Memoirs of the Museum of Victoria 56(2):6 11-615(1 997) 28 February 1997 https://doi.org/10.24199/j.mmv.1997.56.58 THE 'SAFETY' OF BIOLOGICAL CONTROL AGENTS: ASSESSING THEIR IMPACT ON BENEFICIAL AND OTHER NON-TARGET HOSTS

D.P.A. Sands

CSIRO, Division of Entomology, Private Bag No. 3, Indooroopilly, Qld 4068, Australia

Abstract

agents: assessing their impact on Sands, D.P.A. , 1997. The 'safety' of biological control 1-615. beneficial and other non-target hosts. Memoirs of the Museum of Victoria 56(2): 61 Biological control agents considered for introduction into Australia must demonstrate a In high degree of specificity before approval is given for their release from quarantine. assessing -safety', potential agents for weeds are tested to determine if they can damage crops, ornamentals or native plants and for pests, if they can adversely affect on beneficial (including biological control agents) or have a detrimental impact monospecific in native . Effective arthropod biological control agents are rarely of plant or their native range, most being adapted to a group of closely related species potential host range arthropod hosts. When introduced for control of exotic organisms, the taxa closely related to the of agents is usually more restricted due to absence or scarcity of potentially valuable target. While such agents utilise a target species as preferred host, some native species. Host pref- agents will on occasions reproduce on non-target taxa including a laboratory, a problem more erence in agents is difficult to quantify in the confines of often favours frequently encountered in arthropod than weed programs. Confinement the results of such tests attack on a wider range of organisms than under field conditions and where non-target may influence a decision whether to or not, release an organism. In cases agents in quarantine, the benefits ot development is demonstrated by narrowly specific detrimental affects to native biological control of pests must be weighed against any real non-target taxa but no exotic species In Australia some agents occasionally develop on species. Most threats arthropod agents are known to seriously damage populations of native or extinctions quoted overseas are poorly justified.

control agents), or Introduction (including other biological have an significant impact on native arthropod introduced for biological control of pests, exotic agents When spec j es Before release, tests with with broad host ranges (e.g., their several predators are conci ucte(j in quarantine to determine influ- cane toad, snails) have also detrimentally potenUa i host range and to assess the signifi- organisms. They enced beneficial and native canceof any development on non-target hosts or while their often failed to control the target pests prey impact on non-target organisms has led to some Procedures for testing the specificity of agents uninformed distrust in biological control pro- weeds have been widely practiced grams The success and environmental benefits for (Wapshere, 1 974) but tests on agents for arthro- from introductions of narrowly specific exotic public pod pests have not received the same attention aeents are often overlooked with more generalist (Sands and Papacek, 1993). In most countries attention RWen °o the impact by a™° and New Zealand, host Z ZuZ snecies other than Australia ptz^^oT:::Lv^ „**« *» ««*«» ^ «* «*< <****£• pod programs. rFor example in the United States Australia esoecially vertebrates, are not now are only tested if thought likely to attack When biolog cal control agents are agents ™ved organisms (ErOe 1993). Recently, SS&foHntrodurtion, tests must confirm beneficial (Waterhouse, 1991 Anon, 1995) for of host specifictty before per- guidelines a high deg ee have mSn.sgivenbyrelevantauthoritiesfortheir conducting biological control projects acceptable included host specificity procedures for testing release foS quarantine. In assessing to agents for both weed and arthropod targets, evelsofspTchicity, potential agents are tested damage or Though there are no known examples for any determ n fo weeds, if they can undesirable impact on beneficial organisms, crops ornamentals or native reproduce on agents are likely nati ve flora or fauna by narrowly-specific and for arthropod pests, if they nlams practitioners are focusing beneficial insects (Waterhouse, 1991), fo mfluence the aKance of

611 612 D.P.A. SANDS

on developing and refining the methods for tes- pheromones, host size etc.) can be used to guide ting exotic agents with non-target and native the design of tests. invertebrates prior to their release in biological Few agents in arthropod biological control control programs. programs are monophagous. Most are oligopha- gous (e.g., develop on other hosts in the same Selecting non-target and native species for tribe as the target) in their native range. How- testing with exotic agents ever, they may utilise only the target species or one or two others very closely related to the When selecting non-target and native species for target, when introduced to a country where the host specificity testing with exotic agents, target is a pest. Wapshere's (1974) centrifugal (phylogenetic) approach is equally valid for weeds and arthro- Some problems with determining host pod programs. Both exotic beneficial and native specificity of exotic agents species are selected based on their taxonomic relationships with a target organism. For weeds The methods for determining the host speci- projects, once plants related to the target are ficity of potential biological control agents for selected, potted plants can be used in host speci- weeds are relatively well established and prac- ficity tests. However, there are serious con- ticed (e.g., Harley and Forno, 1 992; Waterhouse, straints with native arthropods since collecting 1991; Anon, 1995). However, the methods for and maintaining living species for testing with testing agents for arthropod targets are more agents may be difficult. The biologies of native varied and are still being developed. These are arthropods related to a target are frequently needed to reflect the different means by which unknown, preventing culture of appropriate agents locate, oviposit and develop on/in stages required for tests. Where 100 or more arthropod hosts when compared with agents non-target potted plant species are often tested interacting with plant hosts. with agents in weeds programs the testing of The methods for rearing agents on their host more than 10 species of non-target arthropods may not be appropriate for host-specificity tests may be impractical and is often unnecessary. with non-target taxa. Small containers can often Moreover, the anomalies experienced when tes- be used for agents to induce oviposition, feeding ting agents in the confinement of cages and or development in arthropod hosts. However, insectaries tend to occur more frequently in restricted space often leads to an inaccurate arthropod than in weed programs (Sands, assessment of specificity by disrupting host 1993). recognition and acceptance (Sands and Papacek, Information on the taxa related to a target 1993). When confined, natural enemies of species in its native range is most useful when arthropod pests may oviposit on organisms that selecting a centrifugal range of species related to do not support their development or they may a target pest. However, if tribes are not desig- complete development on hosts or prey not

nated it may be difficult to associate genera or attacked in the field. For example, in Papua New species in groups according to their taxonomic Guinea the egg parasitoid, Ooencyrtus erionotae relationships. Beneficial organisms related Ferrier, is believed to be narrowly-specific to (within the same family) to the target pest should banana , Erionota thrax (Linn.), since it be included when compiling a list of non-target has never been reared from other lepidopteran species for host specificity tests. When con- eggs, even when eggs of the related hesperiid, sidered for release, development of a biological Cephrenes mosleyi (Butler) are deposited less control agent on another beneficial agent is gen- than a metre away on palms. However, its host erally unacceptable. When testing agents on specificity could not be demonstrated accurately native species, selected taxa closely related (fam- in the restricted space of the laboratory since ily, subfamily, tribe or genus) to the target, or in eggs of C. mosleyi and other species readily certain cases, those morphologically similar, are attracted parasitoid oviposition and supported often sufficient to provide adequate information its development (Sands et al., 1991). on the host specificity of an agent, rather than Choice tests with a target host and test species testing extensive lists of species of distantly- caged together may lead to misleading results. related taxa. Additional species can be included Field and Darby (1991) found that the para- ifjustified by the initial test series. The methods sitoid, Sphecophaga vesparum (Curtis) by which taxa related to an agent select their (Hymenoptera: Ichneumonidae), for biological hosts or prey (e.g., certain taxonomic groups, control of the European wasps, Vespula ger- THE 'SAFETY' OF BIOLOGICAL CONTROL AGENTS 613

manica (Fabricius) and V. vulgaris (Lin- is no evidence in Australia for a decline in the naeus)(Hymenoptera: Vespidae), oviposited in abundance of native species which occasionally and developed sparingly on two native wasps, act as hosts to this parasitoid (Gross pers. Ropalidia plebeina Richards and Polistes comm.).

humitis (F.)(Hymenoptera: Vespidae), when Nafus ( 1 993) when studying the natural enem- exposed in presence of V. germanka. However, ies of the butterflies Hypolimnas spp. in Guam, the parasitoids failed to oviposit when exposed found no detrimental impact by biological con- separately to the native species. This attack on trol agents even though introduced parasitoids non-target hosts may have been stimulated by frequently attacked them and one had become the close proximity of V. germanica providing the most significant mortality factor for pupae of the necessary kairomones or other ovipositional a native species. For an exotic agent some devel- cues. This and other studies indicate that choice opment on indigenous fauna or flora may be tests with arthropod target and non-target acceptable, provided that the benefit gained by species exposed to an agent at the same time, controlling a pest outweighs any slight risks of should be interpreted with caution and led to a effects on the abundance of indigenous species. recommendation that choice tests are better The advantage of an oligophagous natural avoided when alternative methods are available enemy was described as a 'lying-in-wait' strategy (Sands and Papacek, 1993). and contrasted with a 'search-and-destroy' strat- Plant material which is substrate to a host may egy of host-specific species (Murdoch et al., be required to stimulate host recognition by an 1985). Monophagous agents may sometimes be agent. Plant kairomones may also influence the considered to be at a disadvantage in an exotic behaviour of natural enemies of arthropods and range since dispersal between pest infestations need to be considered when designing the host can be restricted by a lack of supporting hosts. specificity procedures (Anon, 1995). Olfactom- Every case requires careful assessment based on eters can sometimes be used to contrast an the results of carefully-planned research. agents' response to plant volatiles with their The following criteria may be useful when response to the host alone. Some cage materials assessing acceptable host specificity of an exotic may affect scatter of light entering cages and agent for release: those made of black materials may be necessary 1. Exotic agents are acceptable if narrowly- to stimulate mating or oviposition by certain specific in their native range and shown by natural enemies of arthropods. Several parasi- tests with related beneficial or native species, specific to the target pest in the new toids, e.g., some Tachinidae, require sunlight to be before mating will occur. environment. 2. If an agent completes development in/on any Assessing the impact of agents developing on non-target (beneficial/native) organisms, a native hosts decision must be made as to whether this may have any detrimental effects. information is available on native, non- Some 3. The ability of a narrowly specific agent to agents but very target hosts utilised by exotic develop in/on non-target organisms should their influence little information is available on not automatically preclude a recommen- host populations. For on the density of these dation for release. parasitoid, Trissolcus basalts example, the egg 4. Development in/on some non-target organ- control of (Wollaston), contributed to biological isms may be acceptable provided the host in southern Australia N. viridula in most crops range has been shown to be narrow (i.e., con- from Egypt in after various imports, the first fined to a small group of organisms related to Without parasitisation of 1933 (Wilson. 1960). the target), and provided that the non-target viridula would N. viridula by T. basalis. N. organisms are not preferred to the target. much more serious pest in undoubtedly be a 5. Development in/on some non-target taxa is by no Australia. However, this parasitoid may sometimes be beneficial — when non- where it has means host-specific in Australia target hosts provide a medium for trans- range of native Hete- been reared from eggs of a mission to crop sites (for arthropod pests). roptera, including more than 10 species of Pen- Norris, 1987). tatomidae (Waterhouse and Claims of rarity and extinction induced by provide a While alternative hosts for T. basalis exotic agents reservoir for parasitoid in the absence of eggs of there Howarth (e.g., 1985, 1991) has suggested that N. viridula (Waterhouse and Norris, 1987), 614 D.P.A. SANDS

similar to an many extinctions have followed introductions to Fiji since it behaved in a way native of biological control agents into Hawaii. How- exotic species. L. iridescens had almost no introduction of B. ever, according to Funaski et al. (1988), only one parasitoids prior to the of 30 biological control agents introduced into remota and after its first appearance in pre- plantations, spread Hawaii in the last 1 5 years has been found to viously unaffected the eastern, attack native or beneficial species. Cullen ( 1 989) rapidly from Viti Levu towards has suggested that when agents are introduced copra-producing islands (Paine, 1994). for biological control of weeds, their impact on As those entomologists reported, most of the plants will be limited to very abundant species if zygaenid moths closely-related to L. iridescens the agent is not entirely host specific for the weed occur in countries from New Guinea to Malay- and that an agent will not affect those already sia. Though evidence suggests that L. iridescens held in equilibrium by natural enemies, includ- was most likely exotic in Fiji, its presence else- ing sparsely distributed or endangered species. where remains unknown — unless Kalshoven's L. The same is likely for arthropod targets. (1981) unlikely claim that A. catoxantha and Biological control of the moth, Levuana irl- iridescens are con-specific proves to be correct. descens Bethune-Baker (Zygaenidae) in Fiji, is Whether or not L. iridescens is an exotic species often quoted as an example of extinction of a in Fiji, is now extinct, or still survives in rain- native following introduction of an exotic forest on native palms, this is not an appropriate agent. Control of this coconut pest was achieved example of extinction of a native species caused in 1925 following introduction of the tachinid, by introduction of an exotic biological control Bessa remota (Aldrich) (Diptera: Tachinidae), agent. B. remota would not now qualify as suit- reared from a related Malaysian moth, Cathar- able as a biological control agent since it is not tona (Amuria) catoxantha (Hampson) (Zygaeni- sufficiently specific even to the target family dae) (Tothill et al., 1930). Howarth (1991) (Zygaenidae) (Waterhouse and Norris, 1987). suggested that the last authentic specimen of L.

in 1 929 (with possible iridescens was collected Discussion survival of species to 1 940s). However, there is a specimen in the Koronivia Research Ministry of Improving existing methods and developing Agriculture, Fisheries and Forests Station in techniques for assessing the host specificity of Fiji, collected in August 1941. agents in arthropod biological control programs Paine (1994) reported an outbreak of L. iri- are high priorities for research in Australia. This descens (accompanied by B. remota) on is especially relevant following recent fears over- near Vunindawa, Fiji, in 1956. Paine's identifi- seas relating to threats of extinction of native cation is likely to be accurate since he was fam- invertebrates. A better understanding of the iliar with the moth and its parasitoid. He had methods by which natural enemies select their been associated with Tothill during the biologi- hosts is also needed if host specificity tests are to cal control program on L. iridescens. In view of be meaningful. The criteria used to evaluate the rapid control of L. iridescens, its rarity by agents for arthropod pests differ considerably late 1920s and later survival until 1956 (even from those used for weeds (Goldson and Phil- causing outbreaks), it is probable that L. iri- lips, 1990), for example results of host choice descens has not become extinct. Another likely tests on agents with arthropods which as shown cause for rarity of this moth is change in its orig- by Field and Darby (1991), may be misleading. inal habitat. Coconut plantations have displaced In addition, foraging behaviour of parasitoids is native palms on most of the lowlands in Fiji, important (Lewis et al., 1990) but is difficult to leaving little indigenous habitat for the moth to test in the laboratory. survive on native hosts. The range of cues used by each natural enemy The closely-related Amuria catoxantha to recognise their hosts differ greatly between (Hampson) from which B. remota was collected, species. In the past, the influence of plant kairo- is extremely rare or absent from coconuts mones has not been considered fully when tes- between outbreaks in Java (Kalshoven, 1981). A. ting host acceptance by parasitoids. Plants may catoxantha is said to be confined to native palms be important in stimulating host acceptance or in rainforest but occasionally causes outbreaks rejection by parasitoids. The design and size of in coconuts when transported by wind (Kal- cages used for host specificity tests need to be shoven, 1981). These are phenomena quite poss- evaluated for each taxonomic group of agents. It ible for L. iridescens in Fiji. L. iridescens was not may be necessary to ensure that containers are considered by Tothill et al. (1930), to be native not re-used or ensure that they are free of THE 'SAFETY' OF BIOLOGICAL CONTROL AGENTS 615 residual kairomones or pheromones from a host germanica (Fabricius) and V. vulgaris (Linnaeus) or its plant substrate before non-target organ- (Hvmenoptera: Vespidae) in Australia. New Zeal- 193-197. isms are tested. and Journal of Zoology 18: Funaski, G.L, Lai Po-Yung. Nakahara, L.M., Beards- Fears that an agent might change its host range ley. J.W. and Ota, A.K., 1988. A review of bio- and attack other target taxa after introduction logical control introductions in Hawaii: 1890 to are sometimes expressed. However, there are no 1985. Proceedings of the Hawaiin Entomological recorded examples of monophagous or narrowly Societv 28: 105-160. oligophagous agents changing their host range to Goldson, S.L. and Phillips, C.B.,1990. Biological con- cause damage to beneficials, native plants or trol in pasture and lucerne and the requirements non-target insects (Waterhouse, 1991). Agents for further responsible introduction of entomo- should not be automatically excluded from phagous insects. Bulletin of the Entomological Societv ofNew Zealand 10: 63-74. further consideration when confinement is sus- Harley, K..L.S. and Forno, I.W., 1992. Biological con- pected of causing anomalous results during host trol of weeds. A handbook for practitioners and specificity testing. When all laboratory tests are students. Inkata: Melbourne. 74 pp. inconclusive, it may still be possible to deter- Howarth,F.G., 1985. Impacts of alien land arthropods mine accurately the host range of a potentially and molluscs on native plants and animais in valuable agent by studies in its native range. In Hawai'i. Pp. 149-179 in: Stone, C.P. and Scott, addition, information from the native range J.M. (eds), Hawaii's terrestrial ecosystems: pres- Hawaii: may be useful in determining the adaptation of ervation and management. University of Honolulu. an agent to a particular habitat, e.g., agricultural Howarth, F.G., 1991. Environmental impacts of land and grassland vs rainforest. Very few classical biological control. Annual Review of species are adapted to both environmental situ- Entomology 36: 485-509. ations. Kalshoven, L.G.E., 1981. Pests of crops in Indonesia. In every example where biological control of a Revised and translated by P.A. Van der Laan and weed or arthropod pest is considered the criteria G.H.L. Rothschild. P.T. Ichtiar Baru-Van Hoese: for safety may sometimes differ, based on the Jakarta. 701 pp. results of host specificity tests, either under Murdoch, W.W., Chesson, J. and Chesson, P.L., 1 985. practice. Amer- quarantine conditions alone or by taking into Biological control in theory and 125: 344-366. account information from the agent's indigen- ican Naturalist Nafus, D.M. , 1993. Movement of introduced biologi- ous range. There is an urgent need to examine cal control agents onto nontarget butterflies, interaction of exotic agents introduced for the Hypolymnas spp. (: Nymphalidae). programs, for past biological control arthropod Environmental Entomology 22: 265-272. example, the development of T, basalis on Lewis, W.J., Vet, L.E.M., Tumlinson. J.H., van Len- native Heteroptera. Protocols for assessing teren, J.C. and Papaj, D.R., 1990. Variations in exotic imports of agents should be developed for parasitoid foraging behaviour: essential element arthropods taking account of the often different of a sound biological control theory. Environmen- Entomology 19: 1183-11 93. criteria used in weeds programs. tal Paine, R.W., 1994. Recollections of a Pacific entomo- logist 1925-1966. Australian Centre for Inter- References national Agricultural Research: Canberra. code conduct for the import Sands, D.P.A. and Papacek, D. F., 1993. Specificity Anon , 1995. The FAO of P. 495 and release of biological control agents. Draft requirements for biological control agents. Dall, D.J. Milne. W.M. (eds), (3.1.95) technical guidelines. International Insti- in: Corey, S.A., and sustainable agriculture. CSIRO. tute of Biological Control. Pest control and Canberra. Cullen, J.M., 1989. Current problems in host-speci- Division of Entomology:

(ed.). , 1993. Effects of confinement on para- ficity screening. Pp. 27-36 in: Delfosse, E.S. Sands, D.P.A. Proceedings of VII International Symposium on sitoid/host interactions: interpretation and control of arthropod Biological Control of Weeds, 6-11 March 1988. assessment for biological 196- 199 in Corey, S.A., Dall, D.J. and Rome. Italy. pests. Pp. what the W.M. (edsj, Pest control and sustainable Ertle. L.R., 1993- What quarantine does and Milne, Division of Entomology: collector needs to know. Pp. 53-65 in: Van agriculture. CSIRO. Driesche, R.G and Bellows, T.S (eds). Steps in Canberra. D.P.A., Sands, M.C. and Arura, M., 1991. classical arthropod biological control. Entomologi- Sands, skipper. Erionota thra.x (L.) Lepidoptera: cal Societv of America, Maryland, USA. Banana specificity of Hespcriidae) in Papua New Guinea: a new pest in Field, R.P. and Darby, S.M., 1991. Host (Curtus) South Pacific region. Micronesica Supplement the parasitoid, Sphecophaga vesparum the (Hymenoptera: Ichneumonidae), a potential bio- 3: 95-100. wa.«/w, Vespula Taylor. H.C. and Paine, R.W., 1930. The logical control agent of the social TothilfJ.D., 616 D.I'.A. SANDS

coconut moth in Fiji, A history of its control by Waterhouse, D.F. and Norris, K.R., 1987. Biological means of parasites. Imperial Bureau of Entomo- control: Pacific prospects. Inkata,: Melbourne. 454 logy: London. pp.

Wapshere, A..)., 1 974. A strategy for evaluating the Wilson, I-., 1960. A review of the biological control of safety of organisms lor biological weed control. insects and weeds in Australia and Australian New

Annals ofApplied Biology 11: 201-21 1 Guinea. Commonwealth Agricultural Bureau: 1991. Waterhouse, D.F. , Guidelinesfor biological con- England. 102 pp. trol in lite Pacific. South Pacific Commission:

Suva, I iji. 30 pp.