Introduction
Changes in tillage practices from conventional tillage (CT) to no-tillage (NT) or reduced tillage (RT) can improve the sustainability of an agricultural ecosystem (Lal et al. Reference Lal, Kimble and Cole1999; West and Post Reference West and Post2002). Several advantages of conservation-tillage practices (NT and RT), including timely planting of crops, reduction in soil erosion and nutrient loss, retention of soil moisture, increased stable soil aggregate formation, and improved soil organic matter status, have been documented by researchers (Derpsch et al. Reference Derpsch, Friedrich, Kassam and Li2010; Pimentel et al. Reference Pimentel, Harvey, Resosudarmo, Sinclair, Kurz, Mcnair, Crist, Shpritz, Fitton, Saffouri and Blair1995; Triplett and Dick Reference Triplett and Dick2008). Thus, conservation-tillage practices have been promoted worldwide to improve soil and ecosystem sustainability. Dobberstein (Reference Dobberstein2014) reported that the area under conservation tillage in the United States has increased steadily since 1972 at an annual rate of 2.3%. Estimations made by the U.S. Department of Agriculture in 2012 showed that about 39 million ha of U.S. farmland was under conservation-tillage practices (USDA 2012). Kansas has the largest area under conservation tillage (4.21 million ha), followed by Nebraska, North Dakota, South Dakota, Iowa, and Montana (USDA 2012). In Texas, adoption of conservation tillage is very limited, with only about 0.10 million ha under NT or RT (USDA 2012).
Shifting from CT to conservation tillage can influence weed population dynamics by altering the vertical distribution of weed seeds in soil and impacting weed seedbank persistence and seedling recruitment (Farmer et al. Reference Farmer, Bradley, Young, Steckel, Johnson, Norsworthy, Davis and Loux2017; Young and Thorne Reference Young and Thorne2004). The lack of soil inversion in conservation-tillage systems may lead to the accumulation of weed seeds in the topsoil layer, thus altering their distribution in the soil profile. For instance, Refsell and Hartzler (Reference Refsell and Hartzler2009) found a higher (21 seed cm−3) waterhemp [Amaranthus tuberculatus (Moq.) Sauer] seedbank density at the 0- to 3-cm soil depth in an NT system compared with chisel plowing (10 seed cm−3). The lack of weed seed burial in the NT system may favor the persistence of small-seeded annual weeds (Moyer et al. Reference Moyer, Roman, Lindwall and Blackshaw1994; Swanton et al. Reference Swanton, Shrestha, Roy, Ball-Coelho and Knezevic1999) that are able to emerge from a shallow soil depth compared with large-seeded weeds. In a study conducted in Iowa by Leon and Owen (Reference Leon and Owen2004), greater (>4-fold) A. tuberculatus seedling recruitment was observed in the NT system compared with the CT system. Higher seedbank densities in the topsoil layer and a selection toward small-seeded annuals may subsequently lead to higher weed densities in NT compared with CT. Barberi and Lo Cascio (Reference Barberi and Lo Cascio2001) reported a greater emergence (60%) of winter annual weeds in the NT system compared with the CT system (≤ 43%). Further, the absence of tillage can promote greater persistence of perennial weeds (lack of disturbance to perennial underground structures) in the conservation-tillage systems (Barberi and Lo Cascio Reference Barberi and Lo Cascio2001; Tuesca et al. Reference Tuesca, Puricelli and Papa2001). In Iowa, Buhler et al. (Reference Buhler, Stoltenberg, Becker and Gunsolus1994) observed a higher density (215 plants 0.04 ha−1) of field bindweed (Convolvulus arvensis L.) in a 14-yr NT system compared with moldboard (105), chisel (148), or ridge (70) plowing in a corn (Zea mays L.)–soybean [Glycine max (L.) Merr.] rotation. Likewise, in a 22-yr-long study in Alaska (Conn et al. Reference Conn, Beattie and Blanchard2006), higher seedbank densities of quackgrass [Elymus repens (L.) Gould], a perennial grass species, were recorded in NT (19 seeds m−2) compared with treatments with chisel plowing (0), disking once (9), or disking twice (0) at 0-to 15-cm soil depth.
The impact of tillage regimes on weed population dynamics can be altered by specific cropping systems, and such impacts can be better understood using long-term field studies rather than short-term investigations. At Texas A&M University, a long-term grain sorghum [Sorghum bicolor (L.) Moench ssp. bicolor] experiment was initiated in 1982 to understand the impact of an NT regime on soil properties and health. However, the impact of long-term NT practices on weed population dynamics is yet to be investigated in this experiment. The objective of this study was to compare the effects of long-term NT and CT practices on weed population dynamics and yield characteristics in grain sorghum, an important agronomic crop in Texas.
Materials and Methods
Study Site and Experimental Design
A long-term field experiment was initiated in 1982 along the Brazos River floodplain at the Texas A&M field Research Facility near College Station (30.46°N, 96.43°W). The specific field experiments presented here were carried out during the 2016 and 2017 growing seasons. The soil type of the study site was Weswood silty clay loam (fine-silty, mixed, superactive, thermic Udifluventic Haplustepts) with 29% sand, 42% silt, and 29% clay, and a pH of 8.0. Two tillage treatments (CT and NT) were arranged in a randomized complete block design with four replications (plot size: 4 m by 12 m). Grain sorghum was planted in 1-m-wide rows during mid- to late March and harvested during late July to early August. Glyphosate (Roundup WeatherMax®, Bayer Crop Science LP, 2 T.W. Alexander Drive, Research Triangle Park, NC 27709) was applied as a burndown herbicide at 1,000 g ae ha−1 before planting grain sorghum in both NT and CT systems, and atrazine (Atrazine 4L, Helena Chemical, 225 Schilling Boulevard, Suite 300, Collierville, TN 38017) was applied at 1,120 g ai ha−1 at the time of grain sorghum planting in both NT and CT systems to provide PRE weed control. In the CT system, tillage was performed using a disk harrow (~15-cm depth) after crop harvest, followed by chisel plowing (20- to 25-cm depth) and a second disking before the winter season. The beds were formed subsequently. No land preparation was required at the time of grain sorghum planting in spring, except that the ridge top was knocked off using a cultivator to allow for seed placement in the moisture zone. Interrow cultivation was carried out twice during the early crop growth period for weed control in the CT plots. No soil disturbance was carried out in the NT plots. All plots were fertilized with 135 kg ha−1 nitrogen (NH4NO3) as a band application before grain sorghum planting. Weather data (maximum and minimum air temperature and precipitation) were obtained from a weather station installed near the research site.
Weed Seedbank Dynamics and Seedling Emergence
In this experiment, we studied both extractable seedbank (ESB) and germinable seedbank (GSB) to account for the weeds present in the soil seedbank as well as the ones emerging from the soil, respectively. Studying both GSB and ESB provides comprehensive information about the persistence and viability of weeds in different tillage systems. To estimate weed seedbank size (GSB) and vertical distribution pattern, soil core samples (5-cm diameter) were collected at depths ranging from 0 to 70 cm using a motorized soil auger (AMS, Main Office, 105 Harrison Street, American Falls, ID 83211) a week before grain sorghum planting. Each soil core was divided into five depths (0 to 5, 5 to 15, 15 to 30, 30 to 50, and 50 to 70 cm). The soil samples were washed under a gentle flow of water, and the weed seeds were separated using appropriate sieves (850, 425, and 90 microns). The seeds were then counted under a microscope (AM Scope, Irvine, CA), and placed into Petri dishes to determine the germination potential, followed by a viability test (1% tetrazolium chloride), as described by Patil and Dadlani (Reference Patil and Dadlani2009), conducted on the nongerminated seeds.
To determine weed seedlings emergence pattern (ESB), four quadrats (0.5 m by 0.5 m) were randomly placed within each plot between two grain sorghum rows. Weed seedling emergence was recorded at weekly intervals starting at crop planting in March through the end of June when the majority of seedling emergence was completed. The newly emerged weed seedlings at each observation timing were identified, counted, and removed from each quadrat. The quadrats were covered with a plastic sheet during herbicide applications to the plots to prevent any impact on weed seedling emergence. Total aboveground weed densities per plot were determined from four additional quadrats (0.5 m by 0.5 m) randomly placed in each plot before grain sorghum harvest.
Data Analysis
Data were subjected to ANOVA using PROC GLIMMIX in SAS (SAS Institute, 100 SAS Campus Drive, Cary, NC 27513-2414), and treatment means were separated using the Fisher’s protected Least Significant Difference (LSD) method at α = 0.05. Tillage system and year were considered as the fixed effects in the model, whereas blocks (nested within years) were considered as the random effect. Before performing ANOVA, the normality of residuals was tested using the Shapiro-Wilk test (PROC UNIVARIATE).
Weed Diversity Indices
Weed diversity indices were calculated using the emerged seedlings and weed seed density data. Species richness was calculated by counting the number of weed species present in a treatment (Clements et al. Reference Clements, Weise and Swanton1994). Weed diversity, dominance, and evenness were determined using the Shannon-Wiener index, Simpson’s index, and Pielou’s measure (Equations 1 to 3), respectively. Further, similarity values were estimated using the Jaccard index (Equation 4).
Shannon-Wiener index (Krebs Reference Krebs1985):
$$H = - \sum {p_i}{\rm{ln}}{p_i}$$
where H is the species diversity index, and pi is the proportion of the species i in total number of species.
Simpson index (Southwood Reference Southwood1978):
$$D = \sum {{({n_i}\left( {{n_i} - 1} \right))} \over {\left( {N\left( {N - 1} \right)} \right)}}$$
where n i is the number of individuals of species i, and N is the total number of individuals in a sample.
Pielou’s measure of evenness (Pielou Reference Pielou1966):
$$E = H/\ln S$$
where H is species diversity index (i.e., Shannon-Wiener index), and S is the species richness (number of weed species present in a plot).
Jaccard measure (Janson and Vegelius Reference Janson and Vegelius1981; Southwood Reference Southwood1978):
$${C_j} = {\rm{\;}}j/\left( {a + b - j} \right)$$
where j is the number of species found in both the tillage systems, a is the total number of individuals in CT, and b is the total number of individuals in NT.
Seedling Emergence Data Analysis
Seedling emergence data for each of the dominant weed species were converted into cumulative emergence (%) across the entire duration of emergence. The cumulative seedling emergence data were regressed over the accumulated growing degree days (GDD) (Equation 5) using a three-parameter sigmoidal function (Equation 6). The GDD is a time-based integral of heat accumulation (C) measured daily and is calculated using the following equation (Gilmore and Rogers Reference Gilmore and Rogers1958):
$${\rm{GDD}} = \left( {{{{T_{{\rm{max}}}}{\rm{\;}} + {\rm{\;}}{T_{{\rm{min}}}}} \over 2}} \right) - {T_{\rm{b}}}$$
where T max is the maximum air temperature, T min is the minimum air temperature, and T b is the base temperature for each weed species. Base temperatures of 8.5 C for johnsongrass [Sorghum halepense (L.) Pers.] (Arnold et al. Reference Arnold, Ghersa, Sanchez and Insausti1990), 10 C for A. tuberculatus (Uscanga-Mortera et al. Reference Uscanga-Mortera, Clay, Forcella and Gunsolus2007), and 15 C for prostrate spurge [Chamaesyce humistrata (Engelm. ex A. Gray) Small] (Asgarpour et al. Reference Asgarpour, Ghorbani, Khajeh-Hosseini, Mohammad and Chauhan2015) were used for calculating respective GDD values.
The three-parameter sigmoidal growth function (Equation 6) was fit to the seedling emergence data using SigmaPlot (v. 14.0, Systat Software, 2107 North First Street, San Jose, CA 95131-2026) and took the following form:
$$\left[ {Y = a/(1 + exp - [\left( {x - {x_0}} \right)/b} \right]$$
where Y is cumulative seedling emergence (%) at a given value of x (GDD), a is the upper asymptote (theoretical maximum for Y, normalized to 100%), x 0 is the GDD required for 50% seedling emergence, and b is the slope of the sigmoidal function at x 0.
Model Goodness of Fit
The goodness of fit for the sigmoidal function was tested by estimating the root mean-square error (RMSE) (Equation 7) and the Nash-Sutcliffe model efficiency coefficient (E f) (Equation 8). The R2 is an inadequate measure of goodness of fit for nonlinear models (Spiess and Neumeyer Reference Spiess and Neumeyer2010), but RMSE and E f could be better suited for such functions (e.g., Sarangi et al. Reference Sarangi, Irmak, Lindquist, Knezevic and Jhala2016). RMSE and E f were calculated as follows (Mayer and Butler Reference Mayer and Butler1993; Roman et al. Reference Roman, Murphy and Swanton2000):
$${\rm{RMSE}} = \left[{1 \over n}{\sum\nolimits_{i=1}^n {\left( {{P_i} - {O_i}} \right)} ^2}\right]$$
$${{{E}}_{\it{f}}}=1 - \left[ {\sum\nolimits_{i=1}^n {{{\left( {{P_i} - {O_i}} \right)}^2}/\sum\nolimits_{i=1}^n {{{\left( {{O_i} - \overline {{O_i}} } \right)}^2}} } } \right]$$
where P i is the predicted value, O i is the observed value, and n is the total number of observations. Smaller RMSE values indicate high degrees of model fit. The E f values range between −∞ to 1, and a value closer to 1 indicates a better model fit.
Sum of Square Reduction Test (SSRT)
The differences in cumulative weed seedling emergence (%) between NT and CT were examined through an SSRT (two-curve comparison), as shown by Schabenberger et al. (Reference Schabenberger, Tharp, Kells and Penner1999) for herbicide dose–response data and used in Bagavathiannan et al. (Reference Bagavathiannan, Norsworthy, Smith and Neve2012) to compare weed fecundity data. For performing this test, full (considering tillage as a factor) and reduced models (without considering tillage as a factor) were developed. Model significance was tested based on the test statistic, F obs, calculated using Equation 9:
$$\scale 73.4%{F_{\rm obs} =\\ {[({\rm SS\ Residual})_{\rm Reduced} - ({\rm SS\ Residual})_{\rm Full}]/[({\rm df\ Residual})_{\rm Reduced} - ({\rm df\ Residual})_{\rm Full}]\over ({\rm MS\ Residual})_{\rm Full}}}$$
where SS is the sum of squares, df is the total degrees of freedom, and MS is the mean squares. The calculated F obs was compared with the cutoffs from an F distribution, considering df (Residual)Reduced − df (Residual)Full as the numerator and df (Residual)Full as the denominator df.
Results and Discussion
To our knowledge, this is the first report comparing weed population dynamics between CT and NT systems in a long-term experiment running for more than 35 yr. The tillage-by-year interaction was nonsignificant (P ≥ 0.05) for weed density, weed indices, and cumulative seedling emergence; therefore, data from 2016 and 2017 were combined. The monthly maximum and minimum air temperatures were similar during the 2016 and 2017 growing seasons (May to September) (Figure 1). However, cumulative summer rainfall was greater in 2017 (884 mm) than in 2016 (659 mm).
Monthly maximum and minimum temperatures (C) and rainfall (mm) recorded in 2016 and 2017 in the long-term grain sorghum experiment at College Station, TX.
Seedbank Distribution
The vertical distribution of viable weed seeds in the soil varied between the two tillage systems (Figure 2), and in general the NT system had greater seedbank densities (8% greater) in the topsoil layer compared with the CT system (data not shown). In the NT system, a greater proportion (61% greater) of viable weed seeds (out of the total weed seeds extracted) were observed at the 0- to 5-cm depth compared with the CT system (Figure 2A and B). This corroborates Clements et al. (Reference Clements, Benott, Murphy and Swanton1996), who documented 74% of the total viable weed seeds in the top 5-cm soil profile in the NT system but only 37% in the CT system after 11 yr of a tillage experiment. This could be attributed to the minimal seed burial associated with the NT system. Seed viability levels generally declined at increasing depths irrespective of the tillage system (Figure 2), which could be due to higher levels of seed demise caused by futile germination, viability loss, and/or microbial decay (Conn et al. Reference Conn, Beattie and Blanchard2006; Darlington and Steinbauer Reference Darlington and Steinbauer1961). A relatively higher proportion of viable seeds at greater depths in NT compared with CT could be attributed to seed movement through pronounced soil cracking (PG, personal observation) and root channel formation in NT (Benvenuti Reference Benvenuti2007; Chambers et al. Reference Chambers, MacMahon and Haefner1991).
Vertical distribution of viable weed seeds as affected by 36 yr of conventional tillage (A) or no-tillage (B) in a grain sorghum experiment in College Station, TX.
Seedling Emergence Pattern
The RMSE values for the regression models describing cumulative seedling emergence of S. halepense, A. tuberculatus, and C. humistrata were generally low (ranged between 5 and 29) (Table 1), indicating a good model fit (Roman et al. Reference Roman, Murphy and Swanton2000). Further, the E f values for the cumulative emergence curves of S. halepense, A. tuberculatus, and C. humistrata were 0.9 (Table 1), also indicating a good model fit.
Parameter estimates and measures of goodness-of-fit (RMSE and E f) for the three-parameter sigmoidal function fit to cumulative weed seedling emergence as influenced by different tillage systems in a 36-yr-long grain sorghum experiment in College Station, TX.a, b
a Abbreviations: E f, modeling efficiency coefficient; GDD, growing degree days (C); NS, nonsignificant; RMSE, root mean square error; SE, standard error of the mean; SSRT, the sum of square reduction test.
b Three-parameter sigmoidal function: Y = a/(1 + exp − [(x − x0)/b)], where, Y is cumulative seedling emergence (%); A is the upper limit (theoretical maximum for Y normalized to 100%); X 0 is the GDD required for 50% seedling emergence; and B is the slope of the sigmoidal function at X 0.
c 
$F_{\rm obs} =\\ {({\rm SS\ Residual})_{\rm Reduced} - ({\rm SS\ Residual})_{\rm Full}/({\rm df\ Residual})_{\rm Reduced} - ({\rm df\ Residual})_{\rm Full}\over ({\rm MS\ Residual})_{\rm Full}}$
, Where SS is the sum of squares, df is degrees of freedom, and MS is the mean square. The calculated F obs was compared with the cutoffs from an F distribution considering df (Residual)Reduced − df (Residual)Full as the numerator and df (Residual)Full as the denominator.
The emergence pattern of S. halepense and A. tuberculatus varied between the CT and NT systems, though that of C. humistrata was comparable between the two systems (Figure 3; Table 1). In CT, model-predicted GDD values based on air temperature to obtain 50% emergence (x0) of S. halepense (P < 0.05) and A. tuberculatus (P < 0.05) were 59 and 63, respectively, whereas they were 68 and 75, respectively, in NT. Thus, there was a significant delay in the emergence of certain weed species in the NT system. Both S. halepense and A. tuberculatus are C4 plant species, and soil temperatures were often cooler than air temperatures in NT due to higher water and residue cover (Fabrizzi et al. Reference Fabrizzi, Garcıa, Costa and Picone2005), especially during the early season. This might have delayed seedling emergence. High residue accumulation in NT reflects solar radiation and alters the albedo of the soil surface, leading to a reduction in surface soil temperature (Cox et al. Reference Cox, Zobel, Van Es and Otis1990). Our findings agree with the findings of Refsell and Hartzler (Reference Refsell and Hartzler2009), wherein 50% A. tuberculatus seedling emergence was achieved at 10 d after planting in CT, whereas it took 35 d in NT. In fact, a considerable level of S. halepense and A. tuberculatus seedling emergence occurred even during the late season in NT (Figure 3).
The impact of long-term conventional tillage and no-tillage on cumulative emergence of (A) Amaranthus tuberculatus, (B) Chamaesyce humistrata, and (C) Sorghum halepense in a 36-yr-long grain sorghum experiment in College Station, TX. The growing degree days were calculated based on average air temperatures.
Weed Species Composition
A total of 12 summer and 6 winter weed species were documented in the GSB (i.e., based on seedling establishment aboveground) in the 36-yr-long NT grain sorghum plots; however, only 9 were present in the CT system (Table 2). The eight not observed in CT include annual sowthistle (Sonchus oleraceus L.), bull thistle [Cirsium vulgare (Savi) Ten.], common sunflower (Helianthus annuus L.), hoary bowlesia (Bowlesia incana Ruiz & Pav.), cutleaf evening primrose (Oenothera laciniata Hill), ivyleaf morningglory (Ipomoea hederacea Jacq.), pitted morningglory (Ipomoea lacunosa L.), and sicklepod [Senna obtusifolia (L.) Irwin & Barneby]. Most of the weeds absent in the CT system are small seeded, except common sunflower, I. lacunosa, and I. hederacea, which have difficulty germinating below the 15-cm burial depth typical of a CT system (Burton et al. Reference Burton, Mortensen, Marx and Lindquist2004; Chauhan et al. Reference Chauhan, Gill and Preston2006; Oliveira and Norsworthy Reference Oliveira and Norsworthy2006). For example, Chauhan et al. (Reference Chauhan, Gill and Preston2006) reported that S. oleraceus seedling emergence was 77% at the soil surface and drastically declined with increased soil depth and stopped at 5-cm depth. For I. lacunosa, Oliveira and Norsworthy (Reference Oliveira and Norsworthy2006) found that germination was greater (100%) at soil surface, 50% at 4-cm depth, and approximately 10% at 10-cm soil depth. In general, a lack of soil incorporation and higher soil fertility (especially higher organic carbon content; PG, unpublished data) in the NT system facilitates the germination of small-seeded weeds compared with the CT system. Sorghum halepense was the only perennial weed species observed in this study. This is perhaps due to genetic similarities between grain sorghum and S. halepense and the lack of selective herbicide options for S. halepense in grain sorghum. Further, S. halepense, C. humistrata, H. annuus, and A. tuberculatus were the dominant weed species present in both tillage systems.
Effect of tillage regimes on weed species composition in a long-term grain sorghum experiment in College Station, TX.a
a Weed species data based on 2016 and 2017 observations.
b Growth habit in southeast Texas.
c ✓ = present; × = not present.
Weed Density
Perennial Weed Density
Though S. halepense, the only perennial weed found in the experimental site, occurred in both tillage systems, average S. halepense densities were higher (28 plants m−2) in NT compared with CT (11 plants m−2) (Figure 4; Table 3). Higher densities of S. halepense in the NT system can be attributed to the lack of tillage and improved availability of soil moisture. Tillage can be an effective strategy for controlling S. halepense by exposing rhizomes to sunlight and desiccation (McWhorter and Hartwing Reference McWhorter and Hartwig1965). Conversely, an absence of tillage can allow the proliferation of perennial vegetative structures. Studies have reported higher densities of perennial weeds (spread via vegetative propagules) in NT compared with CT, and attributed this to the absence of tillage (Barberi and Lo Casio Reference Barberi and Lo Cascio2001; Hume et al. Reference Hume, Tessier and Dyck1991). Other perennial weeds did not dominate the system, likely because the herbicide program was effective in managing them.
Impact of long-term conventional-tillage (left) and no-tillage (right) on Sorghum halepense density in a 36-yr-long grain sorghum experiment in College Station, TX.
Impact of conventional tillage (CT) or no-tillage (NT) systems on population densities of Sorghum halepense, Chamaesyce humistrata, Amaranthus tuberculatus, and Lamium amplexicaule in a 36-yr-long grain sorghum experiment in College Station, TX.a
a Data were pooled between 2016 and 2017. The values with different letters are statistically different at P-value < 0.05.
Annual Weed Density
Chamaesyce humistrata and A. tuberculatus were the most commonly found summer annual weeds at the study site, whereas henbit (Lamium amplexicaule L.) was the predominant winter annual weed species. Higher densities of L. amplexicaule, C. humistrata, and A. tuberculatus (117, 4, and 19 plants m−2, respectively) were observed in the NT system compared with the CT system (45, 2, and 5 plants m−2, respectively) (Table 3). Chamaescyce humistrata, A. tuberculatus, and L. amplexicaule all are small-seeded annual weeds that have high levels of fecundity, and the seeds typically remain on the soil surface in the NT system (Table 4). Because of small seed sizes, they have a better ability to germinate and establish from shallow depths. Buhler et al. (Reference Buhler, Mester and Kohler1996) and Steckel et al. (Reference Steckel, Sprague, Stoller, Wax and Simmons2007) have reported that small-seeded annual weeds such as A. tuberculatus and redroot pigweed (Amaranthus retroflexus L.) can be predominant in an NT system. Likewise, Hill et al. (Reference Hill, Renner and Sprague2014) found high densities (10 to 65 plants m−2) of L. amplexicaule in an NT system due to this weed’s ability to germinate readily from the soil surface, supporting the findings of our research.
Average seed size (length and width) of major weeds extracted from the soil seedbank in a 36-yr-long grain sorghum experiment in College Station, TX.a
a Measurements were made using an AM Scope 40×–800× student microscope-LED.
Weed Diversity Indices
In both GSB (aboveground weed densities) and ESB evaluations (belowground seedbank densities), the Shannon-Wiener index (H) and the species richness (S) values were relatively greater in the NT system compared with the CT system (Table 5), showing that tillage had an impact on weed diversity and composition in the 36-yr-long grain sorghum experiment. The H values for the CT and NT systems, respectively, were 0.6 and 0.8 for GSB and 0.2 and 0.4 for ESB, indicating a higher number of weed species in NT than in CT (Table 5). Our findings agree with the trend observed by Legere et al. (Reference Legere, Stevenson and Benoit2011), who reported H values of 1.8 and 2.1 in CT and NT, respectively, in an 18-yr rye (Secale cereale L.) experiment in Canada. Further, the larger S values of 6.2 and 4.0 for GSB and ESB, respectively, in NT (vs. 4.2 and 3.0 in CT) in the current study indicate the generally greater weed densities in the NT system. The greater weed species diversity (H) and species richness (S) in the NT system are probably due to a relatively stable environment, longer persistence of weed seeds owing to lack of incorporation, and higher soil moisture levels compared with the CT system (Govindasamy et al. Reference Govindasamy, Mowrer, Rajan, Provin, Hons and Bagavathiannan2020). In corroboration of this, several studies have found higher H and S values in NT than in CT systems (Dorado et al. Reference Dorado, Del Monte and Lopez-Fando1999; Sosnoskie et al. Reference Sosnoskie, Herms and Cardina2006). Further, repeated tillage in the CT system affects the vertical distribution of weed seeds in the soil profile, which reduces the emergence of several weed species in CT compared with NT (Cardina et al. Reference Cardina, Herms and Doohan2002; Clements et al. Reference Clements, Benott, Murphy and Swanton1996).
Comparison of weed community dynamics indices in conventional-tillage (CT) and no-tillage (NT) systems in a 36-yr-long grain sorghum experiment in College Station, TX.
a H, Shannon-Wiener diversity index; S, species richness; D, Simpson dominance index; and E, Pielou’s measure of evenness. The mean values followed by different letters are statistically different (α = 0.05).
b Germinable seedbank represents emerged seedlings (aboveground); extractable seedbank represents weed seedbank in the soil (belowground).
** P < 0.01.
*** P < 0.05.
Both GSB and ESB evaluations revealed that there were no differences in weed species dominance (Simpson index, D) between the NT and CT systems; however, the measure of evenness (Pielou’s measure, E) differed for ESB, with a greater E value (0.3, P = 0.02) in CT compared with NT (0.2). Redistribution of weed seeds through continuous plowing in CT could have led to a higher E value compared with NT in ESB. Our findings support Pardo et al. (Reference Pardo, Cirujeda, Perea, Verdú, Mas and Urbano2019), who found a higher E value in CT (0.93) compared with NT (0.81) in a long-term (36-yr) tillage experiment in Spain. In general, the lower E values (≤0.4) in both the systems indicated the presence of few dominant weed species in this experiment, which could be attributed to the broad spectrum of activity of the herbicide program followed. Additionally, the Jaccard measure (Cj) showed that 77% of weed species were common in both tillage systems in GSB, whereas it was 82% in ESB evaluations (data not shown).
Grain Sorghum Yield
The impact of tillage systems on grain sorghum yield was weather and weed density dependent. In 2016, higher grain yield was obtained in CT (7,210 kg ha−1) than in NT (2,090 kg ha−1) (Figure 5). Due to the harder soil surface (Govindasamy et al. Reference Govindasamy, Mowrer, Rajan, Provin, Hons and Bagavathiannan2020) and higher weed densities, the establishment and growth of grain sorghum was poor in NT in 2016, while better crop establishment and lower weed densities were observed in CT. In a modeling study conducted in Texas, Ribera et al. (Reference Ribera, Hons and Richardson2004) reported greater grain sorghum yields in a CT system (4,600 kg ha−1) compared with NT system (3,940 kg ha−1). However, sorghum grain yields were comparable between the two systems in 2017, highlighting the importance of good crop establishment and growth conditions for preventing any yield reduction in NT. Findings from this 36-yr-long experiment have clearly demonstrated that tillage regime can influence weed population dynamics, with the NT system favoring greater weed densities and diversity compared with the CT system.
Sorghum grain yield as influenced by long-term tillage practices in College Station, TX. Bars topped with different letters are statistically different at α = 0.05.
The findings from this study are helpful for comprehending the response of different groups of weeds (annual, perennial, small seeded, large seeded, etc.) to the change in the level of soil disturbance. Further, an understanding of the increase or decrease in emergence and density of a particular weed species in response to crop and weed management practices will be helpful for growers to design strategic weed management programs. In particular, the NT system selected for small-seeded annual broadleaf weeds and perennials compared with the CT system (Buhler et al. Reference Buhler, Stoltenberg, Becker and Gunsolus1994; Conn et al. Reference Conn, Beattie and Blanchard2006). Long-term tillage regimes also influence the distribution of weed seeds in the soil, with the majority of weed seedlings recruiting from shallow soil depths in NT. The dominance of small-seeded annual weeds and perennials in the NT system owing to the absence of soil inversion can lead to greater competition with crops for soil moisture, space, and nutrients and eventually decrease crop yield. Therefore, growers need to alter weed management programs that effectively prevent the dominance of small-seeded annuals and perennials; a strategic deep tillage once every 5 to 10 yr will be helpful in burying weed seeds below germinable depths (Blanco-Canqui and Wortmann Reference Blanco-Canqui and Wortmann2020; Dang et al. Reference Dang, Moody, Bell, Seymour, Dalal, Freebairn and Walker2015; McGillion and Storrie Reference McGillion and Storrie2006).
The presence of viable weed seeds beyond 30-cm depth in NT even after 36 yr highlights prolonged viability of certain weed species in the soil seedbank and potential movement of weed seeds through soil cracks. Changes to weed seedling emergence periodicity mean that growers must adjust their weed management practices accordingly. The late-emerging cohorts are less likely to receive any POST application, and such escapes can add a substantial amount of seeds to the soil seedbank. Further, the lack of tillage in NT systems challenges weed control, warranting the development and implementation of robust weed management programs.
Acknowledgments
PG acknowledges funding from the Netaji Subash Chandra Bose International Fellowship offered by the Indian Council of Agricultural Research. Field assistance provided by Vince Saladino and the students and interns of the Texas A&M Weed Science Research program is gratefully acknowledged. The authors declare that no conflict of interest exists.
