Emerging problems of grass weeds related to graminaceous crops

Most of the calories consumed by humans and their livestock come from graminaceous crops: wheat, rice, maize and sorghum. Each of these (except for maize) is threatened by uncontrollable interbreeding weeds, often feral forms of the crop (Gressel Reference Gressel2005). Graminicides (grass-killing herbicides) that selectively kill these weeds in these related crops have not been developed, but the technology of employing genetically engineered maize with resistance to broad-spectrum herbicides (glyphosate and glufosinate) to control grass weeds has been used extensively with success. However, this approach has not yet been deployed in other graminaceous crops, and neither should it be, unless measures are taken to prevent transgenes from establishing in populations of the interbreeding weeds.

Perennial weeds in crop production

Perennial sedges (Cyperus spp.) and Johnsongrass (Sorghum halepense) have long been in the top 10 worst weeds of the world (Holm et al. Reference Holm, Plucknett, Pancho and Herberger1977). They have underground storage organs and while many herbicides will desiccate the above-ground portions, they readily re-emerge from underground buds. These weeds were controlled to some degree by mechanical cultivation and by a few systemic herbicides such as glyphosate. However, they emerged as a greater problem in minimum tillage agriculture, but could be controlled readily in transgenic glyphosate herbicide-resistant crops. The evolution and rapid dispersal of glyphosate-resistant S. halepense, spread both by seed and rhizomes, first in Argentina and more recently in the USA (Heap Reference Heap2010), is ominous.

The last group of systemic herbicides developed by industry (three decades ago), the inhibitors of acetolactate synthase, is highly prone to the evolution of herbicide resistance by mutation of the herbicide target binding site. There are a few herbicide groups that remain available for use in the European Union, too few to have much value in resistance-delaying rotational strategies.

Root parasitic weeds

Parasitic Orobanche species around the Mediterranean basin into Eastern Europe and Asia, and Striga species in sub-Saharan Africa, threaten all major vegetable and grain crops, respectively, in these regions (Gressel & Valverde Reference Gressel and Valverde2009b). The problems are becoming worse, especially in subsistence agriculture in Africa, where cultivation of the host grain crops is intensifying because of food insecurity. However, the result is a decrease in food security, because monoculture has intensified, facilitating parasitic weed dispersal and establishment. There are no really usable widespread technologies for their control except for the deployment of transgenic or mutant herbicide-resistant crops.

Summary: why have weed problems worsened?

Chemical weed control, now coupled with herbicide-resistant crops, has become so successful that there has been complacency and a belief that there would be fewer problems and less need for weed science. The major under-investment in herbicide target discovery (and thereby herbicide chemo-diversity) coupled with the advent of unscientific regulatory systems are major negative developments with dire implications for the future. The movement away from manual labour in a leading crop (rice) in the developing world and decrease in fossil fuel-intensive mechanical cultivation in the developed world has exacerbated weed problems and accentuates the need for new chemical methods. Problems with parasitic weed are spreading, and simple solutions are lacking. The over-dependence on a single herbicide/transgenic herbicide resistance in many crops and its consequences generates concomitant needs for new and better resistances, coupled with other weed control strategies in more comprehensive, integrated management packages. The present regulatory environment fosters the development of chemical and/or biotech solutions only for major crops, leaving minor (often valuable) crops in a weedy vacuum.

New herbicide mode of action discovery

Most major agrochemical companies adopted a genomics approach to the search for new herbicide targets by using Arabidopsis thaliana. This dicotyledonous species is not a competitive weed and therefore is not a biological model for grass weeds or perennial weeds. The chemicals with herbicidal activity that might have been found would almost inevitably be single-target inhibitors, because of the screening for molecules that inhibit a single essential gene product (Gressel Reference Gressel2008). Such new herbicides, if they had been found, would be prone to the rapid evolution of resistance. The failure of this approach to discovery was owing to companies’ lack of lead compounds that might inhibit the potential targets elucidated by genomics.

There are hundreds of well-authenticated lead phytotoxic compounds known that do not inhibit any known target site and are also without known evolution of resistance. However, these compounds have been mostly ignored by industry (Gressel Reference Gressel2002). They are the phytotoxic chemically characterized allelochemicals produced by plants and microbes (Dayan et al. Reference Dayan, Cantrell and Duke2009). One reason that these natural compounds are not prone to the evolution of resistance is probably due to their being multi-site inhibitors. It would seem appropriate for industry to pay more attention to seeking analogues of these complex molecules that could perhaps then be modified to obtain greater crop selectivity. The chemical/molecular tools are now far more sophisticated than in the past; the moiety that binds and inhibits can be ascertained and coupled with chemical groups that crops can naturally metabolize, or that crops can be engineered to metabolize (see further). Hopefully, some of these new herbicides will be naturally systemic, or innovative chemists will learn how to modify them to be systemic so that perennial and root parasitic weeds can be controlled.

As discussed above, the incentive to discover new herbicide targets has been stifled by the availability of inexpensive glyphosate, together with glyphosate-resistant crops taking such a large market share. The rapidly emerging glyphosate resistance in weeds should engender a desire to retain this herbicide as long as possible, which could justify regulations that assist in this goal, while providing incentives to develop new herbicides. Glyphosate resistance could be delayed by requiring glyphosate to be marketed exclusively as a premix with at least one other herbicide.

Engineered crops resistant to contemporary herbicides

Most transgenic crops planted to date are transgenic herbicide (mainly glyphosate)-resistant crops, leading to the problems described above. These problems are not an argument against the developing and deployment of herbicide-resistant crops. On the contrary, the economic success of these crops, together with their beneficial environmental impact, vindicates the approach but warns against overuse of a single herbicide resistance. Some possible ways in which this diversity might be provided are discussed below. There have been efforts to obtain herbicide-resistant crops by mutagenesis or by using the specific gene conversion processes of chimeroplasty, neither of which are defined as transgenic. If the mutations are single base changes, as has been the case with the imidazolinone-resistant crops, then weeds can continue evolving resistance, or the mutant gene can rapidly transfer by hybridization to taxonomically related weeds, ending the utility, as with rice (Shivrain et al. Reference Shivrain, Burgos, Sales, Mauromoustakos, Gealy, Smith, Black and Jia2009) and wheat (Gandhi et al. Reference Gandhi, Mallory-Smith, Watson, Vales, Zemetra and Riera-Lizarazu2006). There are transgenic ways to prevent such gene flow (see further).

Engineered herbicide resistance to newly developed herbicides

The rational design of new herbicides based on multi-site natural allelochemicals (see above) will allow engineering of metabolic resistance to the same herbicides in order to effect crop selectivity. Technologies are now available to isolate genes encoding herbicide metabolism both from culturable and unculturable microorganisms. When such genes are too inefficient to perform the metabolic task with judicious speed (high K _m, low V _max, where high K _m=a high concentration of substrate that leads to half-maximal velocity (V _max); V _max=the limiting reaction velocity as substrate concentrations become very large), gene shuffling can rapidly evolve poor genes to better ones (Castle et al. Reference Castle, Siehl, Gorton, Patten, Chen, Bertain, Cho, Duck, Wong, Liu and Lassner2004).

Thus, novel herbicide discovery technology and novel biotech could combine to alleviate many problems. Cultivating more weed-competitive crops such as barley instead of poor-quality wheat requires removing wheat subsidies, and would enhance the possibilities of designing better weed control strategies.

Resistance management in herbicide-resistant crops

Engineered resistance to less resistance-prone herbicides would be helpful, but resistance can evolve by more than one mechanism, some totally unpredictable. Thus, it is unwise to believe that resistance cannot evolve, as had been the (manufacturer's) predictions for glyphosate (Bradshaw et al. Reference Bradshaw, Padgette, Kimball and Wells1997).

Resistance management can be as simple as hand rouging uncontrolled individuals before they set seed, to complicated rotations or sequential applications. For example, at present the only herbicide mechanism to control parasitic Striga on maize is to use (mutant) imidazolinone-resistant maize, with the seed coated with herbicide (Kanampiu et al. Reference Kanampiu, Karaya, Burnet and Gressel2009). Striga could evolve resistance, especially in long-season maize. Other weeds are hand weeded in this crop, despite women having better things to do. If the mutant imidazolinone resistance were stacked with transgenic glyphosate resistance, a single mid-season glyphosate treatment would control any emerging imidazolinone-resistant as well as susceptible Striga (systemically through the maize) and other weeds, and during the next season the seed treatment would control any glyphosate-resistant Striga that evolved. A more complex scenario of stacking is described in the following section.

Failsafes from gene flow to related weeds

As discussed earlier, there are feral weedy forms of many important crops including rice, sorghum, sunflower, oilseed rape, oats and beets, as well as weeds that are sufficiently related to wheat, sorghum and oilseed rape to cross with them. Selective herbicides do not control these related weeds in these crops, but they can be controlled by transgenically rendering the crop resistant to the herbicide. Alas, the resistance genes can introgress from crop to related weeds.

Strategies proposed for gene containment are all ‘leaky’, e.g. maternal inheritance of chloroplast traits has not prevented gene flow by pollen. Mitigation strategies have been proposed to prevent the establishment of weed–crop hybrids and their back-crosses with weeds (Gressel Reference Gressel1999) and have been demonstrated to work under field conditions (Rose et al. Reference Rose, Millwood, Moon, Rao, Halfhill, Raymer, Warwick, Al-Ahmad, Gressel and Stewart2009). This requires tandem coupling the gene of choice (e.g. herbicide resistance) with a gene that is positive or neutral for the crop but devastating for weeds (e.g. no seed shattering, dwarfism, no secondary dormancy and no premature bolting). Genes in tandem cannot segregate and remain linked, conferring continuous unfitness on the hybrids that are unable to compete. This may not be sufficient for monoculture paddy rice. It was proposed to generate different transgenic herbicide-resistant varieties for rice, each with mitigator genes that confer hypersensitivity to the next herbicide in the rotation, facilitating selective culling of hybrids (Gressel & Valverde Reference Gressel and Valverde2009a). This, coupled with an unfitness trait such as non-shattering of seeds, should keep the hybrids from establishing while controlling the weedy rice forms along with other weeds.

There have been more scientific papers decrying the perils of gene flow than those dealing with it, and it is unfortunate that the biotech seed industry has not so far been inserting failsafe mechanisms into transgenic crops having interbreeding weeds. Industry opens itself to the criticism of being cavalier in their care of weed management issues, and is open to the future loss of a good technology, as was the case of the mutant-derived imidazolinone-resistant rice, which lasted but 2 years in some locales (Gressel & Valverde Reference Gressel and Valverde2009a). Short-term economic gain seems to overcome long-term economic corporate interests as well as agricultural needs.

Engineering parasitic weed control via interference RNA molecules

Small interference RNA molecules (RNA_i) are known to go from the crop to attacking pests. Plant RNA_i, targeting a nematode-specific gene, has been shown to kill nematodes attacking the plant (Huang et al. Reference Huang, Allen, Davis, Baum and Hussey2006). It was proposed to use a similar strategy with parasitic weeds, as RNA traverses short distances into parasitic weeds. A first attempt targeted gene sequences in primary metabolic pathways whose inhibition should kill a plant. RNA_i was constructed for non-consensus sequences of those genes, so as not to injure the crop, but the parasite was not controlled (De Framond et al. Reference De Framond, Rich, McMillan, Ejeta, Ejeta and Gressel2007). As Orobanche stores and uses mannitol and crops do not, a more recent attempt targeted a gene encoding mannitol metabolism and a modicum of control was achieved (Aly et al. Reference Aly, Cholakh, Joel, Leibman, Steinitz, Zelcer, Naglis, Yarden and Gal-On2009), suggesting that this is the right direction. The RNAi approach has applicability only with parasitic weeds that attach to crops.

Engineering weed-stifling crops

There are many levels where a crop might be induced to better compete with weeds. No single level could work like a herbicide, but stacking many such genes might provide a competitive edge that would either preclude the need for herbicides, or allow herbicides to be used as a coup de grace. As noted above, the mechanisms must not reduce yields appreciably. For example, a more extensive root system that assists in competing for water and minerals might be too expensive. The green revolution demonstrated that taller weeds shading growth halved yields, which justified herbicides for control of weeds in dwarf crops.

Traits such as better fertilizer utilization are becoming valuable and if the crop roots were engineered to quickly and efficiently take up minerals, weeds would be stunted. Over-fertilization led to luxurious weeds. Genes are known and available for phosphate solubilization and uptake, iron uptake, etc., and might be used for this purpose.

A more controversial strategy would be to engineer a single variety of a crop to have many different leaf morphotypes and leaf angles, leaving other traits as isogenic (Gressel Reference Gressel2002). This crop would quickly shade the soil surface, keeping light from weeds. Crop yields would increase due to more field area being used for a longer duration by crop photosynthesis. The problem is not in engineering; genes that would fulfil the function are rapidly becoming known. The problem is with varietal regulatory authorities who require morphological uniformity. Such thinking must be overcome.