Efficacy, phytotoxicity and economics of different herbicides in aerobic rice

Abstract

Aerobic rice is a potential water-wise rice production system, but high weed infestation has threatened its sustainability, which demands an efficient and cost-effective weed management technique. Eight commercial herbicide products were applied singly or as tank-mix or in sequence to evaluate their efficacy, rice selectivity and cost-effectiveness in aerobic rice. The study was conducted under field conditions in Malaysia during 2010/2011 following a randomized complete block design. Most of the herbicide treatments provided excellent weed control, and produced much higher net benefit than weedy or weed-free check. None of the herbicides caused significant phytotoxicity to rice plants. Among the herbicide treatments, sequential application of Cyhalofop-butyl+Bensulfuron at early growth stage followed by Bentazon/MCPA at mid growth stage provided the highest weed control efficiency, productivity and net benefit. Application of Bispyribac-sodium at early growth stage followed by Bentazon/MCPA at mid growth stage performed very close to the above-mentioned treatments. Sequential application of Pretilachlor/safener just after seeding followed by Propanil/Thiobencarb at early growth stage also provided satisfactory results in terms of efficacy and economic return. Since manual weeding was not economic, herbicide rotation using the above chemicals may be recommended for effective weed management in aerobic rice.

Keywords:

• Aerobic soil
• dry seeded rice
• herbicide resistance
• net benefit
• relative yield loss
• rice selectivity
• weed control rating

Introduction

Currently, sustainability of water resources is of major concern (Juraimi et al. 2010), and declining water availability threatens the sustainability of traditional flood-irrigated rice ecosystems (Anwar et al. 2010). In Asia, it is predicted that 17 million ha of irrigated rice areas may have “physical water scarcity” and 22 million ha areas may be subject to “economic water scarcity” by 2025 (Bouman and Tuong 2001). It is, therefore, no longer feasible to flood rice fields for better crop establishment and weed control (Johnson and Mortimer 2005). Among different water-saving approaches, aerobic rice cultivation has come up with a huge successes in different parts of the world. Aerobic rice, growing rice in non-saturated and non-puddled aerobic soil, is a promising water-wise technique of rice cultivation under the context of ever-mounting water scarcity (Anwar et al. 2011). But weed menace continues to be a severe problem in aerobic rice systems resulting in up to 90% reduction in grain yield (Gowda et al. 2009a). When direct seeded, rice seeds germinate simultaneously with weed seeds without any “head start” over germinating weed seeds (Zhao et al. 2006), and the initial flush of weeds is not suppressed by flooding (Olofsdotter et al. 2000). Besides, dry-tillage practices and aerobic soil conditions are highly conducive for germination and growth of weeds (Balasubramanian and Hill 2002). Therefore, effective weed management in aerobic rice has become a serious challenge for researchers and farmers (Rao et al. 2007).

Despite some undesirable side-effects such as development of resistant weed biotypes (Rahman et al. 2010) no viable alternative is presently available to shift the chemical dependence for weed management in rice. Many researchers working on weed management in direct-seeded rice opined that herbicide may be considered to be a viable alternative/supplement to hand weeding (Ashraf et al. 2006, Mahajan et al. 2009, Pacanoski and Glatkova 2009, Chauhan and Johnson 2011). Among the herbicides ethoxysulfuron, cyhalofop-butyl, chlorimuron, metsulfuron, bispyribac sodium, penoxsulam effectively controlled weeds in aerobic rice (Mann et al. 2007, Mahajan et al. 2009, Juraimi et al. 2010). Efficacy of a pesticide depends on its ability to produce a desired effect on the target pest. But, efficacy alone is not enough to determine suitability of an herbicide; cost-effectiveness also has to be taken into account before arriving at a decision (Wibawa et al. 2010). Moreover, application time of herbicide is also very important with respect to efficacy, phytotoxicity and ease of application. However, the safest time for applying herbicide may not coincide with the optimum time in terms of maximum efficacy. Before recommending any herbicide, information regarding cost-effectiveness and phytotoxicity need to be considered in conjunction with weed control efficacy for sustainable weed management.

Since aerobic rice is a new concept, information regarding weed composition and management under this system is scarce. Keeping in mind the inevitability of chemical weed control, the study aimed at controlling weeds efficiently in aerobic rice by using herbicides. It involved testing a diverse range of herbicides with a view to identifying potential herbicides with alternative modes of action to avoid risk of developing herbicide-resistant weed biotypes. Other major goals were to evaluate the cost effectiveness and phytotoxicity of different herbicides to rice plants under aerobic soil conditions.

Materials and methods

Site description

The trial was carried out during off season 2010 (May–July) and main season 2010/2011 (November–January) under field conditions at the experimental farm (Field 2), Universiti Putra Malaysia, Malaysia (3^o 00′ 21.34″ N, 101^o 42′ 15.06″ E, 37 m elevation). The soil (0–15 cm) belongs to Serdang series, was sandy clay loam in texture (57.07% sand, 22.21% silt and 20.60% clay) and acidic in reaction (pH 5.8) with 1.43 g cc⁻¹ bulk density, 1.85% organic carbon and 17.36 Milli equivalent (me) 100 g⁻¹ soil Cation exchange capacity (CEC). Soil contained 0.33% total N, 21.2 ppm available P, 143 ppm available K, 794 ppm Ca and 163 ppm Mg. At field capacity, soil water retention was 22.64% (wet basis) and 29.27% (dry basis). The local climate was hot humid tropic with high humidity and abundant rainfall. During the off season, monthly average maximum temperature, minimum temperature and relative humidity were 33.5–35.0 °C, 23.6–24.6 °C and 93.3–93.9%, respectively, while rainfall, evaporation and sunshine hours were 4.2–10.8 mm day⁻¹, 3.65–4.90 mm day⁻¹ and 5.90–7.37 hrs day⁻¹, respectively. During the main season, monthly average maximum and minimum temperature and relative humidity were 31.7–33.3 °C, 22.9–23.8 °C and 94.1–94.6%, respectively, while rainfall, evaporation and sunshine hours were 6.1–9.9 mm day⁻¹, 2.64–4.66 mm day⁻¹ and 3.95–6.34 hrs day⁻¹, respectively.

Plant material

Aerobic rice line AERON 1, sourced from International Rice Research Institute (IRRI), was used as the plant material. The rice line AERON 1 was selected based on its performance in terms of productivity and weed competitiveness under aerobic soil conditions in previous study (Anwar et al. 2010).

Experimental design and treatments

The experiment was conducted in a randomized complete block design with three replicates. Fifteen treatments comprising 13 different combinations of herbicides, season-long weed-free check and season-long weedy check were included in the trial (Table I). Herbicides included one pre-emergence (Pretilachlor), seven early-post emergence (Cyhalofop-butyl, Bensulfuron, Bispyribac-sodium, Quinclorac, Fenoxaprop-p-ethyl, Propanil and Thiobencarb) and two post emergence (Bentazon and MCPA), which were available as eight commercial products. Locally available and widely used herbicides were selected for the trial. The application rates of different herbicides were followed as per manufacturers’ recommendations. All herbicides were applied using 300 L of water per hectare with a 16 L knapsack sprayer. Season-long weed-free plots were maintained through manual weeding. In weedy checks, no weeding operations were done.

Table I. List of herbicide treatments used in the experiments in off season 2010 and main season 2010/11.

		Application
Label	Treatment	Rate	Time (days after seeding)
T1	Bispyribac-sodium	0.03 kg a.i. ha^−1	10
T2	Bispyribac-sodium fb Bentazon/MCPA	0.03 kg a.i. ha^−1 fb 0.6/0.1 kg a.i. ha^−1	10 fb 40
T3	Cyhalofop-butyl+Bensulfuron	0.1 kg a.i. ha^−1+ 0.06 kg a.i. ha^−1	10
T4	Cyhalofop-butyl+Bensulfuron fb Bentazon/MCPA	0.1 kg a.i. ha^−1+ 0.06 kg a.i. ha^−1fb. 0.6/0.1 kg a.i. ha^−1	10 fb 40
T5	Fenoxaprop-p-ethyl/safener	0.06 kg a.i. ha^−1	10
T6	Fenoxaprop-p-ethyl/safener fb Bentazon/MCPA	0.06 kg a.i. ha^−1 fb 0.6/0.1 kg a.i. ha^−1	10 fb 40
T7	Pretilachlor /safener	0.5 kg a.i. ha^−1	1
T8	Pretilachlor /safener fb Bentazon/MCPA	0.5 kg a.i. ha^−1 fb 0.6/0.1 kg a.i. ha^−1	1 fb. 40
T9	Pretilachlor fb Propanil/Thiobencarb	0.5 kg a.i. ha^−1 fb 1.2/2.4 kg a.i. ha^−1	1 fb 10
T10	Propanil/Thiobencarb	1.2/2.4 kg a.i. ha^−1	10
T11	Propanil/Thiobencarb fb Bentazon/MCPA	1.2/2.4 kg a.i. ha^−1 fb 0.6/0.1 kg a.i. ha^−1	10 fb 40
T12	Quinclorac	0.25 kg a.i. ha^−1	10
T13	Quinclorac fb Bentazon/MCPA	0.25 kg a.i. ha^−1 fb 0.6/0.1 kg a.i. ha^−1	10 fb 40
T14	Season long weed-free	–	–
T15	Season long weedy	–	–
/ means herbicides were formulated as a proprietary mixture,+means that the herbicides were tank-mixed and applied at the same time. All herbicides were applied as per manufacturers’ recommended rates in 300 L of water ha^−1 by knapsack sprayer. fb=followed by.

Crop husbandry

Rice seeds were collected from the Malaysian Agriculture Research and Development Institute. The aerobic field was dry-ploughed and harrowed but not puddled during land preparation. Each plot was 5 m long and 3 m wide, and consisted of 12 rows with 25 cm inter-row and 15 cm intra-row spacing. Rice seeds were directly dry-seeded at the rate of six seeds per hill to fulfill the recommended seed rate for aerobic rice of 40–60 kg ha⁻¹ (Singh and Chinnusamy 2006). Each plot was fertilized with triple super phosphate (TSP) and muriate of potash (MP) at 100 kg P ha⁻¹ and 100 kg K ha⁻¹, respectively as basal application; urea was top dressed thrice each at 50 kg N ha⁻¹ at 2, 4 and 6 weeks after seeding. Field was maintained under non-saturated aerobic conditions (−33 kPa or −1/3 bar of hydraulic head or suction pressure) throughout the growing season. In both the seasons, the trial was primarily rain-fed, but supplemental sprinkler irrigation was applied once or twice a week, as necessary. Soil moisture status was monitored daily by using a tensiometer. Overflow canals were maintained to facilitate drainage whenever heavy rainfall resulted in ponding. Plant protection measures were taken, as needed, to minimize confounding effect of competition with insect and/or disease injury.

Data collection

A 25 cm×25 cm quadrate was randomly placed at two spots in each plot for recording weed data at 10, 40 and 75 DAS. Weeds were clipped to ground level, identified, counted by species, and oven dried at 70 °C for 72 hours. Weed density (WD) and weed dry weight (WDW) were expressed as no. m⁻² and g m⁻², respectively. Dominant weed species were identified using the summed dominance ratio (SDR) computed as follows (Janiya and Moody 1989):

Where,

Relative contribution of different weed groups (broad-leaved, grasses and sedges) to the weed community in terms of RD and RDW were also calculated. To evaluate the weed homogeneity, i.e. similarity in occurrence of weed species between treatments, Sorenson's index of similarity (S) was determined using the following formula:

Where,

S=Index of association between treatment A and B,

J=No. of weed species common in both treatment A and B,

A=No. of weed species present in treatment A,

B=No. of weed species present in treatment B

Weed control efficiency (WCE) of different treatments was calculated as follows (Hasanuzzaman et al. 2008):

Where,

DWC=dry weight of weeds in weedy check plots

DWT=dry weight of weeds in treated plots

Crop phytotoxicity rating and weed control rating were assessed visually at 7, 14 and 21 d (Begum et al. 2008) after each herbicide application using a scale of 1 to 5 (Okafor 1986). Central 3 m² area of each plot was harvested at maturity (when 90% grain became golden yellow in color) for recording grain yield at 14% moisture basis (wet weight basis). Percent relative yield loss (RYL) due to weeds and percent yield increase over control (YOC) was calculated as follows:

Economic analysis was performed to determine the cost effectiveness of different treatments following the procedure by Hussain et al. (2008). Two manual weeding sessions were considered sufficient to keep the plots weed-free throughout the growing season. Laborers required for one manual weeding and one round herbicide spraying per hectare were 50 and 2, respectively. The cost for laborers was Ringgit Malaysia (RM) 25 laborer⁻¹ day⁻¹. The amount of commercial product of herbicide required per hectare was calculated and the cost of each herbicide was estimated based on their local market price. Price of paddy was collected from different rice-growing areas and was considered as RM 1000 t⁻¹ for calculating the gross return. The net benefit per hectare for each treatment was calculated by deducting the weed management cost from the gross return.

Statistical analysis

All data were subjected to ANOVA by using SAS statistical software package version 9.1 (SAS 2003). Treatment by season interaction was not significant; therefore, data were averaged across the seasons and used in subsequent analysis. Significant differences among means were calculated using Fisher's protected least significant difference (LSD) test at p ≤ 0.05. Simple regression analysis was conducted using weed control efficiency to predict yield of rice.

Results

Floristic composition of weeds

The experimental field was infested with broadleaf weeds, sedges and grasses. The weed community, mostly dominated by broadleaf weeds, comprised 20 species from 12 different families (Table II). Based on the summed dominance ratio (SDR) values, broadleaf weed species Physalis heterophylla Nees was the predominant species in both main (SDR 19.26) and off (SDR 16.85) seasons. Scoparia dulcis L. emerged as second dominant weed species in the main season (SDR of 13.41) and fourth dominant species in the off season (SDR 10.50). Another broadleaf weed species Cleome rutidosperma ranked third in both main (SDR 10.04) and off (SDR 11.43) seasons. Among the sedges, Cyperus rotundus appeared as the fourth dominant weed in the main season (SDR 9.74) but second dominant weed in the off season (SDR11.62). Grassy weed species Eleusine indica occupied the fifth position in the main season. Leptochloa chinensis was not amongst the top-most dominant weed species in the main season, but appeared as the fifth dominant species in the off season (SDR 8.42). Further analysis showed that the relative composition of the broadleaf, sedges and grasses across the seasons were about 60%, 20% and 20%, respectively. Thus, it is apparent that the present aerobic rice field was dominated by broadleaf weeds and the profound extensive species were P. heterophylla Nees, S. dulcis, C. rutidosperma and C. rotundus. Sorenson's index of similarity among different herbicide treatments (data not shown) ranged between 77.24 and 94.66%, indicating good homogeneity of weed composition.

Table II. Weed composition with summed dominance ratio followed by standard error in off season 2010 and main season 2010/2011 as observed in season-long weedy check.

		Summed dominance ratio
Scientific name	Family name	Off season 2010	Main season 2010/2011
Broadleaf
Alternanthera sessilis (L.) R. Br. Ex DC.	Amaranthaceae	2.33±0.69	1.26±0.26
Cleome rutidosperma DC	Capparidaceae	11.43±3.64	10.04±4.1
Emilia sonchifolia (L.) DC. Ex Wight	Asteraceae	2.43±0.88	0.84±0.19
Euphorbia hirta L.	Euphorbiaceae	1.60±0.58	1.05±0.33
Hedoytis corymbosa (L.)Lam.	Rubiaceae	2.07±0.72	2.08±0.52
Hyptis brevipes Poit	Lamiaceae	0.92±0.25	1.39±0.71
Jussiaea linifolia Vahl	Onagraceae	3.82±1.97	4.73±1.3
Phyllanthus niruri L.	Euphorbiaceae	2.39±0.54	4.25±0.9
Physalis heterophylla Nees	Solanaceae	16.85±4.73	19.26±6.2
Mimosa pudica L.	Fabaceae	3.41±1.10	2.22±0.64
Scoparia dulcis L.	Scrophulariaceae	10.50±4.01	13.41±3.7
Sedges
Cyperus aromaticus	Cyperaceae	1.06±0.21	1.47±0.74
Cyperus rotundus L.	Cyperaceae	11.62±3.27	9.74±2.56
Cyperus sphacelatus Rottb.	Cyperaceae	4.47±1.05	2.25±0.61
Fimbristylis miliacea (L.) Vahl	Cyperaceae	4.05±1.36	5.88±1.95
Grasses
Axonopus compressus (Sw.)Beauv	Poaceae	0.53±0.12	0.32±0.11
Digitaria ciliaris (Retz.) Koel	Poaceae	0.76±0.24	2.67±0.67
Echinochloa colona (L.) Link	Poaceae	4.75±1.25	3.44±1.03
Eleusine indica (L.) Gaertn.	Poaceae	6.59±2.63	7.51±3.25
Leptochloa chinensis (L.) Nees	Poaceae	8.42±3.01	6.19±1.86

Weed control and crop toxicity ratings

Weed control rating was done based on visual observation at 7, 14 and 21 d after application (DAA) (Table III). Visual weed control rating revealed that Pretilachlor/safener provided excellent/satisfactory control of weeds at 7 DAA, although moderate and poor weed controls were rated with this herbicide at 14 and 21 DAA, respectively. On the other hand, application of Cyhalofop butyl+Bensulfuron, Bispyribac-sodium and Bentazon/MCPA resulted in satisfactory weed control at 21 DAA although their performances were poor at 7 DAA and moderate at 14 DAA. Propanil/Thiobencarb application offered moderate weed control at 7 DAA but showed good control both at 14 and 21 DAA. Quinclorac and Fenoxaprop-p-ethyl exhibited very poor control of weeds at 7 DAA, poor control at 14 DAA but moderate control at 21 DAA. Thus, weed control rating varied with time depending on the type of herbicide used.

Table III. Weed control and phytotoxicity rating of different herbicides using 1 to 5 scales (Okafor 1986).

	Weed control rating			Phytotoxicity rating
	Days after application
Herbicide	7	14	21	7	14	21
Bentazon/MCPA	4	3	1	1	2	1
Bispyribac-sodium	3	2	1	1	2	1
Cyhalofop-butyl+Bensulfuron	3	2	1	1	1	1
Fenoxaprop-p-ethyl/safener	5	4	3	1	1	1
Pretilachlor /safener	1	2	3	2	1	1
Propanil/Thiobencarb	3	2	2	1	1	1
Quinclorac	4	4	3	1	1	1
Weed control rating: 1=excellent/ satisfactory, 2=good, 3 =fair, 4=poor, 5=no/very poor control.
Phytotoxicity rating: 1=very slight injury, 2=slight injury, 3=phytotoxic, 4=severely phytotoxic, 5=100% killed.

The phytotoxic effects of herbicides on rice were evaluated based on visual observation at 7, 14 and 21 DAA (Table III). Among the herbicides tested, only Pretilachlor/safener, Bispyribac-sodium and Bentazon/MCPA had a slight phytotoxic effect on rice. Phytotoxicity of Pretilachlor/safener was characterized by slight reduction in plant height and leaf chlorosis found at 7 DAA, which disappeared at 14 DAA. In the cases of Bispyribac-sodium and Bentazon/MCPA, plant growth was stunted to some extent, and leaves failed to expand fully and became yellowish as observed at 14 DAA. However, those symptoms disappeared, and the rice plants recovered within a week. Plants treated with all other herbicides either recovered or showed very slight injuries at any of the observation dates. The crop was constantly being monitored and it was found that phytotoxicity did not persist up to harvest.

Weed control efficiency (WCE)

Herbicide treatments exhibited significant effects on dry weight and weed density at different sampling dates (Figures 1 and 2). At 10 DAS (just before application of early-post emergence herbicides), only the Pretilachlor/safener treated plots produced lower weed dry matter and density compared with those of weedy control plots. On average, weed dry matter and density in the Pretilachlor/safener treated plots were 3.3 g m⁻² and 23 weeds m⁻², respectively; while the same were recorded as 15.6 g m⁻² and 223 weeds m⁻², respectively in weedy plots. Thus, pre-emergence application of Pretilachlor/safener decreased weed biomass by 80% and weed density by 90% compared with untreated weedy plots. Reasonably, weed dry matter and density at 10 DAS recorded with other herbicide treatments which include only early-post with/without post emergence herbicides were similar to those of weedy treatments.

Figure 1.  Weed dry weight at different growth stages of rice variety AERON 1 as influenced by weed control treatments (averaged over seasons). T1=Bispyribac-sodium; T2=Bispyribac-sodium fb Bentazon/MCPA; T3=Cyhalofop-butyl+Bensulfuron; T4=Cyhalofop-butyl+Bensulfuron followed by (fb) Bentazon/MCPA; T5=Fenoxaprop-p-ethyl/safener; T6=Fenoxaprop-p-ethyl/safener fb Bentazon/MCPA; T7=Pretilachlor/safener; T8=Pretilachlor/safener fb Bentazon/MCPA; T9=Pretilachlor/safener fb Propanil/Thiobencarb; T10=Propanil/Thiobencarb; T11=Propanil/Thiobencarb fb Bentazon/MCPA; T12=Quinclorac; T13=Quinclorac fb Bentazon/MCPA; T14=Season long weed-free; T15=Season long weedy.

At 40 DAS (prior to application of post emergence herbicide Bentazone/MCPA), the effects of different early-post emergence herbicides on weed dry matter and density were evident (Figure 1). Weed dry matter reduction due to application of different herbicides ranged from 17–90%; weed density, on the other hand, was reduced within the range of 70–93%. Pretilachlor/safener fb. Propanil/Thiobencarb was the most effective herbicide combination reducing weed dry matter by almost 90% compared with weedy treatment, closely followed by Cyhalofop-butyl+Bensulfuron with/without Bentazon/MCPA (80%), Bispyribac-sodium with/without Bentazon/MCPA (75%) and Propanil/Thiobencarb with/without Bentazon/MCPA (74%). Pretilachlor/safener with/without Bentazon/MCPA also provided good weed control by reducing weed dry matter by 50%. But the performances of Fenoxaprop-p-ethyl/safener with/without Bentazon/MCPA and Quinclorac with/without Bentazon/MCPA were very poor (<20% and 36%, respectively). Response of weed density to different herbicide treatments also followed the same trend (Figure 2). Since Bentazon/MCPA was applied at 40 DAS, its effect on weed biomass/density was not found.

Figure 2.  Weed density at different growth stages of rice variety AERON 1 as influenced by weed control treatments (averaged over seasons). T1=Bispyribac-sodium; T2=Bispyribac-sodium fb Bentazon/MCPA; T3=Cyhalofop-butyl+Bensulfuron; T4=Cyhalofop-butyl+Bensulfuron followed by (fb) Bentazon/MCPA; T5=Fenoxaprop-p-ethyl/safener; T6=Fenoxaprop-p-ethyl/safener fb Bentazon/MCPA; T7=Pretilachlor/safener; T8=Pretilachlor/safener fb Bentazon/MCPA; T9=Pretilachlor/safener fb Propanil/Thiobencarb; T10=Propanil/Thiobencarb; T11=Propanil/Thiobencarb fb Bentazon/MCPA; T12=Quinclorac; T13=Quinclorac fb Bentazon/MCPA; T14=Season long weed-free; T15=Season long weedy.

The WCE based on the weed dry matter at harvest varied significantly among the herbicide treatments. Weed dry matter ranged between 29.77 and 176.43 g m⁻² and density between 89 and 350 plants m⁻². While, in weedy check the respective values were 382.20 g m⁻² and 472 plants m⁻² (Figure 1). Cyhalofop-butyl+Bensulfuron fb. Bentazon/MCPA showed the highest WCE (92%) closely followed by Bispyribac-sodium fb. Bentazon/MCPA (91%), Propanil/Thiobencarb fb. Bentazon/MCPA (88%) and Pretilachlor/safener fb. Propanil/Thiobencarb (87%). Cyhalofop-butyl+Bensulfuron, Propanil/Thiobencarb, Bispyribac-sodium and Pretilachlor/safener fb. Bentazon/MCPA also exhibited high WCE (>80%). Quinclorac fb. Bentazon/MCPA, Fenoxaprop-p-ethyl/safener fb. Bentazon/MCPA, Pretilachlor/safener and Quinclorac, on the other hand, provided satisfactory weed control with WCE values ranging from 70 to 79%. In this study, Fenoxaprop-p-ethyl/safener appeared as the least efficient herbicide with a WCE value of only 53%. Similar findings were observed in the case of weed density (Figure 1).

Rice yield

Rice grain yield was significantly influenced by herbicide treatments (Table IV). All the herbicide treatments yielded significantly higher than weedy check. Cyhalofop-butyl+Bensulfuron fb. Bentazon/MCPA and Pretilachlor/safener fb. Propanil/Thiobencarb performed best in terms of grain yield (3.49 t ha⁻¹ and 3.42 t ha⁻¹, respectively) though statistically similar to that produced by weed-free check (3.68 t ha⁻¹). Among other treatments, Propanil/Thiobencarb fb. Bentazon/MCPA, Bispyribac-sodium fb. Bentazon/MCPA, Cyhalofop-butyl+Bensulfuron and Propanil/Thiobencarb also produced high yield (>3 t ha⁻¹). Fenoxaprop-p-ethyl/safener and Quinclorac provided the lowest grain yield (2.16 and 2.42 t ha⁻¹, respectively) next to weedy check (1.04 t ha⁻¹).

Table IV. Grain yield, weed inflicted relative yield loss and yield increase over control of rice variety AERON 1 due to different weed control treatments (averaged over seasons).

Treatment	Grain yield (t ha^−1)	Relative yield loss (%)	Yield increase over control (%)
T1	3.09de	16.03	197.03
T2	3.25b-d	11.68	212.53
T3	3.14c-e	14.67	201.85
T4	3.49ab	5.16	235.83
T5	2.106i	41.58	107.69
T6	2.62f-h	28.80	151.92
T7	2.55gh	30.98	145.19
T8	2.87ef	22.28	175.96
T9	3.42a-c	7.06	228.84
T10	3.14c-e	14.67	201.92
T11	3.29b-d	10.59	216.34
T12	2.42hi	34.24	132.69
T13	2.75hg	25.27	164.42
T14	3.68a	0.0	253.84
T15	1.04j	72.0	0.0
T1=Bispyribac-sodium; T2=Bispyribac-sodium fb Bentazon/MCPA; T3=Cyhalofop-butyl+Bensulfuron; T4=Cyhalofop-butyl+Bensulfuron followed by (fb) Bentazon/MCPA; T5=Fenoxaprop-p-ethyl/safener; T6=Fenoxaprop-p-ethyl/safener fb Bentazon/MCPA; T7=Pretilachlor/safener; T8=Pretilachlor/safener fb Bentazon/MCPA; T9=Pretilachlor/safener fb Propanil/Thiobencarb; T10=Propanil/Thiobencarb; T11=Propanil /Thiobencarb fb Bentazon/MCPA; T12=Quinclorac; T13=Quinclorac fb Bentazon/MCPA; T14=Season-long weed-free; T15=Season long weedy.
Within a column, means sharing same alphabets are not significantly different at p=0.05 probability level according to LSD test.

Weed-inflicted relative yield loss (RYL) varied widely (5–41%) among the herbicide treatments (Table IV). In weedy check, RYL was very high (72%). Cyhalofop-butyl+Bensulfuron fb. Bentazon/MCPA allowed the least yield penalty (5%) followed by Propanil/Thiobencarb fb. Bentazon/MCPA (10%) and Bispyribac-sodium fb. Bentazon/MCPA (11%). Cyhalofop-butyl+Bensulfuron, Propanil/Thiobencarb and Bispyribac-sodium application resulted in<20% RYL. Fenoxaprop-p-ethyl/safener allowed the maximum RYL (41%). RYLs for the remaining herbicide treatments were moderate, ranging between 22% and 34%. Yield increase over control (YOC) varied due to herbicide treatments (Table IV). The highest YOC (253%) was obtained from season-long weed-free check. Among herbicide treatments, the maximum YOC was achieved through the application of Cyhalofop-butyl+Bensulfuron fb. Bentazon/MCPA (235%) closely followed by Pretilachlor/safener fb. Propanil/Thiobencarb (228%). Propanil/Thiobencarb fb. Bentazon/MCPA, Bispyribac-sodium fb. Bentazon/MCPA, Cyhalofop-butyl+Bensulfuron and Propanil/Thiobencarb registered ≥200% increased yield over weedy check. The lowest YOC (107%) was recorded with Fenoxaprop-p-ethyl/safener.

Figure 3 shows a strong correlation between weed control efficiency and rice yield (R²=0.92). Broadleaf weed control efficiency could predict grain yield more precisely than sedge control efficiency. Broadleaf weed control efficiency could explain grain yield by 94%, while sedge control efficiencies could predict grain yield by 78%.

Figure 3.  Relationship between weed control efficiency and grain yield of rice variety AERON 1.

Economic analysis

Herbicide treatments showed a wide range of economic return (Table V). Economic analysis revealed that the highest net benefit (RM 3106 ha⁻¹) was observed in Cyhalofop-butyl+Bensulfuron fb. Bentazon/MCPA closely followed by Bispyribac-sodium fb. Bentazon/MCPA (RM 2970 ha⁻¹), Pretilachlor/safener fb. Propanil/Thiobencarb (RM 2951 ha⁻¹) and Bispyribac-sodium (RM 2920 ha⁻¹). Fenoxaprop-p-ethyl/safener and Quinclorac appeared as the two least economic herbicides in this study with net benefits of RM 1946 ha⁻¹ and RM 2160 ha⁻¹, respectively. The remaining herbicides gave comparatively lower net benefits but still much higher than that of weedy check. The season-long weed-free plots showed the net benefit of only RM 1180 ha⁻¹which was very close to that obtained from season-long weedy plots (RM 1030 ha⁻¹), and comparatively lower than that of any of the herbicide treatments. The gross income was the highest with season-long weed-free plots (RM 3680 ha⁻¹) followed by Cyhalofop-butyl+Bensulfuron fb. Bentazon/MCPA (RM 3490 ha⁻¹) and Pretilachlor/safener fb. Propanil/Thiobencarb (RM 3420 ha⁻¹). The results clearly revealed that despite the highest gross income, net benefit was very low with season-long weed-free plots because of high cost involvement in manual weeding (RM 2500 ha⁻¹), while the chemical control required lesser investment ranged between RM 167 to 469 ha⁻¹ depending on cost of herbicide used. Thus, it appears that higher economic return could be achieved through herbicide application.

Table V. Cost effectiveness of different herbicide treatments (averaged over seasons).

	Ringgit Malaysia ha^−1
Treatment	Herbicide cost	Laborer cost for spraying/weeding	Total cost	Gross income	Net benefit
T1	110+114	50	274	3140	2866
T2	110+114+60	50+50	384	3490	3106
T3	120	50	170	3090	2920
T4	120+60	50+50	280	3250	2970
T5	210	50	260	2420	2160
T6	210+60	50+50	370	2750	2380
T7	154	50	204	2150	1946
T8	154+60	50+50	314	2620	2306
T9	252	50	302	3140	2838
T10	252+60	50+50	412	3290	2878
T11	117	50	167	2540	2373
T12	117+60	50+50	277	2860	2583
T13	117+252	50+50	469	3420	2951
T14	0.0	2500	2500	3680	1180
T15	0.0	0.0	0.0	1030	1030
T1=Bispyribac-sodium; T2=Bispyribac-sodium fb Bentazon/MCPA; T3=Cyhalofop-butyl+Bensulfuron; T4=Cyhalofop-butyl+Bensulfuron followed by (fb) Bentazon/MCPA; T5=Fenoxaprop-p-ethyl/safener; T6=Fenoxaprop-p-ethyl/safener fb Bentazon/MCPA; T7=Pretilachlor/safener; T8=Pretilachlor/safener fb Bentazon/MCPA; T9=Pretilachlor/safener fb Propanil/Thiobencarb; T10=Propanil/Thiobencarb; T11=Propanil/Thiobencarb fb Bentazon/MCPA; T12=Quinclorac; T13=Quinclorac fb Bentazon/MCPA; T14=Season-long weed-free; T15=Season long weedy.
RM=Ringgit Malaysia. 1 US$ =3 Ringgit Malaysia (RM) (approx).
Market price of commercial herbicides: Sofit N300 EC=70 RM L^−1, Halop 100 EC= 110RM L^−1, Tekong =19 RM 100 g^−1, Nominee 100 SC=98 RM 250 mL^−1, Satunil=42 RM L^−1, Rumpas M=43 RM 250 mL^−1, Quintox=42 RM 100g^−1 and Basagran M60=38 RM L^−1.
Manual weeding cost: 100 laborers ha^−1 for 2 weedings @ 25 RM laborer^−1 day^−1, Herbicide application cost: 2 laborers^−1 ha^−1 round^−1 @ 25 RM laborer^−1 day^−1, Market price of paddy: 1000.00 RM t ha^−1, Gross income=paddy yield (t ha^−1)×market price (RM t ha^−1), Net benefit=gross income – total weeding cost.

Discussion

Weed management is a fundamental practice, failure of which may result in severe losses in terms of yield and economic return. Weeds are dynamic in nature and a shift in their abundance and dominance is likely with changes in management practices. Herbicide is immediate, quick, convenient and the most economic tool against weeds, but recurrent use of the same herbicide for prolonged periods may result in herbicide-resistant weed biotypes. Rotation of herbicides with different modes of actions could safeguard evolution of weed biotypes.

The experiment was conducted under naturally occurring mixed weed populations comprising 20 species. On the basis of summed dominance ratio values, averaged over seasons, dominant weed species could be ranked in the order of: Physalis heterophylla>Scoparia dulcis>Cleome rutidosperma>Cyperus rotundus>Eleusine indica>Leptochloa chinensis>Fimbristylis miliacea. It was also noted that broadleaf weeds constituted more than 50% of the weed population. It is interesting that the weed composition and dominant weed species in the present study are quite different from those of a typical aerobic rice field. Jaya Suria et al. (2011) reported from their trial, conducted at Seberang Perai, Penang, Malaysia with the same aerobic rice variety, that grassy weeds constituted about 80% of total weed community and dominant weed species ranking was in the order of: Eleusine indica>Digitaria ascendens>Cyperus iria>Echinochlo colona>Calopogonium mucunoides>Mimosa invisa. In Karnataka, India, Gowda et al. (2009b) recorded Digitaria sanguinalis, Cynodon dactylon, Panicum repens, Cyperus rotundus, Cyperus iria, Euphorbia hirta, Phyllanthus niruri and Commelina benghalensis as the predominant weed species in aerobic rice fields. Greater abundance of sedges and grasses under dry seeded/aerobic conditions, and abundance of broadleaf weeds under saturated/flooded conditions have been reported by many researchers (Juraimi et al. 2011). The variation in the weed composition and dominance might be due to the differences in the agro-climatic conditions, cropping systems, growing season, cultural practices and weed seed bank composition and periodicity of germination pattern of different weed species among the experimental sites (Juraimi et al. 2010). In our study, Sorenson's index of similarity among different herbicide treatments ranged between 77.24 and 94.66%. Sukarwo (1991) stated that>75% of homogeneity is required to conduct weed control experiments. Thus, the experimental field was suitable for a weed control experiment because of good homogeneity.

The weed densities, averaged over seasons, recorded in the season-long weedy plots were 222, 491 and 472 plants m⁻² at 10, 40 and 75 DAS with corresponding biomasses of 16, 128 and 382 g m⁻², respectively. The high weed density as observed in this study confirms the findings of many researchers (Rao et al. 2007, Gowda et al. 2009a, 2009b, Mahajan et al. 2009, Jaya-Suria et al. 2011) who reported that weed pressure was the highest in direct-seeded aerobic rice among the rice ecosystems. Dry-tillage practices and aerobic soil conditions are highly encouraging for germination and growth of weeds (Balasubramanian and Hill 2002). Besides, when direct seeded, rice seeds germinate simultaneously with weed seeds without any “head start” over germinating weed seeds (Zhao et al. 2006), and initial flush of weeds is not suppressed by flooding (Olofsdotter et al. 2000). Rice ecosystem and cultural practices mostly determine dominant weed species, weed pressure, rice-weed competition and eventually, the weed control strategy. Thus, understanding the weed community along with dominance pattern is necessary for effective weed management.

The present study included 13 combinations of eight commercial herbicide products, most of which provided good weed control, and all of them significantly outnumbered and out-weighed weedy check in respect to yield attributes and yield. Most of the herbicides were found to be effective in arresting weed population and growth as well. The criterion of weed control efficiency (WCE) was considered as the percentage of weed dry weight that is reduced by a particular herbicide treatment in comparison with weedy check. WCEs of the tested herbicide treatments ranged between 53 and 92%. Only Fenoxaprop-p-ethyl/safener with/without Bentazon/MCPA and Quinclorac with/without Bentazon/MCPA exhibited poor WCE, because, both Fenoxaprop-p-ethyl and Quinclorac are graminicides and hence not effective against broadleaf and sedges constituting 80% of the weed community. Pretilachlor/safener also provided low WCE, which might be as this pre-emergence herbicide did not have a long residual effect to control late germinated weed seeds. All the remaining herbicide treatments showed excellent weed control because they contained both graminicides and broadleaf/sedge herbicides providing broad-spectrum weed control. The ultimate reflection of high WCE was the high grain yield. Some of the herbicide treatments like Cyhalofop-butyl+Bensulfuron fb Bentazon/MCPA and Pretilachlor/safener fb Propanil/Thiobencarb produced statistically similar grain yield to weed-free check, which was the consequence of high WCE. In contrast, herbicide treatments with low WCE yielded very poorly. Ashraf et al. (2006) and Hasanuzzaman et al. (2009) also observed herbicide-treated plots gave higher yield than weedy checks, and some herbicides produced as much as weed-free checks. The WCE of a particular herbicide treatment was also reflected in relative yield loss (RYL) and yield increase over control (YOC). The higher the WCE the lower the RYL and the higher the YOC. Reasonably, weed-free check enjoyed the highest YOC and weedy check, on the contrary, allowed maximum RYL. Earlier workers like Begum et al. (2008) and Jaya Suria et al. (2011) while studying with different herbicides in rice explained their results which are quite in line with ours. The removal of competitive effect of weeds reduces inter-specific competition for resources and enables the crop plants to utilize available resources more efficiently throughout the growth cycle, which in turn positively influences crop yield and biomass production (Gowda et al. 2009a).

All the herbicides investigated in the study possess high selectivity to rice without causing any considerable injury. Although a slight injury like leaf chlorosis along with growth stunting were evident with some of the herbicides like Pretilachlor/safener, Bispyribac-sodium and Bentazon/MCPA at 7 to 14 days after application, but thereafter phytotoxicity seemed no obvious impact. At the late-season evaluation, no injury was observed from any herbicide treatment in any season. Rate and application time for all the herbicides were followed as per manufacturers’ recommendation, which might result in no injury to rice plants. Moreover, aerobic soil conditions helped reduce herbicide injury to crop (Olsen et al. 2000). Bhagirath and Johnson (2011) also observed that rice phytotoxicity symptoms were greater when herbicides were applied in saturated soils than in aerobic soils. Jaya Suria et al. (2011) observed that Pretilachlor, Bispyribac-sodium, Propanil, Thiobencarb, Fenoxaprop-p-ethyl, Quinclorac and Bentazon/MCPA caused no injury to rice plants under aerobic soil conditions. Pacanoski and Glatkova (2009) also found no visual toxicity symptoms in rice with the application of different herbicides such as Bentazone, Propanil, Penoxulam and Bensulfuron-methyl. Ntanos et al. (2000), on the other hand reported slight foliar injury in rice with Cyhalofop-butyl and Propanil. Begum et al. (2008) also observed slight injury to rice plant treated with Pretilachlor, 2, 4-D (amine) and Bispyribac-sodium. Bhagirath and Johnson (2011) reported rice shoot biomass reduction due to application of Bispyribac-sodium in both aerobic and saturated soil culture. Thus, response of rice plants to herbicides in terms of phytotoxicity is variable. Crop growth stage, rate of herbicides, soil water content and various other environmental factors might affect the phytotoxicity of herbicide by altering herbicide absorption, translocation and metabolism.

Regression analysis also showed that unit increase in WCE of broadleaf and sedges caused increase in grain yield of 25 and 23 kg ha⁻¹. On average, unit increase in WCE resulted in an increase in grain yield of 26 kg ha⁻¹. This clearly indicates that high weed pressure imposes a serious threat to the productivity of aerobic rice, and effective weed management is crucial for higher yield under aerobic conditions. The increase in rice grain yield by increasing WCE has also been reported by Singh and Singh (2006). Singh et al. (2007) observed that rice yield and WCE against grasses, sedges and broadleaf weeds were positively correlated with correlation coefficients of 0.71, 0.94 and 0.91, respectively; unit increase in WCE of grasses, sedges and broadleaf weeds resulted in an increase in rice yield by 71, 93 and 91 kg ha⁻¹, respectively. This variation might be due to the differences in weed community dominance, rice variety, herbicides and cultural practices between experimental sites.

The cost-effectiveness of herbicides is affected not only by their price but also by their required dose and efficacy (Wibawa et al. 2010). An herbicide with high WCE may not be economic due to its high price and/or high dose. In the present study, cost of different herbicide treatments ranged from RM 167 to RM 469 ha⁻¹ depending upon the price and rate of application, while manual weed control required a high investment of RM 2500 ha⁻¹ for season-long weed-free checks. Although the maximum gross income was obtained from season-long weed-free treatment, due to high cost involvement in manual weeding, the net benefit was much lower (RM 1180 ha⁻¹) than that of any herbicide treatment (ranging from RM 1946 to 3106 ha⁻¹) and was very close to that obtained from season-long weedy check (RM 1030 ha⁻¹). The lowest gross income of only RM 1030 ha⁻¹ was recorded with weedy check due to its lowest productivity. In terms of net benefit, season-long weed-free check generated an additional return of only RM 150 ha⁻¹ over season-long weedy checks. In our study, the net benefit of the herbicidal weed control was two to three times higher than that obtained from manual weed control. Hence, manual weeding is less remunerative than herbicidal control, and keeping aerobic rice field weed-free manually throughout the season is a losing concern, confirmed by many others (Hussain et al. 2008, Gowda et al. 2009b, Mahajan et al. 2009, Mamun et al. 2011).

By economic analysis it was evident that different herbicide treatments varied widely in terms of gross income and net benefit. Highest gross income and net benefit were realized from the sequential application of Cyhalofop-butyl+Bensulfuron and Bentazon/MCPA. Despite the high prices of Cyhalofop-butyl+Bensulfuron and Bentazon/MCPA (RM 110 L⁻¹+RM 190 kg⁻¹ and RM 38 L⁻¹, respectively) this herbicide combination appeared as the most economic because of high WCE (92%) and lower dose. The second most cost-effective herbicide treatment was Bispyribac-sodium fb Bentazon/MCPA. Although Bispyribac-sodium fb Bentazon/MCPA produced lower gross income compared with that of Pretilachlor/safener fb Propanil/Thiobencarb, WCE was higher and application rate was lower in the former herbicide combination which led to higher net benefit. Fenoxaprop-p-ethyl/safener was one of the cheapest (RM 43 L⁻¹) herbicides with low application rate (0.06 kg a.i. ha⁻¹), but its net benefit was the lowest (RM 1946 ha⁻¹) due to the poorest WCE (53%). Economic performance of Quinclorac also was poor (RM 2160 ha⁻¹) as the consequence of lower WCE (70%) accompanied by higher price (RM 420 kg⁻¹) and higher dose (0.25 kg a.i. ha⁻¹). Thus the cost-effectiveness of a herbicide is determined by its efficacy, application rate and price.

The encouraging findings of our study revealed that despite high weed control efficiency, manual weeding is not cost-effective, whilst chemical weed control are highly efficient and economic as well. Among the tested herbicides, Cyhalofop-butyl+Bensulfuron fb. Bentazon/MCPA or Bispyribac-sodium fb. Bentazone/MCPA or Pretilachlor/safener fb. Propanil/Thiobencarb may be considered for their high efficacy and cost-effectiveness for weed management in aerobic rice field. A close look into these selected herbicides indicates their different mode of actions and their alternate use could help resist herbicide resistance in weeds. Therefore, those herbicides could be recommended in the study area and other areas with similar agro-climatic conditions and weed community. As part of a resistance management strategy, long-term changes in weed flora, herbicide efficacy, resistance, and crop productivity should be monitored regularly for sustainable weed management. Further research is needed to develop timing strategies and different tank mixture combinations for applying these herbicides with minimum effective dosages which could be most economic and ecologically desirable weed management approach for aerobic rice.

Acknowledgement

The authors sincerely acknowledge UPM Research University Grant (01-04-08-0543RU) for financial support of the project.