Site Characteristics and Field Preparation

Two field experiments were conducted at the Flakkebjerg Research Center, Denmark (55.33°N, 11.39°E), one in 2019 (EXP₂₀₁₉) and one in 2020 (EXP₂₀₂₀). In both years, trials were carried out in a field possessing a sandy loam soil texture. Soil fertility was amended with 550 kg ha–¹ of YaraMila 21-4-10 (DLG, Axelborg, Vesterbrogade 4A, DK-1620, Copenhagen V, Denmark) to achieve a nitrogen rate of 115.5 kg ha–¹; this fertility rate was selected to ensure that nitrogen would not limit crop growth among the row spacings tested.

Experimental Design and Treatment Specifications

Experiments possessed a full factorial, randomized complete block design with four replications. Experimental factors included (1) interrow spacing at two levels: 15 (RS₁₅) and 20 cm (RS₂₀); (2) crop density at four levels: 200 (CD₂₀₀), 300 (CD₃₀₀), 400 (CD₄₀₀), and 500 plants m^–2 (CD₅₀₀); and (3) weed management treatment at three levels: plots receiving postemergence herbicide treatment (WMT_herbicide), plots receiving no herbicide treatment (WMT_weedy), and plots receiving preemergence tine harrowing (WMT_tineharrow). In total, 96 plots were present in the experiment for each year (2 interrow spacings × 4 crop densities × 3 weed management treatments × 4 blocks). Plot dimensions were 2.5 by 20 m, and a two-row spring malting barley (‘KWS Irina’) served as the test crop.

Postemergence interrow hoeing was implemented in all plots using a Schmotzer hoe (Maschinenfabrik Schmotzer GmbH, Bad Windsheim, Germany) possessing flat rigid shares, designed to limit sideward soil movement. Share width varied to accommodate differences in row spacing among treatments; 8-cm-wide shares for RS₁₅, and 13 cm for RS₂₀, resulting in a 7-cm uncultivated zone surrounding the row at both row spacings. Sinapis alba ‘Lotus’ was sown as a surrogate weed, perpendicular to crop rows, in a 2.5-m horizontal strip through the center of each plot. Surrogate weeds were sown on the same date as barley, at a target density of 40 plants m^–2. A summary of the dates on which key field operations and assessments were performed as well as crop, surrogate weed, and ambient weed growth stages at the time of implementation is provided in Table 1.

Postemergence herbicide applications consisted of Starane XL at 0.5 L ha–¹ (florasulam 1.25 g ai ha–¹ and fluroxypyr 50 g ai ha–¹, Dow AgroSciences Danmark A/S, Abedvej 39, 4920 Søllested, Denmark), Express at 10 g ha–¹ (tribenuron-methyl at 5 g ai ha–1 and tribenuron at 4.82 g ai ha–¹, Du Pont Danmark ApS, Edwin Rahrs Vej 38, DK-8220 Brabrand, Denmark), and Agropol at 0.15 L ha–¹ (spreading adhesive at 150 g ha–¹, UPL Europe, Engine Rooms, Birchwood Park, Warrington, Cheshire, WA3 YN, UK). Herbicides were applied with a custom-built unit, designed specifically for use in plot experiments. The sprayer was equipped with shields and HARDI ISO LD-110 LowDrift nozzles (RAJ & Associates, Crystal River, FL, USA); carrier volume was 150 L ha–¹, boom height was approximately 0.5 m, and pressure was 360 kPa.

Data Collection

Data collected throughout these experiments included crop density, ambient weed density, surrogate weed density, crop biomass, intrarow surrogate weed biomass, intrarow ambient weed biomass, crop yield, grain protein content, grain bulk density, and 1,000-kernel weight (TKW; g). To avoid edge effects, samples were not collected from the outermost rows on either side of each plot or the first 50 cm at the top and bottom of each plot. Achieved in-field densities of the spring barley crop, surrogate weeds, and ambient weeds were recorded before implementing herbicide or interrow hoeing treatments. All crop and weed density (plants m^–2) measures were carried out in 7 by 100 cm quadrats, centered on random sections of crop rows. For all treatments, interrow hoeing was implemented 3.5 cm from either side of the crop row, resulting in an uncultivated 7-cm-wide zone defining the intrarow zone for all data-collection purposes. In each plot, crop densities were recorded within four quadrats, intrarow surrogate weed densities were recorded within two quadrats in the surrogate weed strips, and intrarow ambient weed densities were recorded by species within four quadrats. Weed density counts were not performed in the interrow zone, because the focus of this study is to evaluate intrarow weed effects only.

Table 1.

Summary of dates, and crop (barley), surrogate weed (Sinapis alba), and ambient weed growth stages at the implementation times of critical field operations and data-collection events performed in 2019 (EXP₂₀₁₉) and 2020 (EXP₂₀₂₀).

Intrarow crop, surrogate weed, and ambient weed biomass samples were cut from six (7 by 100 cm) quadrats per plot. In WMT_herbicide plots, six cuts were made randomly throughout the plot and combined into a single sample. In WMT_weedy and WMT_tineharrow plots, four biomass cuts were made outside the surrogate weed strip, and two cuts were made inside the surrogate weed strip; samples from these two zones were kept separate. All plant biomass samples were divided into three categories: (1) crop, (2) surrogate weeds, and (3) ambient weeds. Samples were dried for at least 24 h at a temperature of 80 C and then weighed to obtain plant biomass dry matter measures.

Before barley was harvested, 50 cm from the ends of each plot was cut away, as was the surrogate weed strip in each plot's center. Because the plot width was wider than the combine head, the outermost rows on either side of each plot were not harvested. Final plot length and number of rows were recorded, and crop yields (kg ha–¹) were calculated based on harvested area. Harvested grain was cleaned, and final yield weights were standardized to 15% moisture content. Barley grain quality measures were obtained following harvest. Protein content (%) and bulk density (kg hl–¹) were measured for a subsample of whole barley grains using a near-infrared spectroscopy analyzer (Infratec^TM 1241 Grain Analyzer, Foss A/S, Nils Foss Allé 1, 3400 Hilleroed, Denmark; Buchmann et al. 2001), and TKWs (g) were extrapolated from an average of four 200-kernel weight measures taken per plot. Temperature and precipitation data were retrieved from the Danish Meteorological Institute's weather station located in Flakkebjerg, Denmark (DMI 2022).

Analysis

Experiments possessed an additive design, meaning that for every combination of weed management strategy and row spacing tested, all four levels of the continuous variable, crop density, were represented, thus permitting analysis via linear and nonlinear regression. Notably, data were analyzed separately for each year (EXP₂₀₁₉ and EXP₂₀₂₀). In both years, low crop and surrogate weed densities were observed among five plots possessing the same experimental footprint; plant density measures from each of these plots qualified as outliers, and data were excluded from analyses. Inadequate seedbeds resulting from an outcropping of stones within the field are the likely culprit for poor establishment.

Mixed model variables for the analysis of crop- and weed-response data included the random term, block, as well as fixed terms, crop density, row spacing, weed management treatment, crop density by row spacing, crop density by weed management treatment, row spacing by weed management treatment, and crop density by row spacing by weed management treatment. In all cases, achieved in-field crop density (plants m^–2) per plot served as the continuous explanatory variable during analysis.

Crop-related effects were analyzed using a linear function, and weed-related effects were analyzed using exponential functions. The fit of all regression models was confirmed via the visual assessment of plotted data points and residuals by predicted plots. In addition, comparisons of regression and ANOVA models were made via lack of fit tests, properly classifying crop density as a continuous or categorical explanatory variable (Ritz et al. 2015).

First, full models were fit, initially consisting of 6 lines (2 interrow spacings × 3 weed management treatments) with 12 corresponding parameter estimates for the analyses of crop-related effects, and 4 curves (2 interrow spacings × 2 weed management treatments) with 8 corresponding parameter estimates for analyses of weed-related effects. Second, each model was reduced stepwise, omitting all factors that had no significant effect on parameter estimates (P ≥ 0.05; stepwise reduction procedures are detailed below). Thus, for each figure and table presented, treatments described by the same regression line and parameter estimates are not statistically different, and treatments described by differing regression lines and parameter estimates are statistically different.

Across all data sets, the assumption of normally distributed residuals was evaluated using the Shapiro-Wilk test (Shapiro and Wilk 1965). In the presence of a linear or curvilinear relationship, homoscedasticity was assessed via the visual inspection of residuals by predicted plots. In the absence of a linear or curvilinear relationship, homoscedasticity was assessed using the Levene test (Levene 1960). To resolve issues of nonnormality or heteroscedasticity, log and log(x + 1) transformations were implemented using the transform-both-sides method.

Linear regressions correctly described the relationship between crop density and all crop-related effects, except for spring barley yield in EXP₂₀₁₉ and EXP₂₀₂₀, in which cases three-way ANOVAs were carried out, wherein crop density was treated as a categorical variable, and subsequent means comparisons via Tukey's HSD were made (α = 0.05). In all other instances, the relationship between crop density and crop-related effects, f₁ (x), including spring barley biomass, grain protein, grain bulk density, and TKW, were described using the linear model:

where a represents the level of the dependent crop measure when crop density equals zero, b is the change in crop measure as crop density increases, and c is crop density; r1 and r2 represent RS₁₅ and RS₂₀, and w1, w2, and w3 represent WMT_weedy, WMT_tineharrow, and WMT_herbicide, respectively. Crop response data sets were analyzed with the restricted maximum-likelihood approach in JMP® (version Pro 15.0.0, SAS Institute, Cary, NC, USA). Stepwise model reduction was warranted by nonsignificant effect tests (P ≥ 0.05) and reductions in the model Akaike information criterion (AIC).

Curvilinear relationships between crop density and weed-related effects were also confirmed via the visual assessment of plotted data points and residuals by predicted plots for the fitted regression model. In all cases, the relationship between crop density and weed biomass measures, f₂ (x), including surrogate and ambient weed biomass, was described using the exponential function:

where d represents the amount of weed biomass when crop density equals zero, and e is the rate of weed biomass reduction as crop density increases.

Table 2.

Mean spring barley crop densities achieved in 2019 (EXP₂₀₁₉) and 2020 (EXP₂₀₂₀) across treatment plots with target densities of 200 (CD₂₀₀), 300 (CD₃₀₀), 400 (CD₄₀₀), and 500 (CD₅₀₀) plants m^–2.

Percent reduction in weed biomass as crop density increases, f₃ (x), can then be calculated using the function:

Weed biomass data were analyzed using the nonlinear mixed model PROC NLMIXED in SAS® (v. 9.4, SAS Institute, Cary, NC, USA). Contrasts among parameter estimates, changes in model AIC, and nonsignificant likelihood-ratio tests (P ≥ 0.05) guided stepwise model reduction.

Intrarow Weed Biomass

Outcomes from the present study support our central hypothesis; increasing crop density reliably reduced intrarow surrogate (Figure 1; Table 3) and ambient weed biomass (Figure 2; Table 4) within the hoed cereal system. Our findings complement effects summarized by Mohler (2001), who found that increasing crop density consistently decreased weed biomass across the inter- and intrarow zones among 12 experiments in barley. Notably, the rate of intrarow weed biomass reduction (parameter e in Equations 2 and 3) was very similar across years, irrespective of weed management treatment and row spacing (Figure 3). The only instance where e deviated considerably was for EXP₂₀₁₉ ambient weed biomass in WMT_tineharrow RS₁₅ plots, when the parameter was 2.26 to 2.64 times greater compared with other estimates (Table 4). Excluding EXP₂₀₁₉ WMT_tineharrow RS₁₅ ambient weed biomass results, within the range of crop densities tested (CD₂₀₀ to CD₅₀₀), intrarow surrogate and ambient weed biomass were reduced by 49% to 82% and 53 to 86%, respectively (Figure 3). The effect that increasing seeding rate has on suppressing intrarow weeds is an important result, one that improves our understanding of how to limit potential yield and crop quality losses resulting from remaining competitive intrarow weeds in hoed cereals (Melander and McCollough 2020).

Figure 1.

The relationship between intrarow surrogate weed biomass (g m^–2; Sinapis alba) and crop density (plants m^–2; barley). Observed values represent back-transformed means for two row spacings, 15 cm (RS₁₅) and 20 cm (RS₂₀), and two weed management treatments, receiving no additional weed management treatment (WMT_weedy) and preemergence tine harrowing (WMT_tineharrow), in 2019 (A, EXP₂₀₁₉) and 2020 (B, EXP₂₀₂₀). All plots received interrow hoeing. Data underwent a log(x + 1) transformation.

Percent reduction of intrarow weed biomass was slightly higher for ambient weeds compared with surrogate weeds, comprising differences of 4.5% and 4.3% at CD₃₀₀ in EXP₂₀₁₉ and EXP₂₀₂₀, respectively. The difference in response can be attributed to variation among ambient and surrogate weed groups in their competitive response and timing of weed seedling emergence relative to crop emergence (Håkansson 2003). The ambient group is composed of numerous annual weeds differing in emergence periodicity and competitive ability; both tall and low-growing weeds were among the three most abundant species in each year. In contrast, the surrogate weed, S. alba, was selected to ensure a uniform stand of competitive brassica species, possessing a tall and erect growth habit and resembling problematic wild weeds common in organic cereals: wild radish (Raphanus raphanistrum L.), field mustard (Brassica rapa L.), and charlock mustard (Sinapis arvensis L.). Studies evaluating the effects of crop density on weed biomass originating from both intra- and interrow zones have also observed greater control among weeds emerging later than the crop and less competitive species (Håkansson 2003); therefore, it is unsurprising to observe a similar effect here. The response of both ambient and surrogate weeds is very similar; this strongly suggests that increasing crop density is a reliable management strategy for suppressing intrarow weeds of varying morphologies and competitive abilities.

Notably, the greatest reductions in intrarow weed biomass are achieved by increasing from low to moderate crop densities, with diminishing returns as seeding rate increased. For example, excluding EXP₂₀₁₉ WMT_tineharrow RS₁₅ ambient weed biomass data and rounding to the nearest whole number among all other treatments, increasing from CD₂₀₀ to CD₃₀₀ reduces intrarow surrogate and ambient weed biomass by 15%, whereas increasing from CD₃₀₀ to CD₄₀₀ and CD₄₀₀ to CD₅₀₀ corresponds with 10% and 7% reductions, respectively. Although an economic evaluation is not within the scope of the present study, the severity of weed infestation will determine whether increasing crop density beyond the standard 300 plants m^–2 is viable. For example, in organic spring barley, Kolb et al. (2010) reported that due to the high cost of organic seed, increasing seeding rate as a weed control strategy was only economically viable in situations where weed density was high.

The effect of preemergence tine harrowing on weeds was variable among years. In EXP₂₀₂₀, WMT_tineharrow reduced surrogate weed density and ambient weed density by 43% (P < 0.0001; data not shown) and 20% (P = 0.0003; data not shown) relative to WMT_weedy, respectively. This decrease in weed density corresponded with a decrease in intrarow surrogate (Figure 1; Table 3) and ambient weed biomass (Figure 2; Table 4) of 64% and 52%, respectively, at CD₃₀₀. In EXP₂₀₁₉, WMT_tineharrow did not reduce surrogate or ambient weed density relative to WMT_weedy (P ≥ 0.05; data not shown). Intrarow surrogate weed biomass was not affected by weed management treatment (Figure 1; Table 3), and a 175% increase in ambient weed biomass was observed at CD₃₀₀ for WMT_tineharrow RS₂₀ when compared with WMT_weedy RS₁₅ and RS₂₀ plots (Figure 2; Table 4). Divergent results are emblematic of the variability in treatment outcomes resulting from tine harrowing. The success of harrowing partially depends on weed community composition; both the presence of established weed seedlings before crop emergence and the absence of late-emerging weeds (Lundkvist 2009). Field and weather conditions also play a decisive role in determining weed establishment before harrowing. Harrowing may be an effective weed management strategy, killing 21% to 90% of weed seedlings present (Gallandt et al. 2018; Jabran et al. 2017); however, the uniform shallow soil disturbance caused by tine harrowing can also stimulate weed seed germination (Cirujeda and Taberner 2004; Kees 1962), especially if precipitation follows cultivation. In both experimental years, rainfall likely affected the efficacy of preemergence tine harrowing. Low precipitation in May 2020 (63% less than the 10-yr average; Supplementary Table  2), following preemergence tine harrowing (Table 1), likely contributed to the positive outcome of reduced intrarow ambient weed biomass in EXP₂₀₂₀ WMT_tineharrow plots. In May 2019, total precipitation was 231% greater compared with May 2020, likely contributing the increase in ambient weed biomass observed among EXP₂₀₁₉ WMT_tineharrow plots.

Table 3.

Estimates of parameters d and e for intrarow surrogate weed (Sinapis alba) biomass from Equation 2, where d represents the amount of weed biomass when crop density equals zero, and e is the rate of weed biomass reduction as crop density increases.^a

Figure 2.

The relationship between intrarow ambient weed biomass (g m^–2; assorted species) and crop density (plants m^–2; barley). Observed values represent back-transformed means for two row spacings, 15 cm (RS₁₅) and 20 cm (RS₂₀), and two weed management treatments, receiving no additional weed management treatment (WMT_weedy) and preemergence tine harrowing (WMT_tineharrow), in 2019 (A, EXP₂₀₁₉) and 2020 (B, EXP₂₀₂₀). All plots received interrow hoeing. Data underwent a log(x + 1) transformation.

Notably, ambient weed biomass in WMT_tineharrow RS₁₅ plots responded differently from WMT_tineharrow RS₂₀ in EXP₂₀₁₉; this is the only instance where row spacing impacted weed biomass. In this case, an interaction was present between row spacing and weed management treatment, whereby e, the rate of ambient weed biomass decline in response to increasing crop density, was estimated to be far greater than all other treatments (Figures 2 and 3; Table 4).

Table 4.

Estimates of parameters d and e for intrarow ambient weed (assorted species) biomass from Equation 2, where d represents the amount of weed biomass when crop density equals zero, and e is the rate of weed biomass reduction as crop density increases.^a

Figure 3.

Prediction of percent reduction (%) in intrarow surrogate (gray; Sinapis alba) and ambient weed biomass (black; assorted species) as crop density (plants m^–2; barley) increases. Curves are calculated on the basis of parameter estimates in Tables 3 and 4 for two row spacings, 15 cm (RS₁₅) and 20 cm (RS₂₀), and two weed management treatments, receiving no additional weed management treatment (WMT_weedy) and preemergence tine harrowing (WMT_tineharrow), in 2019 (dotted; EXP₂₀₁₉) and 2020 (solid; EXP₂₀₂₀). All plots received interrow hoeing.

While decreasing row spacing is an effective method for suppressing the growth of weeds (Mohler 2001; Regnier and Bakelana 1995); the effect of increasing crop density from 200 to 500 plants m^–2 should be far greater than that of reducing interrow row distance by 5 cm (Zimdahl 2004), as was observed in the present study. Also, the weed suppressive effect of decreasing row spacing within hoed cereals may be less significant when compared with instances in which full-width tine harrowing or no additional direct control tactics are used (Rasmussen and Svenningsen 1994). When interrow weeds are controlled by hoeing, increasing row spacing while maintaining seeding rate has opposing effects on intrarow weeds, simultaneously (1) delaying crop canopy closure and increasing light penetration into the intrarow zone (Kolb et al. 2012) and (2) increasing intrarow crop density and the crop's competitive advantage over weeds. Because light capture is a primary driver of competition in cereals (Didon and Boström 2003), wide row sowing in hoed cereals can improve growth among surviving intrarow weeds (Kolb et al. 2010; McCollough et al. 2020a).

Crop Yield and Biomass

Spring barley yield (kg ha–¹) did not increase with increasing crop density in either year. While a lack of relationship between crop yield and crop density is unexpected (Mohler 2001), it is not unfounded (O'Donovan et al. 2011; Pageau 1991). Spring barley has the capacity to tiller readily, which contributes to the crop's ability to stabilize yield. As crop density increases, the number of tillers per plant (Simmons et al. 1982), fertile heads per plant, and kernels per head decrease (Pageau 1991; Thomason et al. 2009). Due to a lack of yield response in the present study, we cannot recommend optimized seeding rates for hoed cereals based on Mohler's (2001, p. 276) definition, that optimal density is achieved when “a further increment of seed costs more than the expected increase in yield is worth.” Instead of optimizing for economic return, increasing seed costs must be weighed against other favorable outcomes, including reduced weed biomass and possible weed seedbank withdrawals (Jabbour et al. 2017; Liebman and Gallandt 1997).

Figure 4.

The effect of row spacing and weed management treatment on crop yield (g m^–2; barley). Crop density (plants m^–2) did not affect crop yield in either site year. Observed values represent means of the reduced model for two row spacings, 15 cm (RS₁₅) and 20 cm (RS₂₀), and three weed management treatments, receiving treatment with herbicide (WMT_herbicide), no additional weed management treatment (WMT_weedy), and preemergence tine harrowing (WMT_tineharrow), in 2019 (A, EXP₂₀₁₉) and 2020 (B, EXP₂₀₂₀). All plots received interrow hoeing. EXP₂₀₁₉ data underwent a log₁₀ transformation; back-transformed means are presented.

The main effect of row spacing was consistent across years; RS₂₀ yields were on average lower than RS₁₅ by 7% in EXP₂₀₁₉ and 8% in EXP₂₀₂₀ (Figure 4). Yield loss at RS₂₀ can be attributed to the intensification of intrarow crowding and intraspecific competition as row spacing increases and density is maintained (Weiner et al. 2001). For example, as row spacing widened from RS₁₅ to RS₂₀ within the present study, crop density in the intrarow zone would be expected to increase by 33%. See Supplementary Table  3 comparing full-width and intrarow target and achieved crop densities for RS₁₅ and RS₂₀ plots. Sowing at narrower row spacings typically improves crop yields; however, the effect can be inconsistent (Mohler 2001). Olsen et al. (2004) proposed that the intensity of intraspecific competition resulting from wide row sowing is reduced as weed density decreases due to the improved ability of the crop to readily access interrow resources when weeds are absent. However, in EXP₂₀₁₉, when an interaction between row spacing and weed management treatment was present, yield loss resulting from increasing interrow spacing was greatest in WMT_herbicide plots. Crop yields at RS₂₀ were 8% and 12% less than at RS₁₅ in WMT_tineharrow WMT_herbicide, respectively, and no difference was observed among row spacings for WMT_weedy. In EXP₂₀₁₉, crop yield in RS₁₅ WMT_herbicide was also 8% greater than in RS₁₅ WMT_tineharrow. Results that diverge from the scenario described by Olsen et al. (2004) reflect our working within the hoed cereal system; interrow weeds are controlled across weed management treatments with aggressive cultivation, diminishing the expected benefits of herbicide use. In EXP₂₀₂₀, an effect of weed management treatment was present: crop yields in WMT_tineharrow and WMT_herbicide were 9% and 10% less than in WMT_weedy, respectively; this outcome was not expected.

Crop biomass (g m^–2) was unaffected by treatment variables in EXP₂₀₁₉. Again, spring barley's plasticity and tillering capacity likely contributed to the absence of expected effects in this year. In EXP_2020, in alignment with expected outcomes (Håkansson 2003), barley biomass increased with increasing crop density at a rate of 35.5 g m^–2 per 100 plants m^–2 (Supplementary Figure  1; Supplementary Table  4). Crop biomass was also 6% greater in WMT_herbicide plots than in WMT_weedy and WMT_tineharrow. Crop biomass did not, however, differ among row spacings tested.

Grain Quality Parameters

Grain protein (%) in EXP₂₀₁₉ decreased slightly in response to increasing crop density, at a rate of –0.12% per 100 plants m^–2; however, no correlation was observed in EXP₂₀₂₀ (Supplementary Figure  2; Supplementary Table  5). While it was expected that grain protein would decline with increasing crop density due to the intensification of intraspecific competition, both negative and null effects among cereals have been observed (Boström et al. 2012; McKenzie et al. 2005). Weed management treatment did not affect grain protein in EXP₂₀₁₉, whereas in EXP₂₀₂₀, protein was reduced by 3% and 2% in WMT_tineharrow plots compared with WMT_weedy and WMT_herbicide, respectively. In addition, row spacing did not influence grain protein in either year. Increases in grain protein among cereal crops at wider interrow spacings have been reported, likely due to a corresponding increase in nutrient availability (Boström et al. 2012; Hiltbrunner et al. 2005; Siemens 1963); however, a difference in interrow spacing of 5 cm may not be enough to see results. The interpretation of protein response depends upon the crop's intended use. To meet malting barley quality standards in Europe, grain protein must fall within the range of 9.5% to 11.5% (Pettersson 2006); whereas, barley exceeding 11.5% is regularly sold as livestock feed (Bhatty et al. 1974). Notably, there is a financial incentive to achieve malting quality; selling price was 15% greater than for feed barley in 2020 among reporting member states of the European Union (EuroStat 2021). Given that surplus protein often prevents growers from achieving malting quality standards (Mills et al. 2021), a reduction in protein may be interpreted as a positive effect in many cases.

Barley TKW (g) response to increasing crop density (parameter b) differed among weed management treatments in EXP₂₀₁₉ (Figure 5; Table 5). An increase in 100 plants m^–2 corresponded with a TKW increase of 0.97 g and 0.35 g in WMT_weedy and WMT_tineharrow treatments, respectively; however, for WMT_herbicide, a slight decrease of –0.27 g per 100 plants m^–2 was observed. TKW is expected to decline in the absence of weeds as crop density and intraspecific competition increase (Mills et al. 2021; O'Donovan et al. 2012). However, in the presence of weeds, the negative effects of increasing crop density may be offset by the suppression of intrarow weeds. Melander and McCollough's (2020) results indicate that an increase in intrarow weed density has a greater negative impact on barley TKW than increasing crop density; as the intrarow surrogate weed S. alba increased by 100 plants m^–2, a decrease in TKW of 1.4 to 3.2 g was observed. In the present study, TKW response to row spacing differed among weed management treatments; as row spacing increased, TKW decreased in WMT_tineharrow and increased in WMT_weedy and WMT_herbicide. The observed increase in TKW as row spacing increased from RS₁₅ to RS₂₀ was also 5.8 times greater in WMT_weedy than WMT_herbicide. Inconsistent effects of row spacing on barley TKW have been previously reported (Melander et al. 2018), with both positive (Siemens 1963) and negative (Pageau 1991) effects observed. In EXP₂₀₂₀, differences in b were not observed; across treatments, TKW decreased at a rate of –0.86 g per 100 plants m^–2. Less ambient weed biomass overall in EXP₂₀₁₉ may also help to explain stronger TKW effects in EXP₂₀₂₀. On average, TKW was 2% greater in WMT_herbicide plots than in WMT_weedy and WMT_tineharrow, and no differences among row spacings were observed.

Figure 5.

The relationship between grain 1,000-kernel weight (g; barley) and crop density (plants m^–2). Observed values represent means for two row spacings, 15 cm (RS₁₅) and 20 cm (RS₂₀), and three weed management treatments, receiving treatment with herbicide (WMT_herbicide), no additional weed management treatment (WMT_weedy), and preemergence tine harrowing (WMT_tineharrow), in 2019 (A, EXP₂₀₁₉) and 2020 (B, EXP₂₀₂₀). All plots received interrow hoeing.

Grain bulk density (kg hl–¹) decreased with increasing crop density in both years, an increase in crop density of 100 plants m^–2 corresponded with a reduction of –0.606 in EXP₂₀₁₉ and –0.541 kg hl–¹ in EXP₂₀₂₀ (Supplementary Figure  3; Supplementary Table  6). Results are consistent with anticipated outcomes. Increasing a crop's population in-field typically has a negative effect on grain bulk density (Mills et al. 2021), although variable and null effects have also been observed (Pageau 1991; Thomason et al. 2009). In EXP₂₀₂₀, grain bulk density was only slightly greater in WMT_herbicide (1%) compared with WMT_weedy and WMT_tineharrow; row spacing had no effect in either year.

Additional Intrarow Weed Management Strategies in the Hoed Cereal System

In addition to preemergence tine harrowing and the adjustment of crop density and row spacing, other suitable intrarow weed management strategies exist for the hoed cereals system. Postemergence tine harrowing remains a valuable tactic to be used alone or in combination with preemergence harrowing (Brandsæter et al. 2012); however, crop damage (Lundkvist 2009) and variable efficacy (Gallandt et al. 2018; Jabran et al. 2017) exist as drawbacks to employing this tactic. Stacking hoeing and postemergence tine-harrowing cultivation events has shown promising results, providing some control in the intrarow zone while increasing efficacy in the interrow by 14% to 22% compared with hoeing alone (Gerhards et al. 2020). Combining hoeing and harrowing may also provide a greater boost in efficacy against taprooted weeds, such as scentless mayweed [Tripleurospermum maritimum ssp. inodorum (L.) Appleq.], than shallowly rooted weeds, such as S. media (Rasmussen and Sveningsen 1994).

Hoeing with rigid flat shares, as was done in the present study, limits sideward soil movement into the crop row, reducing crop soil cover and limiting intrarow weeds' burial (Machleb et al. 2018). In contrast, Gerhards et al. (2020) purposefully threw soil into the crop row by hoeing at high speeds with ducks-foot shares, achieving intrarow weed control of 21% to 54% when hoeing at 4 km h–¹ and 31% to 91% at 8 km h–¹. Furthermore, the adoption of new technologies that enable the accurate operation of hoe blades closer to the crop row will also improve weed control by reducing the area of the uncultivated intrarow zone. To attain the degree of intrarow weed control cited earlier, Gerhards et al. (2020) used improved methods for row detection (Tillet et al. 2002). As a result, they successfully hoed between 15-cm rows with 10-cm shares, leaving a narrow uncultivated 5-cm intrarow zone.

Table 5.

Estimates of parameters a and b for spring barley grain 1,000-kernel weight (TKW) from Equation 1, where a represents TKW when crop density equals zero, and b is the change in TKW as crop density increases.

A robust theoretical argument also exists for further optimizing the crop's spatial arrangement (Fischer and Miles 1973; Regnier and Bakelana 1995; Weiner et al. 2001). Instead of planting in standard rows where the crop is crowded into a narrow line, sowing in a random pattern within a 5- to 20-cm wide band may reduce competition among crop plants while increasing intraband weed suppression. However, outcomes have varied among studies evaluating band sowing and hoeing (McCollough et al. 2020a, 2020b; Speelman 1975), and further research is needed to validate the theoretical benefits of this strategy.

Methods evaluated in the present study are also highly relevant within integrated weed management (IWM). By combining herbicide use with physical and cultural control strategies (e.g., interrow hoeing, preemergence tine harrowing, and the alteration of crop density and row spacing), one can delay and reduce the risk of herbicide resistance developing (Riemens et al. 2022). For example, combining interrow hoeing and intrarow band spraying can effectively decrease herbicide use while maintaining crop yields (Abu-Hamdeh 2003; Loddo et al. 2020). Similarly, suppressing weed growth via the alteration of crop density and row spacing and implementing preemergence tine harrowing are each highly applicable for IWM programs.

In summary, increasing crop density is an effective and reliable method for controlling intrarow weeds in hoed cereals; both intrarow surrogate and ambient weed biomass were consistently reduced across years. However, increasing crop density did not improve crop yields and resulted in increased crop biomass in only one of two years. Because yield was not affected by increasing crop density, it is not possible to recommend an optimal seeding rate. It is also important to consider the effects that increasing crop density has on grain quality parameters; protein content and TKW were reduced by crop density increase in one of two years, whereas bulk density was negatively affected in both years. While increasing crop density may decrease average kernel size, malting quality may not be compromised; at higher seeding rates, barley grain kernels possess greater uniformity, an especially important quality parameter of malting barley (Edney et al. 2012; O'Donovan et al. 2011). While the observed decline in grain protein may be a negative outcome for growers producing animal feed, it is likely to be interpreted as beneficial for growers producing malting barley, for whom high protein content is often a limiting factor.

Results of preemergence tine harrowing were variable; both positive and negative effects were observed. Whether or not preemergence harrowing improves weed management outcomes will depend upon site-specific variables, including weather and soil conditions following sowing, crop growth stage, and weed species present.

Increasing interrow row spacing reduced barley yield consistently across years and did not reduce intrarow weeds. Improved yields represent an incentive for the manufacturers of automated hoeing equipment to continue to develop systems capable of accurately functioning at row spacings less than 20 cm. Higher yields may also serve as an incentive for growers to adopt narrow row sowing in the hoed cereal system once these technologies become available. In addition, furthering the advance of precision guidance equipment to function at standard row spacings of 12.5 cm would represent a significant accomplishment that would likely facilitate greater adoption of interrow hoeing among cereal growers.