“It has taken a long time to get to this stage,” said Quentin Paynter, who spearheaded the development of the framework. “And it all started with our work on weed prioritisation for the Australian government.” MWLR and Research Associate Richard Hill first began work on the prioritisation of weed biocontrol targets in 2008, when we were awarded funding from the Australian federal government to prioritise weed targets in Australia. We, as an independent external agency, were contracted to assist with weed prioritisation because the individual Australian states and territories could not reach a consensus on the top priorities.
“The system we developed recognised that weed prioritisation must take three factors into account: (i) weed importance, which is based on the negative impacts a weed creates; (ii) the likely susceptibility of the weed to biological control; and (iii) the cost of implementing biocontrol, to determine which weed target is likely to result in the ‘best bang for your buck’,” explained Quentin.
Weed importance:
“For this aspect of developing the tool, the Australian government had already invested a lot of effort into ranking weeds by their importance when they determined their Weeds of National Significance (or WoNS). So rather than reinventing the wheel, we used the WoNS scoring system to rank weed importance in New Zealand, but focused on developing a system to predict biocontrol impacts,” he added.
Biocontrol success:
To predict biocontrol impacts, Quentin and researcher Chris McGrannachan compiled a list of factors that had been hypothesised to influence biocontrol success, and then assembled a database of published information on the impacts of past biocontrol programmes so that these hypotheses could be tested. The next step was to develop a system that scored weeds higher or lower depending on how many traits they possessed that were correlated with biocontrol success. The researchers found that this worked quite well: biocontrol impacts were invariably high against weeds that had high scores, while biocontrol had often failed against weed targets with very low scores.
Thereafter, funding from the Foundation for Research, Science and Technology was used to refine this approach by calling on the expertise of Jake Overton (MWLR Research Associate). Jake recently helped develop new statistical techniques that we could use to model the impacts of the various traits that were correlated with biocontrol success, alone or in combination, to produce a model with a combination of traits that best predicted biocontrol impact.
The results showed that the success of repeat programmes (biocontrol programmes using agents already developed and released in overseas weed biocontrol programmes) is predicted by the success of the novel, pioneering programme. For novel biocontrol targets, three traits provided a good ability to predict success: weediness of the target weed in its native range, mode of reproduction (sexual or asexual), and ecosystem type (aquatic or wetland versus terrestrial).
Weediness in the native range is important because species that are abundant enough to be considered weeds in their native range may become abundant there because they benefit from disturbances to ecosystems resulting from human activities such as fire or over-grazing. If human-related disturbance drives the abundance of a particular species, then biocontrol is less likely to succeed. On the other hand, if a target plant is uncommon or a minor component of the native flora but is weedy in its introduced range, it may be benefiting from the absence of specialist natural enemies in the introduced range. Biocontrol is therefore likely to be successful.
Mode of reproduction is important because clonal weeds tend to have lower genetic diversity in the introduced range compared to species that reproduce sexually. In extreme examples, such as the case of tradescantia (Tradescantia fluminensis) in New Zealand, a single clone may be present, so there is very limited opportunity for the evolution of resistance to biocontrol to occur. In contrast, genetically diverse, outcrossing, sexually reproducing weeds such as gorse (Ulex europaeus) are more likely to evolve resistance. For example, winter flowering in gorse is likely to have evolved because of selection by seed-feeding biocontrol agents, which are most active in spring and summer.
In terms of ecosystem type, aquatic and wetland weeds are more susceptible to biocontrol than terrestrial weeds, which may be related to habitat stability. For example, in static waterbodies, conditions don’t change much, and floating weeds tend to provide a stable resource for biocontrol agents. On land, however, a disturbance such as a bush fire can temporarily wipe out populations of weeds and biocontrol agents. Weed populations often recover quickly from soil seed banks, but it can take a lot longer for biocontrol agents to reinvade the regenerating weed and build up damaging populations, resulting in patchy biocontrol impacts.
Predicting the cost of biocontrol:
We already knew that repeat programmes are much cheaper than novel/pioneering programmes, because repeat programmes omit costly overseas survey work and most, if not all, host specificity testing. To quantify this, we used MBIE funding to compile a database of New Zealand biocontrol programmes and went through past budgets to calculate the cost of each. Our analysis showed that two factors explained virtually all the variation in programme cost. Pioneering programmes cost about 4.2 times more than repeat programmes, and cost also increases with the number of agents released, indicating that more efficient agent selection should reduce the cost of future programmes.
Once this work was published, ranking weeds by importance was the only remaining task to enable prioritisation. Auckland Council had already developed a tool for ranking weed biocontrol targets, which we refined on the basis of a model developed by Paul Downey and colleagues (University of Canberra). Their model had been used to rank environmental weeds in New South Wales without quantitative data on weed impacts. Our aim was to develop a system (in this case, an Excel file) that produces meaningful data for the NBC without being too onerous for councils to input information. We felt the system should score weed importance according to weed distribution and weed impacts, summed across a range of habitats. The system also needed to consider factors important to the NBC that were not included by Downey, such as the socio-political pressure to control, and the ease and cost of control by existing means, so that species that are difficult and expensive to control score higher than species that are more easily controlled using existing methods.
In September 2021, a workshop was held to discuss the tool and it was agreed that it strikes a good balance between selecting important weeds and good biocontrol targets. One point of discussion was the high ranking of weeds such as gorse, which has been a long-term biocontrol target but with limited impacts to date. This is partly explained by the ecosystem impact scoring favouring widespread terrestrial weeds and the relatively low cost of implementing biocontrol against existing targets compared to novel targets. This led to a modification to the ranking system by amalgamating some terrestrial ecosystems and splitting aquatic ecosystems so that static waterbodies (lakes and ponds) and flowing water (streams and rivers) were scored separately, to reduce the scoring bias towards terrestrial weeds.
Another issue was the ranking of current biocontrol targets. This can potentially be misleading due to inevitable lags between agents being released and successful biocontrol, which can result in weeds being ranked highly when further work on developing biocontrol agents may not be necessary. This is likely to be the case for tradescantia, for example, which ranked number one. Hence, for existing (or past) targets, such as gorse and tradescantia, a discussion is required on a case-by-case basis to decide whether further work is justified or should be abandoned. In other words, the ranking system is designed to help identify the best targets, but it does not need to be followed slavishly.
In conclusion, the NBC prioritisation tool, which now ranks 158 weeds of importance to members of the NBC, is a much better system than previously used to guide decisions about prioritising funds for weed biocontrol research, but the tool is only as good as the information provided. It is crucial that information used to rank weeds is kept up to date and the rankings regularly reviewed, since these could change dramatically; for example, if an overseas programme results in successful control and could be repeated in New Zealand. By following and refining this prioritisation approach we hope to see a tangible benefit to the NBC through increased cost-effectiveness of our biocontrol work in the years to come.
Funding
This project was funded by the National Biocontrol Collective, and Envirolink through the Ministry of Business, Innovation and Employment.
