INVASIVE EXOTIC SPECIES ARE A MAJOR ECOLOGICAL PROBLEM in many of the world's ecosystems (Sala et al. 2000). With some notable exceptions, however, most tropical moist forests appear to be resistant to the impacts of invasive exotic plants (Rejmainek 1996, Denslow & DeWalt in press). Sala et al. (2000), for example, identified land use change, but not biotic exchange, as the major threat to tropical ecosystems. However, the impacts of invasive exotic plants are a leading threat to native ecosystems on oceanic islands, where naturalized exotic plants have doubled the sizes of floras (Sax et al. 2002). On tropical islands, exotic species invade intact forests, alter successional trajectories, impede restoration of endangered plant communities, and contribute to the degradation of fragmented ecosystems (Denslow & DeWalt in press). These impacts are of special concern because tropical islands are repositories of great biodiversity and are characterized by high levels of endemism and large numbers of threatened and endangered species. Furthermore, continental tropical forests that are fragmented or otherwise disturbed also are vulnerable to the impacts of exotic invaders (e.g., Humphries et al. 1992), and control of invasive exotic plants is a major cost for many land managers. Unfortunately, the options available for management of established invasive species are few and costly and carry risk of unintended consequences. A tool with great promise is the introduction of exotic hostspecific arthropods or pathogens to reduce the abundance, rate of spread, and/or habitat distribution of the target plant in its invasive range. Classical biological control is based on the premise that the population explosions of some exotic species in their introduced ranges are due in part to their escape from their native herbivoresthe ecological release hypothesis (Keane & Crawley 2002, DeWalt etal. 2004, Levine etal. 2004). Historically, the primary objective of weed biocontrol has been the direct reduction in density, cover, and range of target weeds that suppress desirable forage species, are toxic or unpalatable to livestock or that impede waterways. Most of this experience has developed in nonforest ecosystems (Coombs et al. 2004) and at temperate as opposed to tropical latitudes; among weed genera targeted by biocontrol worldwide (except Australia which has both tropical and nontropical ecosystems), 51 percent have been exclusively nontropical and only 25 percent exclusively tropical (Julien & Griffiths 1998). In some cases, this approach has met with dramatic success, e.g., St. John's wort: Hypericumperforatum L. (Harris et al. 1969), leafy spurge: Euphorbia esula L. (Harris 1993), various cactus species: Opuntia spp. (e.g., Moran & Zimmermann 1991), water hyacinth: Eichornia crassipes (Martius) Solms-Laubach (Center et al. 1999b), and giant salvinia: Salvinia molesta D. S. Mitchell (Room et al. 1981). The use of introduced arthopods and pathogens to control exotic wildland weeds is more recent but is showing promise in the control of paperbark: Melaleuca quinquenervia (Cav.) Blake in the Florida Everglades (Center et al. 2000), purple loosestrife: Lythrum salicaria L. in the northern (U.S.) wetlands (Blossey 2002), saltcedar: Tamarix spp. along western (U.S.) stream courses (DeLoach & Carruthers 2004), and several Acacia species in South Africa (Impson et al. 2000). Concern has been expressed over the use of biocontrol agents because of their potential to cause'both direct and indirect nontarget impacts (Simberloff & Stiling 1996; Louda et al. 1997, 2003; McEvoy & Coombs 2000). Nontarget impacts of introduced biocontrol agents for weeds can take the form of direct feeding by the agent on nontarget plant species (Pemberton 2000) or indirect interactions such as competition with native herbivores (Louda et al. 1997) and alteration of the community trophic structure (Henneman & Memmott 2001). Recent reviews provide evidence that the risk of direct nontarget impacts is greatest in closely related species and that the agents released following modern, widely accepted host-range testing protocols and review processes rarely affect nontarget species (Pemberton 2000). However, only a small proportion of agents released are effective at reducing the target species, and our understanding of indirect interactions of introduced biocontrol agents is rudimentary (McEvoy & Coombs 2000). As our appreciation grows of the harm caused by invasive plants in native ecosystems, the use of biological control agents in wildlands becomes increasingly attractive. Extensive use of chemical and mechanical control methods in native ecosystems also can have substantial secondary effects, including impacts on nontarget species and ecosystem processes. In addition to their high host specificity, biocontrol agents can be effective over large areas of poorly accessible terrain. In many tropical countries, where available resources-both financial and human-for the long-term control of invasive species in natural areas are small, cost-effectiveness makes biological control an attractive conservation tool. 1 Received 3 March 2005; revision accepted 17 June 2005. 2 Corresponding author; e-mail: jdenslow@fs.fed.us
Read full abstract