The Strength of Biological Control in the Battle against Invasive Pests: a Reply MARK S. HODDLE Department of Entomology, University of California–Riverside, Riverside, CA 92521, U.S.A., email [email protected] Nontarget Impacts of Biological Control habitat fidelity and exploit a wide range of hosts in a vari- ety of habitats are more likely to cause unwanted collat- Louda and Stiling (2004 [this issue]) state that biological eral damage to nontarget species either directly through control is not a simple matter of community re-assemblage direct attack or indirectly through apparent competition because deliberate introductions of exotic natural ene- because of reticulate food-web linkages. mies for suppression of exotic pest populations can have Two plant-feeding insects (Rhinocyllus conicus [inten- reticulate impacts that are difficult, if not impossible, tionally released; Gassmann & Louda 2001] and Larinus to predict a priori. In an insightful retrospective study, planus [an accidental arrival in the United States that was Hawkins et al. (1999) analyzed 68 life-table studies of eliminated as a potential biological control agent because native insects and introduced insect pests to determine of broad host breadth on Carduinae thistles in its home whether biological control is analogous to naturally oc- range; Louda & O’Brien 2002]) mentioned by Louda and curring control (i.e., the action of native natural enemies Stiling as having “unexpected levels of nontarget feeding” on native hosts). Hawkins et al. (1999) show that suc- were anticipated from host-specificity tests because both cessful biological control programs result in less reticulate weevils were known to feed and reproduce on a variety trophic relationships than those seen in natural food webs of thistle species in their home and introduced range. of native insects. The most successful biological control As Louda and Stiling point out, both insects are “thistle programs are those that do not have “natural” food-web specialists,” but thistles are a speciose group with repre- structures because biological control food webs consist sentatives in genera that occur in Europe and North Amer- of short, linear food chains that are devoid of complex ica, and these weevils, due to their broad dietary breadth reticulate trophic interactions. within the thistle group, have greater numbers of signif- This result occurs because biological control systems icant food-web linkages than desirable, thereby making often consist of exotic species that share few ecologi- significant nontarget impacts more likely. The real issue cal or evolutionary links with native biota. Furthermore, here concerns legislation (i.e., redistribution of L. planus control is enhanced in simplified habitats characteristic and R. conicus) and changing social values (i.e., what con- of agroecosystems and, arguably, native systems invaded stituted acceptable damage to nontarget species, and the by exotic plants because both often consist of vast mono- value of native flora and fauna in the 1960s versus 2003), typic stands of exotic vegetation. Host-specific natural en- as opposed to an inherent flaw in the theoretical prin- emies that cause population declines of the target pest ciples underlying concepts of host specificity and host are themselves subject to density-dependent population range assessment as they pertain to biological control. regulation as the biological control agent’s food source is The moth Cactoblastis cactorum, a native of Argentina depleted, and they are unable to adequately exploit other and “poster child” of biological control (Stiling 2002) has hosts in the environment to maintain high population invaded the continental United States from the Caribbean densities because they lack significant trophic linkages to and is attacking native cactii. Louda and Stiling consider other hosts. The more specialized the natural enemy the C. cactorum a “specialist” on the cactus genus Opuntia,a less likely it is to infiltrate native communities and attack group with approximately 200 species in the new world nontarget native species (Hennemen & Memmott 2001). but no representatives in Australia, making C. cactorum Generalist natural enemies that have low levels of host and a specialist on this continent when it was released in Paper submitted April 15, 2003; revised manuscript accepted May 23, 2003. 61 Conservation Biology, Pages 61–64 Volume 18, No. 1, February 2003 62 Biological Control of Invasive Pests Hoddle 1926 for control of weedy Opuntia spp. Following its where. Additionally, mice were lethally sampled and their unwanted incursion, it is not surprising that C. cactorum gut contents examined for remains of fly pupae, and the can feed on native Opuntia spp. in North America, as authors concluded that factors other than food limited these plants are within its known host range. This moth the breeding success of mice over this 2-year study period. was not intentionally released in North America for con- Furthermore, declines in native mammal populations and trol of pestiferous cacti, and should not be classified as native plant communities were not demonstrated in Pear- a biological control agent in this instance. It is probable son et al.’s (2000) study, as claimed by Louda and Stiling. that C. cactorum would not be considered in the current The suggestion by Pearson et al. (2000) that mouse debate if it had not been used as a biological control agent populations can be subsidized and perhaps enhanced by in other countries. preying on abundant but ineffective natural enemies is an Louda and Stiling suggest that adverse effects aris- interesting idea that may merit further work. However, to ing from migratory species like C. cactorum could be accurately address the importance of resource provision- reduced by assessing potential ecological risks associ- ing and the impact it has on native communities would ated with dispersal of natural enemies. This is not a require many consecutive years (>10) of nondestructive novel suggestion, and such measures have already been sampling over multiple, widely separated sites. The result- taken in North America. The Technical Advisory Group ing time-series data on knapweed, fly, and mouse densi- (TAG) consists of representatives from the United States, ties would need to be subjected to sophisticated statistical Canada, and Mexico who assess risk posed by proposed analysis to determine relationships (Elkinton et al. 1996). weed biological control agents and their propensity to The perturbation of ecosystems by ineffective natural cross international borders to threaten nontarget species enemies that form reticulate trophic relationships with (CoFrancesco 1998). The Food and Agricultural Organiza- nontarget species is the most salient point raised by Louda tion (FAO) (1997) and North American Plant Protection and Stiling. Indirect effects mediated through food webs Organization (2000, 2001) provide similar guidelines for are the least recognized, understood, and appreciated. assessing the risk posed by movement of entomophagous Potential impacts via trophic spectra analyses will be dif- and phytophagous natural enemies across international ficult to quantify accurately because long-term financial borders. sponsorship would be needed to undertake these types Virtually all risks posed to native plants by exotic nat- of studies. ural enemies as cited by Louda and Stiling are to those species closely related to target weeds, a fact well recog- nized by biological control scientists as a central tenet in host-range evaluation. Of those natural enemies ex- Recent Studies of Nontarget Populations and ploiting native plants, <1% released for weed control use Evolutionary Change of Natural-Enemy Host Range a native plant unrelated to the target weed (Pemberton 2000). To protect native flora, weed targets should have The concern over nontarget issues raised by prominent few or no native congeners, and only natural enemies ecologists has motivated several recent studies to deter- with narrow host breadths should be considered (Pem- mine whether exotic parasitoids have been responsible berton 2000). For insect natural enemies, in particular for the perceived population declines of native species, parasitoids, the ability to predict risk to nontarget natives as demonstrated by Boettner et al. (2000). Possible pop- is not clear because ecological and biological correlates ulation and range reduction of the native butterfly Pieris have not been identified. This may be due in part to the virginiensis in the northeastern United States has been poor quality of the data sets that are available for retro- attributed to the exotic parasitoids Cotesia glomerata spective analyses (Hawkins & Marino 1997). and C. rebecula released (in 1884 and in the 1960s, re- Louda and Stiling misrepresent Pearson et al.’s (2000) spectively) for biological control of P. rapae. Benson et work on the effect of abundant Urophora spp. larvae (a al. (2003) conclude that P. virginiensis is not threatened tephrid fly released for the control of the spotted knap- by these parasitoids even though its larvae are attacked weed [Centaurea maculosa], a serious weed in native successfully in the laboratory by these parasitoids. In U.S. rangelands) on the population growth of deer mouse New Zealand, the polyphagous pteromalid Pteromalus (Peromyscus maniculatus). Pearson et al. (2000) did not puparum (released 1932–1933), introduced for the con- demonstrate a two- to three-fold increase in mouse popu-