NEEDED: NEW PARADIGMS FOR WEED CONTROL

Twelfth Australian Weeds Conference

462

Abstract Weeds are the major pest constraint in ex-tensive monoculture or near-monoculture agriculture,as exemplified by the wheat-growing areas of Australia,where a weed such as annual ryegrass (Lolium rigidumGaud) can evolve resistance to virtually every wheat-selective and many other herbicides. The evolutionarytrends towards resistance may have been exacerbatedby the penchant for rate cutting that facilitated newtypes of multifactorial resistances, requiring new man-agement models. Herbicide-resistant wheat can onlybe helpful if the resistance is to a rarely used herbi-cide, or one not prone to resistance problems, and/orresistance is to more than one graminicide, for use inresistance delaying mixtures. The likelihood of herbi-cide resistant transgenes introgressing from wheat di-agonally into closely related grass species is quite low,and as those species are not weedy, the risk seems verylong when evaluated with unbiased decision trees. Therisk is somewhat higher for oilseed rapid introgressinggenes into related Brassica weeds, where there is a toogreat a risk. Tandem constructs of herbicide resistancegenes with “antiweediness” genes could alleviate prob-lems of introgression of resistance genes from cropsinto related species without affecting the crop, e.g. tan-dem constructs with genes that abolish secondary dor-mancy, genes that abolish reactions to shading, or genesthat prevent seed shattering. Such genes would remaintightly linked to the resistance genes.

Biotechnology can assist weed control in wheat, andnot just through the development of herbicide-resist-ant wheat. Modifications of the TAC-TIC (Transposonswith Armed Cassettes for Targeted Insect Control)paradigm for insect control using DT’s (deleterioustransposons) carrying kev genes (chemically-inducedsuicide genes) or other deleterious genes would facili-tate weed control without herbicides. Annual ryegrasswould be ideal to test this concept: multicopytransposons bearing desired traits transformed intoryegrass could be seeded in pastures. The transposonswould quickly disseminate into indigenous ryegrassdue to the obligate outcrossing of ryegrass, while notaffecting pastures, but would engender unfitness incompetition with wheat.

Molecular biology and biotechnology have much tooffer in upgrading biocontrol, but are hardly utilizedin inoculum stabilization through to genetic engineer-ing of increased or decreased host range, increasedvirulence, or for introducing failsafes.

INTRODUCTION

Weeds are and have been the major constraint to foodproduction in the world. Staying ahead of Malthus hasintensified agriculture, increasing practices such asmonoculture that are questionably sustainable. Sustain-able should be understood in the context of having theleast possible ecological instability. As can be seen inAustralia, the evolution of the multiply herbicide re-sistant biotypes of annual ryegrass is an extremeagroecological perturbation, and current herbicide usepatterns have proved to be unsustainable.

Similar problems had been happening due to similarpractices around the world. The heavy use ofmonoculture, typically with the same herbicide regime,along with high harvest index, short stature, wimpy(uncompetitive) crops has led to new weed problems.These are not annual or perennial weed problems, theseare millennial weed problems. I include amongst themEchinochloa spp. in rice and other crops, as well asthe Striga species in sub-Sahara Africa that are para-sitic on most crops, halving yields of 100M people(Berner 1995). I also include the grass weeds of wheatssuch as Lolium spp. around the world, and Phalaris inIndia, that are being mimicked by other grasses in theirability to evolve resistances to graminicides, and notnecessarily according to mode of herbicide action. Thisunsustainable instability requires that we analyze whyresistance evolved with such a vengeance, and to as-certain whether practices can be modified to delay evo-lution.

We must also analyze what biotechnology might havein store in herbicide-resistant crops and weedbiocontrol, and adequately analyze the risks these tech-nologies might bring. If the risks of herbicide resist-ance moving from crops to related species is great,failsafes should be designed and instituted instead ofthe two extremes proposed: banning biotechnology, anddenigrating or ignoring the risks. Finally, it is posited

NEEDED: NEW PARADIGMS FOR WEED CONTROL

Jonathan GresselPlant Sciences, Weizmann Institute of Science, Rehovot, IL 76100, Israel

Proverb: If you never change direction, you will end up where you are heading


463

that biotechnology can be used to reverse evolution;e.g. to force weeds such as annual ryegrass to evolveback from being a pernicious weed of wheat to being apasture grass that is not very competitive with wheatand disappears when wheat is grown. Some of thesetechnologies are unlikely to be developed by industry,yet are in the long-term public good, and thus the pub-lic sector with widespread public support will have totake a greater part in such developments.

DOSE AND RESISTANCE

The economics of the extensive cultivation of some-what marginal agroecosystems in Australia have dic-tated lowering input costs to the minimum. This hasresulted in the penchant for lowering herbicide ratesof application to the bare minimum. Nowhere else aresuch low rates so widely used, which together with thebiology of annual ryegrass, to my mind explains theproblems with that species.

When a cut rate that is just lethal to the population(minimum dose lethal=MLD) is used in the field notall the population gets that dose (Figure 1).

Figure 1. MLD (minimum lethal dose) in field condi-tions. Only higher doses provide the MLD to the wholepopulation in the field, low doses provide the MLD tomost individuals, Presumed distribution of diclofop onannual ryegrass the field at 400, 750, 1200 g ae/ha,illustrating the proportion of pests receiving each dose.Double spraying is ignored, as are untouched escapedorganisms (in ‘refuges’). Assumptions: each mutantgene dose provides protection for 50 g/ha beyond thethreshold of 200 g/ha. The cross-hatched area showsthe sensitive population from which one gene dose willbe selected at low rates. Reprinted by permission ofthe American Chemical Society; from Gressel (1995)

Unlike in the laboratory where precision equipment isused to spray synchronous weed populations, field dis-tributions are imperfect. Four-leaf plants are effectivelyunderdosed when the MLD for three leaf plants.

Spraying can be uneven, some weed seedlings are inthe spray shadow of others, weeds such as ryegrassoften have a second flush of germination when muchof the herbicide has dissipated. The underdosed indi-viduals are weakened by the amount of herbicide theyreceive. Any underdosed, weakened individual, bear-ing a mutation that bears even a small modicum ofsurvival value may survive and multiply. This muta-tion could slightly increase the rate the weed degradesthe herbicide, could increase the level of herbicide tar-get, or could slightly protect the weed against the toxicproducts generated in a sick weed. Many such resist-ance systems are known and they can be governed bymany types of genetic modification: point mutationsthat slightly modify herbicide binding or slightly in-crease the V

max of degrading enzymes; mutations in

promoters of the herbicide target enzyme or control-ling the levels of herbicide degradation; amplificationsin the herbicide target gene; and/or the genes control-ling degradation. Such minor mutations are always“positive” and are thus inherited in a semi-dominantmanner. The first individuals with the slight modicumof dominant resistance are heterozygous. They willcross with each other and the homozygous individualswith a double dose of the resistance conferring alleleswill be selected for under the selection pressure of thelow herbicide rate. Any individual bearing a furthermutation will be even more viable and will cross withothers. Thus, under low doses the mean LD

50 for the

population can slowly creep up due to this multifacto-rial mode of resistance, which was loosely termed poly-genic, as population changes are similar to those gov-erned by quantitative inheritance.

A pioneering epidemiological study was performed byIan Heap (1988) in Ron Knight’s lab in Adelaide justafter the first four cases of ryegrass resistance todiclofop were discovered and initially characterizedby brother John Heap (Heap and Knight 1982, 1986).Ian Heap collected a large number of ryegrass seedsamples from farmers who had used diclofop for dif-ferent periods of time. His data (Fig. 2c) have the typi-cal scatter expected from field-derived materials col-lected from farmers with limited records on herbicideuse history, but the pattern is clear as well as highlysignificant statistically: there is a creeping increase inthe mean level of resistance of whole populations thatis a function of the number of seasons diclofop wasused. This is seen by looking at the population slopesthat are indicative of mono-modal creep (Heap pers.comm), and are not bimodal as intimated by Prestonand Roush (1998). This is illustrated schematically aspopulation shifts in Fig. 2D.


464

Figure 2. “Sudden” appearance of major monogene resistance vs slow incremental creep of quantitatively inher-ited resistance. (A) Actual field data on resistance showing changes in weed populations in a monoculture maizetreated annually with atrazine. Amaranthus retroflexus, Echinochloa crus-galli and Digitaria sanguinalis, theforemost weeds, were counted. The maize field was treated with atrazine from 1970 onwards (B). A populationdistribution description of the same data for Amaranthus in (A), where the relative dose rates (R/S) on the hori-zontal axis are arbitrarily plotted. (C) Slow incremental increase in the dose level of resistance in repeatedlytreated Lolium populations. The line showing how the dose required for control may increase was drawn fordemonstration purposes only. Lolium rigidum was treated with a typical annual rate of 375 g/ha diclofop-methyl.The relative dose level needed to control resistance in populations is shown as a function of the number ofdiclofop-methyl treatments. The sensitivity of determination of resistance was lost above a 500 fold increase inrelative dose. The populations of seeds were collected in farmer-treated fields and tested by Ian Heap at the WaiteInstitute. Modified and redrawn from Heap (1988). (D) A population distribution description of the data in (C)where the dose rates on the horizontal axis are arbitrarily plotted. Reprinted by permission of the American ChemicalSociety from Gressel et al. (1996)


465

Unfortunately, no genetic data have been reported onthe inheritance of resistance in ryegrass and there arestill insufficient data on the physiological biochemis-try of the mode(s) of resistance. As predicted >10 yearsago (Gressel 1988) one mode or resistance is clearlycorrelated with a slightly increased diclofop degrada-tion, due to increases in monooxygenase action.

Far more is known about a parallel yet less widespreadweed; Alopecurus myosuroides (blackgrass) evolvedresistance to a wide variety of wheat selectivegraminicides under the selection pressure ofchlorotoluron. This weed also has multiple germina-tion flushes during the long growing season of (true)winter wheat in Europe, and the residual herbicide wasapplied once per season, and soil levels of herbicideslowly decrease.

Blackgrass, like annual ryegrass is an obligateoutcrosser and is highly polymorphic and the geneticvariations in a large number of populations, with vary-ing resistance patterns has been exquisitely studied(Chauvel and Gasquez, 1994). In the “Peldon” resist-ant population (which is the most analogous to themulti-herbicide resistant populations of ryegrass), re-sistance is inhereited by “at least” two separate semi-dominant alleles. Two such semi-dominant alleleswould be able to allow creeping to four different I

50

levels of resistance. In parallel, but without geneticanalysis, the same Peldon blackgrass has been foundto have two distinct biochemical mechanisms confer-ring resistance; cytochrome P-450 monooxygenasetype of herbicide degradation (Hall et al. 1997; Hydeet al. 1997) as well as elevated glutathione peroxidaseactivity that can detoxify the oxygen radicals gener-ated by the herbicide until it is degraded (Cummins etal. 1997a, 1999).

Clearly single cytochrome P450 monooxygenases arenot the sole cause of ryegrass resistance to diclofop:single inhibitors of monooxygenase action only par-tially suppress resistance.

Glyphosate resistance in ryegrass has also been foundwhere extremely cut levels were used; 150g/ha onyoung seedlings (Pratley et al. 1996) and 550 g/ha onolder material in orchards (Powles et al. 1998). In-deed, recurrent selection of the resistant populationresulted in an I

50 shift of the population to a higher

level (J. E. Pratley pers. comm,). The only differencefound between resistant and susceptible populationswas a doubling of the level of EPSPsynthase activity,the target of glyphosate action (Gruys et al. 1999). Adoubled level of EPSPsynthase activity was sufficient

to increase the level of resistance to glyphosate fivefold in Lotus corniculatus, a species where biotyperesistance to glyphosate was found well before that inryegrass (Fig. 3)

Another case of low-level but widespread resistanceof a grass to a graminicide is that of Phalaris minor toisoproturon in India (Malik and Singh 1995). There,farmers who used full rates, evenly spread over thefield, at the right time, without burnt rice straw (=acti-vated charcoal) on the soil surface did not have resist-ance. Those that underdosed by rate cutting, unevenhand broadcasting of herbicide, late treatment (6 leafstage), or with burnt straw were the first to have resist-ance (Gressel et al. 1994). Here too recurrent selec-tion allowed I

50s to creep up (Malik and Singh 1993).

What would have happened if the farmers had usedintermediate or high rates of diclofop or glyphosateon ryegrass? From the herbicide distribution curve(Fig. 1) it is clear that higher rates assure that all thefield receives amounts that are above the MLD

95 of

individuals who have amassed one or even two genedoses for resistance, but not enough for controllingweeds containing more gene doses. Thus, two man-agement elements become clear: (1) evolution ofmultifactorially quantitative inheritance will be pre-vented by using intermediate or higher dose rates; (2)if one waits until there are too many resistant allelesaccumulated to apply an “intermediate” dose, the dosewill act like a low dose insofar is it can now select formore such alleles. This means that proactive resistantmanagement strategies must be instituted, not closingthe stable door after resistance predominates.

What happens using only high doses? At continu-ally high doses major genes for resistance, i.e. genesthat with a single mutation confer a huge modicum ofresistance; often hundreds of fold greater than the wildtype are the only type that can be selected. Such muta-tions are typically (but not exclusively) in the genecoding for the target protein to which the herbicidebinds. In locales where full (high) doses of diclofopare used on ryegrass such major mutations have beenthe only type of resistance found in ryegrass. This isbecause you would have to simultaneously have had alarge number of multifactorial alleles for resistance.The likelihood would be low if each were in the popu-lation at a frequency of 10-5, the likelihood of four to-gether would be 10-20, six 10-30, etc.


466

Figure 3. Relationship between the I50

for glyphosateand the specific activity of EPSP synthasses in differ-ence strains of Lotus corniculatus. (Plotted from datain Table 4 of Boerboom et al. 1990)

Multifactoral resistance can evolve while sequentialaccumulation of such alleles is under the pressure ofMLD treatment. Initially, major gene mutations fordiclofop resistance were not found in the resistantpopulations in Australia, only in North America, whereonly high doses are used. Later they were found in asmall population of the resistant populations in Aus-tralia. This indicates that there are many more multi-factorial genes to choose from than the single majorgene. This view is supported by the varying cross re-sistance patterns for other herbicides found among theryegrass biotypes, even among the few biotypes thatonly saw diclofop in their past history. This indicatesthat each biotype is a mix of different genes, all con-ferring diclofop resistance, but having somewhat dif-ferent abilities to confer resistance to other wheatgraminicides.

Theory (Gressel and Segel 1978), intuition, and somepractice suggest that the higher the herbicide dose rate,the greater the selection pressure eliminating suscep-tible individuals, the more rapid the evolution of pre-dominantly resistant populations.

Can you deal with Catch 22 If this analysis is cor-rect, minimum lethal low doses select for multifacto-rial resistance, high doses rapidly select for major generesistance, and intermediate doses select slowly formajor gene resistance. This is a three-way Catch 22.We have intuitively (Gressel et al. 1996) andmathematically (Gardner et al. 1998) modeled a re-volving dose strategy that mathematically “proves” to

be both cost effective, and holds resistance at bay longerthan using high or low doses alone. The idea is to usethe MLD for 2-3 years (while only a single allele canaccumulate and before a resistance to at least two ormore alleles can accumulate) and then use an interme-diate dose that is capable of controlling individualsthat have accumulated resistance to at least two alleles,i.e. probably 600-800 g/ha in the case of diclofop. Thisshould remove such individuals and allow the farmerto return to the MLD for a few more seasons.

Such mathematically “proven” models can only bevalidated by large scale field testing along with care-ful monitoring. The model allows the best of bothworlds, low or intermediate rates, which select moreslowly for single major gene inheritance; occasionalintermediate herbicide rates, which should set back anymultifactorial type resistance that may be evolving.Low/high revolving doses could be used where majordominant single gene inherited resistance is not ex-pected, as with glyphosate.

The above interpretation of the evolution of multi-her-bicide resistance in ryegrass under selection pressureof diclofop is at variance with that of esteemed Aus-tralian colleagues (Preston and Roush 1998). Unfor-tunately, in the more than a decade of intensive re-search on resistant ryegrass, no genetic studies havebeen reported on the mode of diclofop selected multi-ple resistance and far too little has been published onbiochemistry of resistance. This author feels that Pres-ton and Roush (1998) were too glib in discounting thelarge scale epidemiology (Fig. 2c) and they give thedata obtained an impossible alternative possible ex-planation; that the intermediate I

50 levels are due to

averaging mixed resistant and susceptible populations.The statistical data show that this is not the case (Heappers. comm.); there are moving bell shaped curves asdepicted in Fig. 2D. They also ignore the excellentgenetics (Chauvel and Gasquez 1994) and biochemis-try (Cummins et al. 1997a, 1999; Hall et al. 1997; Hydeet al. 1997) performed on blackgrass, an analogoussystem. Even less fortunate is that they believe thattheir simulations “show” (prove) instead of suggest,and they confuse simulated data with actual data. Mod-els can at best suggest what field experiments shouldbe done. Clearly, much more information is neededabout resistant ryegrass than is presently available: thebiblical warning to “know thine enemy” has not beensufficiently heeded.

I50

Glyphosate (kg/ha)

o I

50 E

PS

P-S

spe

cific

act

ivity

(nm

ol/m

in/m

g)

0.2 0.6 1 1.4

1.5

2.0

2.5

3.0

o

oooo

oo o

o

r 2 = 0.691


467

What to do about resistant ryegrass? The area ofresistant ryegrass in wheat has been increasing for morethan a decade – quite a bit since Ron Knight’s firstcases were discounted as “being unimportant”. In anearly review of the worldwide situation in wheat, Ipointed out that wheat has but one biochemical path-way to degrade herbicides allowing for selectivity –monooxygenases. This is in contrast with the widespectrum of pathways available in other species(Gressel 1988). The evolution of resistance in ryegrassto all wheat selective graminicides under the selectionpressure of diclofop suggested a futility in looking fornew graminicides for wheat. The conclusion was thatnew selectivities would only come from geneticallyengineering genes conferring resistance to graminicidesthat normally kill wheat (Gressel 1988). The initialpremise was partially wrong; “protectants” could beadded to some herbicides that elevated herbicide-degrading glutathione transferases in wheat, confer-ring resistance to herbicides such as fenoxaprop (Talet al. 1993; Cummins et al. 1997b). The utility of thisnovel use of protectants is great, until target site re-sistance evolves, as it has in many grass weeds.

There is a definite need to engineer herbicideresistances into wheat, as both target site and meta-bolic resistance to graminicides is becoming rampant,throughout the world, not just in Australia and India(Gressel 1998; Heap 1999).

INTROGRESSION OF TRANSGENICHERBICIDE RESISTANCE FROM CROPS TO

WEEDS

Risk assessment It is clear that in the Australian con-text, such transgenic crops have much to offer. Fromthe above interpretation of resistance, it is the only hopefor wheat, especially continuous monoculture wheatwhere it is desirable to rotate herbicides. It is also nec-essary when there is a desire to rotate crops. In manyagro-ecosystems the rotational crops are left withryegrass having resistance to a multitude of herbicides,or the rotational crop having hard to solve weed prob-lems of its own. An example is oilseed rape, which hasboth related brassica weeds and ryegrass as problems.Both genetic engineering as well as other biotechno-logical tricks (such as mutation and interspeciescrosses) have been used to generate herbicide-resist-ant crops.

Most countries have a scientifically untenable doublestandard in assessing the hazards from such crops thatseparates those strains derived obtained genetic engi-neering (even when plant genes are used) from others.

Canada no longer practices a double standard and allnewly resistant crops must go through the same hur-dles, which makes sense if the question is the likeli-hood of introgression from crop to weeds.

The needs for BD-HRC Millions of hectares are be-ing planted with biotechnologically derived herbicideresistant crops (BD-HRC), mainly in the western hemi-sphere and Australia. (The term “genetically modified”is scientifically untenable, as all crops have been ge-netically modified.) The farmers perceive the utilityof the use for the BD-HRC, as they have many alter-natives and still repeatedly purchase seed of BD-HRC.These farmers work on very small margins to massproduce the commodities they grow, and the value ofBD-HRC may be only marginally better than previ-ously used cropping practices. Longer-term considera-tions, which evaluate and support such uses are neces-sary.

The real values of BD-HRC come from instances wherethere really are no viable weed control methods, andthe impact of such BD-HRC could be considerabletowards a more sustainable world food production.Both the chemical and biotechnological industries havenot shown particular interest in generating needed BD-HRC.

Two rather different extreme cases where BD-HRC canmake a large impact on Australian production, arebriefly presented below.

Wheat - a crop in need of new resistances Wheat isoften cultivated where few other cash crops are grown,precluding widespread rotations. The high harvest in-dex, large ear, short-stature varieties are poor competi-tors with weeds and can rarely be cultivated withoutusing cost-effective herbicides. Minimum tillage sys-tems requiring more herbicides fit wheat agro-ecosystems with fragile soils.

Grass weeds have evolved morphological and pheno-logical mimicries to wheat for 6000 years (Barrett1993). They have recently evolved biochemicalmimicries that overcome wheat herbicides. The muta-tions confer cross resistance to herbicides that arechemically unrelated to the selector and have differentmodes of action (Gressel 1988; Moss 1992; Powlesand Holtum 1994; Malik and Singh 1995; Singh et al.1998a,b). Such resistances cover 40% of Australianwheatlands and millions of hectares in India. Non-chemical alternatives raise production costs, severelyjeopardizing supplies of the world’s major food grainand the existence of farming in these areas.


468

The immediate answer is to engineer resistances toinexpensive herbicides (Gressel 1988). Because toolittle profit is perceived to come from wheat seed oreven generic herbicides, wheat must be engineered bythe public sector. The situation may have changed inAustralia, where breeders rights have been strength-ened, creating an incentive to create new BD-HRwheats. Glufosinate resistance has been engineered intowheat, more as a marker gene than for utility (Weekset al. 1993). It is a environmentally and toxicologicallysafe herbicide to use, but expensive to manufacture(via dangerous organophosphorus intermediates).Glyphosate resistance has been engineered into wheatand field tested for the niche market of hybrid wheat,where it is less needed. The gene has yet to be madeavailable for use in low-input dryland wheat. Thereare two hazards to assess with glyphosate and otherBD-HR wheats; the risk of ryegrass evolving resist-ance to the herbicide, and the risk of introgression ofthe gene into wheat-related weeds. If glyphosate re-sistance is to be engineered into wheat, it should bealong with (“stacked” with) a second graminicide re-sistance, and herbicide mixtures should always be used.Glyphosate used alone will clearly engender evolutionof glyphosate resistance (Gressel 1996) and/or a shiftin weed spectra towards weeds that have never beencontrolled by glyphosate (Owen 1997). The use ofstacked BD-HR wheat and herbicide mixtures willdelay the resistance of ryegrass to glyphosate. The riskof introgression into wheat-related weeds will be dis-cussed later.

Oilseed rape has become an excellent rotational cropfor use with wheat in many places where wheat isgrown. There are many agronomic advantages of ro-tating a dicot with a monocot, especially vis a visweeds. It should be far easier to clean up grass weedsin oilseed rape than in wheat as there are more selec-tive graminicides for dicots. This is usually correctexcept in Australia, where ryegrass has successfullyevolved resistance to these graminicides.

Ethics, politics and economics of BD-HRC Muchmisinformation, disinformation and widely inaccu-rately-interpreted correct information has been prom-ulgated about BD-HRC (cf. Rissler and Mellon 1995),especially by those with an anti-technology, anti-bio-technology, and/or anti-pesticide bias. Conversely,those with potential commercial gains from sales ofBD-HRC, and/or the increased sales of the herbicidesto be used with them, portray BD-HRC as a risk-freepanacea to agriculture. The detractors often couch theiragenda in political, moral, or environmental terms. Not

all moral philosophers (Kline 1991) or environmen-talists (Lewis 1992) share these radical views.

We are warned that these crops can lead to the evolu-tion of “superweeds” that will inherit the earth (Kling1996). The rapid commercial release of such crops,often without broad-based scientific scrutiny, leads toa certain degree of public skepticism about the needs,utility, risks, and values (beyond profit) associated withthe use of BD-HRC. The severe pressures of the antigroups on policy makers makes it politically incorrectfor pursuing public-sector research in this area, whichaffects obtaining accurate information about the dan-gers. These pressures also prevent generating cropsneeding resistance to herbicides where theagrochemical or seed industry perceives little profit.The situation is further complicated by well-meaningscientists who are drawn into the debates, but lack theknowledge to balance the issues. The agronomic needsfor, and benefits of BD-HRC, have been widely touted,including in a well-balanced book with sections bydetractors (Duke 1996).

Discussions of BD-HRC have often dealt with the pur-ported environmental risks, but have rarely dealt withthe risks from a weed biology/ weed science perspec-tive, yet the major stated risk by the detractors isclaimed to be the BD-HRC becoming volunteer weedsor introgressing with a wild relative rendering itweedier; the ‘super-weeds’ of the mass media. Attemptsat such an assessment based on weed science was re-cently made (Gressel and Rotteveel 1999), using adefined set of uniform criteria set in a decision treeformat. Decision trees, by requiring discrete answersto sequential, stepped questions, lower the bias in ar-riving at conclusions vis a vis the risks deriving from agiven hazard.

Weeds by definition are very versatile, and have man-aged either to evolve resistance, or fill ecological vacu-ums left by species that became extinct as weeds (cf.Haas and Streibig 1982). There have been weeds thathave evolved resistance to every mechanical, chemi-cal, or cropping system and management procedureagriculture has put in their path.

Conversely, there are wild species that are unlikely everto become weeds unless they evolve a large number ofweedy traits (Keeler et al. 1996). Unfortunately, toomany risk studies do not differentiate between weedyrelatives of crops and wild relatives (e.g. Sindel 1997).Risk assessment must be performed on a local or re-gional basis, as the risks from the same BD-HRC willvary greatly from one agricultural ecosystem to another.


469

A second assessment should be done (but has not beendone in the past), about the effects on weed flora ofcountries that import bulk unprocessed commoditiessuch as wheat and oilseed rape (Gressel 1997).

Risks of introgression have been assessed on a case bycase basis. The Canadians were the first to attempt todelineate criteria before even having BD-HRC in place.With oilseed rape they first set out criteria “to evaluateplants with novel traits” (Anonymous 1994a), and thenspecifically evaluated oilseed rape in the context ofthese criteria (Anonymous 1994b). In a series of docu-ments they further evaluated imidazolinone (Anony-mous 1995a), glyphosate (Anonymous 1995b), andglufosinate (Anonymous 1996) resistant oilseed rapes.The decision process was based on their perception ofthe risks to regional agricultural ecosystems in west-ern Canada, and on the scientific knowledge of thetime. They did not consider other regions (includingthe eastern provinces) that may be importing the crops.Internationally the OECD and UNIDO are developinga series of “consensus documents” on the biology ofvarious crops (with regard also to related weeds) sothat there is a common starting point to evaluate eachcropping situation. Their document on oilseed rape(Anonymous 1997) has been released.

The source of a gene was an unimportant factor forCanadian risk analysis. How a gene got there is not asimportant as what the gene does in the crop, and howand whether it will move to weeds, or whether the cropwill become a weed. Much of the stated hazards ofinterspecific introgression from BD-HRC are based onartificial laboratory experiments, which prove thatintrogressions could happen. Thus they show that thehazard exists, but give little indication of risk; howquickly such transfers will occur in the field or how fitrecipeints will be to cope with competition. The timefactor is not inconsequential; if resistance introgressesto produce resistant populations more slowly than natu-ral mutational evolution, what is the significance ofintrogression?

Two types of gene transfer are widely discussed: (1)vertical - within a species (2) horizontal - transferamong unrelated species, usually by asexual means.Biology is not as clear-cut; there can be some sexualtransfer between species in the same genus and closelyrelated genera that are typically included in somediscussions of horizontal gene transfer. Because ex-trapolations are often made from these rare cases ofgene transfer among closely related species to “prove”that all horizontal transfers are possible, we suggest

terming these special cases as diagonal gene transfer,denoting the grey area where they exist.

Vertical and diagonal transfer possibilities are obvi-ous to any plant scientist, but horizontal transfers, withtheir disastrous implications to agriculture are not. Thepossibilities of horizontal transfers are extrapolatedfrom the intergeneric and inter-familial plasmid-me-diated transfer of traits among microorganisms, whichhave allowed transfer of antibiotic resistance (a traitanalogous to herbicide resistance) among unrelatedpathogens. The claim continues that because plasmidsare often used as vectors in the genetic engineering ofcrops, inter familial transfers will become common-place, or at least “inevitable”. This claim does not standup to epidemiological experience with organisms suchas Agrobacterium tumefaciens and A. rhizogenes. Theplasmids for laboratory gene transfers come from theseAgrobacterium spp., which naturally infect a broadrange of dicots, using the plasmid as part of the infec-tion process. If such inter-familial transfer were to oc-cur, it would have been seen over the past 50 yearswith naturally occurring herbicide resistances.

There are no known cases where such genes have trans-ferred inter-familially from any crop to weed viaAgrobacterium, despite the great selective advantagesthat such weeds would have. The more than ten mil-lion hectares of herbicide-resistant weeds that haveappeared in the past thirty years can all be traced tomutational selection evolution and not to plasmid-mediated horizontal gene transfer. Additionally an ex-tensive survey of all the Genebank database found fewAgrobacterium DNA sequence pieces in any of theplant genes, which would have been expected in themillions of years of co-habitation. This matching tasktook hundreds of hours of computer time (Rubin andLevy 1999). Horizontal gene transfer would be a wasteof time to discuss further; diagonal gene transfer is ahazard for which the risks must be estimated.

Generalizing from hazards to risks Because of ge-netic variability of crops and weeds, and chemical vari-ability in herbicides, their effects and modes of action,one cannot make easy generalizations about the risksof introgression of resistance. Each case of predictingthe risk of introgression must be evaluated on its mer-its, often after basic biological, genetic, and epidemio-logical studies. More importantly, other issues mustbe considered:


470

(1) What is the benefit to agriculture of having re-sistance in a certain crop?

(2) What are the possibilities of, and implicationsfrom, having herbicide resistance pass into aweedy or wild species?

(3) What are the possibilities of and implicationsfrom having the BD-HRC becoming a volun-teer weed in agricultural ecosystems, or becomean alien weed in ruderal or more pristine eco-systems?

The final decision is ultimately a balance between sci-ence, economics, local benefits, local values, pressuregroups, as well as local politics. The politicians oftenuse science for clearly political decisions (Powell1997). Still, there is good reason that the criteria forrisk assessment of BD-HRC should be uniform, usinguniversal criteria and processes of examination. Whatis the risk from seeds of a commodity crop for process-ing, when the same seeds would be too risky to sow,yet might escape in the importing country? The onusis on the importing country to demonstrate that it isnot erecting illegitimate, protectionist, and artificialtrade barriers. Indeed, to prevent trade wars, politicalcompromise has led to allowing importation of com-modities initially claimed to have untenable scientificrisks (Powell 1997). “Science” was used for bargain-ing purposes. All governments should use identicalscientific risk assessment criteria for theiragroecosystems to determine if the risks are muchgreater in the potential importing country than in theexporting country. The importer would have to deter-mine whether that the benefits of importing are greaterthan the potential costs of mitigating procedures (i.e.eradication of volunteer or introgressed weeds). Indeedone could envisage an involvement of the insuranceindustry in risk assessment should there be a require-ment that importers or exporters insure themselvesagainst such risks.

Risks of introgression of transgenes to related weedsWheat Genes from wheat easily introgress into thegenomes of some related weeds, weeds that are relatedto the progenitors of wheat. There is much recent in-formation of introgression of genes from hexaploidwheat into Aegilops cylindrica a very problematic weedin the western plain states and the Pacific northwest ofthe USA (Zemetra et al. 1998; Seefeldt et al. 1998).This species shares a D genome with hexaploid wheatand homologous recombination occurs naturally un-der field conditions. It would theoretically be muchharder to obtain transfer from tetraploid (durum) wheatto this species, as durum lacks the D genome. Thus,

homeologous recombination (crossing over betweenrelated but dissimilar chromosomes) would be requiredwhen herbicide-resistant durum is used.

Genes for various traits have been transferred frommany wild grasses to wheat, especially from the genusAegilops. Many Aegilops species are considered to beTriticum (Kimber and Feldman 1987). The only otherAegilops spp. known to have a wheat-homologousgenome is A. squarrosa = Triticum tauschii (Coss.)Schmal. (Kimber and Sears 1987). Only homeologousrecombination can occur between hexaploid anddurrum wheats with the other less related species.

Many intergeneric hybrids generated by breeders donot occur in the field; natural alleles have not passedfrom wheat to wild, related but mainly ruderal spe-cies. The breeders have to resort to forced crosses andthen techniques such as embryo rescue in tissue cul-ture to save the hybrid embryos that would otherwiseabort.

In Australia, relatedness to wheat is relative. Sindel(1997) in his analysis of crop-related weeds lists nonefor wheat. Other Australian weed scientists listThinopyrum juncerforme, Lophopyrum (Thinopyrum)elongatum and Elytrigia (=Agropyon repens as beingpresent in various habitats. The first two are not wide-spread, and rarely in agro-ecosystems, and the last is asummer weed. Thus, time and place would precludemating. As it takes much more than a single gene, evenone for herbicide resistance, to turn wild species into aweed (Baker 1974, 1991) the risks of introgression ofwheat to these weeds are very low.

Oilseed rape Most oilseed rapes cultivated are Brassicanapus, a not too ancient tetraploid derived from theCC genome of B. oleracea and the AA genome of B.campestris = B. rapa (U 1935). Thus, the only weedwhere homologous recombination can occur is B.campestris. Still, homeologous recombination isknown, especially in the laboratory, with many relatedspp. This required hand pollination after emasculationof the weed, male sterility or self incompatibility inthe weed, massive amounts of crop pollen, and /orembryo rescue of the rare progeny, that are mostly ster-ile or runts (see review of Darmency 1994b; Landboet al. 1996; Brown and Brown 1996; Mikkelsen et al.1996; Metz et al. 1997; Lefol 1996a,b; Bing et al. 1996;Scheffler et al. 1995; Conner and Dale 1996, to men-tion a few recent papers). Initially, the fear of BD-HRCprecluded performing such experiments in the fieldespecially where the fears were greatest, in Europe.This hysteria prevents obtaining field data that mightsubstantiate, or more likely allay the fears.


471

The significance of such laboratory studies to the fieldsituation was evaluated by Karieva et al. (1996). Theolder epidemiological / apocryphal reports are actu-ally more relevant to risk analysis than many of theartificial laboratory experiments; the older results couldindicate that such transfers can occur in the field, aswell as the time until predominance, and the competi-tive advantage (if any) of such introgressions.

Most studies rating risks of movement do not differ-entiate between the reports of field transfer and thestudies showing it could occur. In the rush to obtainsuch information, erroneous information has managedto get published in reputable journals. For examplethere was a report of 30% hybridization from transgenicrape seed pollen up to distances of 1 km, >70% in nearproximity (Skogsmyr 1994). This last number is dou-ble the theoretical maximum, as calculated by Connorand Dale (1996). The PCR reactions used to obtainthis fantastic hybridization rate were not controlled byassaying non-transgenic plants, nor were any otherbiochemical or molecular methods used to verify thePCR data, yet this artifactual study is likely to be quotedas fact. Few studies dare to comparatively estimate howlong it will take to have resistance introgress and pre-dominate in field weed populations vs. how long itwould take resistance to evolve by natural selection,vs. the expected commercial lifetime of the herbicide.

Assaying introgression in the field There are waysto ascertain the rapidity of gene movement withoutcausing lasting damage to agriculture and/or the envi-ronment: simply insert a gene for resistance to a rarelyused herbicide. There would be little consequence tothe herbicide becoming extinct due to the resistancedisseminating into the wild. Another way to establishthe rate of transfer of genes is to abolish the doublestandard; most countries including Australia do notscrutinize the use of BD-HRC where the resistance is“natural”, i.e. from selected mutations or artificial ge-netic crosses. ALS-level resistant tissue culture selec-tion in crops was acceptable in the USA and elsewhere(after registration of the herbicide for those crops).Triazine-resistance, laboriously transferred from wildBrassica campestris into oilseed rape, B. napus(Beversdorf et al. 1980) has been under no special scru-tiny in Australia.

Similar genes introduced transgenically are forbiddenin many countries and under scrutiny elsewhere. Atpresent, if a weed becomes resistant to any ALS herbi-cides, it cannot be known whether it evolved resist-ance naturally by mutation, or was introgressed throughcross pollination with a non-transgenic BD-HRC. The

highly mutable ALS gene (ca. 10-6 natural resistancefrequency in populations) (Saari et al. 1994) quicklyevolves naturally in weeds. Engineering the same al-lele with either a two base change coding differencefrom the natural resistance allele, or with differentintrons, would allow easily differentiation of mutationalevents from introgression. This would indicate the rateof evolution due to introgression vs. the rate of evolu-tion from natural mutation.

In the case of triazine resistance, there could be addedbenefits. The trait is maternally inherited, so one mightassume it will never transfer. Maternal inheritance isnot absolute; 0.2% pollen transfer of triazine resist-ance was found with genetic markers (Darmency1994a). Maternal inheritance of chloroplast-encodedtraits is typical, but there are many cases where suchtraits are otherwise inherited (Tilney-Bassett andAbdel-Wahab 1979), so it will be necessary to ascer-tain frequencies of paternal inheritance with each cropsituation where maternal inheritance is predominant.

Indeed, if the herbicide is used in the BD-HRC, thensusceptible weeds growing in its midst will notintrogress the resistant genes; dead weeds don’t havesex. The genes could introgress into nearby unsprayedweeds. Seed set on emasculated plants from oilseedrape was measured 1.5 km from pollen source(Timmons et al. 1996); can the progeny compete, orsurvive in feral populations without selector? With-out emasculation resistant pollen fertilized 24% of con-specific plants in the immediate vicinity but <0.017%just 10 meters away. If BD-HRC oilseed rape seedscarry over as volunteers to other crops where relatedspecies are serious weeds, gene exchange might hap-pen through this route and long distance dispersal be-comes unnecessary.

Many weed scientists and others were surprised thatthe Canadian authorities allowed the field use ofglyphosate, glufosinate, (and soon) bromoxynil-engi-neered transgenic oilseed rapes. The surprise was dueto the known introgression of herbicide-resistant genesinto weeds, including the problematic Brassicacampestris= B. rapa (e.g. Mikkelsen et al. 1996;Kerlan et al. 1993; Lefol et al. 1996a,b). Brassicacampestris has the dubious distinction of being bothdomesticated to become various crops (Polish oilseedrape, turnip, Chinese cabbage, and pak choi) in manyplaces, and the more ancient con-specific wild form isa pernicious weed in other areas (Holm et al. 1997).While botanically identical, they all have differentmorphotypes, with very different phenologies, biolo-gies and competitiveness. Most of the B. campestris


472

crop types are easily controlled. As volunteer weedsthey have never left agricultural or ruderal areas. Theirbotanically, but not phenotypically identical weedy twincan be very problematic in agro-ecosystems, and wasa predominant weed in grains before the advent of se-lective herbicides (Holm et al. 1997). It is hard to pre-dict what will become of the feral populations of Polishoilseed rape, which still remain ruderal. They couldevolve to become more weedy.

Deleterious weed genes have introgressed into bothoilseed rapes from feral populations, lowering yieldand oil quality (e.g. McMullan et al. 1994). A majoruse of herbicide-resistant oilseed rapes is to facilitatecontrol of its wild relatives. The Canadian authoritiessubjected ALS-R rape derived by mutagenesis to fullregulatory scrutiny before release to the market. Theirdecisions (Anonymous 1995a,b, 1996a,b) allow unre-stricted field cultivation, while noting the likelihoodof introgression, and stating that the worst case wouldbe the loss of the particular herbicide to control suchweeds (Anonymous 1995). The decision stated thatintrogression would not increase weediness of crop orrelated weeds outside of agriculture, partly based onrelease studies by Crawley et al. (1993).

B. campestris is the weediest of species related tooilseed rape (Holm et al. 1997), and the one specieswith demonstrated field transfer of genes (Mikkelsonet al. 1996), but not quite as readily as initially pre-sumed (Landbo et al. 1996). The Canadian authoritiesdid not require two safeguards that might have low-ered the risk of transfer of resistance from B. napus toweedy B. campestris:

1. Choice of genomes. If only plants bearing resist-ance on the C genome were used, resistance could onlytransfer to B. campestris by rare homoeologous pair-ing, which seemed to have occurred in an artificialsystem (Metz et al. 1997). Glyphosate resistance couldbe coded for by two genes, one for modified target andthe other for degrading the herbicide. Resistance trans-fer would be delayed if each were inserted on separateC chromosomes, requiring two independenthomeologous transfers. Whether fortuitously or by in-tent, the glufosinate and glyphosate resistances forCanada are on the safer C genome, whereas most ofthe resistances studied in Europe that introgressed toother species were on the B genome (R. K. Downey,pers. comm.).

B. napus is self compatible, and is pollinated by in-sects. In an isolated field experiment, transgenic traitsappeared in <0.02% of seed of non- transgenic oilseedrape planted 200 m away, and in 0.004% at 400 m away

using small plots (Scheffler et al. 1995). The authorsbelieve that the frequency would have been even lessif larger fields were used, based on bee behavior. Theallowable level of cross pollination is 0.1% for “breed-ers basic” and is 0.3% for “certified” seed, >20 timesmore than achieved by transfer at these distances. Thisindicates that the risk is low when compared to alreadyaccepted risks in normal seed multiplication practice;however far less risk may be acceptable in some casesof potential outcrossing of BD-HRC resistance genes.

The interspecific distaste for alien (non con-specific)pollen is a biological barrier to introgression. Thebrassicas are typically considered to be self incompat-ible obligate outcrossers. Yet, when solitary B.campestris plants are grown among B. napus, most ofthe offspring on the B. campestris are not hybrids. TheB. campestris plants overcame self-hatred and polli-nated themselves, despite ample alien pollen (Landboet al. 1996). In some areas the feral B. campestris flow-ers at a different time as the crop, further diminishingthe mating possibilities. Going further afield tointergeneric crosses that have been found in the labo-ratory, six hybrid seeds were obtained in 50,000siliques that formed on male sterile B. napus pollinatedby Sinapis arvensis (Lefol et al. 1996a). The siliqueson the supposedly “male sterile” B. napus producednearly 900 B. napus seeds not hybrids. More impor-tant though is the gene flow in the opposite direction,from B. napus to S. arvensis.

Thus, from all these studies showing that introgressioncan occur, we still have little idea how quickly it willoccur in field situations, and the only way to learn aboutthe rates of field evolution will be epidemiologicallyfollowing large scale field use. Still, hybrids betweenB. napus and wild species are unlikely to perpetuate.Farmers will typically cultivate only certified seed; i.e.seed from totally weed-free fields, having wide weed-free areas (200-400 m) around them. Contaminatedseed ends up at the crusher, being turned into oil andmeal, without introgressed DNA being perpetuated.

Many were surprised that the Canadian decisions didnot call for an active monitoring system to scout outpossible gene or plant movement. Such a system wouldbe costly and cumbersome, and for a few years wouldprobably never find a resistant individual. Still, it wouldonly cost ink and the will to print a request on eachseed that a “hot line” be called if patches of putatively-resistant weeds appeared, or if ruderal populationsspread to areas never before colonized. This, togetherwith a bonded requirement of the commercializers tocover the cost of maintaining a hot line and then as-certaining whether introgression or movement had


473

occurred, as well as have a liability to eradicate feralor resistant populations before they spread, would makeregulatory and biological sense. Much has been writ-ten on how comparatively easy it is to eradicate smallpest populations, and how impossible it is after theyhave reached a critical size (e.g. Moody and Mack1988; Thill and Mallory-Smith 1997), yet few learnthese lessons about the need for early eradication.

Both industry and growers should want to know howquickly and to what extent to expect introgression, andwhat the consequences might be. As described above,this could be done cultivating oilseed rape bearingtransgenic ALS resistance and/or bearing transgenictriazine resistance, as it would be possible to distin-guish between volunteer or introgressed offspring fromnaturally appearing mutants.

In the case of ALS resistant oilseed rape, the transgenictrait would actually be less likely to introgress thanthe mutant. Publications and patents referring to thetransgenic oilseed rape state that the 35S promoter wasused. The 35S promoter comes from the cauliflowermosaic virus (CaMV). The expresson of this promoteris suppressed when herbicide-resistant oilseed rape isinfected by CaMV, and the plants are then sensitive tothe herbicide (Al-Kaff et al. 1997). It has been esti-mated that ca. 80% of Brassica weeds are naturallyinfested with this virus (Cooper and Raybould 1997).Thus most wild Brassica plants that received the geneor became feral would still be sensitive to the herbi-cide, and would have no selective advantage.

The triazine resistance gene coding an amino acid 264transversion from serine to glycine in oilseed rape ishighly unfit (Gressel and Ben-Sinai 1988). Reciprocalcrosses were always less productive with triazine-re-sistant female parents (Beversdorf et al. 1988). Theyare grown in Australia (without regulation), whereoilseed rape gene transfer to wild species is consid-ered “high risk” (Sindel 1997). The amino acid 264transversion of serine to threonine that evolved in po-tato is more fit than serine to glycine (Smeda et al.1993) and should increase yields if transformed intooilseed rape chloroplasts. This should be effective un-til B. campestris again evolves resistance to triazines,as it had in eastern Canada (Maltais and Bouchard1978).

The question of how quickly resistance genes will movefrom B. napus to B. campestris may be moot, as Polishoilseed rape (B. campestris) with various herbicideresistance genes will soon be released in westernCanada (R. K. Downey, pers. comm.). In this regionthey do not consider transfer to the con-specific weed

a problem, but the crop as a volunteer weed could be-come a problem, especially if it becomes multiply re-sistant to all four herbicides introduced in various va-rieties by cross breeding. There is nothing to stop ge-netic engineers from further introducing readily-avail-able genes conferring resistance to other herbicidessuch as 2,4-D (Streber and Willmitzer 1989) in tooilseed rapes. Nothing would then be left to control B.campestris as a volunteer weed in most Canadian plainscrops.

MITIGATING RISKS

There are various failsafe mechanisms that can be usedto mitigate the risk of introgression, when and if it doesoccur.

Apomixis as a failsafe Some apomictic seed is ac-tually of vegetative origin (Koltunow et al. 1995).Apomixis is being developed to establish hybrid vig-our without crosses. If apomictic varieties are pollenfree, then their genes cannot introgress into other spe-cies or into other varieties of the crop or into con-spe-cific weeds. The lack of viable pollen is probably theonly failsafe that would be acceptable to some detrac-tors, who fear intervarietal movement of transgenes,especially to “organic” crops.

Gene Placement Failsafes

Chromosomal Wheat and oilseed rape are composedof multiple genomes derived from different wildsources (Kimber and Sears 1997; U 1935). In any givenlocale it is possible that only one of the genomes ofthe crop is identical to that of a related weed allowingeasy gene transfer. As the D genome of wheat is com-patible with the D genome of Aegilops cylindricatransgenes easily introgress from the B genome ofoilseed rape to many brassica weeds and wild species.One should perform the cytogenetics to assure that thetransgene is on the incompatible A or B genomes ofwheat or the C genome of oilseed rape.

Hybrids A simple failsafe can be used in hybrid wheator rape, when and if they become widely commerical.If a dominant transgene for herbicide resistance isplaced in the male sterile line, there will be no possi-bility of introgression in crop-production areas. Carewill have to be taken in the seed production areas whenthe male sterile line is restored. Such areas must bekept free of related weeds, a typical precaution in seedproduction generally practiced before the advent oftransgenics.

Plastome or chondriome If the transgene for herbi-cide resistance is placed on the mitochondrial or plas-tid genomes, as has been done in tobacco (Daniell et


474

al. 1998; Kota et al. 1999) there should be little possi-bility of gene flow, due to the maternal inheritance ofthese genomes. Species that are often claimed to haveno paternal inheritance often have about 0.1% pollentransfer of traits. Large scale experiments should beperformed with both crops to assure that the level ofpaternal transfer of traits is sufficiently low to justifyusing this strategy. There are technical problems inobtaining plastome specific transformation.

Transgenetic Mitigation (TM) Genetic engineeringcan be used to mitigate any positive effects transgenesmay confer. If the herbicide resistance gene engineeredinto the crop is flanked on either side by a TM gene ina tandem construct, the overall effect would be delete-rious to weeds introgressing the construct from a crop(Fig. 4). This is based on three premises:

a. Tandem constructs of genes act genetically as tightly-linked genes and their segregation from each other isexceedingly rare.

b. There are traits that are either neutral or positive fora crop that would be deleterious to a typical or volun-teer weed, or to a wild species.

c. Because weeds are strongly competitive amongstthemselves, and have large seed outputs, even mildlydeleterious traits are quickly eliminated frompopulations. Even if one of the TM alleles mutates, isdeleted, or crosses over, the other flanking TM genewill remain, providing mitigation.

Other TM traits that could be used are best visualizedwhen observing the differences between crops andweeds. This is best illustrated with two cases: (a) wheatand weedy relatives (b) oilseed rape (Brassica napus)and feral and weedy Polish rape/wild radish B.campestris=B. rapa, as summarized below.

Seed dormancy Weed seeds typically have secondarydormancy with seeds from one harvest germinating bitby bit throughout the following season, and over anumber of years (Vleeshouwers et al. 1995). This evo-lutionary trait is considered to be a risk-spreading strat-egy that maximizes fitness while reducing losses dueto sib competition (Hyatt and Evans 1998; Lundberget al. 1996). Staggered secondary dormancy preventsall the weeds from being controlled by tillage beforethe crop is planted, or controlled by tillage or herbi-cides during crop rotation. Rare mutants lacking sec-ondary dormancy were selectively propagated duringcrop domestication, as the loss of secondary dormancyis desirable to the farmer, who wants uniform germi-nation after planting the crop. Crop seed that germi-nates uniformly after planting gives a uniform harvest.

This can well be seen when comparing crops with theirweedy progenitors and relatives (Ling-Hwa andMorishima 1997).

Genetically abolishing secondary dormancy would beneutral to both crops, but deleterious to the weeds.Tillage, crop rotation, and preplant use of herbicides,all standard practices would control the uniformly-germinating weed seeds lacking secondary dormancyin rotational crops.

Ripening and shattering Weeds disperse their seedover a period of time and much of the ripe seed “shat-ters” to the ground, insuring continuity. A proportionof the weed seed is harvested with crop seed, contami-nating crop seed, facilitating weed dispersal to wher-ever the crop seed will be grown.

Weeds have evolved morphological and phenological“mimicries” to the crop seed (Barrett 1983; Gould1991) necessitating continual evolution and refinementof techniques to remove the contaminating weed seed.Crop varieties have been selected for non-shattering,but recently domesticated crops such as oilseed rapestill suffer from shattering (Simon 1994; Prakash 1988;Price et al. 1996).

The first problem in domestication is control of shat-tering (Young 1991; Levy 1985). In addition to theloss of yield, the shattering of crop seed results in thebecoming a volunteer weed especially in oilseed rape(Lutman 1993).

Uniform ripening and anti-shattering genes would bedetrimental to weeds, but neutral for wheat (because itripens uniformly and does not shatter easily after thou-sands of years of selection), and positive for oilseedrape, which still has a shattering problem. Crop seedcontaminated with low levels of weed or volunteer seedare typically used for feeding or processing, only weedfree “certified” seed grown where introgression isguarded against is (or should be) sown.

Dwarfing. For millenia wheat had been selected forheight, to outgrow weeds, limiting the photosynthateavailable for grain. Weed evolution kept apace, givingrise to taller weeds. The advent of selective herbicidesto kill weeds allowed for genetic dwarfing of thesecrops; more seed harvest, less straw. Some of the na-tive dwarfing genes were tightly linked to genes re-ducing general yield potential, illustrating the poten-tial for single transgene. Still, the lowering of height,precluding the concomitant problem of tall plants lodg-ing and increased yield, especially after fertilizeruse (which previously promoted lodging), allowed


475

countries like India to become self sufficient, despitepopulation increase.

Various new systems of genetically engineered heightreduction are being introduced. These include genesrelating to hormone production (Azpiroz et al. 1998;Schaller et al. 1998; Peng et al. 1999) as well as thosedealing with shade avoidance. Much of stem elonga-tion is in response to shading. This is advantageouswhen competing with other species, but not in a weed-free crop stand where only siblings are competing. Theoverexpression of specific phytochrome genes preventsrecognition of shading and thus the plant remains short(Robson et al. 1996). This is advantageous for a cropand could also be used where the present dwarfinggenes prevent obtaining the highest yields. This traitwould be disadvantageous for a weed that must com-pete with the crops; it would be shaded over by thecrop.

There is considerable debate about the advantages thatwould accrue to weeds from the primary transgenictraits. Resistance by modified site of the herbicide bind-ing to its target should only confer an advantage whenthe herbicide is used. It is unknown whether therewould be pleiotropic advantages of herbicide resist-ance due to introducing genes for metabolic inactiva-tion of herbicides.

Let us assume that the primary transgenic trait confersan advantage to a weed; how much will TM traits ac-tually mitigate that advantage? Weeds are not onlyhighly competitive with crops, they are competitivewith weeds of other species as well as within their ownspecies. Weeds often produce thousands of seeds, insteady state conditions, to replace a single plant, sug-gesting extreme competition to be the replacement; theselection for high competitive fitness is intense. Thishas dual implications in our situation.

Figure 4. Comparison of current and TM technology. In the current technology, on the rare occasions that transgeniccrop pollen (e.g. conferring herbicide resistance, insect resistance, etc.) reaches and fertilizes a live, related weed,there can be some offspring bearing the trait, the proverbial ‘superweed’. In the TM technology, the gene confer-ring the desired trait) is flanked in a tandem construct by TM (transgenetic mitigator) genes that are positive orneutral to the crop, but deleterious to the related weedy species spp. Fertilization of related weedy rices with suchpollen will give rise to offspring that are herbicide resistant, but are also non-competitive wimps


476

After a weed introgresses a transgene and then stabi-lizes (eliminates cytogenetic incompatibilities), the traitwill quickly spread through a population, even if ithas a marginally positive fitness advantage (Thill andMallory-Smith 1997). Conversely, one can balance thedisadvantage of TM traits against the advantage of theprimary trait. This must be done in both in the pres-ence and absence of the reason for having a herbicideresistance trait. Herbicide resistance only provides anadvantage when the herbicide is used. Indeed, whenthe herbicide is not present, the transgene resistancetrait can be disadvantageous; as demonstrated an ALSresistance gene (Bergelson et al. 1996).

The expression of one gene of a tandem construct couldbe lost in the transgenic crop, and sometimes the ex-pression is lost after a few generations. The reasonsfor these losses are not always clear nor relevant forthis discussion, as only stabilized progeny oftransformants are released to agriculture. If all traitsof a tandem construct are expressed after 4-5 genera-tions of backcrossing it is fair to consider it stable, i.e.as stable as any native, tightly-linked adjacent genes.

Each TM trait should work in a balance with the pri-mary trait, and where the primary gene gives a strongadvantage to a weed, it might be necessary to havemore than one TM trait in a construct to obtain bal-ance.

The risk of losing TM traits can be further decreasedby combination with a cytogenetic failsafe, where theseare available. If the tandem construct is located on anon-homologous chromosome in wheat or oilseed rape,then only rare homeologous recombination can moveit. As there is no selective advantage to losing the TMtrait on the non-homologous chromosome, one cancompound the frequency of likelihood of homeologousrecombination with the frequency of loss of the TMtrait(s), reducing risk.

The question of how low a risk must we need to attainis for regulators, but when they do deliberate they mustask; when will the related weed evolve the trait in ques-tion by natural means. The mutations conferring re-sistance to ALS-inhibiting herbicides are naturallyfound in plant populations at a frequency of one in amillion, and ALS-inhibiting herbicides are widely used.If a TM construct had been inserted in tandem with anALS gene, with a likelihood of segregating of 10-10,then the likelihood of getting ALS-resistant weedt rela-tives would not be appreciably changed byintrogression from transgenics. Such analyses shouldbe made wherever possible.

Which TM traits are available as gene sequences?Some possible traits for TM constructs just exist asnamed genes that are inherited, others are also mappedto positions on various chromosomes, and a few areactually characterized as sequenced genes. Thus, notall TM traits are immediately available for insertion intandem constructs. Still, there can be many differentways for a plant to confer a TM trait, and thus, morethan one gene might be available.

Secondary dormancy Unfortunately, Arabidopsis, thetypical source for genes, has already been sufficientlydomesticated that it is unlike cruciferous weeds; thelab strains no longer have strong secondary dormancy(Van der Schaar et al. 1997). A mutant that is insensi-tive to abscisic acid and lacks secondary dormancy wasfound in a wild, undomesticated Arabidopsis strain(Steber et al. 1998). Perhaps a way to find more genesis to use the genetic differences between wildArabidopsis strains and the lab strains presently used,as is being done in other instances (Ackerman et al.1997). Much more basic mechanistic research must beperformed with the crop/weed pairs before the magic‘abolish secondary dormancy” TM genes can be used.

Shattering Physiologically, one way to avoid seedshattering is to have uniform ripening. Early maturingseeds of oilseed rape on indeterminate, continuouslyflowering varieties typically shatter. Determinacy, withits single uniform flush of flowering is one method toprevent shattering, but this often shortens the season,reducing yield. The hormonology of the abscissionzone controls whether shattering will occur and it ispossible that if cytokinins are overproduced, then shat-tering will be delayed. As with secondary dormancy,no sequenced genes are yet at hand, except for cytoki-nin overproduction.

Stature limitation Many of the genes used so far toobtain vertical deprivation (the politically correct termfor dwarfing) seem to have an unknown function. Stillmany genes are known, that control height.

Gibberellins Preventing the biosyntheses ofgibberellins reduces height (Webb et al. 1998), whichis the basis of many chemical dwarfing agents usedcommercially to lower stature and prevent lodging ofwheat. The enzymes and genes controlling varioussteps in gibberellin biosyntheses are known. Copalyldiphosphate synthase, ent-kaurene synthase, and ent-kaurene oxidase are responsible for early stages in thebiosynthesis of all gibberellins (Smith et al. 1998;Yamaguchi et al. 1998, Hedden and Kamiya 1997;Lange 1998; Helliwell et al. 1998). Arabidopsis mu-tations bearing mutations in any of them are dwarfed,


477

with the dwarfing is reversible by gibberellin treatment.Overexpression of a gene coding for ent-kaurene syn-thase, causing co-suppression mimicked the mutantphenotype. Additionally, a gene has recently been iso-lated that confers gibberellin insensitivity when trans-formed into grains (GAI), and thereby induces dwarf-ing (Peng et al. 1999).

Some processes such as flower stalk bolting are con-trolled by specific gibberellins; in radish GA

1 and GA

4

are responsible (Nishijima et al. 1998). It may be nec-essary to characterize the genes coding for themonooxygenases and dioxygenases that are responsi-ble for these later steps (Hedden, 1997). Some of thesegenes have been isolated as well (Kusaba et al. 1998).

Brassinosteroids This new group of hormones alsocauses elongation of stems in many plant species, andtheir absence results in dwarf plants. A 22 d-hydroxy-lase cytochrome P450 has recently been isolated thatcontrols a series of these steps in brassinosteroid bio-synthesis (Choe et al. 1998), and plants missing theenzyme are dwarfed (Azpiroz et al. 1998). Addition-ally, suppressive overexpression of a sterol C24-methyl transferase also causes dwarfing (Schaller etal. 1998).

Shade avoidance Various forms of the pigment phy-tochrome interact to detect whether a plant is beingshaded (Smith and Whitelam 1997; Devlin et al. 1998;Torii et al. 1998). Phytochrome recognition of shad-ing leads to stem elongation, which is unneeded in aweed-free crop. The engineering of suppressiveoverexpression constructs of one of thesephytochromes led to plants that did not elongate inresponse to shading (Robson et al. 1996). Much of thegene isolation has been from Arabidopsis, yet the sup-pressive overexpression was active in dwarfing to-bacco.

Because the advantages of transgenic rape are so great,both industry and the farmers are clamoring fortransgenics. They even do so when they know thatintrogression is imminent. They point out that anotherresistance will then be available. There are a limitednumber of herbicide resistances one can engineer andthus a limited number of times that you can allow re-sistance to introgress.

Still, much basic and applied research is needed to findand use TM genes. Many TM genes may prove to havenear equal worth in increasing agricultural productiv-ity as the primary genes presently being used.

Should there be a regulatory delay in allowing theuse of single primary genes, without TM genes in

situations where there is a strong risk of introgression?Should such decisions on delay be voluntary? My ownview for high risk cases is a paraphrase of a sign aboutplaying of radios, seen in the Edinburgh zoo: wherethere is a high risk of introgression, “consideratebiotechnologists will not release primary genes with-out tandem TM genes; others may not” release singleprimary traits.

FORCING WEEDS TO REVERT TO BEINGINNOCUOUS WILD SPECIES

The approach of the chemical industry to ryegrass andthe weed problems has been to attempt finding newselective herbicides, while that of the biotechnologyindustry to ryegrass and other weed problems has beengenerating resistant crops that can withstand herbicidesthat do yet control these weeds (Gressel, 1998). Agri-culture needs additional paradigms for controlling suchweeds.

A group of us propose an added solution, one thoughtuntil recently to be in the realm of science fiction: togenetically debilitate the weed. (J. Gressel E. Rubin,A. Levy, in manuscript). If this solution is successful,it will integrate with and facilitate other successfulcontrol mechanisms, leading to more durable control.

Modifying pests to facilitate their control is not novel.The release of irradiated male insects to impregnatefemales with sterile sperm has been used successfullyin some venues. Here we propose a transgenic ap-proach, using a wide variety of possible genes that willbe deleterious when turned on; genes that mimic her-bicide action; that inhibit plant growth; that rendersuper susceptibility to herbicides; or modulate hormonelevels.

The TAC-TIC approach Pfeifer and Grigliatti (1996,1997) proposed a means for controlling pests in twoseminal papers called the TAC-TIC model:“Transposons with Armed Cassettes for Targeted In-sect Control”. In their proposal, an insect is transformedwith a gene, which if activated, can debilitate the in-sect. They propose to use a chemically-induced pro-moter to activate genes that would prevent feeding,mating, or otherwise kill the insect. We have termedsuch chemically-assisted-suicide genes as “kev”(Kevorkian) genes. Pfeifer and Griglilatti postulate thatnot many transgenic pests would be needed if thetransgenes are transmitted in a multicopy transposon;they will quickly disperse in the indigenous popula-tion through mating. They suggested that the farmerscould use their normal methods of pest control duringthe period of transposon dissemination. As no method


478

of control kills all pests, the transposon containing andlacking individuals will be similarly affected, and thetransposons will disseminate among the remainingpests – as they would in a population without pesti-cide. After transposon transmission is complete, theuse of a chemical that turns on the chemically-inducedpromoter replaces the pesticide.

Such an approach could be modified and work with aweed such as ryegrass, while is solely or predominantlyoutcrossing. Indeed, if the proper kev genes and pro-moters can be found, and the transposons available;ryegrass can be easily engineered. Most importantly,safety considerations must be dealt with.

The kev genes can be introduced into a transposon cas-sette and transformed into ryegrass to generate par-tially debilitated plants after chemical induction, orpossibly before induction. Transposon proliferationshould enable reaching high copy numbers of the kevgenes in the progeny of these sown parents. The de-bilitating genes will be dominant and high copynumber. Therefore, most, if not all the progeny willrapidly contain them. This kev containing ryegrasscould then be abundantly sown as pasture to increasethe frequency of kev-containing plants in the field.

The Ac/Ds transposon family, originally found in maize(McClintock 1951), has been shown to be active in allthe heterologous plant systems (approx 15 differentspecies) where it has been introduced (See review:Kunze 1996). Ac is preferentially transposed duringDNA replication, and as a result its copy number canincrease while it transposes (Greenblatt 1984)

Potential kev genes In the initial TAC-TIC concept,no specific genes were suggested for use. Perhaps theunavailability of appropriate genes (and promoters)stifled further development of the concept. A plethoraof potential genes are available “off the shelf” for usein ryegrass. The various types have different uses indifferent systems, but here only those appropriate forryegrass are discussed

One would expect kev constructs to be perfectly use-ful on ryegrass that evolved herbicide resistance byenhancing metabolic pathways that degrade a multi-tude of herbicides. One could envisage inhibiting a jointpromoter region for a plethora of cytochrome P450sto suppress multiple herbicide resistances, allowingcontrol by passé herbicides. Kev constructs might notbe as effective on weed populations that have evolvedherbicide resistance by overproducing the same targetgene product as is targetted. It has been claimed that aglyphosate-resistant ryegrass strain

produces double the normal level of EPSPsynthase(Gruys et al. 1999) as do resistant Lotus corniculatusstrains (Boerboom et al. 1990). A kev gene inhibitingEPSP synthase might be less effective on such biotypes.

Chemisterilant kev genes. Chemically induced genesthat will cause pollen sterility a generation hence havebeen proposed for protecting crop varieties (Masood,1998, Crouch 1998, Oliver et al. 1998), the so calledterminator genes of the popular press. Similar con-structs could be considered for use as kev genes, whendisseminated by transposons. This could easily be usedin ryegrass planted in pastures, with the terminator geneturned on prior to planting wheat (but not in areas usedto propagate seed). This would require always sowingpastures.

Debilitating kev genes Ryegrasses have evolved to behighly competitive species, both with the crop as wellas with other weeds and siblings. A typical ryegrassplant produces thousands of pollen grains to fertilizeone ovule, and thousands of seeds to replace one par-ent. Thus, the competition within the species is quitefierce both to fertilize and during the “self thinning”period when seedlings establish. Some of the genesproposed for use in TM constructs could be used asDT genes in ryegrass. They would be neutral or posi-tive in the pasture phase but unfitness that should berapidly and naturally euthanized from the population.

Non-chemical strategies would be appropriate for thethree interbreeding Lolium spp; L. rigidum, L. perenneand L. multiflorum, which are very important pasturegrasses in rotation with or near grain fields. Therewould be nothing wrong with Lolium in pasture phasecontaining potentially self-lethal kev genes, as long asthe chemical inducer is fool proof, i.e. cannot get turnedon in the pasture phase. An additional or alternativeapproach can be considered with Lolium. One can usekev genes that are neutral or even positive during thepasture phase but decidedly deleterious during the crop-ping phase. Thus, kev genes that cause dwarfing byinternode shortening or non recognition of shade wouldnot matter when cattle represent the only shade as theygraze the Lolium and its competitors. Genes that se-verely affect vertical structural stability such as thosestrongly suppressing lignin formation would also in-crease palatability as well as vastly increase the amountpolymeric carbohydrates that can be digested by rumi-nant animals. Each percent of lignin in tissue preventsthe ruminate degradation of three times more cellu-lose and hemicellulose (Jung et al. 1997). Genes sup-pressing auxin biosynthesis or stimulating cytokininbiosynthesis would suppress the dominance of the


479

shoot apex, stimulating branching (tillering) which isgood in pasture but is in competition with crops. Sup-pression of long term (“secondary”) dormancy couldbe added in addition to preclude remaining in the seedbank for more than one season, insuring that seeds willnot remain overlong in the soil, perhaps even requir-ing annual storage and sowing. All these genes will bebeneficial to the pasture species but will render it lesscompetitive to a wheat crop, even a very dwarf wheatcrop.

Ryegrass has an additional advantage. Whereas fewweeds have been transformed, two species of ryegrasshas been shown to be amenable to straightforwardtransformation techniques (Spangenberg et al. 1995;Ye et al. 1997). Clearly these procedures were eluci-dated because of the importance of ryegrass as a pas-ture species.

Chemically induced promoters The TAC-TIC con-cept for insects requires the use of chemically-inducedpromoters, although no specific examples were listedin the initial papers. A wide variety of promoters areavailable for chemically inducing the expression ofgenes in plants (Gatz 1997, Gatz and Lenk 1998). Someare antibiotics or expensive compounds deemed inap-propriate for agronomic use, but copper salts and etha-nol are among the inexpensive, simple promoters.

Biosafety of DT The major and worst weeds are notwild species. They are truly domesticated, man madecontrivances, just like the crops. In their present evo-lutionary state weeds are not much more competentthan crops to exist in the wild. As with crops, whenryegrass becomes feral, it can at best exist only in hu-man-disturbed (ruderal) ecosystems. Tomes have beenwritten on how weeds evolved during crop domestica-tion, mimicing crop seed and seedling shape foolingthe farmer. They evolved similar harvest date to crops,ensuring further spread (Barrett, 1983), as well asevolving biochemically mimicry in modes of crop re-sistance to herbicides (Gressel 1988). With DTs, evo-lution is reversed. Most weeds would be forced to re-vert back from their highly evolved status inagroecosystems to their status as innocuous wild plants.DT constructs would force the evolution of ryegrassback to being a superior pasture grass, but not a weed.When kev genes are used under a chemical promoter,there should be little danger accruing from the DTtransposons entering wild populations of the weed spe-cies or its introgressing relatives. The DT transposonshould have little negative value if the chemical in-ducer of the gene is not found in the wild.

The above are general theoretical considerations, butit is clearly necessary to perform a risk analysis basedon the particular weed, its place in agro- and naturalecosystems, and whether it has readily introgressingrelations. Horizontal gene transfer among plant spe-cies has been exceedingly rare since higher plants ap-peared on earth. One would consider the same to betrue for transposons. One also must consider the po-tential risk from different kev genes under the controlof different promoters. One could consider construct-ing decision trees, similar for those constructed to es-timate the risk of transgenes introgression from cropto weed (Gressel and Rotteveel 1999) to deal with suchissues.

Clearly, despite paper analyss, experiments should bedesigned to safely evaluate the risk of environmentalhazard (if any) from DT weeds.

Public sector involvement The eventual possibilityof establishing DT containing ryegrass to augment andpartially replace herbicides is different from every dayagricultural technology where instant (or at least sea-sonal) gratification is achieved. The willingness of thepesticide industry to become positively involved isdoubted, because herbicides would be replaced bycheap generic chemicals used as inducers, and farmerprofit would be achieved only years after planting DTryegrass. The biotechnology industry could generatethe DT ryegrass, but again there is long delay untilprofit is achieved. It is just such possibilities wherethe long term benefits for farmer, environment, andsociety have little immediate economic justificationwhere the public sector must exert its authority. Thismay not be easy in the current climate of demandingthat the public sector become increasingly dependenton the private sector for support. If there is any valid-ity to ascertaining whether weeds can be pressured tobecome innocuous without pesticides, it can only beexpected to be performed by public sector research withpublic support.

ACKNOWLEDGMENTS

Useful discussions and debates with many colleagueshelped shape the concepts outlined herein. The abilityto develop these concepts was made possible by theGilbert de Botton professorial chair in plant sciences.


480

REFERENCES

Ackerman, M.J., Hirschfeld, M., Wester, L., Weaver,M., Clack, T., Amasino, R.M. and Sharrock, R.A.(1997). A deletion in the PHYD gene of theArabidopsis Wassilewskija ecotype defines a rolefor phytochrome D in red/far-red light sensing. PlantCell 9, 1317-26.

Al-Kaff, N.S., Covey, S.N., Page, A., Kreike, M., Pinder,R. and Dale, P. (1997). Transcriptional andpostranslational plant gene silencing in response toa pathogen. Science 279, 2113-5.

Anonymous (1994a). Assessment criteria for determin-ing environmental safety of plants with novel traits.Directive Dir. 94-08 (Plant Products Division, Ag-riculture and Agri-Food Canada, Nepean Ontario)(http://www.cfia-acia.agr.ca/English/food/pbo/dir9408 html).

Anonymous (1994b). The biology of Brassica napus L(canola/rapessed). Directive Dir. 94-09 (Plant Prod-ucts Division, Agriculture and Agri-Food Canada,Nepean Ontario) (http://www.cfiaacia.agr.ca/Eng-lish/food/pbo/dir9409.html).

Anonymous (1995a). Determination of environmentalsafety of Pioneer Hi-Bred International Inc’simidazolinone-tolerant Canola line Decision Docu-ment DD05-03. 7pp (Plant Products Division, Ag-riculture and Agri-Food Canada, Nepean Ontario).(http://www.cfiaacia.agr.ca/English.food/pbo/dd9503e.html).

Anonymous (1995b). Determination of environmentalsafety of Monsanto Canada Inc’s Roundup herbi-cide tolerant Brassica napus Canola line GT73. De-cision Document 95-02. 7pp (Plant Products Divi-sion, Agriculture and Agri-Food Canada, NepeanOntario) (http://www.cfiaacia.agr.ca/English.food/pbo/dd9502.html).

Anonymous (1996). Determination of environmentalsafety of AgrEvo Canada Inc’s glufosinate-ammo-nium herbicide tolerant Brassica napus Canola lineDecision Document 96-11. 7pp (Plant Products Di-vision, Agriculture and Agri-Food Canada, NepeanOntario) (http://www.cfiaacia.agr.ca/English.food/pbo/dd9503e.html).

Anonymous (1997). Consensus document on the biol-ogy of Brassica napus L. (oilseed rape). Ser. on theharmonization of regulatory oversight in biotech-nology No. 7. Environ. directorate. 32 pp. (Org. forEcon. Co-op. and Devel., Paris).

Azpiroz, R., Wu, Y., LoCascio, J.C. and Feldmann, K.A.(1998). An Arabidopsis brassinosteroid-dependentmutant is blocked in cell elongation. Plant Cell 10,219-30.

Baker, H.G. (1974). The evolution of weeds. Ann. Rev.Ecol. Sys. 5, 1-24.

Baker, H.G. (1991). The continuing evolution of weeds.Econ. Bot. 45, 445-9.

Barrett, S.C.H. (1983). Crop mimicry in weeds. Econ.Bot. 37, 255-82.

Bergelson, J., Purrington, C.B., Palm, C.J. and Lopez-Guitierrez, J.C. (1996). Costs of resistance – a testusing transgenic Arabidopsis thaliana. Proc. Roy.Soc. Lond. B. Biol. Sci. 262, 1659-63.

Berner, D. K., Kling, J. G. and Singh, B. B. (1995).Striga research and control: A perspective from Af-rica. Plant Dis. 79, 652-60.

Beversdorf, W.D., Weiss-Lerman, J. Erickson, L.R. andSouza-Machado, V. (1980). Transfer of cytoplas-mically-inherited triazine resistance from bird’s rapeto cultivated oilseed rape (Brassica campestris andB. napus). Can. J. Genet. Cytol. 22, 167-72.

Beversdorf, W.D., Hume, D.J. and Donnelly-Vanderloo,J.J. (1988). Agronomic performance of triazine-re-sistant and susceptible reciprocal canola hybrids.Crop Sci. 28, 932-4

Bing, D.B., Downey, R.K. and Rakow, G.F.W. (1996).Assessment of transgene escape from Brassica rapa(B. campestris) into B. nigra or Sinapis arvensis.Plant Breed. 115, 1-4.

Boerboom, C.M., Wyse, D.L. and Somers, D.A. (1990).Mechanism of glyphosate tolerance in birdsfoot tre-foil (Lotus corniculatus). Weed Sci. 38, 463-7.

Brown, J., and Brown, A.P. (1996). Gene transfer be-tween canola (Brassica napus L. and B. campestrisL.) and related weed species. Ann. Appl. Biol. 129,513-22.

Chauvel, B. and Gasquez, J. (1994). Relationships be-tween genetic polymorphism and herbicide resist-ance within Alopecurus myosuroides Huds. Hered-ity 72, 336-44.

Choe, S. B., Dilkes, P., Fugioka, S., Takatsuto, S.,Sukarai, A. and Feldmann, K. A. (1998). The DWF4gene of Arabidopsis encodes a cytochrome P450that mediates multiple 22 d-hydroxylation steps inbrassinosteroid biosynthesis. Plant Cell 10, 231-44.


481

Conner, A.J., and Dale, P.J. (1996). Reconsideration ofpollen dispersal data from field trials of transgenicpotatoes. Theor. Appl. Genet. 92, 505-8.

Cooper, J.I. and Raybould, A.F. (1997). Transgenes forstress tolerance: consequences for weed evolution.Brighton Crop Protect. Conf. - Weeds. pp. 265-72

Crawley, M.J., Hails, R.S., Rees, M., Kohn, D. andBuxton, J. (1993). Ecology of transgenic oilseedrape in natural habitats. Nature 363, 620-3.

Crouch, M.L. (1998). How the terminator terminateshttp://www.bio.indiana.edu/people/terminator/html

Cummins, I., Moss, S., Cole, D.J. and Edwards, R.(1997a). Glutathione transferases in herbicide-re-sistant and herbicide-susceptible black-grass(Alopecurus myosuroidos). Pestic. Sci. 51, 244-50.

Cummins, I., Cole, D.J. and Edwards, R. (1997b). Pu-rification of multiple glutathione transferases in-volved in herbicide detoxification from wheat (Triti-cum aestivum L.) treated with the safenerfenchlorazole-ethyl. Pestic. Biochem. Physiol. 59,35-49.

Cummins, I., Cole, D.J. and Edwards, R. (1999). A rolefor glutathione transferases functioning as glutath-ione peroxidases in resistance to multiple herbicidesin black-grass. Plant J. 18, 285-92.

Daniell, H., Datta, R., Varma, S., Gray, S., Lee and S.B.(1998). Containment of herbicide resistance throughgenetic engineering of the chloroplast genome.Nature Biotech. 16, 345-8.

Darmency, H. (1994a). Genetics of herbicide resistancein weeds and crops. In ‘Herbicide resistance inplants’ eds. S.B, Powles and J.A.M. Holtum, pp.263-97. (Lewis, Boca-Raton).

Darmency, H. (1994b). The impact of hybrids betweengenetically modified crop plants and their relatedspecies: introgression and weediness. Mol. Ecol. 3,37-40.

Devlin, P.F., Patel, S.R. and Whitelam, G.C. (1998).Phytochrome E influences internode elongationand flowering time in Arabidopsis. Plant Cell 10,1479-88.

Duke, S.O. (ed) (1996). Herbicide resistant crops: Ag-ricultural, environmental, economic, regulatory, andtechnical aspects. 420 pp. (CRC Press, Boca Raton).

Gardner, S.N., Gressel, J. and Mangel, M. (1998). Arevolving dose strategy to delay the evolution ofboth quantitative vs. major monogene resistancesto pesticides and drugs. Intern. J. of Pest Manage-ment. 44, 161-80.

Gatz, C. (1997) Chemical control of gene expression.Annual Rev. of Plant Physiol Plant Mol. Biol. 48,89-108.

Gatz, C. and Lenk, I. (1998) Promoters that respond tochemical inducers. Trends in Plant Science 3,352-8.

Gould, F. (1991). The evolutionary potential of croppests. Am. Sci. 79, 496-507.

Greenblatt, I.M. (1984) A chromosome replication pat-tern deduced from pericarp phenotypes resultingfrom movements of the transposable element,Modulator, in maize. Genetics 108, 471-85.

Gressel, J. (1988). Multiple resistances to wheat selec-tive herbicides: New challenges to molecular biol-ogy. Oxford Surv. Plant Molec. Cell. Biol. 5,195-203.

Gressel, J. (1995). Catch 22 - Mutually exclusive strat-egies for delaying/preventing polygenically vs. mo-nogenically inherited resistances. In ‘Options 2000-Eighth International Congress of Pesticide Chem-istry,’ eds N.N. Ragsdale, P.C. Kearney and J.R.Plimmer, pp. 330-349. (American Chemical Soci-ety, Washington, DC).

Gressel, J. (1996). Fewer constraints than proclaimedto the evolution of glyphosate-resistant weeds. Re-sistant Pest Manag. 8, 2-5.

Gressel J. (1997). Can herbicide resistant oilseed rapesfrom commodity shipments potentially introgresswith local Brassica weeds, endangering agriculturein importing countries? Pest Resistance Manag.(Winter, pp. 2-5).

Gressel, J. (1998). Biotechnology of weed control. In‘Biotechnology in Agriculture’. ed. A. Altman, pp.295-325. (Marcel-Dekker, New-York).

Gressel, J. (1999). Tandem Constructs: preventing therise of superweeds. Trends Biotech. 17, 361-6.

Gressel, J. and Ben-Sinai, G. (1985). Low intra-specific competitive fitness in a triazine resistant,nearly nuclear-isogenic lines of Brassica napus.Plant Sci. 38, 29-32.


482

Gressel, J. and Rotteveel. A.W. (1999). Genetic andecological risks from biotechnologically-derivedherbicide-resistant crops: decision trees for risk as-sessment. Plant Breeding Rev. 18, 251-303.

Gressel, J. and Segel. L.A. (1978). The paucity of ge-netic adaptive resistance of plants to herbicides:possible biological reasons and implications. J.Theor. Biol. 75, 349-71.

Gressel, J., Gianessi, L., Darby, C.P., Saari, L. and Seth,A. (1994). Herbicide-resistant weeds: a threat towheat production in India. Discussion Paper PS-94-3, (National Center for Food and AgriculturalPolicy, Washington DC).

Gressel, J., Gardner, S. and Mangel, M. (1996) Preven-tion vs. remediation in resistance management. In‘Molecular Genetics and Evolution of Pesticide Re-sistance,’ ed. T.M. Brown, pp. 169-86. (Amer.Chem. Soc., Washington, DC).

Gruys, K.J., Biest-Taylor, N.A,, Feng, P.C.C., Baerson,S.R., Rodriguez, D.J, You, J., Tran, M., Feng, Y,Kreuger, R.W., Pratley, J.E., Urwin, N.A., andStanton, R.A. (1999). Resistance of glyphosate inannual ryegrass (Lolium rigidum) II. Biochemicaland molecular analyses. WSSA Abstracts 39, 163.

Haas, H., and Streibig, J.C. (1982). Changes in weeddistribution patterns as a result of herbicide use andother agronomic factors. In ‘Herbicide resistancein plants’ eds. H.M. LeBaron and J. Gressel, pp.57-80. (Wiley, New-York).

Hall, L.M., Moss, S.R. and Powles, S.B. (1997). Mecha-nisms of resistance to aryloxyphenoxypropionateherbicides in two resistant biotypes of Alopecurusmyosuroides (black-grass): herbicide metabolism asa cross-resistance mechanism. Pest. Biochem.Physiol. 57, 87-9.

Heap, I. (1988). Resistance to herbicides in annualryegrass (Lolium rigidum), Ph.D. Thesis, Waite Agr.Inst., Univ. of Adelaide.

Heap, I. (1999) http://www.weedscience.com/

Heap, J. and Knight, R. (1982). A population of ryegrasstolerant to the herbicide diclofop methyl. J. Aus-tral. Inst. Agri. Sci. 48, 156-8.

Heap, I. and Knight, R. (1986). The occurrence of her-bicide cross resistance in a population of annualryegrass (Lolium rigidum) resistant to diclofopmethy. Austral. J. Agri. Res. 37, 149-56.

Hedden, P. (1997). The oxidases of gibberellin biosyn-thesis: Their function and mechanism. Physiol.Plant. 101, 709-19.

Hedden, P., and Kamiya, Y. (1997). Gibberellin bio-synthesis: Enzymes, genes and their regulation. Ann.Rev. Plant Physiol. Plant Mol. Biol. 48, 431-60.

Helliwell, C. A., Sheldon, C. C., Olive, M. R., Walker,A.R., Zeevaart, J.A.D. Peacock, W.J. and Dennis,E. S. (1998). Cloning of the Arabidopsis ent-kaureneoxidase gene GA3. Proc. Natl. Acad. Sci. U.S.A.95, 9019-24.

Holm, L, Doll, J., Holm, E., Pancho, J. and Herberger,J. (1997). Worlds Weeds: Natural Histories and Dis-tributions. p. 129 (J. Wiley, New York).

Hyatt, L. A. and Evans, A. S. (1998). Is decreased ger-mination fraction associated with risk of siblingcompetition? OIKOS. 83, 29-35.

Hyde, R.J., Hallahan, D.L. and Bowyer, J.R. (1996).Chlorotoluron metabolism in leaves of resistant andsusceptible: biotypes of the grass weed Alopecurusmyosuroides. Pestic. Sci. 47, 185-90.

Jung, H.G., Mertens, D.R. and Payne, A.J. (1997). Cor-relation of acid detergent lignin and klason ligninwith digestibility of forage dry matter and neutraldetergent fiber. J. Dairy Sci. 80, 1622-8.

Kareiva, P., Parker, I.M. and Pascual, M. (1996). Canwe use experiments and models in predicting theinvasiveness of genetically engineered organisms?Ecology 77, 1670-5.

Keeler, K. H., Turner, C. E. and Bollick, M. R. (1996).Movement of crop transgenes into wild plants. In‘Herbicide resistant crops: Agricultural, environ-mental, economic, regulatory, and technical aspects,’ed. S.O. Duke, pp. 303-30, (CRC Press, BocaRaton).

Kerlan, M.C., Chevre, A.M. and Eber, F. (1993).Interspecific hybrids between a transgenic rapeseed(Brassica napus) and related species: cytogeneticalcharacterization and detection of the transgene. Ge-nome 36, 1099-1106.

Kimber, G. and Feldman, M. (1987). Wild Wheats, AnIntroduction. Special Report 353, 146 pp. (Coll.Agric., Univ. of Missouri, Columbia).

Kimber, G., and Sears, E. R. (1987). Evolution of thegenus Triticum and the origin of cultivated wheat.


483

In ‘Wheat and wheat improvement. AgronomyMonograph’ ed. E.G. Heyne No. 13. pp. 154-64.(ASA-SCCA-SSSA, Madison WI).

Kline, A.D. (1991). We have not yet identified the moralissues in agricultural biotechnology. J. Agr. Environ.Ethics 4, 216-22.

Kling, J. (1996). Could transgenic supercrops one daybreed superweeds? Science 274, 180-1.

Koltunow, A.M., Bicknell, R.A. and Chaudhury, A.M.(1995). Apomixis: Molecular strategies for the gen-eration of genetically-identical seeds without ferti-lization. Plant Physiol. 108, 1345-52.

Kota, M., Daniell, H, Varma, S., Garczynski S.F., Gould,F., Moar, W.J. (1999). Overexpression of the Bacil-lus thuringiensis (Bt) Cry2Aa2 protein inchloroplasts confers resistance to plants against sus-ceptible and Bt-resistant insects. Proc. Nat. Acad.Sci. (USA) 96, 1840-5.

Kunze, R. (1996) The maize transposable element Ac-tivator (Ac). In ‘Transposable elements,’ eds. H.Saedler and A. Gierl, pp. 161-94. (Springer, Ber-lin).

Kusaba, S., Fukumoto, M., Honda, C., Yamaguchi, I.,Sakamoto, T. and Kano-Murakami. Y. (1998). De-creases GA(1) content caused by the overexpressionof OSH1 is accompanied by suppression of GA

20

and oxidase gene expression. Plant Physiol. 117,1179-84.

Landbo, L., Andersen, B, and Jorgensen, R.B. (1996).Natural hybridization between oilseed rape and awild relative: hybrids among seeds from weedy B.campestris. Hereditas 125, 89-91.

Lange, T. (1998). Molecular biology of gibberellin syn-thesis. Planta 204, 409-19

Lefol, E., Danielou, V. and Darmency, H. (1996a). Pre-dicting hybridization between transgenic oilseedrape and wild mustard. Field Crops Res. 45,153-61.

Lefol, E., Fleury, A. and Darmency, H. (1996b). Genedispersal from transgenic crops II. Hybridization be-tween oilseed rape and the wild hoary mustard. Sex.Plant Reprod. 9, 189-96.

Levy, A. (1985). A shattering-resistant mutant of Pa-paver bracteatum Lindl: characterization and inher-itance. Euphytica 34, 811-5.

Lewis, M. W. (1992). ‘Green delusions: An environ-mentalist critique of radical environmentalism,’ 288pp. (Duke Univ. Press, Durham NC).

Ling-Hwa, T. and Morishima, H. (1997). Genetic char-acterization of weedy rices and the inference on theirorigins. Breeding Sci. 47, 153-60.

Lundberg, S., Nilsson, P. and Fagerstrom, T. (1996).Seed dormancy and frequency dependent selectiondue to sib competition: The effect of age specificgene expression. J. Theor. Biol. 183, 9-17.

Lutman, P. J. W. (1993). The occurrence and persist-ence of volunteer rapeseed (Brassica napus). Asp.Appl. Biol. 35, 29-36.

Malik, R.K and Singh, S. (1993). Evolving strategiesfor herbicide use in wheat: resistance and integratedweed management. In Integrated Weed Managementfor Sustainable Agriculture ed. R.K Malik, pp. 225-38. (Ind. Soc. Weed Sci: Hisar).

Malik R. K. and Singh, S. (1995). Littleseed canarygrass(Phalaris minor) resistance to isoproturon in India.Weed Tech. 9, 419-25.

Maltais, B., and Bouchard, C.B. (1978). Une moutardedes oiseaux (Brassica rapa) resistante a l’atrazine.Phytoprotection 59, 117-21.

Masood E (1998). Monsanto set to back down over‘terminator’ gene? Nature 396, 503.

McClintock, B. (1951) Chromosome organization andgenic expression. Cold Spring Harbor Symp. Quant.Biol. 16, 13-47

Metz, P.L.J., Jacobsen, E., Nap, J.P. Pereira, A. andSteikema, W.J. (1997). The impact of biosafety onthe phosphinothricin-tolerance transgene ininterspecific B. rapa × B. napus hybrids and theirsuccessive backcrosses. Theor. Appl. Gen. 95,442-50.

McMullan, P.M., Daun, J.K. and DeClercq, D.R. (1994).Effect of wild mustard (Brassica kaber) competi-tion on yield and quality of triazine-tolerant and tri-azine-susceptible canola (Brassica napus andBrassica rapa). Can. J. Plant Sci. 74, 369-74.

Mikkelsen, T.R., Jensen, J. and Jorgensen, R.B. (1996).Inheritance of oilseed rape (Brassica napus) RAPDmarkers in a backcross with progeny with Brassicacampestris. Theor. Appl. Genet. 92, 492-7.


484

Moody, M.E. and Mack. R. N. (1988). Controlling thespread of plant invasions: the importance of nas-cent foci. J. Appl. Ecol. 25, 1009-21.

Moss, S. (1992). Herbicide resistance in the weedAlopecurus myosuroides (blackgrass): the currentsituation. In ‘Achievements and Developments inCombating Pesticide Resistance,’ ed. I. Denholm,A.L. Devonshire and D.W. Hollomon. pp. 28-47(Elsevier, London).

Nishijima, T., Katsura, N., Koshioka, M., Yamazaki,H., Nakayama, M.,Yamane, H., Yamaguchi, I,Yokota, T., Murofushi, N, Takahashi, N., andNonaka, N. (1998). Effects of gibberellins and gib-berellin-biosynthesis inhibitors on stem elongationand flowering of Raphanus sativus L. J Jap.Soc.Hort. Sci. 67, 325-30.

Oliver, M.J., J.E. Quisenberry, J.E Trolinder, N.L.G.and Keim, D.L. (1998). Control of plant gene ex-pression. U.S. Patent 5,723,765.

Owen, M.D.K. (1997). North American developmentof herbicide resistant crops. Brighton Crop Protect.Conf. – Weeds, pp. 955-63.

Peng, J, Richards, D.E., Hartley, N.M., Murphy, G.P.,Devos, K.M., Flintham, J.E., Beales, J., Fish, L.J.,Worland, A.J., Pelica, F., Sudhakar, D., Christou,P., Snape, J.W., Gale, M.D. and Harberd, N.P.(1999). Green revolution genes encode mutant gib-berellin response modulators. Nature 400, 256-61.

Pfeifer, T.A., and Griglilatti, T.A., (1996). Future per-spectives on insect pest management: Engineeringthe pest, J. Invertebrate Pathol. 67, 109-19.

Pfeifer, T.A., and Grigliatti, T.A. (1997). Genetic pestmanagement strategies: A view of targeted pest in-sect management in the 21st century. Agro FoodIndustry Hi-Tech 8, 29-35.

Powell, M. (1997). Science in sanitary and phytosanitarydispute resolution. Discussion Paper 97-50, 31pp.(Resources for the Future, Washington, DC).

Powles, S.B., and Holtum, J.A.M. (eds.). (1994). ‘Her-bicide resistance in plants: biology and biochemis-try,’ 353pp. (Lewis, Boca Raton).

Powles, S.B., Loraine-Colwill, D.F., Dellow, J.J. andPreston, C. (1998). Evolved resistance to glyphosatein rigid ryegrass (Lolium rigidum) in Australia. WeedSci. 46, 604-7.

Prakash, S. (1988). Introgression of resistance to shat-tering in Brassica napus from Brassica junceathrough non-homologous recombination. Z.Pflanzenzuch. 101, 167-8.

Pratley, J., Baines, P., Eberbach, P., Incerti, M. andBroster, J. (1996). Glyphosate resistance in annualryegrass. pp.122. Proc.NSW Grasslands SocietyConf. Wagga-Wagga.

Preston, C. and Roush, R.T. (1998). Variation in herbi-cide dose rates: risks associated with herbicide re-sistance. In ‘Precision weed management in cropsand pastures’. eds. R.W. Medd and J.E. Pratley. pp.128-37. (Proc. of Workshop Wagga, NSW).

Price, J.S, Hobson, R. N., Nealle, M. A. and Bruce, D.M. (1996). Seed losses in commercial harvestingof oilseed rape. J. Agr. Eng. Res. 65, 183-91.

Rissler, J. and Mellon, M. (1993). Perils amidst thepromise - ecological risks of BD-HRCs in a globalmarket. 92 pp. (Union of Concerned Scientists,Cambridge MA).

Robson, P.R.H., McCormac, A.C., Irvine, A. S. andSmith, H. (1996). Genetic engineering of harvestindex in tobacco through overexpression of a phy-tochrome gene. Nature Biotech. 14, 995-8.

Rubin, E. and Levy, A. (1999). Database against data-base: a computational approach to assess thefequency of horizontal gene transfer. The secondSymposium on Plant Genomics. Maagan, IsraelPF. 3.

Saari, L.L., Cotterman, J.C., and Thill, D.C. (1994).Resistance to acetolactate synthase inhibiting her-bicides. In ‘Herbicide Resistance in Plants’ eds. S.B.Powles and J.A.M. Holtum, pp. 83-139. (Lewis,Boca-Raton).

Schaller, H., Bouvier-Navé, P. and Benveniste, P. (1998).Overexpression of an Arabidopsis cDNA encodinga sterol-C24-methyltransferase in tobacco modifiesthe ratio of 24-methyl cholesterol to sitosterol andis associated with growth reduction. Plant Physiol.118, 461-9.

Scheffler, J.A., Parkinson, R. and Dale, P.J. (1995).Evaluating the effectiveness of isolation distancesfor field plots of oilseed rape (Brassica napus) us-ing a herbicide resistant transgene as a selectablemarker. Plant Breed. 114, 317-21.


485

Seefeldt, S.S., Zemetra, R., Young, F.L. and Jones, S.S.(1998). Production of herbicide-resistant jointedgoatgrass (Aegilops cylindrica) × wheat (Triticumaestivum) hybrids in the field by natural hybridiza-tion. Weed Sci. 46, 632-4.

Simon, U. (1994). “Alko” the first seed-shattering re-sistant cultivar of meadow foxtail Alopecuruspratensis L. Acta Hort. 355, 143-6.

Sindel, B.M. (1997). Outcrossing of transgenes to weedyrelatives. In ‘Commercialisation of BD-HRCs: Risk,benefit and trade considerations Coop’. eds. G.D.McLean, P.M. Waterhouse G. Evans and M.J. Gibbspp. 43-81 (Res. Center for Plant Sci. and Bur. ofResource Sci., Canberra).

Singh, S, Kirkwood, R.C. and Marshall, G. (1998a).Effect of ABT on the activity and rate of degrada-tion of isoproturon in susceptible and resistantbiotypes of Phalaris minor and in wheat. Pestic.Sci. 53, 123-32.

Singh, S, Kirkwood, R. C. and Marshall G. (1998b).Effect of the monooxygenase inhibitor piperonylbutoxide on the herbicidal activity and metabolismof isoproturon in herbicide resistant and suscepti-ble biotypes of Phalaris minor and wheat. Pestic.Biochem. Physiol. 59, 143-53.

Skogsmyr, I. (1994). Gene dispersal from transgenicpotatoes to con-specifics: a field trial. Theor. Appl.Genet. 88, 770-4.

Smeda, R.J., Hasagawa, P.M., Goldsbrough, P.B.,Singh, N.K. and Weller, S.C. (1993). A serine-to-threonine substitution in the triazine herbicide-bind-ing protein in potato cells results in atrazine resist-ance without impairing productivity. Plant Physiol.103, 911-7.

Smith, H., and Whitelam, G.C. (1997). The shade avoid-ance syndrome: Multiple responses mediated bymultiple phytochromes. Plant Cell Environ. 20,840-4.

Smith, M.W., Yamaguchi, S., Ait-Ali, T., and Kamiya,Y. (1998). The first step of gibberellin biosynthesisin pumpkin is catalyzed by at least two copalyl di-phosphate synthases encoded by differentially regu-lated genes. Plant Physiol. 118, 1411-9.

Spangenberg, G., Wang, Z.H. Wu, X., Nagel, J. andPotrykus. I. (1995). Transgenic perennial ryegrass(Lolium perenne) plants from microprojectile

bombardment of embryogenic suspension cells.Plant Sci. 108, 209-17.

Steber, C.M. Cooney, S. E. and McCourt, P. (1998).Isolation of the GA-response mutant sly1 as a sup-pressor of ABII-1 in Arabidopsis thaliana. Genet-ics 149, 509-21.

Streber, W.R. and Willmitzer, L. (1989). Transgenictobacco plants expressing a bacterial detoxifying en-zyme are resistant to 2,4-D. Bio/Technology 7, 811-6.

Tal, A., M.L. Romano, G.R. Stephenson, A.L. Schwannand J.C. Hall (1993). Glutathione conjugation: Adetoxification pathway for fenoxaprop-ethyl in bar-ley, crabgrass, oat and wheat. Pestic. Biochem.Physiol. 46, 190-9.

Thill, D.C. and Mallory-Smith, C.A. (1997). The na-ture and consequence of weed spread in croppingsystems. Weed. Sci. 45, 337-42.

Tilney-Bassett, R.A.E. and Abdel-Wahab. O.A.L.(1979). Maternal effects and plastic inheritance. In‘Maternal Effects in Development’ ed. D.R. Newthand M. Balls, pp. 29-45. Cambridge Univ. Press,Cambridge.

Timmons, A.M., Charters, Y.M., Crawford, J.W., Burn,D., Scott, S.E. Dubbels, S.J., Wilson, N.J.,Robertson, A., O’Brien, E.T., Squire, G.R. andWilkinson, M.J. (1996). Risks from BD-HRCs.Nature 380, 487.

Torii, K.U., McNellis, T.W. and Deng. X.-W. (1998).Functional dissection of Arabidopsis COP1 revealsspecific roles of its three structural modules in lightcontrol of seedling development. EMBO J. 17,5577-87.

U, N. (1935). Genome analysis in Brassica with spe-cial reference to the experimental formation of B.napus and the peculiar mode of fertilization. JapanJ. Bot. 7, 389-452.

Van der Schaar, W., Blanco, C. L. A., Kloosterziel, K.M.,Jansen, R.C., Van Ooijen, J. W. and Koornneef, M.(1997). QTL analysis of seed dormancy inArabidopsis using recombinant inbred lines andMQM mapping. Heredity 79, 190-200.

Vleeshouwers, L. M. (1988). The effect of seed dor-mancy on percentage and rate of germination inPolygonum persicaria and its relevance for crop-weed interaction. Ann. Appl. Biol. 132, 289-99.


486

Webb, S. E., Appleford, N. E. J., Gaskin, P. and Lenton,J. R. (1998). Gibberellins in internodes and ears ofwheat containing different dwarfing alleles. Phyto-chemistry 47, 671-7.

Weeks, D., Anderson, O,. and Blechl, (1993). Rapidproduction of multiple independent lines of fertiletransgenic wheat (Triticum aestivum). Plant Physiol.102, 1077-

Yamaguchi, S., Sun, T.P., Kawaide, H. and Kamiya, Y.(1998). The GA

2 locus of Arabidopsis thaliana en-

codes ent-kaurene synthase of gibberellin biosyn-thesis. Plant Physiol. 116, 1271-8.

Ye, X., Wang, Z.H., Wu, X., Potrykus, I. andSpangenberg, G. (1997). Transgenic Italian ryegrass(Lolium multiflorum) plants from microprojectilebombardment of embryogenic suspension cells.Plant Cell Rep. 16, 379-84.

Young, B.A. (1991). Heritability of resistance to seedshattering in kleingrass. Crop Sci. 31, 1156-8.

Zemetra, R.S., Hansen, J. and Mallory-Smith, C.A.(1998). Potential for gene transfer between wheat(Triticum aestivum) and jointed goatgrass (Aegilopscylindrica). Weed Sci. 46, 313-7.

Documents

NEEDED: NEW PARADIGMS FOR WEED CONTROL