Weed Management in Small Holder Rice Production in the Tropics

David E. Johnson 
Natural Resources Institute
University of Greenwich
Chatham, Kent, UK.

"more energy is expended for the weeding of man's crops than for any other single human task" Holm, 1971

Rice is the most important food crop in developing countries, and accounts for 29% of the total calorie intake of these populations. In the tropics, rice is largely grown on small family farms, which are usually less than 4 ha, and in Asia, rice is the most important source of employment in rural areas. Throughout these smallholder systems, weeds are one of the major biological constraints to production.

Rice Cultivation

Rice is grown throughout the tropics in rainfed uplands, seasonally deep flooded areas, and in rainfed and irrigated lowlands. Upland rice describes the cultivation of "rainfed" rice on well drained, non-irrigated fields, and the crop is grown in Asia, Latin America and Africa. Rice grown on seasonally deep flooded areas, "deep water or floating rice", is important in many areas in West Africa and Asia, often along the courses of major rivers. This crop is sown before the flood waters rise and is able to survive the subsequent deep flooding. Lowland rice describes crops where the fields are inundated for at least part of the time between establishment and harvest, and may depend on rainfall or irrigation. Table 1 shows regional rice yields (t/ha) compared to total population and the annual increase in the area cultivated to rice. The average regional yields conceal substantial variations depending upon the production systems and levels of inputs.

Table 1. World Rice Production (1991).


Population (million)

Share of the rice yield (%)

Labour force (million)

Yield (t/ha) in agriculture rice area

Latin America




















Upland Rice

Almost 100 million people depend on upland rice as their staple, including some of the World's poorest farmers. In West Africa, upland rice comprises 57% (1.8 million ha) of the total rice area. Upland rice is grown in diverse systems, ranging from shifting cultivation to relatively intensive systems, utilizing hand, animal or mechanized tillage and rotations with other crops, including cotton, legumes and other cereals. Shifting cultivation occurs throughout the humid forest zone, where land is cleared from forest, usually by slash and burn, and rice is grown for one or more seasons before the land is returned to fallow. Invasion by weeds is a principal reason for abandoning land after periods of cultivation. Upland rice is dibble, broadcast, or row seeded, the former being commonly used in shifting cultivation. The most common method of weed control is by hand, often with the aid of hoes or machetes. In Asia, Africa and parts of Latin America upland rice is typically grown with few, if any, purchased inputs. A wide range of weeds infest upland rice, many of which are pan-tropical, including the grass weeds: Digitaria spp., Echinochloa colona, Eleusine indica, Paspalum spp., and Rottboellia cochinchinensis, and the broadleaf weeds: Commelina spp., Ageratum conyzoides, Portulaca oleracea, Amaranthus spp. and Euphorbia spp. The variability of weed species composition in upland rice tends to be greater than in the other production systems, and is dependent upon ecology, the cropping system and management practice.

Lowland Rice

These systems vary considerably according to the degrees of land levelling, drainage, bunding, to enable either rainfall to be retained or irrigation water supplied to flood the rice crop. The rice crop may either be direct sown or transplanted from a nursery bed. Production systems with effective water control present the greatest technical opportunities for intensification and diversification of the cropping system. In Asia, the majority of land suitable for intensive lowland rice production is already under cultivation. Flooding usually improves soil chemical conditions and the anaerobic soil environment prevents the germination and growth of many weeds. In the direct seeded systems however, where the field cannot be adequately flooded until the crop is established, post-emergent weed control is usually essential. Weed infestations in irrigated rice are frequently exacerbated by poor land levelling and preparation, inadequate water management, irrigation water and rice seed contaminated with weed seeds, direct seeding, and no crop rotation. In West Africa, of the 20-50 million ha of inland valley bottoms and hydromorphic fringes, only about 15% are currently utilized for crop production. Appropriate weed control regimes are a vital component of sustainable development for such areas. The most serious weeds of lowland rice are pan-tropical and the more widespread include sedges, such as Cyperus difformis, C. iria, Fimbristylis spp. and Scirpus maritimus, the grasses Echinochloa crus-galli, E. crus-pavonis, E. glaberescens, E. pyramidalis, Ischaemum rugosum, Leptochloa spp., Oryza barthii, O. longistaminata, O. rufipogon and the broadleaves Ludwigia spp., Eclipta prostrata and Sphenochlea zeylanica. These weeds are well adapted to the aquatic environment, capable of rapid growth and multiplication and are very competitive with rice. The similarity of some weeds to rice, such as Echinochloa spp. at early stages of growth, makes it very difficult for farmers to distinguish them while hand weeding. Some lowland weeds favor particular environments, such as the rhizomatous grasses Paspalum vaginatum and P. distichum in the coastal and inland saline marshes. The distribution to farmers of rice seed contaminated with weed seeds is at least partly to blame for the widespread occurrence of many of the more serious rice weeds. The introduction of weeds such as Oryza spp., Ischaemum rugosum, Echinochloa spp., Rottboellia cochinchinensis, and Euphorbia heterophylla into previously un-infested areas has occurred in this way. In the lowland rice systems, weed seeds may also spread rapidly in irrigation and flood waters.

Losses Due to Weeds

In a survey of upland rice producing countries covering 80% of the total production area, weeds were the most widely reported biological constraint to yields. Upland rice, in particular, competes poorly with weeds and uncontrolled weed growth often results in negligible or zero yield. In West Africa, yields of upland rice with farmers' weed control, were 44% lower than on researcher weeded plots. Losses due to uncontrolled weed growth in upland rice in India were up to 90%, and in both lowland and upland systems in Africa losses were within the range 28-100%. In Belize, Central America, two years after clearing a fallow, yields of upland rice with no weed control were less than 20% of those when the crop was hand weeded twice after sowing. Losses can be particularly severe in direct seeded lowland rice, as the rice and weed seedlings are at similar growth stages. In transplanted rice, the young rice plants have an advantage over germinating weeds and immediate flooding after transplanting limits the establishment of many weeds, hence yield losses to weed competition tend to be less than those in direct seeded rice. In Asia, yield losses due to uncontrolled weed growth in direct seeded lowland rice was reported to be between 45-75%, and for transplanted lowland rice approximately 50%. It is rare however, for farmers not to undertake some weed control and therefore losses on farmers' fields are likely to be considerably less.



Fig. 1. Minimizing weed competition during the early stages of the crop, before it has formed a closed leaf canopy, is particularly important. In upland rice this critical period is approximately 15-40 days after seeding, while in transplanted rice, the crop can form a canopy more rapidly. Where a crop a exposed to prolonged weed competition during this critical period it is not usually able to recover sufficiently to give a good yield.

Weed Control Methods

To a greater extent than in the management of insect pests and diseases, the management of weeds requires integrated strategies to be successful. While in some production systems, herbicides may provide the main means of control, these alone are unlikely to be successful unless combined with good land preparation and, in the lowland systems, good water control. No one weed control method is likely to control all weeds, and in the long term this can lead to a build-up of certain species. The combination of direct weed control methods, such as herbicides or hand weeding, with indirect methods such as land preparation, flooding and a competitive crop, will help prevent this situation. A combination of direct weed control methods such as herbicide use with subsequent hand weeding can result in improved yields (Table 2). The optimum combination of weed control methods will depend on the farming system, economic conditions, and farmers' resource and knowledge base.

Table 2. Effect of different weed control methods on grain yield in lowland dry seeded rice.

Control method

Grain Yield t ha





Butachlor + hand weeding


Pendimethalin + hand weeding


Propanil + MCP


Land Preparation

Tillage serves to provide a suitable soil tilth for a seed-bed and control weeds prior to crop establishment. In smallholder systems, practice varies from zero tillage, as in many of the systems of shifting cultivation, to repeated deep cultivation to remove troublesome perennial weeds, such as Oryza longistaminata which has an extensive rhizome system. Shallow tillage is often ineffective in controlling weeds and, regardless of tillage practice, post-emergence weed control is normally necessary.


The flooding of fields is the most effective method of cultural control of weeds in rice. Flooding to a depth of 10 cm prevents germination of most weed seeds and kills the majority of weed seedlings (Table 3). Normally, flooding is used in conjunction with other control measures, such as herbicides or hand weeding. However, for flooding to be successful, water levels must be maintained and fields well levelled to ensure an even depth of water. In many smallholder schemes, limited irrigation water and poor land development can be major constraints to effective weed control.

Table 3. Effect of water depth on weed growth in rice.

Water depth cm

Weed weight g m-2









Hand weeding Hand weeding is the most widely used weed control method, with availability of labour being the main limitation to its effectiveness. In some areas, adoption of line planting in transplanted rice has allowed the introduction of rotary weeders for cultivation between rice rows, considerably reducing labour requirements for weed control. In India, hand weeding rice at 15 and 30 DAS gave a 60% increase in yield over a single hand weeding at 30 DAS. However, because hand weeding is laborious, with labour sometimes expensive and in short supply, weed control is often imperfect and/or delayed. In a survey of rice farmers in Côte d'Ivoire, 53% said that their fields were not always weeded, with the most common explanation being that weeding was not considered worthwhile due to severe weed infestation, thus effectively abandoning the crop. Further, 80% of farmers said that if weeds were less of a problem they would increase the area of land under cultivation. Several constraints limit the effective use of hand weeding, including household labour constraints, limited cash for hiring labour, and labour not being available for hire during peak periods. In a village survey in Côte d'Ivoire, where no herbicides were used, farmers spent 408 and 506 hours/ha hand weeding upland and lowland rice, respectively. In Asia, weed control in upland rice has been reported to require 32 - 198 man days per ha, representing 17 - 57% of the total labour requirement for the crop. The availability of animal traction can alleviate the constraints posed by reliance on hand labour. Technology including an animal drawn row seeder and hoes, enabling mechanical weed control, may be an appropriate package where animal traction is a possibility, although problems have been encountered with the operation of seeders under farmer conditions.

Fallow systems Man's inability to control weeds is one of the major constraints to continuous cultivation in the tropics, and is a particularly acute constraint for those with limited resources and little access to appropriate technologies. After initial clearance of the forest, weed growth consists largely of broadleaf weeds and forest regrowth, though with repeated cropping, more problematic grasses invade the fields. Studies have shown that weed growth following a fallow of less than three years was almost twice that following a fallow of five or more years. Where the cropping cycle is short compared to the length of fallow, forest rapidly regenerates from seeds, roots and cut stumps. However, repeated cultivation allows invasion by annual and perennial weed species and replacement of native forest species by weeds such as Imperata cylindrica and Chromolaena odorata. While shifting cultivation has been criticized as being wasteful of land and forest resources, it is still the main means of production for many resource-poor farmers. In such systems, the fallow period is an integral part of the cropping cycle. In many areas where shifting cultivation is practiced, the soils are inherently infertile. Cropping intensification on such areas therefore, requires technologies to address increased weed pressure and declining soil fertility, if such systems are to remain sustainable. Research has aimed at improving fallows, often utilizing legumes to enhance soil fertility and particularly nitrogen status, reduce weed growth and protect the soil from erosion. Possible legumes species for rice-based systems include the "creeping" species of Calopogonium, Pueraria and Mucuna, and shrub species such as Sesbania and Aeschynomene.

Crop competition Rice is not very competitive with weeds during the seedling stages, though this can be an important factor during the vegetative and reproductive stages. Crop competitiveness with weeds is particularly important to limit weed infestation after the initial weed control treatments , such as land preparation, herbicides and hand weeding, have been undertaken. Crop and weed interactions largely involve the competition for light, water and nutrients. Where water and nutrients are not limiting, light becomes the most important limiting factor. Weeds however, usually have higher growth rates and nutrient demands than rice, hence the amount and timing of fertilizer application can significantly effect weed-rice competition. In an experiment on moist but not flooded soil in West Africa, an application of 50 Kg N ha-1 to rice doubled the weed biomass in the crop. Suggesting that in such situations, fertilizer application would have to be accompanied by improved weed management. Reducing the distance between rows or hills, has been shown to reduce weed infestation in rice in upland, hydromorphic and lowland ecologies (Table 4). Likewise, increased weed populations can decrease crop yield loss (Table 5). In irrigated lowland systems in West Africa, crop seed rates up to 200 kg/ha have been reported to be used by farmers to improve weed suppression by the crop. It has been noted that some "modern" higher yielding rice varieties are less able to compete with weeds than many of the traditional rice varieties they were intended to replace. Traditional varieties tend to be tall, of longer duration, droopy leaves and high vegetative vigor, while the modern varieties often combine short stature, short growth duration and erect leaves. This suggests that increases in yield potential of modern rice varieties compared to tall, leafy traditional varieties has been achieved by sacrificing their competitive ability with weeds. The lack of weed competitiveness in modern varieties may be one reason that many upland rice farmers have retained traditional varieties. The competitive ability of different rice varieties has become a focus of research, with the intention of combining competitive ability with other desirable characteristics in order to develop varieties suitable to low-input conditions. Studies on different plant types show that tillering ability, height, leaf canopy and root development may be important factors in determining the competitive ability of rice plants. The African rice, O. glaberrima, is a source of a number of these traits which confer competitive ability with weeds, and to produce rice plant types which are more suited to smallholder conditions O. glaberrima/O. sativa hybrids have been produced. Crop varietal development related to weed control also includes host plant resistance to the parasitic weed, Striga, and the possibility of utilizing crop allelopathic activity. Rice cultivars resistant to Striga, a localized problem in some savanna areas of Africa, have been identified and field tested, providing a potential solution for affected areas. Allelopathy is the ability of one plant to interfere with the growth of another by the production of root exudates. Javanica type O. sativa, O. glaberrima, and wild rices have been suggested as possible sources of allelopathic activity, which could be transferred into commercial rice cultivars. While the utilization of allelopathy may have considerable potential for low input weed management regimes, the availability of such technology to the farmer is not likely to occur in the near future.

Table 4. Effect of plant spacing on weed weight in lowland irrigated rice, West Africa.

Plant Spacing cm

Weed weight g m-2

Grain Yield t ha

















Table 5. Grain yield loss due to competition from Echinochloa sp.

Weed density (plants/m-2)

Yield loss %







Chemical control

The importance of herbicide use in the tropics is closely related to the cost and availability of labour. Herbicides are one of the first labour saving technologies to be adopted as labour costs rise. As a consequence, the use of herbicides varies considerably between countries. Many rice crops in Japan receive more than two herbicide applications, while in the Philippines only about 50% of the rice area is treated, and in Bangladesh, where rural labor is relatively inexpensive, there is little herbicide use. Herbicides replace hand weeding and enable direct seeding rather than transplanting, which is less labour demanding. Direct seeding in the lowlands is linked to the use of herbicides, as without their use the weeds grow so rapidly in the stages before the fields can be flooded, that manual means of control are often not feasible. Herbicides are also used in the transplanted systems, though to a much lesser extent, and in the upland systems particularly where rice is grown in rotation with cash crops. The costs involved with herbicide use are likely to remain a major constraint to their widespread adoption in many regions. Herbicides may be classified as non-selective or selective, and pre- and post emergence. Most herbicides used in rice production are selective, controlling some or most weeds, while having a limited effect on the crop. Selectivity is not necessarily dependent upon the compounds, but also on the rates, timing and methods of application, and hence it is important to follow the manufacturers recommendations. Non-selective herbicides such as glyphosate are sometimes used before establishing rice, on weed infestations such as wild rice which are difficult to control with selective herbicides. Pre-emergence herbicides are applied to the soil and control weeds before they emerge, while post-emergence are applied to weeds after they emerge. Among the anilide group are the herbicides butachlor, pretilachlor and propanil. Butachlor can be applied either as pre-emergence or early post emergence to give control over a wide range of annual grasses and some broadleaf weeds. Pretilachlor can be used safely on transplanted rice, or with direct seeded rice if applied with an antidote. Propanil, a contact herbicide with no residual effects, is a well established herbicide which gives control over a range of grasses and some broadleaf weeds at the 2 to 3 leaf stage, though control is less effective at later stages. Propanil is widely used in proprietary herbicides combinations or in tank mixtures. Elsewhere in the world, such as in Latin America, where propanil has been used intensively some resistance to the herbicide has developed among Echinochloa spp. Alkanoic acids (phenoxy acetic acids) include: 2,4-D, 2,4,5-T, MCPA, and triclopyr, and these are foliar applied, translocated herbicides which selectively control broadleaf weeds, sedge and some grass seedlings. These herbicides have been widely used in proprietary and tank mixtures, particularly with propanil, to give a wide spectrum of control; such mixtures are among the more favorably priced herbicide treatments. Due to the high biological activity of some of these compounds eg. 2,4-D, there is degree of risk attached to their use, as spray drift or contamination of application equipment may affect crops such as cotton and a range of the vegetables. In recent years, 2,4,5-T has been withdrawn from many markets due to concern for human health. Thiocarbamates include molinate and thiobencarb, both soil active herbicides. The former requires moisture to be active and once applied the area should be continuously flooded; molinate is particularly effective at controlling Echinochloa spp. Thiobencarb controls a range of grasses and sedges and may be applied pre- or early post emergence. The dinitroanilines include the pre-emergence herbicides butralin and pendimethalin. These are effective against a range of broad leaf weeds and grasses when applied before the germination of weeds, though they have little effect against established weeds. The latter gives good control over Rottboellia cocohichinensis and Setaria spp. Diphenyl ethers include bifenox, flourodifen and oxyflurofen. These are used as pre-emergence or early post-emergence herbicides and are effective against a range of broadleaf and grass species at the seedling stages. Simetryn and dimethametryn are triazine compounds used as selective herbicides in rice, and are applied at pre or early post emergence. The former may be combined with thiobencarb, while the latter is sold as a proprietary mix with either piperophos or pretilachlor. Bensulphuron-methyl is a herbicide of the sulphonylurea group, and can be used in both seeded and transplanted rice. Applied as pre-emergence or early post emergence, the herbicide gives control over a wide range of broad leaf weeds and sedges. In common with other members of the sulphonyl ureas, this herbicide has a high rate of biological activity, and only very small quantities of the active ingredient are required for effective weed control. Bentazon is a contact herbicide which gives good post emergence control over a range of broad leaf and sedges weeds. Bentazon is marketed either as a sole product or as a mix with propanil. Oxadiazon is a pre-emergent herbicide though with limited post emergent action. It controls wide range of weeds and has good persistence in the soil due to low solubility in water and high adsorption in the soil. It is available in various formulations allowing it to applied with conventional sprayers to the soil alone, or with propanil, or using shaker bottles directly to the flood-water in the rice field.

Problems with herbicide application

Smallholder farmers often face a number of problems related to herbicide use, due to either an inadequate rate of herbicide being applied, or the herbicides being applied too late to provide good effect on the weeds. A major cause of this is likely to be the serious lack of information available to the farmer and the poor level of understanding that results. Often only minimal precautions are taken with regard to health and safety and the use of herbicides. While the majority pesticides applied to agricultural crops are probably insecticides, and that herbicides are usually less toxic to man, it is still a matter of concern. Poisoning may occur through ingestion of pesticides, skin absorption, or inhalation, and lack of appropriate protective equipment and training make it almost inevitable. Poor standards of pesticide use and ignorance of the possible dangers could be blamed on low levels of education, but poor packaging with inadequate warnings and a lack of information available to the farmer are also major factors. In the early 1970's in Asia, formulations of 2,4-D and MCPA were recommended for controlling annual weeds in transplanted rice, while granular formulations of the selective herbicides butachlor and thiobencarb were reported effective in direct seeded rice, as alternatives to hand weeding. The subsequent adoption of such herbicides was widespread. Since that time, herbicides have been developed which are safer to use in rice and which have allowed much greater flexibility in application. Formulations allowing the application of herbicides directly to irrigation water without the use of spraying equipment have advantages for the small farmer and have become established practice in many areas. With frequent use of herbicides, herbicide resistance has evolved within some weed populations. In the USA, 30 years of propanil use resulted in resistant Echinochloa sp., and after four years of continuous use, bensulfuron resistance emerged in four aquatic weed species. In Costa Rica, Ixophorus sp. and Eleusine indica were found to have evolved imazapyr resistance after about five years of herbicide use. Studies suggested that selected biotypes of Ixophorus were between 5 to 80 times more resistant to imazapyr than the most susceptible biotypes. The evolution of herbicide resistant weeds is a real threat to effective weed control where herbicides are frequently used. Smallholder systems may be particularly vulnerable as herbicide are often not used at appropriate times or dosages, which may hasten the development of resistance.


There have been tremendous developments in herbicide technology in the past forty years, with a wide range of pre- and post-emergence herbicides now available to farmers. Improved selectivity and formulations allow safer, easier and more flexible application. Herbicide development and the results that can be achieved have been spectacular, but comparatively little research has been focused on cultural control and integrated weed management. While many rice production systems have come to rely on herbicides, the need to reduce costs, and the evolution of new weed problems and herbicide resistant ecotypes, suggests there should be greater emphasis on the judicious use of herbicides, integrated with cultural methods. Further, it is likely that a substantial proportion of smallholder rice farmers, will for the foreseeable future continue to rely on weed control methods other than herbicides. Factors mitigating against adoption will continue to be; insufficient funds, inadequate technical support and supplies, an insufficient cost:benefit ratio to cover the investment risk, and a preference for traditional methods. For such farmers, improved weed control measures which build on traditional practice and which are compatible with farmer resources, will be more appropriate.


The author is funded by the UK Overseas Development Administration, and is based at the West Africa Rice Development Association, Côte d'Ivoire.


  • Ampong-Nyarko, K & De Datta, S. K. 1991 A Handbook for Weed Control in Rice. International Rice Research Institute, P O Box 933, 1099 Manila, Phillipines.
  • De Datta, S. K. 1981. Principles and Practices of Rice Production. John Wiley, New York.
  • De Datta, S. K. 1986. Technology development and the spread of direct seeded flooded rice in South East Asia. Experimental Agriculture. 22, 417-426.
  • Grist, D. H. 1986 Rice. 6th ed. Longman, London
  • Moody, K (1990) Yield losses due to weeds in rice in the Philippines. In Crop Loss Assessment in Rice. International Rice Research Institute, Los Banos. pp. 193-202.
  • Moody, K. (1993) Weed Control in Wet-Seeded Rice. Experimental Agriculture Vol 29, 393-403.