Whodunnit? Solving the Case of the Disappearing St John’s Wort
Unprotected plot in the foreground attacked by beetles, compared with plot behind protected with insecticide.
Although we are often able to provide simple data showing that weeds decline once biocontrol agents are released, this may not provide sufficiently robust evidence of cause and effect for other scientists reviewing our work. We have to be prepared, sometimes, to demonstrate experimentally that the biocontrol agents are responsible for the decline in weed populations and not other factors.
Ronny Groenteman presented a paper at ISBCW about her research to nail the reason for St John’s wort (Hypericum perforatum) decline in New Zealand, which she has been undertaking for 5 years now. Ronny reminded the audience that it is essential to start with appropriate experimental design so that the results can clearly support or refute a hypothesis, and to be able to analyse the results appropriately. “I then showed how multi-model inference analysis can be used as a framework to demonstrate cause and effect in a biological control programme. This type of analysis provides a more holistic approach,” explained Ronny.
When Ronny undertook an insecticide exclusion experiment on St John’s wort, she was able to clearly demonstrate that if St John’s wort beetles (Chrysolina hyperici and C. quadrigemina) are removed from the plants using insecticide, the plant recovers and becomes more abundant. But because Ronny also recorded all the factors that could affect the population dynamics of the plant (e.g. rainfall, temperature, disease, insecticide effects, and herbivory by other insects), she was able to gauge how important these factors were in explaining the results. The multi-modelling approach allows us to test the dominance of biocontrol over all other factors,” said Ronny. And yes it was the St John’s wort beetles that did it!
St John’s wort is a good weed to study because the plants grow quickly and there are good numbers of beetles attacking it. Rather than just assuming their presence, Ronny quantified the background number of beetles present. “It is surprising how often biocontrol scientists neglect to document a basic element like biocontrol agent abundance in an experiment. It takes more resources to undertake this type of work but some of the variables such as changes in ground cover are relatively simple to quantify and can be assessed retrospectively from time-sequence photographs,” added Ronny.
By using population dynamics modelling to analyse the data, Ronny will be able to identify the most vulnerable life-stage of the plant. This is particularly useful in biocontrol studies because agents that affect the plant at the population level are needed rather than agents that just stunt its growth temporarily. The technique can also be used to predict the responses of biocontrol systems to changes in seasonal weather patterns. For example, beetles (Lochmaea suturalis) released to control heather (Calluna vulgaris) in the Tongariro National Park can survive heavy snowfall in the winter but die when subjected to pronounced diurnal temperature fluctuations in the spring. Predicting the response of agents and weeds to variable weather conditions can be important. This allows times when additional resources may be required to manage weeds to be flagged, such as when seasonal conditions are forecast to be less suitable for the agents or particularly favourable for the weed. Predicting responses of agents to weather conditions is also vital when selecting suitable source populations for new agents overseas. New Zealand is already renowned for its variable weather, and climate change may add a new dimension to agent selection if biocontrol agents need to be able to survive more extreme weather events in the future.
Ronny admits that the experimental process she is advocating is intensive but encourages others to adopt similar frameworks when designing and conducting their research so that their results are sufficiently robust and transparent enough to show clear cause and effects of the agents on the target weed. “This ‘data-hungry’ framework is not practical or affordable for every biocontrol experiment,” Ronny added, “but should be seriously considered for flagship programmes and for programmes that are likely to prove particularly challenging.”
This research was funded by the Ministry of Business, Innovation and Employment as part of Landcare Research’s Beating Weeds Programme.