A consideration of specific models requires that the general account of biological control provided in the introductory section is revisited to define the major
Table 1 Summary of the characteristics of major biological control approaches and typical uses of population models (see text for further detail)
Classical biological control
Inundative, augmentative, and inoculative biological control
Conservation biological control
Low numbers of agents released
Agents usually exotic
Weed and arthropod targets common
Control not confined to original release location Inexpensive when it works Risk of ongoing off-target effects Poor success rate Out of farmer's control
Large numbers of agents released Agents often native Targets usually arthropods
(some microbes) Immediacy of response, low need to plan ahead Under farmer's control
Relatively high cost Short duration of control
No agents released, existing populations are enhanced Agents mostly native Targets usually arthropods
Can give prolonged control
Under farmer's control Under-researched Poor immediacy of response, need to plan ahead
Examples of Why many releases fail to establish or modeling bring target under adequate control
Agent release patterns and numbers Which agent to release
Release rates and timing Level of control expected under differing situations Understanding reasons for failure
Which aspects of agent biology most important for success (modeling little used to date)
forms of biological control. The types of question that specific biological control models address differs to a large extent across the forms of biological control used (see Table 1).
This approach relies on the release of relatively small numbers of exotic agent individuals to a new location. Typically, release sizes are in hundreds or thousands per site. Accordingly, there is an expectation that the agents will reproduce in number, establish a self-perpetuating population (or metapopulation), and spread from the original release positions to cover all or most of the target species' range. The major risk associated with this form of biological control is that the introduction of an exotic species into a new geographical location will damage other, nontarget species. The most infamous case of this was the 1930s introduction of the cane toad, Bufo marinus (Linneus), into Australia in an attempt to control the cane beetle. Given the magnitude of this issue, it is possibly surprising that models have not been used to any great extent to help predict the risk of nontarget impacts in classical biological control, especially since modeling has been used to predict risks associated with inundative biological control (see below). The explanation for this is likely to be that the consequences of introducing an exotic species that proves not to be target specific may be so catastrophic that decisions are made on the basis of empirical rather than theoretical or modeling work.
Indeed, experimental specificity testing is a major subdiscipline in this branch of biological control.
Where models have been extensively used in classical biological control is to address the historically low success rate. In the case of arthropod agents against arthropod targets, ~10% of releases give complete control. Understanding reasons for this is of obvious importance and modeling-based efforts have sometimes employed high levels of biological and environmental detail. Work by Gutierrez and co-workers focused on the parasitic wasps released into Africa in classical biological control attempts to control the cassava mealybug (Phenacoccus manihoti Mat.-Ferr.). Modeling helped explain why Epidinocarsis diversicornis (Howard) failed to establish while the related species Epidinocarsis lopezi (De Santis) established and became the most important mortality factor in the pest's population (at least during the dry season). A contrasting example, of a herbivore agent against a plant target, illustrates another way in which modeling may support decision making to enhance the success of classical biological control. In the case of knapweed (Centaurea diffusa Lam.), the biological model was far simpler than the multiple-parameter, differential equation model used in the preceding example. Here, a simple difference equation model provided an indication of the type of agent that needed to be introduced as a follow-up to the initial introduction of several insect herbivores. Since the weed was increasingly resistant to reductions in density as its numbers declined (i.e., a density-dependent response), the new population equilibrium remained above the EIL that applied to the weed's impact on forage production. Accordingly, additional agents that specialize on the fast-growing, prostrate plants typical of low-density infestations were identified as important.
Related and other issues that specific models can address in classical biological control include: (1) whether a specific exotic agent will establish in a new location; (2) what number of individuals should be released per site; (3) what level of control a specific exotic agent will exert over a target; and (4) understanding why a release has failed.
Augmentative, Inoculative, and Inundative Biological Control
These three biological control approaches are considered together since they all are based on the release of relatively large numbers of agents in a specific time and place with the expectation of local control on a relatively short-term basis. Augmentative biological control specifically aims to augment the action of other agents already present. Inoculative biological control typically occurs in protected agriculture where the aim is to establish a breeding population of the agent that will persist for the duration of a specific cropping phase. Inundative biological control is effectively a 'biological pesticide' used in response to escalating pest numbers at a specific time and place though the agent may be an arthropod or a microorganism.
A problem associated with these forms of biological control is the potentially prohibitive cost of using large numbers of short-lived agents. Accordingly, a number of models have considered the level of biological control achieved at differing levels of agent release/application. In the case of work with a granulosis virus against the codling moth (Cydia pomonella Linneus), larval mortality was shown to vary with 1/10 power to the virus concentration with a consequent effect on the level of fruit damage.
These biological control approaches are not, however, without risk, and modeling has also been used to address this issue. A fungal pathogen Chondrostereum purpureum was proposed as an inundative biological control agent for the perennial weed Prunus serotina Ehrh. in forests. Though the pathogen could be applied in a targeted fashion to only the pest trees by formulating it into a mycoherbicide spray, the fungus subsequently produces basidiospores that could disperse and infect nontarget plants, necessitating an analysis of risk. This involved the development of a stratified model to describe spore fluxes for differing layers and showed the significant effect of wind speed and consequent risk of dispersal of inoculum from a treated patch of forest and into commercially valuable Prunus spp. crops.
Other issues that specific models can address for these forms of biological control include: (1) what level of control a specific agent will exert over a target, (2) optimal timing of application or release for the agent to exert maximum impact, and (3) understanding why an agent fails.
Though this form of biological control has its roots in traditional practices such as companion planting and other types of polyculture, it is only in recent decades that it has been rigorously researched. This approach aims to maximize the impact of existing natural enemies rather than accept the costs and risks of introducing and releasing exogenous agents. Reducing the pesticide-induced mortality of natural enemies and habitat manipulation to improve the local availability of resources such as food, shelter, and alternative hosts are increasingly recognized as important. However, reflecting the relative youth of this branch of biological control, there has been very little use of modeling to improve its performance. One important study by Kean and co-workers illustrates the potential value of wider use of modeling in conservation biological control. That work explored which aspects of natural enemy biology had the greatest impact on the target pest population. Modeling indicated that enemy search rate and prey conversion efficiency were the most important parameters. Maximum consumption rate and fecundity were less important while the effect of longevity depended on its interaction with other factors. Finally, the degree of spatial attraction of natural enemies to a location had an almost linear effect on pest suppression. The latter finding is of use for practitioners of conservation biological control in ephemeral habitats (e.g., annual crops), because attracting higher densities of enemies has an immediate effect on pests while other natural enemy parameters - though potentially more important - may be slower to respond and hence are likely to be more appropriate for manipulation in perennial systems such as orchards. At a finer scale, the importance of natural enemy search rate highlighted by the modeling points to the possible advantages that could be gained from conservation biological control methods that targeted this parameter of behavior. An example of this is to provide parasitoid wasp agents with nectar sources since many experimental studies have shown that this increases energy levels and flight propensity, which could, in turn, increase search rate.
Was this article helpful?