Cookies on Plantwise Knowledge Bank

Like most websites we use cookies. This is to ensure that we give you the best experience possible.

Continuing to use means you agree to our use of cookies. If you would like to, you can learn more about the cookies we use.

Plantwise Knowledge Bank
  • Knowledge Bank home
  • Change location
Plantwise Technical Factsheet

water hyacinth (Eichhornia crassipes)

Host plants / species affected
Oryza sativa (rice)

The initial leaves of seedling E. crassipes are elongated and strap-like, but soon develop the familiar spathulate form and, under suitable unshaded conditions, swollen petioles which ensure that, once dislodged, the seedlings will float from the mud into open water. The plant is very variable in size, seedlings having leaves that are only a few centimetres across or high, whereas mature plants with good nutrient supply may reach 1 m in height. Plants in an uncrowded situation tend to have short, spreading petioles with pronounced swelling, while in a dense stand they are taller, more erect and with little or no swelling of the petioles.

The plant system consists of individual shoots/crowns each with up to ten expanded leaves arranged spirally (3/8 phyllotaxy) and separated by very short internodes. As individual shoots develop, the older leaves die off leaving a stub of leafless dead shoot projecting downwards. This may eventually cause the whole shoot to sink and die.

Leaves consist of petiole (often swollen, 2-5 cm thick) and blade (roughly round, ovoid or kidney-shaped, up to 15 cm across). The base of the petiole and any subsequent leaf is enclosed in a stipule up to 6 cm long.

Roots develop at the base of each leaf and form a dense mass: usually 20-60 cm long, though they can extend to 300 cm. The ratio of root to shoot depends on the nutrient conditions, and in low nutrient conditions they may account for over 60% of the total plant weight. They are white when formed in total darkness but often purplish under field conditions, especially in conditions of low nutrients.

Periodically, axillary buds develop as stolons, growing horizontally for 10-50 cm before establishing daughter plants. Extremely large populations of inter-connected shoots can develop very rapidly, though the connecting stolons eventually die.

The inflorescence is a spike which develops from the apical meristem, but tends to appear lateral owing to the immediate development of an axillary bud as a 'renewal' or 'continuation' shoot. Each spike, up to 50 cm high, is subtended at the base by two bracts and has 8-15 sessile flowers (rarely 4-35). Each flower has a perianth tube 1.5 cm long, expanding into six mauve or purple lobes up to 4 cm long. The main lobe has a bright-yellow, diamond-shaped patch surrounded by deeper purple. Once the inflorescence is fully emerged from the leaf sheath, flowers all open together, starting at night, completing the process in the morning and withering by the next night when the peduncle starts to bend down. Each capsule may contain up to 450 small seeds, each about 1 x 3 mm.

The flowers are tristylous. They have six stamens and one style, arranged in three possible configurations (floral trimorphism) - with short style (and medium and long stamens), medium style (short and long stamens) or long style (short and medium stamens). The medium style form is genetically dominant and is by far the commonest form in almost all infested areas. The short-styled form is only known from South America, whereas the long-styled form is found commonly in South America, more rarely in South-East Asia and very rarely in Africa. Only in Sri Lanka is the long-styled the commonest form. Some other tristylous species show incompatibility between the different forms but E. crassipes does not. Hence pollination (mainly by wind) can result in good seed set, though in some populations there may be a higher degree of self-incompatibility.

Prevention and control

Cultural Control

Although the exact nutrient threshold below which E. crassipes will not flourish is not yet clear, it is certain that its vigour is directly related to available levels of nitrogen and phosphorus. Wherever possible, nutrient levels in the water body should be reduced or controlled: for example, by processing sewage or other nutrient-rich water, or by diverting it away from critical areas In South Africa, Coetzee and Hill (2012) suggest that the first step in any control programme should be to reduce the nutrient status of the water body, as a meta-analysis of studies that investigated the combined effect of P and nitrogen (N) water nutrient concentration and control agent herbivory showed that water nutrient status was more important than herbivory in water hyacinth growth.

Mechanical Control

Where E. crassipes is causing the most acute problems (e.g. impeding access for fishermen, or threatening to block harbours or damage hydro-electric installations), an effective solution may be the use of floating booms or fixed barriers to prevent movement into the critical areas. Booms may also be used to try and prevent movement of the weed down rivers, though their success will depend on their design (complicated by the need to maintain navigability along the river), the mass of material involved and the capacity to clear the booms by physical removal of weed.

Physical removal or destruction of the infestation may be achieved on a small scale by manual removal. On the larger scale, machinery is needed, either shore-based, or mounted on boats. Where possible, on smaller water bodies, reliance should be placed on unspecialized shore-based equipment (e.g. drag-lines, excavators, moving-belt elevators etc.), the weed being pushed to the shore by suitably modified boats. For larger water bodies, special boats may be needed with suitable harvesting equipment, together with a means of crushing the weed or otherwise reducing the volume of water. Where the water body is sufficiently large and deep for the weed to be returned to the water after crushing, without risk of decomposition causing deoxygenation, the use of such equipment may be economic. If the weed has to be transported to the land for unloading, the running costs become much greater and such methods may not be economic. 

Julien (2008) reviews biological aspects of E. crassipes related to management, and suggest that containment and eradication from a catchment may only be accomplished if the invasion is very young, small, isolated and accessible, and if the short-term resource commitment is high. Jyoti and Garima (2013) present methods of control including manual pulling and harvesting.

Chemical Control

2,4-D has been widely used for control of E. crassipes. Best results are achieved under conditions of rapid growth, high temperature and high humidity, when most plants of any age will be killed and sink within 2-4 weeks. Under less favourable conditions, some plants may regrow and require repeat treatment. In any case re-treatment is almost inevitably required after a few months as a result of re-infestation from incompletely sprayed plants, re-invasion from outside the sprayed area, or regrowth by seedlings.

Glyphosate has been tested and used for control of E. crassipes. It is much more expensive than 2,4-D but has possible advantages over 2,4-D in not causing taint of drinking water and in causing a slower kill of the weed, apparently reducing the risks of deoxygenation during decomposition (Findlay and Jones, 1996).

Other herbicides that have been used include the contact herbicides paraquat and diquat, but these have high mammalian toxicity and should not normally be used. Diquat use is described by Pitelli et al. (2011), who suggest that night application is more effective than day spraying. Aminotriazol [amitrole], ametryn and terbutryn can each be effective alone, but have been most often used in mixture with 2,4-D. Wersal and Madsen (2010) evaluated the use of penoxsulam, which gave effective control which was not improved by applying in combination with diquat.

New herbicides in the imidazolinone and sulfonylurea groups have been shown to have high activity on E. crassipes, but have not yet been adequately tested. These and other possibilities have been summarized by Price (1993).

Herbicides have rarely been used with complete success, owing to the need for repeated treatment over a long period, requiring dedicated management and organization. Apart from the problems of limited success, the use of 2,4-D and other herbicides can be unsatisfactory in several other respects. Ester formulations of 2,4-D can be highly toxic to aquatic organisms as well as creating a vapour drift problem. While the direct toxicity to aquatic organisms of 2,4-D amine salt formulations and the other listed compounds is largely negligible at the concentrations reached in the water, there can be devastating stress caused by deoxygenation as the weed dies and decomposes. Other problems include those of taint of drinking water by 2,4-D, and, for any herbicide that is used, damage by spray-drift onto non-target crops and other plant life adjacent to sprayed areas.

Biological Control

Seven arthropods and three fungi have been developed and released for the biocontrol of E. crassipes (Harley, 1990; Julien and Griffiths, 1998). The arthropods are the curculionids Neochetina bruchi and Neochetina eichhorniae, the pyralids Xubida infusellus and Niphograpta albiguttalis, the noctuid Bellura densa, the mirid Eccritotarsus catariensis, and the galumnid mite Orthogalumna terebrantis. The fungi are all hyphomycetes: Acremonium zonatum, Cercospora piaropi and Cercospora rodmanii. Additionally, there has been work on the development of the fungus Alternaria eichhorniae as a mycoherbicide (Aneja, 1996; Shabana, 1997). Acremonium zonatum, Cercospora piaropi, Myrothecium roridum, and Rhizoctonia solani are viewed as suitable bioherbicides (Charudattan, 2001). In Africa, an international programme has been established to develop a mycoherbicide for the control of the weed, using fungal isolates that have been found in Africa (Bateman, 2001). Karim Dagno et al. (2012) review the current status of development of mycoherbicides against E. crassipes, but report that biological, technological and commercial constraints have hindered progress. Oil emulsions are recognized as a way to increase both efficiency of application and efficacy of biocontrol agents

The two Neochetina weevils have together given excellent results in the USA, Argentina, India, Australia and Sudan, acting apparently in a complementary fashion. Infestations of E. crassipes have been reduced by 80-90% or more. In Uganda, the two weevils have greatly reduced the problem on Lake Kyoga, and are beginning to take effect on Lake Victoria (Hill, 1999). In Papua New Guinea, N. eichhorniae is reported to be giving 'permanent control' in some areas (Orapa and Atip, 1996). More recently, Orapa and Julien (2001) reported that although control had been achieved in some areas, such as the Sepik River and Waigani Lake, the full impact of biological control by the Neochetina weevils on water hyacinth in PNG is not known.

Some successful control programmes have been recorded in Mexico (Panduro and Domunguez, 1998), Benin, South Africa, Zimbabwe and Malawi. Control takes from 2 to 10 years depending on the location and the environmental conditions, but in some locations (including the countries mentioned) the weevils do not appear able to control the weed.

Adult weevils feed on the leaf and petiole surfaces, preferentially on the youngest leaves (Center, 1985). They make distinctive, almost square, feeding scars. This may cause significant loss of functional leaf surface and also may allow entry of pathogens, with the potential in extreme situations for removing over 50% of the laminar area (Van and Center, 1994). However, the most significant damage is caused by the larval stages. Eggs are laid in the petioles. Upon hatching, the larvae burrow down the petiole into the crown of the plant where they can cause major damage (Patnaik et al., 1988). The weevils pupate underwater in the roots. Under certain circumstances the adults can migrate through flight (Buckingham and Passoa, 1984). This damage to the petiole often results in complete collapse of the leaf and eventually in loss of buoyancy so that the whole plant sinks. Each of the two Neochetina species has small but distinct differences in biology, ecology and feeding habits, which result in additive, complementary effects. N. bruchi are slightly smaller weevils and develop faster but in many locations including Florida, USA, and Benin, N. eichhorniae is the species most commonly encountered in the field. The developmental time is much shorter in the tropics with N. eichhorniae taking 80 days to develop from egg to adult in Florida and about 50 days in West Africa.

The moth Niphograpta albiguttalis is believed to have contributed to the successes in Sudan and the USA. Oke et al. (2012) report that this moth did not successfully establish when released in Benin or Ghana, but that without recorded release of the moth in Nigeria the larvae were found damaging water hyacinth in the infested waterways of Badagry, Ejirin and Epe in Lagos State and Iwopin in Ogun State. The larval instars found were damaging only water hyacinth with bulbous petioles. The other organisms listed above have rarely been effective on their own, but the fungi are often observed to increase the damage caused by insects or by the mite Orthogalumna terebrantis; this has been observed in South Africa.

Chemical control (e.g. using 2,4-D) may be necessary as an extreme measure, for the rapid destruction of large masses of weed which are seriously impeding access or navigation. All the larvae of Neochetina spp. and many adults on the sprayed plants are likely to be lost as a result of complete kill of the weed. This should be considered in deciding the areas to be treated, in addition to the possible problems from deoxygenation when the weed is decomposing. Where Neochetina spp. are being introduced, any herbicide treatment should of course be kept well away from the introduction points. Low doses of 2,4-D, which damage but do not kill the weed are believed to encourage insect attack and will thus be beneficial in the longer term (Haag and Habeck, 1991). Other evidence suggesting that herbicides are not necessarily detrimental to Neochetina spp. is provided by, for example, Findlay and Jones (1996) and Center et al. (1999). Herbicides are also known to encourage certain fungi. Hence chemical and biological control are not necessarily incompatible.

Biological control programmes can readily involve local community groups. In Australia, CSIRO has harnessed the resources of the school system via the formation of the Double Helix Science Club as part of a sponsored initiative to promote science in schools. In 1995, this club released the biocontrol agent Neochetina bruchi (Briese and McLaren, 1997).

A new agent, Cornops aquaticum, is being tested for specificity in South Africa (Oberholzer and Hill, 2001). Coetzee et al. (2011) review biological control efforts in South Africa, but suggest that long-term management of alien aquatic plants in South Africa relies on the prevention of new introductions of aquatic plant species that could replace those that have been controlled, and, more importantly, on a reduction in nutrient levels in South Africa's aquatic ecosystems. 

Sacco et al. (2013) evaluate the potential of the planthopper Taosa logula, native to South America, for control of E. crassipes. Tests showed that individual growth and biomass production of water hyacinth was reduced due to the effect of the insect feeding above five nymphs per cage. The number of new plants produced by clonal reproduction was only significantly different above 15 nymphs per cage. These results suggest that this planthopper could be an effective agent for the biological control of E. crasssipes.

Integrated Control

Although it is hoped that biological control will eventually be capable of achieving the necessary level of control of E. crassipes, there is likely to be scope for the integration of physical and chemical methods with biological methods on a local basis, to help speed the achievement of control. The possible approaches include:

- control of nutrient levels.

- use of booms to control movement of the weed.

- exploitation of variable water levels.

- manual removal of the weed from shores and small channels.

- mechanical removal or destruction by land-based or floating equipment.

- use of biological control agents.

- careful use of herbicide to kill or weaken the weed.

- utilization of the weed.

An example of a well integrated control approach (in Mexico) is provided by Gutierrez et al. (1996). In South Africa, biological control with five arthropod species and fungal pathogens attempted since the mid-1970s has had limited success and it has been suggested that additional control agents may be required as well as implementing site-specific integrated management plans (Hill and Cilliers, 1999). Due to the weed's recent rapid increase in the species' abundance and distribution in Africa and elsewhere, international co-operation has been promoted in order to effectively combat the plant (Julien et al., 1996). Lu et al. (2007) suggest that in China the currently dominant biological control-centered view should be broadened to a sustainability science-based management framework that explicitly incorporates principles from landscape ecology and Integrated Pest Management.

Control of Nutrient Levels

The reduction of nutrient pollution of water bodies, wherever it is at all feasible, should be a high-priority approach. Redistribution of excess nutrient, as an alternative to its prevention, should be considered in some situations.


Where infestations occur in relatively narrow rivers, the removal by manual or land-based machinery is often feasible and, although such removal is expensive, the cost may be at least partly offset by utilization (see below). In larger water bodies, the weed should, wherever possible, be pushed to the shore for harvesting by land-based methods, but floating equipment may be appropriate in some situations.


A range of uses for water hyacinth have been proposed and studied (see Uses) none can be regarded as suitable for large-scale use and at the same time provide a satisfactory means of control. However, some of the uses can be exploited on a small scale, especially in conjunction with manual or mechanical harvesting, to recoup some costs and help to make the procedures more economic. Some of these can help to cover some of the costs of control but in almost no case does the usefulness outweigh the economic problems caused by the weed. The possibilities of incorporating utilization into an integrated system of control are reviewed in detail by Gopal (1987).

Each water body should be considered separately; an ideal combination of measures should be devised for each water body, depending on many factors and in close consultation with all users of the water.

Gopal (1987) ends his book with the warning that 'The interests of mankind can only be safeguarded by seeking effective control of water hyacinth and not by its utilization'.

Summary of invasiveness

E. crassipes, a native of South America, is a major freshwater weed in most of the frost-free regions of the world and is generally regarded as the most troublesome aquatic plant (Holm et al., 1997). It has been widely planted as a water ornamental around the world because of its striking flowers. Wherever it has encountered suitable environmental conditions it has spread with phenomenal rapidity to form vast monotypic stands in lakes, rivers and rice paddy fields. Then it adversely affects human activities (fishing, water transport) and biodiversity. It is impossible to eradicate, and often only an integrated management strategy, inclusive of biological control, can provide a long-term solution to this pest.

Related treatment support
Pest Management Decision Guides
CABI; CABI, 2017, English language
External factsheets
BioNET-EAFRINET Invasive Plant Factsheets, BioNET-EAFRINET, 2011, English language
Sweetpotato DiagNotes Fact Sheets, The University of Queensland, English language
Zoomed image