The Efficacy of Pre-Emergence Herbicides Against Dominant Soybean Weeds in Northeast Thailand

Abstract

Soybean production in Thailand faces significant challenges from malignant weed competition, potentially reducing yields by up to 37% and incurring annual economic losses of approximately USD 3.8 billion. Pre-emergence herbicides are critical for integrated weed management, but their efficacy varies depending on local conditions and soybean varieties. This study evaluates the performance of three pre-emergence herbicides, pendimethalin (1875 g a.i. ha⁻¹), s-metolachlor (900 g a.i. ha⁻¹), and flumioxazin (125 g a.i. ha⁻¹), on weed control efficiency (WCE), soybean growth, phytotoxicity, and yield in Northeast Thailand using a randomised complete block design with two varieties (CM60 and Morkhor60) across rainy (2023) and dry (2024/2025) seasons. Herbicide performance varied seasonally: s-metolachlor showed optimal rainy season results (61.54% weed control efficiency at 63 days after herbicide application (DAA), with a yield of 1036 kg ha⁻¹), while flumioxazin excelled in dry conditions (64.32% WCE, <4% phytotoxicity, and 1243 kg ha⁻¹ yield). Pendimethalin performed poorly under wet conditions but improved in drier weather. Among five dominant weed species, Cyperus rotundus proved the most resilient. CM60 demonstrated superior herbicide tolerance and yield stability, particularly under rainy conditions. These results emphasise that season-specific herbicide selection and variety matching are crucial for herbicide resistance management and effective weed control in Thailand’s rainfed soybean systems.

1. Introduction

Global demand for soybeans continues to grow, particularly for plant-based protein products [1,2]. Thailand’s soybean production is concentrated in the northern and northeastern regions, where improved varieties like CM60 have been widely adopted. CM60 is recognised for its rust resistance and high yield potential of 1875 kg ha⁻¹ with 44% protein content [3]. Recently, Khon Kaen University has developed a new soybean variety called Morkhor60, which shows promising adaptation to Thailand’s growing conditions, with yields of 1250–2063 kg ha⁻¹ across seasons [4]. However, domestic soybean production in Thailand is marginal, with annual output stagnating at around 50,000–60,000 metric tonnes due to unattractive returns compared to other field crops such as corn and cassava. As a result, Thailand continues to rely heavily on imports to meet its soybean requirements, with recent annual imports exceeding 3.4 million metric tonnes. This dependence on foreign supply is further compounded by government policy, as the cultivation of genetically modified soybeans remains strictly prohibited. These factors highlight the importance of improving domestic soybean productivity and sustainability, particularly in key agricultural regions such as Northeastern Thailand [5]. A major constraint to soybean productivity is weed competition, with potential losses ranging from 20 to 90% globally without proper control measures, which is often economically more significant than losses due to insects, pathogens, or other biotic constraints altogether [6]. Weeds such as Imperata cylindrica and Cyperus spp. compete aggressively for nutrients, water, and light while also harbouring pests and reducing post-harvest quality [7]. The critical period for weed control (CPWC) in soybeans spans 18–31 days after emergence (DAE), with weed-free conditions required for up to 61 DAE to prevent yield declines [8,9]. Traditional manual weeding, although common in Thailand, is labour-intensive (40–60% of production costs) and increasingly impractical due to labour shortages [10,11].

Pre-emergence herbicides, such as pendimethalin, s-metolachlor, and flumioxazin, offer sustainable solutions by providing residual soil activity to suppress early weed emergence [12]. These herbicides are particularly critical in light of the widespread development of glyphosate resistance and the limited efficacy of post-emergence options [13,14]. For instance, pendimethalin (1339 g a.i. ha⁻¹) effectively controls grasses and broad-leaved weeds without phytotoxicity in soybeans under suitable conditions [15], while flumioxazin (125 g a.i. ha⁻¹) can achieve up to 96.8% weed control efficiency [16]. However, herbicide efficacy and crop safety are highly variable and contingent on interactions among the herbicide, crop variety, and environmental conditions [17]. Notably, the planting season, whether rainy or dry, profoundly influences edaphoclimatic factors, such as soil moisture and rainfall patterns, which critically impact herbicide phytotoxicity and efficacy [18,19]. Under suboptimal conditions, herbicides can cause significant injury to soybean plants, leading to reduced emergence (19–73%) and yield losses [18,20].

Varietal differences further compound this variability. Early maturation soybean cultivars exhibit heightened susceptibility to herbicide injury compared to medium- or long-cycle cultivars, likely due to less time for stress recovery [21,22,23]. Concerns persist that very early maturation cultivars may be inherently more susceptible [24,25], emphasising the critical need for variety-specific evaluations under local conditions. Despite their advantages, pre-emergence herbicides remain significantly understudied for key local Thai varieties, such as CM60 and Morkhor60, particularly in relation to the distinct environmental pressures of Thailand’s rainy and dry seasons. Therefore, this study aims to evaluate the efficacy and selectivity of several pre-emergence herbicides against weeds in soybean fields, with a specific focus on the varieties CM60 and Morkhor60. The evaluation considers weed control efficiency, soybean growth, phytotoxicity, yield, and yield components, explicitly accounting for the critical influence of planting season on herbicide performance.

2. Materials and Methods

2.1. Experimental Site Description and Experimental Design

Field experiments were conducted during the rainy season of 2023 (planting date: 29 July 2023) and the dry season of 2024/2025 (planting date: 26 October 2024) at the Agronomy Field, Faculty of Agriculture, Khon Kaen University (16.47028° N, 102.80863° E) to study the effect of pre-emergence herbicides and different soybean varieties on soybean productivity, associated weed species, and some physiological characteristics under experimental site conditions. Meteorological data, including rainfall (mm), relative humidity (%), and maximum and minimum temperatures (°C), were monitored throughout both growing seasons and sourced from the Agricultural Weather Station, Faculty of Agriculture, Khon Kaen University (Figure 1). Composite soil samples (0–30 cm depth) were collected to analyse the physical and chemical properties (

Table 1. Physical and chemical characteristics of the 0–30 cm soil layer in the experimental area during the 2023 and 2024/2025 soybean crop seasons.

Soil Properties	Rainy Season of 2023	Dry Season of 2024/2025
Physical Characteristics
Sand (%)	73.36	73.90
Silt (%)	20.87	17.40
Clay (%)	5.77	8.70
Soil Texture	Sandy Loam	Sandy Loam
Chemical Characteristics
pH (1:1)	6.19	5.40
EC (1:5) (dS/m)	0.04	0.04
OM (%)	0.44	0.61
Total N (%)	0.04	0.03
Available P (mg/kg)	15.00	70.00
K (mg/kg)	103.96	71.56
Ca (mg/kg)	590.34	225.18
Mg (mg/kg)	62.39	40.05
CEC (c mol/kg)	3.00	3.36

), revealing a sandy loam texture typical of the experimental site.

The experimental site was prepared through primary tillage using a disc harrow to a depth of 30 cm, followed by land levelling to ensure uniform surface conditions. A basal application of 9.38 kg N ha⁻¹, 4.09 kg P ha⁻¹, and 7.79 kg K ha⁻¹ was administered during field preparation, followed by a top dressing 45 days after planting (DAP) supplying 23.48 kg N ha⁻¹, 10.22 kg P ha⁻¹, and 19.45 kg K ha⁻¹ based on recommendations from the Department of Agriculture, Thailand, and the site’s history of continuous soybean cultivation. Supplemental nitrogen was applied despite the soybean’s nitrogen-fixing capacity to ensure sufficient nitrogen availability during early growth, as biological fixation may be limited by soil acidity and environmental factors. Seeds were not inoculated with N-fixing bacteria because native rhizobia populations were assumed adequate due to prior soybean cropping, as well as to avoid confounding effects on yield and phytotoxicity.

Irrigation was managed using a sprinkler system, with scheduling based on soil moisture monitoring. Soybean seeds were sown at five seeds per hill and thinned to three plants per hill at 14 DAP. The experiment employed a split-plot design within a randomised complete block design (RCBD) with four replications. The main plots consisted of five weed management treatments: weed-free control (hand weeding), weedy control (weedy), and three pre-emergence herbicides (pendimethalin, s-metolachlor, and flumioxazin) applied one DAP (

Table 2. Herbicides and rates used for the treatment of herbicide selectivity.

Treatment	Class	Active Ingredient	Trade Name	Doses
			Manufacturer	(g a.i. ha−1) */
Hand Weeding	−	−	−	−
Weedy	−	−	−	−
Pendimethalin	K1	45.5% w/v CS	Prowl CS/BASF (Bangkok, Thailand)	1875
S-metolachlor	K3	96% w/v EC	Dualgold/Syngenta (Samut Prakan, Thailand)	900
Flumioxazin	E	50% WP	Zumizoya/TJC (Samut Prakan, Thailand)	125

*^/ grams of active ingredient per hectare; hand weeding = plots were kept free of weeds throughout the experiment via manual hoeing; weedy = plots were maintained without weed control. Class = classification of herbicide mechanisms of action according to the Weed Science Society of America (WSSA) and the Herbicide Resistance Action Committee (HRAC) (K1: inhibitors of microtubule assembly; K3: inhibitors of synthesis of very long-chain fatty acids; E: inhibitors of protoporphyrinogen oxidase). CS: capsule suspension; EC: emulsifiable concentrate; WP: wettable powder; and w/v: weight per volume.

). Subplots comprised two soybean cultivars (Morkhor60 and CM60), with individual plots measuring 5 × 5 m and plant spacing of 0.50 m between rows and 0.25 m within rows.

Pre-emergence herbicides were uniformly applied using a calibrated 15 L knapsack sprayer with a flooding fan nozzle (500 L ha⁻¹ spray volume) one day after planting (DAP) during the rainy season of 2023 (30 July 2023) and the dry season of 2024/2025 (27 October 2024). During spraying operations, the air temperature, relative humidity, and soil temperature at a depth of 5 cm were continuously monitored using a mini digital temperature humidity metre (UNI-T UT333) and a glass thermometer. Applications occurred during the early morning under low wind conditions (<5 km h⁻¹), with solutions vigorously agitated before use. Real-time measurements during the 2023 rainy season applications showed averages of 34 ± 4 °C (air), 48 ± 9% RH, and 24 ± 1 °C (soil), while the 2024/2025 dry season exhibited averages of 35 ± 2 °C (air), 62 ± 6% RH, and 31 ± 0.6 °C (soil).

The experiment was conducted on soils with naturally acidic pH levels (5.40–6.19), reflecting typical conditions in Northeastern Thailand’s rainfed farmlands, where sandy, weathered soils and tropical climate result in pH values of 4.5–6.0 and liming is uncommon [26,27]. While optimal soybean growth occurs at pH levels of 6.5–7.0 [28,29], some genotypes—including those used in this study—are adapted to acidic, sandy soils and can perform well without pH adjustment [4]. Thus, maintaining these pH levels ensures the experiment’s relevance to local farming practices and provides realistic insights into herbicide efficacy and crop response.

2.2. Data Collection

2.2.1. Studies on Weeds

Weed parameters, including weed density (plants m⁻²), weed biomass (g m⁻²), the summed dominance ratio (SDR), and weed control efficiency (WCE), were evaluated using a rectangular sampling frame measuring 0.5 × 1 m (0.5 m²). Following the methodology of Aekrathok et al. [30], two random samplings were conducted per plot for each evaluation interval at 21, 35, 49, and 63 days after herbicide application (DAA), with sampling locations systematically varied to avoid re-sampling previously assessed areas. All weed specimens within each quadrat were identified at the species level, counted to obtain weed density, and then oven-dried at 80 °C for 72 h to determine the dry biomass. The SDR values were calculated based on established formulas by Janiya and Moody [31], as cited in Hasan et al. [32]. The five most dominant weed species, as determined via the SDR analysis, were further characterised (

Table 3. Scientific, common, and family names of dominant weeds in soybean crops during the 2023 and 2024/2025 growing seasons.

Categories	Scientific Name	Common Name	Family
Broad-leaved weeds	Trianthema portulacastrum L.	Horse purslane	Aizoaceae
	Oldenlandia corymbosa L.	Diamond flower	Rubiaceae
Grassy weeds	Dactyloctenium aegyptium (L.) P. B.	Crowfoot grass	Gramineae
	Digitaria ciliaris (Retz.) Koeler.	Summer grass	Gramineae
Sedge weeds	Cyperus rotundus L.	Purple nutsedge	Cyperaceae

).

$$\mathrm{SDR}={\displaystyle \frac{\mathrm{Relative} \mathrm{density} \left(\mathrm{RD}\right)+\mathrm{relative} \mathrm{dry} \mathrm{weight} \left(\mathrm{RDW}\right)}{2}}$$

(1)

where

$$\mathrm{RD}={\displaystyle \frac{\mathrm{Density} \mathrm{of} \mathrm{a} \mathrm{given} \mathrm{species}}{\mathrm{Total} \mathrm{density}}}\times 100$$

(2)

$$\mathrm{RDW}={\displaystyle \frac{\mathrm{Dry} \mathrm{weight} \mathrm{of} \mathrm{a} \mathrm{given} \mathrm{species}}{\mathrm{Total} \mathrm{dry} \mathrm{weight}}}\times 100$$

(3)

Weed control efficiency measures the effectiveness of any weed control treatment in comparison to no weed control treatment. Mathematically, it can be expressed as follows [32,33]:

$$\mathrm{WCE} (\%)={\displaystyle \frac{\mathrm{Weed} \mathrm{biomass} \mathrm{in} \mathrm{weedy} \mathrm{plots}-\mathrm{Weed} \mathrm{biomass} \mathrm{in} \mathrm{treated} \mathrm{plots}}{\mathrm{Weed} \mathrm{biomass} \mathrm{in} \mathrm{weedy} \mathrm{plots}}}\times 100$$

(4)

2.2.2. Agronomic Trait

The soybean germination rate (%) was calculated as the proportion of emerged seeds relative to the total sown, evaluated at 3, 5, 7, and 14 DAA [34]. Five plants were randomly selected to measure plant height and the number of soybean nodes at the critical period of soybean plants at 21, 35, 49, and 63 DAA. Herbicide phytotoxicity symptoms on soybeans were observed periodically using the European Weed Research Council (EWRC) rating scale, with damaged plant rates of 0% (no effect on plants) to 100% (total loss of plants and yield), as described by [35].

2.2.3. Yield and Yield Component

Harvesting occurred at physiological maturity, which was determined as 90–95 DAP for CM60 and 95–100 DAP for Morkhor60, depending on the respective environmental conditions during the season of planting. At this stage, five plants per plot were randomly sampled to quantify the pod number (branches and main stem), seed number per pod, and 100-seed weight (g) (determined from a random subsample of harvested seeds). The plant population was determined by counting the number of plants within the harvest area and extrapolating to plants per hectare (plants ha⁻¹). Grain yield was measured from a harvest area of 2 × 2 m and converted to kg ha⁻¹. Grain yield loss (%) was calculated as the percentage reduction in yield compared to the hand weeding treatment using the following formula:

$$\mathrm{Grain} \mathrm{yield} \mathrm{loss} (\%)={\displaystyle \frac{\mathrm{Yield} \mathrm{in} \mathrm{hand} \mathrm{weeding}-\mathrm{Yield} \mathrm{in} \mathrm{weed} \mathrm{control}}{\mathrm{Yield} \mathrm{in} \mathrm{hand} \mathrm{weeding}}}\times 100$$

(5)

2.3. Statistical Analysis

The data were analysed using analysis of variance (ANOVA) to investigate variations in treatment effects, such as weed control efficiency, soybean toxicity, yield, and yield components. All traits were analysed for homogeneity of variance using Levene’s test method, and the results are shown in Table S1. The relationships between the analysed variables (traits) were tested using Pearson correlation coefficients (Figure S1). The data underwent statistical analysis using the Statistix^® version 10.0 (1985–2013) tool (Analytical Software, Tallahassee, FL, USA). Treatment mean differences were determined using the LSD at a 5% probability level.

3. Results

3.1. Effects of Pre-Emergence Herbicides in Weed Data

3.1.1. Weed Control Efficiency

The efficacy of pre-emergence herbicides against weeds in soybean crops exhibited pronounced seasonal variability (

Table 4. Effects of pre-emergence herbicides and soybean varieties on weed control efficiency (%) in the rainy and dry seasons.

Variable	Weed Control Efficiency (%)
	Rainy Season of 2023				Dry Season of 2024/2025
	21 DAA 1/	35 DAA	49 DAA	63 DAA	21 DAA	35 DAA	49 DAA	63 DAA
Weed Control (WC)
Hand weeding	100.00 ± 0.00 a 2/	100.00 ± 0.00 a	100.00 ± 0.00 a	100.00 ± 0.00 a	100.00 ± 0.00 a	100.00 ± 0.00 a	100.00 ± 0.00 a	100.00 ± 0.00 a
Weedy	0.00 ± 0.00 */ d	0.00 ± 0.00 d	0.00 ± 0.00 c	0.00 ± 0.00 d	0.00 ± 0.00 d	0.00 ± 0.00 d	0.00 ± 0.00 c	0.00 ± 0.00 d
Pendimethalin	54.54 ± 8.26 c	41.48 ± 12.37 c	20.22 ± 12.86 c	10.03 ± 13.39 d	65.50 ± 4.80 c	65.94 ± 5.24 c	77.41 ± 7.66 b	81.49 ± 5.90 b
S-metolachlor	87.21 ± 3.96 ab	79.75 ± 2.35 ab	48.30 ± 11.04 b	61.54 ± 6.78 b	85.32 ± 2.14 b	82.59 ± 4.07 b	72.79 ± 3.64 b	64.88 ± 3.92 c
Flumioxazin	76.44 ± 12.55 b	65.17 ± 8.18 b	55.06 ± 8.37 b	29.96 ± 11.59 c	66.30 ± 6.98 c	80.16 ± 5.33 b	78.50 ± 6.83 b	64.32 ± 9.93 c
Variety (Var)
Morkhor60	60.27 ± 9.24	54.63 ± 8.41	33.49 ± 9.72 b	32.15 ± 10.76 b	63.65 ± 8.43	67.40 ± 8.40	63.47 ± 8.45	55.77 ± 8.35 b
CM60	67.01 ± 8.89	59.93 ± 9.30	55.94 ± 8.16 a	48.46 ± 8.39 a	63.20 ± 7.97	64.08 ± 8.12	68.01 ± 8.28	68.51 ± 8.22 a
F-test
Weed Control (WC)	** 4/	**	**	**	**	**	**	**
Variety (Var)	ns	ns	**	* 3/	ns	ns	ns	**
WC × Var	ns	ns	**	ns 5/	ns	ns	ns	ns
CV WC	30.94	32.71	51.91	43.10	18.33	16.96	25.20	19.04
CV WC × Var	25.14	31.52	25.60	59.76	16.18	17.55	18.84	22.47

*^/ mean ± SE, ^1/ Days after application, ^2/ the same letters are not significantly different by LSD at p < 0.05, ^3/ significant at p < 0.05, ^4/ significant at p < 0.01, and ^5/ not significant.

and Figure 2). Manual hand weeding consistently achieved 100% weed control efficiency (WCE) at all assessment intervals (21–63 DAA), serving as the experimental control benchmark. All herbicide treatments demonstrated highly significant differences (p < 0.01) compared to the weedy check in both seasons. In the rainy season of 2023, s-metolachlor (87.21 ± 3.96%, ab) and flumioxazin (76.44 ± 12.55%, b) showed statistically similar efficacy at 21 DAA (p > 0.05), with both outperforming pendimethalin (54.54 ± 8.26%, c; p < 0.05). The superior early performance of s-metolachlor despite heavy rainfall likely accelerated degradation (Figure 2a), although by 63 DAA, its WCE declined to 61.54 ± 6.78% (b) while remaining significantly higher than that of flumioxazin (29.96 ± 11.59%, c; p < 0.05). Pendimethalin’s efficacy plummeted to 10.03 ± 13.39% (d) by 63 DAA, which is consistent with its known susceptibility to leaching under heavy rainfall conditions (Figure 2a).

During the dry season of 2024/2025, s-metolachlor maintained strong initial control (85.32 ± 2.14%, b at 21 DAA), although its efficacy was statistically comparable to pendimethalin (65.50 ± 4.80%, c; p > 0.05) and flumioxazin (66.30 ± 6.98%, c) at this stage. Remarkably, pendimethalin’s persistence improved under drier conditions, achieving 81.49 ± 5.90% (b) WCE by 63 DAA, which was significantly higher than those of s-metolachlor (64.88 ± 3.92%, c) and flumioxazin (64.32 ± 9.93%, c; p < 0.05) (Figure 2b). This reversal highlights how soil moisture modulates herbicide longevity, with pendimethalin’s lipophilic nature favoring adsorption in dry soils.

The soybean variety CM60 consistently enhanced herbicide performance in later growth stages, achieving significantly higher WCE than Morkhor60 at 49 DAA (55.94 ± 8.16% vs. 33.49 ± 9.72%; p < 0.01) and 63 DAA (48.46 ± 8.39% vs. 32.15 ± 10.76%; p < 0.05) during the rainy season (Figure 2c). This trend persisted during the dry season (63 DAA: 68.51 ± 8.22% vs. 55.77 ± 8.35%; p < 0.01) (Figure 2d), suggesting that CM60’s competitive traits (e.g., canopy closure) may suppress late-season weed resurgence. No significant interactions (p > 0.05) between herbicides and varieties were detected, indicating consistent varietal effects across treatments.

3.1.2. Summed Dominance Ratio (%)

Analysis of the summed dominance ratio (SDR) in untreated (weedy) plots showed clear seasonal shifts in weed community composition in Northeast Thailand, with weeds categorised as broad-leaved, grasses, and sedges (

Table 5. Summed dominance ratio (%) in weedy plots at 21, 35, 49, and 63 DAA during the rainy season of 2023 and the dry season of 2024/2025.

Weed Species	Summed Dominance Ratio of Weed Species (%)
	Rainy Season of 2023				Dry Season of 2024/2025
	21 DAA 1/	35 DAA	49 DAA	63 DAA	21 DAA	35 DAA	49 DAA	63 DAA
Broad-Leaved
Oldenlandia corymbosa L.	- 2/	2.19	11.72	6.51	0.03	4.16	9.53	12.03
Trianthema portulacastrum L.	49.65	34.88	11.46	4.18	2.88	1.7	1.38	1.55
Lindernia ciliate (Colsm.) Pennell	-	-	0.37	0.98	0.11	-	0.93	0.19
Cleome rutidosperma	0.23	0.15	0.00	0.00	0.08	0.09	0.25	0.03
Praxelis clematidea R.M.King & H.Rob.	4.80	0.75	2.47	1.66	-	0.04	0.24	0.00
Indigofera hirsuta L.	-	-	-	-	-	0.00	0.14	0.04
Alternanthera sessilis	0.55	0.35	0.27	1.02	0.05	0.23	0.09	0.13
Ipomoea gracillis R.Br.	-	-	-		0.04	0.00	0.04	0.03
Amaranthus viridis L.	1.14	0.46	0.29	0.37	-	0.00	0.00	0.00
Xanthium strumarium L.	-	-	-	-	0.00	0.00	0.00	-
Wrighia arborea (Dennst.) Mabb.	-	-	-	-	0.03	-	0.00	-
Borreria alata (Aubl.) DC.	-	-	-	-	-	-	0.00	-
Ipomoea pestigridis L.	-	-	-	-	0.03	0.04	-	-
Sida cordifolia L.	-	-	-	-		0.00	-	-
Phyllanthus amarus Schumach. & Thonn.	-	-	0.00	-	-	-	-	-
Grasses
Digitaria ciliaris (Retz.) Koeler	29.09	38.28	38.19	37.04	88.51	88.28	81.58	81.24
Dactyloctenium aegyptium (L.) P. B.	0.43	6.02	23.61	32.39	0.00	0.08	0.86	1.65
Eleusine indica (L.) Gaertn.	-	0.67	-	0.00	0.55	0.19	0.21	0.10
Cynodon dactylon (L.) Pers.	-	-	-	-	0.00	1.68	0.00	0.12
Eragrostis pectinacea (Michx.) Nees	-	-	-	-	-	-	0.08	-
Sedges
Cyperus rotundus L.	14.11	16.26	6.44	12.9	7.70	3.53	3.60	2.48
Cyperus compressus L.	-	-	5.17	2.98	-	-	-	-
Cyperus esculentus L.	-	-	-	-	-	-	0.46	0.22
Cyperus difformis L.	-	-	-	-	-	-	0.60	0.21
Total	100.00	100.00	100.00	100.00	100.00	100.00	100.00	100.00

^1/ Days after application, ^2/ No weed species were detected during the observation period.

). SDR quantifies each species’ contribution to the total weed flora, highlighting the most problematic weeds under natural conditions. Across both seasons, the five most dominant species were Digitaria ciliaris (Retz.) Koeler, Trianthema portulacastrum L., Cyperus rotundus L., Dactyloctenium aegyptium (L.) P. B., and Oldenlandia corymbosa L. The specific names and families of each dominant weed species are shown in Table 3.

During the 2023 rainy season, T. portulacastrum was initially dominant (49.65% at 21 DAA) but declined rapidly to 4.18% by 63 DAA. Meanwhile, D. ciliaris increased from 29.09% to 37.04% over the same period, becoming the main competitor later in the season. C. rotundus maintained a notable presence (14.11–16.26% early, above 6% later), while D. aegyptium rose sharply in late season (from 0.43% to 32.39% at 63 DAA). O. corymbosa also increased at specific times, peaking at 11.72% at 49 DAA. In contrast, during the 2024/2025 dry season, D. ciliaris overwhelmingly dominated throughout (over 88% at 21 and 35 DAA, above 81% at 63 DAA), reflecting its adaptability to dry conditions. The other four species contributed much less: O. corymbosa was the only broad-leaved species to increase notably late in the season (from 0.025% to 12.03%), while C. rotundus, D. aegyptium, and T. portulacastrum remained at low but persistent levels.

These findings are highly relevant for the development of effective pre-emergence herbicide programmes. The rapid shift from broad-leaved to grass weed dominance in the rainy season and the persistent grass dominance in the dry season suggest that herbicide selection must prioritise the residual control of D. ciliaris and D. aegyptium while also considering the early emergence of broad-leaved weeds like T. portulacastrum. The continued presence of C. rotundus further emphasises the need for herbicides with activity against sedges or for integrated management approaches.

3.1.3. Weed Density and Weed Biomass

An analysis of weed density and weed biomass in the field trials revealed clear differences among weed control treatments and across seasons, providing important insights into the efficacy of pre-emergence herbicides against dominant weed species in soybean cultivation in Northeast Thailand. In both the rainy and dry seasons, hand weeding consistently resulted in zero weed density and biomass for all assessed species at every observation interval, confirming its effectiveness as a benchmark for complete weed suppression. In contrast, untreated (weedy) plots exhibited the highest weed densities and biomasses, reflecting the natural competitive pressure from the local weed flora (

Table 6. Effect of weed control treatments and soybean varieties on weed density at 21, 35, 49, and 63 DAA.

Variable	Weed Density (plants m−2)
	Broad-Leaved								Grasses								Sedges
	Trianthema portulacastrum L.				Oldenlandia corymbosa L.				Digitaria ciliaris (Retz.) Koeler.				Dactyloctenium aegyptium (L.) P.B.				Cyperus rotundus L.
	21 DAA 1/	35 DAA	49 DAA	63 DAA	21 DAA	35 DAA	49 DAA	63 DAA	21 DAA	35 DAA	49 DAA	63 DAA	21 DAA	35 DAA	49 DAA	63 DAA	21 DAA	35 DAA	49 DAA	63 DAA
Weed Control	Rainy Season of 2023
Hand weeding	0.00 b 2/	0.00 c	0.00 b	0.00 b	0.00	0.00	0.00 b	0.00 b	0.00 b	0.00 b	0.00 b	0.00 b	0.00	0.00 b 2/	0.00 b	0.00 b	0.00 c	0.00 c	0.00 b	0.00 b
Weedy	141.13 a	87.50 a	20.37 a	3.75 b	0.00	7.50	33.50 a	7.50 a	88.62 a	78.00 a	47.12 a	29.12 a	1.25	10.00 a	18.50 a	16.12 a	36.87 b	43.75 bc	14.12 b	11.87 b
Pendimethalin	0.87 b	0.62 c	0.00 b	0.75 b	0.00	0.00	0.00 b	0.75 b	0.12 b	1.37 b	4.25 b	2.75 b	0.00	0.25 b	0.12 b	0.12 b	73.50 a	109.12 a	83.87 a	94.24 a
S-metolachlor	26.50 b	35.12 b	26.37 a	12.12 a	0.00	0.00	0.25 b	0.12 b	4.00 b	0.50 b	1.87 b	2.75 b	0.00	0.00 b	0.00 b	2.00 b	5.87 c	9.25 c	20.37 b	18.00 b
Flumioxazin	2.00 b	1.12 c	6.00 b	3.00 b	0.00	0.00	0.00 b	0.00 b	8.50 b	2.37 b	9.87 b	7.37 b	0.00	0.75 b	1.37 b	2.87 b	22.00 bc	61.75 ab	71.75 a	73.87 a
Variety
Morkhor60	24.4	21.55	9.60	4.75	0.00	3.00	6.65	1.85	23.50	13.95	12.80	10.95	0.25	3.85 a	3.40	3.30	26.25	46.40	39.60	46.15
CM60	43.75	28.20	11.45	3.10	0.00	0.00	6.85	1.50	17.00	18.95	12.45	7.60	0.25	0.50 b	4.60	5.15	29.05	43.15	37.25	33.05
Weed Control	Dry Season of 2024/2025
Hand weeding	0.00 b	0.00 b	0.00 b	0.00 b	0.00	0.00 b	0.00 b	0.00 b	0.00 b	0.00 b	0.00 c	0.00 c	0.00	0.00	0.00	0.00	0.00 d	0.00 c	0.00 b	0.00 c
Weedy	19.88 a	8.63 b	5.88 b	5.00 b	0.13	21.75 a	57.50 a	55.25 a	363.88 a	301.75 a	200.12 a	151.50 a	0.00	0.25	2.00	4.25	14.13 bc	11.38 bc	13.75 b	8.50 bc
Pendimethalin	0.00 b	0.63 b	0.00 b	0.50 b	0.00	0.00 b	0.13 b	0.00 b	5.50 b	5.25 b	12.63 c	18.00 c	0.13	0.00	0.38	0.00	25.88 a	53.88 a	51.75 a	32.63 a
S-metolachlor	7.88 b	22.63 a	16.38 a	12.25 a	0.00	5.00 b	14.88 b	12.75 b	14.50 b	48.75 b	78.00 b	69.88 b	0.00	0.00	1.50	3.25	6.00 cd	3.75 bc	9.75 b	3.13 c
Flumioxazin	0.38 b	0.25 b	0.63 b	1.38 b	0.00	0.00 b	0.00 b	0.00 b	15.75 b	33.00 b	65.00 b	56.00 b	1.63	0.00	3.50	2.63	20.50 ab	19.75 b	25.38 ab	16.88 b
Variety
Morkhor60	5.25	8.50	5.10	4.85 a	0.05	4.05	19.70	22.15 a	79.15	80.70	74.60	45.35 b	0.20	0.00	0.15 b	0.00 b	13.65	16.60	15.40	10.30
CM60	6.00	4.35	4.05	2.80 a	0.00	6.65	9.30	5.05 b	80.70	74.80	67.70	72.80 a	0.50	0.10	2.80 a	4.05 a	12.95	18.90	24.85	14.15

The data were measured per square metre (1 m²). ^1/ Days after application; ^2/ the same letters are not significantly different based on LSD at p < 0.05.

and

Table 7. Effect of weed control treatments and soybean varieties on weed biomass at 21, 35, 49, and 63 DAA.

Variable	Weed Density (plants m−2)
	Broad-Leaved								Grasses								Sedges
	Trianthema portulacastrum L.				Oldenlandia corymbosa L.				Digitaria ciliaris (Retz.) Koeler.				Dactyloctenium aegyptium (L.) P.B.				Cyperus rotundus L.
	21 DAA 1/	35 DAA	49 DAA	63 DAA	21 DAA	35 DAA	49 DAA	63 DAA	21 DAA	35 DAA	49 DAA	63 DAA	21 DAA	35 DAA	49 DAA	63 DAA	21 DAA	35 DAA	49 DAA	63 DAA
Weed Control	Rainy Season of 2023
Hand weeding	0.00 b 2/	0.00 c	0.00 b	0.00 b	0.00	0.00	0.00 b	0.00	0.00 b	0.00 b	0.00 c	0.00 c	0.00	0.00.b 2/	0.00 b	0.00 b	0.00 b	0.00 c	0.00 c	0.00 c
Weedy	17.34 a	35.25 a	13.16 b	4.46 b	0.00	0.25	2.20 a	2.45	7.27 a	49.16 a	68.21 a	78.02 a	0.10	8.46 a	56.05 a	92.83 b	3.85 b	16.01 c	4.77 c	21.83 c
Pendimethalin	0.03 b	0.97 c	0.00 b	4.38 b	0.00	0.00	0.00 b	0.15	6.25 b	1.47 b	25.68 b	9.00 bc	0.00	0.36 b	0.17 b	0.37 b	13.47 a	53.68 a	58.36 a	148.10 a
S-metolachlor	2.47 b	15.67 b	48.16 a	35.52 a	0.00	0.00	0.67 b	0.01	0.20 b	0.22 b	2.76 bc	9.66 bc	0.00	0.00 b	0.00 b	0.37 b	0.76 b	2.50 c	5.71 c	14.76 c
Flumioxazin	0.18 b	1.20 c	5.31 b	7.08 b	0.00	0.00	0.00 b	0.00	0.66 b	1.73 b	9.77 bc	23.55 b	0.00	0.71 b	7.53 b	7.91 b	4.27 b	34.61 b	40.28 b	98.86 b
Variety
Morkhor60	3.24	10.13	14.65	11.29	0.00	0.10	0.84	0.17	1.90	9.48	22.01	27.80 a	0.03	3.06 a	9.05	17.71	4.28	23.19	25.37	70.79
CM60	4.77	11.10	12.00	9.29	0.00	0.00	0.31	0.87	1.35	11.55	20.56	20.29 b	0.00	0.75 b	16.45	22.87	4.66	19.53	18.28	42.63
Weed Control	Dry Season of 2024/2025
Hand weeding	0.00 b	0.00	0.00 b	0.00 b	0.00	0.00 b	0.00 b	0.00 b	0.00 b	0.00 b	0.00 d	0.00 c	0.00 b	0.00	0.00	0.00 b	0.00 c	0.00 c	0.00 b	0.00 c
Weedy	0.79 a	0.81	0.79 b	0.57 b	0.01	1.32 a	2.79 a	2.34 a	48.09 a	164.90 a	238.59 a	194.22 a	0.00 b	0.10	2.01	4.15 a	6.21 bc	4.94 bc	4.72 b	2.64 bc
Pendimethalin	0.00 b	2.09	0.00 b	0.03 b	0.00	0.00 b	0.01 b	0.00 b	3.43 b	11.27 b	9.05 cd	15.77 c	0.05 ab	0.00	4.22	0.00 b	15.41 a	40.31 a	28.63 a	11.32 a
S-metolachlor	0.33 ab	3.88	4.91 a	3.74 a	0.00	0.26 b	0.75 b	0.51 b	4.04 b	20.11 b	57.94 b	59.93 b	0.00 b	0.00	0.29	1.83 ab	3.40 c	1.70 bc	3.81 b	0.83 c
Flumioxazin	0.03 b	0.11	0.05 b	0.15 b	0.00	0.00 b	0.00 b	0.00 b	4.01 b	20.17 b	34.41 bc	53.88 b	0.95 a	0.00	1.13	1.47 b	12.88 ab	11.85 b	9.94 b	5.07 b
Variety
Morkhor60	0.17	1.53	1.02	1.36	0.01	0.29	0.60	0.91 a	12.49	49.30	46.64 b	56.79 b	0.13	0.00	1.69	0.00 b	7.75	12.25	5.75	3.90
CM60	0.29	1.22	1.28	0.44	0.00	0.34	0.82	0.23 b	11.34	37.27	89.36 a	72.73 a	0.28	0.04	1.37	2.98 a	7.41	11.27	13.09	4.05

The data were measured per square metre (1 m²). ^1/ Days after application, and ^2/ the same letters are not significantly different by LSD at p < 0.05.

).

In the rainy season of 2023, the application of pre-emergence herbicides significantly reduced both weed density and biomass compared to the untreated control, although the degree of suppression varied depending on the herbicide and species. S-metolachlor was particularly effective in reducing the densities of both broad-leaved and grass weeds, with T. portulacastrum density dropping to 12.12 plants m⁻² and D. ciliaris to 2.75 plants m⁻² by 63 DAA (Table 6). The corresponding biomasses were 35.52 g m⁻² and 9.66 g m⁻², respectively, indicating substantial, but not complete, suppression. Flumioxazin also reduced the densities and biomasses, particularly of broad-leaved weeds, but was less consistent against grasses and sedges at the later stages. Pendimethalin showed limited efficacy during the rainy season, particularly against sedges, as reflected by high C. rotundus densities (94.24 plants m⁻² at 63 DAA) and biomasses (148.1 g m⁻² at 63 DAA). These patterns closely parallel the WCE data (Table 4 and Figure 2), where s-metolachlor and flumioxazin provided higher and more sustained weed control efficiency compared to pendimethalin, especially in the early and mid-season assessments.

In the dry season of 2024/2025, overall weed pressure for broad-leaved species was lower but much higher for grasses, particularly D. ciliaris, which reached densities of 363.88 plants m⁻² at 21 DAA and maintained high biomass throughout (peaking at 238.59 g m⁻² at 49 DAA in the untreated plots) (Table 7). Pre-emergence herbicides again reduced weed density and biomass, with s-metolachlor and flumioxazin showing the greatest efficacy. For example, s-metolachlor reduced D. ciliaris density to 69.88 plants m⁻² and biomass to 59.93 g m⁻² at 63 DAA, while flumioxazin achieved similar reductions. Pendimethalin, however, was less effective during the dry season, as evidenced by the higher late-season densities and biomasses of grasses and sedges. These results align with the WCE data (Table 4 and Figure 2), where s-metolachlor and flumioxazin maintained high weed control efficiency, particularly in the early to mid-season, while pendimethalin’s efficacy was more variable.

The soybean variety CM60 further modulated outcomes, exhibiting a lower weed biomass for C. rotundus (42.63 g m⁻² at 63 DAA) compared to Morkhor60 (70.79 g m⁻²) in the rainy season (Table 7), likely due to enhanced canopy closure or herbicide retention, as reflected in its higher WCE (68.51% at 63 DAA in the dry season; Table 4 and Figure 2).

3.2. Effects of Pre-Emergence Herbicides on Agronomic Traits

3.2.1. Germination Rate on Soybean

An analysis of the soybean germination rate at 14 DAA revealed significant differences among weed control treatments and varieties, with notable seasonal variation. During the rainy season of 2023, hand weeding resulted in the highest germination rate (93.43%), followed by the untreated (weedy) control (84.41%), pendimethalin (80.00%), and s-metolachlor (73.04%) (

Table 8. Effects of weed control treatments and soybean varieties on germination (%) at 14 DAA.

Variable	Germination Rate (%) at 14 DAA 1/
	Rainy Season	Dry Season
Weed Control (WC)
Hand weeding	93.43 a 2/	95.13
Weedy	84.41 ab	96.25
Pendimethalin	80.00 ab	95.88
S-metolachlor	73.04 b	95.50
Flumioxazin	33.65 c	92.13
Variety (Var)
Morkhor60	63.21 b	92.10 b
CM60	83.00 a	97.85 a
F-test
Weed Control (WC)	** 4/	ns 5/
Variety (Var)	* 3/	**
WC × Var	ns	ns
CV WC	24.17	3.49
CV WC × Var	23.33	3.26

^1/ Days after application. ^2/ The same letters are not significantly different based on LSD at p < 0.05, ^3/ significant at p < 0.05, ^4/ significant at p < 0.01, and ^5/ not significant.

). Flumioxazin, however, caused a marked reduction in the germination rate (33.65%), indicating considerable phytotoxicity to soybean seedlings under these conditions, consistent with the high level of toxicity associated with flumioxazin, as illustrated in Figure 3a. The F-test for weed control treatments was highly significant (p < 0.01), confirming that the choice of pre-emergence herbicide had a strong impact on soybean emergence in the rainy season. Varietal differences were also significant (p < 0.05), with CM60 showing a higher germination rate (83.00%) than Morkhor60 (63.21%), suggesting a greater tolerance of CM60 to herbicide application or environmental stress during early establishment. The interaction between weed control and variety was not significant, indicating that the effect of herbicide treatment on germination was generally consistent across varieties.

During the dry season of 2024/2025, germination rates were uniformly high across all treatments, ranging from 92.13% (flumioxazin) to 96.25% (weedy), and the F-test for weed control was not significant (Table 8). This suggests that under drier conditions, the phytotoxic effects of pre-emergence herbicides on soybean germination are less pronounced or that environmental factors mitigate potential injury. However, varietal differences were highly significant (p < 0.01) during the dry season, with CM60 (97.85%) again outperforming Morkhor60 (92.10%), underscoring the importance of varietal selection for optimal crop establishment.

3.2.2. Phytotoxicity on Soybean

An analysis of phytotoxicity in soybeans following pre-emergence herbicide application revealed substantial differences among treatments and between seasons, with direct implications for crop safety and the practical use of these herbicides in Northeast Thailand. During the rainy season of 2023, all herbicide treatments induced some level of phytotoxicity, with the magnitude and persistence varying significantly by active ingredient (Figure 3).

Among the herbicides, flumioxazin consistently produced the highest phytotoxicity scores, exceeding 65% at all time points from 21 to 63 DAA (Figure 3a). This high and persistent level of crop injury was significantly greater than that observed with pendimethalin and s-metolachlor (p < 0.01), aligning with the marked reduction in soybean germination rate and plant vigour observed during flumioxazin treatment. Pendimethalin and s-metolachlor also caused significant phytotoxicity in the rainy season, but to a lesser extent, with scores ranging from approximately 36% to 42% for pendimethalin and 26% to 29% for s-metolachlor. The statistical analysis confirmed that both weed control treatments and soybean varieties had significant effects on phytotoxicity (p < 0.05), and their interaction was also significant, indicating the importance of considering varietal tolerance to herbicide injury.

Notably, the CM60 variety exhibited lower phytotoxicity scores than Morkhor60 across all treatments and time points in the rainy season, suggesting that CM60 possesses greater tolerance to pre-emergence herbicides (Figure 3c). This varietal difference was statistically significant (p < 0.05) and consistent with the higher germination rates observed for CM60, affecting plant populations and soybean yields (

Table 9. Effect of weed control treatments and soybean varieties on yield and yield components.

Variable	Soybean Yield and Yield Component
	Rainy Season of 2023							Dry Season of 2024/2025
	Pods on Branches	Pods on Main Stem	Seed Number/ Pod	Weight of 100 Seeds (g)	Grain Yield (kg ha−1)	Grain Yield Loss (%)	Plant Population (Plant ha−1)	Pods on Branches	Pods on Main Stem	Seed Number/ Pod	Weight of 100 Seeds (g)	Grain Yield (kg ha−1)	Grain Yield Loss (%)	Plant Population (Plant ha−1)
Weed Control (WC)
Hand weeding	20.21 a	18.31 ab	2.40	17.15 a	1090.50 a	0.00	224,400 a	11.46 a	19.40 a	2.26	17.62 a	1366.50 a	0.00	228,050 ab
Weedy	8.95 d	15.64 b	2.30	16.43 ab	481.30 b	55.86	202,950 ab	2.80 c	8.56 c	2.16	16.94 ab	422.40 d	69.09	230,550 a
Pendimethalin	13.82 c	15.44 b	2.43	15.87 b	365.50 c	66.48	194,850 ab	9.85 b	18.13 a	2.30	16.78 ab	1129.40 bc	17.35	228,350 ab
S-metolachlor	16.50 b 1/	20.90 a	2.48	16.94 a	1036.30 a	4.97	175,950 b	9.84 b	14.89 b	2.28	15.61 c	954.80 c	30.13	227,750 ab
Flumioxazin	17.59 b	20.43 a	2.50	16.93 a	360.20 c	66.97	82,350 c	9.93 b	17.37 a	2.30	16.24 bc	1242.60 ab	9.07	219,850 b
Variety (Var)
Morkhor60	18.68 a	19.10 a	2.48 a	15.48 b	603.05 b	-	152,580 b	11.11 a	14.98 b	2.34 a	15.94 b	1028.7	-	219,980 b
CM60	12.15 b	17.18 b	2.36 b	17.85 a	730.48 a	-	199,620 a	6.44 b	16.36 a	2.17 b	17.35 a	1017.5	-	233,840 a
F-tests
Weed Control (WC)	** 4/	**	ns 2/	*	**	-	**	**	**	ns	**	**	-	ns
Variety (Var)	**	*	* 3/	**	**	-	**	**	*	**	**	ns	-	**
WC × Var	**	**	ns	ns	**	-	ns	**	**	ns	**	ns	-	ns
CV WC	15.53	14.51	6.75	4.21	13.55	-	23.84	13.05	14.39	5.44	4.76	17.70	-	3.48
CV WC × Var	11.75	13.45	5.84	7.57	17.93	-	23.01	13.99	10.96	4.54	4.77	16.5	-	3.35

^1/ Means indicated by the same letters are not significantly different based on LSD at p < 0.05, ^2/ not significant at p < 0.05, ^3/ significant at p < 0.05, and ^4/ significant at p < 0.01.

).

During the dry season of 2024/2025, overall phytotoxicity levels were significantly reduced across all treatments. Flumioxazin, pendimethalin, and s-metolachlor induced only minimal crop injury, with scores below 4% at 21 DAA and declining to less than 1% at later intervals (Figure 3b). No phytotoxicity was observed in hand-weeded or untreated plots. During this season, neither the variety nor the interaction between the variety and weed control had a significant effect on phytotoxicity, as indicated by the non-significant F-tests. This seasonal difference suggests that environmental conditions, such as lower soil moisture and reduced rainfall during the dry season, may mitigate the expression of herbicide-induced crop injury.

3.2.3. Soybean Plant Height

The efficacy of pre-emergence herbicides in soybean cultivation was evaluated using plant height measurements across two growing seasons (rainy season 2023 and dry season 2024/2025) under varying weed control regimes. During the rainy season of 2023, hand weeding consistently produced the tallest soybean plants, which reached 56.15 cm at 63 DAA, significantly outperforming herbicide-treated plots (e.g., pendimethalin: 53.25 cm; flumioxazin: 54.01 cm) (p < 0.05) (Figure 4a). This aligns with its 100% WCE (Table 4 and Figure 2), eliminating weed competition and thereby optimising resource availability for soybean growth (Figure 2a). Conversely, the weedy control group exhibited the shortest plants (53.05 cm at 63 DAA), reflecting severe resource competition from uncontrolled weeds.

Herbicide-induced phytotoxicity (Figure 3a) further modulated plant growth. For instance, flumioxazin, which caused the highest phytotoxicity during the rainy season (67.35% at 21 DAA), resulted in reduced plant height (14.68 cm at 21 DAA) compared to hand weeding (17.16 cm; Figure 4a). However, by 63 DAA, flumioxazin-treated plants recovered to 54.01 cm, suggesting transient phytotoxicity effects. S-metolachlor, with moderate phytotoxicity (28.85–25.90%), maintained competitive plant heights (54.33 cm at 63 DAA; Figure 4a), likely due to its balanced WCE (48.30–61.54%; Table 4 and Figure 2a), which suppressed weeds without severely impairing soybean growth. Varietal differences were pronounced, with Morkhor60 exhibiting taller plants than CM60 across both seasons (e.g., 57.20 cm versus 51.12 cm at 63 DAA in the rainy season; p < 0.05) (Figure 4c). This superiority persisted, even under herbicide stress, highlighting Morkhor60’s resilience, possibly linked to its higher germination rate (92.10% in the dry season versus CM60’s 97.85%) (Table 8), which may enhance early establishment and resource capture.

The decline in WCE for herbicides over time (e.g., pendimethalin’s WCE dropped from 54.54% to 10.03% by 63 DAA in the rainy season; Table 4 and Figure 2a) was correlated with gradual weed resurgence, which likely contributed to reduced plant height gains in the later growth stages. However, during the dry season of 2024/2025, higher WCE values for herbicides (e.g., flumioxazin: 78.50% at 49 DAA) (Table 4 and Figure 2b) corresponded with improved plant heights (36.50 cm at 63 DAA) (Figure 4b), underscoring seasonal variations in herbicide performance and weed pressure.

3.2.4. Number of Nodes

The number of nodes in soybean plants, a critical indicator of vegetative development and yield potential, exhibited significant variations under different pre-emergence herbicide treatments and soybean varieties during the rainy season of 2023 and dry seasons of 2024/2025 (Figure 5). During the rainy season of 2023, weed control (WC) methods significantly influenced the number of nodes at 21 DAA (p < 0.05), with flumioxazin yielding the highest early-stage nodes (4.00 nodes/plant), followed by pendimethalin (3.85 nodes/plant) (Figure 5a). Conversely, s-metolachlor produced the lowest node count (3.28 nodes/plant), likely due to its moderate phytotoxicity, which may have temporarily suppressed meristematic activity. Hand weeding, a non-chemical control, showed intermediate node numbers (3.68 nodes/plant), reflecting the absence of herbicide stress but potential early weed competition. By 63 DAA, s-metolachlor surpassed other treatments (10.25 nodes/plant), suggesting recovery from initial phytotoxic effects, while flumioxazin exhibited a decline (9.9 nodes/plant), aligning with its persistent phytotoxicity.

Varietal differences were pronounced, with Morkhor60 consistently outperforming CM60 in node counts (p < 0.01 at 21 DAA in 2023). For instance, Morkhor60 recorded 3.89 nodes/plant at 21 DAA compared to CM60’s 3.44 nodes/plant (Figure 5a), highlighting the genetic disparities in stress tolerance. This aligns with the germination data (Table 8), where CM60 exhibited higher germination rates (83.00% in the rainy season) but lower subsequent node counts, implying that early vigour may not compensate for herbicide-mediated growth suppression. Morkhor60’s resilience was further evident during the dry season, where it maintained superior node numbers (9.51 nodes/plant at 63 DAA versus 7.98 for CM60; Figure 5d), correlating with its lower phytotoxicity susceptibility (1.58% versus 1.26% for CM60 at 21 DAA in 2024/2025; Figure 3d).

During the dry season of 2024/2025, WC treatments significantly affected node counts at the later growth stages (49 and 63 DAA, p < 0.05) (Figure 5b). Hand weeding and pendimethalin demonstrated the highest node numbers (8.96–9.93 nodes/plant at 63 DAA), attributable to their balanced WCE (77.41–85.32%; Figure 2b) and minimal phytotoxicity (≤3.83% in 2024/2025; Figure 3b). Conversely, weedy plots exhibited the lowest node counts (6.70–7.05 nodes/plant), underscoring the detrimental impact of uncontrolled weed competition. Flumioxazin, despite its high WCE (78.50% at 49 DAA in the dry season; Figure 2b), showed reduced node counts (8.83 nodes/plant at 63 DAA), likely due to cumulative phytotoxic stress (0.85% at 63 DAA; Figure 3b) and residual effects on meristem development.

3.3. Effects of Pre-Emergence Herbicides on Yield and Yield Components

The evaluation of soybean yield and its components under various pre-emergence herbicide treatments in Northeastern Thailand revealed that across both the rainy and dry seasons, hand weeding consistently resulted in the highest seed yields, with values reaching 1090.5 kg ha⁻¹ during the rainy season and 1366.5 kg ha⁻¹ during the dry season (Table 9). This treatment also supported the greatest number of pods on both branches and the main stem, as well as a superior 100-seed weight, reflecting optimal crop establishment and minimal interference from weeds. These outcomes were closely linked to the observance of the highest plant population densities, which were maintainable due to the negligible phytotoxicity effects and effective weed suppression.

Among the herbicide treatments, s-metolachlor demonstrated strong performance, producing yields comparable to those of hand weeding, particularly during the rainy season, when seed yield reached 1036.30 kg ha⁻¹ with only 4.97% yield loss. This treatment also maintained relatively high pod numbers and seed counts, indicating its capacity to balance effective weed control with acceptable crop safety. Flumioxazin, despite inducing the highest phytotoxicity during the rainy season (reaching up to 67.35% at 21 DAA), still supported moderate pod formation and seed development. However, its overall yield was substantially lower (360.2 kg ha⁻¹ in the rainy season), largely due to a significantly reduced plant population caused by early-season crop injury and lower germination rates. This pattern underscores the trade-off between herbicide phytotoxicity and weed control efficacy, where severe early injury can limit stand establishment and consequently reduce yield potential despite effective weed suppression. Pendimethalin exhibited moderate phytotoxicity and declining weed control efficiency over time, as reflected in its intermediate yield levels (365.5 kg ha⁻¹ in the rainy season and 1129.4 kg ha⁻¹ in the dry season). The reduced weed suppression capacity allowed for increased weed competition, which negatively impacted pod formation and seed filling. The weedy control plots consistently produced the lowest yields and yield components across seasons, highlighting the detrimental effect of unchecked weed pressure on soybean reproductive development. Compared to hand weeding, the absence of weed management resulted in a yield loss of 55.86% during the rainy season and 69.09% during the dry season, underscoring the critical importance of effective weed control for maximising soybean productivity.

The relationships between yield components and other agronomic parameters further elucidate the mechanisms underlying these results. Treatments supporting higher germination rates and plant populations (such as hand weeding and s-metolachlor) also promoted greater plant height and node number, both of which are critical for maximising photosynthetic capacity and reproductive sites. Conversely, high phytotoxicity during the rainy season of 2023 in flumioxazin-treated plants suppressed germination and early growth, limiting node development and ultimately reducing yield, despite relatively good pod and seed sets on surviving plants.

Seasonal variations also influenced herbicide performance and crop response. The dry season generally exhibited higher germination rates and yields across treatments, possibly due to more favourable environmental conditions that mitigated herbicide injury and enhanced crop recovery. In contrast, the rainy season’s higher moisture levels may have exacerbated herbicide phytotoxicity, particularly for flumioxazin, thereby reducing stand establishment and yield potential. Varietal differences were apparent, with CM60 achieving higher seed yields and plant populations, while Morkhor60 tended to produce more pods and seeds per pod. These findings suggest that varietal selection should be integrated with herbicide choice to optimise soybean productivity under local conditions.

4. Discussion

This study highlights the importance of effective weed management in optimising soybean yield and its components within the rainfed agroecosystems of Northeast Thailand. The weed flora predominantly consisted of broad-leaved species (62.5%), followed by grasses (20.8%) and sedges (16.7%). Notably, T. portulacastrum, O. corymbosa, D. ciliaris, D. aegyptium, and C. rotundus constituted the majority of weed biomass and density. This observed weed spectrum aligns with previous findings from upland fields in the region [30], reinforcing the necessity for comprehensive weed management strategies.

Hand weeding consistently yielded high WCE, completely suppressing weed growth throughout the rainy and dry seasons (Figure 2). This treatment produced the highest grain yields, reaching 1090.5 kg ha⁻¹ during the rainy season and 1366.5 kg ha⁻¹ during the dry season, as well as the most significant number of pods per branch and main stem, 100-seed weight, and optimal plant population densities (Table 9). These results align with previous studies indicating that manual weed removal remains the benchmark for maximising crop productivity in smallholder systems [36,37]. The superior performance exhibited by hand weeding reflects the effective suppression of weed competition, and the use of this conventional method avoids damage to plants caused by herbicides.

In contrast, the absence of weed control (weedy plots) dramatically reduced yield and yield components. Grain yield declined to 481.3 kg ha⁻¹ during the rainy season and 422.4 kg ha⁻¹ during the dry season, corresponding to a yield loss of 55.86% and 69.09%, respectively, relative to hand weeding. This substantial yield penalty underscores the severity of weed interference, which is consistent with the findings of Jha et al. [38] and Chauhan and Johnson [39], who reported that uncontrolled weed growth can reduce soybean yield by more than 50% in tropical environments. The reduced yield in weedy plots was accompanied by lower pod numbers, seed weight, and plant populations, indicating that weed competition limits resource availability and impairs crop establishment and reproductive development.

Among the herbicides tested, s-metolachlor 96% EC at 900 g a.i. ha⁻¹ provided the most consistent and sustainable weed control, with WCE values of 87.2% at 21 DAA and 61.5% at 63 DAA during the rainy season and similarly high WCE throughout the dry season (Figure 2). This translates to grain yields (1036.3 kg ha⁻¹ during the rainy season; 954.8 kg ha⁻¹ during the dry season) comparable to hand weeding, indicating that s-metolachlor effectively balances weed suppression and crop safety. The correlation between increased grain yield and higher WCE (Figure S1) highlights that early weed suppression was pivotal to achieving yields comparable to hand weeding. This aligns with global studies in which s-metolachlor-based treatments boosted yields by 80–97% [35,40,41]. The superior performance of s-metolachlor is likely due to its longer soil residual activity (15–50 days; [42]) and broad-spectrum efficacy, as reflected in the low weed density and biomass observed in treated plots (Table 6 and Table 7). These findings are consistent with those of Meseldžija et al. [35] and Qadeer et al. [43], who highlighted the yield benefits of s-metolachlor in various cropping systems.

Pendimethalin and flumioxazin exhibited moderate to high initial WCE. However, their effectiveness declined over time, especially against sedge species such as C. rotundus, which demonstrated biological resilience due to its deep tuber system and staggered emergence [44]. This observation is evident in the increased weed density and biomass during the later growth stages, particularly during the rainy season. The limited residual activity of these herbicides, along with their lipophilic nature and heightened leaching during periods of heavy rainfall [45], resulted in diminished weed suppression and moderate yield outcomes (pendimethalin: 365.5–1129.4 kg ha⁻¹; flumioxazin: 360.2–1242.6 kg ha⁻¹). Notably, flumioxazin induced severe phytotoxicity during the rainy season (up to 67.35% at 21 DAA), resulting in poor germination (33.65%) and drastically reduced plant populations (82,350 plants ha⁻¹), which ultimately limited yield despite moderate weed control (Table 8 and Figure 3). Consistent with previous findings, Taylor-Lovell et al. [20] observed that elevated soil moisture levels intensify herbicide-induced damage in soybean plants.

Soil moisture is essential for activating pre-emergence herbicides and ensuring their bioavailability for weed suppression, particularly during the rainy season, when prolonged exposure to cool, wet conditions during crop emergence can hinder soybeans’ capacity to detoxify these herbicides, resulting in heightened phytotoxicity [46,47,48,49]. In addition, during the “soil cracking” stage of emergence, precipitation can cause the splashing of higher concentrations of PPO-inhibitor herbicides onto sensitive parts of the soybean plant, such as the hypocotyl and cotyledons. This can lead to tissue necrosis, further increasing the risk of injury when soil moisture levels are not optimal [46,50].

Weed management practices also influenced soybean growth parameters. Treatments with higher WCE, such as hand weeding and s-metolachlor, supported greater plant height and node number at maturity (Figure 4 and Figure 5). These traits have been associated with yield components, as increased node number provides more sites for pod development [51]. Seasonal differences were evident, with higher plant height and node numbers generally observed during the rainy season, likely due to higher temperatures (28.6 °C versus 26.4 °C) and greater soil moisture, which can enhance vegetative growth [52]. Research by Madhu and Hatfield [53] supports this observation, demonstrating that adequate soil moisture levels substantially improve crops’ photosynthetic rates and total dry matter production. The growth response to seasonal conditions appears complex, as elevated CO₂ levels associated with warmer temperatures may initially slow early vegetative growth before promoting increased leaf number, leaf area expansion, and dry matter accumulation [54]. However, as demonstrated in plots treated with flumioxazin, severe phytotoxicity can counteract these beneficial environmental effects, significantly suppressing early growth and node formation. This suppression ultimately leads to reduced yield potential despite otherwise favourable seasonal growing conditions.

Variations among soybean cultivars significantly influenced yield responses to weed management. The CM60 variety demonstrated higher herbicide tolerance, resulting in greater germination rates and plant populations than Morkhor60, ultimately achieving superior yields (730.5 kg ha⁻¹ versus 603.1 kg ha⁻¹ during the rainy season). The enhanced performance of CM60 was linked to a more effective canopy closure and lower weed biomass in C. rotundus species (42.6 g m⁻² compared to 70.8 g m⁻²), reinforcing the idea that cultivar selection can improve soybeans’ competitive ability against weeds [42,45]. Reinforced by a statement from Bianchi et al. [55] and Bastiani et al. [56], more stem dry matter, greater shoot dry matter, and increased soil coverage by the crop canopy were associated with superior competitiveness. Plant height also played a role in competition, with barnyardgrass showing greater competitiveness compared to short-height soybean cultivars. Conversely, although Morkhor60 exhibited greater height and a higher number of nodes, it experienced more significant yield reductions due to its increased vulnerability to herbicide phytotoxicity.

These findings highlight the critical importance of integrated weed management strategies that combine effective chemical control with the selection of tolerant cultivars to maximise soybean productivity in tropical rainfed systems. The dramatic yield losses observed in the absence of weed control emphasise that timely and effective weed management is indispensable for sustainable soybean production in Northeast Thailand.

5. Conclusions

This study demonstrates that the efficacy of pre-emergence herbicides for weed management in soybean cultivation in Northeast Thailand is season-dependent. S-metolachlor (96% EC, 900 g a.i. ha⁻¹) is optimal for the rainy season, offering weed control efficiency, low phytotoxicity, and high yields. For the dry season, flumioxazin (50% WP, 125 g a.i. ha⁻¹) is recommended, with significantly reduced phytotoxicity compared to rainy season applications and high yields. Both herbicides effectively controlled the most dominant weed species, although Cyperus rotundus L. was less affected. The CM60 variety consistently showed greater herbicide tolerance and yield stability, especially under rainy season conditions. However, these findings are limited to two growing seasons, and multi-year trials are recommended to validate the consistency of herbicide performance and varietal responses. These insights underscore the importance of integrating season-specific herbicide programs with appropriate variety selection to optimize sustainable soybean production in Northeast Thailand.