3.1. Experiment 1. Effect of Temperature on Glyphosate Efficacy
As described above, this experiment was conducted four times in a screenhouse in different months. Two trials (Trial 1 and 4) experienced high average and high minimum temperatures (high-temperature runs) and the other two runs (Trial 2 and 3) experienced low average and low minimum temperatures (low-temperature runs) during their growing duration (
Table 1).
In both runs in the low-temperature regime, there was no survival for the GS biotype at 570 g glyphosate/ha; however, in the high-temperature regime, 83% plants survived glyphosate application at this rate (
Figure 1). Similarly, for the GR biotype, 42 to 58% plants survived the application of glyphosate at 2280 g/ha (three times the recommended rate) when applied during the high-temperature regime (
Figure 1). However, no plants of the GR biotype survived this rate when glyphosate was applied during the low-temperature regime. Similar results were found for
S. oleraceus biomass (
Figure 2). The GR biotype produced 4 to 10% biomass of the control treatment at a glyphosate rate of 2280 g/ha during the high-temperature regime but no biomass was produced at this glyphosate rate during the low-temperature regime. Similarly, at the lower glyphosate rate (1140 g/ha) too, the GR biotype produced greater biomass during the high-temperature regime (12 to 15% of their non-treated control) compared with the low-temperature regime (4 to 5% of their non-treated control). The GS biotype did not produce biomass at a glyphosate rate of 570 g/ha during the low-temperature regime but produced 17 to 28% biomass of the non-treated control treatment during the high-temperature regime at that glyphosate rate.
Undoubtedly, these results suggest that glyphosate efficacy is reduced at high temperatures. Similar results were reported for
E. colona in South Australia, in which glyphosate resistance increased at 30 °C compared with 20 °C [
12]. The authors suggested that a reason for this response at the high temperature was reduced glyphosate absorption, which would reduce the herbicide concentration in the leaf and its ability to enter the chloroplast. This explanation could be true for our study also; however, we did not evaluate the absorption and translocation of glyphosate. A recent study in the USA also reported that all the GR
E. colona plants treated with glyphosate at 840 g/ha died when subjected to 15/10 °C alternating day/night temperature but the GR plants did not show any phytotoxicity at this herbicide rate when subjected to higher temperatures, 25/20 °C and 35/30 °C [
21].
Another study on broadleaf weed species also indicated that the control of
Chenopodium album L. and
Conyza canadensis (L.) Cronquist by glyphosate could be reduced under projected future climatic conditions as both weed species were less sensitive to glyphosate under the higher temperature regime (32/26 °C) compared with the lower temperature regime (18/12 °C) [
22]. These authors also suggested that altered glyphosate translocation might be the basis for reduced weed sensitivity at high temperatures. Our study was conducted in a naturally lit screenhouse, in which photoperiod and light intensity also differed during the four runs. Light intensity may affect the thickness of the leaf and therefore herbicide absorption and efficacy.
Sonchus oleraceus, considered mainly a winter weed in the past, is now a common weed throughout the year in Australia, especially in the Southeast Australian region, where rainfall is distributed throughout the year. In Queensland and New South Wales, growers usually grow one crop in a year, depending on rainfall; therefore, they need to manage weeds during the fallow phase. Due to sustainability of conservation agriculture systems, growers do not like to till their farm to control weeds and they rely on glyphosate during the fallow phase. Our results suggest that summer fallows will experience more glyphosate failures compared with winter fallows.
S. oleraceus plants may survive glyphosate application during the summer months and produce seeds. As this weed can grow throughout the year, plants produced from these seeds may also become harder to control during the winter months, especially the GR populations. Therefore, over-reliance on glyphosate for weed control in summer fallows may result in more weed control failures [
22], especially for weeds like
S. oleraceus, which occur throughout the year.
Additionally, weather conditions in Southeast Australia are quite variable [
23]. There can be hot days during autumn and winter months in this region, which may affect glyphosate efficacy. Therefore, growers need to check temperature conditions before glyphosate applications. Higher rates may be recommended to improve glyphosate activity on plants exposed to high temperatures.
3.2. Experiment 2. Performance of Different Post-Emergence Herbicides
As expected, glyphosate at both rates failed to provide effective control of the GR biotype (
Table 3 and
Table 4). All seedlings of the GS biotype were killed by glyphosate application when applied at the four-leaf stage but a delayed application to the six-leaf stage resulted in 36 to 88% seedling survival (
Table 3). These seedlings, however, produced only 3 to 10% of the biomass in the non-treated control (
Table 4). Similar results were reported for
C. canadensis, which was found to be more susceptible to glyphosate at the seedling stage than at the large rosette stage [
24].
Irrespective of growth stage and herbicide rate, glufosinate and paraquat were the most effective post-emergence herbicides for
S. oleraceus control, resulting in no seedling survival (
Table 3) and biomass production (
Table 4). In the USA, glufosinate has been reported as an alternative herbicide option for control of GR
Ambrosia trifida L. [
25]. A later study also reported glufosinate (700 g/ha) as an option for controlling GR
Sorghum halepense L., with 77% control at soybean harvest [
26]. An early study reported that control of
C. album with glufosinate was most effective when applied at the 10-cm weed height compared to the 15-cm weed height [
27]. In another study, glufosinate provided greater control (88%) of GR
C. canadensis compared with paraquat (55 to 63%) [
28]. Paraquat in our study provided excellent control of
S. oleraceus but weeds may develop resistance to this herbicide due to restricted translocation [
29]. Therefore, plants need to be monitored after paraquat application for possible escapes in the field.
No seedlings of
S. oleraceus survived when 2,4-D + picloram, bromoxynil and saflufenacil were applied at the four-leaf stage. However, a significant number of seedlings survived when these herbicides were applied at the six-leaf stage. The results were more evident for bromoxynil, resulting in 100% seedling survival at both herbicide rates for both biotypes (
Table 3). Compared with the non-treated control, bromoxynil at both rates reduced biomass by 27 to 46% for the GR biotype and 66 to 68% for the GS biotype (
Table 4).
Similarly, late application of saflufenacil resulted in 49 to 89% seedling survival for the GR biotype and 64 to 71% for the GS biotype. However, these surviving plants did not grow very vigorously after herbicide application. Compared with the non-treated control, saflufenacil at both rates reduced 71 to 81% biomass for the GR biotype and 84 to 88% biomass for the GS biotype. A study from the USA suggested that saflufenacil could provide 90% control of rosette and bolted
Parthenium hysterophorus L. at 6 to 27 g/ha, a range similar to our study [
30]. The author also suggested that saflufenacil could be an effective burndown herbicide for control of
P. hysterophorus populations resistant to glyphosate. In another study, the height of
C. canadensis had little effect on saflufenacil efficacy, which provided 95 to 99% control of the weed when applied to small plants or >25-cm tall plants [
31]. The authors suggested that the time of day of application had a greater effect on weed control with saflufenacil than weed height or weed density; however, we did not evaluate the effect of time of application on saflufenacil efficacy.
Irrespective of herbicide rate and growth stage, fluroxypyr and metsulfuron did not reduce plant survival compared with the non-treated control treatment (
Table 3). These herbicide treatments resulted in 0 to 44% biomass reduction for fluroxypyr and 0 to 88% biomass reduction for metsulfuron (
Table 4). The results show that there was a differential response between the two biotypes to metsulfuron. No reduction in biomass was observed for the GR biotype, indicating that the GR biotype may also be resistant to metsulfuron. Both fluroxypyr and metsulfuron are recommended for
S. oleraceus control in Australia but they did not provide effective control of this weed.
S. oleraceus biotypes resistant to chlorsulfuron (acetolactate synthase inhibitor) and 2,4-D (synthetic auxins) have been reported in Australia but no resistant biotypes to fluroxypyr (synthetic auxins) and metsulfuron (acetolactate synthase inhibitor) are known [
10]. Our results suggest that there is a need to screen different populations of
S. oleraceus against these herbicides to evaluate their resistance status.
Except for the higher herbicide application rate at the four-leaf stage for the GS biotype, 2,4-D also proved ineffective in reducing the survival percentage of
S. oleraceus (
Table 3). The effect of 2,4-D was more suppressive on
S. oleraceus biomass than on its seedling survival. At the lower rate, 2,4-D application at the four-leaf stage reduced biomass by 34 and 71% for the GR and GS biotype, respectively (
Table 4). This biomass was further reduced by 66 and 91% for the GR and GS biotypes, respectively, at the higher 2,4-D rate (i.e., 1050 g/ha). Delayed 2,4-D application, however, resulted in only 20 to 44% reduction in biomass, depending on biotype and 2,4-D rate.
As 2,4-D is used to manage a range of broadleaf weeds, its failure at the highest rate warranted a dose–response study. Fortunately, both biotypes were controlled completely at 1400 and 2800 g/ha of 2,4-D (data not shown). The herbicide 2,4-D on its own is not recommended for S. oleraceus control in fallows and our results support this recommendation. Therefore, 2,4-D needs to be mixed with other compatible herbicides to achieve complete control of S. oleraceus.
In general, a delayed herbicide application reduced S. oleraceus control. This was particularly true for bromoxynil and saflufenacil. Our study found a number of alternative herbicide options to control GR biotypes of S. oleraceus; however, these herbicides need to be applied at an early stage to achieve effective weed control. Growers tend to delay herbicide application to maximise weed seedling emergence from the seed bank so that all the weeds can be treated at the same time but this tendency may result in the build-up of herbicide-resistant weed seed banks.
3.3. Experiment 3. Effect of Sorghum Residue Amount on Efficacy of Pre-Emergence Herbicides
The interaction effect of herbicide treatments and sorghum residue amount was significant for the seedling emergence (
Table 5) and biomass (
Table 6) of both biotypes of
S. oleraceus. In the non-treated control treatment, covering the seeds with sorghum residue did not affect the seedling emergence or biomass of both biotypes. Compared with the non-treated control, across residue amount, only a few herbicide treatments were able to reduce seedling emergence but a greater number of herbicide treatments were able to reduce seedling biomass. These results suggest that seedlings that survived herbicide application were not able to grow vigorously.
Isoxaflutole at 75 g/ha provided effective control of seedling emergence of both biotypes when applied without crop residue cover; however, the addition of sorghum residue reduced the efficacy of isoxaflutole at this rate, resulting in an increased amount of seedling emergence (
Table 5). Isoxaflutole at 150 g/ha was required to overcome the problem of increased seedling emergence with the addition of crop residue. These differential responses were not found for seedling biomass of both biotypes (
Table 6). Seedling biomass was similar across both rates of isoxafluotole. Regardless of sorghum residue amount, herbicide rate and weed biotype, isoxaflutole provided 81 to 100% suppression of
S. oleraceus biomass.
Isoxaflutole is a soil-applied isoxazole herbicide used for the control and suppression of selective broadleaf and grass weeds in sugarcane, chickpea and fallows by inhibiting the 4-hydroxyphenyl-pyruvate-dioxygenase (4-HPPD) biochemical pathway. In a previous study, lower degradation of isoxaflutole was found under conservation tillage than under conventional tillage [
32]. Our results suggest that the presence of crop residue in conservation agriculture systems will not affect isoxaflutole efficacy on
S. oleraceus biomass. In fallows, surviving seedlings may grow and produce seeds if follow-up treatments are not applied.
S. oleraceus seedlings may emerge following isoxaflutole application but may not be able to produce enough biomass in competitive sugarcane or chickpea crops. In the USA, isoxaflutole was found to be an option for use in maize, especially for atrazine-resistant weeds [
33]. In Australia, this herbicide needs to be evaluated for weed control in maize; however, a carryover from isoxaflutole applications in maize may require plant back restrictions for certain sensitive crops [
34].
Pendimethalin at 910 g/ha did not affect the seedling emergence of either biotype (
Table 5). Increasing the rate to 1820 g/ha also did not affect seedling emergence in the no-residue treatment; however, integration of pendimethalin at 1820 g/ha with residue retention resulted in decreased seedling emergence of both biotypes compared with the non-treated control treatments. The response was different between the biotypes for seedling biomass (
Table 6). Pendimethalin application resulted in significant reductions (81 to 94%) in the seedling biomass of the GR biotype compared with the non-treated control treatments and there was no difference between the two rates of the herbicide. For the GS biotype, however, seedling biomass was similar between the non-treated control and pendimethalin treatments. These differential responses of the two biotypes to pendimethalin suggest a need to evaluate the performance of this herbicide on several biotypes of
S. oleraceus.
Pendimethalin is a soil-applied dinitroaniline herbicide used for the control of some broadleaf and grass weeds by inhibiting mitosis. A study from the USA suggested that the use of biochar as a soil amendment could decrease pendimethalin efficacy by adsorbing a significant amount of the herbicide [
35]. In another study similar to our study, pendimethalin application (1 and 2 kg ai/ha) in the presence of rice residue cover resulted in lower control of
Cyperus iria L. than in the absence of residue
15. This study suggested that some weed species may escape pendimethalin application in conservation agriculture systems. However, such adverse results of the crop residue on pendimethalin efficacy were not found in our study with
S. oleraceus.
The herbicide
s-metolachlor at 960 g/ha did not affect the seedling emergence of either biotype at different residue levels compared to their respective non-treated control treatments (
Table 5). However, increasing the herbicide rate to 1920 g/ha resulted in decreased emergence in the no-residue cover condition. Covering the seeds with residue cover negated the effect of this herbicide rate on seedling emergence. A similar response was found for seedling biomass; however, both biotypes behaved differently (
Table 6). Application of
s-metolachlor resulted in greater suppression of biomass for the GR biotype compared with the GS biotype. Biomass of both biotypes was similar across
s-metolachlor rates.
S-metolachlor is a soil-applied amide herbicide used for the control of some broadleaf and grass weeds in certain crops by inhibiting very long-chain fatty acid biosynthesis. In a recent study, using sorghum residue resulted in lower
s-metolachlor efficacy on
E. colona and
Chloris virgata Sw., grass weed species [
19]. In our study,
s-metolachlor efficacy was not reduced by the addition of sorghum residue. Differential responses between the two studies could be due to the different irrigation systems used in these studies. Plants in the previous study were sub-irrigated while plants in the current study were irrigated using an overhead sprinkler system. A sprinkler irrigation system could have resulted in washing off the herbicide from crop residue, which was then available to
S. oleraceus. A study from the USA suggested that integration of high-residue cover crops with
s-metolachlor increased
Amaranthus species control [
36], suggesting that residue retention in conservation agriculture systems may not reduce the efficacy of
s-metolachlor.