A preliminary ecological risk assessment of the major weeds on the Magela Creek floodplain, Kakadu National Park Australia

Supervising Scientist Report 194
Walden D, Boyden J, Bayliss P & Ferdinands K
Department of Sustainability, Environment, Water, Population and Communities, 2012
ISSN 1325-1554
978 1 921069 19 2


Print version

About this document

Background and approach

In the late 1990s, the World Heritage Commission expressed concern over possible impacts on environmental and cultural values of Kakadu National Park (KNP) resulting from the proposed Jabiluka uranium mine. A specially formed Independent Science Panel reviewed a Supervising Scientist report to the World Heritage Committee and made 17 principle recommendations. Amongst these recommendations was a need for more comprehensive ecosystem risk assessments at a landscape-catchment scale to differentiate between mining and non-mining impacts, primarily because the region is subject to changes other than those potentially related to mining. Weed invasions are perhaps the most significant non-mining threat to much of the Kakadu landscape including its wetlands.

Background information from existing literature is provided for three highly invasive wetland weed species, Mimosa pigra (mimosa), Salvinia molesta (salvinia) and Urochloa mutica (para grass). The primary focus of this assessment is the current and potential distribution of para grass with reference to its impacts. Field surveys and a remotely sensed QuickBird™ image were used to map para grass on the Magela Creek floodplain. These data were used to determine the current extent of para grass and the average spread rate of the past 20+ years. The data were also used in a Bayesian habitat suitability model to predict the native vegetation communities most susceptible to invasion and the potential extent of para grass on the Magela floodplain. No spatial risk modelling or habitat suitability modelling was conducted for mimosa and salvinia. The Magela floodplain remains under constant threat from mimosa and only a rigorous maintenance control regime prevents re-establishment from residual seed-banks and from new incursions. The spread and impacts of mimosa have been documented for a number of coastal floodplain systems in the Northern Territory. Salvinia, being a floating aquatic fern, is seasonal in its distribution and is subject to an ongoing biological control agent. There are scant data that quantify the impacts of salvinia.

This assessment follows the generic wetland risk assessment model recommended for the Ramsar Convention and comprises six fundamental procedures including: identification of the problem; identification of the effects; identification of the extent of the problem; identification of the risk; risk management and reduction; and monitoring and trend analysis.

Identification of the problem

Mimosa, salvinia, and para grass are very different weeds and, for a variety of reasons, are subject to different management regimes. Since the early 1980s, the threat and consequences of widespread mimosa coverage in the Park were recognised and existing infestations were removed and/or fenced and monitored. A mimosa ‘team’ was also established to prevent new outbreaks through surveillance. Salvinia is already widespread in the Park with the potential for further spread. However, the density and coverage is greatly reduced via an introduced biological control agent, the weevil, Cyrtobagous salviniae. Salvinia will never be eradicated in the Park and is it now considered a permanent component of Kakadu’s flora. Presently, para grass covers a large area of the Magela floodplain. Access to expansive floodplains such as the Magela is often very difficult and the implementation of control and management strategies presents many challenges. 

All three weed species possess many traits that make them highly invasive and successful. Some of these traits include; rapid growth; ready germination or resprouting from seed or small fragments distributed by floodwaters, birds and terrestrial animals; superior competition for resources compared with native species; few or no natural predators; and attributes attractive to humans that facilitate distribution to new areas often over vast distances. Environmental and socio-economic impacts are generally severe for large infestations. Control of all three weeds is usually very expensive and any chemical treatment has the potential to cause damage to non-target flora and in some cases to fauna.

The potential effects

Mimosa and para grass form dense homogeneous shrubland and grassland respectively, greatly reducing flora and fauna biodiversity. With respect to the current para grass infestation on the Magela floodplain, perhaps the greatest concern is for the loss of the wild rice (Oryza spp) grassland which is recognised as an important foraging and nesting resource for magpie geese (Anseranas semipalmata). Salvinia dominates the surface of waterbodies forming dense mats that can alter water quality and nutrient status by effectively excluding light penetration, preventing oxygen transfer from the air and increasing dissolved oxygen consumption. These weeds have the potential to alter hydrological regimes and large infestations impede the access of humans and animals to floodplains and waterbodies, adversely affecting enterprises and activities such as pastoralism, tourism, recreation and traditional Indigenous practices such as hunting and gathering. Para grass can alter fire regimes with more intense and frequent fires that sometimes carry into more sensitive fringing forests. These weeds are also known to provide ideal habitats for pests such as pigs and mosquitoes.

The potential extent

Since 1980, mimosa has been found at a number of sites on the Magela floodplain. Adult mimosa plants at these sites have been removed although seedlings continue to appear at some sites. New incursions are often located each year and these are destroyed and the site closely monitored. This scenario for mimosa occurs throughout the Park. If allowed to establish and spread, mimosa has the potential to dominate a major proportion of Kakadu’s floodplains. A map of the history of mimosa incursions since 1980 is presented for the Magela floodplain. However, no spatial analyses, risk modelling or habitat suitability modelling have been undertaken and no such maps are presented.

Salvinia was first discovered in the Magela system in 1983 and is now present in the East Alligator River system and Nourlangie Creek and elsewhere in the South Alligator River system. Salvinia’s rapid spread through Kakadu indicates that it could occupy all wetlands and waterbodies in a relatively short space of time. At the time of the QuickBird image capture (July 2004), para grass covered approximately 1250 ha of the central region of the Magela floodplain. Satellite infestations with the potential to spread rapidly exist in the north and south regions of the floodplain and along the north-western edge. Bayesian habitat suitability modelling has shown that para grass has the potential to invade about 6360 ha (34%) of the floodplain, displacing the Oryza spp grassland in particular.

Identification of the risks

Risk is a combination of both the effects and extent (or exposure to) of a threatening entity. For mimosa, experience from coastal floodplains elsewhere in the Northern Territory has shown that the majority of land subject to inundation, except for perhaps the very deepest parts of the channels, could potentially be invaded by mimosa. The probability of incursion and establishment remains very high for KNP and continued vigilance is necessary to prevent repeated incursion or emergence from dormancy in existing seed banks. Potential weed vectors such as the high volume of park visitors (eg vehicular and boat traffic), and movements of feral and native animals exacerbate the risks.

Although the risks associated with salvinia infestation have been reduced by the salvinia weevil, research has shown that weevil populations can vary greatly from year to year with a subsequent variation in their efficacy. In the absence of any quantitative data on the effects of salvinia, the actual risks remain uncertain. The risks of salvinia invading other regions in the Park are managed in part by quarantine procedures.

Para grass currently infests large areas of the Magela floodplain and has therefore already displaced large areas of natural habitat. It is known to be spreading at an alarming rate. Habitat suitability modelling has demonstrated that a large proportion of the floodplain, yet to be colonised, is at high risk of infestation which could occur within two to three decades.

The adverse effects of widespread incursion of all three weeds have been acknowledged, although these effects are generally poorly understood with little quantitative information being available.

Uncertainty, information gaps and further research

As the risk of mimosa invasion in Kakadu National Park remains high, a number of research endeavours that may assist managers have been identified. Some of these include: habitat suitability modelling; more detailed GIS assimilation of mimosa plots and new incursions with the possibility of developing remote sensing techniques to identify new outbreaks; documenting the seed stores following control; quantifying the competitive relationship between mimosa and native vegetation; and the precise role of wild and controlled fires.

For salvinia, there is very little quantitative data on the effects on water quality and flora and fauna. Remote sensing techniques could be explored to monitor salvinia distribution at the landscape scale. Continued monitoring of the persistence and success of the salvinia weevil is essential for the control of salvinia.

Some identified areas where further information could assist in the management of para grass include:
Extent: further refinement of the habitat suitability model; more detailed mapping; salinity tolerance; the viability of para grass seeds and the persistence of the seed bank; competition with native species; modification of the floodplain habitat by para grass; tolerance of para grass to water depth and period of inundation; the efficacy of various herbicide treatments utilising trial plots whilst obtaining valuable cost of control data; and the role of fire in management.

Effects: the real impacts upon native habitat displacement; fire (ie increased fuel loads); hydrology (eg reduced flow and increased sediment deposition); extent to which para grass is a physical barrier to larger animals; the effects of herbicides on non-target species.

Management Implications

Ongoing preventative management of mimosa in Kakadu is a labour intensive and costly exercise. Resources are dedicated to i) revisiting old sites where mimosa has been present, in order to monitor and control regrowth from seeds in the seed bank and ii) surveys of suitable habitat to locate new incursions that occur on a regular basis. In the absence of current management strategies, the implication for the Magela and other floodplains of Kakadu is a landscape dominated by mimosa. Removal of mimosa at this scale would cost tens of millions of dollars with considerable and most likely indefinite follow-up control.

Prior to the establishment of the salvinia weevil, considerable resources were allocated to physical and chemical control of the weed, resulting only in short-term gains given the exponential regrowth. After establishment, a decline in weevil populations prompted an intensive study to determine the cause, which appeared to be related to the timing of the onset of the wet season rains and the timing and size of the main flood events. Weevils are sometimes harvested from the wild and distributed to new areas where appropriate.

It is acknowledged that eradication of the larger infestations of para grass is most likely unachievable. Some smaller infestations in the Park have been successfully eradicated and others are currently being targeted for control and/or eradication. Treatment is relatively expensive in floodplain environments as aerial herbicide spraying is often the only feasible method. This is complicated by factors of herbicide drift to non-target species and the uncertain effects on frogs and other fauna. Para grass also grows under paperbark and other fringing forests where aerial treatment is not practical. There has been varying success with past control efforts where some treatments are highly successful with little regrowth whilst other treatments have resulted in substantial regrowth even after multiple treatments over many years. The reasons for this variation in efficacy are not well understood. Initial cost of control modelling suggests that significant reduction in coverage could cost nearly $2 million with nearly $0.5 million per annum in follow-up control.