• fitness costs;
  • resistance management;
  • selection intensity;
  • polygenic resistance;
  • low dose


  1. Top of page
  2. Summary
  3. Introduction
  4. References

Evolved resistance to herbicides is a classic example of ‘evolution in action’. This paper calls for a greater integration of ‘evolutionary-thinking’ into herbicide resistance research. This integration, it is argued, should lead weed scientists to become less focused on simply describing resistance and more driven towards a deeper understanding of the evolutionary forces that underpin resistance evolution. I have attempted in this short paper to initiate a debate into how this might be done. In the first instance, I have highlighted the widespread misunderstanding and mis-measurement by weed scientists of fitness and fitness costs. I have also speculated on the potential for herbicide rotations to exacerbate resistance problems by selecting for generalist (metabolic) resistance. Finally, I have discussed in greater detail the contribution of herbicide rates to resistance evolution and have reported work conducted in Australia which has shown the potential for low herbicide doses to rapidly select for very high levels of resistance in Lolium rigidum. The controversial hypotheses and suggestions put forward need to be tested by field experimentation. They may prove to be unfounded or incorrect, but if they cause us to question and expand the current resistance paradigm they will have been useful.


  1. Top of page
  2. Summary
  3. Introduction
  4. References

It is 50 years since Harper predicted the evolution of resistance to herbicides (Harper, 1956) and today, this foresight is evidenced by confirmed resistance in 183 plant species across six continents (Heap, 2007). The development of resistance to herbicides is a striking example of ‘evolution in action’. As such, the development of strategies to prevent and manage herbicide resistance should be approached by integrating knowledge from population and evolutionary biology into weed science. My aim with this ‘Insights’ paper is to give a personal view of areas where I believe a greater degree of ‘evolutionary-thinking’ could result in significant developments in herbicide resistance research and management. Some of the claims and suggestions put forward are controversial and may prove to be unfounded, and in some cases wrong. My main aim, 50 years since herbicide resistance was first postulated, is to call for a more imaginative approach to some areas of herbicide resistance research and management.

Researching herbicide resistance

In the 5 year period from 2001–2005, 12% (35 of 299) of papers published in Weed Research reported studies on herbicide resistant weeds. Clearly, herbicide resistance is an issue at the forefront of weed science research. Herbicide resistance studies can be grouped into three categories: those that confirm and characterise resistance traits (characterisation); those examining the biological characteristics of resistance, such as mode of inheritance and relative fitness (biological); and those concerned with resistance management (management). Based on this classification, 60% of those articles in Weed Research were concerned with characterisation and only 20% each with biological and management aspects of herbicide resistance.

It is not surprising that much of the earliest herbicide resistance research sought to confirm resistance and to determine the physiological and genetic basis of evolved resistance traits (reviewed in Powles & Holtum, 1994). Understanding the mechanism of resistance and its genetic basis is important, so that the commonalities and differences between resistance cases can be unravelled. However, as we approach 40 years of herbicide resistance research, there would appear to be diminishing returns from published studies that simply report resistance to herbicide A, in species B from location C. Cousens (1999) accused weed scientists of ‘phenomenon fixation’ and a tendency to embark on ‘catalogues of copy-cat studies’ and certainly some of these criticisms apply to herbicide resistance research. Indeed, one may conclude that weed scientists have become overly fixated with describing resistance and less inclined to undertake studies that synthesise this information to more completely understand the population biology of resistance. I make these criticisms knowing that I have published similarly descriptive studies (Neve et al., 2004).

The herbicide resistance paradigm

With few exceptions, one or more of three general mechanisms confers herbicide resistance: an altered herbicide target enzyme, enhanced herbicide metabolism or reduced herbicide translocation. Target-site resistance is the most studied and best understood of these mechanisms and it is generally agreed that this reflects the overall predominance of this mechanism in evolved weed populations. However, it is possible that the relatively simple physiology and genetics of this mechanism and the understandable desire of weed scientists to study the most resistant populations may have resulted in an under representation of other resistance mechanisms in the literature. Notwithstanding this, in general terms, target-site resistance has become the model on which the herbicide resistance paradigm has been constructed. This paradigm states that resistance traits are conferred by modifications to single nuclear genes. These mutations are almost always at least partially dominant and inherited in Mendelian fashion (target-site resistance to the triazine herbicides is a significant departure, being inherited on the chloroplast genome). Given this, evolutionary theory tells us that the rate of resistance evolution will be driven by mutation frequency, the intensity of selection, the dominance and relative fitness of mutations in the presence and absence of the herbicide and by dispersal of resistance alleles within and between weed populations.

Herbicide resistance fitness costs

It is generally expected that mutations conferring resistance to a novel stress will incur a fitness cost in the original stress-free environment (Coustau et al., 2000). It is well established that target-site triazine resistance incurs a substantial fitness cost in the absence of herbicide selection (see Gronwald, 1994). Attempts to detect costs associated with resistance to other herbicide modes of action have been more equivocal. Unfortunately, many of these published studies have misinterpreted, misunderstood or mis-measured fitness costs. As far as is possible, fitness of resistant (R) and susceptible (S) types must be compared in a common genetic background. Studies that compare R and S populations from different locations tell us little about the cost of resistance, because the genetic background is not controlled and differences in growth and other fitness-determining factors may be due to population differences and have nothing to do with the presence or absence of resistance alleles. Fitness costs should also be compared throughout the life cycle (from germination to germination), in different environments, under competitive conditions and in the field where possible. Additionally, where it is known, the mutation(s) endowing resistance should be reported. Failure to satisfy at least some of these criteria leads to the publication of meaningless results. Some recent studies have been more vigilant in addressing these requirements (Purrington & Bergelson, 1997; Roux et al., 2004; Vila-Aiub et al., 2005a,b) and have identified significant costs of resistance. However, it should be acknowledged that fitness costs expressed in laboratory-derived mutants may be quite different from those which evolve in field populations. The failure by weed scientists to integrate evolutionary biology into herbicide resistance research is nowhere more evident than in this area.

Resistance management strategies

Reduction of selection pressure is the main goal of resistance management strategies that rely on herbicide rotations, mixtures and sequences. The efficacy of these strategies has been tested with simulation models (Maxwell et al., 1990; Wrubel & Gressel, 1994) and Diggle et al. (2003) showed that herbicide mixtures, as opposed to rotation, could considerably increase the predicted time to resistance evolution. All of these modelling-based studies operate within the confines of the single gene herbicide resistance paradigm, where discrete single locus mutations confer resistance to a single related class of herbicides.

In an extension to this approach Neve et al. (1999) constructed a three-locus model, where two independent loci conferred target-site resistance to herbicides A and B and a third locus conferred non-target site resistance to both herbicides (cross-resistance). This model predicted that the best way to get the maximum number of control years from the two herbicides was to use herbicide A until resistance evolved and then switch to herbicide B because herbicide rotation resulted in rapid selection of cross-resistance and a lower combined life span for the two herbicides. These results, while untested, raise the interesting possibility that rotation of herbicide modes of action may, in some circumstances, exacerbate resistance problems by selecting for more generalist resistance mechanisms. This result has some foundation in evolutionary ecology which contends that environments that are heterogeneous in time and/or space (cf. herbicide rotation) will lead to the evolution of generalist types (cf. cross-resistance). Conversely, predictable environments (year on year application of the same herbicide) will result in specialist types (see Kassen, 2002). This hypothesis may be contentious, flying in the face of conventional wisdom as it does, but surely it is worthy of some research.

A question of dose

Notwithstanding the reported predominance of single gene Mendelian inheritance of resistance traits, Gressel and co-workers have argued that low herbicide doses favour the evolution of quantitative resistance traits (Gardner et al., 1998; Gressel, 2002). However, many weed scientists have refuted the low dose argument. The question of the impact of dose on resistance evolution reflects another central question in evolutionary biology; does adaptation occur as a result of selection for standing genetic variation (genetic variability that is maintained in the population), or as a result of the selection and fixation of novel mutations (see Hermisson & Pennings, 2005)? The relative importance of these two modes of inheritance has also been keenly discussed in the areas of plant resistance to heavy metals (Macnair, 1991), antibiotic resistance (Lipsitch & Levin, 1997) and insecticide resistance (Roush & Mckenzie, 1987).

Mckenzie (2000) presented a simple theoretical model to assess the potential for monogenic and polygenic resistance responses. This model assumes that within a population there is a normally distributed range of susceptibility to pesticides. When selection acts outside of this range of susceptible phenotypes, the only surviving individuals will be those possessing rare single gene mutations and evolved resistance will be monogenic, resulting in a large change in the resistance phenotype. However, when doses are lower and selection acts within the range of standing genetic variation, polygenic responses will be possible and resistance will evolve by a gradual change in the mean susceptibility of the population.

In Australia, recommended herbicide use rates are lower than elsewhere in the world and many growers further reduce application rates. This has led some to question a role for low use rates in exacerbating the resistance problem in Australia. In 2000, an experiment was initiated to determine the response of a susceptible population of Lolium rigidum Gaudin when herbicide (diclofop-methyl) selection acted within the range of susceptible phenotypes (Neve & Powles, 2005). At 10% of the recommended dose (37.5 g a.i. diclofop-methyl ha−1), approximately one-third of the population survived herbicide application under glasshouse conditions. These survivors were grown to maturity and cross-pollinated and dose–response curves were established for the progeny of this bulk cross. A shift in the dose–response of the population after a single selection indicated an evolutionary response to low dose selection. The population was subjected to two further rounds of recurrent low dose selection and after three generations the GR50 of the most resistant line was 56 times greater than that of the original unselected population (Neve & Powles, 2005). This emphatic result confirms the potential for low herbicide doses to select for standing genetic variation at putative minor genes and for this variation to be rapidly recombined under selection to result in highly resistant phenotypes. Further research confirmed that this resistance was not due to a single gene target-site mechanism. These results cannot be claimed as incontrovertible evidence that low herbicide use rates account for the herbicide resistance phenomenon in Australia. They refer to a single herbicide applied to one population of annual ryegrass and even for this example, the most pressing experiments, those comparing low and high herbicide doses in the field were not conducted.

As I have presented and discussed these results, many have been concerned that they conflict with efforts to optimise herbicide dose for economic and environmental benefits. This is not the case. In fact, these two areas of research are, unwittingly, interested in a very similar question: what, in relation to a species distribution of herbicide susceptibility, is the optimal herbicide dose that can ensure cost-effective control with minimal environmental impact (and least risk of evolved resistance). As Europe and other parts of the world seek to reduce herbicide dose rates, the potential effects that low herbicide use rates have had in other parts of the world should be acknowledged.

Final thoughts

Evolved weed resistance to herbicides is a serious and escalating agronomic problem in many agroecosystems worldwide. The major research effort in this area should be towards the development of economically viable strategies to prevent and manage resistance. Herbicide resistance is a textbook example of rapid plant adaptation to human activity. In my view, more rapid progress towards meeting these management goals can be achieved by a greater integration of evolutionary and population biology with the applied disciplines of weed science and crop-weed agronomy. In this paper, I have provided three examples where greater evolutionary-thinking could inform herbicide resistance research and management. A major challenge will be to test the predictions which arise from ongoing theoretical, modelling and laboratory-based research in the field. This will present significant practical problems and may require investment in large field trials conducted over a number of years. These field trials should begin with susceptible populations (or populations which have been manipulated to contain low frequencies of known target-site and metabolic resistance alleles) of herbicide resistance-prone weeds and the influence and interplay of herbicide use patterns, fitness costs, herbicide doses and other important factors should be investigated in real agronomic situations. These studies should be supplemented by modelling-based research and research that explores the process of herbicide resistance evolution in more experimentally tractable model species (e.g., see Reboud et al., 2007). Some may be sceptical of the specific ideas put forward in this article. This scepticism may ultimately be justified, but herbicide resistance research will be better served if research tests new and challenging hypotheses, rather than continuing to embark on studies whose only aim is to characterise the same resistance mechanisms in new weed species.


  1. Top of page
  2. Summary
  3. Introduction
  4. References
  • Cousens R (1999) Weed science doesn’t have to be a contradiction in terms. In: Proceedings of the Twelfth Australian Weeds Conference, Sep 12–16 (eds AC Bishop, M Boersma & CD Barnes) Hobart, Tasmanian Weed Society, Tasmania, 364371.
  • Coustau C, Chevillon C & Ffrench-Constant R (2000) Resistance to xenobiotics and parasites: can we count the cost? Trends in Ecology and Evolution 15, 378383.
  • Diggle AD, Neve PB & Smith FP (2003) Herbicides used in combination can reduce the probability of herbicide resistance in finite weed populations. Weed Research 43, 371382.
  • Gardner SN, Gressel J & Mangel M (1998) A revolving dose strategy to delay the evolution of both quantitative vs major monogene resistances to pesticides and drugs. International Journal of Pest Management 44, 161180.
  • Gressel J (2002) Molecular Biology of Weed Control. Taylor & Francis, London.
  • Gronwald JW (1994) Resistance to photosystem II inhibiting herbicides. In: Herbicide Resistance in Plants: Biology and Biochemistry (eds SBPowles & JAMHoltum), 2760. CRC Press, Boca Raton, FL, USA.
  • Harper JL (1956) The evolution of weeds in relation to resistance to herbicides. Proceedings of the 3rd British Weed Control Conference, Nov. 5–8, British Weed Control Council, Farnham, UK. 179188.
  • Heap I (2007) The International Survey of Herbicide Resistant Weeds. (9 February 2007).
  • Hermisson J & Pennings PS (2005) Soft sweeps: Molecular population genetics of adaptation from standing genetic variation. Genetics 169, 23352352.
  • Kassen R (2002) The experimental evolution of specialists, generalists and the maintenance of diversity. Journal of Evolutionary Biology 15, 173190.
  • Lipsitch M & Levin BR (1997) The population dynamics of antimicrobial chemotherapy. Antimicrobial Agents and Chemotherapy 41, 363373.
  • Macnair MR (1991) Why the evolution of resistance to anthropogenic toxins normally involves major gene changes: the limits to natural selection. Genetica 84, 213219.
  • Maxwell BD, Roush ML & Radosevich SR (1990) Predicting the evolution and dynamics of herbicide resistance in weed populations. Weed Technology 4, 213.
  • Mckenzie JA (2000) The character or the variation: the genetic analysis of the insecticide-resistance phenotype. Bulletin of Entomological Research 90, 37.
  • Neve P & Powles SB (2005) Recurrent selection with reduced herbicide rates results in the rapid evolution of herbicide resistance in Lolium rigidum. Theoretical and Applied Genetics 110, 11541166.
  • Neve P, Diggle A & Powles S (1999) Modelling the population genetics and dynamics of multiple- and cross-resistance in Lolium rigidum. In: Proceedings of the Twelfth Australian Weeds Conference, Sep 12–16 (eds AC Bishop, M Boersma & CD Barnes) Hobart, Tasmanian Weed Society, Tasmania, 450453.
  • Neve P, Sadler J & Powles SB (2004) Multiple herbicide resistance in a glyphosate resistant rigid ryegrass (Lolium rigidum) population. Weed Science 52, 920928.
  • Powles SB & Holtum JAM (1994) Herbicide Resistance in Plants: Biology and Biochemistry. CRC Press, Boca Raton, FL, USA.
  • Purrington CB & Bergelson J (1997) Fitness consequences of genetically engineered herbicide and antibiotic resistance in Arabidopsis thaliana. Genetics 145, 801814.
  • Reboud X, Majerus N, Gasquez J & Powles SB (2007) Chlamydomonas reinhardtii as a model system for pro-active herbicide resistance evolution research. Biological Journal of the Linnean Society, 91, 257266.
  • Roush RT & Mckenzie JA (1987) Ecological genetics of insecticide and acaricide resistance. Annual Review of Entomology 32, 361380.
  • Roux F, Gasquez J & Reboud X (2004) The dominance of the herbicide resistance cost in several Arabidopsis thaliana mutant lines. Genetics 166, 449460.
  • Vila-Aiub MM, Neve P, Steadman KJ & Powles SB (2005a) Ecological performance of multiple herbicide-resistant Lolium rigidum population: dynamics of seed germination and seedling emergence of resistant and susceptible phenotypes. Journal of Applied Ecology 42, 288298.
  • Vila-Aiub MM, Neve P & Powles SB (2005b) Resistance cost of a cytochrome P450 herbicide metabolism mechanism but not an ACCase target-site mutation in a multiple resistant Lolium rigidum population. New Phytologist 167, 787796.
  • Wrubel RP & Gressel J (1994) Are herbicide mixtures useful for delaying the rapid evolution of resistance - a case study. Weed Technology 8, 635648.