SEARCH

SEARCH BY CITATION

Keywords:

  • Chlamydomonas reinhardtii ;
  • experimental evolution;
  • herbicide mixtures;
  • herbicide resistance;
  • resistance management

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • The widespread evolution of resistance to herbicides is a pressing issue in global agriculture. Evolutionary principles and practices are key to the management of this threat to global food security. The application of mixtures of herbicides has been advocated as an anti-resistance strategy, without substantial empirical support for its validation.
  • We evolved experimentally populations of the unicellular green chlorophyte, Chlamydomonas reinhardtii, to minimum inhibitory concentrations (MICs) of single-herbicide modes of action and to pair-wise and three-way mixtures between different herbicides at various total combined doses.
  • Herbicide mixtures were most effective when each component was applied at or close to its MIC. When doses were high, increasing the number of mixture components was also effective in reducing the evolution of resistance. Employing mixtures at low combined doses did not retard resistance evolution, even accelerating the evolution of resistance to some components. At low doses, increasing the number of herbicides in the mixture tended to select for more generalist resistance (cross-resistance).
  • Our results reinforce findings from the antibiotic resistance literature and confirm that herbicide mixtures can be very effective for resistance management, but that mixtures should only be employed where the economic and environmental context permits the applications of high combined doses.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The establishment of herbicides as the major method of weed control in agriculture (Powles & Shaner, 2001) has resulted in the widespread evolution of herbicide resistance (Powles & Yu, 2010). Mixture strategies that expose a population of a single weed species to two or more herbicides with different modes of action have been widely advocated for resistance management (Gressel & Segel, 1990; Friesen et al., 2000; Powles & Shaner, 2001). These ‘resistance management mixture strategies’ must be differentiated from herbicide mixtures that are applied because they provide control of different weed populations within an agricultural field. Similar resistance management strategies have been proposed for the prevention of insecticide (Denholm & Rowland, 1992), fungicide (Brent & Hollomon, 2007) and antibiotic (Brown & Nathwani, 2005) resistance, and the management of resistance to antiretroviral and anti-cancer drugs (Pastan & Gottesman, 1987). Mixture strategies rely on the assumption that mutations conferring resistance to one component of the mixture do not increase fitness in the presence of the second component. Indeed, the most desirable situation arises when there is antagonistic pleiotropy between resistance mechanisms (Gressel, 2002). Where the assumptions of independent resistance are met, resistance to the mixture can only arise via the spontaneous evolution of resistance mechanisms to both (or all) mixture components (Diggle et al., 2003). The likelihood of this occurring decreases with each additional herbicide in the mixture (Wrubel & Gressel, 1994).

Two broad categories of herbicide resistance mechanisms have been documented: target site and nontarget site (Powles & Yu, 2010). Target site resistance arises from the modification or over-expression of the herbicide target enzyme, and results in resistance that is specific to a single mode of action (specialist resistance; Busi & Powles, 2009; Powles & Yu, 2010). Several target site resistance mutations can accumulate in the same individual, leading to multiple resistance (Powles & Yu, 2010). Nontarget site resistance may be based on the enhanced metabolism of the herbicide, reduced herbicide translocation or sequestration away from the active site of the herbicide. Enhanced metabolism, in particular, can confer resistance to multiple modes of action (generalist resistance) and may require multiple mutations (Powles & Yu, 2010). Generalist resistance may be favoured in more complex, multiherbicide environments, and this may compromise the potential efficacy of mixture strategies.

Mathematical models have been used to demonstrate the potential effectiveness of mixtures for herbicide resistance management (Powles et al., 1997; Diggle et al., 2003; Neve, 2008). However, these models focus predominantly on the evolution of target site resistance. Empirical evidence for the efficacy of herbicide mixture strategies is limited and often anecdotal (Beckie, 2006), although some studies have confirmed the benefits of mixtures over other management strategies (Manley et al., 2002; Beckie & Reboud, 2009). Models exploring the effectiveness of mixtures of insecticides or fungicides for the management of resistance provide conflicting evidence for their benefits (Mani, 1985; Denholm & Rowland, 1992; Russell, 2005), as do experimental studies – some supporting mixtures as an effective method of resistance management (McKenzie & Byford, 1993; Prabhaker et al., 1998), others cautioning against their widespread use (Immaraju et al., 1990; Blumel & Gross, 2001; Castle et al., 2007). It is interesting to compare this with the situation in studies of antibiotic resistance, where clinical trials predominantly report mixtures as effective strategies in slowing resistance evolution (Bergstrom et al., 2004; Brown & Nathwani, 2005; Beardmore & Peña-Miller, 2010).

Increased economic and environmental costs are a major obstacle to the adoption of effective herbicide resistance management mixtures in agricultural settings (Hart & Pimentel, 2002). Short-term economic interests favour the use of a single-herbicide mode of action that achieves a high level of control, as it does not require investment in multiple herbicides (Buttel, 2002). From an environmental perspective, herbicide mixtures raise concerns as they increase inputs of pesticides into the environment (Hart & Pimentel, 2002). In response to these problems, there have been calls to use synergistic mixtures of herbicides, whereby the total combined dose of herbicides in the mixture is reduced (Gressel, 1990). The implications of such strategies for resistance evolution are not well understood. In antibiotic resistance, it has been shown that synergistic mixtures can exacerbate resistance evolution, as the appearance of resistance to one of the components leaves a population exposed to an ineffective dose of the other (Hegreness et al., 2008).

Microbial experimental evolution offers the potential to explore conditions under which herbicide mixture strategies may be effective, overcoming time and space limitations associated with empirical studies with higher plants (Elena & Lenski, 2003). Here, we used the unicellular green chlorophyte, Chlamydomonas reinhardtii, as a model organism. Chlamydomonas reinhardtii grows asexually under laboratory conditions (Harris, 2008), is susceptible to a range of commercial herbicides and has been used previously as a model system for the study of the evolution of herbicide resistance (Reboud et al., 2007). The techniques of experimental evolution (Buckling et al., 2009) are easily applicable to C. reinhardtii and have been adopted to explore a variety of questions relating to herbicide resistance evolution and management (Lagator et al., 2012; Vogwill et al., 2012). We evolved experimentally populations of C. reinhardtii with exposure to mixtures of two or three herbicides with different modes of action (atrazine, glyphosate and carbetamide) at a variety of total combined doses, as well as in single exposures to each of the herbicides. The objectives of this study were to investigate the following: mixtures are effective in delaying and/or preventing the evolution of herbicide resistance; the effectiveness of mixtures is dependent on the total combined dose and the number of herbicides; and an increase in the number of herbicides and a reduction in their combined dose increases the likelihood of evolution of generalist resistance.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Founding population

The Chlamydomonas reinhardtii strain used in the experiment is Seger's CC-1690 wild-type mt+ 21gr, obtained from the Chlamydomonas Resource Center (University of Minnesota, St Paul, MN, USA) core collection. Before selection experiments, the strain had been adapted to liquid Bold's medium through continuous exposure for over 700 generations. Two weeks before the start of selection, 20 μl of the founding population (c. 15 000 cells) were spread on an agar plate. After 7 d of growth, a single colony was picked and used to inoculate a Bold's medium liquid culture. This colony was multiplied for 7 d and was used to found all experimentally evolving populations.

Culture conditions

The culture medium used in all experimental conditions was modified Bold's medium (subsequently BM; Harris, 2008). Populations were cultured in disposable borosilicate glass tubes, in 20 ml of BM and maintained in an orbital shaker incubator, at 28°C and 180 rpm, under continuous light exposure, provided by six fluorescent tubes mounted in the incubator lid (Osram L30 W/21-840, cool white; light intensity measured at the location of the tubes was 161 μmol m−2 s−1; Osram, München, Germany). Cultures were propagated every 7 d (see later), during which time the ancestral population growing in the absence of herbicides would have reached stationary phase (3.1 × 107 cells).

Herbicides

We selected for resistance to three herbicides with different modes of action: atrazine (photosystem II inhibitor), glyphosate (inhibitor of aromatic amino acid synthesis) and carbetamide (mitosis inhibitor). Before selection experiments, we determined the minimum inhibitory concentration (MIC) for each herbicide by growing Chlamydomonas cultures for 7 d in BM supplemented with a range of concentrations of each herbicide (sourced from Sigma-Aldrich, 99% purity). MIC was defined as the lowest herbicide dose that prevented detectable population growth over 7 d, being 0.125, 97.5 and 3.0 mg l−1 for atrazine, glyphosate and carbetamide, respectively. We also determined the ‘MIC equivalent’ value when mixtures of herbicides were used (subsequently MICeq), as a combined dose offering the same level of control as MIC of a single herbicide. The value was obtained as an equal proportion of each herbicide in the mixture relative to its MIC, so that the tested values were 55%, 50%, 45% and 40% of each herbicide's MIC. In all pairwise and three-way herbicide mixtures, the growth inhibitory effects of herbicides were synergistic, such that complete growth inhibition was achieved with each herbicide at 45% of its MIC in a two-way, and at 30% of its MIC in a three-way, mixture.

Selection regimes

Chlamydomonas populations were exposed to a total of 19 selection regimes, including continuous exposure to one of the herbicides at its MIC, as well as two-way and three-way mixtures at a range of doses (see Table 1). Six replicate evolving populations were maintained under each of the 19 experimental conditions for a total of 114 evolving populations. Six populations were propagated in the absence of herbicides and were used as controls and as source populations to provide immigration to the evolving populations. Approximately 125 000 cells (estimated by the absorbance at 750 nm) from the founding population provided the initial population for all selection populations. Transfers into fresh medium containing appropriate herbicides were carried out at 7-d intervals. The absorbance at 750 nm (optical density at 750 nm, OD750) of all evolving populations was measured before transfer. At each transfer, 200 μl of the evolving culture was transferred into fresh medium. If the number of cells in 200 μl of culture medium was estimated as < 125 000, the appropriate number of cells from one of the source populations was added to make the total cell number at the transfer c. 125 000. Therefore, the minimum number of cells at the beginning of each cycle was 125 000. For each of the six populations within a selection regime (replicates), the identity of the source population used for immigration was the same throughout the experiment (and different for each population). The experiment was carried out for 15 transfer cycles (15 wk), at which time 125 000 cells from each population were transferred into BM and allowed to grow for 7 d to multiply evolved populations.

Table 1. A summary of the selection regimes for Chlamydomonas reinhardtii
RegimeHerbicide
Atrazine (A)Glyphosate (G)Carbetamide (C)
  1. A1, G1 and C1 represent continuous exposure to one herbicide at the minimum inhibitory concentration (MIC). AG regimes, for example, refer to mixtures of atrazine and glyphosate at the MIC equivalent combined dose (AGeq) and with 50% (AG1), 75% (AG1.5) and 100% (AG2) of the MIC of each mixture component, respectively.

A1MIC (0.125 mg l−1)
G1MIC (97.5 mg l−1)
C1MIC (3.0 mg l−1)
AGeq0.45MIC0.45MIC
AGx0.5MIC0.5MIC
AG1.50.75MIC0.75MIC
AG2MICMIC
ACeq0.45MIC0.45MIC
ACx0.5MIC0.5MIC
AC1.50.75MIC0.75MIC
AC2MICMIC
GCeq0.45MIC0.45MIC
GCx0.5MIC0.5MIC
GC1.50.75MIC0.75MIC
GC2MICMIC
AGCeq0.3MIC0.3MIC0.3MIC
AGCx0.33MIC0.33MIC0.33MIC
AGC1.50.5MIC0.5MIC0.5MIC
AGC20.67MIC0.67MIC0.67MIC

Cross-resistance assays

To test for the selection of generalist cross-resistance, we assayed the growth of evolved populations at the MIC (determined in the manner already described) of four herbicides to which they had no previous exposure. These herbicides were tembotrione, iodosulfuron-methyl-sodium, fluorochloridone and S-metolachlor with MICs of 65, 8, 2.25 and 1.1 mg l−1, respectively. As previously, 125 000 cells of the evolved populations were inoculated into tubes containing one of these herbicides and the population growth (OD750) was measured after 7 d. Each evolved population (experimental replicate line within each regime) was measured twice in this manner.

Statistical analyses

We wished to address three questions: how do the number of herbicides and the combined dose affect the rates of resistance evolution, and how do the rates of resistance evolution compare between dose treatments within herbicide mixture combinations? None of these questions requires a comparison of all treatment groups. Rather than analysing subsets of the dataset to address the different questions, we analysed the entire dataset, using appropriate factors and nesting structures (the details of the factors and nesting structures for each analysis (hypothesis) are given in Supporting Information Notes S1, together with a detailed description of the fitted models and the full ANOVA tables) to separate treatments of interest from other treatments. This approach ensures that all hypotheses are being tested using the same measure of between-observation variability, and maximizes the degrees of freedom (and hence statistical power) associated with this source of variation. By including all observations in each analysis, comparisons between pairs of regime means are based on the same standard error of difference (from the same underlying estimate of the pooled between-observation variability) across different analyses.

Effects of the number of herbicides

To analyse for the effects of herbicide number, we modelled the population size using a linear mixed model within ANOVA (aov function in R 2.15.0; http://www.r-project.org/). We compared the regimes that evolved in single-herbicide environments with those in mixtures at MICeq doses, as these regimes offered the same initial level of population control, and therefore rates of adaptation could meaningfully be compared. Regimes selected in mixtures at MIC, MIC1.5 and MIC2 were not relevant to this question and were nested appropriately within the model (Notes S1, Table S1). The response variable was population size (measured as OD750 at the end of each transfer period). The random (error) term consisted of time (weeks, 15 levels) nested within each regime (19 levels), nested within replicate (population, six levels). The significance of fixed effects was tested with F-tests.

Effects of combined dose

When investigating the effects of combined dose on the dynamics of resistance, we were only interested in regimes with more than one herbicide as the single-herbicide environments had only one dose (Notes S1, Table S2). To analyse the differences in response between regimes with more than one herbicide, we compared treatments based on the differences between the herbicides included in the mixture (four levels – atrazine–glyphosate (AG), atrazine–carbetamide (AC), glyphosate–carbetamide (GC), atrazine–glyphosate–carbetamide (AGC)), and based on the differences between the combined doses (four different levels). The fixed term also included the interaction between these two factors. The response variable and the error term were the same as above, and the significance of fixed effects was tested with F-tests.

Comparing the time of resistance evolution in selection regimes

To analyse the dynamics of resistance evolution in herbicide mixtures and single-herbicide exposure regimes, we modelled OD750 as the response in a further set of linear mixed models using ANOVA in GenStat (13th edn, VSN International, Hemel Hempstead, UK). We separately modelled resistance for regimes associated with each herbicide mixture (AG, AC, GC, AGC), enabling comparison between all four dose regimes for each mixture, as well as the two or three relevant single-herbicide regimes (i.e. A1 and G1 for the AG mixture, and all three single-herbicide conditions for the AGC mixture), following the nesting approach outlined above. An initial term in each model compared the mean for the six or seven regimes of interest with the mean of the remaining treatments, with nested terms accounting for the variation among the treatments not of direct interest. Each model also included the time term, using a series of linear contrasts to identify the time periods over which there were changes in the level of resistance across the six or seven treatments of interest, and the interaction of these contrasts with the treatment terms identified above, to detect where there were differences in the patterns of resistance evolution between conditions. Each linear contrast assessed the slope of the linear regression over four consecutive time points (the first for weeks 1–4, the second for weeks 2–5, and so on), allowing the identification of both the first point and last point at which a significant change in resistance was seen for each condition. To illustrate, as all regimes started with a linear regression slope that was not significantly different from zero (i.e. no change in resistance), the point at which a slope of one regime started to become significantly different from the slopes of other regimes indicated when resistance in that regime had begun to evolve. It was in this way that we analysed the rates or resistance evolution as a comparison between the linear regression slopes at each of 12 contrasts to assess the time at which each population started to exhibit measurable growth. These 12 linear contrasts are not independent, so that they do not provide a complete partitioning of the between-time variation, and some care is needed in the interpretation of significant effects for overlapping periods.

Cross-resistance

We analysed differences in the cross-resistance profile of selected populations by ANOVA with population growth after 7 d (measured as OD750) as the response variable. OD750 values were log transformed following the addition of 0.0055 (this value being one order of magnitude smaller than the lowest recorded OD750 measurement) to account for zero values. Fixed factors were regime (selection regime, 14 levels, as we excluded regimes that did not give rise to any resistant populations) and the novel herbicide environment (four levels), whilst the error term consisted of the source population. We were particularly interested in the regime × herbicide environment interaction, as this represents the differences in the range of novel herbicides to which a population expressed cross-resistance. A subsequent analysis was conducted using Tukey's honestly significant pairwise tests between the mean OD750 of the populations selected in each regime across all four novel herbicide environments. This test treated cross-resistance as a composite measure that included both the number of herbicides to which a population was resistant and the growth rates achieved in each of these herbicides.

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Dynamics of herbicide resistance

Evolution of resistance

Adaptation to the selection regimes occurred in many experimental populations, under various single- and multiple-herbicide conditions. Resistance (defined here as elevated growth rates in herbicide regimes over time) evolved in all populations under exposure to atrazine and glyphosate, and in two of six populations under carbetamide exposure (Fig. 1). Resistance was observed in all populations exposed to mixtures of atrazine and glyphosate at MICeq, MIC and MIC1.5, as well as in four populations at AG2 (Fig. 1a). Populations exposed to a mixture of atrazine and carbetamide evolved resistance in three populations at ACeq and in two populations at AC. Resistance did not evolve in AC regimes at AC1.5 or AC2 (Fig. 1b). Mixtures of glyphosate and carbetamide gave rise to resistance in all populations evolving at GCeq and GC, in two populations at GC1.5, and was not observed at GC2 (Fig. 1c). In the three-herbicide regimes, resistance evolved in all populations at AGCeq and AGC, in two populations evolving at AGC1.5, and was not observed at AGC2 (Fig. 1d).

image

Figure 1. Mean population size at transfer (measured as the optical density at 750 nm (OD750)) during 15 wk of Chlamydomonas reinhardtii adaptation in selection regimes containing: (a) atrazine and/or glyphosate (AG); (b) atrazine and/or carbetamide (AC); (c) glyphosate and/or carbetamide (GC); (d) atrazine, glyphosate and carbetamide (AGC). The identity of selection regimes with different combined doses is indicated in the figure (see also Table 1) with the number of replicates (of six in total) in which resistance evolved shown in parentheses. Bars are ± SE of the mean.

Download figure to PowerPoint

Effects of herbicide number and combined dose

We identified a significant effect of the number of herbicides in the mixtures on the dynamics of resistance evolution (measured as the mean population size at transfer over the 15-wk selection regime), with resistance evolving more slowly with an increase in the herbicide number (F2,90 = 7.85; < 0.001). We also found that an increase in the total combined dose slowed resistance evolution, as the interaction between herbicide mixture and overall herbicide dose was significant (F9,90 = 6.49; < 0.001).

Rates of resistance between regimes

We analysed the rates of resistance evolution as a comparison between the linear regression slopes at each of 12 contrasts, and we report the F-statistic indicating the differences between all six or seven treatments at each time interval (Tables S4–S7). With regard to comparisons between the AG mixtures and continuous exposure to glyphosate or atrazine (Fig. 1a; Table S4), we first observed resistance to the continuous glyphosate regime (between weeks 2 and 5, F5,90 = 16.50; < 0.001). Resistance in populations exposed to AG and AGeq followed (between weeks 6 and 9, F5,90 = 2.84; = 0.015), with the populations exposed to atrazine (A1) and AG1.5 evolving resistance subsequently (between weeks 10 and 13, F5,90 = 2.43; = 0.004). Resistance evolved most slowly in populations selected at AG2 and, as growth occurred only in four populations near the end of the selection procedure, growth rates (slopes of regression lines) for AG2 populations never became significantly different from zero.

In populations exposed to mixtures of atrazine and carbetamide and the individual component herbicides (Fig. 1b, Table S5), the populations exposed to atrazine evolved resistance first (between weeks 10 and 13, F5,90 = 2.34; = 0.048), closely followed by the populations growing at ACeq (between weeks 11 and 14, F5,90 = 5.07; < 0.001). The slopes of regression lines for exposure to carbetamide (C1), AC, AC1.5 and AC2 never become significantly different from zero.

In the GC comparisons (Fig. 1c, Table S6), resistance evolved most rapidly in the populations exposed to glyphosate only (between weeks 2 and 5, F5,90 = 16.93; < 0.001). Populations exposed to GCeq were the second to evolve resistance (between weeks 9 and 12, F5,90 = 5.05; = 0.001), with the populations exposed to GC exhibiting resistance in the subsequent interval (between weeks 10 and 13, F5,90 = 10.12; < 0.001).

In the AGC comparisons (Fig. 1d, Table S7), resistance evolved most rapidly in the G1 regimes (F6,90 = 15.43; < 0.001), followed by the populations selected at AGCeq and in A1 (between weeks 10 and 13, F6,90 = 6.32; < 0.001). Exposure to AGC of the mixture gave rise to resistance in the subsequent interval (between weeks 11 and 14, F6,90 = 6.21; < 0.001).

Patterns of cross-resistance

We identified an overall effect of the regime × herbicide (genotype × environment) interaction (F42,295 = 4.37, < 0.001), indicating the emergence of phenotypes with different cross-resistance profiles (Table 2). Populations evolving at MIC and MICeq of a three-herbicide mixture were significantly more cross-resistant than all other evolved populations, with the exception of the populations evolved in a mixture of atrazine and glyphosate at MIC (Table S3). There were no significant differences in cross-resistance between populations that evolved in any other regimes.

Table 2. Cross-resistance patterns in selected populations of Chlamydomonas reinhardtii
Selection regimeMean population growth as OD750 in novel herbicides
FIST
  1. The growth (presented as the mean optical density at 750 nm (OD750)) of resistant populations from contrasting selection regimes in the presence of the minimum inhibitory concentrations (MICs) of four herbicide modes of action to which the populations had not been exposed to previously (F, fluorochloridone; T, tembotrione; I, iodosulfuron-methyl-sodium; S, S-metolachlor). Zero values indicate that no cross-resistance has evolved to that herbicide mode of action.

A100.02100
G100/000.055
C100/000
A + Geq00.06700.081
A + Gx00.11800.06
A + G1.500/000
A + G200/000
A + Ceq00/000.232
A + Cx00/000.065
G + Ceq00.04800
G + Cx00.14400
G + C1.500.053200
AGCeq0.0730.08400.186
AGCx0.1390.12700.096
AGC1.50000

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The results indicate that herbicide mixtures may be successful at preventing or slowing the evolution of resistance when all components are used at or close to their MIC. The benefits of increasing the number of herbicides in the mixture depend on the combined dose in the mixture: lower combined doses of a three-way mixture led to significant levels of cross-resistance, whereas higher combined doses were successful at preventing adaptation in these regimes.

Lower combined doses of mixtures do not effectively slow resistance evolution

We observed that, regardless of herbicide identity, populations exposed to the two lowest combined doses (MICeq and MIC) evolved resistance more rapidly to the mixture than when they were exposed to the least resistance-prone component of the mixture at MIC (Fig. 1). At lower combined doses, resistance is likely to evolve rapidly to the more resistance-prone component of the mixture, leaving populations exposed to doses that are lower than MIC of the other herbicide(s). Such dynamics allow populations to rapidly circumvent the effectiveness of mixture strategies, as these elevated growth rates enable rapid population growth and this, in turn, may increase mutation supply rates for rarer mutations that increase population fitness in the presence of the second (and further) herbicide(s) (Drlica, 2003; Busi & Powles, 2009). As such, low-dose mixture strategies may facilitate the accumulation of multiple resistance mechanisms in the same individual (Wrubel & Gressel, 1994). Growth assays conducted at the termination of selection procedures indicated that this was probably the case in this study, as all populations that had evolved resistance to mixture regimes were individually resistant to all mixture components at MIC (data not shown). An alternative explanation is that exposure to lower doses selected for a generalist mutation(s) that provided resistance to all herbicides in the mixture (Neve & Powles, 2005). If the number of mutations required for such a mechanism is low, resistance could emerge as rapidly as we have observed. The appearance of such a mechanism would need to be dose specific, as it is not observed at higher combined doses. Our findings are in line with some previous studies (Immaraju et al., 1990; Birch & Shaw, 1997), indicating that the use of equivalent or lowered MICs poses a significant risk for resistance management, as resistance to these mixtures may evolve more rapidly than to single herbicides at high relative doses (Fig. 1).

Mixtures increase the likelihood of cross-resistance

The requirements for successful mixture strategies (Wrubel & Gressel, 1994) may be overcome if evolution proceeds towards a single generalist phenotype instead of requiring resistance to multiple herbicides through independent mutations (multiple resistance; Gressel, 2002). We observed a significant trend towards cross-resistant phenotypes as the number of herbicides in the mixture was increased (Table 2). An increase in the number of herbicides can lead to a generalist optimum, either because the likelihood of acquiring nontarget site resistance is greater than the likelihood of acquiring multiple resistance mutations and/or because the accumulation of fitness costs associated with each independent resistance becomes too large (Poisot et al., 2011). From an applied perspective, the use of more complex mixtures elevates the risk for management, as wider cross-resistance patterns can reduce the number of available herbicides that can be used for subsequent control.

Mixtures in a wider applied setting

Our results provide empirical evidence in support of the efficacy of herbicide mixtures for resistance management. These results reinforce the findings from previous studies (Gressel, 1997; Diggle et al., 2003; Russell, 2005; Beckie, 2006) supporting the use of mixtures at the full dose of each component herbicide. It is important to acknowledge that there may be some relative differences in the risks of evolution of resistance to individual herbicide modes of action between Chlamydomonas and higher plants; mutations rates may differ, as may the mechanisms of resistance selected. However, estimates of the mutation rate per generation suggest that this is comparable with higher plants (Ness et al., 2012). We would also stress that the application of results from this study are informative of general resistance management principles and should not be interpreted as representing risks associated with mixtures of particular herbicide modes of action. We show that reductions in the combined dose may lead to more rapid resistance and, potentially, to cross-resistant phenotypes, questioning the suitability of mixtures for sustainable management unless these can be applied at high doses. The use of reduced doses of antibiotics in mixtures, particularly synergistic mixtures, has been shown previously to elevate the rates of resistance evolution (Hegreness et al., 2008; Michel et al., 2008). As in medical settings, where high doses of multiple antibiotics need to be balanced against toxicity to patient cells (Gluckman et al., 2011), the use of multiple pesticides in agricultural settings must be considered in the light of environmental concerns and economic constraints (Carroll et al., 2011).

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The authors would like to thank Andrew Morgan, John Lynch and Carol Evered for their much appreciated contributions. The project was funded by the Leverhulme Trust.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information
  • Beardmore R, Peña-Miller R. 2010. Antibiotic cycling versus mixing: the difficulty of using mathematical models to definitively quantify their relative merits. Mathematical Biosciences and Engineering 7: 923933.
  • Beckie HJ. 2006. Herbicide-resistant weeds: management tactics and practices. Weed Technology 20: 793814.
  • Beckie HJ, Reboud X. 2009. Selecting for weed resistance: herbicide rotation and mixture. Weed Technology 23: 363370.
  • Bergstrom CT, Lo M, Lipsitch M. 2004. Ecological theory suggests that antimicrobial cycling will not reduce antimicrobial resistance in hospitals. Proceedings of the National Academy of Sciences, USA 101: 1328513290.
  • Birch CPD, Shaw MW. 1997. When can reduced doses and pesticide mixtures delay the build-up of pesticide resistance? A mathematical model. Journal of Applied Ecology 34: 10321042.
  • Blumel S, Gross M. 2001. Effect of pesticide mixtures on the predatory mite Phytoseiulus persimilis A.H. (Acarina, Phytoseiidae) in the laboratory. Journal of Applied Entomology 125: 201205.
  • Brent K, Hollomon D. 2007. Fungicide resistance in crop pathogens: how can it be managed? Brussels, Belgium: Fungicide Resistance Action Committee, 160.
  • Brown EM, Nathwani D. 2005. Antibiotic cycling or rotation: a systematic review of the evidence of efficacy. Journal of Antimicrobial Chemotherapy 55: 69.
  • Buckling A, MacLean C, Brockhurst MA, Colegrave N. 2009. The beagle in a bottle. Nature 475: 824829.
  • Busi R, Powles SB. 2009. Evolution of glyphosate resistance in a Lolium rigidum population by glyphosate selection at sublethal doses. Heredity 103: 318325.
  • Buttel FH. 2002. Economic and social aspects of pest management. In: Pimentel D, ed. Encyclopedia of pest management. Boca Raton, FL, USA: CRC Press, 221223: 1–3.
  • Carroll S, Kinnison MT, Bernatchez L. 2011. In light of evolution: interdisciplinary challenges in food, health, and the environment. Evolutionary Applications 4: 155158.
  • Castle SJ, Toscano NC, Prabhaker N, Henneberry TJ, Palumbo JC. 2007. Field evaluation of different insecticide use strategies as resistance management and control tactics for Bemisia tabaci (Hemiptera: Aleyrodidae). Bulletin of Entomological Research 92: 449460.
  • Denholm I, Rowland M. 1992. Tactics for managing pesticide resistance in arthropods: theory and practice. Annual Review of Entomology 37: 91112.
  • Diggle A, Neve PB, Smith FP. 2003. Herbicides used in combination can reduce the probability of herbicide resistance in finite weed populations. Weed Research 43: 371382.
  • Drlica K. 2003. The mutant selection window and antimicrobial resistance. Journal of Antimicrobial Chemotherapy 52: 1117.
  • Elena SF, Lenski RE. 2003. Evolution experiments with microorganisms: the dynamics and genetic bases of adaptation. Nature Reviews Genetics 4: 457469.
  • Friesen SLJ, Ferguson GM, Hall C. 2000. Management strategies for attenuating herbicide resistance: untoward consequences of their promotion. Crop Protection 19: 891895.
  • Gluckman PD, Low FM, Buklijas T. 2011. How evolutionary principles improve the understanding of human health and disease. Evolutionary Applications 4: 249263.
  • Gressel J. 1990. Synergizing herbicides. Reviews of Weed Science 5: 4982.
  • Gressel J. 1997. Burgeoning resistance requires new strategies. In: De Prado R, Jorrin J, Garcia-Torres L, eds. Weed and crop resistance to herbicides. London, UK: Kluwer Academic, 314.
  • Gressel J. 2002. Molecular biology of weed control. New York, NY, USA: Taylor & Francis Inc.
  • Gressel J, Segel LA. 1990. Modelling the effectiveness of herbicide rotations and mixtures as strategies to delay or preclude resistance. Weed Technology 4: 186198.
  • Harris E. 2008. The Chlamydomonas sourcebook: introduction to Chlamydomonas and its laboratory use‎. Oxford, UK: Academic Press.
  • Hart K, Pimentel D. 2002. Environmental and economic costs of pesticide use. In: Pimentel D, ed. Encyclopedia of pest management. Boca Raton, FL, USA: CRC Press, 237239: 1–3.
  • Hegreness M, Shoresh N, Damian D, Hartl D, Kishony R. 2008. Accelerated evolution of resistance in multidrug environments. Proceedings of the National Academy of Sciences, USA 105: 1397713981.
  • Immaraju JA, Morsei JG, Hobzn RF. 1990. Field evaluation of insecticide rotation and mixtures as strategies for citrus thrips (Thysanoptera: Thripidae) resistance management in California. Journal of Economic Entomology 83: 306314.
  • Lagator M, Vogwill T, Colegrave N, Neve P. 2012. Herbicide cycling has diverse effects on evolution of resistance in Chlamydomonas reinhardtii. Evolutionary Applications: 17524571.
  • Mani GS. 1985. Evolution of resistance in the presence of two insecticides. Genetics 109: 76107783.
  • Manley BS, Wilson HP, Hines TP. 2002. Management programs and crop rotations influence populations of annual grass weeds and yellow nutsedge. Weed Science 50: 112119.
  • McKenzie C, Byford R. 1993. Continuous, alternating, and mixed insecticides affect development of resistance in the horn fly (Diptera: Muscidae). Journal of Economic Entomology 86: 10401048.
  • Michel J-B, Yeh P, Chait R, Moellering RC, Kishony R. 2008. Drug interactions modulate the potential for evolution of resistance. Proceedings of the National Academy of Sciences, USA 105: 1491814923.
  • Ness RW, Morgan AD, Colegrave N, Keightley PD. 2012. Estimate of the spontaneous mutation rate in Chlamydomonas reinhardtii. Genetics 192: 14471454.
  • Neve P. 2008. Simulation modelling to understand the evolution and management of glyphosate resistance in weeds. Pest Management Science 64: 392401.
  • Neve P, Powles S. 2005. Recurrent selection with reduced herbicide rates results in the rapid evolution of herbicide resistance in Lolium rigidum. Theoretical and Applied Genetics 110: 11541166.
  • Pastan I, Gottesman M. 1987. Multiple-drug resistance in human cancer. New England Journal of Medicine 316: 13881393.
  • Poisot T, Bever JD, Nemri A, Thrall PH, Hochberg ME. 2011. A conceptual framework for the evolution of ecological specialisation. Ecology Letters 14: 841851.
  • Powles S, Shaner DL. 2001. Herbicide resistance and world grains. Boca Raton, FL, USA: CRC Press.
  • Powles SB, Preston C, Bryan IB, Jetsum AR. 1997. Herbicide resistance: impact and management. Advances in Agronomy 58: 5793.
  • Powles SB, Yu Q. 2010. Evolution in action: plants resistant to herbicides. Annual Review of Plant Biology 61: 317347.
  • Prabhaker N, Toscano N, Hennenberry T. 1998. Evaluation of insecticide rotations and mixtures as resistance management strategies for Bemisia argentifolii (Homoptera: Aleyrodidae). Journal of Economic Entomology 91: 820826.
  • Reboud X, Majerus N, Gasquez J, Powles S. 2007. Chlamydomonas reinhardtii as a model system for pro-active herbicide resistance evolution research. Biological Journal of the Linnean Society 91: 257266.
  • Russell PE. 2005. A century of fungicide evolution. Journal of Agricultural Science 143: 1125.
  • Vogwill T, Lagator M, Colegrave N, Neve P. 2012. The experimental evolution of herbicide-resistance in Chlamydomonas reinhardtii results in a positive correlation between fitness in the presence and absence of herbicides. Journal of Evolutionary Biology 25: 19551964.
  • Wrubel RP, Gressel J. 1994. Are herbicide mixtures useful for delaying the rapid evolution of resistance? A case study. Weed Technology 8: 635648.

Supporting Information

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Please note: Wiley-Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office.

FilenameFormatSizeDescription
nph12195-sup-0001-NotesS1-TablesS1-S7.pdfapplication/PDF216K

Table S1 Schematic of experimental conditions (regimes) outlining the statistical testing for the effects of the number of herbicides in the mixtures on the dynamics of resistance evolution

Table S2 Schematic of experimental conditions (regimes) outlining the statistical testing for the effects of the combined herbicides in the mixture on the dynamics of resistance evolution

Table S3 Results of Tukey's pairwise t-test investigating the differences in cross-resistance profiles

Table S4 Comparisons of dynamics of resistance evolution for different dose regimes in atrazine and glyphosate (AG) mixtures

Table S5 Comparisons of dynamics of resistance evolution for different dose regimes in atrazine and carbetamide (AC) mixtures

Table S6 Comparisons of dynamics of resistance evolution for different dose regimes in glyphosate and carbetamide (GC) mixtures

Table S7 Comparisons of dynamics of resistance evolution for different dose regimes in atrazine, glyphosate and carbetamide (AGC) mixtures

Notes S1 Detailed description of the statistical analyses including the nesting models and the resulting ANOVA tables.