What Does Evolution Mean for Biocontrol?
Cinnabar moth.
We are often asked about the implications of evolution for weed biocontrol: can biocontrol agents mutate or adapt over time to thrive better under New Zealand conditions, or perhaps to attack new hosts, and can weeds develop resistance to them? In this story we examine the evidence and consider what evolution might mean for weed biocontrol.
First of all, could successful biocontrol break down over time because the host develops resistance? Resistance to herbicides and pesticides by plants and insects is commonly reported, but resistance to biocontrol agents appears to be rare. Releases of biocontrol agents have been made for over 100 years in different places around the world, but there are few examples of resistance developing. One example, which is close to home, relates to insect biocontrol. There has been some concern recently that Argentine stem weevils (Listronotus bonariensis) in New Zealand pasture are becoming resistant to a parasitic wasp (Microctonus hyperodae) that was introduced as a biocontrol agent in 1990. The parasitic wasp is parthenogenetic, which means it reproduces asexually, producing offspring that are all females. This reproductive strategy has the advantage of not needing to find a mate to breed, but it has the disadvantage of not having the opportunity provided by sexual reproduction to recombine genes and thus provide opportunities for adaptation to changing conditions.
Parthenogenetic agents are not often used for weed biocontrol in New Zealand. Current examples include the hieracium gall wasp (Aulacidea subterminalis) and the soon-to-be-released giant reed gall wasp (Tetramesa romana). However, old man’s beard sawfly (Monophadnus spinolae) females can produce offspring parthenogenetically if mates are in short supply. Ironically, while clonal biocontrol agents might not be ideal, biocontrol of weeds appears more likely to succeed against clonal plants than sexually reproducing weeds. “We hypothesise that there is an ‘evolutionary arms race’ going on and plants that reproduce asexually (e.g. producing clonal offspring) have less opportunity to evolve resistance to agents,” said Simon Fowler. It might be significant that our two fully parthenogenetic agents attack hieracium (Pilosella spp.), which is an apomict (produces seed asexually), and giant reed (Arundo donax), which does not produce viable seed and reproduces clonally. Perhaps these insects and their hosts are clonal because they don’t have to keep up with an evolutionary arms race?
There is some evidence that an evolutionary arms race can at different times favour a biocontrol agent or its host plant. We are studying this relationship in Scotch broom (Cytisus scoparius). Studies have shown that broom seeds in Zealand, while highly variable in size, are still on average around 40% bigger than their European counterparts, which helps explain why broom is so invasive here. In its home range broom relies on disturbance to regenerate or it disappears. Not so in New Zealand, where seedlings happily grow up below existing stands, ensuring populations are perpetuated. There is evidence to suggest that big broom seeds are more successful at producing seedlings that can survive and grow in the shade of existing stands.
“We hypothesise that broom seeds have gradually become bigger in the exotic range because of an absence of broom seed beetles (Bruchidius villosus),” explained Quentin Paynter. Seed size has implications for the seed beetles: the larger the seed, the larger the beetle that emerges, and big beetles are more successful at surviving winter and producing offspring. Now that broom seed beetles are abundant here they may over time create a selection pressure that favours small seeds, as these will reproduce more successfully than the big-seeded plants favoured by the beetles. “This could result in less competitive broom like that seen in the native range,” suggested Quent. However, the broom seed beetle would not do as well under that scenario, so biocontrol could break down, potentially allowing broom to bounce back.
If this were to happen, another seed-feeder that does not appear to rely on large seeds is available, which could be introduced to New Zealand. Larvae of the broom seed weevil (Exapion fuscirostre)feed externally on multiple seeds and are therefore not affected by seed size, and would be fine regardless of seed size. Also, with the way the broom gall mite (Aceria gentistae) is performing, many broom plants may in future not survive to produce much if any seed! We are monitoring broom seed size in New Zealand every 5–10 years to see if changes are occurring, so we will know if any other actions will be needed to stay on track with our goal to successfully biologically control broom.
Another evolutionary consideration is whether agents can rapidly evolve to adapt to new environments, given that evolution is often a slow process. This is the primary focus of Professor Peter McEvoy from Oregon State University, in the USA, who visited New Zealand recently. “I like to test some of the assumptions of ecological theories associated with biological invasions,” said Peter. “It is important to understand the evolution of plant strategies such as dispersal, or their ability to defend themselves against pathogens or predators in relation to biocontrol programmes,” Peter said, adding that it is also important to understand how evolution itself actually works.
Evolution in the form of natural selection occurs only when there is underlying variation in the relevant ecological traits. The variation must also be heritable – passed on genetically from parents to offspring – for persistent change to occur. Also, the change must not compromise any other important traits. (For example, if faster development led to smaller body sizes, that could have negative consequences for survival and reproduction.) Finally, there must be strong and consistent selection pressure so that individuals with a particular set of traits do consistently better. “These can be measured by comparing fitness over time, which increases the ability to contribute to future population growth,” explained Peter.
To test some of the assumptions regarding rapid evolution, Peter and his team have been studying the response of the cinnabar moth (Tyria jacobaeae)to variable season length. The cinnabar moth was introduced to the USA as a biocontrol agent for ragwort (Jacobaea vulgaris) in 1960, readily establishing in the Willamette Valley (87 m a.s.l), Oregon. Not only did the moth expand its geographical range to occupy the Cascade Mountains (up to 1,572 m a.s.l), but it also then encountered and attacked two native plants, which like ragwort are also in the Asteraceae family. This attack was predictable in advance and not a host shift, outlining the importance of considering the potential geographical range an agent can occupy when deciding if it is safe to release. Peter argues that there is a strong case for evaluating the evolutionary potential of candidate agents to adapt to new climates prior to their release.
Peter and his colleagues used altitude as an environmental gradient, comparing the development times of juvenile cinnabar moths from the valley floor with those in the mountains. “In the mountains, the season length is short and the number of days with sufficient temperatures for the moth to develop is reduced,” explained Peter. Essentially, the question they were asking was: can evolution rescue a faltering population at the margin of its range by speeding up development? “We found that as you moved up the gradient of increasing elevation, there was evidence that the moths had evolved a shorter development time to adapt to the shorter growing season,” said Peter. The study concluded that natural selection was a reliable explanation for the observed differences in phenology (timing of lifecycle activities).
“The question of rapid adaptive evolution has been examined closely in the heather (Calluna vulgaris) biocontrol programme here in New Zealand,” said Simon Fowler. For various reasons the population of heather beetles (Lochmaea suturalis)established in New Zealand was extremely small and lacked genetic diversity, creating a genetic ‘bottle-neck’. This limited the ability of the heather beetles to survive in their new, and relatively harsh, environment at Tongariro National Park. “The beetles’ body mass was too small and they had insufficient fatty reserves to survive the conditions they faced over the winter,” said Simon. Since then, larger heather beetles have been brought over from the UK to improve the genetic diversity. Nevertheless, in the interim the original population has expanded hugely, causing significant damage to heather, so it is possible that they may have already evolved to cope better with the conditions without intervention.
Adaptive evolution could also explain some of the lag phases that we see when agents are released. “Often agents are released and they don’t seem to cause much damage to their host plant for quite a few years, and then all of a sudden they become abundant and start doing their job, much to the relief of everyone,” said Simon. An agent released against the Californian thistle (Cirsium arvense) is a good example. The Californian thistle leaf beetle (Lema cyanella)was released widely, and in large numbers, in the 1990s, but only survived and clung on in low numbers at one site in Auckland. Then, around 15 years after first being released at the site, the beetle became common, causing considerable damage to the thistles. It is plausible that the leaf beetle has evolved in some way to perform better, like the cinnabar moth has at higher altitudes in Oregon.
Research into rapid adaptive evolution of biocontrol agents also has important implications for predicting responses to climate change. Predicting the potential for biocontrol agents to adapt and keep pace with climate change will become increasingly relevant. Previously we have studied whether the currently successful biocontrol for ragwort is likely to break down under climate change scenarios. “But there did not appear to be undue cause for alarm,” said Simon Fowler. Generally it seems that if weeds change their distributions, their biocontrol agents will simply follow, as the changes will not be outside the acceptable range for them. Plus warmer temperatures and fewer frosts may even suit some biocontrol agents better.
While adaptive evolution to better exploit local climatic conditions would appear to be a good thing for biocontrol, the ability to evolve to attack other hosts may not be desirable. There is no evidence worldwide to suggest this has occurred for any weed biocontrol agents. Follow-up studies of attack on other hosts invariably finds that the agent was always able to attack it, but that testing had been insufficient to show this beforehand, rather than an evolutionary expansion of host range occurring. The potential to evolve and attack new hosts remains possible, but fortunately it appears that the conditions for this to happen are very rarely met. First, a random mutation needs to occur that is overwhelmingly beneficial. Then conditions would need to favour the mutation becoming more common when, as a rule, a rare mutation will struggle to persist in a population because it is rapidly ‘swamped’ by more common genotypes. It has been calculated that the risk of an agent evolving to change host is between 1 in 10 million and 1 in 100 million – and the risk of a native species unexpectedly becoming a problem is the same.
One safeguard to prevent biocontrol agents from over time attacking non-host plants is limiting the types of organisms used for biocontrol to invertebrates and fungi. Simpler organism like viruses and bacteria are less stable and more easily able to mutate and change, or even to share genetic information with similar species, so we have steered clear of them.However, while the risk of host expansion beyond the fundamental host range (plants that an agent is physiologically able to complete its life cycle on, but may not currently attack in the field) is low, theoretically host expansion within the fundamental host range is more likely. Should we therefore routinely host-range test the offspring of individuals that survived on non-target hosts to determine the potential for the evolution of improved performance on non-target hosts so that they become realised hosts in the field? And if so, how many generations should be considered sufficient?
Simon and Quentin agree that more can, and should, be done here, particularly when using oligophagous insects, which can attack several closely related plant species. “We need a framework for assessing whether a fundamental host is ever likely to become a significant realised host,” agreed Simon. Ideally these tests would be completed in the native range of the agent, but this is not always possible for logistical reasons so we need to be able to predict the risk of attack using more conservative laboratory tests. We might, on the basis of a relative risk score, decide that a plant is unlikely to be a field host. However, if there is adequate selection pressure, an agent might evolve to perform better on that fundamental host and include it in its realised host range. This risk can be tested by rearing an agent for multiple generations on a potential non-target plant to see if there is selection for improved performance. “For example, when we reared the privet lace bug (Leptoypha hospita)for more than one generation on the non-target host lilac (Syringa spp.), the culture died out, indicating that the risk was low,” commented Quentin.
But there are occasionally situations where evolutionary host range expansion might not necessarily be a bad thing. In one example from Australia, eight strains of the blackberry rust fungus (Phragmidium violaceum) were released to try to improve biocontrol of blackberry (Rubus fruticosus agg.). The common name ‘blackberry’ refers to multiple morphologically distinct but closely related species, known as microspecies. These range from highly resistant to susceptible to the strain of blackberry rust illegally released in Australia in 1984, which also quickly self-introduced to New Zealand. The plan behind releasing multiple strains of the rust was that they would interbreed and hopefully develop forms that could attack blackberry, yet remain constrained within the Rubus fruticosus aggregate and not attack indigenous Rubus species. This example highlights the requirement for further testing to understand the implications of evolution for weed biocontrol, and for adequate risk assessment prior to agent release.
Work to better understand the implications of evolution for weed biocontrol in New Zealand is funded by the Ministry for Business, Innovation and Employment as part of Manaaki Whenua – Landcare Research’s Beating Weeds programme. Peter McEvoy is keen to find out whether the cinnabar moth has also expanded its range here. He would greatly appreciate any information from New Zealand about where people have observed cinnabar moths in recent years, or where they are seen this coming summer. Please email Peter at mcevoyp@science.oregonstate.edu
