Control by glyphosate and its alternatives of glyphosate-susceptible and glyphosate-resistant Echinochloa colona in the fallow phase of crop rotations in subtropical Australia
Abstract
Echinochloa colona is the most common grass weed of summer fallows in the grain-cropping systems of the subtropical region of Australia. Glyphosate is the most commonly used herbicide for summer grass control in fallows in this region. The world's first population of glyphosate-resistant E. colona was confirmed in Australia in 2007 and, since then, >70 populations have been confirmed to be resistant in the subtropical region. The efficacy of alternative herbicides on glyphosate-susceptible populations was evaluated in three field experiments and on both glyphosate-susceptible and glyphosate-resistant populations in two pot experiments. The treatments were knockdown and pre-emergence herbicides that were applied as a single application (alone or in a mixture) or as part of a sequential application to weeds at different growth stages. Glyphosate at 720 g ai ha−1 provided good control of small glyphosate-susceptible plants (pre- to early tillering), but was not always effective on larger susceptible plants. Paraquat was effective and the most reliable when applied at 500 g ai ha−1 on small plants, irrespective of the glyphosate resistance status. The sequential application of glyphosate followed by paraquat provided 96–100% control across all experiments, irrespective of the growth stage, and the addition of metolachlor and metolachlor + atrazine to glyphosate or paraquat significantly reduced subsequent emergence. Herbicide treatments have been identified that provide excellent control of small E. colona plants, irrespective of their glyphosate resistance status. These tactics of knockdown herbicides, sequential applications and pre-emergence herbicides should be incorporated into an integrated weed management strategy in order to greatly improve E. colona control, reduce seed production by the sprayed survivors and to minimize the risk of the further development of glyphosate resistance.
Echinochloa colona L. (Link), known as awnless barnyard grass, is a widespread and major weed in summer crops and fallows in the subtropical region of north-eastern Australia (Osten et al. 2007). In fields that were surveyed from 1997 to 2000, Echinochloa spp. were recorded on >70% of all fields (Rew et al. 2005). Furthermore, in a survey of the dryland cotton-cropping systems of subtropical Australia, Echinochloa spp. were by far the most common species in summer fallows and also were major weeds in the summer components of the rotation (Walker et al. 2005). In these surveys, the Echinochloa spp. were not differentiated, but were most likely to be predominantly E. colona and possibly some E. crus-galli (Keenan et al. 2008).
In the subtropical region of Australia, E. colona plants emerge in flushes in the spring and summer (McGillion & Storrie 2006) and thus infest summer crops and fallows. One uncontrolled plant can produce ≤42,000 seeds, many of which are capable of emergence the following spring or summer (McGillion & Storrie 2006). Thus, throughout the fallow, effective control of E. colona is required to stop seed set on the sprayed survivors in order to substantially reduce seedbank replenishment.
Fallowing is a common practice in the subtropical grain region of Australia in order to increase the amount of soil-stored moisture for improved yields of the following crop and to retain stubble in order to reduce the risk of erosion (Wicks et al. 2000). With the introduction of reduced or zero-tillage farming systems, weeds no longer have been controlled with tillage and thus the reliance on herbicides, particularly glyphosate, has increased (Osten et al. 2007).
In a 2001 survey of the subtropical grain region of Australia, the majority of farmers regularly used glyphosate for weed control in summer fallows, whereas less than one-third of them used residual herbicides and about one-third of growers in the region regularly practiced conventional tillage (Osten et al. 2007). Similarly in cotton systems, glyphosate applied alone or in a mixture accounted for ≤94% of the herbicide treatments that were applied in fallow (Walker et al. 2005).
Glyphosate resistance evolves following the frequent and intensive use of glyphosate and where there is little or no diversity in weed management practices (Beckie 2011). Recent risk assessments identified E. colona as highly likely to evolve glyphosate resistance (Walker et al. 2002; Werth et al. 2011), particularly after ∼ ≥15 glyphosate-based summer fallows (Thornby & Walker 2009).
In 2007, glyphosate resistance was first confirmed in a population of E. colona in the subtropical grain region of Australia (Preston 2010). The population was from a field that had been predominantly in a summer fallow and wheat rotation (Cook et al. 2008). Since then, 71 more populations have been identified in this region, mostly in chemical-based fallows (Preston 2013), and one population has been identified in the tropical Ord River region of north-western Australia (Gaines et al. 2012). Glyphosate resistance in E. colona also has been confirmed in Argentina and the USA (Heap 2012).
In order to reduce the risk of glyphosate resistance and to control glyphosate-resistant populations effectively, a wide range of weed management options, aimed at 100% seed-set control in the summer fallow, is required (Storrie et al. 2008). These options should include non-glyphosate knockdown herbicides that are combined with residual herbicides (Beckie 2011) and the double-knock tactic (Storrie et al. 2008). The double-knock tactic refers to the sequential application of two different weed-control methods, applied in such a way that the second option controls the survivors of the first application (McGillion & Storrie 2006), which is commonly glyphosate followed by paraquat at 1–7 days later (Beckie 2011).
To address the effective fallow management of E. colona in the cropping region of subtropical Australia, the options for improving the efficacy of glyphosate (weed growth stage and rate) were investigated and the potential glyphosate alternatives (non-selective postemergence, sequential application and pre-emergence herbicides) for the control of glyphosate-susceptible and glyphosate-resistant E. colona populations were identified.
Materials and Methods
Experiments 1, 2 and 3 (field)
Sites
Three field experiments were conducted in fallow fields that were naturally infested with glyphosate-susceptible E. colona. The experiments were located near Pittsworth (27°43′S, 151°35′E) (Experiment 1), Dalby (27°8′S, 151°16′E) (Experiment 2) and Cecil Plains (27°28′S, 151°13′E) (Experiment 3) on the Darling Downs in north-eastern Australia. At each field site, the soil type was a self-mulching black Vertosol. Experiments 1 and 2 commenced in December 2005 and November 2006, respectively, before the first case of glyphosate-resistant E. colona was confirmed in 2007 (Preston 2013), while Experiment 3 commenced in October 2008.
The initial E. colona densities were moderate for experiments 1 and 2, averaging 41 and 133 plants per m2, respectively, whereas the density for Experiment 3 was higher, with an average of 680 plants per m2. At each site, the growth stage of E. colona ranged from pre- to late tillering (Table 1) at application Time 1. Experiment 1 had a greater proportion of pretillering plants (44%), compared with experiments 2 and 3 (29% and 20%, respectively) that had predominantly early-tillering plants (70% and 79%, respectively). At application Time 2, the pretillering plants had advanced to the early-tillering stage, while similarly the early-tillering plants were late-tillering and the late-tillering plants were flowering.
| Experiment | Pretillering (2–5 leaves) | Early tillering (1–3 tillers) | Mid-late tillering (>3 tillers) |
|---|---|---|---|
| Field experiments (% of average plant density in the untreated plots)† | |||
| 1 | 44 | 54 | 2 |
| 2 | 29 | 71 | 0 |
| 3 | 20 | 79 | 1 |
| Pot experiments (% of all plants)‡ | |||
| 4 | 34 | 64 | 2 |
| 5 | 11 | 84 | 5 |
- † Average density was 41, 133 and 680 plants per m2 in experiments 1, 2 and 3, respectively; ‡ the total number was160 and 240 plants in experiments 4 and 5, respectively.
Herbicide treatments and applications
The herbicides that were used in each experiment were: glyphosate or paraquat (Syngenta Australia, Sydney, Australia) applied alone at different rates and timing or mixed with a pre-emergence herbicide (imazapic [Crop Care Australasia, Brisbane, Australia], atrazine [Nufarm Australia, Melbourne, Australia], metolachlor [Syngenta Australia] or metolachlor + atrazine [Syngenta Australia]), the sequential application of glyphosate followed by paraquat and pre-emergence herbicides mixed with the first or second component of the sequential application (Tables 2 and 3).
| Herbicide | Rate (g ai ha–1) | Biomass reduction (% of untreated plots) | |||
|---|---|---|---|---|---|
| Experiment 1 | Experiment 2 | ||||
| Time 1 | Time 2 | Time 1 | Time 2 | ||
| Glyphosate | 360 | 96.8ab† | 70.8bc‡ | 96.7ab‡ | 26.2f‡ |
| Glyphosate | 540 | 97.6a | 97.3a | 98.8a | 57.3de‡ |
| Glyphosate | 720 | 98.1a | 94.8ab | 99.4a | 83.8c‡ |
| Paraquat | 300 | 88.7ab | 14.1e‡ | 90.4bc‡ | 43.3e‡ |
| Paraquat | 400 | 96.4ab | 22.5de‡ | 95.0ab‡ | 60.1d‡ |
| Paraquat | 500 | 96.1ab | 52.8cd‡ | 98.0a | 80.2c‡ |
| Paraquat + imazapic | 500 + 48 | 98.5a | – | 96.7ab‡ | – |
| Glyphosate + imazapic | 720 + 48 | 99.6a | – | 98.9a | – |
| Glyphosate fb paraquat | 360 fb 300 | 100.0§ | – | 100.0§ | – |
| Glyphosate fb paraquat | 360 fb 500 | – | – | 100.0§ | – |
| Glyphosate fb paraquat | 540 fb 400/300¶ | 100.0§ | – | 100.0§ | – |
| Glyphosate fb paraquat | 540 fb 500 | – | – | 100.0§ | – |
| Glyphosate fb paraquat | 720 fb 300 | – | – | 100.0§ | – |
| Glyphosate fb paraquat | 720 fb 500 | 100.0§ | – | 100.0§ | – |
- † Back-transformed means are presented and lettering assigned to denote significance based on Fisher's Protected Least Significant Difference tests within each experiment; ‡ the upper level of the 97.5% confidence interval is <100%; § the treatment was excluded from the analysis; ¶ a paraquat rate of 400 g ai ha−1 was used in Experiment 1, while 300 g ai ha−1 was used in Experiment 2. fb, followed by the sequential application 7 days later.
| Herbicide | Rate (g ai ha–1) | Biomass reduction (% of the untreated plots) | New emergence (seedlings per m2) |
|---|---|---|---|
| Glyphosate (1) | 720 | 99.5a† | 18.3bcd |
| Glyphosate (2) | 720 | 52.1b‡ | 71.3a |
| Glyphosate fb paraquat | 450 fb 400 | 99.0§ | – |
| Glyphosate fb paraquat | 450 fb 600 | 96.0a | – |
| Glyphosate fb paraquat | 720 fb 400 | 100.0§ | 26.8abc |
| Glyphosate fb paraquat | 720 fb 600 | 99.4§ | 43.2ab |
| Glyphosate + imazapic (1) | 720 + 48 | 99.4§ | 15.5bcd |
| Paraquat + imazapic (1) | 500 + 48 | 97.7a | 15.1bcd |
| Glyphosate + atrazine (1) | 720 + 2250 | 65.3b‡ | 13.3bcde |
| Paraquat + atrazine (1) | 400 + 2250 | 99.2a | 29.9abc |
| Glyphosate + atrazine fb paraquat | 720 + 2250 fb 400 | 94.8a | 5.8def |
| Glyphosate + metolachlor fb paraquat | 720 + 1920 fb 400 | 100.0§ | 1.1gh |
| Glyphosate + imazapic fb paraquat | 720 + 48 fb 400 | 100.0§ | 9.9cde |
| Glyphosate + (metolachlor + atrazine) fb paraquat | 720 + (928 + 1184) fb 400 | 96.4a | 0.0h |
| Glyphosate fb paraquat + atrazine | 720 fb 400 + 2250 | 100.0§ | 6.9de |
| Glyphosate fb paraquat + metolachlor | 720 fb 400 + 1920 | 98.8§ | 1.4fgh |
| Glyphosate fb paraquat + imazapic | 720 fb 400 + 48 | 99.1§ | 6.4def |
| Glyphosate fb paraquat + (metolachlor + atrazine) | 720 fb 400 + (928 + 1184) | 99.2§ | 3.6efg |
- † Back-transformed means are presented and lettering assigned to denote significance based on Fisher's Protected Least Significant Difference tests within each experiment; ‡ the upper level of the 97.5% confidence interval is <100%; § the treatment was excluded from the analysis. (1), Time 1; (2), Time 2; fb, followed by the sequential application 7 days later.
Glyphosate and paraquat were applied at two times, first (Time 1) when the majority of plants had three or less tillers and second (Time 2) when the majority of plants had more than three tillers. The Time 2 treatments were applied 13 days after the Time 1 treatments in Experiment 1 and 7 days afterwards in experiments 2 and 3. The sequential applications had the first component applied at Time 1, with the second component applied 7 days later.
The herbicides were applied by using a tractor-mounted boom at a pressure of two bars and using TT110-01 nozzles (Teejet Technologies, Springfield, IL, USA). The glyphosate and glyphosate mixtures were applied at 60, 73 and 75 L ha−1, using the glyphosate isopropylamine salt 450 g L−1 formulation (Nufarm Australia), and the paraquat and paraquat mixtures were applied at 150, 100 and 105 L ha−1 in experiments 1, 2 and 3, respectively.
The level of rainfall was adequate for weed growth and the incorporation of pre-emergence herbicides in each experiment: 45 and 133 mm in December and January, respectively (Experiment 1), 23, 58 and 123 mm in November, December and January, respectively (Experiment 2), and 57 and 79 mm in October and November, respectively (Experiment 3).
Design and measurements
The experimental design was a randomized complete block with three replications, including two (experiments 2 and 3) or three (Experiment 1) untreated plots in each replication. The plots were 20 m × 2 m for experiments 1 and 2 and 10 m × 2 m for Experiment 3, with the smaller plots being used due to the greater density of E. colona in this experiment.
The growth stages of the populations at herbicide application were measured at Time 1 in the untreated plots in four quadrats (0.5 m2). The level of weed control across the whole treated plot was assessed by using a scale of 0–100% relative to the untreated plots, ranging from 0 = “no visual plant injury” to 100 = “complete plant death”, in increments of five and then in increments of one to between 95 and 100, relative to the untreated plots. The plots were rated 12 days after the Time 1 treatments and 28 days after the Time 2 treatments for Experiment 1, 30 days after the Time 1 treatments and 23 days after the Time 2 treatments in Experiment 2 and 14 days after the Time 1 treatments and 7 days after the Time 2 treatments in Experiment 3.
In order to assess the level of pre-emergence weed control, the newly emerged seedlings were counted in five randomly placed quadrats (0.5 m2) in each plot. The level of emergence was counted in the plots treated with imazapic and in the plots where the imazapic mix partner (glyphosate or paraquat) was applied alone at 40 or 33 days after the Time 1 treatment in experiments 1 and 2, respectively. In Experiment 3, the level of emergence was counted at 30 days after the Time 1 treatment in the plots that were treated with glyphosate and sequential applications with or without a pre-emergence herbicide.
Experiments 4 and 5 (pot)
Site
Two pot experiments, set up outdoors at Toowoomba (27°32′S, 151°56′E) on the Darling Downs of Australia, were started in January 2009 (Experiment 4) and January 2010 (Experiment 5).
Seed for the susceptible population was collected from the same field that was used for Experiment 2, where >97% control was achieved with glyphosate. Seed for the resistant population was collected from near Bellata (29°55'S, 149°47'E) in northern New South Wales, which had been confirmed as having a fivefold resistance to glyphosate in 2007 (Preston 2013).
The E. colona seeds were pregerminated in Petri dishes in an incubator at 28°C with a 12 h day/night cycle. Five days later, two glyphosate-susceptible or glyphosate-resistant seedlings were planted into each pot (13 cm diameter), filled with potting mix (Searles Premium Potting Mix; JC & AT Searle, Kilcoy, Queensland). The plants were watered regularly and a liquid fertilizer (Aquasol; Yates Australia, Sydney, Australia) was added to ensure healthy seedling growth.
Herbicide applications
The herbicide treatments consisted of glyphosate or paraquat applied alone at different rates and timing or as a sequential application of glyphosate followed by paraquat at different rates (Table 4). The herbicides were applied by using a tractor-mounted boom to deliver 75 L ha−1 for the glyphosate treatments and 105 L ha−1 for the paraquat treatments, at a pressure of two bars using the TT110-01 nozzles. Following the treatment, all the pots were returned to the shade house and randomized within their replicate.
| Herbicide | Rate (g ai ha–1) | Biomass reduction (% of untreated plots) | |||
|---|---|---|---|---|---|
| Experiment 4 | Experiment 5 | ||||
| GS | GR | GS | GR | ||
| Glyphosate (1) | 158 | 83.5 | 43.9† | 91.0‡ | 32.5bc†§ |
| Glyphosate (2) | 158 | 99.4‡ | 59.5† | 58.5a† | 14.5c† |
| Glyphosate (1) | 315 | 100.0‡ | 67.0 | 100.0‡ | 49.5ab† |
| Glyphosate (2) | 315 | 98.5‡ | 44.5ab† | ||
| Paraquat (1) | 250 | 100.0‡ | 100.0‡ | ||
| Paraquat (2) | 250 | 98.5‡ | 78.5 | 100.0‡ | 100.0‡ |
| Paraquat (1) | 500 | 100.0‡ | 100.0‡ | ||
| Paraquat (2) | 500 | 100.0‡ | 100.0‡ | ||
| Glyphosate fb paraquat | 158 fb 250 | 100.0‡ | 100.0‡ | 100.0‡ | 100.0‡ |
| Glyphosate fb paraquat | 158 fb 500 | 100.0‡ | 100.0‡ | 100.0‡ | 100.0‡ |
| Glyphosate fb paraquat | 315 fb 250 | 100.0‡ | 100.0‡ | 100.0‡ | 100.0‡ |
| Glyphosate fb paraquat | 315 fb 500 | 100.0‡ | 100.0‡ | 100.0‡ | 100.0‡ |
- † Upper level of the 97.5% confidence interval is <100%; ‡ the treatment was excluded from the analysis; § lettering is assigned to denote significance based on Fisher's Protected Least Significant Difference tests within each experiment. (1), Time 1; (2), Time 2; fb, followed by the sequential application 7 days later.
At the Time 1 herbicide application, the plants were predominantly at the early-tillering stage in both experiments (Table 1). However, there was a lower proportion of pretillering plants in Experiment 5 than in Experiment 4. The growth rates between the glyphosate-resistant and the glyphosate-susceptible populations appeared to be uniform.
The Time 1 treatments were applied 22 or 24 days after transplanting the seedlings in experiments 4 and 5, respectively, with the Time 2 treatments applied 7 days later. For the sequential application treatments, paraquat was applied 7 days after the glyphosate application.
Design and measurements
The treatment structure was a factorial combination of population × herbicide treatments with five replications laid out as randomized complete blocks. The two populations (glyphosate-susceptible and glyphosate-resistant) were treated with eight (Experiment 4) or 12 (Experiment 5) herbicide treatments, with eight common herbicide treatments across both experiments.
The growth stage at the time of the herbicide application was recorded for both plants in each pot. Visual biomass reduction ratings, as used for experiments 1–3, were recorded for all the treatments at 42 days after the second knock was applied in the sequential application treatments.
Statistical methods
The level of biomass reduction was analyzed for all the experiments and emergence counts were carried out for experiments 1, 2 and 3. Experiments 1 and 2 were analyzed together, while experiments 3, 4 and 5 were analyzed individually.
The data were analyzed by using linear mixed models and the variables were transformed if necessary to meet the assumptions of the model. The transformations were arcsin (sin−1√[p/100]) for the biomass reduction data from experiments 1, 2 and 3 and natural logarithm of counts plus one for the emergence data from Experiment 3. The back-transformed means are presented where transformations were used. The means were compared by using Fisher's Protected Least Significant Difference (LSD) test. In general, the results that were reported as not significantly different had P-values of >0.05.
All the treatments with a 100% biomass reduction for the majority or all replications were excluded from the analyses. For the treatments that were included in the analyses, the 97.5% one-sided confidence intervals were calculated for their predicted means in order to compare them with the mean of the treatments that were excluded from the analyses (100%). The treatment-predicted means with a one-sided confidence interval of <100% were considered to be significantly lower than the treatments with 100% control.
For experiments 1 and 2, the emergence means were compared between the treatments with and without the pre-emergence herbicide, imazapic, and for Experiment 3, the emergence means were compared between the double-knock treatments with and without pre-emergence herbicides added to either the first or second component of the double knock. The emergence means were compared by using Fisher's Protected LSD test.
All the analyses were carried out by using the Restricted Maximum Likelihood procedure with GenStat 14th Edition (Payne et al. 2011).
Results
The level of biomass reduction of E. colona differed significantly among the herbicide treatments in experiments 1, 2 and 3 (P < 0.001) and Experiment 5 (P = 0.002), but not in Experiment 4. The level of emergence was significantly different among the herbicide treatments in Experiment 3 (P < 0.001), but not in experiments 1 and 2.
Experiment 1
When applied at the pre- to early-tillering stage (Time 1), the treatments with glyphosate or paraquat applied alone or with imazapic gave effective biomass reductions (88.7–99.6%) (Table 2). There was no significant difference between the treatments at Time 1, although there was a trend of reduced efficacy when paraquat was applied at the lowest rate (300 g ai ha−1).
A delay of 13 days in the treatment application resulted in an average reduction in efficacy of 37%, but there were significant differences between the treatments (Table 2). The level of efficacy for the two highest rates of glyphosate (540–720 g ai ha−1) was not reduced significantly but it was reduced at the lowest rate (360 g ai ha−1) by 26.0%. The greatest reduction was in relation to the paraquat treatments, particularly at the lowest rate, with a 74.6% reduction.
All the sequential herbicide treatments gave 100% control, irrespective of the rates (Table 2). The level of emergence was not reduced by a significant amount when the pre-emergence herbicide, imazapic, was applied with either glyphosate or paraquat (P > 0.05; data not presented).
Experiment 2
When applied at the pre- to early-tillering stages (Time 1), the treatments with glyphosate or paraquat applied alone or with imazapic gave a 95.0–99.4% biomass reduction, except the lowest rate of paraquat (300 g ai ha−1), which provided only 90.4% control (Table 2). A delay of 7 days in the application resulted in significantly reduced efficacy for all the treatments by an average of 46%, with the greatest reduction for the lowest rates of glyphosate and paraquat (Table 2).
All the sequential herbicide treatments gave a 100% biomass reduction, irrespective of the rates (Table 2). As in Experiment 1, the level of emergence was not significantly reduced when imazapic was applied with either glyphosate or paraquat (P > 0.05; data not presented).
Experiment 3
Glyphosate, when applied alone at the pre- to early-tillering stages (Time 1), gave a 99.5% biomass reduction (Table 3). However, this was reduced to 52.1% after a delay in application of 7 days. The addition of atrazine to glyphosate gave poor knockdown control (65.3%), whereas the addition of imazapic to glyphosate or paraquat and the addition of atrazine to paraquat gave a 97.7–99.4% biomass reduction (Table 3).
All the sequential application treatments gave a 94.8–100% biomass reduction with or without an added pre-emergence herbicide (Table 3). The addition of imazapic to glyphosate or paraquat did not reduce the subsequent emergence significantly, compared with glyphosate applied alone at Time 1 (Table 3). This was also the case for atrazine applied with either glyphosate or paraquat. However, the inclusion of a pre-emergence herbicide as part of the sequential herbicide treatments significantly reduced the level of emergence, except for glyphosate + imazapic followed by paraquat (Table 3). This was irrespective of whether the pre-emergence herbicide was mixed with glyphosate or paraquat. The treatments that gave the greatest reduction in emergence were metolachlor and metolachlor + atrazine, with an average reduction of 95% and 87%, respectively, compared to the double-knock treatments without added pre-emergence herbicides.
Experiment 4
Although there was no significant difference between the treatments that were included in the analysis with a <100% biomass reduction, comparisons using the upper confidence interval indicated that the two populations responded differently to some of the herbicide treatments (Table 4). For the glyphosate-susceptible plants, there was no significant difference in the efficacy of the herbicide treatments, with an 83.5–100% biomass reduction (Table 4). However, there was a trend of reduced control of the glyphosate-resistant plants when the low rate of glyphosate (158 g ai ha−1) was applied either at Time 1 or Time 2. In contrast, the higher rate of glyphosate (315 g ai ha−1), paraquat and the sequential herbicide treatments were not significantly different between both the susceptible and the resistant plants.
Experiment 5
Glyphosate at the low rate, when applied at Time 2, gave a significantly lower biomass reduction of the glyphosate-resistant (14.5%), compared to the glyphosate-susceptible (58.5%), plants (Table 4). For the glyphosate-resistant plants, there was no significant difference in the amount of biomass reduction between the glyphosate rates when applied at Time 1, but the amount of biomass reduction was reduced significantly for the lower rate when applied at Time 2.
Comparisons using the upper confidence interval indicated that all the glyphosate-alone treatments were less effective (14.5–49.5% biomass reduction) on the glyphosate-resistant population than paraquat alone and the sequential application treatments (Table 4). All the paraquat and sequential application treatments gave a 100% biomass reduction of both the susceptible and the resistant plants, irrespective of the herbicide rates (Table 4).
Discussion
Improving glyphosate efficacy
Glyphosate was an effective herbicide for the control of young, glyphosate-susceptible E. colona plants at the pre- to early-tillering stages, but an increased rate was required to control the late-tillering plants. Glyphosate consistently provided 83.5–100% control of pre- to early-tillering, glyphosate-susceptible E. colona plants and the herbicide rate had little impact on the efficacy at these growth stages. The efficacy of glyphosate generally was reduced when applied to older plants at the mid- to late-tillering stages, with the greatest reduction at the low application rates. This reduction in control was most evident in experiments 2, 3 and 5, where there were greater proportions of more mature plants (late-tillering to seeding) at the second time of application.
The reduction in control as the result of a larger plant size was overcome to some extent with an increased rate of glyphosate; for example, in Experiment 2, by doubling the 360 g ai ha−1 rate of glyphosate, the level of control was increased by 57.6% for the glyphosate-susceptible plants.
In the subtropical region of Australia, there can be multiple instances of emergence of E. colona in any one season (McGillion & Storrie 2006), thereby resulting in populations with weeds of varying ages and sizes. Also, summer rain can delay access to weed populations, allowing them to grow to larger sizes within a short period of time. In order to effectively control such populations, a higher rate of glyphosate will be required.
An over-reliance on glyphosate alone for the fallow management of E. colona has been the primary driver of the widespread evolution of glyphosate resistance in this species across subtropical Australia (Thornby & Walker 2009). Therefore, even though glyphosate use is an effective tactic for controlling glyphosate-susceptible populations, history suggests that an over-reliance should be avoided and other weed management tactics should be used in conjunction with glyphosate.
Glyphosate applied alone was not effective in controlling glyphosate-resistant E. colona, although an increased rate applied to pretillering plants provided some control. As expected, unacceptable control (14–67%) was achieved with glyphosate on glyphosate-resistant populations, irrespective of the herbicide rate and growth stage. However, a high rate of glyphosate (315 g ai ha−1) applied to pretillering plants provided some control (50–67%). Therefore, glyphosate could play a role in the control of small glyphosate-resistant E. colona plants, but not as a stand-alone application. Other tactics will need to be used in addition to glyphosate to achieve effective control.
Glyphosate alternatives
Paraquat
Paraquat applied alone was an effective treatment on young (pretillering) E. colona plants, irrespective of their resistance status. However, the growth stage at the time of application had a significant impact on its efficacy. This was especially evident in Experiment 1, where a delay of 13 days between applications resulted in a 43–75% reduction in efficacy. A previous study showed a reduction of 21% control when paraquat was applied 8 days later to tillering E. colona plants that were 6–10 cm high (Storrie et al. 2008). In this study, increasing the paraquat rate did not compensate for the reduction in efficacy on older plants. Generally, the level of control that was achieved in the field with paraquat was greatest when it was applied early at a low rate (300 g ai ha−1) than when it was applied later at a high rate (500 g ai ha−1).
Paraquat is an effective alternative knockdown herbicide for the control of both glyphosate-susceptible and glyphosate-resistant E. colona. However, a shift away from glyphosate to an over-reliance on paraquat applied alone poses another herbicide resistance risk. Although no case of paraquat resistance has been confirmed for E. colona internationally, six other weed species, including four grass species, have developed resistance to paraquat in Australia (Heap 2012).
Sequential application
In four out of the five experiments, the sequential application treatments gave 100% control of E. colona. The exception was Experiment 3, where three treatments gave 96–99.4% control. This small reduction in efficacy might have been related to the high density of 680 plants per m2 in this field trial, which could have impeded the level of herbicide coverage.
A similar result was found for Conyza bonariensis, another key problem weed of the subtropical region of Australia, where sequential applications provided better control than glyphosate alone (Werth et al. 2010). This tactic is now an important tool in the management of C. bonariensis.
In this research, the timing of the sequential applications was not explored. The first herbicide application always was applied when most of the plants were pre- to early-tillering. A recent study showed that 1 and 2 month old C. bonariensis plants were controlled better than 3 month old C. bonariensis plants with sequential herbicide applications (Walker et al. 2012). The results of this study also suggest that it is likely that poorer control of E. colona would be achieved if the first herbicide application was delayed until most of the plants were mid- to late-tillering. In this situation, glyphosate would be less effective and more reliance would be placed on paraquat, particularly if applied to glyphosate-resistant plants.
Pre-emergence herbicides
Generally, the addition of a pre-emergence herbicide to either a mixture or a sequential application treatment did not reduce the knockdown efficacy of the treatment. The exception was the addition of atrazine to glyphosate, which reduced the level of knockdown control by 34.2%. Antagonism between these two herbicides has been reported in other studies (Selleck & Baird 1981).
Atrazine or imazapic mixed with either glyphosate or paraquat was not effective in reducing the level of E. colona emergence. In contrast, all the other pre-emergence herbicides significantly reduced the level of emergence when applied as part of the double knock, irrespective of the application in the first or second knock, with the most effective treatments being metolachlor and metolachlor + atrazine.
Pre-emergence herbicides can provide season-long control of weeds (McGillion & Storrie 2006), although this was not assessed in these field experiments. As E. colona can emerge in multiple flushes, pre-emergence herbicides could be useful in reducing the number of E. colona plants in fallow and also in subsequent crops.
Conclusion
This study has identified and defined a number of effective tactics for the control of both glyphosate-susceptible and glyphosate-resistant E. colona populations. These tactics should be used in conjunction with other tactics as part of an integrated weed management strategy to target different stages of the weed's life cycle. Such a strategy will improve weed control, reduce weed seed production, deplete the seed bank and reduce the risk of further herbicide resistance development.
Acknowledgments
The authors wish to thank the Grains Research and Development Corporation, Canberra, Australia, for financial support of this research.
