Surveying for Biocontrol Agents in a Weed’s Native Range

The very first premeditated classical weed biocontrol programme was remarkably unsophisticated: 14 insect species collected by Albert Koebele in Mexico were released on lantana (Lantana camara) in Hawaii in 1902. Agent selection was based on Koebele’s judgement and didn’t even include host specificity testing. Most species Koebele released failed to establish or had only slight impacts, and four species that did establish attacked non-target plants. The discipline has advanced substantially since then!

Nowadays candidate agents must undergo host specificity testing to ensure environmental safety, and current best practice protocols have proven their predictive power over decades of safe introductions. But for a weed biocontrol programme to succeed agents must also be sufficiently damaging. Consequently, work to identify a weed’s natural enemies and prioritise promising candidate agents for further study is arguably the most important stage of a biocontrol programme.

In 1970 Tony Wapshere, a researcher based at CSIRO in Australia, championed using meticulous ecological studies to assist the selection of effective agents. Wapshere described work on rush skeleton weed (Chondrilla juncea), for which data on weed density and natural enemy infestation levels in relation to climate and soil conditions were collected in both the native and invaded range of the weed. Many ensuing programmes followed this approach. For example, in the 1990s a large-scale study compared aspects of Scotch broom (Cytisus scoparius) ecology in its native (England, France) and introduced (Australia, New Zealand) ranges. Principal researcher Quentin Paynter, who took part in that study, noted that “It was very enlightening – we showed that successful seedling recruitment through existing broom stands was common in the introduced range, where plants produce bigger seeds, resulting in more competitive, shade-tolerant seedlings. Consequently, invasive broom populations were much more likely to persist for generations, unlike populations in the native range. Broom seed beetles preferentially attack large seeds – so a selection pressure for small seeds has now been reimposed with the establishment of the seed beetle throughout New Zealand.”

Simon Fowler, another BSI principal researcher who worked on the broom project, noted that research funding only stretches to conducting native range ecological studies on a few focal species, such as broom and tradescantia. This might not be as bad for biocontrol as it sounds, because native range ecological studies are not a panacea. For example, in 1973 Peter Harris, a researcher based in Canada, noted that the multiplicity of species attacking a weed in its native range makes it difficult to distinguish the effects of each. “This was true for Scotch broom,” said Simon, adding that “a classic long-term insecticide exclusion trial revealed natural enemies have a chronic but major long-term deleterious impact on the fecundity and longevity of broom plants in the UK, but the authors of the study could only speculate regarding which natural enemies were likely to be responsible.”

Harris also noted that impacts in the introduced range are hard to predict because biocontrol agents will be introduced without the natural enemies that attack them in their native range.

However, although agents are introduced without parasitoids, some gain them in the introduced range. The likelihood that an agent will be attacked by specialist native parasitoids can now be reliably predicted according to the presence of ‘native ecological analogues’ (defined as a native species that belongs to the same superfamily as the agent and occupies a similar niche on the target weed). Agents that are attacked by parasitoids shared with native analogues have all failed to control target weeds in New Zealand. “Surveying target weeds in New Zealand to identify potential native analogues of candidate agents increases the odds of picking winners by winnowing out candidates that are likely to fail,” said Quentin. “Moreover, guilds with concealed larvae are more likely to be regulated by specialist parasitoids and potentially benefit most from being released from parasitism in the introduced range. This would explain why some species such as the biocontrol agent for Syndey golden wattle – the bud gall wasp (Trichilogaster acaciaelongifoliae) – that are not particularly abundant in the native range proliferate and become hugely successful in the introduced range,” he added.

A recent review of completed weed biocontrol programmes in New Zealand found that agents that were observed to be highly damaging (i.e. completely defoliating or killing plants, or reducing populations in the field) in the native range were almost invariably highly damaging in the introduced range. The review included a prototype scoring system to help prioritise candidate agents, as seen in the table below.

Table 1 - Scoring system for prioritizing novel candidate agents. Predicted impact score = Potential specificity score (Score A) × (Potential damage score (Score B) + Guild score (Score C) + Analogue score (Score D)). Scores B-D are based on mean binomial impact scores for agents with those traits (agents that have “heavy”, “medium”, or “variable” impacts on host plants in NZ were scored 1, and agents that cause “slight” or “no impact” were scored 0). Score A is an additional weighting, promoting agents that are likely to be adequately host specific and penalising agents that are not.

Table 2 - Rankings for candidate arthropod agents for moth plant, prioritised according to the system (excluding species already rejected on the basis of confirmed host records indicating lack of specificity).

Although a simple tool to help select effective agents sounds compelling, the detective skills of a biocontrol researcher cannot be understated: some guilds (e.g. root-feeders) cause cryptic damage, and although agents that only damage reproductive structures rarely control a plant on their own, they should not be discounted entirely because they can enhance overall control in combination with other agents, and reduce weed invasion into new areas or reinvasion after conventional control.

Damaging agents may be overlooked if survey work is inadequate, but the efficiency of native range survey efforts is rarely scrutinised. CSIRO researchers working on a tropical wetland weed Mimosa pigra, noted that 746 collections were made at 277 localities throughout Mexico, the Caribbean, Central and South America. Despite this huge effort, most (11 of 12) insect agents that were eventually released in Australia were found in only 23 collections made at three sites in Mexico. They concluded that if researchers had conducted an early analysis of the results of the collections, then a decision to terminate the survey work could have been made earlier, freeing resources for other research, such as improved agent selection. “I think this study makes a valid point, that promising candidate agents are likely to be widespread and found quite quickly during native range surveys,” said Quentin.

Moreover, had this programme been conducted more recently, advances in genetic matching might have resulted in more targeted surveys. For weed species with large native ranges, genetic matching of native and invasive populations can ensure the target weed is surveyed in an appropriate region, maximising the chance that agents will be compatible with the invasive biotype, thereby cutting survey costs and greatly improving the chance of success.

Finally, the proliferation of easily accessible online data means that promising candidate agents can sometimes be identified without leaving the office, so feasibility studies conducted at the very start of a programme should include a thorough review of published literature on the natural enemies of the target weed.

Although ecological studies are no longer de rigueur, tools that have been developed should ensure candidate agents have a much higher chance of being effective than resorting to the much-derided ‘lottery approach’ of ‘find them, screen them, release them’, assuming that at least one ought to work!

This research was funded by the Strategic Science Investment Fund of the Ministry of Business, Innovation and Employment.