John M. Randall and Mandy Tu
Biological control (biocontrol for short) is the use of animals, fungi, or other microbes to feed upon, parasitize or otherwise interfere with a targeted pest species. Successful biocontrol programs usually significantly reduce the abundance of the pest, but in some cases, they simply prevent the damage caused by the pest (e.g. by preventing it from feeding on valued crops) without reducing pest abundance (Lockwood 2000). Biocontrol is often viewed as a progressive and environmentally friendly way to control pest organisms because it leaves behind no chemical residues that might have harmful impacts on humans or other organisms, and when successful, it can provide essentially permanent, widespread control with a very favorable cost-benefit ratio. However, some biocontrol programs have resulted in significant, irreversible harm to untargeted (non-pest) organisms and to ecological processes. Of course, all pest control methods have the potential to harm non-target native species, and the pests themselves can cause harm to non-target species if they are left uncontrolled. Therefore, before releasing a biocontrol agent (or using other methods), it is important to balance its potential to benefit conservation targets and management goals against its potential to cause harm.
Organisms used to feed on, parasitize, or otherwise interfere with targeted pests are called biocontrol agents. There are several general approaches to using biocontrol agents: 1. ‘Classical’ biocontrol targets a non-native pest with one or more species of biocontrol agents from the pest’s native range; 2. the ‘New Association’ or ‘Neoclassical’ approach targets native pests with non-native biological control agents; 3. ‘Conservation’, ‘Augmentation’ and ‘Inundation’ approaches maintain or increase the abundance and impact of biocontrol agents that are already present, and in many cases native to the area. Classical biocontrol is by far the most common approach for plant pests. Conservation and augmentation approaches show great promise on their own and especially for enhancing the impacts of classical biocontrol and other weed control measures as researchers and managers focus on managing to maximize native biological diversity in invaded ecosystems (Newman et al. 1998).
CLASSICAL BIOLOGICAL CONTROL OF WEEDS
It is hypothesized that some non-native plants become invasive, superabundant and damaging, at least in part because they have escaped the control of their ‘natural enemies’, the herbivores and pathogens that checked their abundance in their native ranges. Classical biocontrol addresses this by locating one or more herbivore and/or pathogen species from the weed’s native range and introducing them so they can control the pest in its new range. These herbivores and pathogens are carefully selected and screened to determine if they will attack crops or other non-target plant species. Successful classical biocontrol programs result in permanent establishment of the control agent(s) and consequent permanent reduction in the abundance or at least the damaging impacts of the weed over all or in part of its introduced range. Classical biocontrol is not expected to eliminate the pest species completely and it often takes years or even decades after the initial release of control agents before their effects are obvious. Classical biocontrol programs may fail for a variety of reasons. Some biocontrol agents never establish, or it may take repeated releases to establish viable populations. Some biocontrol agents may become established, but then have little or no detectable impact on the targeted pest (Greathead 1995).
Some of ‘classical’ biocontrol’s greatest strengths are that once an agent is established, it will persist ‘forever’ and it may spread on its own to cover most or all of the area where the pest is present, generally with little or no additional cost. On the other hand, these strengths can become great liabilities if the agent also begins to attack desirable species (Pemberton1985; Lockwood 1993, 2000; McEvoy and Coombs 2000). Because of this, weed biocontrol researchers take pains to locate and use agents that are specific to the targeted weed and will not attack other “important” plant species. This screening process contributes to the high cost and long time required for the discovery, testing, and approval of new biological control agents.
The selection and screening of candidate classical biocontrol agents
The first systematic biological control projects for weed species began over 100 years ago, and even at that time, potential control agents were tested to make sure that they did not harm agricultural crops. Scientific and public concern for native plant species with no known economic value has increased since then, particularly in the past few decades, and weed biocontrol programs administered by Agriculture Canada and the USDA expanded their host-specificity testing protocols to address these concerns. These programs now require checks for potential impacts on native plants, particularly rare species (DeLoach 1991; Harris 1988). This is in contrast to biocontrol programs that target insects and other arthropod pests, where even today, no host-specificity testing is legally required and few projects voluntarily screen potential control agents (Strong and Pemberton 2000). It has been suggested that this situation prevails because there is little public or professional outcry for the protection of insects, with the exception of non-native honeybees, other biocontrol agents, and possibly some native butterflies.
A key part of the screening process is host-testing, wherein potential control agents are given the opportunity to feed on a variety of crop species and native plants, including those most closely related to the targeted pest. No-choice tests isolate the potential control agent with one or more native species for feeding and/or egg-laying, so that if they do not use the native(s) they will die or fail to reproduce. Other tests give the proposed biocontrol agent a choice between feeding or reproducing on the targeted pest and non-target native species. Today, proposed biocontrol agents are screened for their ability to feed and reproduce on several to many native species, but it is still impossible to test all native species. For programs targeting species such as leafy spurge (Euphorbia esula) with many native congeners (over 100 native Euphorbia spp. in the U.S.), it is not even possible to test all the native species in the same genus. In addition, the tests cannot determine whether the control agents will adapt or evolve over time so that they will become more able or willing to feed on native species. For a more detailed description of the selection and host-testing processes, and suggestions for improving them, see McEvoy (1996).
McEvoy and Coombs (2000) argue that the potential effectiveness of candidate biocontrol agents has been given too little attention in the selection process. They note that ten or more species of biocontrol agents have been released against some weeds. Since there is some risk that each species will have unintended harmful impacts, the overall risk increases with the number of species released. In addition, some relatively ineffective species may actually interfere with and lessen the impacts of species that might be effective in their absence. Therefore, McEvoy and Coombs (2000) urge biocontrol practitioners to instead strive to release the minimum number of agents required to control the weed by first identifying and releasing only those species most likely to be effective. They advocate efforts to systematically identify traits common to successful control agents and the types of insects the target weed is most likely to be vulnerable to, based on its lifecycle and physiological attributes. Similarly, Louda et al. (1997) and Nechols (2000) advocate increased consideration of the interactions a candidate biocontrol agent is likely to have, with control agents and other organisms that are already present in the system.
Use of formal risk assessment procedures, efforts to minimize the number of agents released against a given target, and requiring follow-up studies designed to assess impacts on target and non-target species in order learn how to improve later programs would answer many of the concerns of conservation biologists (Miller and Aplet 1993; Simberloff and Stiling 1994; Strong and Pemberton 2000). The USDA has recently begun requiring post-release studies on the impacts of biocontrol agents for new releases in the U.S. (DelFosse personal communication), and is also considering the use of formal risk assessment procedures. Australia already has a legislative framework that requires a formal risk assessment before releases are granted which is designed to minimize nontarget impacts (McFayden 1998; Withers et al. 2000) and New Zealand is in the process of developing protocols for assessing and balancing risks and benefits of proposed introductions (Barratt et al. 2000)
Successful classical biocontrol projects reduce the abundance or impacts of the targeted pests to acceptable levels across large areas. There have been excellent post-release studies on Klamathweed (Hypericum perforatum) and tansy ragwort (Senecio jacobaea) biocontrol agents (Holloway and Huffaker 1951; Huffaker and Kennett 1959; McEvoy 1985; McEvoy and Rudd 1993; McEvoy et al. 1990; 1991; 1993), which provide quantitative information about reductions in the abundance of the target weeds. In each case significant reductions in the density of the targeted weeds were recorded after biocontrol agents were introduced.
Impacts of the four insects released to control purple loosestrife in the U.S. and Canada have also been monitored. The leaf feeding beetles Galerucella pusilla and G. calmariensis, first introduced in 1992, have apparently reduced purple loosestrife stands at several sites already (Blossey et al., 1994; Scudder and Mayer, 1998). Results from release sites in Ontario, Michigan, and Minnesota indicate Galerucella beetles can significantly reduce above-ground abundance of purple loosestrife in as little as three years (Michigan State University, 1999). In southern Ontario, introductions of Galerucella spp. reduced above ground purple loosestrife biomass from 2,000g/m2 to less than 20g/m2 in 4 years (The Ontario Biological Control Program, 1998). Additional studies found that at high Galerucella densities (200 larvae/plant), plants were entirely stripped of all green tissue and seed production was prevented (Butterfield et al., 1996). Even at lower beetle population densities, adult and early larval feeding destroyed meristematic regions thus, preventing normal growth. Nonetheless, it is not yet clear whether this feeding is significantly reducing the root biomass of established loosestrife stands.
Unfortunately, studies of the impacts of other biocontrol agents released against weeds have been extremely rare. For example, Lym and Nelson’s recent (2000) paper on impacts of two flea beetle species released against leafy spurge is the only published study that quantifies population level impacts of any of the 13 insect biocontrol species released against this widespread pest in the U.S. and Canada. They found that both fleabeetles, Aphthona lacertosa and A. czwallinae reduced leafy spurge stem densities by about 65% up to 16 m from initial release sites within 3 to 5 years. A mixed population of both Aphthona species reduced stem densities by over 95% within 4 years after release. Establishment and rate of spread of these insects were similar regardless of the number of insects released initially. Unfortunately, qualitative before and after biocontrol release assessments of weed abundance are far more common.
Examples of weed biocontrol projects in North America that are regarded as having successfully reduced the abundance of the targeted species to acceptable levels include those to control Klamathweed (Hypericum perforatum), tansy ragwort (Senecio jacobaea), and alligatorweed (Alternanthera philoxeroides). Programs to control leafy spurge (Euphorbia esula) and purple loosestrife (Lythrum salicaria) appear to be on their way to at least partial success (Anderson et al. 2000). On the other hand, programs to control Canada thistle (Cirsium arvense), spotted and diffuse knapweed (Centaureamaculosa and C. diffusa) and yellow starthistle (Centaureasolstitialis) have not yet been successful, despite years of effort and releases of several insect species against each one.