Introduction
Annual sowthistle (Sonchus oleraceus L.) is an emerging weed of global importance that is increasingly prevalent in the Australian cropping regions (Chauhan et al. Reference Chauhan, Gill and Preston2006; Gomaa et al. Reference Gomaa, Hassan, Fahmy, González, Hammouda and Atteya2014; Hassan et al. Reference Hassan, Gomaa, Fahmy, González, Hammouda and Atteya2014; Manalil et al. Reference Manalil, Werth, Jackson, Chauhan and Preston2017; Osten et al. Reference Osten, Walker, Storrie, Widderick, Moylan, Robinson and Galea2007). Abundant seed production, small seed size, and wind dispersal of seeds are some of the features that are contributing to the invasive success of this weed (Chauhan et al. Reference Chauhan, Gill and Preston2006; Widderick et al. Reference Widderick, Walker, Sindel and Bell2010). Studies on germination ecology indicated the potential of S. oleraceus to germinate under a broad range of temperature conditions and varying environmental conditions, including pH, salinity, and water stress (Chauhan et al. Reference Chauhan, Gill and Preston2006; Manalil et al. Reference Manalil, Werth, Jackson, Chauhan and Preston2017). This weed was once considered to be only a winter weed, as it was mostly confined to the winter season, but currently it is present in summer, winter, and fallow phases of cropping due to its potential to emerge under a broad range of temperature conditions (Manalil et al. Reference Manalil, Werth, Jackson, Chauhan and Preston2017; Werth et al. Reference Werth, Boucher, Thornby, Walker and Charles2013; Widderick et al. Reference Widderick, Walker, Sindel and Bell2010). Among the three agroecological grain-cropping regions of Australia, the western region (Western Australia) and the southern region (Victoria, Tasmania, and South Australia) are characterized by a Mediterranean-type climate with winter-dominant rains and cropping mostly confined to the winter season. Conversely, the northern region (New South Wales and Queensland) is characterized by well-distributed rainfall and intensive cropping during both the summer and winter seasons, providing enough opportunities for this weed to proliferate in the region (GRDC 2018). When present in fallow phases without competition from a crop, this weed could deplete substantial soil moisture (GRDC 2009). Interference studies in different agroecosystems indicate substantial yield loss due to S. oleraceus in crops (Hassan et al. Reference Hassan, Gomaa, Fahmy, González, Hammouda and Atteya2014; Song et al. Reference Song, Kim, Im, Lee, Lee and Kim2017). Suppression of other plants could be due to the depletion of resources and the allelopathic properties of this weed (Gomaa et al. Reference Gomaa, Hassan, Fahmy, González, Hammouda and Atteya2014; Peerzada et al. Reference Peerzada, O’Donnell and Adkins2019). Although S. oleraceus is quite prevalent in South Australia, Queensland, and New South Wales, specific studies on weed interference in crops are lacking (Manalil et al. Reference Manalil, Werth, Jackson, Chauhan and Preston2017; Osten et al. Reference Osten, Walker, Storrie, Widderick, Moylan, Robinson and Galea2007; Werth et al. Reference Werth, Boucher, Thornby, Walker and Charles2013). Various surveys indicate an increase in prevalence of this weed in the grain crops and cotton (Gossypium hirsutum L.) production regions of Australia (Manalil et al. Reference Manalil, Werth, Jackson, Chauhan and Preston2017; Osten et al. Reference Osten, Walker, Storrie, Widderick, Moylan, Robinson and Galea2007; Werth et al. Reference Werth, Boucher, Thornby, Walker and Charles2013).
Sochus oleraceus can thrive well under moisture-limiting conditions and under a wide range of pH, saline, and nutrient-deficient environments (Chauhan et al. Reference Chauhan, Gill and Preston2006; Manalil et al. Reference Manalil, Ali and Chauhan2018; Widderick et al. Reference Widderick, Walker, Sindel and Bell2010). This weed can germinate under a wide range of temperature conditions and can grow up to an altitude of around 2,500 m, indicating the potential to emerge and survive under cooler environments (Peerzada et al. Reference Peerzada, O’Donnell and Adkins2019). Although S. oleraceus is categorized as a meso-xerophytic plant, it will flourish in moist fertile soils under noncompetitive fallow by using residual soil nutrients and soil moisture (Peerzada et al. Reference Peerzada, O’Donnell and Adkins2019). The increase in the occurrence of this weed has been particularly noted in cropping systems that use conservation tillage and glyphosate-tolerant crops (Manalil et al. Reference Manalil, Werth, Jackson, Chauhan and Preston2017; Peerzada et al. Reference Peerzada, O’Donnell and Adkins2019; Werth et al. Reference Werth, Boucher, Thornby, Walker and Charles2013). Germination ecology studies indicate low germination of this weed in darkness compared with illuminated conditions, indicating the likelihood of limited establishment in shaded environments or under residue cover (Chauhan et al. Reference Chauhan, Gill and Preston2006; Manalil et al. Reference Manalil, Ali and Chauhan2018; Widderick et al. Reference Widderick, Walker, Sindel and Bell2010). Additionally, germination was quite high at the soil surface, and no germination was observed at 6-cm depth (Manalil et al. Reference Manalil, Ali and Chauhan2018; Widderick et al. Reference Widderick, Walker, Sindel and Bell2010). Despite the documentation of increased occurrence of S. oleraceus in the northern grain region of Australia, no specific study has explored the interference pattern of this weed.
Evolution of herbicide resistance necessitates the inclusion of all possible nonchemical strategies for weed management (Chauhan et al. Reference Chauhan, Namuco, Ocampo, Son, Thu, Nam, Phuong and Bajwa2015; Owen et al. Reference Owen, Beckie, Leeson, Norsworthy and Steckel2015; Preston et al. Reference Preston, Wakelin, Dolman, Bostamam and Boutsalis2009; Riar et al. Reference Riar, Norsworthy, Steckel, Stephenson, Eubank, Bond and Scott2013). Surveys of glyphosate-tolerant cotton systems in Australia indicate a rapid progression of S. oleraceus as a major broadleaf weed and point to either a natural tolerance or resistance to glyphosate (Manalil et al. Reference Manalil, Werth, Jackson, Chauhan and Preston2017; Werth et al. Reference Werth, Boucher, Thornby, Walker and Charles2013). Adkins et al. (Reference Adkins, Wills, Boersma, Walker, Robinson, McLeod and Einam1997) observed chlorsulfuron-resistant populations of S. oleraceus in the 1990s. In Australia, Cook et al. (Reference Cook, Davidson and Miller2014) observed glyphosate-tolerant populations of S. oleraceus, and Boutsalis and Powles (Reference Boutsalis and Powles1995) reported resistance to acetolactate synthase–inhibiting herbicides. Werth et al. (Reference Werth, Thornby and Walker2011), in a risk assessment study, included S. oleraceus in the “high-risk category” of weeds for its potential to develop resistance to glyphosate based on the biology and current management practices followed in the cotton production region.
There are many instances when major weeds fail to compete with crops and crops may smother weeds in competition, and such information is quite valuable for framing nonchemical weed management strategies (Cholette et al. Reference Cholette, Soltani, Hooker, Robinson and Sikkema2018; Lazzaro et al. Reference Lazzaro, Costanzo and Bàrberi2018; Mwendwa et al. Reference Mwendwa, Brown, Wu, Weston, Weidenhamer, Quinn and Weston2018; Reiss et al. Reference Reiss, Fomsgaard, Mathiassen and Kudsk2018). Emergence, growth, flowering, seed production, and seed dispersal in weeds in relation to crop phenological phases offer valuable inputs to frame appropriate weed management strategies (Andersen, Reference Andersen1992; Beckie et al. Reference Beckie, Blackshaw, Harker and Tidemann2017; Chauhan et al. Reference Chauhan, Gill and Preston2006; Devlaeminck et al. Reference Devlaeminck, Bossuyt and Hermy2005; Hassan et al. Reference Hassan, Gomaa, Fahmy, González, Hammouda and Atteya2014). Weeds that mature and disperse seed after crop harvest and have high seed retention during crop harvest provide opportunities for management through seed capturing and destruction during harvest operations, thereby reducing seedbank enrichment (Walsh et al. Reference Walsh, Broster, Schwartz-Lazaro, Norsworthy, Davis, Tidemann, Beckie, Lyon, Soni, Neve and Bagavathiannan2018; Walsh and Powles Reference Walsh and Powles2014). Although substantial S. oleraceus seeds could be dispersed by wind (Peerzada et al. Reference Peerzada, O’Donnell and Adkins2019), a delayed maturity of weed compared with the crop may allow the capture of weed seeds during crop harvest. At present, knowledge gaps exist in regard to the competitiveness and seed set of S. oleraceus in winter crops in Australia. Therefore, a study was conducted to explore the pattern of interference of S. oleraceus in wheat (Triticum aestivum L.).
Materials and Methods
To explore the interference of S. oleraceus, field studies were conducted from May to October in 2016 and 2017 at the Research Farm of the University of Queensland (27.543°S, 152.334°E), Gatton, Australia. The soil of the experimental site was a heavy clay with a pH of 7.5 and an organic matter content of 2.7%. The nitrogen, phosphorus, and potassium concentrations of the soil were 62, 87, and 412 kg ha−1, respectively. The long-term average rainfall of the site was 772 mm. In 2016, there was 562 mm of rainfall, of which 202 mm was during the crop-growing months of the winter season (May to September 2016). Only 82 mm of rainfall was received during the winter growing season in 2017, although annual rainfall was 797 mm (Figure 1).
Figure 1. Walter-Lieth climate diagram for Gatton, Queensland, for 2016 (A) and 2017 (B). Mean monthly temperature (C) and precipitation (mm) are shown on the left axis (red) and right axis (blue), respectively. Letters “J” to “D” on x axis indicate January to December (experiment was carried out from May to October). Dry and wet months are represented by area speckled in red and blue vertical lines, respectively. Dark blue bars on the x axis indicate months with possibility of frost. Top and bottom black numbers on the left axis are the mean maximum temperature of the hottest month and mean minimum temperature of the coldest month, respectively.
The field was cultivated two to three times using a rotory cultivator to ensure a stale seed bed at the time of planting. Wheat (‘Spitfire’, Pacific Seeds, Toowoomba, Australia) was seeded at 60 kg ha−1 with an 18-cm row spacing using a hoe drill (John Deere, Moline, IL, USA). Sonchus oleraceus seeds were collected from a wheat field in Queensland (27.559°S, 152.324°E) and were established at four densities. The targeted densities were weed-free (0 plants m−2), low (10 to 20 plants m−2), medium (30 to 40 plants m−2), and high (50 to 70 plants m−2). The trial was established in using a randomized block design with three replications and plots measuring 5.0 m by 2.3 m. After wheat sowing (on the same day), weed seeds were mixed with dry soil and uniformly hand broadcast. Differential weed seeding rates were used based on the laboratory germination data so as to have different initial emergence rates to provide the required weed densities.
Plots were sprinkle irrigated four times on alternate days beginning at seeding to ensure weed and crop emergence and during crop flowering in August (twice). In Australia, wheat is grown both under rainfed and irrigated environments; in this study, irrigation was provided initially (mainly to ensure uniform crop and weed emergence) and during anthesis. Diammonium phosphate at 25 kg N ha−1 and 28 kg P ha−1 was broadcast before irrigation. The plots were continuously hand weeded to remove all other weeds except the target weeds. Weed density and dry biomass were recorded within a 60 cm by 54 cm quadrat at two locations within each plot at the time of wheat anthesis and again before crop harvest. Wheat panicles were counted within 1-m row length of the crop in two places. Numbers of grains were averaged from 20 randomly picked panicles per plot, and 1,000-grain weight was measured from the harvested sample. Harvesting was carried out with a plot harvester, and grain yield (kg ha−1) was adjusted at 12% moisture content.
Sonchus oleraceus seed production was computed by randomly picking 20 flower heads from each plot that were ready to open. Flower heads were dissected, and seeds were counted using a magnifying glass. Total flower heads (both dispersed and ready to disperse) were counted from the quadrated weed sample coinciding with wheat maturity. Based on this, total seed production and percentage of seed dispersal coinciding with crop maturity were computed. Emergence, flowering, and maturity of wheat and S. oleraceus were related to growing degree days base 5 (GDD5).
$$\eqalign{{\hskip-3pc{\rm{GDD_5} = 	\sum \lbrace [ ( \rm{Maximum \ daily \ temperature}}} \cr 	{{{\hskip-10pc+ \rm{Minimum\ daily\ temperature} )/2 ] - 5 \rbrace}$$
([1])ANOVA was performed to identify differences between treatments (R Development Core Team 2018) on yield attributes. Bartlett’s test and a Shapiro–Wilk test were used to evaluate the homoscedasticity and normality assumptions before performing ANOVA. All experiments were carried out twice. Data were analyzed differently for the two field trials, as results were significantly different. Nonlinear regression analysis was performed to explore the relationship between weed density and crop yield, and weed density and weed seed production. ANOVA was performed on yield components, including panicles per square meter, grains per panicle, and 1,000-grain weight, and means separation was accomplished using Fisher’s LSD (α = 0.05).
A Walter-Lieth climate diagram was prepared using the Climatol package (R Development Core Team 2018).
To assess the relationship between weed density and crop yield loss, several commonly used models were compared using the Akaike information criterion (AIC) (Table 1). The model with the lowest AIC was a two parameter modified exponential decay regression model:
$$g{\rm{ }} = {\rm{ }}a\left( {1{\rm{ }} - {\rm{ }}ex{p^{( - b*x)}}} \right)$$
([2])where g is the crop yield loss (%) to weed density (x), a is the maximum crop yield loss (%), and b is the rate constant (slope of regression) (Archontoulis and Miguez Reference Archontoulis and Miguez2015).
Table 1. Regression models tested and their Akaike information criterion (AIC).a
a Asterisks (*) indicate model with lowest AIC values included in this study.
Similarly, a four-parameter logistic model was fit to assess the relationship between weed biomass and crop yield loss:
$${\rm{g }} = {\rm{a}} + {{d - a} \over {1 + {\rm{exp}}\{ b\left[ {\log \left( x \right) - \log \left( e \right)} \right]\} }}$$
([3])where g is the crop yield loss (%) to weed biomass (x), e is the magnitude of independent variable producing a response halfway between upper limit d and lower limit a, and b denotes relative slope around e.
A two-parameter hyperbola model was used to explore the relationship between weed density at harvest and weed seed production (Archontoulis and Miguez Reference Archontoulis and Miguez2015):
$$g = a*x/\left( {b{\rm{ }} + {\rm{ }}x} \right)$$
([4])where g is weed seed production corresponding to weed density (x), a is the maximum weed seed production as estimated by the model, and b is the rate constant (slope of regression).
Results and Discussion
Wheat emerged at 9 (GDD5 = 125) and 12 (GDD5 = 120) d after seeding in 2016 and 2017, respectively. Emergence of S. oleraceus was observed at 8 to 10 d after seeding in both years. Sonchus oleraceus flowered at 85 (GDD5 = 914) and 89 (GDD5 = 889) d after seeding in 2016 and 2017, respectively. The crop was mature and ready to harvest at 136 (GDD5 = 1,523) and 139 (GDD5 = 1,647) d after seeding in 2016 and 2017, respectively.
Wheat yielded 6,720 and 5,711 kg ha−1 in the weed-free control plots in 2016 and 2017, respectively (data not shown). Yield decreased exponentially with increasing densities of S. oleraceus (Figure 2). There were 55% and 57% reductions in crop yield in the high-density plots in 2016 and 2017, respectively (Figure 2). Based on the regression model, weed densities at anthesis corresponding to a 50% yield reduction were 43 and 52 plants m−2 in 2016 and 2017, respectively. The values for weed biomass (dry) corresponding to the 50% yield reduction were 98 and 140 g m−2 in 2016 and 2017, respectively (Figure 3). Among yield components, a significant reduction was observed for the number of panicles as a response to increasing weed density (P < 0.001); at the high density, there were 47% and 55% reductions for the number of panicles per square meter in 2016 and 2017, respectively. Reductions were also observed for grains per panicle (P < 0.001); there were 23% and 27% reductions in 2016 and 2017, respectively, due to maximum weed interference. However, significant differences were not observed for 1,000-grain weight (Table 2).
Figure 2. Effect of Sonchus oleraceus density on wheat yield in 2016 and 2017. The lines represent a modified exponential decay regression model fit to the data by year, and symbols represent weed density. The selected model had the lowest Akaike information criterion among several commonly used models tested.
Figure 3. Effect of Sonchus oleraceus biomass on wheat yield in 2016 and 2017. The lines represent a four-parameter logistic model fit to the data by year, and symbols represent weed biomass. The selected model had the lowest Akaike information criterion among several commonly used models tested.
Table 2. Changes in wheat yield components due to competition from Sonchus oleraceus. a
a Means separation was carried out by Fisher’s LSD (α = 0.05; n = 3); within columns, means followed by different letters indicate significant difference.
b Values in parentheses are mean weed densities (plants m−2) in 2016 and 2017.
The trials were conducted using a high seeding rate (60 kg ha−1) and a narrow row spacing (18 cm) to ensure maximum possible interference from wheat, and the trial practices represent those common in the northern region of Australia. Average wheat seedling density was 105 and 108 plants m−2 in 2016 and 2017, respectively. In the absence of weed interference, wheat yield exceeded 5,500 kg ha−1 in both years, representing a high output for wheat in this region. The lower wheat yield (P = 0.015) in 2017 can be attributed to less rainfall received during the crop-growing season in 2017 compared with 2016 (Figure 1). Interference from S. oleraceus caused a maximum yield reduction of 55% to 57%. Weeds from the Asteraceae family are highly competitive due to vigorous growth and their potential to exploit available resources (Brant et al. Reference Brant, Pivec, Zábranský and Hakl2012; Hassan et al. Reference Hassan, Gomaa, Fahmy, González, Hammouda and Atteya2014; Song et al. Reference Song, Kim, Im, Lee, Lee and Kim2017). The results clearly indicate a high level of interference by S. oleraceus in wheat as a winter weed resulting in a significant wheat yield penalty.
There was a hyperbolic relationship between weed density and seed production (Figure 4). Weed seed production increased as weed density increased, and there was a maximum of 1,82,940 and 1,92,657 seeds m−2 in 2016 and 2017, respectively. The amount of seed dispersal that occurred by crop harvest was 97% and 95% in 2016 and 2017, respectively (data not shown). Sonchus oleraceus flowered at 85 to 89 d after seeding, and a large proportion of seeds dispersed before harvest. Reproductive success of Asteraceae weeds are attributed to high seed production, seed dispersal by diversified means, and a prolonged reproductive phase (initial flowering to final seed dispersal) (Beckie et al. Reference Beckie, Blackshaw, Harker and Tidemann2017; Devlaeminck et al. Reference Devlaeminck, Bossuyt and Hermy2005). For many weeds, high seed retention at harvest offers an opportunity to minimize future infestations by destroying the weed seeds through seed capturing. Although rigid ryegrass (Lolium rigidum Gaudin) and wild radish (Raphanus raphanistrum L.) can produce a substantial amount of seeds, high seed retention at crop harvest gives an opportunity to reduce seedbank enrichment through employing harvest weed seed control (Walsh et al. Reference Walsh, Broster, Schwartz-Lazaro, Norsworthy, Davis, Tidemann, Beckie, Lyon, Soni, Neve and Bagavathiannan2018; Walsh and Powles Reference Walsh and Powles2014). This study indicates that seed-capturing techniques may not have similar desired results in reducing the infestation of this rapidly emerging weed in the northern region.
Figure 4. Seed production of Sonchus oleraceus in 2016 and 2017. The lines represent a hyperbolic model fit to the data by year, and symbols represent weed density. The selected model had the lowest Akaike information criterion among several commonly used models tested.
Studies exploring germination ecology of this weed indicated that it has a low level of seed persistence (Widderick et al. Reference Widderick, Walker, Sindel and Bell2010). In a study in the northern grain region, only 2% of seeds were viable at the soil surface after 6 mo, and 12% seeds remained intact at 10-cm depth after 30 mo (Widderick et al. Reference Widderick, Walker, Sindel and Bell2010). In addition, poor emergence was observed when seeds were buried more than 2 cm below the soil surface (Chauhan et al. Reference Chauhan, Gill and Preston2006) and when a thick crop residue cover was present (Manalil et al. Reference Manalil, Ali and Chauhan2018). Our results indicate that S. oleraceus can cause interference in winter wheat that leads to considerable reduction in yield. The majority of weed seeds disperse by harvest, so weed seed capture is not a feasible management option. Integrating tillage or employing residual cover along with POST and residual herbicides may help in containing this weed from further increasing in prevalence.
Acknowledgments
This work was supported by a grant from Grains Research Development Corporation (GRDC), Australia, under project UA00156. No conflicts of interest have been declared.
