My favorite definition of “biological control” is a graphical definition. Figure 1 shows a Levins’ diagram, where circles represent a negative interaction and arrows represent a positive interaction; solid lines represent a direct interaction and dashed lines represent an indirect interaction. Levins’ diagrams were introduced to me by Dave Andow; my PhD advisor, George Heimpel, drew my attention to their usefulness in defining biological control.
Essentially, biological control is a situation where a biological population has an indirect positive effect on something we value (the dashed line in Fig 1). This positive indirect interaction is a result of the population’s direct negative effect on a pest population which itself has a direct negative effect on the thing we value. The “thing we value” might come from a third trophic level, like grain from a population of Triticum aestivum, or it may be something abiotic, like the throughput volume of water pipes (e.g. something that invasive zebra mussels can adversely impact by creating clogs).
One of the cool things about Levins’ diagrams is that the sign of an indirect interaction (positive or negative) can be determined arithmetically by multiplying the direct interactions between it. In Fig 1, two direct negatives equal an indirect positive. While the sign of the interaction is obvious in Fig 1, the utility of this arithmetic property becomes more valuable in more complex systems.
Classical biological control involves the importation and release of a reproducing population of predators, parasites, parasitoids, or herbivores to manage a pest species. The goal is to permanently establish a reproducing population of the biological control agent which will suppress the invasive population to harmless low densities. Essentially, classical biological control is meant to establish a new indirect positive interaction.
When we discuss risk in classical biological control, we are typically referring to the risk of our biological control agent attacking a non-target species. Figure 2 describes one possible scenario for non-target impact. Not only does the biological control agent attack the non-target species, but because the target population has a positive impact on the biological control agent, there is an indirect negative impact of the pest species on the non-target species – a direct negative times a positive is an indirect negative.
This got me wondering: Are there many documented classical biological control introductions where an indirect negative interaction caused a non-target species to go extinct?
I made the following ISI Web of Science search: ts=(extinction*) and ts=(“classical biological control” or “importation biological control” or “classical biocontrol” or “importation biocontrol”), which generated only 15 results.
One of the results, a review article by Francis Howarth (1991), provides a good summary (although see Follet et al. 2000, discussed below). The summary appears to still be up to date, because, as far as I can tell, all of the extinctions were caused by introductions that occurred prior to the 1990s.
What did Howarth have to say…?
Possible extinctions caused by classical biological control:
1. Levuana iridescens (the coconut moth) was native to Fiji and was attacked by the tachinid fly Bessa remota from Malaysia. This example isn’t really a non-target extinction, though, because the tachinid was intended to attack L. iridescens. The coconut moth was considered a pest in coconut plantations, even though it was probably an endemic species in Fiji [Actually, it was probably NOT endemic – see correction below (4-16-15)]. The biological control agent is still present on the island, parasitizing other local moths, and it has likely caused extinctions of other native species too.
3. The large blue butterfly, Maculina arion, has been extirpated from several parts of Europe and was thought to be extinct for awhile. One hypothesis for the cause of M. arion declines is the release of the Myxoma virus to control rabbits. Weird, right? The decline in rabbits changed the plant community to be less habitable to certain ant species. The larval stages of M. arion are obligate parasites in ant colonies. This was a totally unexpected indirect effect of biological control (if we count viruses as biological control agents).
4. Euglandina rosea, the rosy wolf snail, has been introduced to several Pacific islands to control the invasive giant African snail, Achatina fulica. While it has not done a good job of controlling this target pest, it has decimated dozens of species of endemic snails. Tree snails in the family Partulidae have lost 56 of 61 species endemic to the Society Islands (Coote & Loève 2003)! The rosy wolf snail is native to tropical North America, and was first used as a classical biological control agent in Hawaii, where it has also devastated native tree snail species.
Okay, I should stop numbering these, because the list goes on a bit longer than I’d expected…
Howarth (1991) mentions several other potential examples, including damselflies in Hawaii attacked by Gambusia fish introduced for mosquito control; Hawaiian shrimp extirpated by tilapia introduced for weed and mosquito control; a bunch of other Hawaiian Lepidoptera have likely been driven extinct by the slew of generalist predators and parasitoids released on the islands; some native Hawaiian natural enemies that lost their prey/hosts due to introduced natural enemies have likely gone extinct; some reptiles on Caribbean islands were driven to extinction by mongoose released to control rats….
Clearly an introduced biological control agent can cause an extinction. Most of these examples occurred on islands. At least one of the extinctions was of the target species, which should properly be considered “neo-classical” biological control (i.e. where the target is a native species and the biological control agent is an introduced species) [but see correction below (4-16-15)]. However, a paper by Follet et al. (2000) points out that much of the evidence presented by Howarth (1991) is in the form of “casual” correlations, and that a causal connection between these biological control agents and non-target population declines is not always clear. However, at least in the case of the rosy wolf snail, I would argue that case for causal link between the biological control agent and tree snail declines is difficult to deny. The same pattern of colonization by the rosy wolf snail and decline of endemic snails has been repeated on multiple islands.
To answer my original question, an indirect negative interaction could have been important in all of these cases. All of the putative extinctions, except in the case of the Myxoma virus, were cause by biological control agents with wide host/prey breadth. And with Myxoma, the interaction was certainly due to indirect effects. But bearing in mind the points made by Follet et al. (2000), inferring a causal connection for the indirect effect on non-target decline remains, admittedly, elusive. I should add that my reading into the cases mentioned above has not been exhaustive.
Of course, extinctions can be caused by introduced populations without indirect interactions. This may be particularly likely in cases of introduced pathogens with high transmissions rates (Fisher et al. 2012).
All of this information serves as a reminder that careful host range testing of biological control agents and detailed taxonomic knowledge of local biodiversity is critical if we are to safely conduct classical biological control projects. I’d also suggest that neo-classical biological control is rarely, if ever, a good idea. I do maintain, however, that classical biological control can be a valuable management tactic. In managing invasive species we are confronted with difficult choices, and inaction involves its own uncertainties and adverse consequences.[correction (4-16-15): I did some follow-up reading on the Levuana moth, and it seems that it was likely NOT endemic to Fiji. Kuris (2003) presents a compelling case that the moth was invasive and was spreading through the Fijian archipelago. Interestingly, Kuris reports that the motivation for releasing Bessa remota was to preserve the coconut palm which was integral to Fijian culture, and that it was not released for solely economic reasons like helping coconut plantations, as I had presumed. Hoddle (2006) furthers the argument and makes the case that it is not clear Levuana has been extirpated from Fiji, let alone reached a global extinction. In his review of obscure literature, Hoddle paraphrases JD Tothill (the PI leading the release of Bessa remota in the early 20th century) who notes that two remote islands still had occasional Levuana outbreaks (at least in the late 1920s) and this was likely because alternative lepidopteran hosts were unavailable. This is a strong suggestion that indirect effects played a roll in suppressing Levuana. Unfortunately, Hoddle’s attempt to find Levuana on these islands was thwarted by the Fijian military, which has converted the islands into prison camps for the alleged participants of the May 2000 attempted coup d’etat.]