Impact Evaluation of Weed Biological Control Agents

This document was originally published on the website of the CRC for Australian Weed Management, which was wound up in 2008. To preserve the technical information it contains, the department is republishing this document. Due to limitations in the CRC’s production process, however, its content may not be accessible for all users. Please contact the department’s Weed Management Unit if you require more assistance. #1

Impact evaluation of weed biological control agents

What is impact evaluation?

Impact evaluation of biological control agents is performed either before or after their release in the introduced range of the target weed.

Pre-release impact evaluation is carried out in laboratories and glasshouses or in the field in the weed’s native range Best practice guide to predict impact of candidate agents on individual plants or populations and to assist in selection of the most promising agents.

Post-release impact evaluation measures the effectiveness of agents Post-release impact evaluation is crucial to understanding the overall success rate of a biological in reducing target weed populations • control program. Here the effects of grazing, herbicides and biological control agents (Longitarsus Impact evaluation of weed biological control agents in the introduced range and quantifies echii and Mogulones larvatus) are being monitored on Paterson’s curse. the benefits for associated plant Photo: CSIRO communities, ecosystems, the economy and society in general. program, ultimate success can only and improve efficiency and effectiveness In some instances, the direct effects be demonstrated by undertaking post- of future programs for other weeds. of released agents may be monitored release impact evaluation. Claiming For example, the data gathered may on non-target plant species identified success of a biological control program inform practitioners that other agents as possible hosts during host-specificity without quantitative assessment is affecting different life stages of the tests prior to release: but this is not the questionable because other factors target weed will need to be introduced focus of this guide. such as variation in climate, land use to exert additional stresses on weed and / or management practices may populations or that change in other Why perform post-release have contributed to a weed management practices may increase impact evaluation? population’s decline. agent impact. Rigorous assessment of programs also provides data to Measuring impacts of agents on demonstrate the effectiveness of While confirming establishment, target weed populations can provide biological control and hence justify prevalence and abundance of released the necessary information to fine-tune continued investment in research agents over space and time is a crucial a biological control program’s strategy milestone for any biological control and implementation. Who should undertake necessary data for subsequent analyses. against which effectiveness can be post-release impact Post-graduate student projects are also later evaluated. These targets estimate evaluation? a possible avenue for more theoretical the required decline in weed density intensive studies to be undertaken, (or other population level parameters) The research group that initiates the such as developing an understanding over a defined timeframe to reduce biological control program often takes of the indirect impacts of agents on impact of weed populations below responsibility for post-release impact the ecosystem. Community groups and an economic or ecological threshold. evaluation of agents, but rarely gets state and local government officers For example, a 25% reduction in the continuation of funds necessary may also become involved, particularly weed cover within 10 years may be to undertake long-term evaluation. when monitoring protocols and considered as necessary to obtain Sufficient resources and funding should sampling regimes are simple and not a satisfactory increase of desirable be negotiated and put aside at the too time-consuming. A program can vegetation at infested sites. More start of programs to support at least greatly benefit from such involvement generally, determining the relationship a minimal level of impact evaluation because it increases the number of between a quantitative measure of activities soon after an agent is released. sites monitored at regional scales. the weed population (including per Additional funding should be accessible capita assessment) and the economic later to support long-term monitoring. Prerequisite benchmarks or ecological impacts may be more Multi-disciplinary teams comprising for evaluation helpful because it allows an observed ecologists, entomologists / plant level of decline to be equated to a pathologists, economists and modellers Performance targets defined at the degree of impact and hence benefits. are most effective in devising appropriate beginning of biological control programs Devising such relationships however, monitoring programs to gather the provide quantitative benchmarks requires knowledge of the economic and / or ecological impacts of the weed and what level of control is necessary to ameliorate these impacts. Partnerships between economists, plant ecologists and plant population modellers are recommended to document impacts of the target weed in order to devise realistic performance targets for a program. Such a priori explicit criteria against which agents or programs can later be judged are still rarely made, to the detriment of biological control as a discipline.

Economic impact evaluation

Cost-benefit analyses are fundamental to evaluating the economic impacts of biological control programs quantitatively. The output of such analyses, expressed as a benefit-cost Impact evaluation of weed biological control agents agents control weed biological of evaluation Impact ratio or as a net present value, is a • measure of the return on investment from the program. A benefit-cost ratio is derived by dividing the value of the losses avoided by affected industries or other stakeholders (in other words what economic damage would have Impact evaluation of the bridal creeper rust fungus (Puccinia myrsiphylli) and responses of associated vegetation. been caused by the weed in the Best practice guide practice Best Photo: Peter Turner absence of the project, including 2 costs for control), by the research and » Define performance targets of development costs of the biological biological control program control program. A net present value » » » » » is an indicator of how much cash value Select appropriate approach to a project adds to the value of the monitor ecological impacts of agents industry concerned. Both costs and benefits are discounted at a specified Determine number of sites and years rate to account for differences in the to monitor Use experience to improve impact time when they were incurred; costs evaluation of current and future biological control programs are up-front while benefits increase Determine parameters (variables) steadily as agents gradually impact to monitor on the weed.

Cost-benefit analyses are more robust Implement monitoring schedule when carried out after there has been ample time for the field impact of Analyse and report (publish) results » agents to be realised and sufficient long-term ecological impact data to Figure 1: Key steps for developing a monitoring program to evaluate post-release ecological impacts of biological control agents. be assembled. They are also more Adapted from Roni et al 2005 convincing when based on accurate data on research and development each has pros and cons (Table 1). To be changes in density because they can costs of programs, economic and social most effective, post-release ecological be analysed with high precision using costs of the target weed and its rate impact evaluation should be planned image analysis software. Decrease in of spread and distribution, before and well before any agents are released weed density may not be easily detected after releases of agents. to ensure that crucial pre-release in photographs if other vegetation is Unpriced values for benefits such as measurements are gathered. This replacing the controlled weed and the preservation of native biodiversity includes determining the design, therefore additional information on the Best practice guide and scenic amenity, improved scale and duration of the monitoring type of plants present at photopoints recreational access to land and water, program and the types and timing has to be collated. Photographs also reduction of fire hazard, decreased of data collection (Fig. 1). do not provide evidence that the exposure of environment to herbicides, decline in weed populations is due and costs such as ‘collateral damage’ Approaches to the agent(s), unless it is coupled on non-target plants (if applicable), with an assessment of agent damage are not included in the analysis unless Before- and after-release and density. • Impact evaluation of weed biological control agents a definitive dollar value has been assessments Comparing historical and estimated. Nevertheless, unpriced Photopoints: The simplest and contemporary data: Published and environmental and social benefits and cheapest approach to evaluate impact unpublished historical information, costs should be captured and described of biological control agents is to establish such as field ecological data and aerial to provide a broader perspective to the fixed reference points at weed-invaded or satellite photos of specific weed- economic analysis. sites before or soon after the release infested sites, collected regardless of of agents and to take a series of the initiation of a biological control photographs from those at regular Ecological impact program, can provide baseline data to intervals after release (see the evaluation compare with contemporary data taken parthenium case study). This approach at the same sites. Aerial and satellite Several different approaches, varying in is suitable when impact of the agent(s) imagery can be particularly powerful their level of complexity and amount of is visually conspicuous and a dramatic by providing a long-data series prior, resources required for implementation, decline in weed density and cover is during and after release of biological have been used to evaluate short- and observed. It can be difficult however, control agents over a large, spatial long-term ecological impacts of biological to measure modest changes in weed scale. The use of historical data for control agents. Some approaches are density from such a photo series. Where comparison is most useful when the more effective or suitable than others appropriate, photographs of vertical agent is widespread and if data of under certain circumstances (eg views of fixed areas (eg quadrats) are equal quality exist for several sites. terrestrial versus aquatic systems), and more suitable for measuring slight 3 Table 1: Summary of advantages and disadvantages of various approaches to evaluate post-release ecological impacts of biological control agents.

Approach Description Advantages Disadvantages Photopoints Comparison of Simple and cheap Cannot definitively attribute weed photographs taken at fixed Suitable when dramatic decline in decline to agent reference points at weed- weed populations is observed Difficult to measure modest changes invaded sites in weed density May be misleading if other vegetation replaces controlled weed Comparing Comparison of historical Relatively cheap to implement Cannot definitively attribute weed nts

e historical and data on weed with similar because it takes advantage of decline to agent contemporary contemporary previous work May be difficult to locate quality pre data measurements Suitable if agent is already release records assessm widespread and pre-release data e exist for several sites Stakeholder Comparative surveys of Relatively cheap to implement Cannot definitively attribute weed releas - surveys land managers’ perceptions Useful for large-scale evaluation decline to agent of prevalence of the target Provides only qualitative (and possibly weed biased) assessments of program outcomes Comparing Comparison of pre- and Persuasive if a long-term, sustained Requires more resources than above quantitative post-release measurements reduction in the weed correlates with methods. data on weed and agent(s) presence of agent Difficult to separate effects of agent from Before- and after taken across multiple sites Can improve study by including other confounding factors (unless control and years control plots (with no agent) at plots included) monitored sites Needs to be carried out over long periods Suitable to measure changes in if high variability in weed parameters is associated ecosystem observed between years Comparing sites Comparison of sites or plots Suitable if agent has already been Difficult to separate effects of agent from or plots with and where agents are present released, but is not widely other confounding factors that affect without agents with others where they are established weed growth at different sites / plots not (no pre-release data Suitable to measure changes in Needs to be carried out across several collected) associated ecosystem if agent-free similar sites and over multiple years sites are retained over long periods Not suitable if agent disperses rapidly. Correlative studies Correlation between agent Suitable if agent has already been Difficult to separate effects of agent from density or damage severity released and is widely established other confounding factors that affect and weed performance Suitable for assessment at individual weed growth at different sites plant level Needs to be carried out across several similar sites and over multiple years Agent exclusion Comparison of plots where Allows assessment of relationship Resource- and time-consuming to experiments agent is excluded, using between agent densities or damage establish and maintain cages or pesticide severity and weed performance, Generally carried out on a small scale applications, with others without external confounding factors and over the short term where they are present Unintended procedural effects can Exclusion treatments can alter the system be managed by measuring effects of investigated (eg insecticide can exclude the exclusion treatments without pollinators and affect weed reproduction) agent present and use this to Difficult for medium to large woody weeds interpret experimental results Demographic Models to demonstrate Useful to exclude other explanations Requires specialised expertise to develop modelling mechanistic connection of population decline of the weed models between agent impact and over time Data needed to parameterise models weed densities Useful to predict future dynamics of may not be available weed populations under pressure Inevitable trade-off between simplicity of from agent and other control tactics model and how well it fits the ‘real’ data Useful when no or limited pre Impact evaluation of weed biological control agents agents control weed biological of evaluation Impact release data are available • Useful to assist planning of timeframe of monitoring activities Best practice guide practice Best

4 quantitative studies to provide hard data to support findings.

Comparing quantitative data: Quantitative data collected on selected populations of the target weed, associated plant communities and ecosystem processes in the immediate years before the release of biological control agent(s) can be compared with similar data gathered after their release (see the mimosa and bridal creeper case studies). Such studies have to be carried out at multiple sites across the weed distribution and over several years to capture the often high temporal variations in population levels and to 2005 ascertain that observed changes are primarily due to the agent(s). The inclusion of control plots at monitored sites, where agents are not released or are excluded (see below), can also provide additional information to identify possible confounding effects. These control plots however, may be short-lived if the agents rapidly colonise

them, or exclusions cannot be adequately Best practice guide maintained over long periods.

Comparing sites or plots with and without agents In situations where an agent has already been released but has not widely

established or spread, it is possible to •

compare weed population parameters Impact evaluation of weed biological control agents 2006 at areas where the agent is present with Photopoints at an infestation of bridal creeper at Gairdner River in Western Australia. Top photograph was taken in September 2005 less than 2 months after the rust fungus (Puccinia others where it is not (see the mimosa myrsiphylli) was released, while the photograph on the bottom was taken in September 2006, and parthenium case studies). Results 15 months after the release. Photos: Peter Turner of such comparison studies however, may be influenced by differences in site or plot characteristics such as unsuitable As for photos taken at fixed reference performed years after the release of abiotic or biotic conditions that prevent points, this approach cannot attribute, agent(s). Such surveys are a relatively agents establishing and / or affect plant with certainty, the decline in weed inexpensive way to obtain a qualitative growth (eg competition with other populations to the agent(s). assessment of the outcomes of a vegetation). These studies should be program, but may be deceptive because Stakeholder surveys: Perceptions of conducted across several similar sites stakeholders’ perceptions can be land managers on the prevalence of and over multiple years to account for influenced by whether or not they are a target weed and the economic and / these possible confounding effects. It supporters of biological control or or social problems that it causes can may not be possible however, to perform perceptions may change independent be surveyed before the initiation of comparative studies for many years if of the biological control program itself. a biological control program. They an agent population builds-up rapidly Therefore surveys should be coupled can then be compared with answers and individuals disperse to control sites. with more detailed site-specific obtained in subsequent surveys

5 Correlative studies plant size, age and genotype. The main Demographic modelling Correlative studies are effective to relate advantage of such a pot or planting Models are useful to demonstrate a agent densities and rate of attack to method is the ability to retrieve entire mechanistic connection between agent changes in growth and reproduction individual plants at the end of the impact and reduced weed densities parameters of individual weeds. They experiment, making it easier to examine in order to exclude other explanations can demonstrate the overall impact on agent impact on below-ground biomass of population decline of the weed over the weed population and associated (see the bridal creeper case study). time. With data on weed growth rates vegetation if permanent sites are Procedural difficulties however, are and fecundity gathered before and monitored over several years. They can often associated with these types of shortly after agent release, a population also be used to clarify contributions experiments. For example, physical model can be developed to predict the from multiple agents. Changes in plant barriers, such as nylon mesh used future dynamics of weed populations, performance recorded in correlative in cages, can reduce light levels and including rate of spread, under pressure studies may not be due solely to the evapo-transpiration rates, which in from the agent (see the mimosa case effect of the agents but also to other turn may alter plant performance. study). Comparison of model predictions external factors, such as moisture Cage and insecticide exclusion with long-term post-release data on availability and microhabitat experiments can also affect plant plant demography will evaluate whether characteristics. The density and levels performance by excluding pollinators the agents are affecting the population of attack of an agent may also be leading to reduced seed production dynamics of the weed and to what altered by the nutritional quality, size in plants within treated plots and extent agent damage is having an and vigour of the target weed. To producing unrealistic results (see the impact on population densities. overcome some of these problems, mimosa case study). The exclusion Modelling based on results from correlative studies should be carried out treatment may also influence the exclusion experiments or correlative over several sites and years and ideally system by affecting populations of studies can also facilitate evaluation concurrently with exclusion experiments endemic pathogens, predators and of biological control programs for (see below), to irrevocably demonstrate parasitoids. It is recommended to use which no or limited prior data are cause and effect and account for a pesticide with a restricted mode available, even years after they were possible confounding factors. of action that is efficient against the initiated. Such an approach provides specific agent to overcome non-target Agent exclusion experiments some form of quantitative evidence impacts on associated microbial and to support claims that a program has Exclusion experiments using cages or insect communities. The pesticide used been successful or not. achieved through regular applications to exclude the target agent may have of insecticides or fungicides are unwanted fertiliser or phytotoxic Quantitative data collection performed to relate agent densities effects that affect plant growth. Such or severity of damage directly to unintended procedural effects do not The types of quantitative data and time plant performance, independent of present an insurmountable problem of collection depend on the performance confounding external factors (see the if they can be quantified. Experiments targets set for the biological control parthenium and bridal creeper case that compare the performance of program, the approach selected for studies). These types of experiments plants without the agents in caged or impact evaluation and the key life history can also be used to isolate the effect chemically-treated plots versus uncaged stages of the target weeds and agents of single agents (eg pathogen versus or untreated plots, can be used to (see the parthenium and bridal creeper arthropod). They are generally more estimate the effect of applying the case studies). For example, it is necessary suitable for annual weeds, whose exclusion treatment, which can be to measure changes in seed production population density levels often taken into account when interpreting to assess impacts of a seed-feeding fluctuate widely between years, than agent: although it is also important to

Impact evaluation of weed biological control agents agents control weed biological of evaluation Impact results of exclusion experiments.

• for medium to large woody weeds verify if this parameter alters the size Exclusion experiments are very time- (see the mimosa case study). They of the seedbank and recruitment of consuming and costly to carry out and may sometimes involve the transfer seedlings. Timing of data collection therefore are generally conducted on of potted plants of the target weed may also be influenced by logistic a small scale for only a short period or transplantation of standardised considerations, such as limited access of time. These experiments may also plants to the field to enable detailed to sites during floods in wet seasons be unsuitable to measure changes examination of responses to agents (see the mimosa case study). in associated vegetation due to their Best practice guide practice Best under natural conditions, while inherent short-term nature. Ideally, agent ecological impacts should controlling for differences in soil and be measured at the plant, population 6 Table 2: Possible measurements and most suitable approaches to evaluate post-release ecological impact of biological control agents at different levels.

Level What to measure How to measure Plant – growth (number, size and biomass of above- and below- – before- and after-release assessments ground parts) – agent exclusion experiments – reproduction (number and biomass of flowers, fruits – correlative studies and seeds) – survival – agent density and damage levels Population – photo references – before- and after-release assessments (including – weed density or cover stakeholder surveys and comparison with historical data) – age structure – agent exclusion experiments – seedling recruitment and survival – comparing sites / plots with and without agents – viable seedbank density – weed stand size and spread – agent density and damage levels Ecosystem – species richness and abundance of desirable plant species During: – biomass productivity (eg pasture carrying capacity, – before- and after-release assessments crop yield) – agent exclusion experiments (but ensure procedures – species richness and abundance of undesirable plant do not confound any ecosystem level comparisons) species (weeds) – comparision of sites / plots with and without agents – species richness and abundance of beneficial arthropods and microbes – ecosystem characteristics (eg soil nutrient levels, plant community structure, water quality) and processes (eg rate of nutrient cycling and decomposition)

and ecosystem levels (Table 2), but to be taken, and enable subsequent fast populations build up to damaging Best practice guide availability of resources may considerably evaluation to take place across many levels, how long weed populations limit the extent of data collection. more sites. For example, inflorescence and the associated ecosystem take Practitioners are encouraged to measure length and tap root diameter of to respond and the overall variability agent abundance and responses of Paterson’s curse (Echium plantagineum) of the system. Continued, regular individual plants, and wherever possible were found to give a good estimate monitoring (in some cases for up to to also monitor changes in weed of seed production, which is labour 30 years) is often necessary to account populations, by measuring parameters intensive and time consuming for confounding factors such as spatial •

such as seed production, viable to measure. and temporal climatic variability. For Impact evaluation of weed biological control agents seedbank, seedling recruitment, plant example, populations of a weed (such Recording a range of abiotic (eg rainfall, density, cover and stand size. Monitoring as an annual) and agents that fluctuate temperature and soil nutrients) and of associated plant communities (species widely between years need to be biotic parameters (eg presence / richness and abundance) and other monitored more regularly to capture abundance of predators, parasitoids ecosystem characteristics (eg rate of this variability. Terminating impact or competitors of the agents) may litter decomposition, soil nutrient levels, monitoring before agents have had be useful for later interpreting and water quality in aquatic systems), is also a chance to achieve their full potential explaining variability in results. recommended to assess whether the provides misleading results and Monitoring several representative ultimate goals of the program have understates the success of biological sites across the weed distribution to been achieved. control programs. examine the effect of the agents across Resource constraints may also restrict different climatic regions and habitats Lack of resources is often a key factor the number of sites that can be included may also assist to identify limitations that prevents annual monitoring from in the impact evaluation phase. on agent impact (see the bridal creeper being carried out for long periods of Establishment of relationships between case study). time. In such instances, sites can be key plant performance parameters and re-assessed at longer intervals (ideally the impact that agents have on these Long-term monitoring every 2–3 years and up to 10 year relationships at a few representative intervals) to capture long-term impacts The duration of long-term impact sites can help streamline the number of agents by comparing with data monitoring of agents depends on how and type of measurements that need collected before or soon after releases. 7 8 Best practice guide • Impact evaluation of weed biological control agents weed invasion. to avoidatreadmill ofcontinuous alert landmanagerstotheseissues control programs, itis important to this isbeyondthescopeofbiological disturbance, highnutrientlevels).While continued weedinvasion(eg or tosomeotherfactorsthatfacilitate loss (ordepletion)oftheirseedbank at controlled sitesmaybeduetothe of naturalrecovery ofdesirablespecies other non-desirableplantspecies.Lack controlled targetweedisreplaced with activities, are necessaryifa biologically- various managementandrevegetation Follow-up interventions,suchas Site rehabilitation case study). the targetweed(seemimosa integrated managementstrategyfor term impactofagentswithinabroader also assistindeterminingthelong- and non-biologicalcontrol tacticscan dynamics, disturbanceandbiological processes suchasweedpopulation Models thatincorporatemultiple a successfulcombinationoftactics. strategies isoneapproach to identify control agentswithothercontrol control. Empiricaltestingof biological regimes mayincrease overalllevelsof integration withothermanagement populations ofthetargetweed,their partially successfulatreducing indicates thattheyhaveonlybeen evaluation ofbiologicalcontrol agents In caseswhere post-release impact management Integrated weed Other considerations monitoring activities. planning thetimeframeoflong-term parameters ofaweed,andassistin may taketoaffect keygrowth used toestimatehowlonganagent Demographic modellingcanalsobe Entomology, Canberra,ACT. Australian Weed Management,CSIRO Adele ReidandLouiseMorin,CRCfor Compiled by crops inT applications. andpractical of weeds:theory and White,G.(eds)Biologicalcontrol control agentsonweeds.In: Julien,M. Measuring theimpactofbiological Farrell, G.andLonsdale,M. (1997). Journal ofAppliedEcology41:547–560. invasive shrubintropical Australia. integrated weedmanagementofan and Lonsdale,M.(2004).Modelling Buckley, Y.M., Rees,M.,Paynter, Q. Further reading for publication. for editingandpreparing the guide Weeds CRCisgratefullyacknowledged study. Thecommunication teamofthe his commentsonthemimosacase Landcare Research, NewZealandfor grateful toDrQuentinPaynterof on adraftofthisguide.We are also CRC staff andstudentswho commented and toallcurrent andprevious Weeds participants forstimulatingdiscussions Thanks are extendedtoworkshop information inthisguidewasdiscussed. March 2006,where someofthe evaluation ofbiocontrol’ inBrisbane, organising theworkshopon‘Impact acknowledged fortheirassistancein Primary IndustriesandFisheries)are Day (QueenslandDepartmentof of PrimaryIndustries)andMichael Royce Holtkamp(NSWDepartment Acknowledgements Management, Adelaide. No. 13,CRCforAustralianWeed and biologicalcontrol.Technical Series (2007). Putten, E.I.andMcFadyen,R.E.C. Holloway, R.J.,Chatterton,W.S., Van Ireson, J.E.,Davies,J.T., Friend,D.A., Canberra, pp.105–118. AgricultureInternational Research, W asmania: economicimpacts eeds ofpasturesandfield Australian Centr e for 100: South AfricanJournalofScience plants: casestudiesfrom SouthAfrica. (eds) (2003).Improvingtheselection, Spaffor pp 13–34. Fisheries Society, Bethesda,Maryland, watershed restoration. Methods formonitoringstreamand for aquaticrestoration. In:Roni,P. (ed.) a monitoringandevaluationprogram Steel, E.A.(2005).Stepsfordesigning Roni, P., Liermann,M.,Jordan, C.E.and Australian Weed Management,Adelaide. Technical SeriesNo.10,CRCfor Australian weedbiologicalcontrol. Economic impactassessmentof Page, A.R.andLacey, K.L.(2006). biological contr M.B. (2004).Costsandbenefitsof I.M., Ndala,S.,Brown, B.andRapholo, Anderson, H.J.,LeMaitr van W 45: 253–258. Australian JournalofEntomology A scientificappr van Klinken,R.D.andRaghu,S.(2006). 22: 447–453. T way ofcontrolling invasive plants? Are exoticnaturalenemies aneffective Thomas, M.B.andReid,A.M.(2007). pp. 189–230. Academic publishers,Amsterdam, control: measuresofsuccess. Gurr of terr (2000). Successinbiologicalcontr Syrett, P., Briese,D.T. andHoffmann, J.H. Management, Adelaide. Series No.7,CRCforAustralianW biological controlagents.Technical testing andevaluationofweed rends inEcologyandEvolution , G.andWratton, S.(eds) 113–122. estrial weedsbyarthr ilgen, B.W., M.P., deWit, d Jacob,H.andBriese,D.T. ol ofinvasivealien oach toagentselection. American e, D.C.,Kotze, opods. In: Biological Kluwer ol eed ...case study Mimosa: a leguminous shrub

Mimosa (Mimosa pigra) is an invasive The post-release impact evaluation prickly shrub that grows up to 6 m of mimosa agents has so far primarily high and forms impenetrable thickets involved the use of four different over vast stretches of the Northern approaches. Territory in tropical Australia. It grows all year round and produces pods from 1. Before- and after-release February to April, which release seeds assessments – comparing quantitative data until June. Its dense stands threaten Challenges faced to monitor impact of mimosa native biodiversity in Kakadu National biological control agents at some sites in the From 1984 to 1986, before most Northern Territory include difficulty in access, Park and other wetland areas and also pictured here, and many others such as agents were released, data on leaf fall interfere with human land use by crocodiles, mosquitoes, leaches and humidity. and seed and pod production were Photo: CSIRO affecting pastures, livestock and access collected at one site along transects set to water. up from the edge to the centre of a evaluate impact of established agents. dense mimosa stand. Similar data were A total of 13 insects and two fungi Firstly, limited pre-release data were collected again at the same site from have been released on mimosa in collected before agents were released. 2001 to 2003, years after populations Australia, of which at least seven have Most agents rapidly dispersed of released agents had built-up. The established. Of these, six agents cause throughout the weed’s range after density of the various agents and obvious damage to mimosa: the their release, making comparison of damage levels were also recorded. seed-feeding bruchid (Acanthoscelides sites with and without agents difficult puniceus) released in 1983, the stem- From 2001 to 2003, leaf litter fall was (however see the exception below). mining moths (Neurostrota gunniella 20% higher and seed rain 47% lower Regular monitoring is difficult to and Carmenta mimosa) both released than data collected in the mid 1980s. undertake in the wet season due to in 1989, the flower-feeding weevil From these data, the flower-feeding flooding, which restricts access to field

(Coelocephalapion pigrae) released weevil C. pigrae was estimated to Best practice guide sites. The culling of feral Asiatic water in 1994, the leaf-eating beetle destroy only 10% of flowers. Seed rain buffalos in the late 1980s to early (Malacorhinus irregularis) released was found to have declined by ~60% 1990s, may have contributed to reduced in 2000 and the defoliating moth where the highest densities of the mimosa invasiveness, potentially (Macaria pallidata) released in 2002. stem-mining moths N. gunniella and confounding the impact of agents. C. mimosa occurred. The density of Lastly, mimosa is a long-lived woody Post-release impact evaluation agents and reductions in seed rain shrub, which means impact of agents

were highest at the edge of the mimosa • Several factors inherent to the biological at the weed population level may not stand. This study however, could not Impact evaluation of weed biological control agents control of mimosa make it difficult to be detectable for decades. attribute with certainty the decline in weed reproductive outputs to the agents as the removal of buffalos may have contributed to the suppression of mimosa recruitment at this site and hence a decline in seed production over time.

A stem-mining (Carmenta mimosa) larva eating Mimosa infestation in the Northern Territory. into a mimosa stem. Photo: CSIRO Photo: CSIRO

9 case study Another survey conducted at this site however, were only surveyed annually, Further reading in 2005, after the release and which may be insufficient to detect Buckley, Y., Rees, M., Paynter, Q. and establishment of the defoliating moth an impact from these other agents. Lonsdale, M. (2004). Modelling M. pallidata revealed that mimosa Admittedly, because sites with and integrated weed management of an seedbanks have continued to decline without C. mimosa were not randomly invasive shrub in tropical Australia, and are now approximately 10% of chosen, results may have been Mimosa pigra. Journal of Applied the levels before the release of biological influenced by differences in site Ecology 41: 547–560. control agents. characteristics that affect agent-plant interactions. Heard, T.A. and Paynter, Q. (in press). 2. Comparing sites with and Biological control and integrated without agents 3. Agent exclusion experiments management of Mimosa pigra. In: Weed Litter and seed fall, seedbanks, Although attempted, insecticide biological control with arthropods in the tropics – towards sustainability. vegetation cover, size, density and age exclusion experiments were not Cambridge University Press. structure of mimosa stands at nine successful to assess impact of insect sites where the stem-mining moth agents on mimosa. In early trials, an Lonsdale, W.M. (1988). Litterfall in an C. mimosa was present were compared even application of insecticide onto fully Australian population of Mimosa pigra, with similar data taken at eight other grown shrubs was difficult to achieve an invasive tropical shrub. Journal of sites where the agent was absent. This due to their size and consequently Tropical Ecology 4: 381–392. evaluation method was appropriate for shrubs were not maintained totally free Lonsdale, W.M. and Farrell, G. (1998). this agent because it is a slow disperser, of agents. The use of insecticide was Testing the effects on Mimosa pigra of allowing sites free of the agent to be also found to disrupt pollinator insects a biological control agent Neurostrota easily located. None of the sites were and hence mimosa reproductive gunniella (Lepidoptera: Gracillaridae), free of the other agents, which dispersed outputs. Another insecticide exclusion plant competition and fungi under field rapidly after release, and consequently experiment was later performed using conditions. Biocontrol Science and densities and damages caused by all mimosa seedlings transplanted to the Technology 8: 485–500. agents had to be surveyed and correlated field, but unfortunately, several months Paynter, Q. (2004). Evaluating Mimosa to plant performance across all sites. after initiating the experiment, the pigra biological control in Australia. systemic insecticide used was found not Mimosa seed rain was reduced by In: Julien, M., Flanagan, G., Heard, T., to be effective in excluding C. mimosa. more than 90% at sites colonised by Hennecke, B., Paynter, Q. and Wilson, It is worth emphasising that insecticide C. mimosa where the highest damage C. (eds) Research and management exclusion experiments, using newer, severity caused by the agent was of Mimosa pigra. CSIRO Entomology, more specific insecticides and different recorded. Seedbanks were reduced Canberra, pp. 141–148. application methods may be found to at C. mimosa-infested sites compared be appropriate to assess agent impact Paynter, Q. (2005). Evaluating the with sites where the agent was absent. on mimosa. impact of the biological control agent The cover of other vegetation also Carmenta mimosa on the woody increased under stands where C. mimosa 4. Demographic modelling wetland weed Mimosa pigra in was present, inhibiting mimosa seedling Models based on the two most Australia. Journal of Applied Ecology establishment and also apparently effective agents, the stem-mining moths 42: 1054–1062. increasing the susceptibility of mimosa N. gunniella and C. mimosa predicted to fire, by boosting fuel loads beneath Paynter, Q. (2006). Evaluating the that it would take from 12–29 years stands. It was concluded that C. mimosa impact of biological control against to reduce the cover of mimosa from has the potential to reduce dramatically Mimosa pigra in Australia: Comparing 90% to <5% at sites where the agents litterfall before and after the introduction the abundance of mimosa on Northern have established. Explicit modelling of biological control agents. Biological

Impact evaluation of weed biological control agents agents control weed biological of evaluation Impact Territory floodplains, provided of integrated weed management • Control 38: 166–173. overgrazing does not reduce the level strategies then revealed that other of fuel for fires. Routley, B.M. and Wirf, L.A. (2006). approaches such as burning and Advancements in biocontrol of Mimosa Limited impact could be attributed mechanical control were needed to pigra in the Northern Territory. In: to the other biological control agents reduce the short- to medium-term Preston, C., Watts, J.H. and Crossman, C. pigrae, N. gunniella and A. puniceus detrimental impacts of mimosa. N.D. (eds) Proceedings of the 15th during this study. Mimosa stands Australian Weeds Conference. Weed

Best practice guide practice Best Management Society of South Australia Inc., Adelaide, pp. 561–564. 10 ...case study Parthenium: an annual pasture weed

Parthenium (Parthenium hysterophorus) levels. Studies have been performed 2. Comparing plots with and is an annual herb, with a deep tap root at sites where livestock was excluded without agents and an erect shoot reaching up to 2 m to avoid potential confounding effects Three parthenium-infested sites were high. It germinates in spring–early due to grazing pressure, which positively surveyed for 2–3 years to primarily summer and flowers within 6–8 weeks. interacts with parthenium invasiveness. evaluate impact of the leaf-feeding A fully grown plant can produce more beetle Z. bicolorata, although the other 1. Before- and after-release than 15,000 seeds and the seedbank agents, E. strenuana and L. setosipennis assessments – photopoints persists in soil for long periods, with were also prominent at the sites. At nearly 50% of seeds remaining viable Photographs of a field heavily infested each site, parthenium patches were for up to 6 years. It generally senesces by parthenium taken before, during selected based on whether or not in late autumn, but under suitable and after an outbreak of the leaf- Z. bicolorata was present and defoliating conditions such as high rainfall, it can feeding beetle Z. bicolorata showed plants. Within each patch, ten 0.25 m2 grow and reproduce all year round. a decline in weed population and the quadrats were randomly selected and Parthenium is mainly found in recovery of desirable pasture species a range of plant and agent Queensland, where it out competes (Fig. 2). This series of photographs measurements recorded: plant height, useful pasture species and causes provided a visual impression of the plant density, flower production, total significant human health problems impact of this agent, which was plant biomass, viable soil seedbank, such as hay fever and asthma. supported with subsequent quantitative Z. bicolorata defoliation levels and, impact studies. Nine of the 11 biological control agents incidence of E. strenuana galls and released on parthenium in Australia L. setosipennis larval damage. Each year, since 1980 have established: the leaf- measurements were made in different feeding beetle (Zygogramma bicolorata) parthenium patches. released in 1980, the seed-feeding

weevil (Smicronyx lutulentus) released Best practice guide in 1981, the stem-galling moth (Epiblema strenuana) and stem-boring weevil (Listronotus setosipennis) both released in 1982, the leaf-mining moth (Bucculatrix parthenica) released in 1984, the winter rust fungus (Puccinia abrupta

var. partheniicola) released in 1991, • the stem-galling weevil (Conotrachelus Impact evaluation of weed biological control agents albocinereus) released in 1995, the root-boring moth (Carmenta ithacae) released in 1998 and the summer rust fungus (Puccinia melampodii) released in 1999. Of these agents, Z. bicolorata and E. strenuana appear so far to have Nov 1996 the greatest impact.

Post-release impact evaluation

Four main approaches have been used to evaluate post-release impact of the key agents released for parthenium biological control. They explored how climatic, spatial and temporal variability, Jan 1997 July 1998 plant competition and multiple agents Figure 2: Series of photographs of a parthenium-infested field at Mt Panorama in Central Queensland taken at a fixed reference point before (November 1996), during (January 1997) affect parthenium at the plant (including and after (July 1998) an outbreak of the leaf-feeding beetle (Zygogramma bicolorata). different life stages) and population Photos: K Dhileepan

11 case study In patches defoliated by Z. bicolorata, transplanted into the ground of each Similar field cage exclusion experiments parthenium density, height, biomass, sub-cage. Two sub-cages were randomly were also performed to measure impact flower production, germinable seeds allocated to a control treatment, which of E. strenuana and Z. bicolorata, from soil samples and seedling remained insect-free. Thirty insects although for E. strenuana both potted emergence, were reduced by up to were introduced in each of the other plants and parthenium seedlings, which 100% percent compared to that of six sub-cages when parthenium was at grew naturally from seeds in the field, unattacked patches. Nonetheless, the rosette, pre-flowering or flowering were used instead of transplanted because plants measured at each site stages (two replicate sub-cages per seedlings. were not randomly chosen and the growth stage). At the first signs of The field cage experiments for the three agents were not experimentally parthenium senescence, all plants were agents however, produced different manipulated, it is difficult to claim harvested and height, root length, stem results to those obtained in insecticide with certainty that changes in plant diameter, and number of branches, exclusion experiments (see below), performance were only due to the leaves, flowers, oviposition and larval with cage exclusion results generally agents. For example, plants in some feeding sites were recorded. Root, indicating a higher impact than patches may have been more susceptible stem and leaf dry weight biomass was insecticide exclusion for E. strenuana, to attack by Z. bicolorata, potentially also measured. lower impact than insecticide exclusion leading to confounded results. The Flower production was reduced by for L. setosipennis, and no difference negative effect of Z. bicolorata 38% percent when the agent was in impact between cage and insecticide defoliation on plant performance introduced at the rosette stage. No exclusions for Z. bicolorata. The parameters however, was observed reduction was observed when insects discrepancy observed for two of the over several sites and years, increasing were introduced at the pre-flowering agents may have been due to confidence in results. and flowering stages of parthenium. differences in insect population levels, The impact on flower production duration of experiments, influence of 3. Cage exclusion experiments measured in this experiment however, cages on the system and the use of The effect of the stem-boring weevil was only about half of that measured potted plants and seeded plants in the L. setosipennis on different growth in previous experiments involving potted cage exclusion experiment instead of stages of parthenium was assessed plants in a glasshouse. The stem-boring a natural field infestation of the weed. using a cage exclusion experiment weevil did not have any impact on total in the field. Three insect-proof cages plant height and biomass at the end 4. Pesticide exclusion experiments (12 m x 4 m) were set up outdoors and of the experiment, irrespective of the A pesticide exclusion experiment was each cage was further divided into eight parthenium stages at which insects carried out between 1996 and 2000 sub-cages using insect-proof partitions. were introduced. at two sites infested by parthenium Six parthenium seedlings were Impact evaluation of weed biological control agents agents control weed biological of evaluation Impact •

The leaf-feeding beetle (Zygogramma bicolorata) and the stem-galling moth (Epiblema strenuana) so far appear to have the greatest impact on controlling parthenium. Here (left) is a close-up of Z. bicolorata as an adult, feeding on a parthenium leaf, and (right) the larva of the stem-galling moth within a parthenium stem. Best practice guide practice Best Photos: K Dhileepan

12 case study Further reading

Dhileepan, K. (2003). Seasonal variation in the effectiveness of the leaf- feeding beetle Zygogramma bicolorata (Coleoptera: Chrysomelidae) and stem-galling moth Epiblema strenuana (Lepidoptera: Tortricidae) as biocontrol agents on the weed Parthenium hysterophorus (Asteraceae). Bulletin of Entomological Research 93: 393–401.

Dhileepan, K. (2003). Current status of the stem-boring weevil Listronotus setosipennis (Coleoptera: Curculionidae) introduced against the weed Parthenium hysterophorus (Asteraceae) in Australia. Close-up of a parthenium stem-boring weevil (Listronotus setosipennis) adult. Photo: K Dhileepan Biocontrol Science and Technology 13: 3–12. to measure impact of agents, primarily was also cut at ground level and its dry Dhileepan, K. (2007). Biological E. strenuana and Z. bicolorata. At each weight measured. control of parthenium (Parthenium site, eight 2.25 m2 plots were maintained hysterophorus) in Australian The effectiveness of E. strenuana and free of agents by monthly applications rangelands translates to improved Z. bicolorata was significantly affected of insecticides (monocrotophos pasture production. Weed Science by weather conditions, with major [Azodrin®] a systemic organophosphate 55: 497–501. impacts on parthenium seen in only applied as a foliar spray and carbofuran 1 of the 4 years, due to below average Dhileepan, K. and McFadyen, R.E. [Furadan®] a broad spectrum carbamate rainfall. By the end of the experiment, (2001). Effects of gall damage by the Best practice guide applied as granules to soil). A fungicide a significant reduction in the soil introduced biocontrol agent Epiblema (Mancozeb) was also applied at one seedbank of plots where agents were strenuana (Lepidoptera: Tortricidae) on site in 2000 following an outbreak of present was observed at one of the sites. the weed Parthenium hysterophorus the rust P. melampodii. Another eight In 1997, a 40% increase in associated (Asteraceae). Journal of Applied plots at each site were left exposed to grass biomass was observed at one site Entomology 125: 1–8. the agents and sprayed with water only. as a result of heavy defoliation and Dhileepan, K., Setter, S. and McFadyen, Densities of rosettes, pre-flowering galling of parthenium by Z. bicolorata •

R.E. (2000a). Response of the weed Impact evaluation of weed biological control agents and flowering plants, plant height and and E. strenuana. At the other site, grass Parthenium hysterophorus (Asteraceae) flower production were recorded in all biomass increased by 52% in 1998 to defoliation by the introduced plots at monthly intervals during the mostly due to parthenium galling by biocontrol agent Zygogramma bicolorata growing season. The proportion of E. strenuana and by 45% in 2000 (Coleoptera: Chrysomelidae). Biological plants with E. strenuana galls and the because of the combined effects of Control 19: 9–16. number of galls per plant, as well as the E. strenuana and the rust P. melampodii proportion of plants with Z. bicolorata on parthenium. Dhileepan, K., Setter, S. and McFadyen, damage and an estimate of leaf area R.E. (2000b). Impact of defoliation The variability in agent effectiveness defoliated, were also recorded. Soil by the introduced biocontrol agent due to climate would not have been samples were collected four times per Zygogramma bicolorata (Coleoptera: detected if the experiment had not year to determine the number of viable Chrysomelidae) on parthenium weed been conducted over multiple years. parthenium seeds present; a crucial in Australia. Biocontrol 45: 501–512. Maintaining pesticide exclusion parameter to measure in an annual weed treatments for 4 years however, was whose populations fluctuate widely both expensive and labour intensive. between years. In mid autumn, at the The pesticides also had to be applied end of each growing season, parthenium more frequently during warm or wet plants were pulled from the ground weather, which further increased the and total dry weight biomass assessed. costs of the experiment. The associated pasture grass in plots

13 ...case study Bridal creeper: a geophyte climbing vine

Bridal creeper (Asparagus asparagoides) the leaf beetle (Crioceris sp.) released plants were transferred to the field in is a climbing vine that smothers large in 2002, has so far established at only one experiment to assess agent impact areas of natural vegetation and a few sites. on rhizome and tuber production. threatens biodiversity. It is also found Three different approaches so far have climbing trees in irrigated orchards in Post-release impact been used to evaluate impact of New South Wales and Victoria, where evaluation agents, primarily the rust fungus, on chemical control is difficult to implement. Studies to evaluate impacts of biological bridal creeper. It develops extensive below-ground mats control agents on bridal creeper had of rhizomes and tubers, which provide to take into consideration the weed’s 1. Before- and after-release energy for shoot growth as well as a climbing habit, which makes it difficult assessments – comparing buffer against adverse conditions. In to distinguish between individual quantitative data winter rainfall areas, it begins to grow plants. Consequently, percentage One to 3 years before releasing the in late summer – early autumn, produces cover and number of shoots and leafhopper and / or rust fungus in the fruits in mid to late spring and above- biomass produced per surface area early 2000s, support structures (trellises), ground foliage naturally senesces at were recorded instead of the usual 2 m in height and 90 cm wide made the beginning of summer. Bridal creeper plant density measurements. Artificial with meshed wire attached to two steel is common in the temperate zone trellises or stakes had to be used to fence pickets, were set up at the edge of Australia. support climbing shoots on which of 3 m2 plots (three to four plots per Three biological control agents of South most fruit production occurs. Peak site) at 15 bridal creeper infested sites African origin have been released on measurements and sampling had to across southern Australia. Since then, bridal creeper in Australia since 1999. be performed in early to mid spring bridal creeper growth and reproductive The leafhopper (Zygina sp.) released before fruits had ripened, to prevent parameters, such as percentage cover, in 1999 and rust fungus (Puccinia removal by birds, and before shoots number and dry weight of shoots myrsiphylli) released in 2000 have had started to senesce. Due to climbing trellises and growing within successfully established across most difficulties in sampling the below- a randomly selected, permanent 1 m2 of the range of the weed. In contrast, ground biomass, standardised potted quadrat, and number of fruits produced have been recorded in mid-spring each year. The incidence and severity of damage caused by the agents have also been measured each year since their release.

Bridal creeper populations have drastically declined since the release of the rust fungus at most sites located in coastal areas (Fig. 3). Assessments made in 2007 revealed that most bridal creeper at these sites has been replaced by native plants or other weeds. A similar decrease in bridal creeper populations however, has not been recorded at drier inland sites since release of the

Impact evaluation of weed biological control agents agents control weed biological of evaluation Impact agents. Lower incidence and severity • of damage caused by the rust fungus has consistently been observed at these sites, where bridal creeper growth and reproductive parameters have considerably fluctuated from year to The rust fungus (Puccinia myrsiphylli) adversely impacts bridal creeper across most of its range. year. Although present at most sites, Photo: CSIRO leafhopper populations have been Best practice guide practice Best

14 case study vegetation was recorded, possibly due to the persistent below-ground tuberous mats of bridal creeper preventing recruitment. Longer-term monitoring of these plots is required to fully evaluate the response of other plant species to the biological control of bridal creeper and determine whether revegetation will occur naturally. The increase of exotic species in some plots where bridal creeper cover decreased points to possible future hurdles to rehabilitate these sites.

3. Fungicide exclusion experiments Using potted plants placed in the field: In late autumn bridal creeper plants with a similar number of tubers contained in pots infected or not with the rust fungus, were placed into a 2000 2005 natural bridal creeper infestation near Figure 3: Example of a trellis set up at Broulee, NSW, where bridal creeper has been monitored before and since the release of the rust fungus (Puccinia myrsiphylli) in 2001. Sydney, New South Wales, where agents Photos: CSIRO had not yet been released. This approach allowed for a robust examination of erratic and patchy across sites and over released in the centre of each bridal the effects of the rust fungus on the time, and thus may only play a minor creeper stand at each site. The sites below-ground biomass to be performed Best practice guide role in regulating bridal creeper were revisited in September 2006, in the field. The healthy plants were populations. Results from these before bridal creeper naturally maintained rust-free using monthly assessments are currently being analysed. senesced, to perform another series fungicide (Tebuconazole [Folicur®]) of vegetation surveys. applications, while rust-infected 2. Before- and after-release plants were sprayed with water only. In both vegetation surveys, the mean assessments – response of This fungicide was found not to affect number of native plant species and the associated plant communities the growth of bridal creeper in a

percentage cover of all native under- • following agent release previous experiment. After 6 months, storey plants were significantly lower Impact evaluation of weed biological control agents Four bridal creeper-infested sites in half the plants were harvested in the in bridal creeper areas compared to Western Australia were surveyed during spring and numbers and dry weight reference areas. Average bridal creeper 2004–2006 to assess the response of of shoots determined. Tubers were cover decreased from about 50% at plant communities following an epidemic counted, length of rhizomes measured, the time of the first vegetation survey of the rust fungus on bridal creeper. At and total below-ground biomass dried to 10% in the last survey, about 15 each site, plots (10 m x 1 m) were and weighed. The remaining plants were months after the rust fungus was established in stands of bridal creeper left in the field until they were harvested released within the bridal creeper stands. and in nearby reference areas, within the following autumn. During that time There was little change in native plant native vegetation with little or no bridal their foliage naturally senesced at the cover in these plots. At one site creeper present (two to three paired end of spring and new shoots were however, the cover of native climbers plots per site). All vascular plant species produced in late summer. increased from 0.07% to 5.0%, while were identified and percentage shoot the cover of other exotic species The above- and below-ground biomass cover of the understorey (to a maximum increased from 0.03% to 23.4%. of rust-infected plants harvested in height of 1.5 m) was estimated visually the spring was significantly lower than While there was a reduction in bridal for each species within each plot. In that of rust-free plants. Many of the creeper cover due to the impact of the June 2005, approximately 8 months rust-infected plants that remained in rust fungus during the period plots were after the first vegetation survey of plots the field until the following autumn was carried out, the rust fungus was surveyed, limited recovery of native

15 case study did not re-shoot and had few (if any) tubers, compared to rust-free plants which produced luxuriant foliage and had a well-developed below-ground biomass.

Using plots in a natural field infestation: Permanent 1 m2 quadrats (10 per site) with a central 100 cm long stake were set up in bridal creeper infestations at six sites across New South Wales and Western Australia where the rust fungus was the dominant, if not sole, agent present. Half of the quadrats at each site were maintained rust-free using monthly fungicide applications. The bridal creeper leafhopper (Zygina sp.) can severely damage bridal creeper but its populations The remaining quadrats were sprayed have been erratic and patchy across sites and over time. Photo: CSIRO with water only. The percentage cover of bridal creeper and other plant species within quadrats was measured infected quadrats was considerably less Stansbury, C.D., Batchelor, K.L., Morin, three times during the growing season. than in control quadrats and gradually L., Woodburn, T.L. and Scott, J.K. Numbers of bridal creeper shoots, fruits, decreased over years, indicating a (2007). Importance of support for fruit seeds and seedlings, and dry weight possible cumulative adverse effect production by the invasive climber, of above-ground biomass were also on below-ground organs. There was Asparagus asparagoides. Weed measured in September before bridal a consistent increase in leaf litter and Technology 21: 820–824. bare ground in the rust-infected plots. creeper naturally senesced, but after Turner, P.J. (2003). Interactions between When other plants did establish, bridal green fruits had developed. an insect, a pathogen and their host creeper was replaced by both native plant, bridal creeper. Honours Thesis, This experiment has so far been and exotic species, although other weeds Impact evaluation of weed biological control agents control weed biological of evaluation Impact conducted for 3 years. Bridal creeper University of Canberra, Canberra. • were more prevalent at disturbed sites. growth parameters in quadrats where Turner, P.J., Scott, J.K. and Spafford, the rust fungus was not excluded were Further reading H. (2008). Implications of successful significantly reduced compared to biological control of bridal creeper Morin, L., Neave, M., Batchelor, K.L. measurements made in rust-free (Asparagus asparagoides (L.) Druce) and Reid, A. (2006). Biological control: quadrats. Bridal creeper cover and the in south west Australia. In: van Klinken, a promising tool for managing bridal number of shoots produced in rust- R.D., Osten, V.A., Panetta, F.D. and creeper in Australia. Plant Protection

Best practice guide practice Best Scanlan, J.C. (eds) Proceedings of the Quarterly 21: 69–77. 16th Australian Weeds Conference. Queensland Weeds Society, Brisbane, pp. 390–392.

Disclaimer While every care is taken to ensure the accuracy of the information in this publication, the CRC for Australian Weed Management takes no responsibility for its contents, or for any loss, damage or consequence for any person or body relying on the information, or for any error or omission in this publication.