Biological Control of Red Water Fern in South Africa
Report No KV 158/05

Jan 2005


Red water fern, Azolla filiculoides Lamarck (Azollaceae) is one of the five declared aquatic weeds in South Africa. This plant is native to South America and was first recorded in South Africa in 1948. Until the 1980s, the fern was confined to small streams and farm dams in the Colesburg area of the Northern Cape Province. However, a combination of phosphate-rich waters and the lack of natural enemies lead to its inevitable spread throughout the country. Dense mats of the weed (up to 30cm thick) severely degraded aquatic ecosystems and impacted all aspects of their utilization. The failure of mechanical control and the undesirability of herbicide control in the aquatic environment made red water fern an ideal candidate for biological control in South Africa.

The frond-feeding weevil, Stenopelmus rufinasus Gyllenhal (Coleoptera: Curculionidae) was collected from Azolla caroliniana Willd. in Florida, USA, and following host specificity screening it was released on A. filiculoides in South Africa in December 1997. Fears that this insect was a new association on A. filiculoides and therefore might not be all that effective and the fact that is was collected from a tropical area and might not be able to establish in the cooler regions of the country, prompted two courses of action. First, surveys were conducted on A. filiculoides in Argentina in an attempt to more closely match both the host species and climate compatibility. Second, an additional insect, the flea beetle, Pseudolampsis guttata (LeConte) (Coleoptera: Chrysomelidae) was collected on A. caroliniana in Florida and imported for host specificity screening.

However, both of these courses of action proved to be failures as the weevil collected in Argentina was a different species, Stenopelmus brunneus (Hustache), and the quarantine culture was subsequently destroyed without being released. In addition, S. rufinasus had not only established in some of the the coldest areas of South Africa but was having an impressive impact on the weed in these areas (see below).

The flea beetle (P. guttata) was, however, still screened for possible release. Favourable biological characteristics of this species included: long-lived and mobile adults; short immature development times; high rate of increase, and high per capita feeding rates. In the laboratory, this species was a superior agent to the weevil (S. rufinasus). Host specificity screening, however, indicated that the flea beetle is an oligophagous species, capable of utilizing several species in the genus Azolla and could pose a threat to native, southern African species. The flea beetle was therefore rejected as a possible biological control agent for red water fern in South Africa.

The taxonomy of the genus Azolla remains difficult as it relies on differentiation of the reproductive structures. The morphology of the sporophyte is very plastic and can change dramatically depending on day length, temperature and, most importantly, water chemistry. The most reliable characteristic appears to be the structure of the sporocarps. For the purposes of the host specificity screening of the flea beetle, we had material identified by an expert from Portugal. In addition to this, we sent material to an expert in Hong Kong. Opinions of the experts differed in that the one from Portugal identified three taxa of Azolla in southern Africa: Azolla filiculoides (introduced), Azolla pinnata var. africana (from Zambia and Malawi) and Azolla pinnata var. asiatica (also introduced) from a pan in the Bluff Nature Reserve near Durban. The expert from Hong Kong confirmed the identification of Azolla filiculoides and Azolla pinnata var. africana but was uncertain of the identification of Azolla pinnata var. asiatica. Therefore the taxonomy of southern African Azolla species warrants further study.

The second part of this project was to conduct a thorough post-release evaluation of the weevil, Stenopelmus rufinasus on red water fern. The first release was made on a one hectare dam in a bird sanctuary in Pretoria in December 1997. Nine hundred weevils were released on the dam, which was 100% covered by a 5 cm thick mat of the weed. By February 1998 (2 months later) the red water fern mat had collapsed and from a 2m2 sample of decaying material in excess of 30 000 weevils were reared. This was an astonishing phenomenon, which has been successfully repeated throughout the country over the last four years.

To date, the weevils have been released (usually in batches of 100 adults) at some 112 sites throughout South Africa. The current information available on these sites is that the weevil has been responsible for clearing 91 of them completely. For the remaining 21 sites, either the weed has been washed away during flooding, they have not been revisited, or are in the final stages of control. All of the sites have cleared in less than one year. In addition to this, the weevil has migrated to other sites, sometimes up to 300km away from the point of release. It is uncertain if the weevil has been transported on weed by waterfowl, or if there has been short distance dispersal onto other dams with the weed, or if it is as a result of long-range dispersal by the adults. At 7% of the sites the weed has returned up to 2 years after the initial clearance. The weevil has located 90% of these and the weed is again under control.

In order to quantify the impact of the weevil on red water fern, field cage experiments were conduced at five sites in Gauteng Province during summer and winter. At each site, three floating cages (50cm x 50cm x 50cm) were erected and each was inoculated with 1kg of the fern. Two of the cages were gauze covered, while the third only had a small gauze skirt around the base. Ten male and 10 female weevils were introduced to the experimental cage while the other gauze covered cage served as a control for the effect of the weevils. The third cage served as a control for the affect of the gauze on the growth of the fern. Two samples of red water fern were taken from each cage weekly. One of the samples was used to determine the populations of the weevil while the other sample was used to quantify the impact of the weevil on the dry weight of the fern. At all five field sites, total weed clearance was achieved within a period of seven weeks in the summer trial and 14 weeks in the winter trial. The gauze had no significant effect on the growth of the fern which grew normally in both control cages. These cage experiments confirmed the field-release experiments (see above) in that weevil populations are capable of a rapid increase resulting in a dramatic crash of the fern mat.

Although the S. rufinasus that was released in South Africa originated in the tropical climes of southern Florida, USA, it was able to establish and have a major impact on the weed even in the coolest areas of South Africa (southern and eastern Free State). We therefore decided to undertake a series of laboratory trials to investigate the thermal tolerance of this weevil in order to be able to predict if there were any areas of the country where it might not establish. The critical thermal limits, or temperatures at which the insect became immobilized varied between 0C and 5C (lower limit) and between 45C and 48C (upper limit). The lethal temperature at which 50% of the population was killed after prolonged exposure to that temperature was –12.1C (lower limit) and 36.5C (upper limit). The laboratory results confirmed what we had noted in the field, in that although many of the adults would become immobilised during the night in the winter months, and therefore might be exposed to predation and frost, the low winter temperatures would not prevent establishment. Futhermore, the eggs, larvae and pupae were buffered from the air temperatures in that they are endophytic (within the plant tissue). However, their duration of development is likely to be much longer during the winter months. The predictive strengths of two models (CLIMEX and degree-day) were also tested, with both confirming that the establishment of the weevil in South Africa would not be restricted by climate.

Biological control is often cited as the most cost effective weed control option and yet very few cost-benefit analyses have been conducted on weed biological control programmes. The fairly well defined nature of the Azolla filiculoides problem in South Africa coupled with the dramatic success of the weevil afforded us the opportunity to conduct a cost-benefit analysis of the programme to date. Questionnaires were sent to 30 land owners (farmers, municipalities, golf courses) that had been affected by the weed to assess the economic impact of the weed. Costs included loss of livestock, loss of irrigation potential, burning out of pumps, construction of alternative water supply facilities and the cost to control the weed, either mechanically or through the use of herbicides. The cost of the weed to aquatic biodiversity was impossible to calculate, but it was no doubt huge. The cost to develop the biological control programme was then offset against the total cost of the weed at all known sites in South Africa over the last 20 years. The benefit to cost ratio of the biological control programme on red water fern in South Africa was calculated at 2.5:1 for 2000, increasing to 13:1 in 2005 and 15:1 in 2010 as the annual costs of the programme decrease. This indicates an enormous return on investment into this research.

The biological control programme against red water fern is a unique example of biological weed control. Seldom, if ever, has there been a biological control programme that has resulted in such a rapid, and drastic decline in a weed population. Obviously long-term monitoring is required to determine the resurgence in the weed populations and how the weevil is able to locate and reduce these. The initial data suggests that no matter where red water fern appears, the weevil will locate and control it.

The biological control of red water fern in South Africa has been as successful as the programmes on three other aquatic weeds, water lettuce, salvinia and parrot’s feather, which leaves water hyacinth as the final aquatic weed to bring under effective biological control. However, the biggest challenge facing aquatic ecosystems in South Africa remains eutrophication, of which invasion of aquatic weeds is only a symptom. It is highly likely that in many of the systems in which red water fern has been controlled, unless the levels of eutrophication are reduced, other and possibly worse aquatic weeds will take hold.