Suppression of Lolium multiflorum Lam. with Vicia villosa Roth combined with residual herbicides

Lolium multiflorum Lam. (annual ryegrass) is one of the most conspicuous and invasive weed species in winter cereal crops worldwide. In Argentina, this weed affects more than 1.7 million ha of wheat fields producing yield losses of 20-30% at a density of 100 to 250 plants m-2 (Scursoni et al., 2012, 2014). In Argentinian wheat crops, annual ryegrass management is strongly dependent on pre and post-emergent herbicides, such as diclofop-methyl, clodinafop-propargyl, pinoxaden, iodosulfuron-mesosulfuron plus metsulfuron-methyl, pyroxsulam plus cloquintocet-mexyl, flucarbazone sodium and flumioxazin (Gigón et al., 2017). In addition, chemical-based management has become increasingly complex and expensive due to the occurrence of multiple resistance biotypes, mainly to EPSPS, acetyl-CoA carboxylase (ACCase) and/or acetolactate-synthase (ALS) inhibitors (Heap, 2019). In addition, herbicide ‘escapes’ or delayed emergence of annual ryegrass can easily replenish the soil seedbank due to its high fecundity (González-Andujar and Fernández-Quintanilla, 2004). Chiefly, as a result of the increment of resistance cases, Argentinian farmers are urgently seeking for alternative management options to reduce both economic losses and environmental externalities (Gigón et al., 2017; Scursoni et al., 2019).

From an Integrated Weed Management (IWM) perspective, cultural management practices can help to increase crop competitiveness (Cornelius and Bradley, 2017). In this context, the inclusion of cover crops (CC) is a viable alternative for weed suppression in agricultural rotations. Cover crops can potentially generate two distinct weed suppressive effects, by competition during their active growing period, and also due to allelopathic and/or mulching effects generated by vegetal residues that could inhibit or delay germination of many weed species in the field (Pittman et al., 2019).

Hairy vetch (HV; Vicia villosa Roth) is an annual legume, very well adapted to temperate areas of Argentina (Renzi et al., 2019). The inclusion of HV within cereal crops rotations could benefit soil fertility due to its nitrogen fixation capacity. In addition, HV has been proposed as a ‘good CC candidate’ for weed suppression in semiarid environments (Renzi and Cantamutto, 2013; Akbari et al., 2019). A negative aspect for the use of HV as CC is its low initial growth rate (Holderbaum et al., 1990) which might derive into an incomplete suppression of highly invasive weeds, such as annual ryegrass. Thus, the use of HV could be improved if combined with the application of residual herbicides during autumn before wheat sowing. So far, no studies have been conducted to evaluate neither the suppressive effect of HV nor its tolerance to commonly used residual herbicides for annual ryegrass control in the southern region of Buenos Aires province (SBA). Therefore, the objectives of this study were to evaluate: (i) annual ryegrass suppression by HV combined with autumn-applied residual herbicides, and (ii) the tolerance of HV to such herbicides.

Study sites

Field experiments were conducted on cereal crop fields of the SBA from 2017 till 2019. The experiments were performed at Coronel Dorrego (-38.832°S, -61.267°W) during 2017, Coro­nel Suárez (-37.444°S, -61.811°W) in 2018 and at Coronel Suárez, Coronel Pringles (-38.074°S, -61.158°W) and Claromecó (-38.803°S, -60.164°W) during 2019.

The soils at the experimental area are Typic argiudolls. Soils samples (0-20 cm) were taken in each experimental field and subjected to typical soil routine analysis. For C. Dorrego, the soil was a sandy loam, slightly alkaline (pH = 7.5), low organic matter (2.0%) and high in phosphorus (P) content (17.2 ppm P Bray & Kurtz). For C., the soil was a silty clay loam, neutral (pH = 6.8), medium OM (4.0 %), and high in phosphorus (P) content (18.0 ppm). For C. Pringles, the soil was a silty clay loam, neutral (pH = 7.2), medium OM (3.0 %), and high in phosphorus (P) content (19.0 ppm). For Claromecó, the soil was a sandy clay loam, neutral (pH = 6.4), medium OM (3.8 %), and high in phosphorus (P) content (26.2 ppm).

Experimental design

Hairy vetch cultivar ‘Patagonia INTA’ was sown at 25 kg pure live seed ha^-1 in rows 17.5-cm apart at a seeding depth of 20-30 mm under no tillage system. Such a sowing rate is the typical density used by farmers. HV sowing was performed during autumn at all experimental fields, on June 7^th, 2017 (C. Dorrego), March 3^rd, 2018 (C. Suárez) and March 25^th, 27^th and 29^th 2019 in C. Suárez, C. Pringles and Claromecó, respectively. Seeds were inoculated with the appropriate (commercially available) rhizobium group in a peat-based inoculum (Rhizobium leguminosarum bv viciae). Herbicides were applied one week before crop sowing in all pre-sowing cases.

The specificity of each experiment is provided below:

Experiment 1. A factorial design including twelve residual herbicides applied on HV pre-sowing (May 30^th, 2017) (table 1) under high and medium annual ryegrass infestations (867 and 373 plants m^–2) was evaluated in C. Dorrego.

Experiment 2. A factorial design including seven residual herbicides applied at HV pre-sowing or pre-emergence (February 26^th and March 3^rd, 2018 respectively) (table 1) under a low annual ryegrass infestation level (66 plants m^-2) were studied in C. Suárez.

Experiment 3. A completely randomized design (n=3) was implemented to study the tolerance of HV to the three most promising pre-emergent herbicides selected from EXP 1 and 2 (table 1). This experiment was performed during 2019 at three localities (C. Suárez, C. Pringles and Claromecó).

Table 1

List of pre-sowing and pre-emergent herbicides used in the different experiments PRES: pre-sowing and PREE: pre-emergent.

Experiments 1 and 2 were designed following a randomized complete block design with a split-plot arrangement (EXP 1: herbicides x L. multiflorum densities; EXP 2: herbicides x application timing). Main plots were herbicide treatments. Subplots were annual ryegrass densities and application timing in EXP 1 and 2 respectively. EXP 1 and 2 were carried out in 4 × 20 m plots, and EXP 3 in 3 × 5 m. Eleven herbicides were evaluated in EXP 1, seven in EXP 2, and three in EXP 3 (table 1) plus the untreated control (without herbicide). In addition, in EXP 1 a treatment without HV nor herbicide was included under a high level of weed infestation.

Herbicides were applied using self-propelled sprayers equipped with XR 8002 flat-fan nozzle tips (140 L ha⁻¹ at 117 KPa and 6 km h^-1). HV was terminated with a rolled crimper previous to corn sowing in mid-November (EXP 1) and with glyphosate+2.4-D amine previous to sunflower seeding in mid-September (EXP 2).

Data collection

In Experiments 1-2, the stand of both HV and annual ryegrass established plants were assessed 4-5 weeks after sowing (WAS) using 0.5 m² quadrats randomly distributed in each plot (n=4). HV and annual ryegrass biomass at crop termination were assessed at random using 0.16 m² quadrats (n=3). In Experiment 3, HV stand and biomass were evaluated after five and thirteen weeks after herbicide application (WHA). Samples were oven-dried at 65°C for 72 h. Percent stand and biomass reduction values were calculated by dividing the differences between treated and untreated plots by the untreated plot values.

The L. multiflorum seedling densities and biomass were plotted as a function of L. multiflorum and HV biomass, respectively. A regression analysis was performed using GraphPad Prism Software version 6.0 (GraphPad, San Diego, California, USA).

Weather data from each experiment were recorded with a meteorological station located < 1 km. Monthly rainfall totals and average monthly temperature for each year are presented in table 2.

Table 2

Accumulated monthly rainfall (mm), average monthly temperatures (°C) and sum/average of precipitation/temperature during the hairy vetch (HV) growing season (Jan-to-Oct) compared to the historic average values.

Statistical analysis

The analyses of variance (ANOVA) were performed using InfoStat software (Di Rienzo et al., 2013). For analyses, the percent stand and biomass reduction were arcsine-square root transformed, and the untransformed data are presented in the tables for clarity. Regression analysis was performed to determine the significance of the relationship between L. multiflorum density and biomass, and between annual ryegrass and HV biomass. Goodness of fit values was performed using the normalized root mean square error (NRMSE) (Araya et al., 2017).

As observed in table 2, registered rainfall in 2017 was higher than the average historical value (+11%). The high amount of precipitation registered during late summer and autumn did not allow for HV early sowing, thus the growing period was shortened to winter-spring seasons in EXP 1. During 2018-2019, rainfall values were lower than historic records; however, no detrimental effects on HV establishment were observed (EXP 2-3).

Evaluation of annual ryegrass and HV stand densities

For untreated controls, annual ryegrass densities were more than 5-fold greater in EXP 1 (867 and 373 plants m^–2 for high and medium infestation respectively) compared to EXP 2 (66 plants m^-2) (P<0.05) (figure 1). HV densities of untreated controls were 61, 83 and 52 plants m^-2 in EXP 1, EXP 2 and EXP 3, respectively.

Figure 1

Lolium multiflorum biomass (kg ha^-1) as a function of seedling density evaluated at hairy vetch (HV) termination (a), and HV biomass related to L. multiflorum biomass (b). Black indicates the untreated controls.

As observed in table 3, during 2017 (EXP 1) no significant interactions were observed between weed infestation level and herbicides treatments for both annual ryegrass (P = 0.92) and HV stand densities (P = 0.85). Among the evaluated herbicides, oxyfluorfen and acetochlor reduced ryegrass stand compared to the untreated control (table 3). The former also reduced the HV stand (P<0.01) not affecting its biomass (P>0.05). Neither acetochlor, imazamox, imazethapyr, imazapyr+imazethapyr, S-metolachlor, metribuzin nor pendimethalin affected HV stand (table 3).

Table 3

Percent reduction of hairy vetch (HV) and annual ryegrass stand densities (seedling m^-2) and biomass at crop termination as a result of residual herbicide treatments applied at pre-sowing of HV (EXP 1, 2017). Epigraph: ‘**’ indicates significance at p<0.01, ‘*’ at p<0.05 and ´ns´ not significant difference (p > 0.05). † Relative to untreated control.

The interaction between herbicide treatment and application timing was highly significant for HV stand (EXP 2, table 4). Annual ryegrass density was reduced by > 95% compared to the untreated control except for the case of diflufenican. Stand reductions of HV were < 25% (compared to the untreated control) using acetochlor, diflufenican, pyroxasulfone+flumioxazin or trifluralin (in pre-sowing) or pyroxasulfone, acetochlor and S-metolachlor (in pre-emergence) (table 4 and figure 2). However, both acetochlor and trifluralin applied in pre-emergence increased the damage over HV compared to pre-sowing applications (table 4).

Table 4.

Percent reduction of hairy vetch (HV) and annual ryegrass stand densities (seedling m^-2) and biomass at crop termination as a result of residual herbicide treatments applied at pre-sowing or pre-emergence of HV (EXP 2, 2018). Epigraph: ‘**’ indicates significance at p<0.01, ‘*’ at p<0.05 and ´ns´ not significant difference (p>0.05). † Relative to untreated control.

Figure 2

Influence of residual herbicide on hairy vetch (HV) stands densities (seedling m-2) 3 to 4 weeks after sowing and biomass reduction at 90 days after application (mean±SE) in EXP 3. ‘**’ indicates significance at p<0.01, at p<0.05, and ´ns´ not significant difference (p>0.05) compared to the untreated control.

Evaluation of annual ryegrass and HV biomass at crop termination

Annual ryegrass biomass increased with plant density. A polynomial model was adjusted to describe weed biomass related to seedling density (figure 1a). Annual ryegrass biomass was significantly influenced by herbicides and it was negatively correlated to HV biomass (figure 1b). Under a high ryegrass infestation, HV reduced the weed biomass by half (7060 vs 3627 kg ha^-1; P<0.01) without the application of herbicides. HV biomass at medium and high ryegrass infestation levels were not affected by herbicides’ treatments (P = 0.62, 2503 kg ha^-1).

Among the evaluated herbicides, acetochlor significantly reduced ryegrass biomass compared to the control (table 3, 4). For the rest of the cases, the reduction was on average < 50% in 2017 (EXP 1) but higher than 65% in 2018 (EXP 2). Acetochlor, S-metolachlor, pyroxasulfone and flumioxazin plus pyroxasulfone reduced annual ryegrass biomass > 85%, in both application timing, compared to the untreated control (table 4). The herbicides were more effective at a low-medium ryegrass infestation compared to the highest infestation level (table 3, 4). No significant differences in HV biomass among herbicide treatments, except for pre-sowing and pre-emergent applications of diflufenican or trifluralin in pre-emergence (table 3, 4; figure 2).

For the environmental conditions of the SBA, annual ryegrass densities at high, medium or even low infestation levels can reduce wheat yield in more than 60, 30 and 10% respectively (Gigón et al., 2017).

Pre-sowing and pre-emergent herbicides applied under adequate soil conditions and with low level of residues (≈ 30% cover) were capable of reducing annual ryegrass stand by 65-76% (acetochlor and oxifluorfen) in 2017, and more than 95% (acetochlor, S-metolachlor, flumioxazin, pyroxasulfone and pyroxasulfone plus flumioxazin) compared to the control in 2018 (table 4).

Our findings agree with Boutsalis et al. (2014), Cornelius and Bradley (2017), and Khalil et al. (2019) who reported that pyroxasulfone provided substantial control on annual ryegrass. Tharp and Kell (2000) observed reductions of 94% on annual ryegrass stands with S-metolachlor applications. Observed differences between suppression levels with acetochlor and S-metolachlor in 2017 and 2018 were possibly due to contrasting infestation levels. As indicated by Boutsali et al. (2014), a lower annual ryegrass density resulted in better overall control with pyroxasulfone applied at pre-sowing and pre-emergence in south-eastern Australia wheat fields.

No significant differences in the HV biomass between medium and high L. multiflorum infestation levels (P = 0.62, 2503±1020 kg ha^-1) could be explained by the principle of ´facilitation´ (Vandermeer, 1989). The presence of tendrils allows HV to climb over ryegrass plants thus improving competition for light (Aarssen et al., 1986). Highly competitive species and/or cultivars generally have better ability to access resources such as light, soil moisture and nutrients, thus suppressing the growth of weed species (Latif et al., 2019; Akbari et al., 2019).

HV was less sensitive to the evaluated herbicides than annual ryegrass. Cornelius and Bradley (2018) showed that HV proved to be one of the CC species least affected by herbicide carryover applied in soybean (e.i. S-metolachlor, acetochlor, pyroxasulfone, flumioxazin, sulfentrazone, and imazethapyr). As observed in our work, only diflufenican (at pre-sowing and pre-emergence) and trifluralin (at pre-emergence) reduced HV biomass. However, no phytotoxic effects on HV were mentioned with diflufenican in post-emergence (Renzi and Cantamutto, 2013) and trifluralin in pre-sowing (Graham, 2006). Despite some herbicides reducing the HV stand, no biomass reduction was observed at crop termination compared to the untreated control (table 3, 4; figure 2). This response may have occurred due to the compensatory growth of HV and also to some extent by the sowing density typically used by farmers in the SBA. This sowing rate almost doubles the minimal seedling densities recommended by previous study conducted by Renzi et al. (2017). Therefore, further studies should be performed to evaluate the sowing density effect on HV compensatory behaviour.

HV as cover crop plus residual herbicides provided acceptable suppression of annual ryegrass. Thus, it could be a valid alternative for cereals and summer crops rotations of the SBA. Our results point out an opportunity for growers and farmers to select HV as cover crop based on additional ecosystem benefits, mainly nitrogen fixation (Renzi et al., 2019). A weed-suppressive cover crop can limit weed seed rain, reducing population growth and ultimately weed pressure in future cash crop (Baraibar et al., 2018. Annual ryegrass seedbank viability is no more than two years (Narwal et al., 2008), so CC+residual herbicides could be a highly effective management strategy to reduce the presence of dense ryegrass infestations in the short-term.

Among the evaluated alternatives, acetochlor, S-metolachlor and pyroxasulfone showed the best performance without significant HV biomass reductions. Despite the remaining chemical options provided partial controls (30-40%) of annual ryegrass, as well as, affecting the HV stand they should not be discarded as they may allow the selection of wide spectrum of herbicides with different modes of actions (MOAs) mainly when a typical high sowing density is adopted by farmers. Alternative MOAs combinations with HV could also be used for managing ‘difficult to control weeds’ that are actually emerging in the southern region of Buenos Aires province.

The work reported in this manuscript was supported by the Instituto Nacional de Tecnología Agropecuaria (PE-E4-I086; PE-E6-I142), Criadero el Cencerro (TT 25092), Albertina S.A, Agencia Nacional de Promoción Científica y Tecnológica, MINCyT (PICT-2016-1575) and Universidad Nacional del Sur (PGI 24/A225). Appreciation is extended to Julián Castillo and Agustín Gurruchaga for assistance in establishing the field experiments.