet if reciprocal introgression exists. Using molecular evidence, such as differences and similarities in allozymes and chloroplast DNA, Doebley et al. (1987) could not demonstrate significant introgression of maize genes into the teosintes. Contrary to what would be expected had significant introgression occurred, teosintes growing near maize exhibited no more similarity to maize than the teosintes that did not grow near it.

These findings suggest that gene flow occurs from teosinte to maize. Estimates derived from population structure analysis suggest that the genome of Mexican landraces that developed in proximity to the teosinte ssp. mexicana, are composed of 0.2-12% ssp. mexicana germplasm (Pritchard et al., 2000).

Experimental pollinations showed that teosinte pollen applied to maize silks readily formed kernels on maize ears, but that crosses in the reciprocal direction were often unsuccessful (Baltazar and Schoper, 2002a; Baltazar et al., 2003). Kermicle and Allen (1990), Evans and Kermicle (2001), have investigated incompatibility between teosinte and maize. Cross-compatibility factors know as gametophyte factors (ga) are numerous in maize and teosinte. However, a system unique to teosinte, Tcb1 (Teosinte crossing barrier1), was described by Evans and Kermicle in 2001. They hypothesized that Tcb1 is responsible for recognition between pollen and silks, and could play a significant role in isolating teosinte from maize reproductively.

Investigation of the biology of pistillate and staminate inflorescences of maize and teosinte revealed that teosinte pollen grains were smaller (76 microns) than maize pollen (103 microns) (Baltazar and Schoper, 2001ab). Teosinte pollen also was more disposed to desiccation under dry atmospheric conditions. Maize pollen remained viable between one and two hours as previously described (Luna et al., 2001), but teosinte pollen grains started loosing turgidity after 15 minutes and died after one hour. Despite the short pollen viability of teosinte pollen grains, teosinte pollen grains were able to fertilize maize ovules as demonstrated by the formation of kernels at harvest when pollinations were made in less than 15 minutes after pollen was released from the plants. The resulting kernels were shriveled and smaller compared to kernels formed on ears pollinated with maize pollen.

4.3   Assessing the Consequences of Gene Flow from Maize

4.3a   Impacts of Gene Flow from Commercially Improved Maize to Local Cultivars and Landraces

Landraces of maize have been the focus of several studies in Mexico (Benz, 1988; Sanchez and Ordaz, 1987; Wellhausen et al., 1952). An early survey by Wellhausen and co-workers (1952) reported 25 races and three sub-races. Currently, more than 50 races can be differentiated (Sanchez et al., 2000), based on multivariate analysis of morphological, cytological, isozymes and DNA analysis. The largest differentiation on the number of landraces from the 1950s to present may be attributed to more sensitive techniques, increased efforts in collecting, or to the creation of new local varieties as farmers continue to select for specific environmental conditions and food uses.

Research on maize varieties in Mexican agricultural ecosystems have indicated that seed exchange among farmers in small communities and pollen flow are equally important to facilitate gene flow (Louette, 1997; Castillo and Goodman, 1997; Pressoir and Berthaud, 2004). Current evidence suggests that many small-scale subsistence-oriented farmers have taken improved varieties and planted them alongside their local varieties (Bellon and Brush, 1994).

Also important is a farmer's willingness to deliberately promote gene flow among improved open pollinated varieties and hybrids into their local landraces (Aguirre-Gomez et al., 2000) to transfer valuable genes into their local varieties. Hybridization and subsequent introgression between local and improved maize through empirical plant breeding practices is highly valued by farmers in Mexico and Central America (Almekinders et al., 1994; Bellon and Risopoulos, 2001). At present, it is widely accepted that gene exchange among improved open pollinated varieties, hybrids and land races has been common practice in Mexico and Central America (Castillo and Goodman, 1997; Louette, 1997). Yield advantages of improved landraces from the same region compared to old landraces suggests that introgression of genes through farmers intervention has been beneficial to Mexican landraces (Castillo and Goodman, 1997).

Concerned with maintaining genetic diversity in local cultivars and landraces of maize, Iltis (1974) proposed the creation of in situ conservation to complement the ex situ conservation of maize germplasm systematically practiced since the 1940's-50's. It is estimated that 80% of the teosinte genetic diversity in Mexico has been collected (Serratos et al., 1997) and preserved ex situ. In situ conservation refers to preserving entire ecosystems, within certain agricultural areas, by prohibiting changes in cultural practices or the introduction of foreign genetic material. This concept has been criticized widely on the grounds that it is not feasible for socio-economic reasons. In their studies of maize varieties and seed sources in the traditional farming community of Cuzalpa (Mexico), a reserve established in part for in situ conservation of maize landraces, Louette and Smale (1996) found that farmers enhance the genetic diversity of local cultivars with introductions of both improved cultivars and landraces from farmers in other areas. Genetic material from introduced varieties was incorporated into local cultivars to improve yields and, as a result, was a source of increased diversity and not genetic erosion. Traditional agricultural systems in this area are not closed and isolated with respect to the flow of new genetic material but are open and evolving. Therefore, in situ preservation and collection of landraces represents only a snapshot of the diversity present at that moment.

One needs to differentiate the intentional introduction of genes from "improved" maize varieties/hybrids into local landraces/varieties from the unintentional introduction of genes via pollen flow from "improved" varieties/hybrids. No matter what the route, however, the consequences are likely to be similar if the same agronomic forces dictated by the local farmers are consistently applied.

4.3b   Impacts of Gene Flow from Biotech Maize Varieties to Local Cultivars and Landraces

In late 2001, a paper was published in the journal Nature that claimed to present evidence for introgression of biotech DNA into Mexican landraces (Quist and Chapela, 2001), despite the fact that Mexico had banned the cultivation of biotech varieties. While some considered the finding a threat to maize genetic diversity, the scientific evidence presented in the paper was widely criticized by scientists. Nature editors later retracted the paper.

Despite the controversy over the paper's technical merits, its conclusion that DNA from plants derived through biotechnology was detected in Mexican maize was confirmed in early 2004 by the Mexican Biosecurity Commission Cibiogem. Follow up studies commissioned by Cibiogem revealed that the DNA was detectable in seven percent of 4000 samples from 188 localities in the states of Oaxaca and Puebla. Preliminary analysis suggests that some indigenous farmers planted imported grain that contained biotech corn, together with local cultivars. The result was not surprising as it is generally thought that introgression of transgenes from commercial hybrids into landraces is possible if the plants are grown in proximity to one another. The most important issue is, what impact introgression of transgenes has on biodiversity. In the position of the Mexico-based International Maize and Wheat Improvement Center (CIMMYT) on the issue of transgenes in Mexican landraces, Director General Iwanaga said: "When transgenes are present in Mexican maize landraces grown by farmers, does this mean that an important resource is lost forever? As scientists, we would answer 'no,' because the landraces may have changed, as they do all the time, but they have not disappeared. On the contrary, with the addition of a transgene, they could actually be considered more diverse. This additional diversity may not be desirable, however. It is precisely this issue that the Mexican government must resolve.

4.3c   Impacts of Gene Flow between Biotech Maize Varieties and Wild Relatives

Most teosintes are sexually compatible with maize (Doebley, 1990) and are able to form hybrids. As described earlier in the document, experimental data suggest that cross-pollination events are more likely to happen when pollen flows from teosinte to maize rather than vice versa. If the maize plants were derived through biotechnology, the resulting hybrids would carry transgenes. Doebley (1990) identified several conditions that would facilitate introgression of transgenes in teosinte. First, introgression of transgenes into teosinte populations is possibly only when biotech maize is grown in an area where the wild relative is native (Mexico/Central America.)

Second, genes from maize cannot interfere with traits, such as seed dispersal, that allow teosinte to reproduce. Humans have bred maize plants to retain seeds together on single structure. The lack of seed dispersal differs from wild maize (teosinte), in which each seed is designed to separate from the others. This is meant to maximize reproductive success by ensuring most seeds have enough space, light, nutrients, water, and other resources to develop. Whenever cultivated maize is crossed with teosinte, the seeds do not disperse from the hybrid. Wilkes (1972) states that although hybrids of maize and teosinte have been found among all known teosinte populations in Mexico and Guatemala, the gene flow has not altered the strong selection pressure for a hard enclosed teosinte fruitcase. This significantly reduces the possibility that the hybrids will produce progeny that will survive to reproduction. Therefore, maintenance of genetic material from cultivated maize in the teosinte population is difficult, whether the cultivated maize is a biotech or non-biotech variety.

Third, as described above, the transgene must confer an adaptive advantage to the hybrid that increases its fitness beyond those teosinte plants lacking the transgene. In the case of Bt, for example, wild teosinte populations would not benefit by having a Bt gene unless lepidopteran pests have a significant impact on the size of the teosinte populations in that area. Given the lack of adaptive advantage conveyed by the Bt transgene in conjunction with the adaptive disadvantage of significantly decreased seed dispersal, the Bt trait would be lost in just a few generations. Phenotypic traits provided by the transgene must result in a plant's enhanced ability to survive in order for the transgene to be actively selected.

Ellstrand et al. (1999) reviewed the literature and presented evidence for natural hybridization and introgression for a number of important crops. For maize, they cited the following research papers.

1. Work by Wilkes (1977) describing plants morphologically intermediate between maize and teosintes that were observed in and near Mexican maize fields when teosinte is abundant.
2. Doebley's (1990) allozyme studies that demonstrated very low levels of introgression from maize into the seven teosinte species or subspecies they studied.
3. Kato's work (1997) indicating that cytogenetic analyses "…offer no evidence of …maize-teosinte introgression in either direction…"

According to Ellstrand et al. (1999), no one has conducted genetic analyses of wild intermediates using currently available molecular techniques to determine whether the intermediates are true hybrids, introgressants, or crop mimics. Determining this would help address Ellstrand's (1997) concern of the possibility of extinction by hybridization. To him, the problem of extinction by hybridization depends upon whether hybridization occurs at a substantial rate, and whether the wild relative is locally rare compared to the nearby crop. The "nearby crop" could be a landrace, a commercially improved variety developed by conventional breeding, or a biotech variety. The potential impact on the wild relative will be the same.

The key issue here is that both outbreeding depression (resulting from detrimental gene flow that reduces the fitness of a local rare species) and swamping (which occurs when a local rare species loses its genetic integrity) are frequency-dependent phenomena and show positive feedback. Ellstrand and Elam (1993) showed that with each succeeding generation of hybridization and backcrossing, genetically pure individuals of the locally rare species became increasingly rare until extinction occurred. An unpublished example from Ellstrand showed that if 900 individuals of a locally common species mate randomly with 100 individuals of the locally rare one, extinction by outbreeding depression and/or swamping could occur in two generations. His example also showed that if the rare species mated with its own kind even five times more frequently than the rate expected under random mating, the time to extinction is expected to double relative to random mating. Ellstrand goes on to point out that in contrast to the issue of enhanced weediness, the problem of extinction by hybridization does not depend upon relative fitness, but only on the patterns of mating.

According to Serratos et al. (1997), the principal threats to teosinte populations are changes in land use through increased grazing and urbanization. They view biotech maize as a marginal threat compared with the effects of urban growth. The impact of biotech maize, or any improved maize, can be controlled through known practices of isolation. Finally, some have questioned whether transgenes from maize would enhance teosinte weediness. According to Ellstrand et al. (1999), to date, gene flow from non-biotech varieties to teosinte has not increased its weediness.

4.3d   Impact of Gene Flow via Pollination on Neighboring, Non-biotech Maize

The seed industry follows internationally determined and accepted standards for genetic purity. For example, these standards call for the use of specific distances between fields to manage gene flow. These distances are necessary to meet the standards for purity.

The seed industry is applying the lessons learned from working with these standards, as well as our own experiences, to understand and manage gene flow issues related to crops derived through biotechnology. In addition to using isolation distances before marketing these crops, we surround the field with a number of border rows of non-biotech plants that serve as pollen-catchers. This further reduces the chances for cross-pollination. In spite of the many measures (discussed below) used during seed production to minimize the presence of biotech genetic material in non-biotech seed batches and vice versa, it is recognized by many that trace amounts of biotech genetic material or seeds may be found in some fields containing non-biotech varieties.

Because maize is a cross-pollinated crop with wind-dispersed pollen, cross-pollination can occur unless steps are taken to avoid it. If non-biotech maize plants in one field have receptive silks at the time of biotech pollen shed in a neighboring field, some cross-pollination of the non-biotech hybrid with the biotech hybrid will probably occur. The amount of this cross-pollination depends upon many factors, including the distance of the fields from each other, presence of non-biotech crop borders, intensity and duration of pollen shed of the non-biotech maize (pollen competition), and the predominant wind speed and direction.

U.S. state seed certifying agencies are responsible for official standards for certified seed (Table 1). The standards, which vary somewhat by state, generally set minimum distances for isolation. These distances are modified by additional border rows, size of the field and production block, adequate natural barriers (in some states), and differential flowering dates (in some states).

 Table 1. General isolation distances set by U.S. state seed certifying agencies for single-cross
 seed production. Requirements designed to target genetic purity of 98-99 percent.

When no border or one-border rows are used, minimum distances from 410-660 feet are typically required between the female parent of the hybrid being produced and any other corn of the same seed color, maturity or endosperm type. If the corn that might serve as a potential adventitious pollen source has different kernel color or endosperm type, isolation distance of 660 feet is required. Recognizing the inevitability of adventitious presence of different genetic material, the Organization for Economic Co-operation and Development (OECD) suggested purity levels for parent seed at 99.0-99.5%. For hybrid maize seed, the U.S. Federal Seed Act (USDA) requires 95 percent purity in order to label it as a single hybrid. These widely accepted standards for certification of varietal purity could also be useful when trying to establish thresholds for purity of non-biotech maize.

It is important to realize, however, the situation for seed production, which is what seed companies do, and grain production, which is what producers do, are different. These same isolation standards that apply to seed production fields may result in much higher purity in hybrid grain production fields. A hybrid grain field produces a pollen load that is many times, perhaps 10 to 100 times, greater than that of a single-cross seed production field. In addition, the timing of pollen shed in a hybrid field is usually highly synchronized with silk emergence. The large pollen load and synchronous timing of pollen and silks serve to greatly limit pollen mixing from outside sources.

Growers should also keep in mind that most biotech hybrids, including those containing the Bt trait, have heterozygous (or hemizygous) dominant gene systems. Only half of the pollen produced in a Bt commercial field contains the Bt gene because only one parent contains the transgene. This reduces the chance of co-mingling due to pollen drift from the biotech field to one half of what would otherwise be expected if all of the pollen possessed the gene.

The use of border rows is another technique widely used to minimize the probability of biotech pollen drift into a non-biotech field. Table 2 provides the minimum number of border rows that can be substituted for distance in the isolation of a non-biotech field from a biotech field for grain production.

Finally, in addition to distance and/or border row isolation, segregation of grain at harvest also can improve purity. If non-biotech and biotech varieties are grown in plots next to each other, harvesting a number of rows (12-16) from the side of the non-biotech plot that is nearest the biotech field, and then separating this grain from the rest of the non-biotech grain can help the grower achieve a much higher purity level in the grain harvested from the remainder of the non-biotech field.

It is obvious that there is an advantage to larger production fields and to greater isolation distances. It also should be clear, however, that if the grower has a market or use for the grain produced in the border rows, removing 12-16 outside rows should be adequate to insure high purity grain. Pollen mixing has always been a normal occurrence in corn production. The accepted practice has been that each grower is responsible for any isolation that may be required for various types of crops.

 Table 2. Minimum number of border rows recommended for grain production in Nebraska, Iowa,
 Indiana, and Wisconsin to maintain 1% or less non-biotech purity in a hybrid field at various
 distances (in feet) from a biotech field. (Information prepared by Dr. Joe Burris, Emeritus
 Professor Iowa State University under commission from the American Seed Trade Association.)

Return to Table of Contents

5.1   Background - Pollination in Soybeans

Soybean (Glycine max) is a self-fertile and self-pollinated crop. Pollinations typically occur before the flowers open. This phenomenon essentially prevents cross-pollination (Poehlman, 1987). In cool weather, flowers are more likely to stay closed. Because of the unique reproductive biology of soybeans, the primary cause and route of multiple genotypes in different soybean varieties in agricultural settings results from physical mixing of different seed types during harvesting, post-harvest handling and subsequent planting. However, cross-pollination in cultivated soybeans can occur (Ahrent and Caviness, 1994; Caviness, 1966; Garber and Odland, 1926; Kiang et al., 1992).

5.2   Assessing the Potential for Gene Flow from Soybeans

5.2a   The Likelihood of Gene Flow through Cross-pollination of Soybean Varieties

As described above, the likelihood of gene flow via cross-pollination in cultivated soybeans is very low because of the propensity for self-pollination. This is confirmed by a number of studies in different locations. Garber and Odland (1926) presented the results of several studies across two years in Wisconsin that showed natural cross-pollination to range from 0.14% to 0.36% from one year to the next. Caviness (1966) observed cross-pollination rates ranging from 0.03% to 0.44% in a three-year study in Arkansas. With distances of more than 15-feet from the pollen source, however, natural cross-pollination was rare and did not vary greatly. Caviness observed one natural cross at a distance of 33-feet and one at 45-feet across the three years of his experiments. Within 15 feet, however, seasonal differences in many environmental variables, such as temperature, moisture and cloudiness, caused wide fluctuations in the amount of natural cross-pollination.

Ahrent and Caviness (1994) looked at a broader spectrum of germplasm to determine if genotypes differed significantly for the rate of cross-pollination and if they should be managed differently to produce genetically pure seed in small plots bordered by other genotypes. They used 12 cultivars differing in maturity across two growing seasons to determine the extent of cross-pollination under field conditions where both honeybees and indigenous insect populations were present. In general, they found a cross-pollination rate about 1% higher than previously reported, with a range of cross-pollination from 0.17 to 2.55%. This probably resulted from the following elements of their experimental design.

• A beehive was placed at the edge of the experimental area to ensure there was an adequate number of pollinators throughout the flowering period, and
• Pollen donor plants and plants with the recessive trait were grown in adjacent rows, favoring cross-pollination by small insects present in the field.

Cultivars differed significantly in rates of cross-pollination, and therefore, isolation distances vary with genotypes. This study also showed significant differences in cross-pollination rates for some environments.

The highest rate of cross-pollination was recently reported from Ray et al. (2003) when two cultivars, a white flowered and a purple flowered, were alternately grown 15.2 cm apart within a row. The dominance of purple flower color over the white color was used to identify cross-pollination and the rates ranged from 0.65 to 6.32%, with an average of 1.8%. Significantly lower was the rate of cross-pollination, when the white flower color cultivar was grown at a distance of 5-6 meter from the purple color cultivar (0.03 to 0.05%).

Due to the reproductive characteristics of soybean, most cross-pollination is likely facilitated by insects (Erickson et al., 1978; Rust et al., 1980). Caviness (1966, 1970) showed that honeybees are responsible for the occasional cross-pollination, but that thrips are ineffective pollination vectors.

The unlikelihood of cross-pollination in soybeans is recognized by Certified Seed Regulations. For Foundation seed, the most stringently regulated category, soybeans can be grown zero distance from the nearest contamination source, as long as the distance is adequate to prevent mechanical mixing.

5.2b   Assessing Gene Flow from Cultivated Soybeans to Wild Soy or among Wild Soy Populations

The center of origin for soybeans is northeastern China. Populations of wild soybeans (Glycine soja and G. gracilis) exist in only five regions of the world: central/northeast China, Taiwan, certain regions of Siberia in Russia, Korea, Japan, and Australia (Hermann, 1962). There are no soybean wild relatives in the United States (Rissler and Mellon, 1993).

Natural cross-pollination of cultivated G. max to wild G. soja or vice versa should be very infrequent (less than 1%), even with insect vectors present. It is difficult to hybridize these species under controlled conditions and should be even more difficult under natural field conditions. We could find no written documentation of gene flow through natural cross-pollination from commercially improved varieties of G. max to wild soy populations. Kwon (1972) has documented cross-pollination between wild soybeans and Korean landraces of soya. Ellstrand et al. (1999) cited personal communication from Dr. Reid Palmer that wild plants morphologically intermediate to G. max and G. soja occur spontaneously near Chinese and Korean soybean fields when G. soja is present. However, there is no information available that documents natural cross-pollination between domesticated (either landraces of commercially improved varieties) and wild soy in China.

A stabilized hybrid taxon from northeastern China, G. gracilis, that is intermediate to, and interfertile with, both G. max and G. soja apparently exists. Population-level analyses have not yet been conducted to determine the extent, if any, to which crop alleles have moved into natural populations. It is not yet clear whether G. gracilis represents a stabilized hybrid-derived taxon or a subset of the variation found within hybrid swarms between G. max and G. soja. Its position in the taxonomy of Glycine remains controversial (Hymowitz and Singh, 1987).

Very little information exists on the rate of cross-pollination among wild soybeans. Kiang et al. (1992) sampled four natural populations growing along the Kitakami River in Iwate Prefecture, Japan. They estimated the rate of cross-pollination at 2.3%, based on the observed number of heterozygous seeds in these populations, which is consistent with the cross-pollination rates among cultivated G. max cited above.

5.3   Assessing the Consequences of Gene Flow from Soybeans

5.3a   Potential Ecological and Evolutionary Impacts on Wild Relatives of Gene Flow from Biotech Varieties

Controlled hybridization of G. max to wild Glycine species has been extremely difficult, so the impact of gene flow from cultivated soy to wild soy populations or from commercially improved, non-biotech or biotech varieties to soy landraces should be essentially non-existent.

Because of the low frequency of hybridization between cultivated and wild soy and the relatively small area of overlap between wild and cultivated soy, the amount of transgene introgression into wild soy populations is expected to be minimal. When biotech soybeans are grown in areas containing wild soy populations, fairly simple techniques could be used to further minimize the already limited opportunity for gene transfer. A minimal increase in either isolation distances or border row numbers will reduce the possibility of cross-pollination, resulting gene flow and possible introgression to virtually zero.

As discussed in the introduction to the paper and also in the section on maize, many factors determine whether gene flow from cultivated crops to wild populations will result in an ecological or evolutionary change. In soybeans, the rate of gene flow is greatly minimized, first and foremost, by the reproductive biology of the soybean plant. In addition, ecological impact depends on the nature of the recipient plant and the adaptive value of the newly acquired trait. To date, the only transgene in soybean that affects yield (i.e., potential reproductive success) is resistance to the herbicide glyphosate. Because of the very low frequency of hybridization between wild and cultivated soy, in order for the glyphosate resistance transgene to become established in the wild soy population and have an ecological impact, there must be intense selective pressure placed on the wild soy population that would favor plants that had acquired the gene for glyphosate resistance. In other words, the wild soy population must be frequently treated with the herbicide glyphosate. Having a well-planned herbicide application program in place before growing biotech crops in centers of origin would be an expected practice.

5.3b   Impact of Gene Flow via Pollination on Neighboring, Non-biotech Soybeans

Cross-pollination of soybeans is generally infrequent (<1%) although some have documented higher rates under optimal conditions (Ahrent and Caviness, 1994; Ray et al., 2003). Most likely, cross-pollination will not occur if biotech and non-biotech fields are 10 m apart. The OECD purity standard of 99 99.5% should be easily met at this distance, unless farmers inadvertently mix seed during planting, harvesting or in shipping and handling. The Association of Official Seed Certifying Agencies (AOSCA), which sets seed purity standards for virtually all crops, does not have specific guidelines for isolation distances for certified soybean production. Adventitious presence of genetic material is minimized during certified seed production by following their only guideline for isolation: "Fields of soybeans shall be separated from any other variety or uncertified seed of the same variety by a distance adequate to prevent mechanical mixture" (AOSCA, 1996). Adventitious presence can result from growers or subsequent handlers accidentally mixing seed by not completely cleaning harvest machinery, storage bins or seed handling systems.

Return to Table of Contents

6.1   Background - Pollination in Oilseed Brassica

Three Brassica species are important oilseed crops globally - Brassica napus (rapeseed, oilseed rape), B. rapa (turnip rape, sarson), and B. juncea (Indian mustard, oriental mustard). The first two are referred to collectively as rapeseed, while B. juncea is considered a mustard. The term, "canola," describes grain from any of the above species that has less than 2% erucic acid (C22:1) in seed oil, and less than 30 micromoles per gram of aliphatic glucosinolates in the oil-free meal (Canola Council of Canada website - www.canola-council.org ). Grain from any of these species that does not meet canola quality standards is referred to as "rapeseed," or in the case of B. juncea, as "mustard grain."

The oilseed Brassica species can be hybridized sexually to varying degrees. B. juncea and B. napus can self fertilize, but B. rapa is self-incompatible. As Figure 1 shows, the three diploid species possess three basic genomes, A, B, and C. These genomes are combined in the three tetraploid species in pair-wise combinations, i.e. AB, AC, BC. The A genome is common to the three major oilseed Brassica species, explaining the success of interspecific crossing, and the ability to transfer genes among these species.

Pollination in the three Brassica crops is primarily by insects, although pollen can also be wind-borne.

Figure 1: Relationships among cultivated Brassica species in U's Triangle (Mizushima, 1972)

6.2   Assessing the Potential for Gene Flow from Brassica

6.2a   The Likelihood of Gene Flow through Cross-pollination

The three oilseed Brassica species are cultivated species within the family, Brassicaceae. In the crops that are self-compatible (B. juncea and B. napus), on average, 80% of the seeds arise from self -fertilization (Rakow and Woods, 1987). B. rapa is self-incompatible and, therefore, the potential for gene flow between biotech and non-biotech varieties is greater for this crop.

The vast majority of the pollen produced by B. napus does not move very far: male sterile plants 1.2m from pollen-producing plants set 26% less seed than those 0.6m from the pollen source, while those 1.8m away produced 40% fewer seeds (Downey, 1992). However, pollen from Brassica species can be spread considerable distances by wind and insects (Scheffler et al., 1995; Raybould, 1999; Rieger et al., 2002).

According to Downey (1992), given the ease with which B.rapa, B.juncea and B.napus hybridize in both the greenhouse and the field, and the possibility of pollen movement over a considerable distance, gene transfer among non-biotech varieties of these three species may well have occurred at a low frequency for a number of years.

The Brassicaceae also contains a number of major weeds, including those in the genera Sinapis, Capsella, Thlaspi, Erucastrium, Raphanus, and others. Concerns have been raised about the potential for the transfer of transgenes from the cultivated oilseed Brassica species to their weedy relatives in Europe and North America where Brassica crop species are widely grown. Some Brassica crops and their wild relatives will hybridize under artificial conditions in laboratories or highly contrived field conditions. Others will hybridize at very low rates under natural conditions (Barton and Dracup, 2000; Raybould, 1999).

Another concern that has been raised is that oilseed crop species themselves might become "weeds" in agricultural settings, usually where they are rotated with cereals or other broadleaf crops. However, extensive work by Crawley and his colleagues in the UK have clearly demonstrated that biotech oilseed rape does not survive well in the wild and is, therefore, highly unlikely to invade other habitats (Crawley et al., 2001).

6.2b   Assessing Gene Flow from Cultivated Brassica to Wild Relatives

Numerous researchers have investigated the potential for oilseed Brassica species to cross with wild relatives (for a tabular summary, see Wolfenbarger and Phifer, 2000).

Bing (1991) were not successful in attempting to hybridize wild mustard (Sinapsis arvensis) with B.rapa, B.juncea or B.napus, even though the cross-pollination was attempted under optimal laboratory conditions. The low probability of gene flow from B. napus to S. arvensis was confirmed after greenhouse studies from Moyes et al. (2002), and low frequency of hybrids were obtained only when S. arvensis was the maternal parent. If none of the three Brassica crops hybridize readily with wild mustard even under the most favorable conditions, it seems unlikely that such gene transfer would occur in the field at considerable frequencies. In support of this, Bing (1991) could not find a naturally occurring hybrid of S. arvensis with B. rapa, B. juncea or B. napus when the species were cross-cultivated in field plots over a three-year period. Several other studies could not detect gene transfer from B. napus to S. arvensis (Bing et al., 1996; Chevre et al., 1996; Lefol et al., 1996a; Moyes et al., 2002).

Bing et al. (1991) and Downey (1992) describe similar work on gene transfer between the three non-biotech Brassica crops and the wild relative, B. nigra. If B. nigra served as the pollen source, hybrid seed was obtained from 3% of the crosses with B.juncea and 0.9% of the B.napus crosses; the reciprocal crosses, in which the crops served as pollen donors, were successful in 0.5% and 0.1% crosses, respectively. However, attempts to backcross the B. juncea X B. nigra hybrids to B. nigra were not successful, nor were the attempts to self-pollinate the B. juncea X B. nigra hybrids that occurred in the field. When the hybrids were exposed to B.juncea, a few seeds were obtained. Therefore, if B. juncea X B. nigra hybrids occur in the field, they will backcross to the crop and not to the weed. The researchers observed a similar situation in the B. napus X B. nigra crosses: hybrids from the reciprocal crosses were highly sterile and no seed was set when selfed or backcrossed to B.nigra. However, a backcross to B.napus produced a few seeds.

Under co-cultivation in the field, the only interspecific Brassica hybrids Bing (1991) observed were between B.napus X B.juncea and B.napus X B.campestris.

Similar studies to determine the potential for gene transfer under field conditions have been conducted in the 1990s in Europe, Canada and Australia. The results are similar to those described previously: interspecific hybridization between Brassica crops and related weedy species will be "rare, sporadic and unpredictable" (Raybould, 1999). A review of the literature in this area is contained in the proceedings of the 10th International Rapeseed Congress, held in Canberra, Australia, in September, 1999 (Champolivier et al., 1999; Downey, 1999).

The transfer and dissemination of transgenes within and among genotypes of the same and related oilseed species have also been investigated experimentally. In Canada, where biotech herbicide resistant canola has been used commercially for eight years, and nearly 70% of the canola area was planted to these product types in 2003, large-scale studies have also been conducted.

A number of studies (Eber et al., 1994; Lefol et al., 1996ab; Chevre et al., 1996; Scheffler et al., 1995; Chevre et al., 2000) have shown that hybrids between biotech herbicide resistant B. napus and/or B. rapa and various weedy relatives could be produced at very low frequency under field conditions. For example, in a study of gene flow of transgenes for herbicide tolerance between B. napus and wild radish, Chevre et al. (2000) found 1 radish seedling with the transgene in 189,084 seedlings. From this, they calculated the 95% confidence limits for interspecific hybridization to range from 10-7 to 3.1-5. However, the reciprocal cross, in which the wild radish served as the pollen donor, resulted in a hybridization frequency with a 95% confidence range of 2.1-5 to 5.1-4 (n = 73,847 oilseed rape seedlings).

In virtually all cases in which gene flow was observed between biotech crops and wild relatives, the gene flow did not occur from the biotech crop to the wild relative. The interspecific hybrid seed was produced on the biotech parent and the weed species served as the pollen donor. In addition, the resulting F1 seed was viable, readily grew to flowering, but produced no viable pollen and set no seed from self-pollination. Backcrossing of the interspecific hybrid with the weedy parent (using the F1 as the female) produced little to no seed. In successive backcrosses, some authors reported progressively lower frequencies of seed set, and loss of the herbicide resistance trait (Downey, 1999), although successful transfer of a transgene has also been documented for Brassica (Champolivier et al., 1999).

In summary, based on evidence to date, the likelihood of successful transfer of a transgene from an oilseed Brassica to a wild relative is extremely low.

6.2c   Assessing Gene Flow among Cultivated Brassica

In the case of gene flow of a transgene within or among cultivated Brassica species, there are no barriers to transfer within B. napus, and transfer between B. napus and B. rapa occurs at relatively high frequency (Scheffler and Dale, 1994; Paul et al., 1995; Bing et al., 1996; Downey, 1999). Warwick et al. (2003) detected a relatively high frequency of gene flow (approximately 7%) from B. napus carrying a herbicide resistance trait or green fluorescent protein as markers, to B. rapa that was grown within and in the margin of the B. napus field. Even higher was the average gene flow frequency from B. napus to B. rapa, that were collected near the margin of a commercial herbicide resistant B. napus field (approximately 13.6%). Rieger et al. (2002) studied the pollen-mediated movement of herbicide resistance between commercial canola fields in Australia. Samples from 63 conventional canola fields growing near non-biotech herbicide-resistant fields were taken and analyzed for acquired resistance. The study revealed that cross-pollination between commercial fields occurs at low frequencies (to a maximum of 0.2%) but to considerable distance (to a maximum of 3km). These findings were supported by a study commissioned by the British Department for Environment, Food & Rural Affairs (Ramsey et al., 2003). Male-sterile oilseed rape plants were used as pollen recipients and placed at various distances from commercial B. rapa and B. napus fields. The study indicated the following. The amount of pollen-mediated gene flow rapidly declines over tens of meters from the pollen source;

• Gene flow from a large area of plants to a neighboring field of fully fertile plants is of the order of 0.1% (one seed in a thousand contains DNA from both crops);
• Long distance pollen-mediated gene transfer can occur, but is rare. This means that relatively small separation distances can reduce gene transfer through cross-pollination in fields of fully fertile oilseed rape to low levels (around 0.1%, or below);

Because of the lack of competing self-pollen using male-sterile plants, the authors conclude that the data over-estimate the flow of a marker gene into a normal male-fertile population at least tenfold.

The occurrence of multiple herbicide resistance transgenes in canola (B. napus) has been reported in Canada (Orsen, 2002; Beckie et al., 2001). Two of the resistance traits were acquired from biotech canola plants and one from a herbicide resistant canola variety derived through conventional breeding. Orsen (2002) recommends methods for controlling multiple herbicide resistant Brassica varieties that have minimal or no impact on biodiversity. None of this is unexpected, as both species are partially or completely cross-pollinated, and transfer of genes via cross-pollination between the two species is well established.

6.3   Assessing the Consequences of Gene Flow

6.3a   Impact of Gene Flow from Biotech Brassica to Wild Relatives

Successful transgene transfer from oilseed Brassica species to Sinapis arvensis (Lefol et al., 1996a), Hirschfeldia incana (Lefol et al., 1996b) and Raphanus raphanistrum (Chevre et al., 1999; Rieger et al., 1999) has been reported. All of these transfers involve herbicide resistance genes. It has been demonstrated that in semi-natural environments, herbicide tolerant plants may be at a disadvantage compared to individuals without that trait (Crawley et al., 1993, 2001; Snow et al., 1999). However, the situation may be different with other types of transgenes (insect/disease resistance, cell cycle genes, etc.) that might confer a selective advantage (Stewart et al., 1997). Defoliation of rapeseed caused by artificially introduced insectivore led to increased reproduction in favor of insect resistant (Bt) plants. Therefore, no blanket statements about a lack of ecological impacts can be made at this time.

6.3b   Impact of Gene Flow from Biotech Brassica to Neighboring, Non-biotech Brassicas

Transfer of a herbicide resistance transgene from a herbicide resistant crop to a neighboring, non-herbicide resistant field has implications for both the purity of the resulting non-herbicide resistant crop, and for volunteer management in successive crops in the rotation. Co-mingling of the non-herbicide resistant crop is only a purity issue if the harvested material is pedigreed seed, or if it is to be used as grain that is "certified as non-biotech." In both cases, the establishment of standard thresholds, similar to those applied to other crops on levels of transgene permitted in non-biotech crops provides a solution.

The impact of herbicide resistant canola volunteers in the rotation is a more complex issue. Volunteer canola is a common problem in successive cereal crops. However, it is easily controlled with any of a number of herbicides. Glyphosate-resistant volunteers present a potential problem in reduced or no-till systems where glyphosate is used extensively for pre-emergent weed control. However, once again, tank-mixing of glyphosate with any of several herbicides to which canola is susceptible addresses this issue. Volunteers with multiple herbicide resistance genes can also be controlled, although the available herbicide options will be fewer. In general, this is a management issue within crop production systems, rather than an environmental issue. (Warwick et al. 2003)

Return to Table of Contents

Measures to prevent gene flow from biotech crops to non-biotech crops and weedy/wild relatives are either physical isolation approaches or molecular approaches for gene containment. Physical approaches include additional border rows, size of the field and production block, adequate natural barriers and differential flowering dates. Border rows and isolation distances discussed for corn in previous sections are required from federal agencies and widely practiced.

Molecular strategies stopping for biotech gene flow have been reviewed recently (Daniell, 2002; Lu, 2003). An overview of current and future technologies is summarized below (adopted from Daniell, 2002). Most of the technologies listed below for gene containment will not be practical for all biotech crops, but some might offer a valuable addition to currently used strategies to prevent gene flow, especially for those crops with higher risk for out-crossing.

Return to Table of Contents