Introduction
Palmer amaranth (Amaranthus palmeri S. Watson) is one of the most problematic weeds infesting midsouthern U.S. row crops (Riar et al. Reference Riar, Norsworthy, Steckel, Stephenson, Eubank and Scott2013) and can have severe economic impacts if left unmanaged (Bensch et al. Reference Bensch, Horak and Peterson2003; Culpepper et al. Reference Culpepper, Grey, Vencill, Kichler, Webster, Brown, York, Davis and Hanna2006; Korres et al. Reference Korres, Norsworthy and Mauromoustakos2019a). The high fecundity, extreme growth patterns, and extensive emergence period of A. palmeri are just a few biological attributes that enable the weed to be superior in competing with other plant species for available resources (Burke et al. Reference Burke, Schroeder, Thomas and Wilcut2007; Chahal et al. Reference Chahal, Irmak, Jugulam and Jhala2018; Horak and Loughin Reference Horak and Loughin2000). Additionally, the negative effects of A. palmeri emergence extend beyond the current growing season, because its high seed dispersal ensures successful establishment in multiple growing seasons (Schwartz et al. Reference Schwartz, Norsworthy, Young, Bradley, Kruger, Davis, Steckel and Walsh2016; Sparks et al. Reference Sparks, Barrentine, Burgos and McClelland2003). These characteristics, in combination with the ability of A. palmeri to evolve resistance to many herbicide sites of action (Heap Reference Heap2024; Norsworthy et al. Reference Norsworthy, Korres, Walsh and Powles2016), create a need for multiple weed management strategies (cultural, biological, chemical, and mechanical) to combat the weed (DeVore et al. Reference DeVore, Norsworthy and Brye2012; Norsworthy et al. Reference Norsworthy, Ward, Shaw, Llewellyn, Nichols, Webster, Bradley, Frisvold, Powles, Burgos, Witt and Barrett2012).
Regarding rice (Oryza sativa L.) production, a flood-irrigated system has traditionally been the dominant practice for most rice producers in the midsouthern United States (Hardke Reference Hardke2022). Nonaquatic weeds, such as A. palmeri, fail to survive season-long in flood-irrigated rice due to the anaerobic conditions associated with the production system (Norsworthy et al. Reference Norsworthy, Scott, Bangarwa, Griffith, Wilson and McCelland2011); however, a strong shift toward furrow-irrigated rice has occurred in recent years due to decreased input costs and convenient rotation to other row crops, such as soybean [Glycine max (L.) Merr.] (Hardke et al. Reference Hardke2022; Nalley et al. Reference Nalley, Massey, Durand-Morat, Shaw, Parajuli and Tsiboe2022). Amaranthus palmeri can thrive in a furrow-irrigated rice system due to an extended emergence period provided by consistently moist conditions favoring weed growth and development (Norsworthy et al. Reference Norsworthy, Scott, Bangarwa, Griffith, Wilson and McCelland2011). Therefore, creating successful weed management strategies in furrow-irrigated rice systems has become more challenging due to the lack of a continual flood helping suppress most terrestrial weeds (Bagavathiannan et al. Reference Bagavathiannan, Norsworthy and Scott2011; Norsworthy et al. Reference Norsworthy, Griffith and Scott2008).
In addition to the timing of weed emergence and specific weed species involved, weed density plays a vital role in determining crop yield loss (Swanton et al. Reference Swanton, Nkoa and Blackshaw2015). For example, research conducted by Klingaman and Oliver (Reference Klingaman and Oliver1994) demonstrated that just two A. palmeri plants m−1 of row emerging simultaneously with soybean can cause a 48% reduction in crop yield. Therefore, producers must understand weed biology and emergence behavior to successfully implement weed management programs and reduce the economic impact of problematic weeds (Jha and Norsworthy Reference Jha and Norsworthy2009; Myers et al. Reference Myers, Curran, VanGessel, Calvin, Mortensen, Majek, Karsten and Roth2004). Cultural control methods and other weed management practices have been encouraged to reduce reliance on herbicides for weed control in row-crop production systems (Chauhan and Johnson Reference Chauhan and Johnson2010; Norsworthy et al. Reference Norsworthy, Griffith, Griffin, Bagavathiannan and Gbur2014; Shekhawat et al. Reference Shekhawat, Rathore and Chauhan2020). Additionally, manipulating the competitive ability of the crop could prove beneficial in reducing crop yield loss from established weeds (Gibson et al. Reference Gibson, Fischer, Coin and Hill2002).
Crop seeding rate is an integral part of any crop production system that inevitably alters the competitive ability of both the crop and weed (Swanton et al. Reference Swanton, Nkoa and Blackshaw2015). Adjusting the crop density changes canopy formation (Korres et al. Reference Korres, Reddy, Rouse and King2019b); hence, reduced weed germination and emergence can occur due to less daily light reaching and warming the soil surface (Bell et al. Reference Bell, Norsworthy and Scott2015; Korres and Norsworthy Reference Korres and Norsworthy2017; Norsworthy Reference Norsworthy2004). Additionally, fluctuating temperatures significantly enhance the germination and emergence of small-seeded summer annuals (Baskin and Baskin Reference Baskin and Baskin1987). Consequently, crop canopy formation reduces diurnal temperature fluctuations and has a greater effect on suppressing A. palmeri than changes in light quality (Jha and Norsworthy Reference Jha and Norsworthy2009). Researchers have concluded that the germination response of many Amaranthus species, including redroot pigweed (Amaranthus retroflexus L.), smooth pigweed (Amaranthus hybridus L.), and common waterhemp [Amaranthus tuberculatus (Moq.) Sauer; syn. Amaranthus rudis Sauer), is influenced by a reduction in the red:far-red (R:FR) ratio, which can occur from crop canopy formation (Gallagher and Cardina Reference Gallagher and Cardina1998; Leon and Owen Reference Leon and Owen2003; Norsworthy Reference Norsworthy2004). Furthermore, light interception influences a weed’s competitive ability with the crop by affecting the growth and development of the weed. Therefore, altering crop planting density may assist in reducing weed interference from plants that emerge after the crop (Jha et al. Reference Jha, Norsworthy, Riley, Bielenberg and Bridges2008; Steckel et al. Reference Steckel, Sprague, Hager, Simmons and Bollero2003).
In crop–weed interactions, successful crop competition is attributed to innate biological attributes that are influenced by the presence of weeds (Jha et al. Reference Jha, Kumar, Godara and Chauhan2017). While the relationship between increased crop density and weed emergence has been examined (Bell et al. Reference Bell, Norsworthy and Scott2015), the correlation between rice plant density and weed emergence for specific weeds, such as A. palmeri, remains unknown. Additionally, the impacts of competition and yield associated with A. palmeri emergence in rice have not been investigated. Therefore, this research was conducted to (1) determine the effect of increasing rice density on A. palmeri emergence, crop canopy formation, and relative yield; and (2) quantify the influence of A. palmeri density and aboveground biomass on furrow-irrigated rice grain yield.
Materials and Methods
Effect of Rice Density on Amaranthus palmeri Emergence, Crop Canopy Cover, and Relative Yield
A field trial was initiated at the Lon Mann Cotton Research Station in Marianna, AR (34.72560°N, 90.73485°W), in 2022, and at the Milo J. Shult Research and Extension Center in Fayetteville, AR (36.09366°N, 94.17414°W), in 2023, to determine the effect of rice density on A. palmeri emergence in a furrow-irrigated rice system. The experiment was designed as a randomized complete block with four replications. The soil at the Marianna site was a Convent silt loam (coarse-silty, mixed, superactive, nonacid, thermic Fluvaquentic Endoaquepts) consisting of 9% sand, 80% silt, 11% clay, and 1.8% organic matter with a pH of 6.5, and the soil in Fayetteville was a Leaf silt loam (fine, mixed, active, thermic Typic Albaquults) composed of 18% sand, 69% silt, 13% clay, and 1.6% organic matter with a pH of 6.6. Before rice planting, the field was cultivated and shaped into 96- and 91-cm-wide beds in Marianna and Fayetteville, respectively. Plots were 1.9-m wide (two beds) by 5.2-m long and 1.8 m-wide (two beds) by 5.2-m long in Marianna and Fayetteville, respectively.
Before planting, the soil for each trial was amended for fertility based on soil test values from the Marianna Soil Test Lab recommendations. On May 11, 2022, and May 8, 2023, a hybrid, long-grain rice cultivar, ‘Full Page RT 7321FP’ (RiceTec, Alvin, TX 77512), was planted and later emerged on May 17, 2022, and May 25, 2023, respectively. Rice was planted at 0, 32, 68, 100, 137, 168, 205, 237, 274, and 310 seeds m−2 at a 1-cm depth with 19 cm between rows. The commercial standard seeding rate used by Midsouth furrow-irrigated rice producers when planting hybrid cultivars is approximately 190 seeds m−2 (Hardke et al. Reference Hardke, Barber, Bateman, Butts, Drescher, Hamilton, Henry, Mazzanti, Nicolli and Norsworthy2024).
Clomazone (Command® 3ME, FMC, Philadelphia, PA 19104) was broadcast applied immediately after rice planting at 336 g ai ha−1 at each location to provide residual control of summer annual grasses. All spray applications were made with a CO2-pressurized backpack sprayer utilizing four 10015 AIXR nozzles spaced 48 cm apart (TeeJet® Technologies, Springfield, IL 62703) and calibrated to deliver 140 L ha−1 at 276 kPa.
After planting, two 1-m2 quadrats were randomly established in each plot. Amaranthus palmeri densities were recorded from each quadrat every 7 d, beginning 1 wk after rice emergence and ending 10 wk after emergence. At 2 wk after rice emergence, rice plants in two 1-m row sections were counted in each plot. After newly emerged A. palmeri were recorded each week, the weeds were either chemically terminated from the quadrats or removed by hand if necessary. A previous greenhouse experiment confirmed that propanil did not affect the emergence of A. palmeri (JKN, personal observations). Therefore, the trial was over-sprayed with propanil (STAM®, UPL, King of Prussia, PA 19406) to keep the trial free of weeds other than A. palmeri throughout the growing season. Once the rice reached V5, the crop was irrigated every 3 d using conventional furrow-irrigated rice methods, and nitrogen was applied at a total of 135 kg N ha−1, in the form of urea (460 g N kg−1), in three separate applications at intervals of 2 wk (Barber et al. Reference Barber, Bateman, Butts, Hamilton, Henry, Lorenz, Mazzanti, Norsworthy, Roberts, Wamishe, Watkins, Hardke and Chlapecka2021).
Singular crop canopy photographs (in red-green-blue) were captured at 40 m above the crop by an unmanned aerial system (DJI Mavic 2, DJI Technology, Nanshan, Shenzhen, China 518053) at 5, 7, and 10 wk after emergence (WAE) to assess rice canopy cover. Green pixel counts from each image were measured using Field Analyzer (Green Research Services, Fayetteville, AR 72701) and used to determine crop percentage canopy cover at each evaluation timing after the removal of A. palmeri (King et al. Reference King, Norsworthy, Butts, Barber, Drescher, Fernandes and Avent2025a). The entire plot was harvested once the rice reached maturity using a Kincaid 8-XP (Kincaid, Haven, KS 67543) plot combine, with a header width of 1.8 m. Rice grain collected from each plot was adjusted to 12% moisture to report yield.
Influence of Amaranthus palmeri Density and Biomass on Furrow-irrigated Rice Yields
A field experiment was conducted at the Milo J. Shult Research and Extension Center in Fayetteville, AR (36.09301°N, 94.16882°W) in 2023 and 2024 to determine the effect of A. palmeri density and aboveground biomass on rice grain yield. The soil was a Leaf silt loam (fine, mixed, active, thermic Typic Albaquults) composed of 18% sand, 69% silt, 13% clay, and 1.6% organic matter with a pH of 6.6.
The experimental area was field cultivated before planting and shaped into 91-cm-wide beds, which then received clomazone applied preemergence at 336 g ai ha−1 to reduce annual weedy grass emergence. On April 22, 2023, and April 16, 2024, a hybrid, long-grain cultivar, Full Page RT 7321FP (RiceTec), was planted at the recommended rate of 36 seeds m−1 of row at a 1-cm depth with 18 cm between rows, and rice emerged on May 9, 2023, and April 30, 2024. Plot dimensions were established to be 3.7-m wide (four beds) by 5.2-m long, with a 0.9-m alley between replications. Once the rice plants reached the V5 growth stage, the trial was irrigated every 3d using standard furrow-irrigation techniques, and nitrogen, as urea (460 g N Kg−1), was administered at a cumulative rate of 135 kg N ha−1 in three split applications at intervals of 2 wk (Barber et al. Reference Barber, Bateman, Butts, Hamilton, Henry, Lorenz, Mazzanti, Norsworthy, Roberts, Wamishe, Watkins, Hardke and Chlapecka2021).
The experiment was a completely randomized design with a total of 80 interference plots across 2 yr (60 plots in 2023 and 20 plots in 2024) in the same field. Natural populations of A. palmeri were allowed to germinate and emerge in each plot after rice planting. Applications of fenoxaprop (Ricestar® HT herbicide, Gowan, Yuma, AZ 85364) and halosulfuron-methyl (Permit® 75WG, Gowan) were later applied to remove other undesired weed species that emerged within the trial without affecting A. palmeri establishment. Additionally, at 6 wk after rice emergence, an application of propanil (STAM®, UPL) was used to control newly emerged cotyledon A. palmeri plants under the assumption that interference was already occurring from larger plants, and these larger plants, although slightly injured would survive the application in a production scenario, further competing with the crop and producing seed. All herbicide applications were made utilizing a CO2-pressurized backpack sprayer calibrated to deliver 140 L ha−1 at 276 kPa, using four 110015 AIXR nozzles positioned 48 cm apart (TeeJet® Technologies). For comparison purposes, weed-free plots were randomly assigned to the experiment each year and which were maintained by hand weeding throughout the growing season in addition to the previously mentioned herbicide applications.
All data were collected from the center 1.8 m of each plot to mitigate interference from weeds in adjacent plots. Amaranthus palmeri density at rice harvest was determined by counting plants in a 1-m2 quadrat in each plot. Amaranthus palmeri plants were then cut at the soil surface within the quadrat, bagged, and placed in an oven at 66 C for 2 wk and dried to constant mass to record dry aboveground biomass. After A. palmeri removal, rice panicles were harvested in the quadrat using handheld rice sickles. The rice panicles were placed into an Almaco small bundle thresher (Almaco, Nevada, IA 50201) to obtain a weight that could be converted to rice grain yield. The grain yield collected was adjusted to 12% moisture.
Statistical Analyses for Effect of Rice Density on Amaranthus palmeri Emergence, Crop Canopy Cover, and Relative Yield
All data were analyzed in JMP Pro 17.0 using regression analysis (SAS Institute, Cary, NC 27513). An exponential three-parameter (3P) curve (Equation 1) was used to model the relationship between A. palmeri emergence at 5 WAE and beyond and crop canopy cover after achieving the lowest Akaike information criterion (AIC = 703.34; Figure 1).
$$Y\; = {\rm{Asymptote}} + {\rm{Scale\;}} \times {\rm{Exp}}\left( {{\rm{Growth\;rate\;}} \times {\rm{Canopy\;cover}}} \right)$$
Figure 1. Three-parameter exponential model
$\left[ {y = a + b*{\rm{exp}}\left( {c*{\rm{rice\;canopy\;cover}}} \right)} \right],$
where
$a$
= asymptote (P = 0.0916),
$b$
= scale (P < 0.0001), and
$c$
= growth rate (P = 0.0427), to determine the relationship between Amaranthus palmeri emergence at 5 wk after emergence (WAE) and beyond and rice canopy cover at 5 WAE. The solid line represents the fit of the exponential 3P model, and the dotted lines represent the 95% confidence interval of the fitted line.
where
$Y$
is the dependent variable (cumulative A. palmeri emergence m−2); asymptote, scale, and growth rate are parameters; and crop canopy percentage is the explanatory variable.
Conversely, an exponential 2P curve (Equation 2) was used to model total A. palmeri emergence as a function of rice density (AIC = 684.16; Figure 2).
$$Y\; = {\rm{Scale}} \times {\rm{Exp}}\left( {{\rm{Growth\;rate\;}} \times {\rm{Density}}} \right)$$
Figure 2. Two-parameter exponential model
$\left[ {y = a*{\rm{exp}}\left( {b*{\rm{rice\;density}}} \right)} \right],$
where
$a$
= scale (P < 0.0001) and
$b$
= growth rate (P < 0.0001) to determine the relationship between Amaranthus palmeri emergence at 5 wk after crop emergence and beyond and rice density. The solid line represents the fit of the exponential 2P model, and the dotted lines represent the 95% confidence interval of the fitted line.
where
$Y$
is the dependent variable (cumulative A. palmeri emergence m−2), scale and growth rate are parameter estimates, and rice density per square meter is the explanatory variable.
The relationship between crop canopy cover and rice density at 5, 7, and 10 WAE was nonlinear; hence, a Weibull growth model (Equation 3) was used to fit the relationship (AIC = 1,874.54; Figure 3).
$$Y\; = {\rm{Asymptote}} \times \left\{ {1 - {\rm{Exp}}\left[ { - {{\left( {{{{\rm{Density}}}}\over{{{\rm{Inflection\;point}}}}} \right)}^{{\rm{Growth\;rate}}}}} \right]} \right\}$$
Figure 3. (A–D) Weibull growth curves
$(y = a*\left\{ {1 - \exp \left[ { - {{\left( {{{{\rm{rice\;density}}}}\over{b}} \right)}^c}} \right]} \right\}$
, where
$a$
= asymptote,
$b$
= inflection point, and
$c$
= growth rate, to estimate rice canopy cover as a function of rice density at 5, 7, and 10 wk after emergence (P < 0.0001). (A–C) The individual prediction line for each canopy cover assessment and corresponding 95% confidence interval, highlighted by the solid and dotted line, respectively; (D) the predicted lines of the entire model.
where
$Y$
is the dependent variable (percent canopy cover) at each emergence timing; asymptote, inflection point, and growth rate are parameters; and rice density is the predictor variable.
Because the experiment was initiated in different locations, a three-parameter exponential model was also utilized to determine the maximum yield potential in 2022 and 2023 as a function of rice density to help account for year-to-year variation (data not shown). The analysis included Weibull, Gompertz, and logistic curves, with the exponential 3P model achieving the lowest AIC (1,285.98). The maximum yield potential in each year was 11.1 Mg ha−1 and 7.7 Mg ha−1 for 2022 and 2023, respectively, and is based on the asymptote of the model.
The rice grain yield from each plot was then made relative to the maximum yield potential within each year. To determine relative yield (%) as a function of rice density (plants m−2), Equation 1 was used to model the relationship in the Fit Curve Platform in JMP Pro 17. Within the model, relative yield is the dependent variable; asymptote, scale, and growth rate are parameters; and rice density is the predictor variable (Figure 4).
Figure 4. Three-parameter exponential model
$\left[ {y = a + b*{\rm{exp}}\left( {c*rice\;{\rm{density}}} \right)} \right],$
where
$a$
= asymptote (P < 0.0001),
$b$
= scale (P < 0.0001), and
$c$
= growth rate (P < 0.0001), to predict relative rice yield as a function of rice density. The solid line represents the fit of the exponential 3P model, and the dotted lines represent the 95% confidence interval of the fitted line.
Statistical Analyses for Influence of Amaranthus palmeri Density and Biomass on Furrow-irrigated Rice Yields
Regression analysis was used to determine the impact of A. palmeri density and aboveground biomass on furrow-irrigated rice yields in JMP Pro 17 (SAS Institute). Site-year was treated as a random effect to make universal conclusions across multiple environments (MacRae et al., Reference MacRae, Webster, Sosnoskie, Culpepper and Kichler2013; Midway Reference Midway2022). A strong nonlinear relationship occurred between rice yield loss and A. palmeri density and aboveground biomass. The maximum rice yield potential was calculated by averaging rice yields of the weed-free plots, which helped account for yield variability throughout the field. The rice grain yield from each interference plot was then made relative to the average weed-free yield to determine rice yield loss from A. palmeri interference.
To determine rice yield loss (%) as a function of A. palmeri density, two- and three-parameter curves were fit using the Fit Curve Platform in JMP Pro 17. The exponential 3P model (Equation 4) was determined to be the best fit after achieving the lowest AIC (654.51).
$${\rm{Yield\;loss\;}} = {\rm{Asymptote}} + {\rm{Scale\;}} \times {\rm{Exp}}\left( {{\rm{Growth\;rate\;}} \times {\rm{Density}}} \right)$$
where yield loss is the dependent variable; asymptote, scale, and growth rate are model parameters; and A. palmeri density is the independent variable (Figure 5). Equation 4, having the lowest AIC (560.97), was also used to describe the relationship between A. palmeri biomass and rice yield loss, with the independent variable being A. palmeri biomass (in g m−2) (Figure 6).
Figure 5. Three-parameter exponential model
$\left[ {y = a + b*{\rm{exp}}\left( {c*{\rm{weed\;density}}} \right)} \right],$
where
$a$
= asymptote (P < 0.0001),
$b$
= scale (P < 0.0001), and
$c$
= growth rate (P < 0.0001), to estimate rice yield loss as a function of Amaranthus palmeri density. The solid line represents the fit of the exponential 3P model, and the dotted lines represent the 95% confidence interval of the fitted line.
Figure 6. Three-parameter exponential model
$\left[ {y = a + b*{\rm{exp}}\left( {c*{\rm{weed\;biomass}}} \right)} \right],$
where
$a$
= asymptote (P < 0.0001),
$b$
= scale (P < 0.0001), and
$c$
= growth rate (P < 0.0001), to predict rice yield loss as influenced by Amaranthus palmeri aboveground biomass. The solid line represents the fit of the exponential 3P model, and the dotted lines represent the 95% confidence interval of the fitted line.
There appeared to be a linear relationship between A. palmeri density and aboveground biomass. Hence, a linear regression model was used to fit the relationship in the Fit Curve Platform in JMP Pro 17 after conducting a lack of fit test (P-value > 0.05) to ensure that the model was appropriate for explaining the relationship (Equation 5; Figure 7).
$$Y = a + bX$$
Figure 7. Relationship between Amaranthus palmeri biomass and density. The solid line represents the fit of the linear regression model (P < 0.0001), and the dotted lines represent the 95% confidence interval of the fitted line. R2 value displays the percentage of variability explained by the fit of the line.
where
$Y$
is the dependent variable (A. palmeri biomass in g m−2),
$b$
is the slope of the line,
$X$
is the independent variable (A. palmeri density m−2), and
$a$
is the intercept when
$X$
is equal to zero.
Results and Discussion
Amaranthus palmeri Emergence as a Function of Rice Canopy and Density
As rice canopy cover at 5 WAE increased, A. palmeri emergence beyond this point decreased (Figure 1). In both years, more than 75% of the A. palmeri emerged in the first 4 wk after rice emergence (data not shown); hence, rice density had no effect in the early stages of the growing season due to the lack of an appreciable canopy. As a result, total A. palmeri emergence data were based on weeds that emerged at 5 WAE and beyond. In this period, total A. palmeri emergence ranged from 0 to 132 plants m−2. Based on the exponential 3P model explaining the relationship (Figure 1), A. palmeri emergence from 5 WAE and beyond was reduced from 47 to 15 plants m−2 when canopy cover was 0% and 12%, respectively. However, weed emergence was not further reduced when canopy cover exceeded 12%, suggesting that A. palmeri emergence will not cease even at the maximum canopy cover of 38% at 5 WAE. Jha and Norsworthy (Reference Jha and Norsworthy2009) reported that 75% light interception by soybean canopy affected cumulative A. palmeri emergence, which ceased once soybean canopy formation reached 90%.
Total A. palmeri emergence decreased exponentially as rice density increased to 192 plants m−2, as explained by the exponential 2P model (Figure 2). Amaranthus palmeri emergence was greatest in plots not containing rice, with 53 (±6) A. palmeri m−2 emerging, on average. Conversely, a rice density of 50, 100, and 150 plants m−2 resulted in 27 (±3), 13 (±3), and 7 (±2) total A. palmeri plants m−2, respectively; however, rice densities beyond 150 plants m−2 did not aid in suppressing additional emergence of A. palmeri plants. These results display the extreme competitive ability of A. palmeri even at higher than recommended rice seeding rates and that successful A. palmeri suppression will likely depend on canopy formation as the growing season progresses rather than rice density alone early in the season. Additionally, these findings indicate that additional suppressive or management efforts beyond manipulating rice density will need to be implemented in an integrated approach for successful A. palmeri management.
Crop Canopy Cover as Influenced by Rice Density
Crop canopy cover increased as the seeding rate and subsequent rice density increased (Figure 3D). A similar relationship was shown by research conducted by Ottis and Talbert (Reference Ottis and Talbert2005), who reported that canopy cover increased linearly as rice density increased; however, the seeding rates used in that study exceeded those in this experiment, which may be attributed to differences in recommended plant densities as new rice cultivars have become available. Crop canopy cover also increased as rice matured, reaching maximum values of 38% (±4.34%), 63% (±3.06%), and 80% (± 2.53%) at 5, 7, and 10 WAE, respectively (Figure 3A–C). At a rice density of 135 plants m2, 38% canopy cover was achieved at 5 WAE. To reach 38% canopy cover at 7 and 10 WAE, only 50 and 10 plants m−2, respectively, were required, indicating that fewer plants are needed in subsequent weeks.
Impact of Rice Density on Relative Rice Yields
Rice yields were determined to be statistically greater in 2022 than in 2023 after using nonlinear regression analysis with rice density set as the independent variable (data not shown). Within the model, the 95% confidence intervals associated with the asymptote and inflection point model parameters did not overlap; however, the growth rate parameter estimates were similar across site-years according to the nonlinear regression analysis, which suggests that yield responses were analogous at both locations. Therefore, the grain yield from each plot was made relative to the maximum yield potential for each year, and the relationship between relative yield and rice density was best explained with an exponential 3P model, which accounted for 88.8% of the variation in relative yield.
Relative rice yield increased exponentially as rice density increased up to a maximum of 193 plants m−2 and 100% relative yield (Figure 4). Rice yields were 39% (±5.60%), 77% (±6.0%2), and 91% (±3.67%) of the maximum yield potential at densities of 5, 15, and 25 plants m−2, respectively, and maximum rice yield potential was achieved at 30 plants m−2. Although the ideal final plant stand for hybrid rice cultivars is 65 to 110 plants m−2 (Hardke et al. Reference Hardke, Barber, Bateman, Butts, Drescher, Hamilton, Henry, Mazzanti, Nicolli and Norsworthy2024), optimal grain yield can still be achieved at lower plant densities with appropriate crop management (Wu et al. Reference Wu, Wilson and McClung1998). Research has confirmed that tiller number is influenced by plant density and begins to decrease as rice density increases due to plants competing for available resources (Counce et al. Reference Counce, Wells and Gravois1992; Liang et al. Reference Liang, Muo and Ran1986), which subsequently has an impact on grain yield (Counce and Wells Reference Counce and Wells1990). Therefore, a suboptimal plant density still reaching maximum yield potential, as observed here, may result from rice plants compensating in other yield components such as spikelets per panicle, kernel weight, or tillering capacity (Gravois and Helms Reference Gravois and Helms1992; Miller et al. Reference Miller, Hill and Roberts1991).
Furrow-irrigated Rice Yield Loss as Influenced by Amaranthus palmeri Density and Biomass
When rice yields in interference plots were made relative to the weed-free control, the furrow-irrigated rice yield loss response as a function of A. palmeri density was best explained with a nonlinear, exponential 3P curve. With A. palmeri density set as the explanatory variable, the exponential 3P model accounted for 64.0% of the variation in rice yield loss. Previous research has identified that the relationship between weed density and crop yield loss is often nonlinear, indicating that this approach is acceptable for quantifying the interference potential of various A. palmeri densities in rice (Blackshaw et al. Reference Blackshaw, Stobbe and Sturko1981; Cousens Reference Cousens1985; Zimdahl Reference Zimdahl1980). Overall, rice yield loss increased as A. palmeri density increased up to a maximum of 47 plants m−2 and 100% yield loss (Figure 5), which is a well-understood relationship among many researchers (Burke et al. Reference Burke, Schroeder, Thomas and Wilcut2007; Knezevic et al. Reference Knezevic, Weise and Swanton1994; Massinga et al. Reference Massinga, Currie, Horak and Boyer2001).
Rice yields were reduced by 10% (±4.43%), 50 (±2.91%), and 90% (±7.75) at A. palmeri densities of 1, 7, and 24 plants m−2, respectively (Figure 5). However, the yield loss data reported here are influenced by multiple cohorts of A. palmeri emerging at various times relative to the crop throughout the growing season. Hence, A. palmeri biomass provides a better relationship with rice yield loss, because later-emerging, smaller plants produced less biomass, which cannot be accounted for when looking solely at weed density. Additionally, there were few observations of weed densities greater than 30 plants m−2; hence, the relationship is weaker due to less variation being accounted for at those higher densities. The degree of interference on rice and subsequent reduction in yield would likely be elevated if a single A. palmeri emerged with the crop due to the absence of intraspecific competition (King et al. Reference King, Norsworthy, Butts, Fernandes, Drescher and Avent2025c). Furthermore, rice yield loss responses were similar once A. palmeri densities exceeded 15 plants m−2 based on the mean 95% confidence intervals overlapping with additional plants included in the model. The response likely exists because each individual A. palmeri plant at a higher density has less impact on rice yield, which indicates intraspecific competition (Smith Reference Smith1998). The results in this study are slightly different from those observed by Rowland et al. (Reference Rowland, Murray and Verhalen1999), who reported that A. palmeri densities >8 plants m−1 of row triggered intraspecific weed interference when transplanted at the 1–true leaf growth stage within a 5-cm distance from cotton (Gossypium hirsutum L.). However, fewer A. palmeri plants seeded simultaneously near an emerging crop are likely more uniform and have greater interference potential with the crop than those emerging at various times throughout the growing season (Swanton et al. Reference Swanton, Nkoa and Blackshaw2015); hence, differences in competition thresholds between the two studies likely differ due to the variability in timing of A. palmeri seedling establishment.
There was also a strong relationship between rice yield loss and A. palmeri aboveground biomass accumulation (R2 = 0.78, P-value < 0.0001). Similar to the relationship between rice yield loss and weed density, an exponential 3P curve was best suited for describing the influence of A. palmeri biomass on crop yield loss (Figure 6). Rice yield loss increased exponentially as A. palmeri biomass increased to a maximum biomass quantity of 1,313 g m−2. Amaranthus palmeri is extremely efficient at conducting photosynthesis due to the weed being a C4 plant (Wang et al. Reference Wang, Klessig and Berry1992; Ward et al. Reference Ward, Webster and Steckel2013); therefore, A. palmeri can produce extreme quantities of biomass, which aids in the competitive success of the weed with multiple crops (Ehleringer and Forseth Reference Ehleringer and Forseth1980; Fast et al. Reference Fast, Murdock, Farris, Willis and Murray2009; Jha and Norsworthy Reference Jha and Norsworthy2009).
Amaranthus palmeri biomass of 26, 235, and 1,050 g m−2 caused 10% (±3.31%), 50% (±3.06%), and 90% (±7.81%) reduction in rice yield, respectively, based on the model predictions of the regression analysis (Figure 6). A previous experiment also identified a nonlinear relationship between A. palmeri biomass and cotton lint yield loss, which suggested that crop yield loss increases considerably once A. palmeri commences the exponential stage of vegetative development (Fast et al. Reference Fast, Murdock, Farris, Willis and Murray2009). The innate ability of A. palmeri to maximize photosynthetic rate and accelerate its life cycle provides the weed with a competitive edge over other plant species that utilize a C3 photosynthesis pathway, such as rice (Ehleringer and Forseth Reference Ehleringer and Forseth1980; Shell and Lang Reference Shell and Lang1976). Amaranthus palmeri biomass was shown to be an acceptable predictor of rice yield loss, and the correlation between biomass and crop yield loss was greater than that of weed density; however, weed density may serve as the more practical option for determining crop yield loss due to ease of implementation in the field (Klingaman and Oliver Reference Klingaman and Oliver1994).
Additionally, A. palmeri biomass increased linearly with A. palmeri density (R2 = 0.51, P-value < 0.0001) (Figure 7). The weaker correlation here is further evidence that multiple A. palmeri cohorts emerged, rather than plants only emerging with the crop. Amaranthus palmeri biomass ranged from 0 to 1,313 g m−2 and increased by 47 g m−2 for every additional A. palmeri plant from 0 to 25 plants m−2. A similar, linear relationship was reported in another study in which A. palmeri biomass increased 105 g m−2 with every single plant increase within 2 to 10 plants per 6.1 m of cotton row (MacRae et al., Reference MacRae, Webster, Sosnoskie, Culpepper and Kichler2013). The linear relationship between aboveground biomass and weed density would usually indicate a lack of intraspecific competition; however, the results observed here may be attributed to the variable emergence timings and subsequent size of A. palmeri cohorts, considering earlier-emerging plants have greater interference potential (Knezevic et al. Reference Knezevic, Weise and Swanton1994; Radosevich and Holt Reference Radosevich and Holt1984).
Because A. palmeri is among the most common and problematic weed species in the midsouthern United States (Norsworthy et al. Reference Norsworthy, Griffith, Griffin, Bagavathiannan and Gbur2014), producers must implement control strategies that minimize crop yield loss. The interference potential of A. palmeri in furrow-irrigated rice has not been explored, and the interactions that occur between the crop and weed are not well understood. Findings from this research support that A. palmeri density and aboveground biomass accumulation negatively impact crop yield loss (Rowland et al. Reference Rowland, Murray and Verhalen1999); however, assessing yield loss as a function of weed density is more practical from a producer standpoint. The results from this interference research may overestimate the impact of A. palmeri on rice yields due to the absence of preemergence herbicides affecting the earliest-emerging cohorts; however, findings from this experiment still provide growers with valuable information when looking to build management guidelines for A. palmeri in a furrow-irrigated rice system. With the yield loss data presented here, a producer could apply a herbicide based on the number of A. palmeri plants present in the field, which will potentially help mitigate severe crop yield loss through reduced biomass production and subsequent interference with the crop. Although A. palmeri fecundity was not quantified in this study, the extreme seed dispersal by the weed ensures high populations of offspring in future growing seasons that can potentially cause significant yield loss.
Regarding the rice seeding rate experiment, there was limited benefit in reducing A. palmeri emergence and increasing rice yield with increased crop seeding rates. Based on the 2024 Arkansas enterprise budget, the estimated cost for planting imidazolinone-resistant hybrid rice at a commercial seeding rate (28 kg ha−1) is US$474 ha−1; hence, an above-standard seeding rate would result in increased input costs and reduced returns. Additionally, producers should consider planting earlier, likely before A. palmeri peak emergence, which will allow the rice canopy to have greater potential to suppress the emergence of the weed. These findings also indicate that preemergence and postemergence residual herbicides will be necessary to give rice a competitive advantage until canopy closure in furrow-irrigated fields infested with A. palmeri. Fluridone has been shown to provide excellent residual control of A. palmeri when applied postemergence in a furrow-irrigated rice system (King et al. Reference King, Norsworthy, Butts, Barber, Drescher and Godar2025b). From 5 to 10 WAE, increased rice density and subsequent canopy coverage reduced A. palmeri establishment; however, none of the evaluated rice densities and canopy cover assessments prevented A. palmeri emergence. Based on these results, postemergence herbicides will also be needed to control late-emerging A. palmeri, considering weed emergence is more frequent in a furrow-irrigated rice system (Beesinger et al. Reference Beesinger, Norsworthy, Butts and Roberts2022). Overall, both experiments display the competitive nature of A. palmeri in rice and the need to adopt multiple weed management tactics to maintain rice yield.
Funding statement
This research was not supported by any funding agency, commercial enterprise, or nonprofit entity.
Competing interests
The authors declare no conflicts of interest.
Open access
