Download Download Tech Report #129. Smith, C. W., J. Denslow, and S. Hight. Sept. 2002. Proceedings of a workshop on biological control of invasive plants in native Hawaiian ecosystems

PACIFIC COOPERATIVE STUDIES UNIT
UNlVERSlN OF HAWAII AT MANOA
Department of Botany
3190 Maile Way
Honolulu, HI 96822
Technical Report 129
PROCEEDINGS OF WORKSHOP ON BIOLOGIC
ECOSYSTEMS IN HAWAI'I
JUNE 2000
Edited by:
Clifford W. Smith, Julie Denslow, and Stephen Hight
September 2002
NATIVE
TABLE OF CONTENTS
PREFACE
ENHANCING SUCCESSFUL BIOLOGICAL CONTROL OF WEEDS BY
EXPANDING AND IMPROVING OVERSEAS RESEARCH. Joe Balciunas .
BlOLOGlCAL CONTROL OF IVY GOURD, COCClNlA GRANDIS
.
(CUCURBITACEAE), IN HAWAI'I. Marianne E. Chun .
CLASSICAL BIOLOGICAL CONTROL OF CLlDEMlA HlRTA
(MELASTOMATACEAE) IN HAWAI'I USING MULTIPLE
STRATEGIES. Patrick Conant. . .
HOST SPECIFICITY TESTING OF BIOCONTROL AGENTS OF WEEDS.
Tim A. Heard . .
HOST SPECIFICITY AND RlSK ASSESSMENT OF HETEROPERREYIA
HUBRICHI, A POTENTIAL CLASSICAL BIOLOGICAL CONTROL AGENT
OF CHRISTMASBERRY (SCHINUS TEREBINTHIFOLIUS) IN HAWAI'I.
Stephen D. Hight . .
BIOLOGICAL CONTROL POTENTIAL OF MlCONlA CALVESCENS USING
THREE FUNGAL PATHOGENS. Eloise M. Killgore .
BIOLOGICAL CONTROL OF GORSE IN HAWAI'I: A PROGRAM REVIEW.
George P. Markin, Patrick Conant, Eloise Killgore, and Ernest Yoshioka
SETTING PRIORITIES FOR THE BlOLOGlCAL CONTROL OF WEEDS: WHAT
TO DO AND HOW TO DO IT. Judith H. Myers and Jessica Ware..
HOST SPEClFlClTY TESTING FOR ENCARSIA SPP., PARASITOIDS OF THE
SILVERLEAF WHITEFLY, BEMlSlA ARGENTlFOLll BELLOWS & PERRING,
IN HAWAI'I. Walter T. Nagamine and Mohsen M. Ramadan .
PREDICTABLE RISK TO NATIVE PLANTS IN BlOLOGlCAL CONTROL OF
WEEDS IN HAWAI'I. Robert W. Pemberton .
REVIEW AND PERMIT PROCESS FOR BlOLOGlCAL CONTROL RELEASES IN
HAWAI'I. Neil J. Reimer . .
FOREST PEST BIOLOGICAL CONTROL PROGRAM IN HAWAI'I.
.
Clifford W. Smith .
A RESOURCE MANAGER'S PERSPECTIVE ON THE ROLE OF BIOCONTROL
IN CONSERVATION AREAS IN HAWAI'I. J.T. Tunison . .
STRAWBERRY GUAVA (Psidium cattleianum): PROSPECTS FOR BIOLOGICAL
CONTROL. Charles Wikler and Clifford W. Smith . .
SYNTHESIS AND CONCLUSIONS; HAWAII BIOCONTROL WORKSHOP.
BIOLOGICAL CONTROL OF INVASIVE PLANTS IN NATIVE
HAWAIIAN ECOSYSTEMS
PREFACE
The importation of alien insects and pathogens to control invasive alien weeds raises
justifiable concern among land managers and conservationists. Do we risk
compounding the problem by introducing yet another alien species for which we have
only an imperfect assessment of its risk of becoming invasive itself? What is the
likelihood that an imported control agent will affect non-target species or expand beyond
expected habitats and host species? For the Hawaiian archipelago the dangers are
particularly acute.
Hawai'i has many endemic species, a substantial percentage of which are at risk of
extinction. The Hawaiian vascular plant flora includes about 1302 taxa (including
subspecies and varieties) of which 1158 are endemic (Wagner et a/. 1990). Some 37%
of these taxa are endangered or at risk of becoming extinct, representing 38% of all
federally listed endangered species in the United States (Loope 1998). Islands,
moreover, appear to be particularly vulnerable to invasive species. Over 900 nonindigenous plant species have become naturalized in Hawai'i, more than 90 of which
constitute substantial problems for conservation because they compete with native
species or so alter ecosystem processes that whole communities are changed (Vitousek
and Walker 1989). In spite of the magnitude of the invasive weed problem in Hawai'i,
we are unable to predict with any confidence which new plant introductions are likely to
become problems in future years. Beyond those species whose invasive tendencies
have been demonstrated elsewhere, our understanding of what combination of species
traits and ecosystem characteristics produce explosive, habitat-altering population
growth is rudimentary. There are good reasons for caution in the use of alien insects
and pathogens as control agents for invasive weeds.
Nevertheless biological control offers one of the most cost-effective and enduring
mechanisms for the control of persistent weeds that have become widely invasive in
natural habitats. Chemical and mechanical approaches to the control of weed
populations require perpetual maintenance, may inflict unwanted side effects on nontarget species and communities and are of limited use in large diverse ecosystems.
Extensive infestations in poorly accessible terrain require considerable long-term
investment in personnel and resources, expenditures that may be difficult to justify when
short-term economic returns are not apparent. Biological control offers the possibility for
~m1~~a~relyeradi~ati0n]~sive-ed~ov~mi~~acreageand-ime~~ibleterrain in perpetuity. Yet the technique is far from a panacea. Many years of exploration
and host-range testing are necessary before a potential control agent can be brought to
the point of release. Limitations of quarantine space and personnel mean that only a
handful of agents can be under investigation at any one time. While the numbers of
releases resulting in unpredicted impacts on non-target hosts have been low in recent
times (J. Balciunas this volume, R. Pemberton this volume), many releases have been
less than successful because the agent either fails to establish viable populations and/or
is ineffective in limiting populations of the target plant over part or all of its range.
Financial constraints frequently inhibit our ability to conduct the necessary studies on the
biology of a species in its native environment.
Clearly the challenge to the community of scientists and managers seeking to use
biological control agents in Hawaii is to make the most efficient use of limited space,
personnel, and financial resources in bringing the safest yet most effective insect and
pathogen agents on line. The most productive research strategies for meeting that goal
was the topic of the 2000 Conservation Forum of the Hawai'i Secretartiat for
Conservation Biology: Biological Control of lnvasive Plants in Native Hawaiian
Ecosystems. Presenters and discussants were invited to provide both breadth of
international experience in a diversity of plant-herbivore-predatorsystems and depth of
understanding of the particular idiosyncracies of island ecosystems. They were charged
to take from the theory and patterns of evolutionary and population biology and from the
experience gained in Hawai'i to recommend a framework of research priorities and
strategies. Such strategies should not only improve the efficiency with which we bring
new control agents to the point of release, but also Increase the likelihood that released
agents are both effective at reducing population sizes of target species and unlikely to
threaten non-target plants. This volume is a compendium of historical syntheses,
examples of effective research strategies, and detailed case studies from the Hawaiian
experience in biological control. It is capped by a synthesis that arose from discussions
of strategies for exploration and country-of-origin studies, of lessons from Hawaiian
releases, of protocols for host-range testing, and of appropriate pre-and post-release
assessments of impact.
It was our hope that the forum would be not only a stimulus for discussion and
information exchange, but also a source of renewed energy, direction, and cooperation
among the diverse community of scientists and managers concerned for the future of
native Hawaiian ecosystems. We are thus grateful for the participation of representatives from many state and federal agencies, of land managers, and of community groups
and for the contributions of scientists from the US mainland and abroad who contributed
enthusiastically in all aspects of the proceedings. All these components were brought
together in a smoothly-run meeting through the efforts of the late Nancy Glover, Director
of the Secretariat for Conservation Biology, and her assistant, Moani Pai, who oversaw
the conference logistics, and through the excellent management of Audrey Haraguchi
and her assistant Olivia Rivera of the Institute of Pacific Islands Forestry, who arranged
travel for international and mainland participants. The productivity and quality of the
meeting would have been much diminished without their dedicated efforts. We are
grateful for financial support from US Department of Agriculture Forest Service
International Programs and the lnstitute for Pacific Islands Forestry, from the Secretariat
for Conservation Biology, and from the US Geological Service-Biological Resources
Division Pacific Cooperative Studies Unit, University of Hawai'i.
LITERATURE CITED
Loope, L.L. 1998. Hawall and Pacific lslands. pp. 747-774, In: M.J. Mac, P.A. Opler,
C.E. Puckett Haecker, and P.D. Doran (eds), Status and trends of the nation's
biologicalresources, Volume 2. U.S. Department of the Interior, U.S. Geological
Survey, Reston, VA.
Vitousek, P.M., and L.R. Walker. 1989. Biological invasion by Myrica faya in Hawai'i:
plant demography, nitrogen fixation, ecosystem effects. Ecological Monographs 59:
247-265.
Wagner, W. L., D. R. Herbst, and S. H. Sohrner. 1990. Manual of the Flowering Plants
of Hawai'i. Honolulu, University of Hawaii and Bishop Museum Presses.
STRATEGIES FOR EXPANDING AND IMPROVING OVERSEAS
RESEARCH FOR BIOLOGICAL CONTROL OF WEEDS
Joe Balciunas
USDA Agricultural Research Service, Exotic & lnvasive Weed Research Unit, Western
Regional Research Center, 800 Buchanan St., Albany, CA 94704
Email: [email protected]
Abstract The followmg recornmendat~onsare made to Improve overseas research on blolog~cal
control agents Conduct more long-term overseas evaluations Place much greater emphasis on
field host range studies Carry out more research on Impact and efficacy of potential agents
Perform more ecological research overseas on target weed Adhere to lnternatlonal standards for
b~olog~ca~control
stud^& Document all findings including fallures
Key words: pre-release impact assessment, code of best practices, field host range
NEED FOR MORE OVERSEAS RESEARCH
For many, but not all, invasive exotic pests, classical biological control is a management
option that should be considered. For pests that are widespread or established in remote
areas, classical biological control may be the only viable control technology. The
introduction, release, and establishment of the natural enemies of exotic weeds obviously
require research in the region where the pest is native. I began conducting evaluations of
potentlal weed biocontrol agents in their native range shortly after receiving my Ph.D. in
entomology at the end of 1977 (Balciunas and Center 1981). Most of my career has been
spent in international exploration, and I've been fortunate to meet and work with many
other entomological explorers from many countries, whose task has been to identify
potential biocontrol agents in the native range of the host plant. I've also worked closely
with quarantine scientists in USA and Australia. Through this experience, I have gained a
personal appreciation not only for what types of overseas research works and what
doesn't, but also for which approaches are most likely to be effective.
The dramatic increase in international travel and commerce regrettably has been
accompanied by an increase in the introduction of new weeds. Development of biological
control agents for some of these new weeds has been a high priority, but the increase in
numbers of targets has required the expansion of investment in international research. In
some cases, this has been as simple as choosing a location and cooperator and making
arrangements for shipment of potential agents. In other cases, extensive and mtens~ve
research on the potential agent in its native range is not only desirable, but necessary.
Practitioners of the biological control of weeds, unlike those involved in the biological
control of insect pests, have long sought to minimize impacts on non-target species.
Evaluations of host-specificity have been routine for almost four decades. Initially, the
concern was primarily non-target impacts on commercial crops, but today, potential nontarget impacts on native plant species receive the most attention. Host specificity with
respect to native plant species usually must be evaluated in quarantine where the release
is proposed. This has exacerbated a bottleneck to the expeditious evaluation of potential
biological control agents because quarantine space and resources are limited and
expensive. Moreover, adequate testing to secure approval for release of a new agent
usually will require many years of evaluation in quarantine. In contrast, a single weeklong visit to the native host range could easily reveal a dozen species feeding on the
target plant. It is more efficient to screen these potential agents for host specificity in
their native range, than to tie up scarce quarantine resources in initial evaluations for
polyphagy or in nurturing small, non-viable laboratory colonies.
OVERCOMING RESTRICTIONS TO EXPANDED INTERNATIONAL RESEARCH
Although international research on new targets, as well as expanded research on current
targets, is highly desirable, a number of factors restrict an expansion in intemational
research. I briefly review two of these restrictions and offer some suggestions to
overcome them.
Insufficient Funding for International Research.
Current funding for international research to develop biological control agents for invasive
weeds is not only far from optimal; it is below critical levels. We lack the financial
resources both to launch projects on nWweeds and to SuccSssfully address current
targets. Significant new funding is seldom available, and current funding is stretched
thin. A string of successes in weed biocontrol would stimulate development of further
funding; however, few funds are currently available to ensure such future success. The
best near-term solution would be to reallocate current resources to increase the likelihood
of success for key projects. Fewer, better-funded projects more hkely would generate
such successes. However, this would mean terminating or delaying projects that appear
to offer little chance for success. A weed that has many closely related species native to
the area where an agent would be released will require far more effort and funds, and will
most likely have fewer acceptable agents, than a weed with few or no close relat~vesin
the region of release.
Weed targets for which other countries already have developed successful biological
control agents should be a high priority for support, because much of the intemational
research will have been completed. In 1996, we launched a project targeting Scotch
thistle (Onopordum acanthium L. Asterales, Asteraceae). While this was a new target
for North America, the Australians had been conducting intemational research on Scotch
thistle for many years, and, by 1996, had cleared and released several promising species.
For the U.S. we could concentrate on evaluation of the most promising of these
(Balciunas et a/. 1998). Unfortunately, the first two potential agents evaluated, the weevils
Lixus cardui Olivier (Coleoptera, Curculionidae) and Trichosirocalus sp. nov., were
unsu~tablefor release in the US because they readily attacked native American thistles
(Balciunas, unpublished data). Thus, the use of transfer agents developed for release In
other countries is not a panacea; significant additional testing will likely be necessary,
especially if the agents had received only cursory initial screening. For instance, the
moth, Cactoblastis cactorum (Berg) (Lepidoptera: Pyralidae), provided spectacular control
of Opuntia cactus in Australia, but in the Caribbean, it was found to attack endangered
native Opuntia cactus in Florida (Pemberton 1995).
-
Scarcity of Suitable Overseas Laboratories o r Collaborators
While ecological theory and the successes of biocontrol research suggest that effective
biocontrol agents are most likely to be found in the native range of the target weed, few
overseas laboratories specialize in biological control research, and they seldom are
located near appropriate areas for survey. War, civil unrest, and natural catastrophes,
such as a massive earthquake, flood, or volcanic eruption may restrict access to
appropriate areas. If sufficient funds are available, stationing an American scientist in the
appropriate region is often the most efficient strategy. Research and exploration can be
focused on the priorities of the home laboratory and the scientist easily can be relocated
when the project is completed or the location proves unrewarding or inhospitable. Since
it is difficult to persuade mid- or late-career scientists to commit to a lengthy overseas
assignment, young scientists are frequently assigned such projects. Domestic surveys of
natural enemies already present for the weed in its new range will familiarize the scientist
with the target host and appropriate collecting techn~ques,and will help himlher to
discriminate among types of damage to the target before helshe goes overseas. It will
also allow a supervisor to assess the capabilities of the scientist before assignment to a
distant, foreign location. When lack of funds or staffing constraints prevent assignment of
a staff scientist, local scientists may be contracted to conduct the desired research.
Although salaries and research costs are likely less, more supervision and
communication are required to assure successful completion of the project. The
supervisor should plan annual slte vislts and the local sclentlst should vslt the US
laboratory to better understand the status of the pest and its biology where it is invasive
and to interact with U.S. cooperators. When local biological control experts are available,
their collaboration can facilitate a successful project considerably to the benefit of both
cooperatinglaboratories.Tor example, Stefan Neser, at South Africa's Plant Protection
Institute, has contributed substantially to the search for potential biological controls for
Cape Ivy (also known as German ivy, (Delairea odorata Lemaire, synonym Senecio
mikanioides Asterales, Asterales)) in its native home. With his guidance this project
quickly compiled a complete list of herb~voresand began evaluations of some of the most
promising (Balciunas 2000a)
If local scientists are unavailable, trained scient~stsfrom a third country may be hired
to conduct the research in the desired region. USDA-ARS maintains biological control
laboratories in Australia, Argentina, and France and their staff regularly conduct surveys
in areas far distant from their laboratories. Likewise, biological control specialists from
CAB1 Bioscience, based in London and Switzerland, can be contracted to conduct
research almost anywhere. However, those organizations may require reimbursement for
all costs, including salaries and overhead.
Some projects rely on short exploration trips by staff scientists to the native range of
the target host to produce likely candidates for quarantine research. I do not recommend
this approach because it is not an efficient use of scarce quarantine resources and
because it is less likely to produce effective biological control agents.
-
RECOMMENDATIONS FOR IMPROVING OVERSEAS BIOCONTROL RESEARCH
Growing concern about the safety of biological control, coupled with the need for more
biocontrol successes mandate that international research be appropriate and efficient.
Since research funds are scarce, it is also important that international research not only
be productive in the development of new biocontrol agents, but also that agents are
successful at limiting the spread or impact of the target plant. The following
recommendations are made for conducting overseas biological control research that is
both efficient and effective.
Invest in long-term research.
Overseas research can be divided into two categories: 1) opportunistic, short-term forays,
and 2) long-term research. While I sometimes dismiss short-term forays as "grab-andrun", there will always be a role for quick studies. Frequently, the best sites for intensive
surveys and long-term research cannot be determined from the literature making personal
inspection of potential long-term study sites necessary. For example, because the native
range of hydrilla is broad comprising the tropical parts of Asia, Australia, and Africa, I
made three, 6-month trips between 1981-82, repeatedly collecting natural enemies in
these regions (Balciunas 1985). Northern Australia was selected as the best location for
long-term research on several potential agents.
However, projects that rely solely on short trips to the nat~verange of the target plant
to supply potential agents and preliminary data are (in my view) 'penny-wise and pound
foolish.' When untested agents are only sporadically available, their evaluation in
quarantine will proceed slowly. Only the most easily collected and reared are likely to
receive sufficient evaluation to support a release request. This approach gambles that a
suitable agent will be found before funds and interest in the project fall below critical
levels.
By contrast, long-term projects can complete the extenswe surveys necessary to
document completely the natural enemies of the target host. A short l~stof potential
agents can then be prepared and methodically evaluated, both in the f~eldand in
quarantine For example, a dyear survey of melaleuca (Melaleuca qurnquenenia (Cav.)
Blake - Myrtales, Myrtaceae) trees in Australia yielded over 400 herbivores (Balciunas et
a/ 1995),we were able to recommend nearly 30 as deserving further study.
Once an overseas laboratory is estab-lismd, host-specificity testing can usually be
conducted far more eas~lyin the native host range than under the restricted, containment
conditions of a quarantine facility. Extensive host range tests overseas elmmate
inappropriate agents, and speed up the quarantine testing for good agents
Emphasize Field Research Overseas.
While laboratory host-range tests conducted in the native host range are desirable, field
evaluations of the host range are even more valuable. In Australia, we routinely not only
tested the host range of a potential agent under laboratory conditions, but also regularly
conducted field surveys of related species and other potential hosts for these same
insects (Balciunas et a/. 1994, 1995, 1996). These field data were critical clarifying
ambiguous laboratory tests (such as a low reproduction rate on a non-target host) and
eventually in gaining release approval for agents that would have been eliminated
otherwise.
Conduct More Research on the Efficacy of Potential Agents.
Most practitioners of weed biocontrol feel that an agent is safe if it has a negligible impact
on non-target species. However, ecologists are now finding that some of these
presumably safe agents can have deleterious ecological impacts. For example, gall flies
(Urophora spp. Diptera, Tephritidae) released to control some knapweeds have become
extremely abundant on the knapweed target, but have failed to reduce the knapweed
populations. Fortunately, there has been no evidence of a direct impact on non-target
plants. However, there is now good evidence that field mice (Pemmyscus sp. Rodentia,
Rattidae) are using the abundant overwintering Urophora pupae as their primary winter
food, and that this has led to markedly higher populations of field mice (Pearson et a/.
2000).
To me this confirms the need to emphasize an agent's efficacy as much as we do its
safety. An ineffective agent not only can cause unpredictable changes to the food web
and the environment, but its development also wastes scarce biocontrol resources that
could have been devoted to better agents. Determination of probable impact is difficult to
do in the lab, but sometimes possible under field conditions overseas. In Australia, we
demonstrated that insects at natural, ambient levels, qulckly inhibit growth of Melaleuca
saplings by comparing the growth rates of saplings that had been sprayed with
insecticides, with those that had not been sprayed (Balciunas and Burrows 1993).
-
Conduct More Research on the Biology and Ecology of Target Weed in its Native
Range.
In its native range, the target weed is frequently innocuous and sometimes uncommon;
almost always little known about its biology and ecology. Research in the native range of
the weed could reveal the weak points in its life cycle and allow us to choose more
effective agents. Better knowledge of the weed's distribution and environmental
requirements in its native range can help us predict additional localities that are
susceptible to invasion (Balciunas and Chen 1993).
Adhere to International Standards for Research in Host Countries
Overseas researchers must familiarize themselves with the rules and regulations for
collecting, testing, and shipping specimens in each of the countries that they visit. Failure
to do so may lead to unpleasant outcomes, not only for them and their project, but to
scientists who follow.
Recently, I helped to formulate a Code of Best Practices for workers in the field of
biological control of weeds. In July 1999, delegates attendmg the Xth International
Symposium-for Biological Control of Weeds ratified this Code (Balciunas 2000b). Like
other practitioners, overseas researchers should adhere to the principles outlined in the
Code (Table 1). We are receiving increased scrutiny from ecologists and the general
public. Adherence to the Code not only will make our subdlsclpllne safer, but will assure
its greater acceptance by the public.
Document Findings and Failures.
Each weed biocontrol project may require decades of research and turnover of staff is
inevitable. Likewise, old targets may receive new attention, especially when they invade
new regions. Thus old research is valuable to new scientists. While all scientists have
an obligation to document their research, this is especially critical in our discipline. Much
research may be repeated because the original findings, including failures, were not
documented adequately in accessible literature. While refereedjournals are preferred,
symposium and conference proceedings are a good outlet for data and observations that
are of local interest.
ACKNOWLEDGEMENTS
I thank the organizers for allowing me to share these personal observations.
LITERATURE CITED
Balciunas, J. K. 1982. Insects and other macroinvertebratesassociated with Eurasian
watermilfoil in the United States. Technical Report A-82-5. U.S. Army Engineer
Waterways Experiment Station, Vicksburg, MS. 87 pp.
Balciunas, J. K. 1985. Final report on the overseas surveys (1981-1983) for insects to
control hydrilla. Technical Report A-85-4. U.S. Army Engineer Waterways
Experiment Station, Vicksburg, MS. 60 pp.
Balciunas, J. K. 2000a. Biological control of Cape ivy project reaches milestone.
CalEPPC News. pp. 93-94.
Balciunas, J. K. 2000b. A proposed Code of Best Practices for classical biological
control of weeds. pp. 435-436, In: N. R. Spencer (ed.), Proc. X Int Symp. Biol.
Control of Weeds, 5-9 July 1999, Bozeman, MT. Montana State University,
Bozeman, MT.
Balciunas, J. K. and D. W. Burrows. 1993. The rapid suppression of the growth of
Melaleuca quinquenervia saplings in Australia by insects. Journal, Aquatic Plant
Management 31: 265-270.
Balciunas, J. K., D. W. Burrows, and M. F. Purcell. 1994. Field and laboratory hostranges of the Australian weevil, Oxyops vitiosa (Coleoptera: Curculionidae), a
potential biological control agent for the paperbark tree, Melaleuca quinquenervia.
Biological Control 4: 35 1-360.
Balciunas, J. K., D. W. Burrows, and M. F. Purcell. 1995. Australian insects for the
biological control of the paperbark tree, Melaleuca quinquenervia, a serious pest of
Florida, U.S.A., wetlands. pp. 247-267, In: E. S. Delfosse and R. R. Scott (eds.)
Proc. Vlll Int. Symp. Biol. Control of Weeds, 2-7 February 1992, Lincoln University,
New Zealand. CSlRO Publishing, Melbourne, Australia
--
~~
Balciunas, J. K., D. W. Burrows, and M. F. Purcell. 1996. Comparison of the
physiological and realized host-ranges of a biological control agent from Australia for
the control of the aquatic weed, Hydrila verticillata. Biological Control 7: 148-158.
Balciunas, J. K. and T. D. Center. 1981. Preliminafy host specificity tests of a
Panamanian Parapoynx as a potential biological control agent for hydrilla.
Environmental Entomology 10: 462-467.
Balciunas, J. K., K. Chan, K. Do, M. Pitcairn, and D. Isaacson. 1998. Host specificity
testing of Lixus spp. for biological control of Scotch thistle. pp. 72-73, In: (D. Woods
ed.), Biological Control Program Annual Summary, 1997. California Dept. Food &
Agriculture, Plant Health and Pest Prevention Services, Sacramento, CA.
Balciunas, J. K. and P.P. Chen. 1993. Distribution of hydrilla in northern China:
implications on future spread in North America. Journal, Aquatic Plant Management
31: 105-109.
Balciunas, J. K. and M. C. Minno. 1984. Quantitative survey of the insects and other
macrofauna associated with hydrilla. Miscellaneous Paper A-84-2. U.S. Army
Engineer Waterways Experiment Station, Vicksburg, MS. pp. 5-172.
Balciunas, J. K. and M. C. Minno. 1985. Insects damaging hydrilla in the USA. Journal,
Aquatic Plant Management 23: 77-83.
Pearson, D. E., K. S. McKelvey, and L. F. Ruggiero. 2000. Non-target effects of an
introduced biological control agent on deer mouse ecology. Oecologia 122: 121-128.
Pemberton, R. W. 1995. Cactoblastis cacforum (Lepidoptera: Pyralidae) in the United
States: An immigrant biological control agent or an introduction of the nursery
industry? American Entomology4: 230-232.
Table 1. Code of Best Practices for Classical Biological Control of Weeds (as
approved July gth, 1999, by the delegates to the X International Symposium on
Biological Control of Weeds, Bozeman, Montana)
Ensure that the target weed's potential impact justifies release of nonendemic agents
Obtain multi-agency approval for target
Select agents with potential to control target
Release only safe and approved agents
Ensure only the intended agent is released
Use appropriate protocols for release and documentation
Monitor impact on target
Stop releases of ineffective agents, or when control is achieved
Monitor impacts on potential non-targets
Encourage assessment of changes in plant and animal communities
Monitor interaction among agents
Communicate results to public
BIOLOGICAL CONTROL OF IVY GOURD, COCCINIA GRANDIS
(CUCURBITACEAE), IN HAWAI'I
Marianne E. Chun
Hawai'i Department of Agriculture, 1428 S. King St., Honolulu, HI 96814, U.S.A.
E-mail: [email protected]
Abstract. Three insect blologlcal control agents collected in Kenya have beer1 introduced ir~to
Hawai'i to combat the exotic weed ivy gourd (Coccinia grandis). The clearwing moth, Melittia
oedipus, was released in 1996. The larvae of this moth bore into the mature vines and roots of
ivy gourd. It is now established in Hawai'i. Two additional agents, which belong to a group
known as the African melon weevils, were released in 1999. The first, Acflhopeus burkharkrum,
forms galls on young shoots. The second, A. cocciniae, mines ivy gourd leaves. Aspects of the
project, including host range testing are discussed.
Key words: Acythopeus burkhadomm, Acythopeus cocciniae, Coccinia gmndis,
Cucurbitaceae, Curculionidae, Meliffia oedipus, Sesiidae
INTRODUCTION
Ivy gourd, Coccinia gmndis (L.) Voigt (Violales, Cucurbitaceae), is an aggressive vine
that has become a serious weed in lowland areas of Hawai'i, particularly on the island of
O'ahu and on the Kona coast of the island of Hawai'i. It is native to Africa and has been
present in the Indo-Malayan region of Asia for many centuries (Burkhart 1993, Singh
1990). It is also naturalized in parts of Australia, the Caribbean, the southern United
States and several Pacific islands (Linney 1986, Telford 1990). Ivy gourd was first
collected on the slopes of Punchbowl, Honolulu, In 1968 and Its presence in the state
can probably be attributed to several independent introductions by immigrants from
Southeast Asia where ivy gourd is used for food and medicinal purposes (Nagata 1988).
Though probably still used to some extent by Southeast Asian immigrants, the general
public in Hawai'i has not adopted it as a common food item.
Ivy gourd is a dioecious perennial herb with tuberous roots and thick stems, which
can grow to several meters in length and up to 12 cm in diameter. These succulent
stems allow ivy gourd to survive defoliation caused by occasional outbreaks of powdery
mildew and by the recent drought experienced in Hawai'i. It has white flowers and small
cucumber-like fruits which turn red when ripe, making them attractive to birds who
distribute the seeds to new locations.
During the 1970's and 80Js,ivy gourd spread rapidly and began to blanket trees and
other vegetation. Unlike weeds which flourish at higher elevations and escape
widespread notice, the rampant growth of ivy gourd in residential neighborhoods and
agricultural' areas prompted many complaints from the general public and the Outdoor
Circle, a volunteer organization devoted to maintaining the natural beauty of Hawai'i.
The heavy vines hanging from telephone and electrical wires became a problem for
utility companies. In addition, ivy gourd fruits were found to be an excellent host of the
melon fly, Bactrocem cucurbitae (Coquillet) - (Diptera, Tephritidae). There was
particular interest at the time in reducing or eliminating fruit flies to facilitate the export of
Hawaiian agricultural crops. Proliferation of ivy gourd increased the melon fly
population. In response to the above, there was legislative interest in finding a way to
control ivy gourd, and the Hawai'i Department of Agriculture (HDOA) began work on the
project.
EXPLORATION
The vines and tuberous roots of ivy gourd were unresponsive to widespread application
of herbicides, and mechanical management was prohibitively expensive. Consequently,
the decision was made to look for biological control agents. In 1990, exploration had
been planned for Southeast Asia to look for natural enemies of the insect pests Thrips
palmi Karny (Thysanoptera, Thripidae) and the banana aphid Pentalonia nigronervosa
Coquerel (Homoptera, Aphidae). Ivy gourd was added to the list of target species based
on literature that considered Asia to be part of its native range and on consultations with
Asian scientists who believed effective control agents were present in the region.
How-ever, all of the insects and diseases collected during-subsequent work in Thailand,
Malaysia, and Indonesia were known to attack other cucurbits and therefore were not
cons~deredfor use in Hawai'i.
Two years later, exploration for natural enemies of ivy gourd shifted to Africa, the
center of diversity for the genus Coccinia (Singh 1990). New information suggested that
C. grandis is native to north central East Africa and perhaps Arabia (C. Jeffrey, pers.
comrn. to R. Burkhart). It likely moved into Asia in trade centuries ago. The other 29
species of Coccinia are confined to tropical Africa (Singh 1990). A list of collection
localities for ivy gourd was obtained from the Kew Herbarium, London.
During the summer of 1992, Burkhart, HDOA exploratory entomologist, collected
over 30 species of insects and several pathogens of ivy gourd in Kenya and Tanzania.
Other cucurbits were examined and preliminary host range tests were conducted on
promis~ngspecies. Based on these tests and field observations, five insect species and
three pathogens were selected as potential biological control agents and sent to Hawai'i
for further testing.
HOST RANGE TESTING
All cucurbits found in Hawai'i were included in host specificity tests on the eight potential
biological control agents. These included commercially grown cucurbit crops, naturalized
weedy cucurbits, and representative species of the endemic genus Sicyos as listed by
Telford (1990). Additional test plants included species in the Order Violales, several
plant species of economic importance, and key endemic species that are major
components of native Hawaiian ecosystems. Plants other than cucurbits that contain
cucurbitacins, secondary plant compounds characteristic of the family Cucurbitaceae
(Metcalf and Rhodes 1WO), were not tested, since none of the listed species are known
to occur in Hawai'i.
Early attempts to identify candidate insects met with failure, a problem common to
many biological control projects when agents are collected in parts of the world with
poorly known faunas. Eichlin (1995) identified the sesiid moth just as host range testing
was completed. The identity of the two weevils was delayed until a systematist willing to
help was located. An inltlal mlsldentlflcatlon incorrectly placed them in the wrong genus
resulting in our selecting the wrong plants for host screening. Eventually, they were
described by O'Brien and Pakaluk (1998), and with the names provided, it was possible
to apply for release.
Further delays were due to the time required for input and approval from multiple
state and federal agencies. The length of time required for the process is unpredictable
and may vary due to changing procedures and problems within the approving agencies.
In the case of this project, it contributed to a three-year gap between the release of the
first agent and the last. The possibility of delays due to this unpredictability should be
considered during the planning stage of a project, as they may add greatly to the cost
and can tie up personnel and quarantine space for long periods of time.
RELEASEPHASE
Three of the eight African insect species that underwent host range testing were found to
be host specific and were released in Hawai'i. The biology of the three insects and
release information is presented below.
Melittia oedipus Oberthiir (Lepidoptera: Sesiidae).
Larvae of this clearwing moth feed inside mature stems and tuberous roots of IVY gourd.
Females lay
Adults are dlurnal,emerging in the mornlng and mating soon thereafter.
eggs soon after mating is completed and are most active in full sunlight. Eggsarelaid
singly on all parts of the ivy gourd plant, from ground level to vines covering the tops of
trees. Newly hatched larvae bore immediately into the stems and are exposed only
briefly. Larval development and pupation all take place w~thinthe vine, and adults
emerge in two to four months. Other females are attracted to sites w~thprevious
infestations, and larvae of different sizes are commonly found in close proximity.
Repeated attacks cause vines to break and decay. Other insects attracted to the
decaying vines, such as the banana moth, Opogona sacchan (Bojer) - (Lep~doptera,
Pyralidae), appear to increase the damage. Meljftia oedipus was first released on the
slopes of Punchbowl, O'ahu, in August 1966. From that date to August 1999,
approximately 21,600 adults and 16,000 larvae were released on O'ahu. The moth is
now well established and vines have thinned out substantially
-
--
Acythopeus burkhartorum O'Brien (Coleoptera: Curculionidae)
This small black beetle is part of a group known as the African melon weevils. Adult
females lay their eggs in meristematic tissue at the tips of young shoots. As the shoot
elongates, galls form at the juncture of the stem with leaves and tendrils. When mature,
the larva excises the proximal end of the gall, causing it to fall from the plant. The larva
seals this cut end with plant fibers obtained from the gall lining. It then turns and excises
the crther end, forming a smooth cylinder, 10-13 mm in length. Pupation takes place
within this protective case, and the adult emerges three to four months later. Burkhart
(1993) surmised that the long pupal stage is probably an adaptation for the long dry
season in East Africa when suitable host plants are lacking. The adult weevils feed on
young ivy gourd leaves but cause only minimal damage. While this species is not
capable of killing ivy gourd, the galls are an energy sink and may slow down growth of
young vines. It was first released in August 1999 in Waimanalo on the island of Oahu
and in December 1999 in Kailua-Kona on the island of Hawai'i.
Acythopeus cocciniae O'Brien (Coleoptera: Curculionidae).
This weevil is similar in appearance to A. burkhartorum but is smaller, being only 2 to 2.5
mm in length, versus 5 to 6 mm for A. burkhartorum. A. cocciniae develops as a
leafminer in ivy gourd leaves. Developmental time is approximately 33 days from
oviposition to adult emergence. The adults also feed on the leaves. Combined damage
from larval and adult feeding can be quite severe. Acythopeus cocciniae was first
released on O'ahu in November 1999 and in Kona one month later. It has been
recovered at all release sites on both islands.
BIOTIC INTERFERENCE
One of the concerns with any potential biological control agent is the likelihood that
natural enemies will attack the agent itself. In the case of Melittia oedipus, there was
concern that a moth might not be effective, as lepidopterans in Hawai'i are often heavily
attacked by parasitoids. However, a reference to sesiid pest species in North America
noted that they are not well controlled by natural enemies (Solomon 1995). In particular,
there was no mention of attack by Trichogramma spp., egg parasitoids that have been a
limiting factor for many Lepidoptera in Hawai'i. This may be due to the ~lnusuallythick
chorion of sesiid eggs. In Africa, parasitism of specimens collected by Burkhart was light
and only a few specimens of a large braconid, Hyrfanommatium crassum Enderlein
(Hymenoptera, Braconidae), emerged from several pupae (Burkhart 1993). Neither this
spesies-nor-any-congener-s-o~cur-inMawai'iand,so-far,none-of-the-field-sollested--larvae, or pupae has shown any signs of attack by parasitoids. However, in 2001, a few
male eupelmids (Eupelmus sp. - Hymenoptera, Eupelmidae) were reared from fieldcollected eggs.
Ants were anticipated predators of immature stages of the moth. Some fieldcollected eggs show signs of predation, and it is probable that ants kill neonate larvae as
they emerge from the egg. However, once larvae bore into the stem, they are relatively
well protected, as entrance holes are blocked by frass.
A factor not considered prior to release was the high rat population present in the
Hawaiian Islands. Since the laboratory rearing of M. oedipus is labor intenslve and
HDOA personnel and laboratory space are limited, initially it was thought that larval
releases would be the most efficient means of getting large numbers of moths into the
field. Concentrated numbers were placed in close proximity to increase the probability of
emerging adults finding mates. This technique was effective at first. However, rats soon
discovered this new food source and tore open vines to extract the larvae. The moth,
however, became established, but the high rate of predation impeded a rapid population
buildup. We therefore switched to releasing adults instead of larvae since female moths
can scatter eggs over a wide area and the rats find a lower percentage of the immatures.
It is still too early to determine how detrimental natural enemies will be to
populations of the two Acythopeus weevils. In Kenya, both weevils were commonly
found early in the wet season in May. However, by the beginning of the dry season in
September 40-50% of the leafminers, and almost 100% of the gall-formers, were
parasitized (Burkhart 1993). A eurytomid (Eurytoma sp. - Hymenoptera, Eurytomidae)
and a eupelmid (Neanastatus sp. prob. mfatus Feniere - Hymenoptera, Eupelmidae)
were collected from parasitized galls. Neither of these two parasitoid species occurs in
Hawai'i. However, Eupelmus cushmani (Crawford) (Hymenoptera, Eupelmidae), has
been reared from A. cocciniae on both O'ahu and Hawai'i, and birds appear to be
opening A. burkhartorum galls. In addition, there is a likelihood that the melon fly will
oviposit in these galls (M. Ramadan, pers. comm.).
LITERATURE CITED
Burkhart, R. M., 1993. Unpublished report on out-of-state travel to Chairperson, Board
of Agriculture, Hawaii Department of Agriculture files. March 24, 1993.
Eichlin, T. D., 1995. New data and a redescription for Melittia oedipus, an African vine
borer (Lepidoptera: Sesiidae). Tropical Lepidopterist 6: 47-51.
Linney, G., 1986. Coccinia grandis (L.) Voigt: A new cucurbitaceous weed in Hawaii.
Hawaii Botanical Society Newsletter 25: 3-5.
Metcalf, R. L. and A. M. Rhodes. 1990. Coevolution of the Cucurbitaceae and Luperini
(Coleoptera: Chrysomelidae): basic and applied aspects. In: D.M. Bates, R.W.
Robinson and C. Jeffrey (eds), Biology and Utilization of the Cucurbitaceae. Cornell
Univ. Press, lthaca & London.
Nagata, K. M., 1988. Notes on some introduced flora in Hawaii. Bishop Museum
Occasional Papers 28: 79-84.
O'Brien, C. W. and J. Pakaluk, 1998. Two new species of Acythopeus Pascoe
(Coleoptera: Curculionidae: Baridinae) from Coccinia grandis (L.) Voight
(CuTwb-itaceae)-i"Kenya.-Pmceedingso ~lhe~Ento6m~log5~aaI~S~~iety,
f
WaasMngton.100: 764-774.
Singh, A. K., 1990. Cytogenetics and evolution in the Cucurbitaceae. In: D.M. Bates,
R.W. Robinson and C. Jeffrey (eds), Biology and utilization of the Cucurbitaceae.
Cornell Univ. Press, lthaca & London.
Solomon, J. D., 1995. Guide to Insect Borers of North American Bro~dleafTrees and
Shrubs. Agriculture Handbook 706. U.S. Department of Agriculture, Forest Service.
Washington, DC. 735 p.
Telford, I. R. H., 1990. Cucurbitaceae. pp. 568-581, In: W. L. Wagner, D. R. Herbst and
S. H. Sohmer (eds), Manual of the flowering plants of Hawaii, University of Hawai'i
and Bishop Museum Presses, Honolulu.
CLASSICAL BIOLOGICAL CONTROL OF CLIDEMIA HIRTA
(MELASTOMATACEAE) IN HAWAI'I USING MULTIPLE
STRATEGIES
Patrick Conant
Hawai'i Department of Agriculture, 16 E. Lanikaula St., Hilo HI 96720, U.S.A.
Email. [email protected]
Abstract. Biological control of
Clidernia hirta in Hawai'i has been episodical in its application
over the last 50 years driven more by economics than by biology. Four phases (mid 1950s, late
I9'0sS Ilate_19_80s, mid_1_99&) ~_arecdes_cri
bed. A l l L b u t o n e w a s t h e resultLoflobb_ying-by
concerned interests. A number of strategies have been tried over the years and their
contribution to the control of C. hirta is examined. Six different insect natural enemies and one
pathogen have been released up to the present. Evaluation of effectiveness has been
completed for only one of the insects and no study has been made of the combined effects of
the agents on the growth or reproduction of the plant. Although better control is still needed in
infested natural areas, the two most recently released moths that attack reproductive parts may
have good impact potential.
Key words: biocontrol, Antiblemma acclinalis, Ategumia matutinalis, Buprestidae,
Carposinidae, Cerposina bullafe, Colletofrichum gloeosporioidesf.s. clidemiae,
Liothrips urichi, Lius poseidon, Melastomataceae, Momphidae, Mompha trithalama,
Noctuidae, Phlaeothripidae, Pyralidae.
INTRODUCTION
Biological control of Clidemia hilta D. Don (Myrtales, Melastomataceae) in Hawai'i
began almost 50 years ago and has had a complex history of periodically active work.
A variety of approaches have been employed, ranging from lobbying for funds to
employing different release methods for the biocontrol agents. Over the decades, and
as support has waxed and waned, time has allowed the testing of a wide variety of
strategies. By examining the successes and failures, we can plan future weed
biocontrol projects that are better tuned to unique problems inherent in Hawaiian
ecosystems and in implementation of biocontrol programs.
Nakahara ef a/. (1992) summarized the history of the classical biological control
program against C. hirta, commonly known in Hawai'i as clidemia or Koster's curse.
Smith (1992) reported on the spread and ecosystem-altering capacity of this weed.
These authors made recommendations regarding biological control of clidemia. l will
review these recommendations later, but I especially want to emphasize what we have
learned along the way in our efforts to control a well-established invasive forest weed in
Hawai'i. A chronological account of the clidemia biocontrol effort shows that this
program changed course several times as interest and funding came and went over the
years. My purpose here is to show the various strategies tried and how they
contributed to the classical biocontrol program for this weed.
BIOLOGICAL CONTROL
First Phase.
Clidemia was first reported as established in Wahiawa, O'ahu, (Krauss 1954) at a
meeting of the Hawaiian Entomological Society. However, at the same meeting, a
forester of the Board of Agriculture and Forestry, Mr. Karl Korte, reported seeing it there
in 1941. Also reported in the meeting were personal observations by C. E. Pemberton
of a noticeable reduction in clidemia between his visits to Fiji in 1920 and 1937 that he
attributed to control by the thrips, Liothrips urichi Karny (Thysanoptera,
Phlaeothripidae). Importation and release of the clidemia thrips in 1953 began the
biological control program for clidemia that is still ongoing (Table 1).
lmportation of the clidemia thrips is an example of what I will call "mail order
classisal biocontrol". It is a fast-and generallycheap way to import natural enemies.
For obvious reasons, there are some prerequisites to being able to "place the order",
such as:
1 A known natural enemy that has somehow shown some potential for biocontrol,
and,
2. A reliable, affordable, cooperator where the weed and natural enemy occur.
Fulfilling item two is not as easy as it might sound. It is usually not difficult to find
cooperators In foreign countries, but it can be very difficult to find one that is both
affordable and reliable. The thrips had given good results in Fiji (Simmons 1933) and a
cooperator was available.
It was not until 1982, however, that Reimer and Beardsley (1989) conducted an
evaluation of the effectiveness of the thrips. The insect only occurred in sunny or partly
sunny areas and did not affect production of flowers or berries. However, it stunted
vegetative plants, causing significant termmal leaf abscission, even killing some plants.
Once the thrips was released, interest in further biocontrol work waned (perhaps
because it was no longer considered a significant weed of pastures and plantations).
This is a typical scenario for long-term biocontrol projects in Hawai'i and probably
elsewhere.
Secand phase.
It IS unclear what spurred Davis to collect a pyralid leaf roller, Ategumia matutinalis
(Guenee) (Lepidoptera, Pyralidae) on clidemia in Puerto Rico and Trinidad, West lndies
but it was released in Hawai'i in 1969 (Davis 1972). Perhaps the growing publicity
about invasive forest weeds that came with early 1970's environmental consciousness
and determined lobbying efforts at the state legislature by the Sierra Club, the
Conservation Council for Hawai'i and others gained popular support for control of
~nvasiveforest weeds during the mid 70's. The continued spread of clidemia in recent
decades (Wester and Wood 1977) has made ~ta hlgh pnor~tytarget. Newspaper
articles appeared and chronicled the releases of clidemia natural enemies by members
of the Sierra Club and its High School Hikers. Without the determined lobbying efforts
ot several people (e.g., Betsy Gagne, Lorrm Gill, Dana Peterson), the dormant clidemia
biocontrol program might not have been revitalized. Keeping the broad conservation
community informed and involved in natural area weed biocontrol can make a big
difference In fundlng allocations.
Funding obtained by the lobbying effort and the publicity generated led the
governor to issue a mandate requiring the use of available departmental funds to find
effective methods of control, including biocontrol By 1978, the University of Hawai'i
and the Department of Land and Natural Resources (DLNR) were involved in clidemia
research and control and the Hawai'i Department of Agriculture (HDOA, formerly the
Board of Commissioners of Agriculture and Forestry) was designated the lead agency
(Nakahara et a/. 1992).
Third phase.
A classical biocontrol exploration program was proposed by HDOA and implemented in
1979. Robert Burkhart conducted a typical exploratory trip of three months duration in
South America. A number of natural enemies were sent back but could not be kept
alive in quarantine. Dead specimens of each species were retained in the HDOA
collection. Insects attacking reproductive parts could not be cultured due to the poor
ambient light in the 1960's vintage quarantine building. Clidemia plants did not thrive,
flowers aborted-and? ruit dropped. The-inadequacy of the quarantinefacilityfor
growing certain target weeds through their life cycle led to another change in the
approach to clidemia control. Virtually all classical biocontrol exploration done by the
HDOA previously was based typically on a 3-month or less collecting trip. In contrast,
the 1980 exploratory trip was a 6-month trip to Trinidad jointly funded by DLNR and
HDOA. A local technician was hired there and trained to ship insects to Hawai'i after
Burkhart left. Unfortunately, for the second time, none survived in quarantine in
Hawai'i. Then in 1982, Burkhart set up house-keeping in Trinidad at the
Commonwealth Institute of Biological Control (CIBC) for a longer stay to collect, rear
and test the specificity of the natural enemies of clidemia. Burkhart was a welcome
guest because he had taken natural enemies of crop pests to Trinidad for exchange.
On this trip, a new, almost revolutionary approach for HDOA was tried: The host
specificity tests were performed outdoors under natural conditions to avoid all the
problems associated with the dark and highly artificial quarantine environment. Was
this "country of origin" method of exploration effective in this case? Was it worth the
cost? To answer the first question, yes it was indeed effective. Regarding cost, this
trip was a somewhat special case in that the costs were minimized by exchanging
natural enemies for services and by combining funding from other exploratory
biocontrol projects he performed simultaneously. This same country-of-origin work
today can be quite expensive.
In his clidemia work at CIBC Burkhart used two basic approaches:
1) choice and no choice tests in outdoor cages; and,
2) exhaustive sampling of representative taxa of sympatric non-target plants to
delineate the natural host range of the more promising natural enemies.
His studies identified 14 species that appeared to have potential for Hawai'i
(Nakahara et a/. 1992). Two of these, Eurytoma sp. "blackn(Hymenoptera:
Eurytomidae) and Penestes sp. (Coleoptera: Curculionidae), were found to be of no
value (Burkhart 1986, 1988). His tests led to the eventual release of four more species.
Within this small complex of insects lie more lessons to be learned from this project.
Among the myriad dilemmas facing an exploratory entomologist working alone in a
foreign country is "the quick fix vs. the long slow, difficult, but presumably more
effective fix" dilemma. The two approaches are not mutually exclusive but the quick fix
does tend to favor species that are easy to handle in quarantine and which have not
been evaluated as the most effective of the suite of species available. However, an
exper~encedbiological control specialist can often choose effective agents after a short
visit to the native country of a target weed. The evaluation of all host-specific species
for impact and then working out their biology and conducting host screening
---
experiments prior to their introduction into quarantine can be extremely expensive and
time-consuming. Administrative pressure to justify spending money out of state as well
as the need for success in order to justify the foreign travel are never far from one's
thoughts.
Lius poseidon Napp (Coleoptera: Buprestidae) and Anfiblemma acclinalis Hubner
(Lepidoptera. Noctuidae) are examples of the "quick fix". Lius poseidon is a buprestid
beetle that mines the leaves in the immature stage while adults are defoliators.
Antiblemma acclinalis larvae roll up leaves and feed within. Unfortunately both species
are attacked by parasitoids already present in Hawai'i.
Mompha trithalama Meyrick (Lepidoptera: Momphidae) and Carposina bullata
Meyrick (Lepidoptera: Carposinidae) are examples of the long, slow method Mompha
trithalama larvae feed primarily on the seeds within berries and C. bullata larvae feed
primarily-on-flowers,(these roles-overlap-somewhat). The-formeFis-established in
Hawai'i but it is too early to determine any impact. The latter has not been established.
Biotic interference is a significant problem in biological control programs,
particularly in Hawai'i. Its effects have been studied in the Hawaiian program against
clidemia. Reimer and Beardsley (1986) found that larvae of the leaf roller (A.
matutinalis) were parasitized by 4 species of hymenoptera. Their sampling methods
did not include egg or pupal parasitiods, although they did rear out one species of
Trichogramma from an egg. Percent parasitization of the larvae was consistently high
suggesting that "These high levels of parasitization by parasitoids may be a major and
at the very least an important factor contributing to low (A. matutinalis) field
populations" (Reimer and Beardsley 1986). Effectiveness of the leaf roller has never
been evaluated, but its rarity in the field suggests that it has little impact on the plant.
Damage in the field is readily recognized by the rolled up sub-terminal leaves in which
the larvae feed. Reimer (1988) found that ants and an anthocorid bug preyed on the
thrips and caused significant mortality.
The two control agents released and evaluated for biocontrol of clidemia both
suffer from biotic interference. Antiblemma acclinalis Hubner (Lepidoptera, Noctuidae),
first released in 1995, may be suffering a similar fate since the moth and its damage
are rarely seen, even at former release sites. Young larvae feed on leaves at night,
and rest under leaves during the day. Third instar and older larvae migrate down to the
ground during the day and climb back up to feed on foliage at night (Burkhart 1987).
My collections of Lius poseidon larvae produced adults of Chrysocharis parksi
Crawford (Hymenoptera: Eulophidae), a purposefully introduced parasitoid of Liriomyza
(Diptera. Agromyzidae) leaf miners on vegetable crops. It is possible that other leaf
miner parasitoids are attacking this beetle and they may be attacking a Gracillariid moth
that has been released to control Myrica faya Aiton (Myricales, Myricaceae). How much
biotic interference occurs with other natural enemies released for biocontrol of weeds in
Hawai'i? More field collections of immature stages of biological control agents are
needed to assess this problem. In fact, it is a significant need in the evaluation of all
Hawaiian biocontrol introductions.
In 1985, the Hawai'i State Legislature adopted another tactic, the use of a plant
pathogen. A leaf spot fungus (Colletotrichum gloeosporioides f. sp. clidemiae Trujillo
(Deuteromycotina, Melanconiaceae) from Panama appeared to have good potential
(Trujillo, Latterell and Rossi 1986). It was only the second time a pathogen had been
used in classical weed biocontrol in Hawai'i, and the first one against a natural-area
weed. There always had been considerable resistance to the use of pathogens but the
demonstrated potential of the fungus overcame opposition. The fungus is now
established on most islands infested with clidemia. Defoliation can be extensive over
contiguous areas when weather conditions are favorable (cool, windy and rainy). Its
effects on the weed have not been quantitatively evaluated as yet so it is difficult to
assess its long-term impact, but it does appear to defoliate and stress the plant at least
seasonally.
Lius poseidon was approved for release by the Board of Agriculture in early 1988.
Specimens from each shipment were first sent to Dr. Minoru Tamashiro, University of
Hawai'i, to check for pathogen infection of the agents prior to their release. The insect
is now established on Maui, O'ahu, Kaua'i and Hawai'i. The effectiveness of ne~ther
the leaf feeding adults nor leaf-mining larvae has been quantified. Damage to young
plants appears to be greater than to mature plants, particularly in combination with
thr~psdamage.
Fourth phase.
The clidemia project became dormant again after the release of L. poseidon. The
fourth attempt to use biological control began in the mid 1990s. Hurricane lniki (1992)
caused extensive damage to forested areas on Kaua'i, which became vulnerable to
weed invasion. U.S. Forest Service (USFS) funds became available in 1995 for forest
restoration with control of invasive forest weeds a high priority. Since clidemia already
infested the wetter low to mid elevation forests there, some of these funds were used to
import another natural enemy for clidemia. Antibiemma acclinalis had been approved
for release years earlier but no funds had been available to import it. It was first
released in 1995 but remains uncommon, probably due to parasitism.
Rearing and release of A. acclinalis was still ongoing in 1998 when USFS Special
Technology Development Program funds were obtained to import C. bullata and M.
trithalama. Burkhart had finished all the host specificity tests in Trinidad many years
earlier, but the results and petition to release had never been prepared because
funding had lapsed. State and Federal approval was obtained for release of both
species in 1995 and releases began that year. However, releases were very small and
establishment was doubtful. Funding from the U.S. Army coincidentally became
available in 1997 for Burkhart to rear both species in Tobago, West lndies and ship
them to the HDOA in Honolulu as pupae. Samples were sent to Dr. Gerard Thomas in
Berkeley for pathogen diagnosis prior to release. Releases of M. trithalama were made
at Schofield Barracks East Range (Schofield-WaikaneTrailhead), Lyon Arboretum and
Kahana Valley on O'ahu, and at Pohoiki and Waiakea Forest Reserve on the island of
Hawai'i. M. trithalama now appears to be established at Kahana Valley and Pohoiki.
C. bullata has yet to be recovered, but surveys will continue at all release sites.
Redistribution will be made to other islands and other clidemia infestations within
islands once either moth is firmly established.
FUTURE PROSPECTS
The consensus among environmentalists and land managers appears to be that
clidemia is still not under adequate control. It is too early to know the effects of the two
most recently released lepidoptera (Table 1). No formal impact evaluation studies are
undenvay other than checking for establishment. Importation and release of new
agents in the future may warrant consideration. Nakahara et a/. (1992) mention five
species (among others) of clidemia natural enemies in Trinidad identified by Burkhart
as having potential for biocontrol in Hawai'i. Two of these are lepidoptera that attack
both clidemia and Miconia sp. (Melastomataceae) flowers. The remaining three are a
eurytomid gall-forming wasp that attacks the berries, an unidentified cecidomyiid midge
that attacks flowers and an unidentified stem-boring cerambycid beetle, which proved
difficult to work with. All three of these might be at low risk of biotic interference and
the former two could be useful additions to the complex of natural enemies now
established. However, the use of plant pathogens should be reconsidered now that
HDOA has a quarantine facility with a plant pathologist on staff. Pathogens have some
advantages over insects: they can be tested more quickly than insects, take up less
space and are, in general, easier to propagate.
Table 1. Natural enemies of Clidemia hirta (Melastomataceae) released in Hawai'i.
Species
Liothrips urichi
Ategumia matufinalis
Colletotrichum gloeosporioides f .s. clidemiae
Lius poseidon
A ntiblemma acclinalis
Mompha trithalama
Carposina bullata
Part of plant attacked
TERMINALS
LEAVES
LEAVES
LEAVES
LEAVES
FLOWERSIFRUIT
FRUITIFLOWERS
Year Released
1053
1969
I986
1988
1995
1955
1995
Using biological control against any plant in the family Melastomataceae in Hawai'i
could also be beneficial to the control of clidemia. Since virtually all of these species
are known to be weeds in Hawai'i, host specificity of a biocontrol agent needs only to
be limited to the family of the host plant. The biocontrol program for Miconia
calvescens could conceivably aid in control of clidemia as well as other
melastomataceous weeds in Hawai'i.
SUMMARY OF STRATEGIES USED AND LESSONS LEARNED IN CLlDEMlA
BIOCONTROL
Education and publicity can lead to popular support and funding of
individual programs.
"Mail
order" natural enemies (biocontrol agents readily available from a
2)
foreign cooperator) can be a quick method of importing agents, if a
reliable cooperator at a reasonable cost can be found.
Traditional
short-term (three months or less) classical biocontrol
3)
exploration can be useful if natural enemies that can be reared easily
in quarantine are found.
Long-term
work in country of origin by a Hawai'i-based explorer can be
4)
effective but may be expensive. This strategy allows host-range and
specificity tests to be done under natural conditions.
5) The advantages and disadvantages of "Fast-tracking" easy-to-rear
foliar feeders vs longer-term efforts for harder-to-rear flowerlfruit
feeders should be evaluated
6) The likely impact of biotic interference should be evaluated. Leaf
feeding lepidoptera with exposed diurnal larvae may have lower
1)
probability of success due to biotic interference by parasitoids. Leaf
mining larvae may also be at risk from parasitoids.
7) Plant pathogens should be considered seriously in any biological
control program against weeds.
LITERATURE CITED
Burkhart, R. M. 1986. Progress report on exploratory studies on Clidemia hirta in
Trinidad, West Indies. June 1984-June 1985. Hawai'i Department of Agriculture
files, Honolulu. 52 pp.
Burkhart, R. M. 1987. Supplemental report on the host range and life history of
Anfiblemma acclinalis Hubner (Lepidoptera: Noctuidae). Hawai'i Department of
Agriculture files, Honolulu, Hawai'i. 13 pp.
Burkhart, R. M. 1988. Supplemental report (Part II) of investigations in Trinidad of
insects feeding on the flowers and berries of Clidemia hirta. June 1985-June 1986.
Hawaii Department of Agriculture files, Honolulu. 51 pp.
Davis, C. H. 1972. Recent introductions for biological control in Hawaii. Proceedings,
Hawaiian Entomological Society 21 : 59-62.
Krauss, N. H. 1954. Notes and Exhibitions. Proceedings, Hawaiian Entomological
Society 15: 263-265.
Nakahara, L. M., R. M. Burkhart and G. Y. Funasaki. (1992). Review and status of
biological control of clidemia in Hawai'i. pp. 452-465, In: C.P. Stone, C.W Smith
and J.T. Tunison (eds.), Alien plant invasions in native ecosystems of Hawai'i:
management and research. University of Hawai'i, Department of Botany,
Cooperative National Park Resources Studies Unit, Honolulu.
Reimer, N. J. 1988. Predation on Liothrips urichi Karny (Thysanoptera:
Phlaeothripidae): a case of biotic interference. Environmental Entomology 17:
132-.134.
-Reime~N~cl~and-d~VV,-Beafd~ley~-1986.-Some-notes-onparasitization~aL~~~~~
Blephatomastix ebulealis (Guenee) (Lepidoptera: Pyralidae) in Oahu Forests.
Proceedings, Hawaiian Entomological Society 27: 91-93.
Reimer, N. J, and J. W. Beardsley. 1989. Effectiveness of Liothrips urichi
(Thysanoptera: Phlaeothripidae) introduced for biological control of Clidemia hirta
in Hawaii. EnvironmentalEntomology 18: 1141- 1146.
Simmons, H. W. 1933. The biological control of the weed Clidemia hirta D. Don in Fiji.
Bulletin of Enfornological Research 24: 345-348.
Smith, C. W. 1992. Distribution, status, phenlology, rate of spread, and management
of clidemia in Hawai'i. pp. 241-253, In: C.P. Stone, C.W Smith and J.T. Tunison
(eds.), Alien plant invasions in native ecosystems of Hawai'i: management and
research. University of Hawai'i, Department of Botany, Cooperative National Park
Resources Studies Unit, Honolulu.
Trujillo, E. E., F. M. Latterell and A. E. Rossi. 1986. Colletotrichum gloeosporioides, a
possible biological control agent for Clidemia hilta in Hawaiian forests. Plant
Disease 70: 974.
Wester, L. and H. 6.Wood. 1977. Koster's Curse (Clidemia hilta), a weed pest in
Hawaiian Forests, Environmental Conservation. 4: 35- 41.
HOST SPECIFICITY TESTING OF BIOCONTROL AGENTS OF
WEEDS
Tim A. Heard
CSlRO Entomology, Long Pocket Laboratories, 120 Meiers Rd, lndooroopilly 4068,
Brisbane, Australia
Email: [email protected]
Abstract. A range of test designs is available to biocontrol practitioners. Their suitability
depends on the biology of the potential agent being tested. Tests are conducted on many
aspects of the biology of agents including oviposition, adult feeding, larval feeding, larval
development, adult longevity and fecundity, and field utilization under natural conditions.
Commonly used designs are choice, no-choice and choice-minus-target, in parallel or in
sequence. Many behavioral factors affect the results of host specificity testing. These factors
can be divided into 1. the sequential behavioral responses to host plant cues, 2. learning and 3.
the effects of time-dependent factors. These behavioral effects can be complex, can be caused
by a variety of mecnanlsms and can produce opposlng effects. Biocontrol workers should be
familiar with the behaviors that affect the results of these tests and apply their knowledge of
them appropriately. That is, biocontrol workers responsible for host specificity testing should
recognize that they are applied behavioural biologists as well as applied ecologists.
Keywords: host range testing, risk assessment, applied behavioral theory.
INTRODUCTION
Host specificity testing is a critically important step in the process of introducing natural
enemies for classical biological control. Most direct non-target effects can be predicted
if a sound understanding of the host specificity of potential biocontrol agents is gained.
Host specificity testing provides the basic information upon which the safety of a
proposed biocontrol agent can be assessed. Sound host specificity testing also
prevents another problem-the rejection of safe and potentially effective agents
because of an inability to prove their high level of specificity. Hence in host-specificity
testing, we are trying to avoid false results, both false positive results and false
n e ~ a t ~ - o ~ . - E a l s ~ e ~ p ~ srefer
i t i v to
e sthe attack of a plant in the test when there is no
potential for attack on that plant in nature. That is, the host range is over-estimated.
False negative results indicate no attack in the test where there is potential for attack in
the field; the host range is thereby under-estimated (Marohasy 1998).
In this paper, I review experimental designs used in host-specificity testing of
potential biological control agents. I then list some of the more important behavioral and
biological factors that affect the validity of tests and discuss the strengths and
limitations of the various tests used in light of these behavioral and biological factors. I
do not cover the topic of the selection of plants for the host test list as this is well
covered in the literature. This talk focuses on arthropod agents of weeds but the results
and analysis are applicable to host specificity testing of predators and parasitoids of
arthropod pests. Host specificity testing of pathogens will continue to rely on inoculation
of test plants because spore dispersal of most potential weed control agents is passive.
However, where insects are the vectors of pathogens, appropriate testing must account
for insect behavior.
REVIEW OF METHODS USED IN HOST-SPECIFICITY TESTING
Aspects of biology examined
Many experimental approaches to determination of the host range are used. Various
experiments examine aspects of host selection and use by agents as indicated by
oviposition, adult feeding, larval feeding, larval development, adult longevity and
fecundity. Some tests are conducted in the field and determine which hosts are used
under natural conditions (Heard & van Klinken 1998).
Test designs
Tests are traditionally divided into two designs-- choice and no-choice--dependingon
whether the organism can select from a range of species or is restricted to a single
species. In a no-choice test, groups of insects are placed on each plant species in
separate arenas such as cages, jars, petri dishes, etc (Figure 1). In choice tests,
insects are located in arenas in the presence of a choice of plant species including the
target weed (the control). A separate category of design is the choice-minus-target (or
choice-minus-control) test that includes a choice of test plant species excluding the
target. The results of no-choice tests allow us to predict the realized field host range in
the absence of the target weed. Choice tests do the same in the presence of the weed.
Both of these scenarios are possible in nature so both types of tests have a rule.
No-choice and choice-minus-targettests are subdivided into sequential or
simultaneous depending on whether the target species and the test species are offered
in sequence to the same insects or at the same time to different insects. If done
simultaneously, separate groups are placed on test plants at the same time. Sequential
tests differ in that the same group of adults is moved from plant to plant. In the
sequential tests, the insects are not naive but have experience of other plants. The
significance of this will be discussed in the section on learning below.
Choice-minus-target tests are useful and perhaps under-used designs. A powerful,
efficient test with few behavioral shortcomings is one in which the target weed is
offered as no-choice and done simultaneously with the choice of test plants.
We can gain a clearer picture of the suitability of different host specificity tests by
viewing tests as a combination of biological responses measured under different
experimental designs (Table 1). The test design used depends on the biological
Table 1. Frequency of occurrence of test designs versus biological response
m e a s u r e ~ ~ s d e t e F m i n e d f r ~ m a U t e ~ a t u ~ v i ~ h ~
past 20 years (Heard & van Klinken 1998).
Oviposition
~ d u lfeeding
t
Larval feeding
Larval development
Adult longevity
Adult fecundity
Field utilisation (from surveys)
Field utilisation (from open-jield tests)
Total
Nochoice
15
13
8
30
9
7
Choice
26
15
6
Choiceminus-target
6
3
5
82
4
56
9
Total
47
31
14
30
9
7
5
4
response under investigation. Tests on larval development, adult longevity and
fecundity can be only of the no-choice design, while field-survey and open field tests
are necessarily choice tests under natural or semi-natural conditions. Other responses
(oviposition, adult feeding and larval feeding) may be addressed in any of the test
designs. From the final column, we see that tests commonly examine oviposition and
larval development. In many studies these two tests are done together for a particular
agent to assess the behavioral preferences of adults (acceptability) and the suitability
of the plant for larval development and production of viable adults. If both responses
are measured together, results may be confounded. If, for example, the eggs are laid
internally in the plant and not easily scored, larval presence or damage or adult
emergence are the only indications of oviposition. In this case, one cannot discern
whether a negative score IS the result of unacceptability of the host for oviposition, or
unsuitability for larval development, or both.
When these two responses are measured separately, some practitioners subject
all test plants to both tests. Others only subject species accepted in oviposition tests to
larval development tests. A third option is to assess larval development on all plant
species and to assess suitability for oviposition only on those species that support
larval development. Because most immature insects have limited dispersal ability, larval
development trials are most important for plant species on which oviposition occurs.
Withers (1999) recommends that all test plants be subjected to both tests to fully
assess risk.
Larval development tests are prone to false positive results, because the
developmental host range is often wider than the range of plants used in nature. Hence
by themselves, larval development tests can result in the rejection of safe agents.
Sometimes, when tests assessing oviposition behavior are difficult (e.g. in many
Lepidoptera and Hemiptera), larval development tests cannot be done in conjunction
with oviposition tests. Larval development tests are still routinely used in the USA
where they often are called starvation tests. Where standardized testing procedures
are required, it is the only test that can be applied to nearly all agents.
Adult feeding tests are commonly investigated. These trials are restricted to
species that feed destructively and so are not used with those species that do not feed
(e.g., Cecidomyiidae) or that feed non-destructively (e.g., most Lepidoptera, Bruchidae,
Tephritidae). Adult feeding may not be important per se, in that only minimal damage
can be done to plants if larvae cannot develop on them. However, even cosmetic
-damagecanaffectpublicperception~o~biocontroI~results.~I~dia,~a~small_amountof~
feeding on sunflower by Zygogramma bicolorata Pallister (Coleoptera: Chrysomelidae)
a biocontrol agent of Parthenium hysterophorus L. (Asterales, Asteraceae) has almost
shut down the use of biocontrol agents to control weeds in that country.
Adult fecundity tests (also known as oogenesis tests) examine the ability of a plant
to support egg production. Their use is restricted to insects that depend on adult
feeding for continued egg production (Schwarzlaender et a/. 1996). They differ from
adult feeding or oviposition tests that test the acceptability of a host. Adult fecundity
tests examine both the acceptability of the food (behavior) and its suitability for egg
maturation (physiology). These tests are relevant only for plant species that adults
accept for feeding. Adult longevity tests often are performed also with adult fecundity
tests.
Field surveys and open-field tests are being used only occasionally but
increasingly. In open-field tests, the test plants are placed In the fleld in the native
range of the agent. Sometimes the local abundance of agents is increased. The
absence of a cage in these tests allows the assessment of all aspects of the host
selection process, so the results are more likely to predict the realized host range when
the agent is introduced into a new geographic range. Field surveys are fundamentally
the same as open-field tests, but in the latter plant spatial arrangements and agent
densities are not manipulated.
A STRATEGY FOR HOST-SPECIFICITY TESTING
This array of test combinations must be integrated into an overall experimental
strategy. More than one test usually is needed to determine host specificity. The
sequence in which tests are employed may vary; field tests can be done early, for
preliminary screening, or late, for clarification of equivocal lab results. Several tests may
be applied to all plant species on a list or one type of test can be used to eliminate
plants from further testing before a second test is applied to the reduced set of plants.
Alternatively a subset of plants may be tested initially to decide whether the agent is
sufficiently promising to justify further testing.
Some attempts have been made to combine the results of several tests to
calculate an index of plant suitability - a bottom line (Wan & Harris 1997). This is a step
towards a risk assessment approach, in which the risk to non-target plant species is
quantified. For example, the probability that a plant is accepted for oviposition
multiplied by probability of larval development estimates the probability of an adult
producing a pupa on the host plant. All plants on the test list should be tested for both
these components to make comparable calculations. Some practitioners resist this
approach because it would not seem necessary to test the suitability of a plant for larval
development if it is never accepted for oviposition. (Withers 1999). It may not be
desirable to develop a formulaic approach to host specificity testing. The current lack of
standardization in the type of test used reflects the unique aspects of the biology of
each agent as well as associated practical considerations.
BIOLOGICAL FACTORS AFFECTING TESTING STRATEGIES
Insect behavior involved in host selection is expressed during the conduct of tests and
can influence the results. Understanding the host selection behavior of potential agents
is the key to more effective host specificity testing.
Sequential behavioral responses t o host plant cues
There is a long held and widely accepted recognition that insects use a sequence of
behavioral responses to cues in host selection. The sequence of steps in host selection
includes habitat location, host location, host acceptance, and host utilisation (Keller
1999).
Consideration of the sequential behavioral steps to host selection raises a number
of issues that have consequences for host specificity testing. Possibly the most
important point is inability to express the early steps of the host selection sequence in
many experimental arenas If each step eliminates a certain number of potential hosts,
then the removal of that step from a host test may generate false positive results. This
process has been demonstrated with Drosophila magnaquinaria (Diptera,
Drosophilidae) (Kibota & Courtney 1991). In nature, this insect first selects its habitat
(low lying areas); within that habitat it then selects its only natural host (skunk
cabbage). If the insect is confined to a cage, it selects many more plant species for
oviposition than it does in nature, producing false positive results. Testing for host
range in insects such as this should incorporate this early step in the host selection
behavior.
Use of larger, more natural arenas, field surveys and open-field testing may
alleviate this problem. Less well-known methods to minimize the false positives in lab
tests include the use of wind tunnels or olfactometers, or simply provision of good
airflow through cages. Long-distance cues used in host location often rely heavily on
olfaction and the still air in cages does not allow for the upwind response of insects to
olfactory cues. If a potential agent gives a negative response to a plant species in an
olfactometer, that plant species may be eliminated as a potential host. That plant
species may have been accepted when the insect was confined to it in a cage.
Increased airflow through cages may provide a simple solution to allow some insects to
include olfactory cues in the host selection process (Keller 1999).
It is thus important to understand the critical steps in the host selection process for
the insect being tested; and include these steps in the test design. Some insects are
very specific in their habitat choice, e.g. the Drosophila on skunk cabbage. Others,
such as many Lepidoptera, use distant, pre-alighting host plant cues. Still other insects,
such as Aphis fabae Scop. (Homoptera, Aphididae), passively locate plants, but show
high specificity for chemo-tactile cues such as surface chemicals. Insects that depend
on pre-alighting host plant cues probably will not be tested accurately in small arenas,
but insects that passively locate hosts can be.
Learning
Learning can be defined as the modification of behavior due to the effects of prior
experience. The effects of experience (learning, memory and forgetting) are important
behavioral components of the host-selection process. Hymenopterous parasitoids are
well known for their abilities to learn. A smaller proportion of phytophages are known to
learn; they include Lepidoptera (adults and larvae), tephritid flies, Orthoptera and
Coleoptera (Papaj & Lewis 1993).A number of mechanisms are involved in learning; I
will examine two-- habituation and induction of preference.
Habituation is the decrease in response to a stimulus with repeated exposure to
that stimulus. Habituation to feeding deterrents is a common phenomenon in which an
insect initially is deterred but, after more exposure, begins to accept a plant.
Habituation can have consequences for host specificity testing. For example, insects
-may-habituateto-dete~~ts-ehon~ost~~gkFepeated~onta&-whil~onfine~withthem in cages resulting in eventual acceptance of those plants. False positive results
are produced in such cage tests. In nature, the insects are likely to leave the plant
before habituation occurs.
lnduced preference is the effect of experience on changes in food or oviposition
preferences. For example, in many species, newly hatched larvae will only accept
plants they first experience for further feeding. lnduced oviposition and adult feeding
preferences are influenced by early experience in many insects. For example, an adult
may respond to cues she learned when she emerged from her pupa to show
oviposition or feeding preference for the same species of plant on which she emerged.
lnduced larval feeding preferences can lead to false negative results, if insects begin
their feeding on the target weed and are then transferred to test species.
Other mechanisms of experience and learning can affect the results of host
specificity tests (Marohasy 1998, Heard, 2000). In general several tactics help avoid
unwanted consequences. First, whenever possible use nai've insects that have no
experience of any plants. Second, be aware of the false results certain tests can
generate and incorporate this information into their interpretation. Third, use different
test designs and compare results to assess which mechanisms may be operating.
Time-dependent effects
Time-dependent effects are reversible changes in the responsiveness of an organism
to a potential host resulting from food or oviposition site deprivation. As an insect
becomes more deprived it may become more likely to accept lower ranked hosts. A
hungry insect may feed on a wider range of hosts than a satiated one. An insect that
has recently laid an egg may reject a lower ranked host for a considerable period
(Withers et a/. 2000).
The main consequence of this effect is a false negative result in choice tests. For
example, Bruchidius villosus (F.) (Coleptera, Bruchidae), a seed feeder introduced into
New Zealand and Australia against broom, is attacking tagasaste (Chamaecytisus
palmensis), a non-target plant in New Zealand. This attack on tagasaste does not
represent a host range expansion, but a failure of host specificity testing to predict field
host range (Fowler et a/. 2000). Testing relied on choice tests alone and under the
conditions of this test, tagasaste was not attacked. Choice tests can generate false
negative results due to time-dependent effects. Insects may never reach a sufficiently
deprived state in the presence of the host plant to accept lower ranked hosts. However,
when in a no-choice situation, as happens in the field because tagasaste fruits are
available before broom, the insect may oviposit and feed on non-target species
A recent review of open-field testing has shown that this effect can take place
under natural conditions (Briese 1999). The results of open-field tests vary depending
on the experimental design. For example when the density of test plants is too low,
false negatives results are obtained because the insects never reach a sufficiently
deprived state.
To minimize these consequences, temporal patterns of feeding and oviposition
should be understood for each insect. This information should be incorporated into the
design of the host specificity tests. Second, no-choice trials should be continued for the
whole of the insect life to ensure that the insects become sufficiently deprived to accept
lower ranked hosts. Third, in open-field tests and field surveys, the target plant should
be removed to achieve a deprived state in insects.
I thank Stephen Hight and Julie Denslow of the USDA Forest Service, for providing the
opportunity to visit Hawai'i to present this paper. The useful comments of Lindsay
Barton Browne, Mic Julien and an anonymous reviewer improved the manuscript.
LITERATURE CITED
Briese, D. 1999. Open field host-specificity tests: Is "natural" good enough for risk
assessment? pp. 44-59, In: Host Specificity Testing in Australasia: Towards
Improved Assays for Biological Control. T. M. Withers, L. Barton Browne and J.
Stanley (eds), Scientific Publishing Queensland Department of Natural Resources,
Brisbane.
Fowler, S. V., P. Syrett, and P. Jarvis. 2000. Will the Environmental Risk Management
Authority, together with some expected and unexpected effects, cause biological
control of broom to fail in New Zealand? pp. 173-186, In: Proceedings of the ldh
International Symposium on Biological Control of Weeds. Bozeman, Montana,
USA, July 4-14, 1999. N. Spencer and R. Nowierski (eds). Montana State
University, Bozeman?
Heard, T. A., 2000. Concepts in insect host-plant selection behavior and their
application to host specificity testing pp 1-10, In- Host-specificity testing of exotic
arthropod biological control agents: the biological basis for improvement in safety.
R. G. Van Driesche, T. A. Heard and A. McClay (eds), Forest Health Technology
Enterprise Team, USDA Forest Service, Morgantown, West Virginia.
--
- - --
--
-
--
---
-
Heard, T. A. and R. D.Klinken van. 1998. An analysis of designs for host range tests of
insects for biological control of weeds. pp. 539-546, In: Pest Management - Future
Challenges, Sixth Australasian Applied Entomology Research Conference, Vol 7.
M. P . Zalucki, R. A. I. Drew, and G. G. White (eds), The University of Queensland
Printery. Brisbane.
Keller, M. A. 1999. Understanding host selection behaviour: the key to more effective
host specificity testing. pp. 84-92. In: Host Specificity Testing in Australasia:
Towards Improved Assays for Biological Control. T.M. Withers, L. Barton Browne
and J. Stanley (eds). Scientific Publishing Queensland Department of Natural
Resources, Brisbane.
Kibota, T. T. and S. P. Courtney. 1991. Jack of one trade, master of none: host choice
by Drosophila magnaquinaria. Oecologia 86: 251-260.
Marohasy, J. 1998. The design and interpretation of host-specificity tests for weed
biological control with particular reference to insect behaviour. Biocontrol News and
Information 19: 13N-20N.
D. R. Papaj and A. C. Lewis. 1993. Insect Learning: Ecology and Evolutionary
Perspectives. Chapman and Hall, New York.
Schwarzlaender M, H. L. Hinz, and R. Wittenberg. 1996 Oogenesis requirements and
weed biocontrol: an essential part in host-range evaluation of insect agents or just
wasted time? pp. 79-85, In: Proceedings of the 9th international symposium on
biological control of weeds, January 19-26 1996. V. C. Moran and J. H. Hoffmann
(eds). University of Cape Town, Rondebosch.
Wan F. H. and P. Harris 1997. Use of risk analysis for screening weed biocontrol
agents: Altica carduorum Guer. (Coleoptera: Chrysomelidae) from China as a
biocontrol agent of Cirsium arvense (L.) Swp. in North America. Biocontrol
Science and Technology 7 : 299-308.
Withers, T. M. (1999) Towards an integrated approach to predicting risk to non-target
species. pp. 34-40, In: Host Specificify Testing in Australasia: Towards lmproved
Assays for Biological Control. T . M . Withers, L. Barton Browne and J. Stanley
(eds), Scientific Publishing Queensland Department of Natural Resources,
Brisbane.
Withers, T. M., L. Barton Browne and J. Stanley. 2000. How time-dependent processes
can effect the outcome of assays used in host specificity testing. pp. 41, In: Hostspecificity testing of exotic arthropod biological control agents: the biological basis
for improvement in safety. R. G. Van Driesche, T. A. Heard and A. McClay (eds),
Forest Health Technology Enterprise Team, USDA Forest Service, Morgantown,
West Virginia.
Figure 1. Some test designs used in host specificity testing of biocontrol agents. Circles
represent plants, squares represent cage, and arrows represent the addition of a group
of insects.
Nochoice sequential
Choice
Nochoice simultaneous
~
-- -
Choiceminus-target simultaneous
plant 2
HOST SPECIFICITY AND RISK ASSESSMENT OF
HETEROPERREYIA HUBRICHI, A POTENTIAL CLASSICAL
BIOLOGICAL CONTROL AGENT OF CHRISTMASBERRY
(SCHINUS TEREBINTHIFOLIUS) IN HAWAI'I
Stephen D. Hight
U S. Forest Service, Pacific Southwest Research Station, Institute of Pacific Islands
Forestry and University of Hawai'i, Hawai'i Volcanoes National Park Quarantme Fac~l~ty,
P.O. Box 236, Volcano, HI 96785-0236, U.S.A.
Email: h ~ / ~ ~ ~ e lcom
t a _ p
Current Address U S D A , Agricultural Research Service & Center for Biological Control,
Room31OrPerry Paige Buildi~g;-South~Florida-A&M-University~allahassee,F~32307,
U.S.A.
Abstract. Heteroperreyiahubrichi, a foliage feeding sawfly of Schinus terebinthifolius, was studiedto
assess its suitability as a classical biological control agent of this invasive weed in Hawai'i. No-choice
host-specificitytests were conducted in Hawaiian quarantine on 20 plant species in 10 families. Adult
females oviposited on four test species. Females accepted the Hawaiian native Rhus sandwicensisas
an oviposition host equally as well as the target species. The other three species received dramatically
fewer eggs. Neonate larvae transferred onto test plants successfully developed to pupae on S.
terebinthifolius (70% survival) and R. sandwicensis (1% survival). All other 18 test plant species failed
to support larval development. A risk assessmentwas conducted to quantrfy the suitability of non-target
plants as a host to H. hubrichion the basis of the insects' performance at various stages in its life cycle.
Risk to all plant species tested was insignificant except R. sandwicensis. Risk to this native plant relative
to S. terebinthifoliuswas estimated at 1%. Currently this is too high a risk to request introductionof this
insect into the Hawaiian environment. Detailed impact studres in tne name range of S. tere6inthmliu.s
are needed to i d e n t i the potential benefit that this insect offers. Also, field studies in South America
with potted R. sandwicensiswould give more reliable analysis of this plants riskfrom natural populations
of H. hubrichi.
Key Words: Schinus terebinthifolius, Heteroperreyia hubrichi, Rhus sandwicensis,
Brazilian peppertree, Christmasbeny, classical biological control, host specificity, risk
assessment, non-target impacts.
INTRODUCTION
Schinus terebinthifolius Raddi (Sapindales, Anacardiaceae), locally known as
Christmasberry in Hawai'i, or Brazilian peppertree in Florida, is an introduced perennial
plant established throughout the Hawaiian Islands (Yoshioka and Markin 1991). This
species is native to Argentina, Brazil, and Paraguay (Barkley 1944, 1957) and was
brought to Hawai'i as an ornamental before 1900 (Neal 1965). As early as 1928, state
foresters began planting this tree in reforestation efforts on state forest reserves
throughout four of the main Hawaiian Islands (Skolmen 1979). The plant is a dioecious,
evergreen large shrub to small tree that has compound shiny leaves. Flowers of both
male and female trees are white and the female plant is a prolific producer of bright red
fruits. The green foliage and bright red fruits have been popular in Hawai'i for Christmas
wreaths and decorations (Wagner et al. 1990). A less common Hawaiian name for this
plant is "wile-laiki", named for Willie Rice, a politician who often wore a hat lei made of the
fruits (Neal 1965).
In Hawai'i, S. terebinthifolius has become an aggressive, rapidly spreading weed that
displaces native vegetation (Bennett et al. 1990, Cuddihy and Stone 1990). The plant
occurs from near sea level to about 920 m (Wagner et al. 1990). As early as the 1940's,
S. terebinthifolius was recognized as an important invader of dry slopes on Oahu (Egler
1942). Hawai'i Department of Agriculture recognizes the plant as a noxious weed (Morton
1978). Conservation organizations consider Christmasberry a high priority target in
Hawai'i because it is already widespread and has great potential to increase its range
even farther (Randall 1993). The U.S. Fish and Wildlife Service (1998) identified S.
terebinthifolius as one of the most significant non-indigenous species currently
threatening federally listed threatened and endangered native plants throughout the
Hawaiian Islands.
Naturalization of S. terebinthifolius has occurred In over 20 countries worldwide
throughout subtropical (15-30") areas (Ewel et al. 1982). Attributes of the plant that
contribute to its invasiveness include a large number of fruits produced per female plant,
an effective mechanism of dispersal by birds (Panetta and McKee 1997), tolerance to
shade (Ewel 1978), fire (Doren eta/. 1991), and d r o u g h t s e n and Mullerg80),andGiii
apparent allelochemical effect on neighboring plants (Medal et a/. 1999).
As a member of the Anacardiaceae, S. terebinthifolius shares its allergen causing
properties with other members of the family. While not affecting as many people as some
of the more notable members of the Anacardiaceae (poison ivy, poison oak, and poison
sumac), the plant sap can cause dermatitis and edema to sensitive people (Morton
1978). Resin in the bark, leaves, and fruit have been toxic to humans, mammals, and
birds (Ferriter 1997, Morton 1978). The lumber industry has deemed this plant of little
value due to its relatively low quality, its poor form due to the multiple, low stems, and the
poisonous, resin byproducts (Morton 1978). The sawdust and smoke are particularly
dangerous to sensitized people.
No control method is currently available against large, dense populations of S.
terebinthifolius. Mechanical removal with heavy equipment or chainsaws can be
acceptable along accessible areas, such as ditch banks, utility rights-of-ways, or other
disturbed areas (Ferriter 1997). Several herbicides and application methods are available
that a ~ d
in the control of S. terebinthifolfus (Ewel et al. 1982, Gioeli and Langeland 1997,
Laroche and Baker 1994, Woodall 1982). However, these non-biological methods are
labor intensive, expensive, and provide only temporary control due to the plant's
regenerative capability (Medal et a/. 1999). In addition, mechanical and chemical controls
are unsuitable over a large scale and in most natural settings because they are
detrimental to non-target organisms. The plant is intolerant of heavy shading and has
been know to die out under some plants, e.g., Schefnera actinophylla (Endl.) Harms
(Apiales, Araliaceae) (C. Smith, personal communication).
Classical biological control against Christmasberry was initiated in Hawai'i in the mid1900's (Yoshioka and Markin 1991). Surveys were conducted in South America (primarily
Brazil) for potential biological control agents (Krauss 1962, 1963). Three insect species
native to Brazil were released into Hawai'i: a seed-feeding beetle, Lithraeus (=Bmchus)
atronotatus Pic (Coleoptera, Bruchidae), in 1960 (Davis 1961, Krauss 1963); a leaf-rolling
moth, Episimus utilis Zimmerman (Lepidoptera, Olethreutidae), in 1954-1956 (Beardsley
1959, Davis 1959, Krauss 1963); and a stem-galling moth, Crasimorpha infuscata
Hodges (Lepidoptera, Gelechiidae), in 1961-1962 (Davis and Krauss 1962, Krauss 1963).
The first two species became established but were reported to cause only minor damage
(Clausen 1978, Yoshioka and Markin 1991). A seed-feeding wasp, Megastigmus
transvaalensis (Hussey) (Hymenoptera, Torymidae), accidentally introduced from South
Africa, has been found attacking seeds of Chrlstmasberry in Hawai'i since early 1970's
(Beardsley 1971, Yoshioka and Markin 1991).
Recent classical biological control efforts against S. terebinthifolius have been
focused in Florida since the late 1980's. This plant is listed as a Florida noxious weed
(FDACS 1994); it is displacing native vegetation in parks and natural areas (Bennett and
Habeck, 1991) and is estimated to infest over 4050 kmz (Habeck 1995). Exploratory
surveys for natural enemies in Brazil identified at least 200 species of arthropods
associated wlth S terebinthifolius (Bennett et al. 1990, Bennett and Habeck 1991, Medal
et a/. 1999). Based on field observations of thelr damage and lack of records that
indicate an association with cultivated plants in Brazil, several insects were selected as
biological control candidates for further study in Florida. Host specificity studies were
conducted on the sawfly Heteroperreyia hubrichi Malaise (Hymenoptera. Pergidae) in
Brazil and at a Florida quarantine facility (Medal et a/. 1999). Larval development and
female oviposition tests of H. hubrichi were conducted on 36 plant species in 15 families.
The insect was determined to be host specific to S. terebinthifolius and a request for
release of this insect into the Florida environment is currently under evaluation by Animal
and Plant Health Inspection Service, USDA (Medal et a/. 1999).
capitalizing on biological studies and host specificity tests conducted in Brazil and
Ftorida~a-bidogicaI-controlprojectwasinitiated to-evaluatefhe-potential-offWhubrichras
a control agent of S. terebinthifolius in Hawai'i (Hight et a/. in press). This paper presents
a synopsis of the investigation on the host range of H. hubrichi in Hawaiian quarantine
and a risk assessment for non-target plants.
MATERIALS AND METHODS
Twenty plant species underwent host specificity testing in the Volcano Quarantine
Facility. The selected plants belonged to one of three groupings: taxonomically
associated plants, habitat associated native plants, and habitat associated agricultural
plants (Table 1). Plant relatedness is based on the phylogenetic system of Cronquist
(1981). The order Sapindales has 15 families and four of these families (Anacardiaceae,
Rutaceae, Sapindaceae, and Zygophyllaceae) have native as well as introduced
members in Hawai'i. The single, native, Hawaiian species of Zygophyllaceae, Tribulus
cistoides L., was not tested because it occurs only in coastal habitats below 50 m
elevation (Wagner et el. 1990). Of the remaining 11 families, only members of the family
Meliaceae have been introduced into Hawai'i. Plants that make up the second group are
native plants that occur in the same habitat and are therefore likely to be exposed to any
introduced biological control agent. The second group is not as closely related to S.
terehjnthifolius, although members in three families (AraliaceaeIApiales,
MyrtaceaelMyrtales, and FabaceaeIFabales) are in the same subclass (Rosidae). The
third group contains two important, woody, crops that are found associated with S.
terebinthifolius habitat. These two species are in the same subclass as S.
terebinthifolius.
Insect Material.
Two shipments of H. hubrichi were imported from Brazil into the Hawai'i Volcanoes
National Park Quarantine Facility. The first shipment was received 19 November 1998
and consisted of 236 neonate larvae, which eclosed from four egg masses, and 192 late
instar larvae. The second shipment arrived 22 March 1999 and contained 101 late instar
larvae. Individuals of both shipments were collected in southern Brazil around the city of
Curitiba, Parana State. Quarantlne host speclflclty tests were conducted from
subsequent generations reared in captivity.
Both male and female adults can fly, although the male is a stronger flyer. Neither
the male or female adult H. hubrichi feed. However, both sexes were observed drinking
from small water droplets.
Adult Oviposition Tests.
No-choice oviposition tests were conducted in the quarantine facility on cut shoots for
each of the 20 test plant species. Tests were conducted in plastic containers holding a
single stem of a test plant (with 2 to 4 leaves). A mated female H. hubrichi was placed
on the test plant and if she oviposited, she remained inside the container with her eggs.
If a female did not oviposit on the test plant within 48 to 60 hr, she was removed and
placed in a new oviposition arena with a stem of S. terebinthifoliusto evaluate her
fecundity. Number of eggs laid and viability of eggs were recorded. For all plant species,
tests were replicated at least four times. Citms sinensis (L.) Osbeck (Sapindales,
Rutaceae) was not tested because of lack of plant material. Tests on this plant in Florida
and Brazil found this to be an unacceptable host plant.
Oviposition tests were conducted on potted plants of five test species on which
oviposition has occurred and/or on whlch larvae had developed on cut shoots. A mated
female was placed on the caged test plant until she died. The number of eggs laid and
viability of eggs was recorded. Each plant species was replicated at least six times.
No-Choice Larval Development Tests.
All test plants were evaluated as to their ability to support larval development under nochoice conditions. Unfed, neonate larvae less than two hours old were transferred to
small cut shoots of the test plant stuck into moistened florist-foam-filled vials and reared
in 480 ml plastic containers. Since larvae feed gregariously, 15 larvae were transferred
into each container with a fine tip brush. Each test plant was replicated at least six times.
For each family of larvae used in the tests, 2-3 replicates of 10-15 larvae were reared on
S. terebinthifolius to insure the vitality of each egg mass. Containers were cleaned,
larvae were fed, and mortality was assessed on the third day after transfer and then every
fourth day. Containers were evaluated every day after larvae became sixth instars.
Larval development tests were also conducted on potted plants of five test species
because of oviposition activity and/or larval development on cut shoots. Each plant had
an egg mass of H. hubrichi either naturally ovlposlted on the stem or tied onto a stem
from a successful oviposition on S. terebinthifolius. The number of larvae that
successfully developed on each test plant was recorded. The test was replicated on
each plant species at least three times.
Relative Host Suitability
In an attempt to quantify potential suitability of non-target plant species for agent
development, Wan and Harris (1997) developed a scoring system that compares the
suitability of non-target species to that of the target species. I have followed their
approach to obtain estimate host suitability. The index of suitability of a non-target host
plant for H. hubrichi use is R, x R2x ... R,, where R is the performance of the insect at
various life stages on the test plant relative to that on S. terebinthifolius. Suitability
parameters estimated for each test plant species included the proportion of females that
oviposited on the plant, number of eggs oviposited, proportion of larvae that survived, and
development time of larvae fram eggs to pupae (Table 2). For purposes of calculation,
zero values for any parameter (complete rejection or failure) were taken to be 0.001.
RESULTS
Insect Biology.
The adults of H. hubrichi are generally black with yellow legs. A female and male H.
hubrichi mate on the surface of soil or plants, although females do not need to mate for
oviposition to occur. Each female oviposits her eggs in a single mass just into the
surface of non-woody stems. Eggs in a mass are arranged in rows and the female
"guards" her eggs until she dies, just before the eggs hatch. Eggs hatch in 14 days.
Neonate larvae feed gregariously on both surfaces of young leaflets at the tip of shoots.
As they grow they move as a group onto new leaflets and larger leaves until the third to
fourth instar when they disperse throughout the plant and feed individually. A larva is
green with red spots and black legs. After reaching the seventh instar, the larva moves
into soil and pupates. Insects reared on S. terebinthifolius took 26-42 days from egg
natch to pupation. Pupation lasted two months for 80% of pupae and the longest
successful pupation occurred in seven months.
Adult Oviposition Tests.
Female H. hubrichi oviposited on cut shoots of five different test plant species (Table 1).
All females that were placed on S. terebinthifolius and R. sandwicensis oviposited on
their test plant. Less than half of the females placed on the other four test species
su c c e ~ s h l l y ~ o ~te-6onnt
f p ~ s i h~ir-t~~t~pta~nt~(Tabte~2)~o~e'~e~~aII-n-on~0vip~~itinsfema1es
successfully oviposited once they were moved onto S. terebinthifolius afler the 48-60 hr
test period. This indicated that the females were capable of ovipositing on the test plant
but rejected that plant species as an oviposition host.
Mean number of eggs oviposited by females on each test plant species is presented
in Table 1. There was no significant difference in the number of eggs deposited on R.
sandwicensis and on S. terebinthifolius (t-test; p> 0.05; t(43)= 1.762). Oviposition on the
three Sapindaceae test plant species was highly variable with most tests receivifiq i ~ c ,
eggs. In those plants receiving eggs, the average number deposited was high:
Dodonaea viscosa 57 eggs; Litchi chinensis 78 eggs; and Euphoria longan 56 eggs.
Oviposition was more restricted on potted plants than on cut shoots. Mated H.
hubrichi females oviposited on only three of the five species of potted test plants (S.
terebinthifolius, R. sandwicensis, and E. longan). Females did not oviposit on potted D.
viscosa or L. chinensis, even though oviposition did occur on cut shoots of D. viscosa
and L. chinensis.
- --
-- -
-
-
-
No-Choice Larval Transfer Tests.
Neonate larvae successfully developed on cut shoots of only two test plant species, S.
terebinthifolius and R. sandwicensis. Larvae on most of the other test plant species were
deac! within seven days (Table 1). Although cut shoots of two additional plant species
supported some larval development for more than two weeks, (Mangifera indica
(Sapindales, Anacardiaceae), 23 d and E. longan (order, family) 19 d), no larvae survived
to pupation.
Successful larval development on the five potted plant species was similar to
development on cut shoots. Larvae developed only on potted S. terebinthifolius and R.
sandwicensis. The proportion of larvae successfully developing on S. terebinthifolius and
R. sandwicensis potted plants was slightly higher to the proportion on cut shoots of those
two test plant species (78% and 4%, respectively).
Relative Host Suitability
The relative host suitability of the test plant species is shown in Table 3. Suitability
estimates are calculated only for the five plant species that received eggs from
ovipositing females. Scores for all four non-target plants were lower than for S.
terebinthifolius, measured at 1.O. All other 15 tested species were unacceptable host for
both oviposition and larval development and are not at risk by the release of H. hubrichi
into Hawai'i.
DISCUSSION
Field observations in Brazil and laboratory feeding tests in Florida indicated that H.
hubrichi was highly host specific and safe to release into the Florida environment (Medal
et a1 1999). Additional host specificity studies in quarantine primarily on native Hawaiian
plants confirmed a highly limited host range for H. hubrichi. Tests at all locations showed
that S. terebinthifolius was the preferred, if not the only host plant of H. hubrichi.
However, the potential host range in Hawai'i appears to be slightly broader than that
identified in Florida and Brazil. Tests in Florida evaluated two North American species of
sumac (R. copallina and R, michauxir) and found them unsuitable for H. hubrichi
oviposition and incapable of supporting larval development (Medal et.al. 1999, J. Cuda,
personal communication). Hawaiian tests indicated that the Hawaiian sumac (R.
)did support larval development and was hlgllly attractive to the female for
oviposition. Chemicals still present in ancestral, continental species that deter
herbivorous insects may have been lost over time in the Hawaiian sumac. Of the five
varieties of S. tembinthifolius recognized in South America (Barkley 1944), H. hubrichi
prefers the most pubescent variety (M. Vitorino, personal communication). The dense
pubescent nature of R. sandwicensis may stimulate female oviposition regardless of the
quality of the plant for larval development. Both S. terebinthifolius and R. sandwicensis
were comparable in their acceptance by ovipositing females as measured by proportion of
females that oviposited on the test plant and the number of eggs that a female laid. But
R. sandwicensis was a dramatically poor host for H hubrichi larvae in both performance
characteristics of larval survival and development time.
To identify the potential non-target effect that native R. sandwicensis might be
exposed to because of the introduction and release of H. hubrichi into Hawai'i a host
suitability assessment was conducted. To arrive at realistic estimates of host suitability,
both physiological and behavioral processes must be estimated (McEvoy 1996).
Estimates of host suitability were determined by quantifying crucial stages in the sawflies
sequence to locate, accept, and develop on the host, i.e., oviposition by the female, larval
development time, and larvae survival rate from egg to pupa in no-choice tests.
Relative host suitability of non-target species was evaluated for the five test plant
specres that experienced any establishment and/or damage from H. hubrlchi in the host
specificity tests. Four plant species had extremely low levels of suitability (Table 3). In
fact, since all four of these plants completely failed to support larval development it may
be argued that their suitability for H. hubrichi development is zero. The life cycle of H.
hubhchi would be interrupted If the insects were to colonize any one of these plants and a
population of H. hubrichi would fail to establish. A low suitability level was measured for
R. sendwicensis (approximately 1 94).
Introduction of H. hubrichi into Hawai'i will not be requested at this time because of
the apparent risk to R. sandwicensis. However, additional information is being sought
which may reverse this decision. Field experiments in Brazil with potted R. sandwicensis
are being proposed to evaluate the risk of this non-target plant under more natural
settings. The observed host range of herbivorous insects is often wider under laboratorybased tests than open-field tests (Cullen 1990. Briese 1999). In addition. the risk inherent
in introducing a biological control agent may be outweighed by its benefit. Therefore,
detailed impact studies are needed in Brazil to evaluate the effect H. hubrichi has on S.
terebinthifolius fitness. Neither of these types of tests is currently funded.
Additional surveys for phytophagous insects of S. terebinthifolius should be
conducted in northern Argentina, the most likely center of origin of this species (Barkley
1944). Virtually all previous South American explorations by scientists from Hawai'i
(Krauss 1962, 1963) and Florida (Bennett et a/. 1990, Bennett and Habeck 1991) have
-
p
taken place in southern Brazil. Although this work has identified several promising
biological control candidates, additional surveys may be more successful in Argentina.
For example, on a 10-day survey in January 2000 of S. terebinthifolius natural enemies in
the state of Missiones, Argentina, two species of stem boring Cerernbicidae and a bark
girdling Buprestidae were collected (S. Hight, unpublished data). Identifications of these
insect species are pending. No stem boring or bark girdling insects were identified from
Brazilian surveys.
ACKNOWLEDGMENTS
I thank Clifford Smith and Julie Denslow for comments on earlier drafls of this manuscript.
I am grateful to Jonathan Chase, Donovan Goo, Baron Horiuchi, Brian Kiyabu, Patti
Moriatsu. Roddy Nagata, Wendell Sato, and Alan Urakami for collecting, propagating,
-and/or-donating-testpIa~t~~(se&in-this-resear.MaccelleVitorinoand-Hendque--Pedrosa-Macedo are acknowledged for their assistance in collecting H, hubrichi in Brazil.
The National Park Service is acknowledged for their support and use of the Hawaii
Volcanoes National Park Quarantine Facility. A special thanks to Ivan Horiuchi for
technical assistance throughout this study and to Clifford Smith for his sustained help and
guidance.
LITERATURE CITED
Barkley, F. A. 1944. Schinus L. Brittonia 5: 160-198.
Barkley, F. A. 1957. A study of Schinus L. Lilloa Revista do Botanica. Tomo 28.
Universidad Nacional del Tucumen, Argentina.
Beardsley, J. W. 1959. Episimus sp. Pmceedings, Hawaiian Entomological Society 17:
28.
Beardsley, J. W. 1971. Megastigmus sp. Proceedings, Hawaiian Entomological Society
21: 28.
Bennett, F. D., L. Crestana, D. H. Habeck, and E. Berti-Filho. 1990. Brazilian peppertree
- prospects for biological control. pp. 293-297, In: Proceedings of the VII
lntemational Symposium on Biological Control of Weeds, 6-11 March 1988, Rome,
Italy. E.S. Delfosse (ed), lstituto Sperimentale per la Patologia Vegetale (MAF),
Rome, Italy.
Bennett, F. D. and D. H. Habeck. 1991. Brazilian peppertree - prospects for biological
control in Florida. pp. 23-33, In: Proceedings of the Symposium of Exotic Pest
Plants, 2-4 November 1988, Miami, FL. T. D. Center, R. F. Donn, R. L. Hofstetter,
R. L. Myers, and L. D. Whiteaker (eds), U.S. Department Interior, National Park
Service, Washington, DC.
Briese, D. T. 1999. Open field host-specificity tests: Is "natural" good enough for risk
assessment? pp. 44-59, In: Host Specificity Testing in Australasia: Towards
Improved Assays for Bialogical Control. T. M. Withers, L. Barton-Browne, and J.
Stanley (eds), CRC for Tropical Pest Management, Brisbane, Australia.
Clausen, C. P. (ed). 1978. Introduced Parasites and Predators of Arthropod Pests and
Weeds: A Wodd View. Agriculture Handbook 480. Agricultural Research Service,
USDA, Washington, DC.
Cronquist, A. 1981. An Integrated System of Classification of Flowering Plants,
Columbia University Press, New York.
Cuddihy, L. D. and C. P. Stone. 1990. Alteration of Native Hawaiian Vegetation: Effects
on Humans, Their Activities and Introductions. University of Hawai'i Press, Honolulu,
HI.
Cullen, J. W. 1990. Current problems in host-specificity screening. pp. 27-36, In:
Proceedings of the VII International Symposium on the Biological Control of Weeds,
6-11 March 1988. E. Delfosse (ed), lstituto Sperimentale per la Patologia Vegetale,
Rome, Italy.
--
Davis, C. J. 1959. Recent introductions for biological control in Hawaii - IV. Proceedings,
Hawaiian Entomological Society 17: 62-66.
Davis, C. J. 1961. Recent introductions for biological control in Hawaii - VI. Proceedings,
Hawaiian Entomological Society 17: 389-393.
Davis, C. J. and N. 1.H. Krauss. 1962. Recent introductions for biological control in
Hawaii VI I. Proceedings, Hawaiian Entomological Society 18: 125-129.
-
Doren, R. F., L. D. Whiteaker, and A. M. LaRosa. 1991. Evaluation of fire as a
management tool for controlling Schinus terebinthifolius as secondary successional
growth on abandoned agricultural land. Environmental Management 15: 121-129.
Egler, F. E. 1942, Indigene versus alien in the development of arid Hawaiian vegetation.
Ecology 23: 14-23,
Ewel, J. J. 1978. Ecology of Schinus. pp. 7-21, In: Schinus: Technical Proceedings of
Techniques for Control of Schinus in South Florida: A Workshop for Natural Area
Managers, December 2, 1978. The Sanibel Captiva Conservation Foundation, Inc.,
Sanibel, FL.
Ewel, J. J., D. S. Ojirna, K. A. Karl, and W. F. DeBusk. 1982. Schinus in Successional
Ecosystems of Everglades National Park. South Florida Research Center Report T676. USDI, National Park Service, Washington, DC.
-
FDACS Florida Department of Agriculture and Consumer Services. 1994. Biological
Control Agents: Introduction or Release of Plant Pests, Noxious Weeds, Arthropods,
and Biological Control Agents. FDACS, Gainesville, FL.
Ferriter, A. (ed), 1997. Brazilian Pepper Management Plan for Florida:
Recommendationsfrom the Brazilian Pepper Task Force, Florida Exotic Pest Plant
Council. The Florida Exotic Pest Plant Council, Florida.
Gioeli, K, and K. Langeland. 1997. Brazilian Pepper-tree Contml, SS-AGR-17
University of Florida, Cooperative Extension Service, Gainesville, FL.
Habeck, D. H. 1995. Biological control of Brazilian peppertree. Florida Nature 68: 9-11.
Hight, S. D., I. Horiuchi, M.D. Vitorino, C.Wikler, and J.H. Pedrosa-Macedo. 2002.
Biology, host specificity, and risk assessment of the sawfly Heteropemyia hubrichi, a
potential biological control agent of Schinus terebinthifolius in Hawaii. Biological
Control. Accepted.
Krauss, N.L.H. 1962. Biological control investigations on insect, snail and weed pests in
tropical America, 1961. Proceedings, Heweiien Entornologicel Society 18: 131-1 33.
Krauss, N. L. H. 1963. Biological control investigations on Christmas berry (Schinus
tembinthifolius) and emex (Emex spp.). Proceedings, Hawaiian Entomological
Soc7efyt8:-281287.
--
--
Laroche, F. B. and G. E. Baker. 1994. Evaluation of several herbicides and application
techniques for the control of Brazilian pepper. Aquatics 16: 18-20.
McEvoy, P. B. 1996. Host specificity and biological pest control. BioScience 46: 401405.
Medal, J. C., M. D. Vitorino, D. H. Habeck, J. L. Gillmore, J. H. Pedrosa, and L. D. De
Sousa. 1999. Host specificity of Heterupenwyia hubrichi Malaise (Hymenoptera:
Pergidae), a potential biological control agent of Brazilian Peppertree (Schinus
tembinthifolius Raddi). Biological Control 14: 60-65.
-
Morton, J. F. 1978. Brazilian pepper its impact on people, animals and the environment.
Economic Botany 32: 353-359.
Neal, M. C. 1965. In Gardens of Hawaii. B.P. Bishop Museum Special Publication 50.
Bishop Museum Press, Honolulu, HI.
Nilsen, E. T. and W. H. Muller. 1980. A comparison of the relative naturalization ability of
two Schinus species in southern California. I. Seed germination. Bulletin Tomy
Botanical Club 107: 51-56.
Panetta, F. D. and J. McKee. 1997. Recruitment of the invasive ornamental, Schinus
terebinthifolius, is dependent upon frugivores. Australian Journal Ecology 22: 432438.
Randall, J. M. 1993. Exotic weeds in North American and Hawaiian natural areas: The
Nature Conservancy's plan of attack. pp. 159-172, In: Biological Pollution: The
Contml and Impact of lnvasive Exotic Species. B.N. McKnight (ed). Indiana Acad.
Sciences, Indianapolis, IN.
Skolmen, R. G. 1979. Plantings on the Forest Reserves of Hawaii, 1910-1960. U.S.
Forest Service, Institute of Pacific Islands Forestry, Honolulu, HI.
U.S. Fish and Wildlife Service. 1998. Draft Recovery Plan for Multi-Island Plants. U.S.
Fish and Wildlife Service, Portland, OR.
Wagner, W. L., D. R. Herbst, and S. H. Sohmer. 1990. Manual of the Flowering Plants of
Hawai'i. University of Hawai'i Press, Honolulu, HI.
Wan, F. and P. Harris. 1997. Use of risk analysis for screening weed biocontrol agents:
Altica carduorum Guer. (Coleoptera: Chrysomelidae) from China as a biocontrol
agent of Cirsium arvense (L.) Scop. in North America. Biocontml Science and
Technology 7 : 299-308.
Woodall, S. L. 1982. Herbicide Tests for Control of Brazilian-pepper and Melaleuca in
Norida. USDA Forest Service Research Note SE 314. Southeastern Forest
Experiment Station, Asheville, NC.
l o shiaka,-E-R .-ancLGGGP4arki
n~199~tI:ff~d~~0_fbi~L0~~al~~0~ol~ofChri~stma~sber
(Schinus terebinthifolius) in Hawaii. pp. 377-387, In: Pmceedings of the Symposium
of Exotic Pest Plants, 2-4 November 1988, Miami, FL. T. D. Center, R. F. Doren, R.
L. Hofstetter, R. L. Myers, and L. D. Whiteaker (eds), U.S. Department Interior,
National Park Service, Washington, DC.
I
Euphoria
Alectryon
Nephelium
I
Citrus sinensis
B. Habitat Associated Native
Plants
Araliaceae
Reynoldsia
sanalvicensis*
Myrtaceae
Metmsidems
polymorpha*
Fabaceae
Acacia koa*
Sophora
chrysophylla*
Myoporaceae
Myoporum
sandwicense*
Table 3. Relative host suitability analysis for Heteropemyia hubrichi performance on
various host plant test species in Hawaiian quarantine.
Performance ~easure'
1
2
3
4
Relative Suitability
S. terebinthifolius
1-000
1 000
1-000
1.OOO
1.OOO
R. sandwicensis
1.000
1.026
0.778
0.011
0.009
D. viscosa
0.038
0.043
0.001
0.001
.
x
3.6
0.438
0.296
0.001
0.001
1.3 x
0.417
0.200
0.001
0.001
8.3 x
Plant Species
L. chinensis
E. longan
'
-.~
la9
Performance measure estimates are relative to S. terebinthifolius. 1 = proportion of
females that oviposit; 2 = mean number of eggs oviposited; 3 = mean development time of
larvae; 4 = proponion of eggs that survive to pupae.
BIOLOGICAL CONTROL POTENTIAL OF MICONIA CALVESCENS
USING THREE FUNGAL PATHOGENS
Eloise M. Killgore
Biological Control Section, Hawai'i Department of Agriculture, P. 0. Box 22159,
Honolulu, Hawai'i 96823-2159, U.S.A.
Emai I: eloiseki/[email protected]
Abstract. Biological control of miconia (Miconia calvescens) became a management option as
soon as the severity of its threat to Hawaiian ecosystems was recognized. No weed in Hawai'i
has received as much publicity, attention, and funding for control. Three fungal pathogens
have been considered as potential agents. Co/letotrichum gloesoporioides f. sp. miconiae was
assessed within six months, the petition for release approved within eight months and the
fungus released on the islands of Hawai'i and Maui in 1997. It is established on Hawai'i and has
spread to other areas. Its effectiveness is under evaluatlon. Pseudocercospora tamonae
causes extensive damage to leaves, attacks other melastomes and the seedlings of some
Myrtaceae but only fruits on miconia. It is very uncertain whether this species will be approved
for release. Coccodiella myconae produces large wart-like growths that deform leaves
considerably. It appears to be an obligate parasite of miconia but hyperparasitized by another
species tentatively identified as Sagenomella alba. It has proven difficult to transfer from one
plant to another where it does not sporulate. Further work on this species is in progress.
Key words: Biological control of weeds, Coccodiella myconae, Colletotrichum
gloeosporioides f. sp. miconiae, invasive species, Melastomataceae, Miconia
celvescens, Pseudocercospora tamonae, Sagenomella alba.
THE TARGET WEED, MICONIA CALVESCENS
Early Considerations
Davis (1978) first expressed concern about the threat of Miconia calvescens DC
(Myrtales, Melastomataceae) in Hawai'i. In January 1991, a miconia tree with seedlings
was discovered in a plant nursery near Hana, Maui. By midyear, surveys revealed
additional populations in the surrounding area including mature flowering trees (Gagne
et a/. 1992). The threat from this plant generated considerable publicity (Altonn 1991,
Eager 1995, Tanji 1996a,b, Yohay & Fukutomi 1997). The resilience, dominance of
emergent miconiap1mts;form~0~rsotgpi~stmd~their-fecundity~widespreactdissemination by frugivorous birds, the erosion associated with the shallow-rooted
miconia trees and its deleterious impact on native forests of Tahiti and Moorea were
described by Meyer (Meyer 1994,1996). Medeiros et a/.(1997) noted that the similarity
in stature, composition, and requirements of climate of Hawaiian and Tahitian forests
strongly suggested similar vulnerability. The publicity stimulated many private and
public agencies to support control activities
Melastome Action Committees
The miconia problem generated considerable support but it lacked leadership. Maui
Land and Pineapple Company had similar concerns with another weed Tibouchina
hehacea (DC) Cogn. (Myrtales, Melastomataceae). The combined interest led to the
organization of the Melastome Action Committee (MAC) committee by Tri-Isle Resource
Conservation and Development Office, U.S. Department of Agriculture (Medeiros
1998). Many private and government agencies participated: Biological Resources
Division U.S. Geological Survey; U.S. Fish and Wildlife Service; U.S.D.A. Forest
Service; National Park Service; Hawai'i Army National Guard; Hawai'i Department of
Agriculture (HDOA); Hawai'i Department of Land and Natural Resources; University of
Hawai'i; Maui County's Office of Economic Development; and, The Nature
Conservancy (Conant 1996).
This proactive committee furthered public awareness and planned a broad control
strategy against miconia principally on the island of Maui. MAC focused primarily on
the containment of miconia using chemical and mechanical eradication because the
bulk of the funds were dedicated for local use only. Other funding was obtained for
initial exploratory work on biological control.
A somewhat similar committee was established later on Hawai'i with funding again
focused on local control efforts. Different control strategies were taken because the
principal infestations on Hawai'i were on small parcels of private property whereas
those on Maui were on large areas of primarily state lands.
Within a few years MAC became the Maui lnvasive Species Committee. The
Hawai'i committee was later integrated with the Big Island lnvasive Species Committee.
The focus on the management of melastomes diminished in light of other problems
resulting in a much less focused effort against miconia. Funds for biological control
were decreased significantly and would have ceased but for the persistence of one or
two people. Their efforts have established new funding for renewed exploratory and
biological studies for potential invertebrate agents in Costa Rica and Brazil.
BIOLOGICAL CONTROL
The search for natural enemies of miconia initially was the responsibility of HDOA.
Their staff included an experienced exploratory entomologist and many biological
control researchers as well as limited insect and pathogen quarantine facilities.
Exploration
The initial exploratory studies between July 1993 and September 1995 were conducted
in Costa Rica, Brazil, Paraguay, and Trinidad and Tobago. Many insect feeders and
several diseases were collected and sent into quarantine in Hawai'i (Burkhart 1996).
Burkhart recommended that biological control studies focus on the pathogens because,
apart from some processionary caterpillars (Lepidoptera, Riodinidae), the insects did
-notapeea~~ave-m~~pa~-~~-the~~e\l-werepr~bab~-~~~host-sp
In 1996, Barreto began exploration for diseases of miconia in Brazil. He shipped many
fungi that were isolated from miconia plants into quarantine in Hawai'i. He also traveled
to the Dominican Republic and Costa Rica to relocate three of the fungi mentioned by
Burkhart. He later collected the same fungi in Brazil and recommended all three
pathogens for further screening.
Further exploration was conducted in 1996 and 1998 in Guatemala where plants
closer to the Hawaiian biotype were thought to exist (Killgore 1996, Ramadan 1998).
Very sparse populations of miconia were found throughout the country. Failure to
establish collaboration with Guatemalan scientists dimmed prospects for further
research on several prospective biological control insects, particularly a Margardisa sp.
(Coleoptera, Chrysomelidae), found in the Peten and Coban areas of the country. No
new pathogens were found.
Barreto (2000) explored several regions of Ecuador for diseases of miconia but
found nothing new. With little hope for new diseases the search for other collaborators
to reconsider insects in Central America resulted in the establishment of a cooperative
agreement with the University of Costa Rica. A psyllid has already been recommended
and interest in at least one species in the Riodinidae rekindled.
Host Range Testing
The widely accepted protocol for selecting host range test plants follows the centrifugal
phylogenetic method first proposed by Wapshere (1974). Members of the family
Melastomataceae were challenged first, followed by species in other families within the
Order Myrtales.
All members of the Melastomataceae in Hawai'i are naturalized alien species
-(Ahrreda-t990)~Mostareweedyandnoxio~ndmch;rd~idemiaf7trt~L+D;Bo~-Tibouchina herbacea (DC) Cogn., T. longifolia (Vahl) Baill. ex Cogn., T. unMeana (DC)
Cogn., Adhrostema ciliatum Pav. ex D.Don,Oxsypora paniculata (D.Don) DC, Medinilla
venosa (Blume) Blume, Dissotis rotundifolia (Sm.) Trian, Heterocentron subtriplinervium
(Link & Otto) A. Braun & C. Muell., Melastoma candidurn D. Don, M. sanguineum Sims,
Pterolepis glomerata (Tottb.) Miq. Tetrazygia bicolor (Mill.) Cogn., and Trembleya
phlogifomis DC. Because none of these plants is endemic, indigenous, or
economically important in Hawai'i except as ornamentals, a pathogen of M. calvescens
capable of infecting these melastomes could still remain a candidate for biological
control. A pathogen, that did not infect any other melastome, was not expected to
infect any plant species outside of the family and was deemed highly specific and
desirable for biological control purposes. The need for careful, thorough screening
(Watson 1985) was confirmed by results with Pseudocercospofa famonae (see below).
Other than the family Melastomataceae, there are only 19 genera within the Order
Myrtales that occur in Hawai'i (Wagner et a/. 1990). Some are endemic, indigenous or
have economic importance including Teminalia L. (Combretaceae), Cuphea P. Br.
(Lythraceae), Lyfhnrm L. (Lythraceae), Eucalyptus L'Her. (Myrtaceae), Eugenia L.
(Myrtaceae), Metrosideros Banks ex Gaertn. (Myrtaceae), Psidium L. (Myrtaceae).
Syzygium Gaertn. (Myrtaceae), and Wikstroemia Endl. (Thymelaceae). Susceptibility of
any of the plant species belonging to these genera would necessitate further
evaluation.
Host range tests are typically performed under the most advantageous conditions
for the pathogen. For fungi, these conditions would normally include a very high
concentration of spores of 1 x -lo5per ml or higher, and a relative humidity of 100% for
16 to 48 hours. Host range testing under these conditions often indicates a broader
host range than is found under field conditions (Watson 1991), and may distort the
evaluation of maximum risk posed by a biological control candidate resulting in the
rejection of agents that are really safe and useful.
The pathogens
Colletotrichum qloeosworioides (Penz.) Sacc. f. sp. miconiae (Phvllachorales,
Phllachoraceae).-This fungus causes leaf spots on miconia resulting in leaf yellowing
and premature defoliation. It reproduces by asexual spores, or conidia, produced in
acervuli that arise on the abaxial leaf surface. Setae are sometimes observed in these
structures. The conidia are 14.7 -17.5 pm long and 5.0 4 . 2 5 pm wide, straight,
cylindrical, with rounded ends. Conidia of Colletotrichum fungi are produced under high
humidity conditions and are disseminated by "wind-driven rain" (Trujillo, pers. corn.). All
test plants were inoculated using a spore concentration of 1 x lo5 conidialml in sterile
water and incubated for 48 hours in an enclosed chamber at 100% relative humidity.
Host range testing, concluded in November 1996, showed that this fungus was
restricted in pathogenicity at least to the genus Miconia. The most closely related
species, Clidemia hirta, was not susceptible. This form of the fungus did not infect any
other species in the Melastomataceae, which occur in Hawai'i, or any member of the
other families in the order Myrtales. Based on the host specificity tests, the fungus was
subsequently described as f.sp. miconiae of Colletotrichum gloeosponoides (Killgore et
al. 1999).
The request to release C. 9. miconiae from the containment facility was submitted
in December 1996 to HDOA for state approval, which was granted on March 3, 1997,
and forwarded to USDA APHIS PPQ. The Federal permit was signed on July 11,1997
- ( K i U g o r e e t a l , I 9 9 8 ) , - a ~ d - t h e f u n g u s c e l e a ~ r ~
and on Maui later that year. Attempts to establish the disease on Maui have failed so
far. The fungus requires rain and wind for spore dissemination and recent drought
conditions may have limited its spread. Studies on the impact of the fungus are
ongoing.
Under aseptic laboratory conditions the fungal pathogen attacked germinating
miconia seeds and also kills newly emergent seedlings (Su Que Leong, pers. corn.). If
these observations are confirmed in the field they suggest that C. g. miconiae could
play an even more important role as a biological control agent for M. calvescens.
Pseudocercospora tamonae (Chupp) Braun (Deuteromvcotina, Dematiaceae1.-This
fungus was isolated and identified by Barreto from leaves of Miconia phanerostila Auth.
(Myrtales, Melastomataceae) and also from M, calvescens from Rio de Janeiro, Brazil.
The minute leaf spots did not appear to be damaging, but when pathogenicity tests on
miconia were completed, the pathogen proved to be more aggressive than C. g.
miconiae. Besides causing numerous leaf spots, infected leaves became deformed and
defoliation occurred three weeks after inoculation.
The conidia are pale brown, cylindrical, tapering towards the apices, 53-90 x 3 -5
pm with 6-1 1 septa. They are produced on conidiophores, which emerge from
diseased leaf tissue. High humidity favors spore production and spores are easily
disseminated by air currents.
Unlike the C. g. miconiae, however, Pseudocercospora tamonae was not as host
specific. It infected other species within the Melastomataceae including Clidemia hirta,
Arthrostema ciliata, Dissotis mtundifolia, Tibouchina urvilleana and T. herbacea. It also
infected members of the Myrtaceae, including Metrosideros polymorpha Gaud.,
Psidium cattleianum Sabine and Syzygium malaccense (L.) Merr. & Perry. More leaf
spots developed on M. calvescens leaves than on any other susceptible plant species,
and sporulation of P. tamonae was only observed on miconia. Leaf yellowing and
premature defoliation occurred only on those susceptible genera of the
Melastomataceae. Only immature leaves of M. polymorpha, P. caftleianum and S.
malaccense became infected, and there was no defoliation. The request to release this
pathogen has been submitted. It is expected that the wider host range suggested by
the host range test conducted for this fungus will be accepted as representing an
aberration due to unnatural experimental conditions. Infections are obtained in the test
only when leaves are exposed under very high humidity. The pathology is not
ecologically significant as the fungus was unable to complete its life cycle from spore to
spore on the non-target plants.
Coccodiella myconae (Dubv) Hino & Katuamoto (Phyllacorales. Phyllacoraceae).Coccodiella myconae was collected in Costa Rica and Brazil Barreto, who noted that it
should be transferred to Coccodiella. It is extremely common in Brazil but always
associated with the hyperparasite provisionally identified as Sagenomella alba W.
Gams & Soederstrom (Eurotiales, Trichocomaceae). Infection produces wart like
growths on the adaxial leaf surface with a corresponding concavity on the abaxial side
resulting in the infected leaf becoming deformed and falling prematurely. Dark brown
to black stromata line the concavities in a somewhat circular pattern. Perithecia
develop within the stromata and produce asci (80-110 x 6-9 pm) with unicelluar
ascospores (6-9 pm).
Diseased samples were sent on numerous occasions to the HDOA pathogen
-qua~anti~&as~in-Mond~u-b~t-thefung~cscould-n~becultu~ed~4s~o~pse~
collected from diseased leaves were inoculated onto Hawaiian miconia plants, but the
transfer was successful in only three of seven attempts. In each case, however, the
fungus failed to develop mature fruiting structures and the fungus was never recovered.
C. myconae is almost certainly an obligate parasite and since most obligates are
host-specific, the outlook for this fungus as a biological control agent is extremely
favorable. Research into the susceptibility of the Hawaiian miconia biotype, the
conditions for infection and disease development, and the life cycle of the fungus are
underway in Brazil.
DISCUSSION
Concurrent research on the biological control of a weed at the same time as
conventional control measures are being developed is unusual. The history of this
species in Tahiti and its behavior in Hawai'i clearly illustrated its threat to Hawaiian
forests (Gagne et al. 1992). Conventional control technologies were viewed as a
containment strategy awaiting the development of biological control agents. Funding
for each approach had to be sought from different agencies because of differing
jurisdictions and constraints. Some weed control specialists believed that miconia had
become too well established in Hawai'i and the eradication projects would require too
many years of dedicated funding and leadership (Conant, pers. corn.). Others felt that
miconia could be eradicated in Hawai'i in part from a general opposition to biological
control but also from a lack of appreciation of the difficulty of eradicating a species
using conventional technologies. The lack of consensus delayed an aggressive
approach to biological control.
Although C. g. miconiae can slow the growth of established miconia plants, and
also cause dieback of young seedlings, it alone will not control this weed. Preliminary
observations of field-release plots have shown that the pathogen may decrease
population densities.
The lack of host specificity in tests with the miconia pathogen, Pseudocercospora
tamonae, may be cause for its exclusion as a biological control agent. However, under
laboratory conditions, this fungus is a more devastating pathogen than C. g. miconiae
and it is not dependent on wind-driven rain for dissemination. Observations of disease
development on all susceptible test plants showed that P. tamonae was a primary
pathogen of miconia and other melastomes but not other families in the Myrtales where
it causes minor disease symptoms. Since one of these hosts is Metrosideros
polymorpha, an endemic dominant of most montane forests in Hawai'i, support for
releasing P. tamonae will be difficult to obtain. The validity of the concern should be
evaluated. Infection only occurred under conditions of maximum humidity and very
high spore densities over several hours' exposure, an improbable situation in the field.
However, if infection occurred under normal field conditions then the pathogen might,
through time, adapt to Metrosideros.
Research on the miconia pathogen Coccodiella myconae has not progressed due
to difficulties in disease transmission. The potential of this fungus, however, precludes
abandonment of this agent at this stage. Ongoing research into the life cycle and
mechanism of infection of this fungus may lead to solutions to the transmission
problems. Isolation of the fungus from its hyperparasite can be achieved.
Total eradication of miconia is and always will be the primary goal of
environmentalists in Hawai'i (Conant et a/. 1996). Biological control is basically a
contr~deand-vvilInd-bjtit~f~r~i~temismia,A-t-best,it-wiII-reduce-tbegrovvthand spread of this weed to the extent that the conventional control means will be
successful in critical areas. Meanwhile the search for additional biological control
agents continues.
ACKNOWLEDGEMENTS
Thanks to Dr. C. W. Smith for continuous encouragement on biological control as the
only viable long-term solution to miconia. The financial support of the following
agencies is acknowledged: Government of French Polynesia, U.S. Geological Service
Biological Resources Division, State of Hawai'i, Maui Water Board. Logistical support
and other cooperation from the following has assisted the program enormously: U.S.
Forest Service, National Park Service, U.S. Geological Service Biological Resources
Division, Hawai'i Department of Agriculture, The Nature Conservancy Hawai'i.
LITERATURE CITED
Almeda, R., 1990. Melastomataceae. pp. 903-917, In: Manual of Flowering Plants of
Hawai'i. W. L. Wagner, D. R. Herbst, and S. H. Sohmer (eds.). University of
Hawai'i Press and Bishop Museum Press, Honolulu.
Altonn, H., 1991. Isle forest periled by evil weed's explosive growth. Honolulu Star
Bulletin. May 31: A-8.
Barreto, R. W. 1999. Fungal Pathogens and Natural Enemies of Miconia calvescens
and Tibouchina herbacea: Report of a survey made in the Dominican Republic
and Cost Rica during December 1998 and January 1999. File report, Pacific
Cooperative Studies Unit, University of Hawai'i, Honolulu. 7 pp.
Barreto, R. W. 2000. Fungal Pathogens, Natural Enemies of Miconia calvescens:
Report of a survey made in Ecuador during May 2000. File report, Pacific
Cooperative Studies Unit, University of Hawai'i, Honolulu. 10 pp.
Burkhart, R. 1996. The Search For Biological Control Of Miconia calvescens:
Photographic Documentation of Natural Enemies of Miconia calvescens
(Melastomataceae) found in Central and South America between July 1993 and
September 1995.
http:llwww.botanv.hawaii.edulfacultvlcw smithlmc control.htm
Conant, P., A. C. Medeiros and L. L. Loope, 1996. A multiagency containment program
for miconia (Miconia calvescens), an invasive tree in Hawaiian rain forests. pp.
249-254, In: Assessment and Management of Plant Invasions. J.O. Luken and
J.W. Thieret (eds). Springer Verlag. New York.
Davis, J. 1978. Miconia calvescens. Newsletter, Hawaiian Botanical Society 17: 63.
Eager, H., 1995. Council members hear detail of miconia threat. The Maui News.
March 31: A-3.
Gagne B., L. L. Loope, A. C. Medeiros and S. J. Anderson, 1992. Miconia calvescens:
a threat to native forests of the Hawaiian Islands. Pacific Science 46: 390-391.
Gardner, D. E., 1990. Role of biological control as a management tool in national parks
and other natural area. Technical Report NPSINRUHINRTR-Q0101. United States
Department of the Interior. Washington, D.C. 41pp.
Killgore, E. M. 1998. Trip report, HDOA files.
Killgore, E. M., L. S. Sugiyama, and R. W. Barreto, 1998. Prospective biological control
of Miconia calvescens in Hawaii with a non-indigenous fungus Colletotrichum
gloeosponoides (Penz.) Sacc. f. sp. miconiae. pp. 65-71, In: Proceedings of the
First Regional Conference on Miconia control, August 26-29, 1997, Papeete,
Tahiti, French Polynesia. J.-Y. Meyer and C. W. Smith (eds), Gouvernement de
Polynesie franpiseJUniversity of Hawai'i at ManoaICentre ORSTOM de Tahiti.
Killgore, E. M., L. S. Sugiyama, R. W. Barreto, and D. E. Gardner, 1999. Evaluation of
Colletotrichum gloeosporioides for biological control of Miconia calvescens in
Hawai'i. Plant Disease 83: 964.
Klingman, D. L. and J. R. Coulson, 1982. Guidelines for introducing foreign organisms
into the U.S. for biological control of weeds. Plant Disease 66: 1205-1209.
Medeiros, A. C., L. L. Loope, P. Conant and S. McElvaney, 1997. Status, ecology, and
management of the invasive plant Miconia calvescens DC (Melastomataceae) in
the Hawaiian Islands. Bishop Museum Occasional Papers No. 48.
Medeiros, A. C., L. L. Loope, and R. W. Hobdy, 1998. Interagency efforts to combat
Miconia calvescens on the Island of Maui, Hawai'i. pp. 65-71, In: Proceedings of
fhe First Regional Conference on Miconia control, August 26-29, 1997, Papeete,
Tahiti, French Polynesia. J.-Y. Meyer and C.W. Smith (eds). Gouvernement de
Polynesie franqaiseluniversity of Hawai'i at ManoaJCentre ORSTOM de Tahiti.
Meyer, J.-Y., 1994. Mecanismes d'invasion de Miconia calvescens DC. en Polynesie
Franqaise. Ph.D. thesis. Academie de Montpellier - Sciences et Techniques de
Languedoc, Universite Montpellier 11. 126 pp.
Meyer, J.-Y., 1996. Status of Miconia calvescens (Melastomataceae) in the Society
Islands (French Polynesia). Pacific Science. 50: 60-78.
Ramadan, M. 1998. Trip Report, HDOA files.
Singh, B. P., 1999. High security containment facilities in the United States for fungal
plant pathogens of quarantine significance. pp. 194-204, In: Containment Facilities
and Saieguards for Exotic Plant Fathogens and Pests. R.P. Khan and S.B. Mathur
(eds.). APS Press, St. Paul.
Tanji, E., 1996. Governor declaring war on alien tree. Honolulu Advertiser. April 8: A-3.
-
-
Tanji, E., 1996. Volunteers tackle foreign invader. Honolulu Advediser. April 14: A-25.
Wagner, W. L., D. R. Herbst and S. H. Sohmer. 1990. Manual of flowering plants of
Hawai'i. University of Hawaili Press and Bishop Museum Press. Honolulu.
Wapshere, A. J., 1974. A strategy for evaluating the safety of organisms for biological
weed control. Annals of Applied Biology 77: 201-21 1.
Watson, A, K., 1985. Host specificity of plant pathogens in biological weed control. pp.
577-586, In: Proceedings of the IV Symposium of the Biological Control of Weeds,
August 19-25, 1984. E. S. Delfosse (ed). Agriculture Canada, Vancouver,
Canada.
Watson, A. K., 1991. The classical approach with plant pathogens. pp. 3-23, In:
Microbial Control of Weeds. D. 0. TeBeest, (ed). Chapman and Hall. New York.
Yohay, J. B. and B. Fukutomi, 1997. At war with a plant. Honolulu Star Bulletin. May
22: A-9.
BIOLOGICAL CONTROL OF GORSE IN HAWAI'I:
A PROGRAM REVIEW
George P. arki in,' Patrick on ant,^ Eloise ~ i l l ~ o rand
e,~
Ernest yoshioka4
1
USDA Forest Service, Forestry Sciences Laboratory, Rocky Mountain Research
Station, 1648 South 7'hAvenue, Bozeman, Montana 59717-2780
Email: gmarkinefs.fed.us
2
Hawai'i Department of Agriculture, 16 East Lanikaula Street, Hilo, Hawai'i 96720
Email: [email protected]
3
Hawai'i Department of Agriculture, 1428 King Street, Honolulu, Hawai'i 96813-5524
-
-4
Emall: elolseklllgore@yat~oo.corn
Hawai'i Department of Agriculture ( R ~ ~ E a S € L a n i l t a u E S t ~ t : H i l o ~ H a w a i J i 96720
Abstract. Gorse (Ulex europaeus), a spiny, leguminous shrub, has invaded pasturelands and
natural ecosystems on Maui and Hawal'i. An interagency effort to implement long-term control
of gorse included support for a biological control effort. Between 1984 and 2000, seven insect
natural enemies and one plant pathogen were tested, six of which were eventually released in
Hawai'i. This paper reviews the history, organization, and cost of this program and the lessons
we learned in an attempt to identify information that might be useful in planning similar, future
programs.
Keywords: Apion scutellare, A. ulicis, Agonopterix ulicetella, Anisoplaca ptyoptera,
Chlorophorus trifasciatus, Cydia lathyrana, Dicfyonota strichnocera, Pempelia
genistella, Sericothrips staphylinus, Sitona spp., Tetranychus lintearius, Ulex
europaeus, Uromyces pis; f. sp. europaei.
INTRODUCTION
Gorse (Ulex europaeus L.; Fabales, Fabaceae), a spiny, leguminous shrub from
Western Europe, was used extensively to form hedges for containing livestock before
the invention of barbed wire. Gorse was distributed widely throughout the world for this
purpose but in almost all new localities it quickly escaped from cultivation and became
a serious weed (Holm et a/. 1979). It probably was introduced to Hawai'i around the
turn of the century (Degener 1975) and by 1925 was recognized as a serious weed on
Maui. At that time, biological control was attempted but none of the introduced agents
established (Julien & Griffiths 1998). By the 1950s, the change from sheep- to cattleranching on the island of Hawai'i resulted in the realization of gorse as a major weed
and another biological control program was undertaken. Fourteen insects were
evaluated; most could not be reared in quarantine or failed specificity testing. Only
three were released, all weevils of the genus Apion (Chrysomelidae: Curculionidae).
Only the seed weevil, A. ulicis (Forster), became established on Maui but with no
noticeable impact on the spread of gorse (Markin & Yoshioka, 1989).
In the absence of an effective complex of biological control agents, prisoners
stationed at the Olinda correctional facility were used to control gorse on Maui by
manual removal and by planting pines as shade trees. On the island of Hawai'i,
chemical control was undertaken between 1976 and 1978, funded through the federal
Comprehensive Employment Training Act. This program was so successful that by
1980, gorse was considered controlled and further management efforts dropped on
Hawai'i. Unfortunately, no one considered the long-lived seeds in the soil. By 1983,
gorse had recolonized the entire original area. In 1984, Department of Hawaiian Home
Lands and Parker Ranch attempted to organize a new gorse management program,
which included a renewed effort to develop effective biological control agents. By 1996,
initial work to find, test, and release a new complex of insect agents had been
completed; at least four insect biological control agents were established which
attacked different parts of the plant (Markin et el. 1996). A plant pathogen was
released in early 2000. The program has now shifted to monitoring the impact of these
agents on gorse.
A description of the agents and their release in Hawai'i has been presented
elsewhere (Markin et at. 1996). The purpose of this paper is to review the history and
organizafi~f-thepragral.nand-to-identif+thelessons-leacnecCin-conductinga-biological control program in Hawaiian natural ecosystems.
HISTORY OF PROGRAM: 1980-2000
On October 27, 1983, the "Big Island Resource Conservation and Development
Committee", a local program of the USDA Soil Conservation Service (SCS), met at
Mauna Kea State Park to discuss the spread and management of gorse on the Big
Island. The meeting was followed by a field trip to the Humu'ula infestation on the
southeast slopes of Mauna Kea (for a map of the gorse infestations in Hawai'i, see
Markin et a/. 1988). The committee was so impressed by the massive resurgence of a
plant everyone believed had been controlled that they formed a gorse-control
committee to look into the implementation of a new management program. In
recognition of the fact that herbicides would probably be ineffective in limiting the
spread of gorse because of the massive seed bank, the committee recommended a
renewed biological control effort.
The first official meeting of the Hawai'i Steering Committee on Gorse Control was
held in 1983. The Committee reviewed research results including testing of new
herbicides, management through burning and grazing, biological control, and a longterm integrated control program. The Committee comprised representatives from
Hawaiian Homelands, Parker Ranch, several adjacent ranches infested or threatened
by gorse, Hawai'i Department of Agriculture (HDOA), the Hawai'i Division of Lands and
Natural Resources, and the U.S. Army Pohakuloa Training Area. The USDA-SCS
Resource Conservation and Development (RC&D) Office at Waimea accepted the
committee and its program as an official RC&D program, allowing the Committee to
solicit funds, write contracts, and submit grant proposals.
For the first year, most effort focused on increasing local awareness of the problem
with some attempts at dlrect control and containment. A previously released agent, the
gorse seed weevil (A. ulicis),previously established on Maui, was introduced to Hawai'i
(Markin & Yoshioka 1989).
Ernest Yoshloka, HDOA entomologist on Hawai'i with previous gorse experience
during the 1950's biological control effort and George Markin, entomologist with the
USDA Forest Service Institute of Pacific Island Forestry (IPIF) became involved in the
gorse control project at this time. While primarily a weed of pastures, gorse also was
invading the lower edge of the pukiawe shrub zone on Mauna Kea and its seeds were
carried down watercourses into the lower elevation rain forest. Therefore it was judged
a suitable weed for study by IPIF. The National Park Service (NPS) was interested in
supporting the program also because gorse was an invasive weed in Haleakala
National Park on Maui. Because HDOA had been unable to rear gorse insects in their
Honolulu quarantine facility, the gorse biological control agents were evaluated in the
new high-elevation quarantine facility at Hawaii Volcanoes National Park.
In the 19801s,the most active biological control program on gorse was being
conducted in New Zealand (NZ) under the leadership of Dr. Richard Hill, NZ
Department of Scientific and Industrial Research (DSIR, presently Landcare Research).
From field exploration in England, he had selected ten insects with the most potential
as biological control agents of gorse (Hill 1982). NZ was supporting initial evaluation of
several agents in England and Agonopterix ulicetella (Stainton) (Lepidoptera,
Oecophoridae) was already in quarantine. In 1985, an informal cooperative effort with
New Zealand was established and the first shipment of A. ulicetella arrived in Hawai'i in
1986.
By 1986, the Gorse Steering Committee had increased to 15 active members,
mostly local landowners and representatives of state and federal land management
agencies, with a mailing list of over 50 interested participants on Maui and Hawai'i. The
Committee actively supported biological control. Grants from the County of Hawai'i and
U.S. Fish and Wildl~feService [USFWS]) were used to rear, screen, and test A.
ulicetella and other potential gorse biological control agents. HDOA directed the
program as a continuation of earlier biological control efforts against gorse, and
selected test procedures and species for specificity tests In 1987, another small grant
was received from the County of Maui, and the State of Oregon became involved as a
partner in 1987. Gorse is a major problem along the west coast of Oregon where it
hinders forest management (Hermann and Newton1968) and recreation and creates a
significant fire danger (Holbrook 1943).
By this time, the Gorse Steering Committee had obtained two influential members:
Ken Autry from the USDA-SCS office in Waimea provided organization and dynamic
leadership and Francis Pacheco, a local consultant for the sugar industry, brought the
Committee political contacts within the state. Under the guidance of these two
sometimes-conflicting personalities, the Gorse Steering Committee undertook new
public education campaign through newspaper releases, public field days, and an
aerial tour for legislators. Through this effort, the Committee was able to attract the
support of local state legislators who facilitated its 1988 approach to the Governor's
Agricultural Coordinating Committee (GACC) and the legislature for funding. With wide
public awareness, support of local legislators, and testimony from the Gorse Steering
eommittee~tha-bill-passed-anctwas-ftlII y-f u n d h e ~ t a t d w e e m ~_ o r
exceeded by IPlF and NPS contributions in salaries and facilities.
Hawai'i now was able to participate in the development of biological control agents
in Europe Hawai'i and New Zealand established a formal agreement and funded
Commonwealth Agricultural Bureau (now CAB1 Bioscience) to conduct the needed
studies. By supplementing the research funding from NZ, a member nation of CABI,
Hawai'i and Oregon were able to support more research in Europe than would have
been possible otherwise.
The first new agent, A. ulicetella, was released in the fall of 1988. IPIF, in
cooperation with New Zealand, coordinated the foreign work, conducted the final host
testing in quarantine in Hawai'i, and obtained release permits. HDOA established mass
rearing facilities to produce large numbers of insects, which were released throughout
gorse infestations in Hawai'i. Permanent study plots were set up to monitor growth of
gorse and impact of the biological control agents.
The next agent, the gorse thrips, Sericothrips staphylinus Haliday (Thysanoptera,
Thripoidae) was released in 1990. At about this time, two agents were rejected
because they showed some ability to feed on two native trees, koa (Acacia koa Gray Fabales, Fabaceae) and mamani (Sophom chrysophylla (Salisb.) Seem. - Fabales,
Fabaceae) (Table 1).
In 1991, the program was delayed when HDOA's Plant Quarantine Branch stopped
'releases of all new weed agents while new regulations were written and implemented.
The new regulations, in place by 1994, established the requirement of a federal
environmental assessment for all new releases and increased the permit review time
from six months to 1-2 years. Under the new review process, the gorse spider mite
(Tetranychus lintearius (Dufour) Acari, Tetranychidae) was released in 1995 and the
moth, Pempelia genistella (Duponche) (Lepidoptera, Pyralidae), in 1996.
-
Table 1. Summary of insect biological control agents in et el. (1996).
Agents
Year
Released
Year
Status
Established
1984*
1989
1988
1990
1985
b&uesM
Apion (Exapion) ulicis
Apion (Pempion) scutellare
Agonopterix ulicetella
Sericothrips staphylinus
-
1989
1992
Tetranychus lintearius
Pempelia genistella)
Dictyonota strichnocem
Anisoplaca ptyoptera
1995
1995
1996
Not released
Not released
Uromyces pisi t.sp. europaei
2000
78% of pods attacked
Not established
Well established
Established,
spreading
Well established
Not established
Fate unknown
Rut
Cydia lathyrana (root moth from England)
~~p~~oot-weevii~mmEnglarrct)
Chlorophorus trifasclatus (root-feeding beetle from Portugal)
Released on island of Hawai'i, already established on Maui.
Unfortunately, by the time the new regulatory process was implemented, the state
was encountering severe financial problems and in 1993 the legislature stopped
funding the program. Several root-feeding insects whose development had been
delayed were therefore never tested (Table 1). Loss of funding greatly reduced the
HDOA effort toward mass rearing, release, and redistribution of agents in the field and
halted a long-term monitoring program. With IPlF and Oregon state funding, the gorse
program was able to obtain release permits for the last two agents. In 1995 the mite, T.
lintearius, was released from quarantine and further work in Hawai'i discontinued.
Quarantine work on the last insect, P. genistella, was continued at Bozeman, Montana
during 1995-96 and testing of new insects discontinued after the release of this agent
in 1996.
During this period HDOA constructed a new plant pathogen quarantine facility at
Honolulu. One of the first pathogens to be brought into it was Ummyces pisi J. Schrot.
f.sp. europaei (Uredinales, Puccinaceae), a rust fungus obtained from England. This
agent was released in the spring of 2000.
At least four of the agents are now established (Table 1). Superficial observations
from entomologists and land managers familiar with the Mauna Kea gorse infestations
suggest that the biological control agents may have reduced flower production and
annual shoot length. Plants frequently appear sickly and yellowing with numerous dead
and dying branches often covered with webbing from the mite. The ultimate impact of
these agents probably will not be known for another 5-10 years.
CONCLUSIONS
While it is too early to determine the success of this program, the original goal of
finding, testing, and establishing a complex of at least four biological control agents in
Hawai'i was accomplished. We can, however, estimate what a new biological control
program might cost, how long it might take, and identify information that might be useful
in planning, organizing, and conducting programs in Hawai'i.
Cost.
Establishing the cost for the insect portion of this program is difficult because of the
many different sources of funding that supported it and the contributions in salary, time,
and materials made by the different agencies. A rough breakdown over the 1I-year
period indicates that the insect work required almost $1,500,000 (Table 2). An
additional $200,000 was spent over 7.5 years evaluating the insect pathogens, for a
total cost of roughly $1,700,000. Cooperation with NZ yielded an additional $1million
benefit, since they paid the cost of all the preliminary work in Europe to find, select, and
study the agents that were eventually brought into our quarantine. Also, utilizing the
data obtained in NZ's host testing of these agents significantly reduced quarantine
studies. Finally, an additional $250,000 would probably have been necessary to run a
long-ten post-release monitoring study to complete this program.
A similar biological control program on a totally new weed of Hawaiian natural
ecosystems is expected to cost around $3 million. While this may seem high, it falls
-within_the-rangeof_estimates_within_theraogeof_estimates_far_otherwefar-~the~weed~iolo~~~~co
costs for a complete weed biological control program 20 years ago were between $1 to
2 million (Andres 1977, Harris 1979).
Conclusion: To undertake a totally new weed biological control program in Hawai'i
will probably cost around $3 million.
Time.
The insect component of this program took over 11 years from its conception in 1984
until the release of the last insect in 1996. During this period, two years' delay were
necessitated while the governing regulations were changed. An additional year's delay
occurred due to a request by USFWS for additional testing of endangered species of
Hawaiian plants. However, the program was shortened by the fact that NZ had
previously spent several years doing the preliminary studies necessary to identify the
natural enemies of gorse. Quarantine evaluation of the species proceeded fairly
Table 2. Program Costs. Estimates of funding or values of contributed services that
supported the insect portion of the Hawai'i Gorse Biological Control Program from 19841996.
U.S. Forest Service, Institute of Pacific Island Forestry (estimated that 80% was for salaries)
$ 775,000
National Park Service Contribution (Operation, Electricity, and Maintenance of Hawai'i
National Park Quarantine)
Volcanoes
$
45,000
Salaries, Hawai'i Department of Agriculture, Plant Pest Control Personnel. Hilo
$ 121,000
State of Hawai'i, Legislature-appropriatedFunds
$ 450,000
Oregon Department of Agriculture
$
50,000
Use of Montana State University Insect Quarantine Facility to study and ship two agents to
$2Qiuu
Hawai'i
Outside Grant $ 35,000
TOTAL $1,496,000
smoothly, each taking 2-3 years. Initially, approval for their release could be obtained
in as little as 6 months. However, with the need now to do an environmental
assessment and with the new regulations that allow outside agencies such as the
USFWS to request re-testing or additional testing, it is expected that each insect would
now require 3 to 5 years.
While the scientific work and review process for each agent proceeded reasonably
smoothly, this long time span (11 years) had several major undesirable effects. The
program was totally dependent on the gorse steering committee to provide the political
----supportto-raise-moneyY-~~~~9~199~~thetee~a~_excellent
support from this
committee, but once money from the state had been obtained and the first agents were
in the field, interest and participation in this committee dropped off. When the budget
crisis hit Hawai'i in 1993, the strong local committee needed to see that the program
continued was no longer available. The second problem was that the legal regulatlons
under which this work was conducted constantly changed in response to interacting
laws and regulations.
Conclusion: Future programs (at least the finding, testing, and release of the new
agents) should be tightly organized and conducted within as short a timeframe as
possible. Drawing out the timeframe will allow burnout and loss of your local outside
supporters and expose the program to changes in regulations, causing delays and
increasing costs. Ideally, the testing in quarantine, even if it means rearing and testing
a large number of different species of insects simultaneously, and the approval process
should be concentrated, if at all possible, In a 5-year period.
Steering Committee.
The chance of finding a single, permanent, long-term source of funding for a biological
control program for a weed of Hawaiian natural ecosystems is probably impossible
Therefore, the need to locate and use multiple sources of funding will probably be the
norm. The most effective approach is by means of a local steering committee
composed of land managers, landowners, and other interested parties who are already
fighting the weed and willing to commit their time and effort to fund raising. Any weed
biological control program needs the political support of the local community, therefore,
anotner function of the steering committee is education and increasing public
awareness. The steering committee can also coordinate the work of different agencies
involved. Besides supporting biological control, the gorse steering committee
supported testing other management efforts, including testing of new herbicides,
shadmg by reforestation, effectiveness of burning, grazing by goats, and the use of
different grazing regimes. The long-term solution to the gorse problem in Hawai'i will
probably require a combination of several management approaches.
Conclusion: Local land managers and private landowners should be recruited into
a steering committee during conceptual discussions of a new biological control program
to solicit the necessary funding and provide support to the scientists.
Leadership.
Researchers should not be expected to provide leadership for multi-agency, politically
driven programs such as the gorse program. They are too involved in day-to-day work
with insects and the research involved. There is also a potential conflict of interest.
The outside leadership contributed substantially to the success of the scientific program
providing political support, obtaining funding, maintaining a focus in the research and
demanding results and reports. The success was largely the result of the commitment
of the steering committee, particularly the chairman.
Conclusion: A successful program needs leadership provided by well-informed,
politically savvy outsiders (non-researchers) committed to seeing the program succeed.
Outside Cooperators.
The complexity of the scientific research that identified, evaluated, and tested the
complex of insects required the involvement of many cooperators in other countries.
Previous work in NZ and England and cost sharing considerably benefited this project
by saving time and research. Finding another country that was interested in gorse
control~.-~blZ~n~unninga-closelyelyc~011dina~ec/hi~y110~~tati~e-p~0~
ca9m-withPely~
them is probably one of the major factors that contributed to our success.
Conclusion: Look for outside cooperators.
Lead Scientist.
The studies necessary to progress from locating a natural enemy in its homeland to its
release as a biological control agent in the field in Hawai'i have become so
complicated, time-consuming, and costly that conducting a program for a single weed
can demand the full time and attention of the responsible scientist. The majority of the
time involved in a new biological control program will be spent in interaction with other
people such as members of the local steering committee, foreign scientists, and the
administrative bureaucracy that must be navigated to obtain permission to ultimately
release the new agents. Finally, the scientist in charge must be able to train and
supervise the technicians who will do most of the routine handling of the insects, from
their arrival in quarantine to their release.
Conclusion: Each weed targeted for biological control needs a single, full-time
scientist responsible not only for the necessary entomological studies, but for
coordinating and guiding all stages of the program. Since overall administration and
coordination will be more time-consuming and critical than any single, scientific study
that might be conducted, the lead scientist should be judged on the number of new
biological control agents released, established in the field, and attacking the weed,
instead of on the number of peer-reviewed papers published in a scientific journal.
Expect outside criticism.
While in general the program experienced excellent local support, there was criticism.
In conducting biological control, it is common to take the shortsighted view that it is just
another management tool. It is difficult to realize that many people outside the field do
not understand the goals, and through ignorance, fear, or for personal motivations, will
attack it.
During this program, Howarth (1983, 1991) published his articles questioning the
safety of biological control. He primarily looked at problems with many of the earlier
biological control programs aimed at insects and did not address weed biological
control, however, several mainland scientists used his findings to question the practice
of weed biological control in general, and this program specifically. These outside
criticisms were beneficial in that they required a re-examination of the program, testing
procedures, and the need to educate another group of people, outside scientists. The
second objection came from the USFWS, some members of which felt that under the
Endangered Species Act insufficient attention was paid to the potential threat to
endangered plants in Hawai'i. Their objections did not significantly harm this program just necessitated that another group of plants be included in the host range testing but it did emphasize that the regulations under which biological control is conducted are
not fixed in stone but are constantly changing.
Conclusion: A continually new audience of people are watching and reviewing
the work and will be questioning its values and safety. Future biological control of
weeds programs, therefore, must expect some criticism, be flexible enough to identify
the basis for the attacks, and be prepared to work to resolve them.
LITERATURE CITED
Degener, 0. 1975. Plants of Hawaii National Park illustrative of plants and customs of
the South Seas. Ann Arbor, MI: Braun Broomfield, Inc. 312 pp.
Harris, P. 1979. Cost of biological control of weeds by insects in Canada. Weed
Science. 27: 242-250.
Hermann, R. K, and M. M. Newton. 1968. Tree planting for control of gorse on the
Oregon coast. Research Paper 9. Oregon State University, Forest Research
Laboratory, Corvallis, OR.
Hill, R. L. 1982. The phytophagous fauna of gorse (Ulex eurvpaeus L.) and host plant
quality, Doctoral dissertation, Imperial College, University of London, Silwood
Park, Ascot, U.K.
Holbrook, S. H. 1943. The gorse of Bandon. Chapter 14, In: S. H. Holbrook (ed.),
Burning an empire. Macmillan Press, London.
Holm, L. G., J. V. Pancho, J. P. Herberger, and D. L. Plucknett. 1979. A geographic
atlas of world weeds. John Wiley and Sons, NY.
Howarth, F. G. 1983. Classical biological control: panacea or Pandora's box.
Proceedings, Hawaiian EntomologicalSociety. 24: 239-244.
Howarth, F. G. 1991. Environmental impacts of classical biological control. Annual
Review of Entomology. 36: 485-509.
Julien, M. H., and M. W. Griffiths, eds. 1998. Biological control of weeds. CAB1
Publishing, Brisbane, Australia.
Markin, G. P., L. A. Dekker, J. A. , Lapp, and R. F. Nagata. 1988. Distribution of gorse
(Ulex europaeus L.), a noxious weed in Hawaii. Newsletter, Hawaiian Botanical
Society. 27: 110-117.
Markin, G. P., and E. R Yoshioka. 1989. Present status of biological control of the
weed gorse. pp. 357-362, In: E.S. Delfosse (ed.), Proceedings: VII International
Symposium on Biological Control of Weeds, March 6-11, 1988. lstituto
Sperimentale per la Patologia Vegetale, MAF, Rome.
Markin, G. P., E. R. Yoshioka, and P. Conant. 1996. Biological control of gorse in
Hawaii. pp. 371-375, In: V.C. Moran, and J.H. Hoffman (eds.), Proceedings:
IX International Symposium on Biological Control of Weeds, January 19-26, 1996.
University of Cape Town, Stellenbosch, South Africa.
SETTING PRIORITIES FOR THE BIOLOGICAL CONTROL OF
WEEDS: WHAT TO DO AND HOW TO DO IT
Judith H. Myers and Jessica Ware
Centre for Biodiversity Research, Dept. Zoology and Faculty of Agricultural Sciences,
6270 University Blvd., University of British Columbia, Vancouver, V6T 124, Canada
Email: [email protected]
Abstract. Three options are available for dealing with non-indigenous plant species that either
may become or already have become invasive weeds; keep them out, eradicate introduced
populations while they are still small, or finally attempt biological control of established
populations. By far the best approach to controlling potentially invasive foreign weeds is to limit
the introduction of plants to new areas. Better communication of the consequences and
environmental costs of these species may help balance the pressure applied on regulatory
agencies by Industries to permit commercial plant importations under lew limitations.
Eradication attempts must be bold and fast. Proponents of the program should not make
unrealistic promises because eradication is so difficult to achieve. Finally, biological control
does have a potential role in the management of non-indigenous weeds. However, finding
agents that are capable of reducing the densities of plants is not an easy task. Successful
biological control agents are capable of killing or greatly reducing the vigor of their host plants at
life stages for which little compensation can occur. Greater focus on efficacy can help reduce
the number of non-indigenous species that are introduced in biological control prngrams
Keywords: eradication, invasive plants, seed predators, plant introductions, quarantine,
noxious weeds
INTRODUCTION
Over the last 500 years transoceanic travel and commerce has led to a global
redistribution of non-native plants (Moody and Mack 1988). Without natural enemies,
many alien plants have established, invaded and displaced native vegetation (Mack ef
a/. 2300). Three options are available for dealing with alien plant species that either
may become or already have become invasive weeds; keep them out, eradicate
introduced populations while they are still small, or finally attempt biological control of
established populations.
-
PRIORITY ONE KEEP THEM OUT
Although exclusion is the most effective way to prevent potentially invasive weeds,
regulations to stop the introduction of weeds has been woefully ineffective. In the USA
at least 2000 non-native plants are invading managed and natural systems. This
includes at least 235 woody plants and 600 herbaceous plants, including grasses and
aquatic species (Reichard and Campbell 1996). A majority of non-indigenousweeds
were introduced intentionally to areas where they were not native and where they have
become serious environmental or agricultural pests. In the USA 85% of the 235
species of woody plant invaders were introduced as ornamentals or for landscaping
(Reichard and Campbell 1996). In Australia approximately 46% of noxious weeds have
been introduced for ornamental or other purposes (Panetta 1993). In the city of Zurich
alone 300 plant species have naturalized, 52% for ornamental purposes, and these are
thought to have led to the disappearance of more than 150 native species (Landolt
1993).
It is not surprising that the types of plants valued by horticulturists are also those
with characteristics that preadapt them to be weeds. Species favored by
horticulturalists plants are those that produce many flowers, begin to flower early in the
season, are easy to propagate, and that grow well in disturbed sites (White and
Schwarz 1998). The prevention of weed introductions therefore creates a tension
between agencies regulating plant introductions and the horticultural and agricultural
industries. In Australia the government restricts the entry of plants via the Quarantine
Act of 1908 that prohibits taxa from 19 genera and 66 species (Panetta 1993).
However, plants being introduced by nurseries and the gardening industry are not
included on these schedules.
Given the pressure by commercial ventures to introduce foreign plants, and the
need for quarantine regulations to protect against invasive weeds, a procedure for
predicting the potential invasiveness of non-indigenous plants is required. Schemes for
assessing plant species prior to introduction have been proposed by Panetta (1993),
Reichard and Hamilton (1996), and White and Schwarz (1998). White and Schwarz
(1998) compare the traits of plants desired by gardeners and those of invasive weeds.
They list the following traits to be in common; rapld growth, early and many flowers,
prolific seed production, good seed germination, and no major pests. Adaptations for
efficient dispersal and vegetative spread are also likely to apply to both situations.
Rechard and Campbell (1996) developed a scheme for predicting the invasiveness of
plants by comparing the traits of 235 species of plants known to be invaders to those of
87 noninvasive plants. These comparisons were based on plant growth rates, juvenile
penods, germination requirements and the ablllty to reproduce vegetatively. Under this
system plants can be categorized into those to be rejected, accepted or held for further
examination. White and Schwarr (1998) tested the ability of Reichard's scheme to
predict the lnvasiveness of known plant invaders and found that 85% would have been
rejected by the scheme, 13% would have been held for further examination and 2%
would have been accepted.
Panetta (1993) proposed a scheme based on earlier work of (Hazard 1988). He
assigned a point value to different characteristics; plants receiving a value of twenty or
more points are rejected while those with a value of 12 or less are accepted, and those
in-between are further investigated (Table 1). This system rejects outright aquatic
plants, potentially causing friction with the major industry selling aquatic plants for
aquaria and ponds.
Another system for evaluating the invasiveness of plantsIsAAustralian Weed
risk assessment scheme proposed by Pheloung (1995 cited in White and Schwarz
1998). This system relies on a number of plant attributes including whether the plant is
a known invader, as well as other biological characteristics having to do with climatic
requirements, reproduction, dispersal, persistence, noxiousness, and distribution.
lnvasiveness elsewhere is also considered. Overall there are 49 questions divided into
8 categories. Nonweedy traits receive a score of 0 or -1, unknown traits a score of 0,
and weedy traits a score of 1. Some scores are weighted differently depending on the
answer to the question. Plants receiving a score of 0 or less are considered safe for
introduction while those with a score greater than 7 are considered to be potentially
weedy invasives. This system focuses on vegetative growth and the ability of plants to
tolerate a range of conditions and high levels of damage. It also includes
characteristics such as the possession of parasites, toxicity and the seed type
produced.
White and Schwarz (1998) tested the ability of this system to reject plant species
already known to be invasive in Australia and of the current invaders, 84% would be
rejected by this scheme, 16% fell into the category of requiring future study and none
would have been accepted. More recent versions have questions dealing with
dispersal attributes.
Criterion
1 ISthe species free-floating aquatic or can
Point value
20
it survive and reproduce as a free-floating aquatic
1
Is it a weed elsewhere
20
Are there close relatives with a history of
10
invasion to similar habitats
Is it spiny
10
Are diaspores spiny
10
Is the species harmful to animals
8
Does the plant produce stolons
5
Does the plant reproduce vegetatively
8
Are diaspores wind-dispersed
8
Are diaspores dispersed by mammals or machinery
8
Are diaspores dispersed by water
5
I Are diaspores dispersed by birds
5
I
I
Table 1. Evaluation scheme proposed by Panetta (1993) to predict potential
invasiveness of plants.
These evaluation schemes may not be perfect but they do suggest criteria that
could be used effectively to slow the continued introduction of potentially invasive
-plants~Thereisa_great-needto-applyp01itical~pressll[1e~to~~htenregulatbns~ontbe~~
purposeful introduction of plants. Currently the interests of the nursery and ornamental
plant industry are served with little regard for the environmental impacts of invasive,
non-indigenous species. Greater publicity on the cost of these weeds, and better
education of landscape architects, plant breeders, home and commercial gardeners,
pet store owners and those with aquaria and garden ponds are a necessity.
PRIORITY TWO - ERADICATION
Following the introduction and establishment of a non-indigenous plant there may be a
short time-window in which the species might be eradicated (Myers et a/. 2000).
Eradication is particularly difficult for plants that produce many seeds because dispersal
can be rapid. Also the establishment of a long-lived seed bank makes total elimination
difficult and continued vigilance over many years imperative. Usually the opportunity
for eradication is lost by the time the problem is recognized and the project considered
or finally put into place. A vivid example of a lost opportunity was the failure to
eradicate the tropical marine alga Caulerpa taxifolia (Vahl) (Chlorophyta,
Chlorophyceae) from Monaco where it was first recognized in 1984 (Meinesz 1999).
Eradication was called for in 1991 but to no avail. This "aquarium" plant is now
widespread in the Mediterranean where it forms dense stands.
Successful eradications of small populations of recently introduced nonindigenous species may not be recorded in the literature and therefore it is difficult to
judge how often this is successful. One such success in southeastern Queensland was
the eradication of Eupatonum serotinum Michx. (Asterales, Asteraceae) in the 1950's
(R. McFadyen pers. comm.). An on-going project in Australia is the attempt to
eradicate Siam weed, Chromolaena odorata (L.) R.M. King and H. Robinson (Asterales,
Asteraceae). Although densities have been greatly reduced, and the distribution has
been limited to a 50 km radius, total eradication will be a slow process (R. McFadyen,
personal communication).
In a recent review (Myers et a/. 2000) 6 factors were proposed as necessary for
a successful eradication program:
1) Sufficient resources to complete the project;
2) Clear lines of authority for decision making;
3) A target organism for which the biological characteristics are compatible with
eradication (easy to find and kill, no seed bank);
4) Easy and effective means to prevent reinvasion;
5) Easy detection of plants at low densities; and,
6) Plans for restoration management if the species has become dominant in
the community; there is little value in replacing one weed with another.
For widely established weeds, area-wide management may still be possible. This will
involve continuous efforts to suppress populations of the plants chemically and
mechanically in all locations. Working along the borders of the plant distribution may
help to reduce the spread of the invasive weed, but metastatic spread of an invasive
species fallowing seed dispersal by animals or movement of plants by human activities
works against successful containment.
In conclusion, rapid action following the identification of a newly established
species may allow successful eradication, but procrastination can cause the window of
opportunity to slam shut.
For well established non-indigenous species biological, chemical, or mechanical control
are the only mechanisms to reduce plant vigor and density. In most instances,
chemical and mechanical control are too expensive and damaging to other members of
the plant community and to the environment for widespread use. Although biological
control usually is better focused on the target weed species than are the other control
procedures, it does require the introduction of another species. In addition, in several
cases biological control agents have moved onto non-target species of native plants
(Louda et a/. 1997, Louda 1999, Pemberton 1995). Therefore, careful consideration
must be given to which species of biological control agents are introduced. Each
introduction of an insect or plant disease will carry with it some chance that an
unwanted non-target impact will result (Cory and Myers 2000, Follett and Duan 1999).
Selecting those species of natural enemies with the greatest potential for success is a
challenge to the biological control practitioner.
Biological control, if successful, will reduce the density of the plant to a level at
which other control is not required (McFadyen 2000), but it will not eliminate the weed
species totally. However, biological control is not always successful. Estimates of how
successful biological control is have varied from 10 to 30% of the weed species for
which it has been attempted (Crawley 1989, McFadyen 1998, Myers, et a/. 1988) to
80% success in Australia and South Africa (McFadyen 2000). Some of thrs variation
may be influenced by how success is evaluated. And how much effort has been put
into programs. If successful control requires a whole complex of natural enemies to
reduce the density of the host plant populations, achieving success may be slower and
more difficult than if single agents are able to reduce the densities of the target weeds
In an attempt to evaluate the question of whether a complex of agents is necessary
for successful biological control, several studies have examined successful biological
control programs recorded in the literature to determine whether success was attributed
to several agent species or to just one. In the first of these studies (Myers 1984), 81%
of successful biological control programs were attributed to one species of agent. In a
more recent review (McFayden 2000) 41% of 41 successful biological control programs
of weeds involved only one agent or success was attributed to one agent. Another
study, based on data from Julien and Gnffiths (1998), found that 54% of programs
involving the introduction of several agents attributed success to a single agent (Denoth
et al. in press). These studies are difficult to interpret because most often careful
evaluation of the impacts of the various agents is not done. Therefore, attributing
success to a particular agent or group of agents may be a bit arbitrary. There may be a
tendency for biological control practitioners to attribute some success to every agent
that has been introduced. This may explain why of 8 biological control successes
recorded for Hawai'i (Opuntia cordobensis, Spegazzini (Caryophyllales, Cactaceae), 0.
fiscus-indica (L.) Miller, Agemtina riparia (Regel) R.M. King & H. Robinson (Asterales,
Asteraceae), Emex spinosa (L.) Campd. (Polygonales, Polygonaceae), Rubus argutus
Link (Rosales. Rosaceae), Tribulus cistoides L. (Zygophyllales, Zygophyllaceae)
(native) and T. tenestris L.) only 2 weeds (25%) are thought to have been controlled by
a single agent and the other 6 successes, including the control of a native species, are
attributed to a combination of agents (information from (Julien and Griffiths 1998)).
This contrasts to 8 successful weed control programs in South Africa for which success
in 5 programs (62%) is attributed to a single agent (Olckers and Hill 1999).
4istori~a~Cy-onawrags5t~age~1t~ha~ebee~htr0d~~8cCf0r-ea~h-bi0I0gical-control program although some weed species such as lantana have been the target for
over 20 species of natural enemies (Broughton 2000). Of the thousands of biological
control agents introduced globally, only about 10% have had sufficient impact on the
host plants to be considered effective. The record of outcomes of agents introduced
for biological control of weeds in Hawai'i is shown in Figure 1 based on data reviewed
by Julien and Griffiths (1998).
This review of introductions and successes in biological control indicates that a
number of non-indigenous agents are being introduced to control other non-indigenous
species. Because the introduction of each species is associated with the possibility
that it will have non-target impacts, it is important to be parsimonious in choosing
agents for introduction. If certain types of potential biological control agents have had a
poor record of success in the past, they should probably not be considered for
introduction in future programs. We propose that one group of natural enemles that
-
-
are unlikely to have much impact on host plant density are those that merely reduce
seed production.
Figure 1. Fate of the 81 species of agents introduced to Hawai'i for the biological
control of weeds.
Not
established
UnsuccessfuI
Successful
Other
Seed predators are unlikely to be successful biological control agents because
seedling survival is frequently related to the density of the seeds; high densities survive
less well than low densities. This principle of "self-thinning" is well known in plant
ecology (Silverton and Lovette Doust 1993). The ability of plants to compensate for
reduced seed production and lower seedling numbers makes the impacts of biological
control programs difficult to predict (Myers et a/. 1988). Most serious weeds have higtiseed production and therefore are likely to be able to absorb high seedling mortality
from effective andlor numerous seed predators. Therefore, it would seem that seed
predators have little potential as biological control agents for plants that are able to
compensate for the loss of seeds and for those for which seedling establishment is
limited by the number of "safe sites" available (Andersen 1989).
Although seed predators may not reduce the density of invasive weeds it is a
widely held view that they may slow the spread of the target plants. In South Africa
biological control practitioners consider seed predators to be useful in programs that
use both chemical and mechanical control in addition to biological control (Impson eta/.
1999). However the contribution of seed predators in these programs has not been
evaluated. If the spread of plants is by diffusion, reduced seed production is more
likely to be translated into a reduced rate of spread than if plants disperse as a
metastatic process, with new establishments occuning outside the initial introduction
site and serving as new foci for spread of the species (Clark et a/. 1998, Kot et a/.
1996, Moody and Mack 1988). It cannot be assumed that seed predators will have a
measurable impact on the rate of spread of plants. However, experimental analysis of
this could be very useful for target weed species.
One way to evaluate the effectiveness of seed predators as biological control
agents is to determine if they ever have been successful. Tephritid flies are seed
predators, and they have been introduced in 25 biological of weed programs (Julien
and Griffiths 1998). They never have been shown to be successful control agents. A
review of South African weed control programs lists seed predators as only being
successful when in combination with other agents (Olckers and Hill 1999). In these
programs stem borers were the most successful agents followed by sap-suckers.
Rees and Paynter (1997) developed a model of scotch broom, Cystisus scoparius
(L.) Link (Fabales, Fabaceae), from which they concluded that a reduction in plant
fecundity could reduce broom density when the disturbance rate is high, and plant
fecundity and seedling survival low. However, these situations do not hold for most
introduced populations of broom and Paynter eta/. (1996) concluded that seed
predators were unlikely to reduce broom populations in New Zealand. They did
suggest that seed predators might slow the spread of the species, but this is
contradicted by observations in Oregon where the introduced insect Exapion fuscirostre
(F.) (Coleoptera, Curculionidae) reduced seed production by 85% but did not reduce
either broom density or spread (Andres and Coombs 1992). A recent matrix model of
scotch broom showed that in prairie in Washington State, USA 99.9% of the seed
would have to be destroyed to suppress invasion of the plant, but in urban sites with
poor soil and disturbance only 70% of the seeds would have to be reduced to slow
invasion (Parker 2000). These studies suggest that seed predators are not likely to be
effective biological control agents for broom.
Several studies of native plants have indicated a relationship between seed
predators and plant recruitment and density (Louda 1SJ82a1l98Zb, 1983, 1999, Louda
and Potvin 1995). This difference in the impact of seed predators on populations of
introduced and native plants may be related to the role of competing plant species in
the native plant communities. Serious weeds often occur at such high densities that
intraspecific competition is more important than interspecific competition. The potential
relevance of competition to the impact of seed predators is seen in the one example
that we know of in which biological control success has been attributed to a seed
feeder. This is the control of nodding thistle, Carduus nutans L. (Asterales,
Asteraceae), in North America by the weevil Rhinocylus conicus Froelich (Coleoptera,
Curculionidae) (Harris 1984, Kok and Surles 1975), a biological control agent that has
also been shown to have an impact on rare, native thistles (Louda 1999, Louda and
Potvin 1995).
Nodding thistle is an annual plant and depends on disturbance for establishment,
but once established it can dominate a site, at least temporarily (Wardle et a/. 1995). A
comparison of nodding thistle biology to another weed, diffuse knapweed, Centaurea
diffusa Lam. and Frauenfeld (Asterales, Asteraceae), may indicate why a seed feeder
is successful for nodding thistle but not for more typical weeds that are longer lived and
better competitors. Knapweed is a short-lived (2-5 years) perennial and is able to
invade sites that have experienced little disturbance (Berube and Myers 1982).
Knapweed has very high summer survival of the rosette stage (90 -95%) while rosette
survival of nodding thistle can be low (10-30%) (Sheppard et a/. 1994). A simulation
model of knapweed population growth indicated that because seedling survival was
-
-
density dependent, the only way to reduce weed density would be via an agent that
killed rosettes (Myers and Risley 2000). Similarly, Kelly and McCallum (1995) found
that density-dependent survival from seedling to flowering in nodding thistle plants
compensated for seed loss, the higher mortality among rosettes in nodding thistle may
mean that reduced seed production is sometimes translated into reduced plant density.
Although seed predators may be effective in reducing the density of annual plants that
are poor competitors, most invasive weeds are longer-lived and good competitors.
Seed predators may be attractive as biological control agents because they are
relatively easy to find, particularly if they develop in the seed heads. Also, biological
control success is often measured simply by the number of plants attacked rather than
impacts on plant density (McFadyen 2000). This also makes seed predators attractive
as candidates for biological control programs because they can be readily evaluated.
The over-use of seed predators is demonstrated in the biological control program
targeted against the yellow star thistle, Centaurea solstitialis L. Six species of seed
predators have been introduced to North America against this weed (Pitcairn et a/.
1999). Although at some sites seed and seedling density have been reduced for
several years, no reduction of plant density has been reported for this annual plant.
Would one species of seed predator have been as effective as 6?
Predicting the type of an agent that will be successful in reducing the dens~tyof a
target weed is difficult. Study of the biology of the target plant may give some clues to
"weak" points in the life cycle. If a plant produces lots of seeds it is unlikely that
reduction in seed production will be translated to a reduction of plant density.
Experiments in which the ability of plants to compensate for various types of damage
may give clues of the type of agents that are likely to be effective. In addition, Force
(1972) and Zwolfer (1973) have proposed that the most effective biological control
agents are likely to be those that are not very common in the native distribution of the
plant. If a species of natural enemy occurs at low density in the native habitat because
it is a poor competitor or has a high level of parasitism it may demonstrate a good
reproductive response when introduced to a new habitat, free of competitors and
parasitoids. Therefore, being rare in the native habitat and having a high reproductive
rate when reared In the absence of natural enemies or competitors may be
characteristics to look for in potential biological control agents.
Because the introduction of non-indigenous species dilutes the native biodiversity
of an area, the introduction of new species should be undertaken in a conservative
manner. Just because a species has passed the host specificity tests does not
necessarily mean it should be introduced. The possibility of non-target impacts should
always be a concern. Therefore, efficacy should be another c o n s i d e r a t i ~ o o s i ~
potential biological control agents. A study of the biology of the target weed, including
its ability to compensate for the loss of various life stages, should be a prerequisite for
biological control introductions. A good example of compensation for herbivory is
shown by lantana, Lantana camara L. (Lamiales, Verbenaceae), which survived
experimental defoliation for 2 years (Broughton 2000). Defoliators are unlikely
candidates to successfully control this plant species. Prediction of the impact of natural
enemies on host plant density is certainly not easy. One way to evaluate control
agents may be to create high-density patches in the native habitat and determine which
species of natural enemies move onto the plants. By evaluating the impacts of
potential agents in the native habitat, informed decisions can be made prior to
introduction in the new habitat. Better evaluation of on going biological control
programs could also provide information to allow improved understanding of what
things work and why. From current information, seed predators do not appear to be
effective agents. Therefore their further introduction in biological control programs is
unlikely to be a parsimonious approach to biological weed control.
CONCLUSION
By far the best approach to limiting potentially invasive weeds is limiting the introduction
of plants to new areas. Better communication of the consequences and environmental
costs of non-indigenous species may help balance the pressure applied on regulatory
agencies by industries involved in commercial plant importations. Eradication attempts
must be bold and fast. But because eradication is so difficult to achieve, proponents of
the program should not make unrealistic promises. Finally, biological control does have
potential for controlling the impact of foreign weeds. However, finding agents that are
capable of reducing the densities of plants is not an easy task. Successful biological
control is associated with agents that are capable of killing or greatly reducing the vigor
of their host plants at a life stage for which little compensation can occur. A greater
focus on the efficacy of proposed agents can help reduce the number of nonindigenous species that are introduced in biological control programs.
ACKNOWLEDGEMENTS
JHM wishes to thank the U.S. Forest Service, USDA for providing support to attend the
Hawai'i Conservation Forum on Biological Control of lnvasive Plants in Hawaiian
Natural Ecosystems.
LITERATURE CITED
Andersen, A. 1989. How important is seed predation to recruitment in stable
populations of long-lived perennials? Oecologia 81: 310-315.
Andres, L., and E. Coombs. 1992. Scotch Broom Cytisus scoparius (L.) Link
(Leguminosae). pp. 303-305, In: Biological Control in the U.S. Western Region:
Accomplishments and Benefits of Regional Research Project W-84 (19641989). J. Nechols, L. Andres, J. Beardsley, R. Goeden and C. Jackson (Eds).
Division-of-Ag~su#ur88~d-NaturaCRes0~r~e~,~ive~~ity~of~
alifornia,Berkeley-,
CA.
Berube, D., and J. Myers. 1982. Suppression of knapweed invasion by crested wheat
grass in the dry interior of British Columbia. Journal of Range Management 35:
459-6 1.
Broughton, S. 2000. Review and evaluation of lantana biocontrol programs. Biological
Control 17: 272-286.
Clark, J., C., Fastie, G. Hurtt, S. Jackson, C. Johnson, G. L. King, M., and J. Lynch.
1998. Reid's paradox of rapid plant migration. BioScience 48: 13-24.
Cory, J., and J. Myers. 2000. Direct and indirect ecological effects of biological
control. Trends Ecology and Evolution 15: 137-139.
Crawley, M. 1989. Insect herbivores and plant population dynamics. Annual Review of
Entomology 34: 531-564.
Denoth, M., L. Frid, and J. H. Myers. 2002. Multiple agents in biological control:
Improving the odds? Biological Control (In Press).
Follett, P.A. and J.J. Lynch. (eds.). 1999. Nontarget Effects of Biological Control.
Kluwer Academic Publishers, The Netherlands
Force D.C. 1972. r- and K- strategists in endemic host-parasitoid communities.
Bulletin, Entomological Society of America 1 8: 135-137.
Harris, P. 1984. Carduus nutans L., nodding thistle and C , acanthoides L., plumeless
thistle (Compositae). pp. 115-126, In: Biological Control Programs Against
Insects and Weeds in Canada 1969-1980. J . Kelleher and M. Hulme (Eds).
Commonwealth Agricultural Bureau, Slough, U.K.
Hazard, W. 1988. Introducing crop, pasture and ornamental species into Australia the risk of introducing new weeds. Australian Plant Introduction Review 19:1926.
Impson, F., V. Moran, and J. Hoffman. 1999. A review of the effectiveness of seedfeeding bruchid beetles in the biological control of mesquite, Prosopis species
(Fabaceae), in South Africa. pp. 81-88, In: Biological Control of Weeds in South
Africa (1990-1998). T . Olckers and M. Hill, (Eds). Entomological Society of
Southern Africa, Johannesburg, SA.
Julien, M., and M. Griffiths. 1998. Biological control of weeds: a world catalogue of
agents and their target weeds. CAB International: Wallingford, Oxon.
Kelly, D., and K. McCallum. 1995. Evaluating the impact of Rhinocyllus conicus on
Carduus nutans in New Zealand. pp. 205-211, In: Vlll International Symposium
anSiologieatC--ORfmtaff W e e d s + D e l f o s s e - a ~ b R ~ S m H ~ E d ~ S l W
Re,~ l
Melbourne, Australia, Canterbury, New Zealand.
~
Kok, L., and W. Surles. 1975. Successful biological control of musk thistle by an
introduced weevil, Rhinocyllus conicus. Environmental Entomology 4:10251027.
Mot, M., M. Lewis, and P. van den Driessche. 1996. Dispersal data and the spread of
invading organisms. Ecology 77: 2027-2042.
Landolt, E. 1993. ~ b ePflanzenarten,
r
die sich in den leMen 150 Jahren in der Stadt
Ziirich stark ausgebreitet haben. Phytocoenologia 23: 651-663.
Louda, S. 1982a. Limitation of the recruitment of the shrub Haplopappus squarrosus
(Asteraceae) by flower- and seed-feeding insects. Journal of Ecology 70: 43-53.
Louda. S. 1982b. Distribution ecology: variation in plant recruitment over a gradient in
relation to insect seed predation. Ecological Monogmphs 52: 25-41.
Louda, S. 1983. Seed predation and seedling mortality in the recruitment of a shrub,
Haplopappus venetus (Asteraceae) along a climatic gradient. Ecology 64: 511521.
Louda, S. 1998. Population growth of Rhinocyllus conicus (Coleoptera: Curculionidae)
on two species of native thistles in prairie. Environmental Entomology 27: 834841.
Louda, S. M. 1999. Negative ecological effects of the musk thistle biocontrol agent,
Rhinocyllus conicus Foel. pp. 215-243, In: Nontarget Effects of Biological
Control. P . A. Follett and J. J. Duan, (Eds). Kluwer Academic Publishers, The
Netherlands.
Louda, S., and Potvin, M. 1995. Effect of inflorescence-feeding insects on the
demography and lifetime fitness of a native plant. Ecology 76: 229-245.
Louda, S., D. Kendall, J. Connor, and D. Simberloff. 1997. Ecological effects of an
insect introduced for the biological control of weeds. Science 277: 1088-1090.
Mack, R., D. Simberloff, W. Lonsdale, H. Evans, M. Clout, and F. Bazzaz. 2000. Biotic
invasions: causes, epidemiology, global consequences, and control. Ecological
Applications 10: 689-710.
McFadyen, R. E. 1998. Biological control of weeds. Annual Review of Entomology 43:
369-393.
McFadyen, R.E. 2000. Successes in biological control of weeds. pp. 3-14, In:
Proceedings X International Symposium Biological Control of Weeds. N.
Spencer, (Ed). Montana State University, Bozeman, MO.
Moody, M., and R. Mack. 1988. Controlling the spread of plant invasions: the
importance of nascent foci. Journal of Applied Ecology 25: 1009-102 1.
Myers, J. 1984. How many insect species are necessary for successful biocontrol of
weeds? pp. 77-82, In: Proceedings VI. International Symposium on Biological
Control of Weeds. E. Delfosse, (Eds). Agriculture Canada, Ottawa.
Myers, J., and C. Risley. 2000. Why reduced seed production is not necessarily
translated into successful biological weed control. pp. 569-581, In: Proceedings
X International Symposium Biological Control of Weeds. N. Spencer, (ed).
Montana State University, Bozeman, MO.
Myers, J., C. Risley, and R. Eng. 1988. The ability of plants to compensate for insect
attack: Why biological control of weeds with insects is so difficult. pp. 67-73, In:
VII. International Symposium on Biological Control of Weeds. E . Delfosse (Ed).
lnstituto Sperimentale per la Patologia Vegetale, Rome Italy.
Myers, J., D. Simberloff, A. Kuris, and J. Carey. 2000. Eradication Revisited: dealing
with exotics. Trends in Ecology and Evolution 15: 316-320.
Olckers, T., and M. Hill. 1999. Biological Control of Weeds in South Africa (1990-1998).
African Entomology Memoir No.1. Entomological Society of Southern Africa:
Johannesburg, SA.
Panetta, F. 1993. A system for assessing proposed plant introductions for weed
potential. Plant Protection Quarterly 8: 10-14.
Parker, I.M. 2000. Invasion dynamics of Cystisus scoparius: a matrix model approach.
Ecological Applications 10 : 726-743.
Pemberton, R. W. 1995. Cactoblastis cactorum (Lepidoptera: Pyralidae) in the United
States: An immigrant biological control agent or an introduction of the nursery
industry? American Entomologist 41: 230-232.
Pheloung, P. 1995. Determining weed potential of new plant introductions to Australia.
A report on the development of a weed risk assessment system commissioned
and endorsed by the Australian Weeds Committee and the Plant Industries
Committee. Agricultural Protection Board, Western Australia.
Pitcairn, M., D. Woods, D. Joley, C. Turner and J. Balciunas. 2000. Population buildup
and combined impact of introduced insects of yellow starthistle, Centaurea
solstitialis, in California. pp. 747-751, In: X Symposium, Biological Control of
Weeds. N . Spencer (ed). Bozeman, MO.
Reichard, S., and F. Campbell. 1996. Invited but unwanted. American Nurseryman
187: 39-45.
--Reichad,~,,and-C,HamiIton~l996.~Predicting-invasionsoff~oOo~d~plantsidr~duced~f~
into North America. Conservation Biology 11: 1993-203.
Sheppard, A., J. Cullen, and J. Aeschlimann. 1994. Predispersal seed predation on
Carduus nutans (Asteraceae) in southern Europe. Acta Oecologica 15: 529541.
Silverton, J., and J. Lovette-Doust. 1993. Introduction to plant population biology.
Blackwell Scientific Publications, Oxford.
Wardle, D., K. Nicholson, M. Ahmed, and A. Rahman. 1995. Influence of pasture
forage species on seedling emergence, growth and development of Carduus
nutans. Journal of Applied Ecology 32: 225-233.
White, P., and A. Schwarz. 1998. Where do we go from here: the challenges of risk
assessment for invasive plants. Weed Technology 12: 744-751.
Zwolfer, H. 1973. Competition and coexistence in phytophgagous insects attacking the
heads of Carduus nutans L. pp. 74-80, In: I1 International Symposium, Biological
Control of Weeds. P. H . Dunn (ed). Rome, Italy.
HOST SPECIFICITY TESTING FOR ENCARSIA SPP.,
PARASlTOlDS OF THE SILVERLEAF WHITEFLY, BEMISIA
ARGENTIFOLII BELLOWS & PERRING, IN HAWAI'I
Walter T. Nagamine and Mohsen M. Ramadan
Hawai'i Department of Agriculture, Biological Control Section, 1428 South King Street,
Honolulu, Hawai'i 96814, U.S.A.
Email: Walter-T-Nagamineaexec.state.hi.us
ABSTRACT. We describe host specificity studies for four Encarsia spp. (Hymenoptera:
Aphelinidae) parasitoids used in the biological control of the silverleaf whitefly, Bomisia argentifolii
Bellows & Perring (Homoptera: Aleyrodidae), in Hawai'i. The problems encountered in
determining rion-target test species are described with respect to the potential impact on
Hawaiian Lepidoptera. The misidentification of an Encarsia sp. in the scientific literature
suggested that one of the lour Encarsia spp. parasitoids would parasitize lepidopteran eggs. This
information created confusion and cast doubt on our test results. An authority on Encarsia
parasitoids re-examined the misidentified species and corrected its identity.
Key words: Aleyrodidae, Aphelinidae, Bemisia, biological control, Encarsia, host
specificity tests, lepidopterous eggs, non-target species, silverleaf whitefly
The silverleaf whitefly, Bemisia argentifolii Bellows & Perring (Homoptera: Aleyrodidae),
is a major pest of many vegetable crops in Hawai'i. In 1992, the Hawai'i Department of
Agriculture began a biological control program for the silverleaf whitefly. Some species
of Encarsia (Hymenoptera: Aphelinidae) are well known as biological control agents of
whiteflies and other Homoptera and are generally host specific. The department's
exploratory entomologist collected four species of Encarsla parasitoids; E. Iutea (Masi)
and E. mineoi Viggiani from Egypt in 1992, and E. hispida DeSantis and E. pergandiella
Howard from Brazil in 1994.
There is an increasing awareness and concern regarding the potential of biological
control agents to attack non-target hosts (Howarth 1991). In Hawai'i, host specificity
studies usually require testing of non-target species. Since all whitefly species in Hawai'i
are immigrants (there are no natives or beneficial species), testing was not considered
necessary as long as the potential agent was known to be specific in host selection.
~ 0 ~ e ~ ~ ~ E 1 ~ ~ a r s i a ~ / u t i ~ ~ - h a d - b ~ m e ~ ~ ~ a n ~ a g ~ ~ ~ t 0
Heliothis zea (Boddie) (Lepidoptera, Noctuidae) and the cabbage looper, Trichoplusia ni
(Hubner) (Lepidoptera, Noctuidae) in Arizona cotton fields (Stoner and Butler 1965).
Female parasitoids emerged from whiteflies, while male parasitoids emerged from the
moth eggs.
Egg parasitism in this genus is not as rare as had been thought. Polaszek (1991)
summarized lepidopteran egg hosts of Encarsia and found records of 18 species of
Lepidoptera from six different families that were parasitized by Encarsia. Encarsia and
other genera in the family Aphelinidae have complex natural histories in which males
and females of the same species have different host relationships (Walter 1983). In
some Encarsia species, females will develop almost always as primary parasitoids of
whitefly hosts. Males, however, can develop in one of three ways:
1. as primary parasitoids of whitefly hosts;
2. as hyperparasitoids, developing at the expense of female larvae or pupae of their
own species, other primary parasitoid species, or both; and,
3. as primary parasitoids of lepidopteran eggs.
The potential non-target host species included exotic, beneficial, and native species.
We conducted tests to determine if the presumed E. lutea from Egypt would parasitize
lepidopteran eggs to produce male parasitoids as the E. lutea from Arizona was reported
to do. The four species of Lepidoptera used were the cabbage looper, Trichoplusia ni
(Hubner); tobacco budworm, Heliothis virescens (Fabricius); tomato pinworm, Keiferia
lycopsersicella (Walsingham) (Lepidoptera, Gelichiidae); and diamondback moth,
Plutella xylostella (L.) (Lepidoptera, Plutellidae).
Our results showed that the Egyptian strain of E. lutea did not parasitize the eggs of any
of the Lepidopteran species tested, raising the question of why the Arizona strain but not
the Egyptian strain was able to parasitize lepidopteran eggs. Further examination of the
voucher specimens from the 1965 Arizona study found that this Encarsia sp. had been
misidentified; it was not E. lutea, but an undescribed species (Williams and Polaszek
1996). Excellent taxonomic work and related studies resolved the dilemma and cleared
E. lutea as a parasitoid of potential harm to Hawaiian Lepidoptera.
LITERATURE CITED
Howarth, F.G.1991. Environmental impacts of classical biological control. Annual
Review of Entomology 36: 485-509.
Polaszek, A. 1991. Egg parasitism in Aphelinidae (Hymenoptera: Chalcidoidea) with
special reference to Centrodora and Encarsia species. Bulletin of Entomological
Research 81: 97-106.
Stoner, A. and G.D. Butler. 1965. Encarsia lutea as an egg parasite of bollworm and
cabbage looper in Arizona cotton. Journal of Economic Entomology 58: 1148-50.
Walter, G.H. 1983. Divergent male ontogenies in Aphelinidae (Hymenoptera:
Chalcidoidea): a simplfied classification and a suggested evolutionary sequence.
Biological Journal of the Linnaean Society 19: 63-82.
Williams, T. and A. Polaszek. 1996. A re-examination of host relations in the
Aphelinidae (Hymenoptera: Chalcidoidea). Biological Journal of the Linnaean
Society 57: 35-45.
PREDICTABLE RISK TO NATIVE PLANTS IN BIOLOGICAL
CONTROL OF WEEDS IN HAWAI'I.
Robert W. Pemberton
lnvasive Plant Research Lab, USDA-Agricultural Research Service
3205 College Ave. Ft. Lauderdale, FL 33314, U.S.A.
Email: [email protected]
ABSTRACT. The analysis examines the use of non-target native plants in Hawai'i resulting from
biological control projects on target weeds with close relatives compared with projects on target
weeds that lack close relatives. Target weeds with close relatives are riskier targets for biological
control than are weeds without close relatives in Hawai'i. The two projects conducted against
weeds with close relatives resulted in non-target use of native species; four of the five insect
species established in these projects now use native plant species as hosts. Only one of 18
(5.0%) projects against Hawaiian weeds that lack close relatives has produced native plant use.
Overall, 53 of 54 agents established for weed control exhibit predictable and highly stable host
ranges. This pattern of non-target plant use indicates that the risk to the native flora can be
judged reliably before introduction. The degree of risk is directly related to the relative
relatedness of the targeted weeds and the natlve flora and the speclnclty of the natural enemies
employed.
Keywords: biological control of weeds, non-target use, insectlplant interactions
INTRODUCTION
Biological control is a valuable method of controlling introduced pests in agriculture and
in natural areas. Biological control is currently being employed against invasive weeds
at a number of United Nations World Heritage Sites including South African Cape
Fynbos, Kakadu National Park in northern Australia, and Everglades National Park in
Florida (Center 1995). It has been an important tool in the fight against introduced
weeds in Hawai'i for almost 100 years (Funasaki eta/. 1988).
Like other pest control technologies it carries some risk. The associated risks relate
primarily to organisms targeted for biological control and host specificities of the
biological control agents employed. The safety of introduced biological control
organisms to non-target native organisms is an important issue in biological control
(Follet & Duan 2000, Wajnberg 2001). In the 1980's some practitioners of biological
l
control of weeds reported the use of native plants by introduced b i o l o g i c a l ~ t r oagent5
(Andres 1985, Pemberton 1985, Turner 1985, Turner et al. 1987). Howarth (1983,1991)
challenged the safety of biological control in general and specifically in Hawai'i, claiming
harm to native insects by introduced biological control parasitoids. Hawkins and Marino
(1997) examined the use of North American native insects by introduced parasitoids and
found that 16% of these parasitoids adopted native insects as hosts. Louda et a/. (1997)
reported population level damage to a native Cirsium thistle in Nebraska by the
biological control weevil Rhinocyllus conicus (Froelich) (Coleoptera, Curculionidae).
Because of the concerns for and documented cases of non-target impacts, reform of
biolo~icalcontrol practice and regulation to ensure greater attention to environmental
safety is needed (McEvoy and Coombs 2000, Strong and Pemberton 2000).
An important part of the biological control safety debate concerns the predictability
and stability of the host ranges of introduced biological control agents. Understanding
the predictability and stability of the host ranges of introduced agents is hampered by the
lack of general assessments of non-target host usage. This paper draws upon my
recent analysis of the use of non-target native plants by introduced biological control
agents of weeds in the United States, the Caribbean, and Hawai'i (Pemberton 2000).
Presented here is the information for Hawai'i.
MATERIALS AND METHODS
The analysis examines the use of non-target native plants in Hawai'i resulting from
biological control projects on target weeds with close relatives compared with projects on
target weeds that lack close relatives. By "use" I mean a completed life cycle of the
introduced agent on the non-target plant species. "Use" does not imply impact that is
unstudied. Close relatives are defined as congeneric species in the native flora. The
data set includes the establishment of 54 agents on 20 target weeds in Hawai'i. The first
releases were against Lantana camam L. (Lamiales, Verbenaceae) in 1902. The last
introductions resulting in establishment included in the analysis were in 1994; later
releases were excluded because I judged that insufficient time had passed for agent
population growth and dispersal to non-target species. Overall agents established on
weeds with close relatives and on weeds without close relatives have been released for
similar mean lengths of time (47 vs. 50 years, respectively).
The source of information on biological control of weeds projects in Hawai'i is Julien
and Griffiths' 1998 Bioiogical Control of Weeds: A World Catalogue of Agents and Their
Target Weeds. The principal source of information on the use of non-target native plants
is the entomological literature supplemented with personal communications with
researchers familiar with the projects.
RESULTS AND DISCUSSION
Target weeds with close relatives are riskier targets for biological control than are weeds
without close relatives in Hawai'i. The two projects conducted against weeds with close
relatives resulted in non-target use of native species; four of the five insect species
established as biological control agents in these projects now use native plant species
as hosts (Tables 1, 2). The project to control an introduced blackberry, Rubus argutus
Link - Rosales, Rosaceae) led to the establishment of three insect species in the 1960's;
all three use the two native Hawaiian species, Rubus hawaiensis A. Gray and R.
macraei A. Gray, (Funasaki et a/. 1988, George Markin, personal communication). The
other project in this category, control purple nutsedge, Cyperus rotundus L. - Juncales,
Cyperaceae, established two insect species, one of these, a weevil (Athesapeuta cypen
~~l~r;oieoptera,CZurcuIiim5dae)~intt~duce~nrt9~s8~s-arrative~e&
-~ ~ y p e ~ - - - -
polystachyos Rottb. (Poinar 1964).
By comparison, only 1 of the 18 (5.6%) projects against Hawaiian weeds that lack
close relatives has produced native plant use (Tables 1, 2). In these projects, only 1 of
49 (1.6%) established biological control agents now uses a native Hawaiian host. The
lacebug Teleonemia scrupulosa Stal (Hemiptera,Tingidae), introduced for control of
Lantana camam L. (Lamiales, Verbenaceae), was reported to use naio. Myoporum
sandwicense (DC) Gray (Lamiales, Myoporaceae), an endemic shrub (Maehler and Ford
1955, Bianchi 1961). All five biological control insects that have adopted native nontarget plants as hosts were released prior to 1970, before risk to native plants was
seriously considered by Hawaiian biological control researchers (Ken Terrarnoto,
personal communication). In Hawaiian biological control projects, 53 of 54 established
agents exhibit predictable and highly stable host ranges.
Teleonemia scmpulosa was collected in Mexico and released in Hawai'i in 1902,
without host specificity testing. The insect has been thought to be a Lantana specialist
(Winder and Harley 1983). The Myoporaceae and Verbenaceae are now considered to
be in the same order- the Lamiales (Angiosperm Working Group 1998), but lantana and
naio are not closely related. Changes in our understanding of plant phylogenetic
relationships brought about by molecular research (e.g., DNA sequence data:
Angiosperm Working Group 1998) suggest that it will be Important to evaluate the weed
and its relatedness to the Hawaiian flora in this light. The true host range of T.
scrupulosa is unclear. When introduced to Uganda for lantana control, it fed on and
damaged sesame, Sesamum indicum L. - Larniales, Pedaliaceae), and reproduced on
the plant to a limited extent (Davies & Greathead 1967). This report, as well as other
unverified records on target hosts (a Lippia sp. - Verbenaceae) in the Antilles, ebony,
Diospyros sp. Ebenales, Ebenaceae) in the U.S. (Drake and Ruhoff 1965), and
Xanthium sp. (Asterales, Asteraceae) in Hawai'i (Funasaki et a/. 1988), suggest that the
insect may not be the specialist that it was presumed to be. Recent searches on the
island of Hawai'i, where both naio and lantana grow closely together, found much T,
scrupulosa damage to lantana but none to naio (S. Hight and P. Conant, personal
communication).
This pattern of non-target plant use by introduced biological control agents indicates
that the risk to the native flora can be judged reliably before introduction. The degree of
risk is directly related to the relative relatedness of the targeted weed and the species in
the native flora. Species in the native flora can be protected by selecting target weeds
that are related only distantly to species in the flora and by employing agents with diets
narrow enough to avoid damaging native plants in the flora.
Hawai'i's flora is taxonomically circumscribed, with many common plant families
absent or with limited distribution (Wagner eta/. 1999). Most invasive weeds are
distantly related to native species, which suggests that biological control programs
against these weeds would unlikely harm native species. Of the 20 targeted weeds for
which biological control agents were released prior to 1994, only two have close
relatives. These weeds were targeted because of the problems they caused,
independent of the presence of native relatives. The Hawai'i Department of Agriculture's
Priority lists of weeds for FY 2000 (Nakahara 1999) lists 30 plant species for which
chemical/mechanical or biological control activities will be directed. Seven of these
plants belong to non-native families, while 17 others belong to non-native genera
(Wagner et a/. 1999). Only six of these plants have congeneric native relatives that
could be put at risk by biological control. These are species of Acacia, Caesalpinia,
~enchrus,Rubus, ~01anum;and possibly Digitaria (one species may be native) (Wagner
et a/. 1999).
Most of the seriously disruptive weeds in Hawai'i lack close relatives in the native
flora. For instance, Hawai'i has many invasive weeds in the Melastomataceae, including
the dangerous Miconie celvescens DC (Myrtales, Melastomataceae) (Medeiros et a/.
1997), but no native members of this family. Similarly, Hawai'i has no native gingers
(Zingiberales,Zingiberaceae) so biological control of Kahili ginger (Hedychium
gardnerianum Sheppard ex Ker-Gawl.), which can dominate the understory of rain
forests at mid-elevations, should be of low risk to the native flora. However, cultivated
gingers in Hawai'i are closely related and must be considered. Although this paper
deals with the risk to native plants, risk to other valued plants (agricultural, horticultural,
and cultural) related to the target weed also should be considered, as they traditionally
have been. Likewise, the lack of native species of Psidium, Senecio, and Paederia
suggest that weeds in these genera are appropriate targets. Since all three genera
belong to families containing native plants, it is important to evaluate the degree of
-
relatedness of the weeds to their confamilial Hawaiian relatives. lnvasive weeds with
close relatives, such as Himalayan blackberry (Rubus ellipitcus Sm.), would be much
riskier targets for biological control. Specialist insects typically use host plants limited to
a circumscribed taxonomic range (Strong et a/. 1984), e.g., within a plant family, within a
tribe within a family, a genus, subgenus, section or even a species. However, single
species specificity is less common than genus or subgenus host specificity). Plant
pathogens may have narrower host plant ranges than insects, w~thsome forms llm~tedto
subspecific taxa of plants, as with the rust Puccinia chondrillina Bubak & Snow
(Uredinales, Pucciniaceae) used to control rush skeletonweed, Chondrilla juncea L.in
California (Plper and Andres 1995). Careful determinatlon of field host range of the
candidate biological control organism in its native area coupled with rigorous host plant
specificity testing will predict the agent's potential host range in the area of introduction.
The specificity required depends directly on the degree of relatedness of the target weed
and species in the local native flora (Pemberton 2000). Biological control agents
employed against melastomaceousweeds in Hawai'i need be tested against native
species only at the family level to assess their likely use of native species. By contrast,
agents employed against Rubus weeds should be tested against individual species of
Rubus to avoid introducing species that might feed on Hawai'i's two native Rubus
species. The natural enemy pool from which to select biological control agents will be
larger for potential agents that require testing only at the family level. Species level
specialists, which may be needed for weeds with congeneric native relatives, may not
exist or may be difficult to find. Moreover, projects on weeds with close relatives will be
more expensive because more exploration and host specificity testing will be needed to
identify narrower specialists.
Given enough resources and time to identify specialist enemies and to confirm their
specificities, projects on weeds with close relatives can still be viable. The biological
control effort against leafy spurge (Euphorbia esula L.) in North America is an example
of a successful program on a weed with many native relatives in North America
(Nowerski and Pemberton, in press). This program was successful despite the many
native Euphorbia species in North America for a number of reasons. First, most of the
native species were actually not very closely related the target weed; most belong to
subgenera other than the subgenus Esula to which the target weed belongs. Second,
funding for the primary research programs continued for more than 25 years, which
enabled the examination of large numbers of candidate agents. This enabled the narrow
specialists to be identified and employed and the candidates with broader host ranges to
be discarded. Third, large numbers of narrow specialists that are also very damaging to
target weed, the Aphthona flea beetles (Chrrysornelidae), had evolved with subgenus
Esula plants.
Given the constraints on funding for biological control, the limited quarantine space
and low number of qualified biological control researchers, only a small portion of
invasive weeds can be subject to full biological control programs. Potential targets are
then necessarily prioritized by the seriousness of the problems they cause (kinds of
impacts, rates of spread, etc.), the control potential and cost of biological control, and
risk associated with such projects. Weeds with close relatives reasonably should be of
lower priority. Because biological control can be so effective against invasive weeds that
are frequently difficult to manage by other methods, there is a tendency to view all such
weeds as appropriate targets. But biological control may not be the most appropriate
control method for weeds with close native relatives. The risk to native plants
associated with biological control projects on weeds with close relatives should be
considered in relation to the risks associated with other control methods or with the
continued spread of the weed. In the fight against aggressive invasive weeds, absence
of control is not without risk as well. Fortunately, most Hawaiian weeds appear to be
safe targets for biological control.
ACKNOWLEDGEMENTS
I thank Pat Conant and Ken Teramoto (Hawai'i Department of Agriculture), and Frank
Howarth (Bishop Museum) for help with Hawaiian literature. Stephen Hight (US Forest
Service) and Pat Conant kindly shared their field observations of Telelonemia
scrupulosa damage to Lantana camara but not to Myoporum sandwicense. I am
particularly grateful to George Markin (US Forest Service) for allowing me to use his
unpublished observations of the use of native use of native Hawaiian Rubus by
introduced biological control agents.
LITERATURE CITED
Andres, L. A. 1985, Interaction of Chfysolina quadrigemina and Hypericum spp. in
California. pp. 235-239. In: E. S. Delfosse (ed), Proceedings, VI International
Symposium Biological Control of Weeds. Agriculture Canada.
Angiosperm Working Group. 1998. An ordinal classification for the families of flowering
plants. Annals Missouri Botanical Garden 85: 531-553.
Bianchi, F. 1961. Teleonemia scrupulosa. Proceedings, Hawaiian Entomological
Society 17: 313.
Center, T. D. 1995. Selection criteria and ecological consequences of importing natural
enemies. Biodiversity and Conservation 4: 885-526.
Davies J. C. and Greathead, D. J. 1967. Occurrence of Teleonemia scrupulosa on
Sesamum indicum Linn. in Uganda. Nature 230: 102-103.
Drake, C. J. and Ruhoff, F. A. 1965. Lacebugs of the world. Bulletin U. S. National
Museum 243: 384.
Follet, P. A. and Duan, J. J. (eds.) 2000. Non-Target Effects of Biological Control.
Kluwer, Dordrecht, The Netherlands.
Funasaki, G.Y., Lai, P-Y., Nakahara, L.M., Beardsley, J., and Ota, A. K. 1988. A review
of biological control introductions in Hawaii: 1890-1985. Proceedings, Hawaiian
EntomologicalSociety 28: 105-160.
Hawkins, B. A,, and Marino, P.C. 1997. The colonization of native phytophagous
insects in North America by exotic parasitoids. Oecologia 112: 566-571.
Howarth, F. G. 1983. Biological control: panacea or Pandora's box? Proceedings,
Hawaiian EntomologicalSociety 24: 239-244.
Howarth, F. G. 1991. Environmental impacts of classical biological control. Annual
Review Entomology 36: 485-509.
Julien, M. H, and Grlfflths, M. W. (eds.) 1998. Biological Control of Weeds; A World
Catalogue of Agents and Their Target Weeds. Edition 4. C.A.B. International,
Wallingford, UK.
Louda, S. M., Kendall, D., Connor, J., and Simberloff, D. 1997. Ecological effects of an
insect introduced for the biological control of weeds. Science 277: 1088-1090.
McEvoy, P. B, and Combs, E. M. 2000. Host specificity and biological pest control. pp
15-30. In: P. A. Follet and J. J. Duan (eds), Nontarget Effects of Biological
Control. ( Kluwer, Dordrecht, The Netherlands.
Maehler, Mr., and Ford, Mr. 1955. Teleonemia scnrpulosa. Proceedings, Hawaiian
Entomological Society 15: 377.
Medeiros, A.C., Loope, L. L., and Conant, P. 1997. Status, ecology, and management
of the invasive plant, Miconia calvescens DC (Melastomataceae) in the Hawaiian
Islands. Bishop Museum Occasional Papers 48: 23-36.
Nakahara, L. 1999. Priority lists of weeds for FY 2000. (unpublished memorandum, Nov.
16). Hawai'i Department of Agriculture.
Nowierski, R.M. and R.W. Pemberton. Leafy spurge (Euphorbia esula L.). In: R. Van
Driesche, B. Blossey and M. Hoddle, S. Lyon and R. Reardon (eds.) Biological
control of invasive plants in the eastern United States. US Forest Service Forest
Health Technology Enterprise Team-2002-04, Morgantown, West Virginia. (in press)
Pemberton, R. W. 1985. Native plant considerations in the biological control of leafy
spurge. pp. 365-390. In, E. S. Delfosse (ed), Proceedings VI lnternational
Symposlum Biological Control of Weeds. Agriculture Canada.
Pemberton, R. W. 2000. Predictable risk to native plants in weed biological control.
Oecologia 125: 489-494.
Piper, G. L., and Andres, L. A. 1995. Rush Skeletonweed. pp. 252-255, In: J. R.
Nechols, L. A. Andres, J. W. Beardsley, R. D. Goeden, and C. G. Jackson (eds),
Biological control in the western United States. University of California Division
of Agriculture and Natural Resources Publication 3361, Oakland, CA.
Poinar, G. 0. Jr. 1964. Observations on nutgrass insects in Hawaii with notes on the
host range of Bactra truculenta Meyrick and Athesapeuta cypen Marshall.
Proceedings, Hawaiian Entomological Society 18: 4 17-423.
Strong, D. R., Lawton, J. H., and Southwood, R. 1984. Insects on Plants. Harvard Univ.
Press, Cambridge.
Strong, D. R., and Pemberton, R. W. 2000. Biological control of invading species: risk
and reform. Science 288:
Turner, C. E. 1985. Conflicting interests and biological control of weeds. pp. 203-225.
In: E. S. Delfosse (ed), Proceedings VI lnternational Symposium Biological
Control of Weeds. Agriculture Canada.
Turner, C. E., Pemberton, R. W., and Rosenthal, S. S. 1987. Host utilization of native
Cirsium thistles (Asteraceae) by the introduced weevil Rhinocyllus conicus
(Coleoptera: Curculionidae) in California. Environmental Entomology 16: 1II-
lls.
Wagner, W. L., Herbst, D. R., and Sohmer, S. H. 1999. Manual of Flowering Plants of
Hawai'i Vol. 1 and 2. (revised edition). University Hawai'i Press and Bishop
Museum Press, Honolulu.
Wajnberg, E. , Scott, J. K., and Quimby, P. C. 2001. in press. Evaluating Indirect
Ecological Effects of Biological Control. International Organizationation of
Biological Control, Montpellier, France.
Winder J. A. and Harley, K. S. 1983. The phytophagous insects on Lantana in Brazil
and their potential for biological control in Australia. Tropical Pest Management
29: 346-362.
TABLE 1. Known non-target native ost plants of introduced biological control a(:ents of weeds in Hawai'i.
Target Weed
Non-target plant host
Eiolagical control agent
R
Cyperus rotundus
(purple nut sedge- Cyperaceae)
Cyperus polystacbyos
(manyspike flatsedge-Cyperaceae)
Athesapeuta cypen'
(Coleoptera: Curculionidae)
Poinar, 1
Lantana camara
(lantana- Verbenaceae)
Myoporuum sandwicense
(naio- Myoparaceae)
Teleonemia scrupulosa
(Hemiptera: Tingidae)
Maehle
Bia
Rubus argotus
(prickly Florida blackberry-Rosaceae;
Rubus hawaiensis
(Hawaii blackberry- Rosaceae)
Croesia zjmmermani
(Lepidoptera: Tortricidae)
Funasak
Markin p
Priophorus mono
(Hymen0ptera:Tenthredindae)
G. Mark
Schreckensteinia festaliella
(Lepidoptera: Heliodinidae)
G. Mark
Croesia zimmerrnani
(Lepidoptera: Tortricidae)
G. Mark
Priophorus mono
(Hymen0ptera:Tenthredindae)
G. Marki
Schreckensteinia festaliella
(Lepidoptera: Heliodinidae)
G. Marki
Rubus macraei
('akala)
TABLE 2. Comparison of non-target use of native plants by introduced agents in
biological control projects on target weeds with close relatives against projects on
t a r ~ eweeds
t
that lack close relatives. Close relatives are plant species that belong to
the same genus as the weed.
Target Weeds
Percent of projects with non-target
With native relatives
Without native relatives
100
5.6
(2 of 2)
(1of 18)
80.0
2.0
(4 of 5)
(1 of 49)
3
I
Total agents
54
use
Percent of agents adopting native
hosts
Number non-target plants used
Percent using non-target native plants 9.3
Percent of unpredicted use
1.6
REVIEW AND PERMIT PROCESS FOR BIOLOGICAL CONTROL
RELEASES IN HAWAI'I
Neil J. Reimer
Hawai'i Department of Agriculture, Plant Quarantine Branch, 701 Halo St., Honolulu, HI
96813, U.S.A.
Email: [email protected]
ABSTRACT. A review of the permitting process for the introduction of biological control agents
into Hawai'i is presented. The effects and results of the permitting process on the screening and
establishment of host specific biocontrol agents are discussed.
Keywords: Biocontrol introductions, permit, host specificity
INTRODUCTION
Hawai'i has had a long history in the use of biological control to reduce population levels
of introdwxd pests (Funasaki etal. 1988). The first use of classical biocontrol in Hawai'i
was in 1893 by Albert Koebele, entomologist for the Republic of Hawai'i, against the
cottony cushion scale (Icerya purchasi Maskell (Homoptera: Margarodidae) (Timberlake
1926). The project was successful and the reaction to subsequent pest problems in
Hawai'i often has been the introduction of natural enemies of the pest (Table 1). Diverse
fauna and flora have been used to combat equally diverse pests. Examples include
insects, fungi, viruses, bacteria, snails, bats, birds, fish, toads, and frogs to control
insects, plants, snails, and other organisms (Table 2). Many of these introductions
appear to have been successful in that the pest populations eventually did drop to
acceptable levels, although scientific evaluations of the effectiveness of these
introductions have been virtually non-existent.
The result of these introductions has been the establishment of 266 alien insect
species. This is a small percentage (3%) of the total insect fauna of ca. 7700 species
(Nishida 1994)); however, some of these introductions have had dramatic effects on a
few of the 5000 endemic species (Howarth 1985). These negative impacts of biocontrol
introductions primarily have been due to the lack of pre-release risk analyses, poor to
nonexistent host specificity studies, and the absence of adequate import regulations.
The U. S. Department of Agriculture - Animal Plant Health Inspection Service - Plant
Pest Quarantine (USDA-APHIS-PPQ) and the Hawai'i Department of Agriculture
(HDOA) currently regulate the importation of biocontrol agents into Hawai'i. These two
agencies have different jurisdictions and mandates, with some overlap.
The USDA has statutory authority under the Plant Quarantine Act (1912),the
Federal Plant Pest Act (1957), and the Federal Noxious Weed Act (1974) to prevent the
introduction and dissemination of plant pests (7 CFR 371.(c)(2)) The USDA only has the
authority to regulate an organism if it feeds on, infects, or parasitizes living plant tissue
or plant products, transmits plant pathogens, attacks a natural enemy of an herbivore or
plant pathogen, attacks pollinators, or attacks organisms that control weeds. The
agency does not have the authority to regulate biocontrol agents that are not plant pests.
4
All biocontrol agents imported for weed control attack plants and are by definition plant
pests. They are, therefore, regulated by USDA.
The USDA requires separate permits for
1) Importation of a plant pest into the U.S.;
2) Movement of a plant pest between States; and
3) Release of a plant pest into the environment.
The federal permitting process requires the submission of PPQ Form 526
(Application for Release) that is forwarded to the HDOA for review and
recommendations. All applications to date, for which HDOA has recommended
rejection, have also been denied by the USDA. If approval is recommended by HDOA,
USDA then reviews the application. This process usually involves review by the
Technical Advisory Group; however, Hawai'i applications are exempt from TAG review
due to the thoroughness of the HDOA review process. A draft environmental
assessment (EA) is requested from the applicant for any requests for the release of
weed biocontrol agents. The USDA prepares the final EA. If endangered or threatened
species potentially are affected by the release of a biocontrol agent then the application
is sent to the U.S. Fish and Wildlife Service for review. A release permit is issued if the
evaluation of the EA produces a finding of no significant impact (FONSI).
The HDOA permitting system differs from the federal system, in part, because of
fundamental differences in purpose as defined in the State statutes. In contrast to
federal law, Chapter 150A of the Hawai'i Revised Statutes (HRS) regulates the
importation of any plant or animal, regardless of whether or not it is a plant pest. The
HRS addresses the importation of non-domestic animals (including reptiles, mammals,
birds, arthropods, and mollusks), microorganisms, and plants. The HDOA permitting
system was not designed specifically for the regulation of biocontrol agents, yet it does
govern their importation and release. Specifically, the HRS prohibits the importation of
all non-domestic animals and microorganisms urliess approved by the Board of
Agriculture.
The regulation of animal importations into Hawai'i has had a long history beginning
in 1890 with the establishment of the "Laws of the Hawaiian Islands" by King David
Kalakaua. This law included language to prevent the introduction of plants and animals
that may become harmful to agriculture. Actual inspections of organisms proposed for
importation were not conducted until 1902 and there were no official reviews of
introductions between 1902 and 1944. In 1944, the Department of Agriculture and
Forestry (now the HDOA) established a policy for Board of Agriculture (BOA) review of
importation requests. This was followed in 1965 by a policy in which advisory
~ s u b _ c m i t t e ~ s s c ~ m p ~ofsspecialists
ed
reviewed the importation applications and
advised the BOA. In 1975, the HRS mandated that BOA review all importation
applications; importation permits were based on the recommendations of a Plants and
Animals Advisory Committee. This law was revised in 1990 to specify three lists of
organisms to be reviewed prior to imponatlon. These were the Prohibited, Restricted,
and Conditionally-Approved Lists. Organisms were not allowed to be imported into
Hawai'i until they had undergone review and had been placed on one of these lists.
Animals placed on the Prohibited List are not allowed into the State under any
conditions. Those on the Restricted List may be imported by government agencies,
municipal zoos and aquariums under fairly restrictive conditions. Organisms on the
Conditionally-Approved List may be imported by all of the above organizations and
additionally by businesses and individuals under specific conditions. Biocontrol agents
always have been placed on the Restricted List. As a result, only government agencies
have been able to import biocontrol agents since the list was established in 1990 (Table
1). This may change in the near future. There are discussions currently underway to
place host-specific agents already established in Hawai'i on the Conditionally-Approved
List. If approved, this modification would allow anyone to import and release these
agents into Hawai'i. A major concern in relaxing importation restrictions on biocontrol
agents is controlling the quality and purity of shipments arriving from insectaries on the
mainland. Mechanisms should be developed for verification of the species identity of the
biocontrol agents shipped and for determination of stock purity for each shipment. This
may be accompl~shedthrough either an insectary certification process or an import
inspection procedure.
Table I.lndlvlduals and Agencies involved In the Introduction of blocontrol
agents into Hawai'i.
Year
IndividualIAgency*
Number of
introductions
68
1893-1900
ROH. sugar plantations. private individuals
1900-1909
HSPA, TOH
84
1910-1919
HDOA, HSPA
45
1920-1929
HDOA, HSPA
89
1930-1939
HDOA, PRI, USDA
81
1940-1949
HDOA, PRI, UH, UC, USDA
53
1950-1959
1 HDOA, HDOH
161
1960-1969
HDOA
101
I 1970-1979
HDOA
96
1980-1989
HDOA, UH, USDA Forest Service
45
1990-1999
HDOA, UH, USDA Forest Service
22
HDOA = Hawai'i Department of Agriculture, HDOH = Hawai'i Department of Health, HSPA = Hawai'i Sugar
Planters Association, PRI = Pineapple Research Institute, ROH = Republic of Hawai'i, TOH = Territory of
Hawai'i, UC = University of California, USDA = United States Department of Agriculture.
The review process for a State importation permit application involves 6 steps.
First, the application is submitted to the HDOA with all of required and pertinent
information, including information on host specificity, distribution, preferred habitat,
temperature requirements, etc. Host specificity studies may be carried out either in the
country of origin or in one of the three approved containment facilities in Hawai'i. The
Advisory Subcommittee then reviews the application. The recommendations from this
subcommittee are passed on to the Plants and Animals Committee for their
recommendations to the BOA. The BOA either approves or disapproves the application.
If approved, the application is submitted to a public hearing process. Comments from
the public are brought back to the BOA for discussion, followed by final approval or
disapproval of the application. If approved, a State permit is issued. The organism may
be imported and released if both State and Federal permits have been issued and permit
conditions are met by the importers.
The HDOA review process for the introduction of biocontrol agents has evolved into
an effective system that screens agents for host specificity and potential negative
impacts on other species. None of the agents introduced since the review process was
initiated in 1975 have attacked any native or beneficial plant or animal species. This
was not the case for introductions before 1975.
IMPORTATIONS
A total of 708 natural enemies were released between 1890 and 1999, of which 286
have become established (Table 2). The majority (237) of these established agents
have contributed to the control of the target pest species. However, 33 (13.6%) also
attacked a different pest or native and/or beneficial non-target species. Native insects
were attacked by 20 (8.2%) of these introduced biocontrol agents. Before 1944, the
year that the BOA began reviewing the applications, only 54.7% of the introduced agents
were host specific. Between 1944 and 1975 when the BOA reviewed all permit
applications, the percentage of host-specific agents introduced increasedto 77.4%. After
1975 host specificity for all released biocontrol agents was 100%; in that year the three
committees (Entomology/Microorganism Subcommittee, Plants and Animals Committee,
and BOA) began reviewing all applications.
Table 2. Types of biological control agents introduced into Hawai'i between 1890 and
I-Insects
I Others
I R:".%,
I
I
I
Number
Established
This change in the host specificity record likely is not due exclusively to changes in
the regulatory process. Public attitude to environmental concerns has changed
dramatically since the 1970's and this also may have strongly influenced decisions made
by the review committees. Prior to the 1970Js,environmental impacts such as host
specificity were of minor concern or of no concern to the BOA and subcommittees
reviewing applications and were rarely if ever discussed. The Board focused on
agricultural concerns almost exclusively. Environmental impacts and host specificity
issues are now often the primary concerns addressed by these review committees.
The current review process has been effective in limiting introductions to hostspecific biocontrol agents, but not very efficiently. The process can take from six months
to one year after the application is received before a permit is issued. This delay is
caused primarily by the requirement for a public hearing. The public hearing process
currently Is changing to a public notification process for which public hearings on each
island will no longer be required, an expensive and time-consuming process. Instead a
call for comments from the public can be made by a notification in the Hawai'i
newspapers. This change is expected to decrease the duration of the review process
from one year to 3-4 months. The results will be the retention of the same high quality of
application review and, hopefully, the continuation of the excellent record of 100% host
specificity in biocontrol agents released since 1975.
LITERATURE CITED
Funasaki, G. Y., P-Y. Lai, L. M. Nakahara, J. W. Beardsley, and A. K. Ota. 1988. A
review of biological control introductions in Hawai'i: 1890-1985. Proceedings
Hawaiian Entomological Society 28: 105-160.
Howarth, F. G. 1985. Impact of alien land arthropods and mollusks on native plants and
animals in Hawai'i. pp. 149-179, In Hawai'i's Ternstrial Ecosystems: Preservation
and Management. C.P. Stone and J.M. Scott (Eds.). Cooperative National Park
Resources Studies Unit, University of Hawai'i at Manoa. University of Hawai'i Press,
Honolulu.
Idishida, G. M. 1994. Hawai'ian ternstrial arthropod checklist, Bishop Museum
Technical Report No. 4.
Timberlake, P. H. 1926. Biological control of insect pests in the Hawai'ian Islands.
Proceedings Hawaiian Entomological Society 6: 529-556.
FOREST PEST BIOLOGICAL CONTROL PROGRAM IN HAWAI't
Clifford W. Smith
Department of Botany, University of Hawai'i at Manoa, 3190 Maile Way, Honolulu HI
96822, U.S.A.
Email: [email protected]
Abstract. Forest weeds were not considered to be a major management problem in Hawai'i until
the latter quarter of the last century. Most previous biological control programs in the state were
against agricultural pests. An interagency committee (U.S.D.A. Forest Service, National Park
Service, Hawai'i Department of Agriculture, Hawai'i Division of Forestry and Wildlife, and the
University of Hawai'i) was established to encourage studies of forest pests, develop biological
control agents and foster the implementation of their recommendations. The progress of
biological control efforts against the following weeds in Hawai'i are presented: Clidemia hirta,
Hedychium gardnerianum, Miconia calvescens, Myrica faya, Passitlora mollissima, Psidium
cattleianum, Rubus ellipticus, and Tibouchina herbacea. Recommendations are made for the
establishment of oversight (action) committees for each targeted species, long-term commitment
to a program once started, more thorough studies in the native range of the target species using
local experts and students and a full-time advocate scientist for forest pest biological control in
the Islands.
Keywords: Clidemia hirfa, Hedychium ganlnerianum, Melastomataceae, Miconia
calvescens, Myrica faya, Myrtaceae, Psidium caffleianum, Passiflora mollissima,
Passifloraceae, Rosaceae, Rubus ellipticus Tibouchina herbacea, Zingiberaceae.
INTRODUCTION
Biological control has been an integral part of forest management in Hawai'i for 100
years. Forest weeds, however, were not considered to be a major management
problem until the latter quarter of the last century. This indifference was in large part the
result of the influence of Charles Lyon who had promoted the introduction of species to
the Islands for watershed reforestation. Also, until quite recently, most naturalists were
interested in native species, particularly the endemics. While they decried the weeds,
they generally did little to control them even in the most critical areas let alone consider
biological control as a management approach. In the early 1980's attitudes began to
change. The National Environmental Protection Act required that federal agencies
develop resource management plans for resources under their jurisdiction. This
formalized planning and revlew process resulted in a professional transformation and
expansion of the National Park Service (NPS) natural resources management program.
Somewhat similar but less extensive modifications occurred in state programs.
The state biological control program operated by the Hawai'i Department of
Agriculture (HDOA) was willing to assist in the development of biological control agents
for forest pests but only as an adjunct to their own mandates. In addition, their
quarantine space was limited and located at sea level in Honolulu, unsuitable for species
from high elevations the typical habitat of the forest weeds that were initially targeted.
Forest managers also realized that a more focused program to promote development of
biological control agents targeting forest weeds was needed. The issue came to a head
during an annual performance review of NPS natural resource management programs
and the Cooperative Pacific Science Unit by NPS Regional Chief Scientist Dennis Fenn
in 1983. He requested that interested agencies meet to discuss the problem. Realizing
that no single agency could support such a program on its own, five agencies committed
to cooperate in a forest pest management program that focused primarily on biological
control.
A Memorandum of Agreement was established between NPS, USDA Forest Service
(USFS), Hawai'i Department of Agriculture (HDOA), Hawai'i Department of Land and
Natural Resources (DLNR), and the University of Hawai'i (UH). The NPS agreed to
convert one of their greenhouses at Hawaii Volcanoes National Park into a quarantine
facility as well as provide a plant pathologist to work on the development of agents. The
US Forest Service agreed to provide a biological control specialist to work on insects in
the quarantine facility as well as act as the quarantine officer for the facility. HDOA was
an important contributor because of its legislated mandate to oversee all biological
control efforts in the state. DLNR proposed to lead the committee and fund programs.
UH agreed to conduct research particularly on the post-release fate of agents through a
monitoring program. All agencies agreed to fund biological research whenever possible.
The Memorandum of Agreement was signed in 1985. Since then eight projects have
come under the review and sponsorship of the Committee to varying degrees. They are
summarized below.
SPECIES TARGETED
-
Clidemia Clidemia hirta (L.)D. Don (Myrtales, Melastomataceae).
See Conant (this volume).
Clidemia is substantially controlled in open ranchland by the thrips, Liothrips urichi
Karny (Thysanoptera, Phlaeothripidae). The leaf spot fungus (Colietotrichum
gloeosporioides f. sp. clidemiae Trujillo Deuteromycotina, Melanconiaceae) has reduced
some populations in rainforest areas. Clidemia is now spreading into lowland dry forest.
Control is by no means complete and further agents are still needed for this species. As
Conant (ibid) notes there are several potential insect agents. None, however, show
much potential to control this weed. It may well be that there is no realistic hope to
contain it in rainforest situations.
The negative impact of this species needs to be reevaluated in light of recent
introductions. It may be too early to tell if the seed predators are having any impact but
Myers' (this volume) comments on their potential efficacy suggests that we should not
expect any dramatic effects. Further studies should be conducted in Central America
and directed at stem borers and defoliators. The studies should be long-term,
conducted year-round, and focus on forested areas.
-
Bao_an~~pokaPassiflora mollissima (Kunth.&.H. Bailey (Passifloraceae).
This project has been led by DLNR since the early 1980's with considerable involvement
of U.S.F.S. in the 1990's. Pemberton (1989) conducted the initial exploratory research
in South America and noted that there was considerable potential to manage this
species with biological control agents. Later exploratory research was focused in
Colombia on his recommendation. However, the political instability of the region, a lack
of leadership in the program and the absence of an oversight committee has hampered
the project. The USFS sponsored research in Merida, Venezuela, for several years.
The following insects have been studied and some released.
Pyraustra perelegans Hampson (Lepidoptera: Pyralidae) feeds on leaves and buds.
It has been released in 1881 with little effect. It is established on the Big Island but
population levels are extremely variable. Most people assume that the insect was
unable to overcome the many generalist lepidopteran parasitoids in the Islands. R. Leen
(pers. comm.) suspects that a species of the fungus Metschnikowia (Ascomycete:
Saccharomycetales) is responsible for the poor performance of the insect.
Unfortunately, no definitive study has been conducted to differentiate between these two
hypotheses. However, other hypotheses need to be considered also, e.g., the climatic
conditions are unsuitable. Unfortunately,the reason why a released insect does not live
up to its potential is rarely studied formally. A few anecdotal notes are sometimes
published.
Cyanotrica necyria Felder (Lepidoptera Notodontidae), a leaf feeder from Ecuador
and Colombia was released in 1988. It has established but has had no demonstrable
effect. Further work on this species is desirable because it has a high potential
completely defoliating plants.
Josia fluonia Druce (Lepidoptera: Notodontidae), a defoliator, has been
recommended for release but is awaiting final approval. One experiment suggested that
it could complete its life cycle on apple but the few insects that did complete their life
cycle were in very poor condition. Recent experiments have shown that it can survive
on the edible passionfruit (P. edulis Sims f, flavicarpa Deg.) suggesting that the proposal
for release should be reconsidered. Further work on this species is not recommended
because the insect does not appear to have a significant impact on the target plant.
Josia ligata Walker (Lepidoptera: Notodontidae), a defoliator, was brought into
quarantine but the colony did not survive.
Zapriotheca nr. nudiseta (Diptera: Drosophilidae) larvae feed on flower buds. It has
passed host specificity testing, but has not been proposed for release yet. This colony is
certainly highly inbred. It appears to have considerable potential in disrupting the
reproductive cycle of banana poka. Further importation of the insect is recommended to
overcome genetic problems and enable host screening to be completed. It will be
extremely difficult to assess the impact of this insect because large plants are needed.
The logistics of handling such plants in quarantine are unrealistic and field studies in
South America would be extremely difficult under current political conditions.
A fungus, Septoria passiflome Sydenham (Deuteromycetes, Dothidiaceae),was
released 1986 and has had an apparently dramatic defoliating effect in Laupahoehoe,
Hawai'i Island (D. E. Gardner, pers. comm.). Confirmation of the cause of defoliation is
important in this case because previous defoliation events were attributed to drought
conditions. Thorough evaluation of the effects of previous releases in the banana poka
biological control program should be conducted before further work is considered.
Species of Odonna (Lepidoptera, Oecoriphoridae), a root crown borer, and Dasyops
(Diptera, Lonchaeidae), a stem borer, should be studied in South America to obtain data
on life history, host specificity, and impact. The Dasyops has been brought into
quarantine in Hawai'i where though the insects failed to mate they laid eggs profusely.
-TheseinsectsaFekno-whtoattackbana~apoka~~~r~&~sidebe~usefa~~~es-for
handling them experimentally were not available at that time.
Two other species are becoming serious weeds the sweet granadilla (P. ligularis
Juss.) and yellow granadilla (P. laurifolia L.). Unlike the established rnelastomes, all of
which can be targeted because the whole family is considered noxious, one member or
the family, P. edulis, is a marginal agricultural crop. Many people, however, harvest it in
the wild for desserts, jams, etc.
P. mollisima is a species which illustrates the weakness of the recent approach to
biological control against forest pests in the Islands. During the 1950-70's there was
considerable enthusiasm for the establishment of a forest industry in the state. Banana
poka was a threat to the prized koa timber market because it smothered the natural
regeneration of the forest as well as damaging large trees due to the weight of the vines,
especially when wet. When it was realized that large-scale forestry was unfeasible
interest in forest problems declined and with it support to combat banana poka. It is still
a serious problem in native forests on the Big Island and has also become established in
Kula, Maui. Other weeds, e.g., miconia, strawberry guava have supplanted interest and
financial support for banana poka. The whole program is now in abeyance.
Cooperation with similar control efforts in New Zealand is possible.
-
Himalayan raspberry Rubus ellipticus Sm. (Resales, Rosaceae).
A small cooperative exploratory program was established with the Chinese Academy of
Agricultural Sciences Institute of Plant Protection, Beijing, In 1996 to look for diseases
and insects that attack this species as well as R. niveus Thunb. in the Himalayan region
of China. Earlier attempts in India to identify potential agents targeting this species
failed due to various problems but particularly the remote locations of most known
collection sites. No evaluation of potential agents in northern Thailand has been
attempted.
Attempts to establish a Rubus action committee have not been successful because
nobody wants to lead it even though there is a strong interest to control the plant in the
conservation, hunting and recreation communities. An action committee to coordinate
the project, provide the necessary oversight and develop funding is necessary if this
project is to move forward.
The danger of non-target impacts is significant because of the coexistence of two
native congeners, R. hawaiiensis Gray and R. macraei Gray. Previous agents
introduced against R. argutus Link have attacked these species (Pemberton this volume)
although with little apparent negative effect. Since no long-term monitoring was
established when the insects were released, retrospective evaluation is virtually
impossible. It is somewhat surprising that nobody has used the release of biological
control agents to study the epidemiology of new arrivals in the Islands. Excellent
opportunities for studying fundamental principles of island biology are being missed.
Biological control is also without fundamental information that would probably enhance
the success of future releases, particularly when so many previous releases failed.
-
Fayatree Myrica faya Ait. (Myricaceae).
A previous attempt to control fayatree failed (Hodges & Gardner 1985). Eucosma
smithiana (Walsingham) (Lepidoptera, Tortricidaae) was released in 1956. It is
established on M. cerifera but not M. faya.
The current project, coordinated by the Fayatree Action Committee, was led by the
"Big Island Resource Conservation and Development Committee" in 1987, a local
program of the USDA Soil Conservation Service (SCS), US. Department of Agriculture
Resource Conservation and Development Agency on Hawai'i Island. The committee
eventually stopped meeting in 1995 after many years of effective work soon after the
RC&D lead person left the islands. Strong political support from E g Island legislators
continued funding through the Governor's Agricultural Coordinating Committee that was
ultimately subordinated into the Hawai'i Department of Agriculture. Some of this funding
was later reprogrammed into similar work on melastomes at the suggestion of agency
personnel.
Caloptilia nr. schinella (Lepidoptera, Gracillaridae) from the Azores and Madeira
was released in 1991. It is established but has had no demonstrable effect. It is
possible that leaf miner parasitoids may be attacking this moth (Conant, this volume).
Cooperative programs with the University of the Azores failed to find any further suitable
biological control agents for M. faya in its native range in the Azores or Madeira. Most
insects found on fayatree had alternate hosts, such as Vaccinium, that made them
unsuitable agents. A Septoria leaf spot fungus did cause premature leaf fall, but we
were unable to obtain fertile material. A similar species, Septoria hodgesii Gardner
(Deuteromycetes, Dothideaceae),from M. cerifera L. in the eastern US was released at
Volcano, Hawai'i Island in 1998, but with no noticeable impact to date. Trujillo (pers.
comm.) has suggested that acid rain around the Volcano area inhibits spore
germination, and that the fungus may be more successful if tested elsewhere.
We have a single species in quarantine from Madeira, Phyllonorycter myricae
Deschka (Lepidoptera, Gracillaridae) that is may be suitable for release. However, the
colony has undergone considerable inbreeding and it has not been possible to replenish
the stock. In addition, establishment success of this microlepidopteran likely would be
jeopardized by parasitoids already established in the Islands. Furthermore, we have
little evidence that this agent would have a significant impact on its target. This species
was probably a poor choice as a potential agent using Balciunas' cr~teria(this volume).
The absence of significant agents in Macaronesia may be explained by the fact that
the Macaronesian colonies of fayatree were components of an isolated vegetation type
now an outlier of the once more wldely distributed laurisllva (sub-tropical rainforest)
widespread in the Mediterranean and near East during the Tertiary. Insects adapted to
fayatree may not have reached the distant Macaronesian Islands that are at least 1000
km from the Iberian Peninsula. Populations of fayatree in many areas of Portugal north
of Lisbon appear to have been planted. Natural populations from the Algarve, Portugal,
and the Atlas Mountains were not relocated.
We have also started work on the herbivores of related species of fayatree in
Venezuela where several potential agents, including a promising stem borer, have been
identified. Further development of these species is on hold awaiting their shipment to
Hawaii. There is little hope of getting the necessary permits until the political situation in
the country has settled down.
Fayatree continues to expand in natural areas and ranchland. Though not a high
profile weed, Vitousek and Walker (1989) have shown that it modifies ecosystem
processes. These modifications are of such magnitude that fayatree remains among the
highest priority weeds for the Forest Pest Biological Control Program.
Kahili ginger - Hedychium gardnerianum Roscoe (Zingiberales, Zingiberaceae).
Anderson and Gardner (1999) have studied a strain of Ralstonia solanacearum (E.F.
Smith) Yabuuchi et at. (Bacteria, Pseudomonaceae) that attacks kahili ginger. They
expressed considerable optimism that this fungus has the potential to bring about longterm control of this pest including the suppression of seedling establishment.
The slow-acting nature of this pathogen may be beneficial in that native species will
have a chance to recover before weeds overwhelm them. The establishment of the
bacterium is somewhat difficult generally requiring physical damage. Nevertheless, the
bacterium has been established in some populations in the Islands. Evidence suggests
that the density of plants in these areas is declining. Perhaps the most encouraging
aspect of this disease is that seeds do not germinate or damp off soon thereafter where
the bacterium is present in the soil. EPA approval may be required before the fungus
can be broadcast as a biocide. In the meanwhile, use of this pathogen is limited to local
application only though experiments on mass culture, optimal dosage, and alternative
inoculation techniques are undenrvay.
A potential conflict of interest is that it attacks edible ginger (Zingiber officinele
Roscoe - Zingiberales, Zingiberaceae). The difficulty of dissemination and
establishment of the bacterium suggests that this concern is not a significant problem.
The infestations of kahili ginger are well above the areas were ginger is grown
commercially.
spectrum of interests, both pro and con, regarding the management of the weed. The
committee chair should be willing and able to meet frequently, write testimony, lobby,
enable the various interest groups to interact effectively, etc. Without good leadership
these committees soon dlsassoaate. The comm~ttee'sfocus should be management of
the weed using all effective strategies not just biological control. It should support
research on all promising methodologies. Biological control projects require oversight to
provide focus and an Interlace with a community increasingly opposed to any species
introductions. The action committee also provides an important review of the scientific
program. The research team is more effective if freed from fund-raising and other
administrative duties. See Markin (this volume) for further discussion on action
committees.
The program on biological control of forest weeds in Hawai'i now is in need of a
scientific leader, who is prepared to champion biological control, discuss the issues, lead
the research program, and work in foreign countries supervising and collaborating with
foreign cooperators. The head of an action committee and the scientific leader,
however, should not be the same person. From my own experience, I found it difficult to
lobby for money for projects I was supervising. It creates the appearance of ethical
problems due to a conflict of interest. The scientific leader should be available as a
resource person to the chair of the action committee during the lobbying process.
In my view, the broadening of interest of invasive species committees has become
one of the biggest impediments to effective development of biological control agents for
forest pests. Island-based invasive species committees compete for state funding.
There is now a Big Island lnvasive Species Committee in addition to the original
committee on Maui. Committees for other islands are being formed. These will compete
for resources because each island's needs are somewhat different. lnvasive species
committees also focus on all pest species, not just weeds, so their attention is spread
thinly over an ever-increasing array of species. Biological control is likely to become a
lower priority management strategy when such a broad array of pests is considered.
Some people disapprove of biological control because of the problems associated with
the introduction of predators for mongoose, carnivorous snails, etc., many years ago.
Unfortunately, their anathema carries over to the biological control of weeds where the
track record of success with few side effects is very good (Pemberton, this volume,
Reimer, this volume). Because these broad-based committees are effective solicitors of
funds, their success may effectively reduce funding for biological control projects.
Nevertheless, the invasive species committees may replace single species action
committees with good results if they are prepared to foster biological control studies over
the necessary 7-10 year research and implementation period.
Another major problem facing biological control of natural area weeds is the lack of
a critical mass of specialists. All too often programs disintegrate once a key individual
moves elsewhere or retires. The gradual dissolution of the Fayatree Action Committee
was the result of the chairman's rotation back to the continental U.S. Active members in
the program felt uncomfortable stepping into his shoes. Other members did not have the
time or inclination to manage the necessarily frank discussions. The research program
itself is also susceptible to disruptions from personnel changes. The banana poka
project died when the principal investigator took another job. His replacement had a
different research agenda. Having someone take over a program midstream is
extremely disruptive. Without very careful job performance management the project can
fall apart rapidly.
Perhaps the most important element of a successful biological control program is an
effective advocate: someone who will promote biological control to all audiences (from
administrators to legislators to land managers to conservationists); a person who
maintains a network of colleagues around the world; someone prepared to manage
cooperative agreements with researchers in the countries of origin of target organisms;
someone who can work with federal, state and NGO funding agencies or individuals;
someone who appreciates the biotic interference problems in the state. It might be said
that such a person could no longer be a research scientists. However, there are ample
opportunities for collaborative work especially with foreign collaborators as well as
preparing synthesis papers. Without such a person the "Forest Pest Biological Control
Program in Hawai'i" will lurch along perhaps with some successes, more likely not, until
it disappears.
LITERATURE CITED
Anderson, R., & Gardner, D. E. 1999. An evaluation of the wilt-causing bacterium
Ralstonia solanacearum as a potential biological control agent for the alien kahili
ginger (Hedychiumgardnerianum) in Hawaiian forests. Biological Control 15: 89-96.
Burkhart, R. 199.6. The Search For Biological Control Of Miconia calvescens:
Photographic Documentation of Natural Enemies of Miconia calvescens
(Melastomataceae) found in Central and South America between July 1993 and
September 1995. http://www.botanv.hawaii.edulfacultvlcw smith/
mc control.htm
Conant, P. this volume. Classical biological control of Clidemia hitta (Melastomataceae)
in Hawai'i using multiple strategies
Hodges, Jr., C. E., & D. E. Gardner. 1985. Myrica Faya : Potential Biological Control
Agents. Technical Report 54. Cooperative National Park Resources Studies
Unit, University of Hawai'i at Manoa.
Killgore, E. M. this volume. A case study in weed biological control practices: Miconia
calvescens DC and three fungal pathogens
Pemberton, R.W. 1989. Insects attacking Passiflora mollissima and other Passiflora
species: field survey in the Andes. Proceedings, Hawaiian EntomologicalSociety
29: 71-84.
Pembert~~fhisuolume~~Predictable_risk-t~ati~-ep1ants~l-cont~oLofweeds in Hawai'i.
Smith, C. W. 1985. Impact of alien plants on Hawaii's native biota. pp. 180-250, In:
Hawaii's Ternstrial Ecosystems: Pmservation and management. Stone, C. P. and
J. M. Scott (ed). Cooperative National Park Resources Studies Unit, University of
Hawaii, Honolulu.
Trujillo, E. E. 1985. Biological control of Hamakua parnakani with Cercosporella sp. in
Hawaii. pp. 661-671 in Proceeding of the VI symposium on the biologicalcontrol of
weeds. August 19-25, 1984. E . S . Delfosse, ed. Vancouver, Canada. Agriculture
Canada.
Trujillo, E. E., M. Aragaki, and R. A. Shoemaker. 1988. Infection,diseasedevelopment,
and axenic culture of Entyloma compositarum, thecause of Hamakua pamakani
blight in Hawaii. Plant Disease 72: 355-357.
Vitousek, P.M., and L.R. Walker. 1989. Biological invasion by Myrica faya in Hawai'i:
plant demography, nitrogen fixation, ecosystem effects. Ecological Monographs 59:
247-265.
Wikler, C. & Smith, C.W. this volume. Strawberry Guava (Psidium Cattleianum):
Prospects for Biological Control
A RESOURCE MANAGER'S PERSPECTIVE ON THE ROLE OF
BIOCONTROL IN CONSERVATION AREAS IN HAWAI'I
J.T. Tunison
National Park Service, US Department of Interior, Hawaii Volcanoes National Park, P.O.
Box 52, Hawaii National Park, Hawai'i 96718, U.S.A.
Email: Tim_ [email protected]
Abstract. Resource managers in Hawaiian conservation areas view biological control as an
important component of forest weed management. Biological control is not regarded as a
panacea for all widespread weeds. However, it is perceived as an additional tool for selected
appropriate species for which conflicts with economic interests or non-target species are not
present. Biological control is employed as a control method, along with exclusion, rapid
response, control of disturbance sources, chemical and manual control, and cultural practices
such as seeding or outplanting. The role of biological control in the management of forest weeds
in conservation areas is described.
Key words: biological control, Hawaiian forest weeds, control strategies
INTRODUCTION
Many widespread alien plants have invaded protected natural areas in Hawai'i. Nearly
100 species have been identified as threats to natural areas because of their ability to
form monospecific stands or to markedly alter ecological processes (Smith 1985). Many
of these disruptive weeds are already distributed widely in some natural areas,
particularly in the lowlands. Others are incipient pests inside natural areas or nearby.
Resource managers with the National Park Service, US Fish and Wildlife
Service, The Nature Conservancy of Hawai'i, the Hawai'i State Division of Forestry and
Wildlife Natural Area Reserves, and private reserves have developed strategies to
manage weeds in their natural areas. These approaches emphasize chemical, manual,
or cultural control techniques. Manual and chemical techniques of weed control
practiced in conservation areas are not applicable on a landscape level, at least at
current and foreseeable funding levels. In addition, follow-up treatments are needed at
regular intervals for the indefinite future for widespread weed species. Resource
managers perceive biological control of widespread weeds as an attractive tool for longterm control of widespread weeds on a landscape level. Even though an interagency
Memorandum of Agreement was developed In the early 1980s, to date, biological control
has not contributed effectively to the control of weeds in natural areas of Hawai'i.
WEED CONTROL STRATEGIES
Statewide and island wide strategies.
The potential importance of biological control in the management of Hawai'i's forest
weeds can be characterized by describing its role in the overall strategy for alien plant
control in natural areas (Figure 1). Biological control may have a role in the control of a
number of widespread weed species, but is not appropriate for certain species just
becoming established in Hawai'i.
Control strategies begin, ideally, with exclusion on national, statewide, or
islandwide bases. Exclusion is the most cost-effective strategy in alien plant
management, keeping potential weed species out of Hawai'i or preventing their
movement among islands. However, the system of exclusion and quarantine practiced
by federal and state agricultural agencies lacks the regulatory teeth or funding resources
effectively to keep potentially invasive species out of Hawai'i. Current lists of prohibited
species, screening, and quarantine are highly selective and exclude few species from
entering Hawai'i. Effective exclusion would require major legislative and policy shifts at
both the federal and state levels to modify methods by which proposed introductions are
assessed for their potential risk and impacts. Natural areas managers and researchers
with the US Geological Survey-Biological Resources Division (BRD) are preparing data
bases on current or potentially invasive species to reduce potentially invasive species'
introductions to the state.
A rapid response capability is needed to eradicate or contain incipient invasive
species. A rapid response approach has begun on an island-wide basis. Haleakala
Natlonal Park and BRD lnltiated an effort on Maui to contain or eradicate Miconia
calvescens DC (Myrtales, Melastomataceae), a stand-forming tree from tropical
America, to prevent its establishment in the park. The rapid response effort on Miconia
control evolved into the Maui lnvasive Species Committee (MISC), funded by local,
state, and federal agencies and set up to contain or eradicate other incipient invasive
plant species island-wide on Maui. lnvasive species committees have developed more
recently on Hawai'i Island and Kauai to address the rapid response needs on these
islands. However, a committee approach is not viable in the long term. One state
agency should be given the mandate for rapid response to incipient invaders and to
which and interagency committee could contribute.
Biological control is probably not the appropriate tool for management of some
newly established species, but is generally more appropriate for widespread species.
However, biological control should be considered for serious weed pests that appear to
be uncontainable. The current response to Miconia provides an example. There is little
doubt that Miconia is a serious threat to both managed and natural ecosystems in
Hawai'i and that a situation similar to that in Tahiti should be avoided if at all possible
Considerable effort has been put into containing the infestation while biological control
research was undertaken. One agent, a fungal pathogen (Killgore, this volume), has
been released, but further funding for exploratory research has been limited. Now that it
is generally accepted that eradication of Miconia is not feasible and containment efforts
over broad areas will continue indefinitely (Duane Nelson and Lloyd Loope, personal
communication), a renewed effort at developing effective biological control agents has
begun. Miconia is a highly suitable target for biological control in that potential nontarget hosts and conflicting economic interests are minimal. Closely related species in
the family Melas&mataceae are all introduced; it is a family that contains many
undesirable invasive species.
Even if exclusion of new introductions is tightened and rapid response
capabilities are strengthened, new invasive species will arrive in Hawai'i and become
widespread and the populatlons of other species will expand and become Invaslve.
Moreover, many ecologically disruptive species already have become widespread in
Hawai'i (Smith 1985). Biological control is an appropriate strategy for only a subset of
these species. The most suitable candidates for biocontrol are those for which minimal
non-target impacts are expected and no conflict with economic interests exist (Davis and
Gardner 1982). There are also greater chances of developing successful biological
control agents if the native range, taxonomy and biology of the target species are well
known. Most study of the biology, handling techniques and efficacy assessment ideally
should be studied in the native range of the target host y wherever possible.
---
Conservation area strategies
Other alien plant control strategies are implemented in parks, refuges, preserves,
reserves, and other conservation areas (Figure A), once invasive weed species become
established. A cultural technique of weed control practiced by all conservation agencies
is to remove sources of disturbance; in Hawai'i this entails exclusion of alien ungulates
such as feral goats and pigs, axis deer, or mouflon sheep. Hawai'i's flora evolved
without selective pressure from large mammalian herbivores and lacks defenses against
browsing. Hawai'i's introduced ungulates severely browse native species, disturb soil,
and disperse weed seeds, thereby facilitating the spread of introduced plant species.
The importance of alien ungulates in the spread of alien plant species is shown in
natural ungulate exclosures on cliff faces, in pit craters, or on isolated mountain tops, all
areas without a history of ungulate browsing where native vegetation flourishes. Such
areas are largely free of weed species (Loope and Scowcroft 1985, Belfield 1998). In
fenced exclosures from which alien ungulates have been removed, native vegetation
often recovers and the spread of weeds is inhibited (Loope and Scowcroft 1985).
Most managers of conservation areas practice a form of rapid response for newly
introduced species known to be invasive in other areas. Often they arrive in
conservation areas in predictable corridors of invasion such as along roadsides, trails, or
developed areas. Monitoring and attacking new populations while they are still localized
is a highly cost-effective approach, because species are controlled while they still can be
eradicated readily, rather than attempting control after they have become widespread.
Containment rather than eradication is implemented for some localized or even
widespread species. The goal of containment is to prevent further spread and typically
is executed by eradicating satellite populations first and then attacking the more typically
dense core populations (Tunison et a/. 1994).
Control of species widespread in conservation areas is often intractable because
of limited financial resources. One approach pioneered by national parks in Hawai'i is to
control widespread weeds in high priority sites within conservation areas even though
these species are not controllable throughout the protected area. In national parks, the
targeted areas are called Special Ecological Areas (SEA) (Tunison & Stone 1992).
These areas typically are selected for their intact native plant communities, high degree
of manageability, high biological diversity, concentrations of rare species, particularly
representative or unique characteristics, potential for research and interpretation, and
other criteria. All ecologically disruptive, widespread alien plants are controlled in SEAS
by manual and chemical control methods.
The SEA approach has a number of shortcomings, indicating the need for a
- e o n e t l m ~ i e l o g i ~ k ~ ~ p p ~ o a 6 h ~ ~ ~ ~ @ i e d - o ~ a - landscape level; it is most effective where weed densities are low. Recurring control
efforts are needed on an annual or multiyear basis, for the indefinite future. Unmanaged
lands in which the target widespread weeds are present often surrounded SEA'S. Even
when the local seed bank in the SEA is exhausted, seed dispersal into the target area
will continue.
Another alien plant cultural control technique being developed in some of
Hawai'i's conservation area is restoration of native communities by outplanting and
seeding native plant species. Manual and chemical control is effective in bringing about
native plant recovery in conservation areas, when carried out in areas where alien plants
occur at low densities in otherwise largely intact communities. In those cases, native
plants colonize the small gaps created by removal of scattered alien plants. However,
outplanting and seeding are now being used, along with alien plant control, in sites
highly altered by alien species. The newly established native vegetation reintroduces
seed sources and provides competition for alien species.
For conservation areas, biological control is an important tool to be implemented
in conjunction with manual, herbicidal, and cultural control techniques, rather than
instead of these techniques. Only a subset of the widespread weeds affecting natural
areas are suitable candidates for biological control; most widespread weed species will
need conventional approaches. For those species which are suitable candidates for
biological control, finding, testing, and releasing effective biological control agents may
take a number of years. More time may be required for populations of biocontrol agents
to build up to have an impact on the target weed species. In the meantime, valuable
biological resources in conservation areas need require protection from the impacts of
ecologically disruptive alien plants. This requires chemical and manual control. Also,
biological control may be only partially effective and vary in effectiveness by habitat.
Chemical and manual control should be used to reduce densities of the target alien plant
species to acceptable levels Funding for the necessary research for development of
biological control agents will be difficult to obtain until the efficacy of the approach can be
clearly and concisely demonstrated.
LITERATURE CITED
Belfield, T. R. 1998. Botanical survey of Kilauea Volcano East Rift craters, Hawaii
Volcanoes National Park. Technical Report 122. Cooperative National Park
Resources Studies Unit, University of Hawai'i at Manoa.
Gardner, D. E. and C. J. Davis. 1982. The prospects for biological control of nonnative
plants in Hawaiian National Parks. Technical Report 45. Cooperative National
Park Resources Studies Unit, University of Hawai'i at Manoa.
Loope, L. L. and P. G. Scowcroft, 1985. Vegetation response within exclosures in
Hawai'i: A review. pp 377-400, In Hawai-i's Ternstrial Ecosystems:
Presen/ation and Management. C. P . Stone and J. M. Scott (Eds.) Cooperatlve
National Park Resources Studies Unit, University of Hawai'i at Manoa. University
of Hawai'i Press, Honolulu.
Smith, C. W. 1985. Impact of alien plants on Hawai'i's native biota. pp 183-249, In
Hawai'i's Terrestrial Ecosystems: Preservation and Management. C.P. Stone
and J.M. Scott (Eds.). Cooperative National Park Resources Studies Unit,
University of Hawai'i at Manoa. University of Hawai'i Press, Honolulu.
An app~oadrtcalierr
plant control in Hawaii Volcanoes National Park. pp. 781-798, In Alien Plant
Invasions in Native Ecosystem of Hawai'i: Management and Research. C. P.
Stone, C. W. Smith, and J. T. Tunison (Eds.). Cooperative National Park
Resources Studies Unit, University of Hawai'i at Manoa. University of Hawai'i
Press, Honolulu
T-lIt,,,,,,9-a--
Tunison, J. T. N. G. Zimmer, M. R. Gates, R. M. Mattos. 1994. Fountain grass control
in Hawai'i Volcanoes National Park, 1985-1992. Technical Report 91.
Cooperative National Park Resources Studies Unit, University of Hawai'i at
Manoa.
107
FIGURE 1. ALIEN PLANT CONTROL STRATEGIES FOR
CONSERVATION AREAS
I RAPID RESPONSE I
ESCAPES
CONTAINMENT
Statewide or Island-wide Strategie
Conservation Area Strategies
WIDESPREAD
WEEDS IN
STRAWBERRY GUAVA (Psidium cattleianum):
PROSPECTS FOR BIOLOGICAL CONTROL
Charles ~ i k l e rand
'
Clifford W. smith2
' UNICENTRO, Central-Westem State University1FUPEF, P.O. Box 21, BR 153 - Km
07, Bairro Riozinho, 84500-000, Irati, PR Brazil.
Email: Charles@irati unicentro.br
Department of Botany, University of Hawai'i at Manoa, 3190 Maile Way # 400,
Honolulu, HI 96822 USA.
E-mail: [email protected]
Strawberry guava - Psidium cattleianum Sabine (Myrtales, Myrtaceae) was
introduced to Hawaii about 1825 and quickly escaped from cultivation. It is now found on all
inhabited islands where it is one of the most important forest weeds. The small trees form
extremely dense monotypic stands up to 10 m in height. Four insect species were found that
have significant deleterious effects and, therefore, considerable potential. A leaf gall produced by
Tectococcl~sovefusHempel (Homoptera, Eriococcidae) is the most promising due to the damage
caused and the ease of handling. Bud galls, precocious developments of the bud that terminate
shoot growth, are formed in response to Dasineura gigantea (Diptera, Cecidomyiidae). A seed
gall induced by Eurytoma psidii Thurdczy and Wikler, in ed. (Hymenoptera, Eurytomidae)
cements groups of seeds together and prevents germination of all seeds in the fruit. These latter
two species have a significant impact but are more difficult to manipulate in quarantine. A shoot
gall produced by Eurytoma cattleianii or Eurytoma desantisi (Hymenoptera, Eurytomidae)
terminates further growth of the shoot. The species are difficult to handle in the laboratory. The
biological control potential of three additional species is discussed: a leaf gall formed in response
to Neotrioza tavaresi (Hemiptera, Psyllidae) which, though species specific, has little impact on
the plants; a sawfly, Haplostegus epimelas Konow (Hymenoptera, Pergidae) is unsuitable
because it attacks commercial guava occasionally; and, a chrysomelid Lamprosoma azureurn
Germar (coleoptera, Chrysomelldae) IS not recommended because it attaCKS a number OT
myrtaceous species.
Abstract.
Keywords: Dasineum gigantea, Eurytoma, Haplostegus epimelas, Lamprosoma
azureum, Myrtaceae, Neotrioza tavaresi, Psidium, Tectococcus ovatus.
INTRODUCTION
-Strawbe~fy+avapsiditlm-cattIeianumSabne-Myrtales,Myrtaceae)-was-intcoducedin the Hawaiian Islands around 1825 (Wagner ef a/. 1990). Since then it has spread into
mesic and wet environments from sea level up to 900 meters. Smith (1985)classified it
as one of the top ten weeds of Hawaiian native forests, a status it still maintains. It is a
weed of many tropical and subtropical islands throughout the world. It does not tolerate
frost. Though it can be found in tropical continental environments and has naturalized in
some places it only becomes a pest in insular ecosystems.
This fruit is eaten and occasionally made into a juice though the commerc~al
'strawberry guava juice' is a mixture of strawberry and guava juice. The plant is
frequently featured in Japanese-style gardens for its smooth multicolored bark
contrasting with shiny, dark green leaves and toleration of pruning and shaping. These
potential conflicts of interest are negated by its invasiveness, its tendency to form
monotypic stands and the abundant fruit host several species of fruit flies. These flies
preclude the export of untreated soft fruit from the Hawaiian Islands, a serious
impediment to the development of a tropical fruit industry in the Islands.
Manual control efforts are extremely labor intensive. The plants resprout readily
from cut stumps. Grubbing it out of the ground is the only effective mechanical control, a
method with profound, generally undesirable ecological consequences and
impermissible in areas of archaeological interest. Using herbicides is not very effective.
Spraying the trees requires the use of surfactants to wet the leaves but the shoots soon
send out new shoots. Cut-stump and girdling techniques are not very effective except in
areas where there is a drought season. Treatment of plants in wet areas appears to be
effective but within two years shoots, initially d~stortedbut later more normal, are
produced. There are no known ecological techniques to control this weed. Biological
control is, therefore, the last resort.
This report documents the results of the search for potential biological control
agents of strawberry guava. One of the primary charges to the research team was that
all agents must not attack the commercially important P. guajava. Initially, this stricture
appeared to be insurmountable but three gall forming insects were discovered that were
restricted to the target species. Preferred study sites had P. cattleianum and P.guajava
growing in close proximity to one another. The laboratory and garden experiments were
conducted at the Forest Protection Laboratory, Federal University of Parana, Curitiba.
Sampling & Observations
Psidium caffleianum - yellow and red varieties, common guava (Psidium guajava L). and
other myrtaceous species including Surinam Cherry (Eugenia uniflora L.), gabiroba tree
(Campomanesia xanthocarpa Berg) and Eucalyptus spp. were studied in the field.
These species were also grown in large park areas of the Centro Politecnico where the
potential agents were later released for further host range evaluation under natural
environmental conditions.
Once the habits of the gall-forming insects were better understood, all associated
plant species in the immediate environment of strawberry guava populations were
examined for the insects or their galls. From this study, a checklist of species on which
the galls were not found was produced thereby providing an initial evaluation of host
range specificity.
The biology of each gall-forming insect included evaluations of hyperparasites and
predators. The potential impact of natural control agents against the candidate species
was assessed from observations but no quantitative studies were conducted. This
information was then used in recommending the priority of each candidate for
importation to the control area.
Investigation in the area of origin
Early exploration for strawberry guava was not encouraging. Hodges (1988) provided
the first inforrnatlon on potential agents and recommended restricting further studies to
southern Brazil, suggesting Parana State as the best place for further research. Some
of the species that he thought might be potential agent, (e.g., the bark beetle Scolytopsis
bmsiliensis Eggers (Coleoptera, Scolytidae), were later found on other species.
The yellow-fruited variety of the plant is relatively common in the restinga as
scattered individuals or in small groups, never in thickets as in Hawai'i. The red-fruited
variety was first found two years later and then only on the First Plateau. The plants are
similar to several closely related species resulting in considerable initial uncertainty in
determining the habitat and range of the species particularly when fruit are absent. It
has been difficult to find large populations of P. cattteianum.
RESULTS
Pathogens
Few pathogenic fungi were found on P. cattleianum and none of them were host
specific. None had any significant impact on the plants (Hodges 1988). Subsequent
fieldwork by Barreto (Federal University Vi~osa,MG) has also proven negative.
There is a very common leaf spot disease that also occurs on P. longipetiolatum. It
has not been identified nor has it been cultured. No similar disease has been reported
previously on P. caffleianum. It has not been found on P. guava but it effects were
minimal and it has no potential for biological control.
Hodges (1988) also found a tarspot fungus, Phyllachora subcircinans Speg.
(Phyllachorales, Phyllachoraceae), on leaves of a few trees near Paranagua. This
fungus was previously reported on P. caffleianum but we have not found it.
The only fungus collected by Hodges on P. guajava was the rust Puccinia psidii
Wint. (Uredinales, Pucciniaceae) that he did not observed on P. caftleianum. It is,
however, very common on P. cattleianum in Curitiba. It is very common in many plants
of the Myrtaceae family and cannot be used in biological control due to its lack of
specificity.
Insects
During an initial survey 133 species from 80 families in 13 orders were found feeding on
strawberry guava. The most common orders were Coleoptera, Diptera, Hemiptera,
Hymenoptera, Orthoptera, Homoptera and Lepidoptera. The Coleoptera were the most
diverse and abundant; 39% of all families, 47% of the species and 53% of the
collections. The next most abundant were the Hemiptera with 14% of the families and
18% of species. Only 31 species were restricted to the Myrtaceae. Of these only five
species are specific enough to be considered for biological control.
Leaf Gall - Tectococcus ovatus HemDel, 1900 (Homoptera. Eriococcidae).-The gall is
convex oval on one side of the leaf, and acuminate oval on the other. The acuminate
portion is generally on the upper side of the leaf whether or not that is the abaxial
surface. Occasionally galls may have acuminate or convex forms on both sides of the
leaf. The size of the galls varies depending on the developmental stage and sex of the
insect, those containing adult males are narrower and more acuminate than females.
The galls are the same color as the leaf though the tips are frequently red (Vitorino
1995). The insect was easily handled in the laboratory and readily transferred from one
plant to another. It is multivoltine.
~e~~t~~o~~us~ovat~sT~~distrib~te-d~t
hroughoutth~-ra - n g e - d I-but it is much more frequent on the red-fruited form on the First Plateau.
There is one parasitoid, Metaphycus flavus (Hymenoptera, Encyrtidae). The
percentage parasitism in Brazil is 49%. There is also an ectoparasite Aprostocetus sp.
(Hymenoptera, Eulophidae) but the percentage parasitism is only I%. Both species are
present in Hawaii. There is also a predator, Hypemspis delicata Massuti and Vitorino
(Coleoptera, Coccinellidae) (Almeida & Vitorino 1997). Even with parasite levels as high
as 50% the impact of the galls on leaves remains high. Consequently, Tectococcus is
the most highly recommended of the potential agents against strawberry guava because
of its impact on the host plant, its ease of handling and it multivoltine life cycle. Very
heavy infestations can kill trees or branch systems but the most common result of
infestation is a decrease in vigor, deformation of leaves that senesce prematurely and
lack of flowering of infected shoots.
Bud Gall - Dasineum aiuantea h e l o & Maia 1999 (Diptera, Cecidomviidae).Infestation of a bud results in the formation of small leafy rosettes up to 3 cm diameter
somewhat like a double flower due to precocious production and development of the
leaves that ultimately outpaces the ability of the meristem to produce embryonic tissue.
The rosettes are initially green but turn yellow as they age and finally turn a deep
magenta before drying out and turning brown. The insect attacks both the red and
yellow-fruited forms (Angelo 1997) throughout the host's range.
Galls are formed in both terminal and lateral buds of flushing shoots and
occasionally in flowers and more rarely fruits. The production of the gall terminates the
activity of the apical meristem of the affected shoot killing it. Heavily infested shoots do
not grow any further. Any future activity of that branch results from the growth of a lower
bud released from the apical dominance of the terminal parts. Leptacis sp.
(Hymenoptera: Platygasteridae)was found associated to D. gigantea but it effects was
not significant. This agent is very promising because it stunts the trees. It is relatively
easy to handle the only problem being coordinating the emergence of the insects from
galls with flushing shoots. It is essentially univoltine though a secondary flush of shoots
during the year may be infested. Shipment of the galls requires no special precautions.
The wide ecological range of the species makes it a generally useful agent. There is a
potential cultural use of the galls in Hawal'i where they could be substituted for green
roses frequently used in leis.
Seed Gall - Eurvtoma psidii Thuroczv & Wikler, in ed. (Hvmenoptera, Eurytomidae).Fruit containing seed galls have a lumpy, blotched appearance and are generally larger
than normal, smooth fruit. The seeds are cemented together in large masses or a few
smaller masses. The pulp of galled fruit is substantially reduced and almost dry.
Apparently normal seeds may be found in galls but they and the seed masses fail to
germinate (Wikler 1999). Galled fruit falls to the ground at the same time as normal fruit.
Galls can be found in the leaf litter under trees for at least two years after fruit drop.
The insect is found throughout the range of Psidium cattleianum. It does not
demonstrate any preference for fruit type or ecological situation. It was also found in
Psidium longipetiolatum.
There is one natural enemy, the parasitoid Torymus psidii Thuroczy and Wikler, in
ed. (Hymenoptera, Torymidae).
The prospects of this insect as a biological control agent are good. However, only
10-50% of fruit are attacked (Wikler 1999). The level of infestation is weakly correlated
with plant population size suggesting that the insects may have a much greater impact in
-thel~e-mo~&~i~~~~dsi~Mawaii~i~sect-that-reducesseecCviability~n\~ever,isa long-term control agent that may not be suitable for managers looking for more
immediate reductions in population levels of the weed. It should, however, have a
strong effect in controlling dissemination in areas where range expansion is still
occurring.
Difficulties should be anticipated coordinating the emergence of insects with floral
bud production in quarantine. he insect lays its-eggs in young buds and open flowers
up until pollination but not those where the floral organs are beginning to deteriorate.
The insects, however, emerge from infested fruit for several months after the first flush.
There is a potential conflict of interest with horticulturalists who grow the plant for its
fruit. Galled fruit look distasteful and the rough texture of the seed masses is somewhat
unpalatable. This markedly reduced pulp may make the fruit less suitable as a host for
fruit flies, a potential benefit.
Stem Gall - EuNtoma caffleianii Thuroczy & Wikler in ed. and EuMoma desantisi
Thuroczv & Wikler in ed. (Hymenoptera, Eurvtomidae).-The insect attacks emerging
shoots producing a gall at the base of the shoot. The galls are predominantly lightly
dilated to round, 2-3 times the diameter of the stem. It is the same color as the stem,
initially green but slowly turning brown with age. Leaf development is normal but no
flower buds are formed. At the end of the growing season the shoot distal to the gall
dies terminating growth of that branch. Heavily infested plants, therefore, are somewhat
stunted when compared with adjacent uninfected plants.
The species is confined to the First Plateau of Parana State (800-1100m). It is
thought that E. desantisi is parasitic on E. cattleianii, the exact relationship is under
investigation. After emergence, some generalist insects and birds attack the gall-former.
The gall-former is species specific attacking both the yellow and red-fruited forms
but shows a marked preference for the red-fruited form.
This species is the lowest ranked of the recommended four potential biological
control agent. It has not been managed through its life cycle under controlled conditions.
Coordinating the availability of insects with plants with flushing shoots at the proper
stage of development could prove difficult. Its restricted occurrence to the higher
elevations of the distribution of P. caffleianum means that it has limited potential as a
universal agent against this plant. Hawaiian forest managers, however, may be
interested because the critical areas for conservation of Hawaiian native forests are all at
higher elevations where potential conflicts of interest with fruit fanciers and horticulturists
are minimal. This is also the area where the weed is still spreading.
Lamprosoma azureum Germar. 1824 (Coleoptera, Chrvsomelidae).-Both larvae and
adults damage the plants feeding on the young, unsclerified bark of the shoots
frequently girdling it (Caxambu 1998). Even if the girdling is incomplete the shoot is
severely stunted and very susceptible to attack by pathogens. The beetles attack young
trees only and have never been observed on plants above 1.8 m tall. The highest
number of insects recorded on one plant was 8 where they caused extensive damage.
Unfortunately, this species is not specific to strawberry guava. It has been found
feeding on other species of Myrtaceae, e.g., Eugenia uniflora L., Campomanesia
xanthocarpa Berg, Psidium guajava L., P. spathulaturn Mattos and Acca sellowiana
Berg. The scatoshell were also found on two species of Melastomataceae: Tibouchina
sellowiana (Cham.) Cogniaux and T. urvilleana (DC.) Cogniaux.
Neotrioza tavaresi Crawford. 1925 (Hemi~tera.Psyllidae).--The insect produces large,
round, green galls on leaves and is found throughout the range of P. caffleianum. The
gall is round up to 5 mm average diam. The maximum number of galls observed was
70lleaf but most leaves have fewer than 20. Psyllids emerge during October-November
(Spring). They copulate 5-10 minutes after emergence. The adults live around 5 days
and feed on sap in leaves. Oviposition occurs mainly in the leaf margin where the eggs
are attached by a pedicel. The nymphs hatch and after short dispersal they attach
themselves to the adaxial surface and start feeding (Butignol, pers. comm.).
Wasps, ants, flies, spiders and birds attack the adults. The nymphs are protected
inside the galls but can be attacked by parasitoids also, as yet unidentified.
This species, though apparently confined to P. caffleianum, will not be an effective
biological control agent. The damage that it causes does not result in premature leaf
drop or any reduction in flowering or growth of the plants except in extreme infestations.
Haploste~use~imelasKonow. 1901 (Hvmenootera. Peraid&.-This
sawfly can cause
extensive damage to young shoots and mature leaves of P. caffleianum. Eggs are laid
subepidermally along one side of a young shoot slowing growth of that side
considerably. Damage and fungal growth can kill twigs. Young nymphs feed on the
undersurface of leaves that in conjunction with oviposition kills a larger percentage of
shoots. Later instars consume large quantities of mature leaves defoliating shoots
(Pedrosa-Macedo, 1998). Sawflies can be found throughout the range of P. cattleianum.
An unknown mite damages the eggs. The potential of this sawfly as a biological
control agent is poor. Previous reports indicated that the sawfly also attacks P. guajava,
a result confirmed as a rare event in this study.
DISCUSSION
Exploration.
Strawberry guava is not a very common plant in Brazil. It is normally a minor subcanopy
species except in the restinga. Its physical similarity to a number of related species
caused considerable confusion and it was not until the second year of exploration that
sizeable populations of the plant were found. Even then the red-fruited form was
discovered only in the third year of the study. Though found in disturbed areas and
abandoned fields these areas were poor study sites because the trees were cut back
occasionally. Plants in disturbed areas were generally infested with only one gallforming insect, if at all.
The requirement that potential biological control agents not attack the congeneric
common guava was a constraint that was counterproductive initially because the two
species are sympatric only as weeds in abandoned fields and disturbed areas. Though
the focus was always on strawberry guava considerable time was wasted looking for
common guava In gardens and small farms. Consequently, the full complement of galls
was not discovered until the third year of the project and it was only then that the
potential of the project was clearly understood.
The ability to support year-round studies in native habitat was instrumental to the
success of the project. As insects were discovered populations were established on
plants in a small orchard on the university grounds where they were able to attack
several closely related species. The fact that there were insects that did not attack a
closely related species suggested that the host-specificity of these gall-forming insects
was highly discriminating.
Two potential agents, Lamprosoma azureum and Haplostegus epimelas, that did
not form galls were studied in great detail because of their significant impact. It was
discovered later that under certain conditions they would attack common guava so they
ar~otl.eeommended-as-pobnfial-age&.
Cost.
The total cost of the Brazilian studies was $500,000.00 ($300,000 from US sources and
$200,000.00 from CNPq and CAPES (Brazilian equivalent of NSF) who provided 2-year
M. Sc. Scholarships for 5 students and one for Ph. D. including 1 year at IIBC-UK.
Being able to use students to study the life histories of gall-forming insects as graduate
student thesis topics was very important. Not only were the costs low but students were
also abte to study the insects over several years. For gall-forming insects this is the best
way to approach the problem. During the first year the study established the possible
life cycle that is confirmed or corrected during more focused studies in the second year.
During the second year it was also possible to try to manipulate the insect in the
laboratory or laboratory garden. These latter studies are an important element in the
development of the management techniques that are necessary for later manipulation of
the insects in quarantine.
Host ranges.
Once the habits of the gall-forming insects were understood and the potential of the
insect as a biological control agent established, associated plant species in the
immediate environment of populations were examined for galls. Searches for each
insect were conducted particularly when the galls were developing on the strawberry
guava plants. From this study, a checklist of species on which the galls were not found
was established. This list can be used as a broad host range screening. For strawberry
guava the list contains over 198 species in 59 families.
Insect biology.
The biology of each gall-forming insect was studied in the field and the university
gardens. Hyperparasites and predators were studied in slmllar detail. The potential
impact of natural control agents against candidate species was established which should
enable entomologists in Hawaii to estimate the potential of known Hawaiian parasitoids.
This information was important in establishing the priority of candidates for importation.
Release from parasitoids, etc., should enable a potential agent to be more effective.
Tectococcus ovatus, however, is parasitized by two species, congeners of which are
already present in Hawaii. Even with parasite levels as high as 50% in Brazil the impact
of the galls on leaves remains high. Consequently, T. ovatus is still the most highly
recommended of the potential agents.
Life cycle studies.
Host-range testing of the potential agents has been conducted with species that are
easily manipulated such as Tectococcus. However, there are major problems with
several other gall-forming species that require narrow windows of the plant phenophase
in order to infect the plant. There are major problems coordinating the emergence of the
insects in the laboratory with the necessary phenophase of potential hosts. It is
extremely difficult to expose even strawberry guava to the gall-forming insects at the
correct stage in a laboratory or greenhouse whereas garden plants are much easier to
handle. This problem suggests that manipulating some of the potential agents in
quarantine is going to be a severe challenge.
Quarantine priorities.
Our approach to establishing the priority of each insect was to consider the following
criteria:
2actthegeate&eeim~a~tof
the insect on some aspect of the plant's biologythe higher the recommendation.
Handling - The greater the ease of experimental manipulation of the insect the
higher the recommendation.
Parasitism and predation - The greater the number of parasitoids and predators the
lower the recommendation.
The priority is:
Tectococcus ovatus (leaf gall). Already in quarantine in Hawaii and has passed
all host range challenge tests to date;
Dasyneura gigantea (bud gall). Plan to attempt to establish this species in
quarantine in 2001;
Eurytoma sp. (seed gall). One attempt made to introduce it into quarantine failed;
and,
Eurytoma sp. (stem gall). No quarantine studies anticipated at present.
Comprehensive evaluation of biological control potential.
We were not able to evaluate the impact of these potential agents on the target plant
quantitatively. Vandalism was a continual problem during the study even within the
school compound. The cost of increasing the security of the area was too high.
Qualitative assessments in the field and of garden plants were relatively easy for
defoliators. Tectococcus ovatus infestation results in premature leaf drop as the
population increases on the plant. Major branches and even whole trees can be
defoliated. Neotrioza tavaresi infestation, however, does not result in premature leaf
drop and is, therefore, not recommended for further consideration. The impact of
species affecting the shoots was entirely subjective. We were unable to manipulate the
insects sufficiently to conduct controlled studies of impact. The stunting of growth by
these insects was pronounced enough to suggest that the competitivity of infect plants
would be reduced considerably. No attempt was made to assess the impact of the seed
gall.
Insects have been the focus of this biological control study. The ident~ficationof five
gall-forming insects all of which displayed considerable host specificity suggested that
there was adequate potential for the development of a successful biological control
program without further studies. Pathogens were evaluated early in this project but were
excluded when it became apparent that most were not specific. The one species, a leaf
spot fungus, that might have been considered was never found fertile and was only
found in one population for a brief period. We did encourage plant pathologists to
continue searching for potential agents but subsequent studies by Barreto have proven
negative also.
The prospects for a successful biological control program against Psidium
cattleianum are considered to be very good. The four gall-forming insects
recommended attack different aspects of the biology of the plant offering a multi-faceted
approach to plant control. The insects are relatively free from attack by hyperparasites
and predators though novel attacks in other countries should be anticipated. The host
range specificity of the insects is very narrow; they do not attack the commercially
important P. guajava. It appears that this weed, so widespread particularly on tropical
and subtropical islands, can be controlled by the introduction of a few insects.
ACKNOWLEDGMENTS
The authors gratefully acknowledge Dr. Simon Elliot and MSc. Milton Mendon~aJr.
+or their-helpfutcomments-on-the drafi-manuseript;-Sincere-thanks-aredueto-tke-IateDr. Luis de Santis and Dr. Csaba Thuroczy for their assistance in the attempts to identify
the species. We thank our colleagues Dr. J. H. Pedrosa-Macedo, A. Angelo, C.
Butignol, M. Caxambu. Dr. M. Vitorino and Dr. L. Sousa from the Biological Control
Laboratory of the Forest Sciences Department of UFPR for access to their information
and the staff of the Parana Forest Research Foundation - FUPEF who provided
substantial administrative assistance. We are very grateful for the funding from U.S.
National Park Service (CA8034-2-9004) via its Cooperative Studies Unit at the University
of Hawaii and U S . Geological Survey, as well as that from CNPq and CAPES.
LITERATURE CITED
Almeida, L. M., and Vitorino, M. D. 1997. A new species of Hypemspis Redtenbacher
(Coleoptera: Coccinellidae) and notes about the life habits. The Coleopterists
Bulletin. 51: 213-216.
Angelo, A. C. 1997. A galha dos botdes do ara~azeiro- Psidium cattleianum
SABINE, 1821 (Myrtaceae), e insetos associados. Curso de Pos-Gradua@o
em Ciencias Biologicas. Universidade Federal do Parana. MSc. Dissertation.
95 p.
Caxambu, M. G. 1998. Morfologia e aspectos bioecologicos de Lamprosoma azureum
Germar, 1824 (Chrysomelidae, Larnprosomatinae)associado a Psidium cattleianum
Sabine, 1821 (Myrtaceae). Pos-Gradua@o em Ciencias Biologicas. Universidade
Federal do Parand. MSc. Dlssertatlon. 65 p.
Hodges, C. S. 1988. Preliminary Exploration for Potential Biological Control Agents for
Psidiurn cattleianum. Technical Report 66. Cooperative National Park Resources
Studies Unit, University of Hawaii, Honolulu.
Pedrosa-Macedo, J. H. 2000. Biology and behavior of the strawbeny guava sawfly,
Haplosteaus epimelas Konow 1901 (Hymenoptera: Pergidae), in the Southern
Brazil. Proceedings, Entomological Society Washington. 102: 129-134.
Smith, C. W. 1985. Impact of alien plants on Hawaii's native biota. pp. 180-250, In:
Hawaii's Terrestfial Ecosystems: Preservation and management. Stone, C. P. and
J. M. Scott (ed). Cooperative National Park Resources Studies Unit, University of
Hawaii, Honolulu.
Vitorino, M. D. 1995. Aspectos Biologicos e de Especificidade de Tectococcus ovatus
Hempel, 1900 (Homoptera: Eriococcidae) para o Controle Biologico do Amcazeiro Psidium cattleianum Sabine 1821. (Myrtaceae). Curso de Pos-Gradua~goem
Ciencias Biologicas. Universidade Federal do Parana. MSc dissertation. 55 p.
Wagner, W. L., D. R. Herbst, 8 S. H. Sohmer. 1990. Manual of the Flowering Plants of
Hawai'i. University of Hawaii and Bishop Museum Presses, Honolulu.
Wikler, C. 1999. Distribui~iiogeoghfica mundial de Psidium cattleianum Sabine
( M y r t a c e a e ! e um ceciddpeno com possibilidades de utiliza~ioem contmie
bioldgico. Ph.D dissertation, School of Forestry, Universiaade Federal do Parana.
135 p.
BIOLOGICAL CONTROL OF INVASIVE PLANTS IN
NATIVE HAWAIIAN ECOSYSTEMS
SYNTHESIS AND CONCLUSIONS
Biological control can be an effective approach to control widespread ecologically and
economically harmful species, but it is only one tool of many that should be employed to
mitigate the impacts of invasive weeds on natural and managed ecosystems. For
example, the magnitude of personnel and financial resources invested in weed
management could be substantially reduced if there were adequate programs to prevent
inapprapriate introductions. Similarly, early action to eradicate newly-established,
potentially invasive species would also reduce the need for expensive control programs.
To date, effective preemptive action, such as screening species proposed for
introduction, has not been supported in Hawai'i, but the political will for their
establishment is building. The development of an integrated state-wide strategy to
reduce new introductions and control existing problem species was highly recommended
by Forum participants, land managers, conservationists, and the research community.
In reality, the magnitude of the invasive species problem in Hawai'i is so severe, that
concerted, focused, state-wide measures are needed to protect its natural heritage.
Biological control of invasive wildland weeds is an important component of any such
strategy. In a series of working groups and plenary discussions, forum participants
discussed ways in which the development and use of biological control agents for
wildland weeds could be made more efficient, more effective and safer. Here we
summarize those remarks.
HISTORICAL PERSPECTIVE OF BIOCONTROL IN HAWAI'I.
Several speakers critically examined past attempts at biological control of weeds in the
United States. They found several examples of success (Pemberton this volume,
Delfosse this volume) as well as reason for caution (Myers this volume). Unforseen,
non-target impacts among biocontrol agents released to control weeds were scarce
among Post World War II releases. Herbivorous insects are relatively host specific in
comparison to predators and parasitoids released to control insect pests. Hawai'i has a
relatively good track record in this regard. Most examples of non-target impacts by
biological control agents have been among predatory insects and vertebrates and most
- o c c t t r r ~ f o r e W o r I d - M I-wbmpre-releaseremrekon-thep&ntMtagents-wss-l
less thorough and when the health of native species and ecosystems less of a concern.
Nevertheless, examples of unanticipated host range extensions by some agents raise
justifiable concerns about the fates of biocontrol agents once they have been released.
While successes in one part of the world may suggest promising agents for another,
potential local hosts should always be tested for new introductions. Moreover agents
once released are not limited by political boundaries and may be spread unintentionally
across water, inhospitable environments, and regional boundaries.
Unfortunately, the success rate is also low. Probable causes for low success rates
include inadequate release procedures, selection of ineffective agents, and limited
growth or establishment of the agent population due to local predator, pathogen, and
parasitoid loads. Research to improve effectiveness of biocontrol agents should include
a better understanding of factors contributing to their population control in the country of
origin, controlled release studies, and better post-release monitoring of the fates,
impacts, and effectiveness of newly released agents.
No introduction of insects or pathogens is done without risk. However, in
comparison to other alternatives the risks are small. lnvasive weeds can alter native
communities and ecosystems within a very few years threatening all species of fauna
and flora dependent on them as well as altering their utility for human inhabitants.
Outside of high priority management areas such as parks and nature reserves where
chemical and mechanical control is feasible on limited target areas, practical tools to
protect the integrity of wildlands are scarce and expensive. Nevertheless, the recent
requirement for ecological impact analyses prior to release is an important step in further
decreasing unanticipated negative impacts or ineffective agents.
THE ROLE OF BIOLOGICAL CONTROL IN NATURAL AREA MANAGEMENT
Natural area managers are reluctant to engage in biological control for both practical and
ethical reasons. Managers need rapid solutions to weed management; the lengthy delay
required to ensure the safety of the potential control agents and their relatively low
success rate make biocontrol a risky strategy. There is considerable reluctance on the
part of some managers to add further alien species to the Hawaii's biota even when the
risk to native species is small. Future problems well may arise through synergies with
other species There is also a suspicion that introduced agents might adapt to other nontarget species in time. Some of this apprehension is the result of highly publicized
mistakes in the past where misguided or political decisions allowed the release of agents
that themselves are now significant problems for native species.
Most managers consider biological control as the last resort after all other efforts at
control have failed. Few programs systematically target those species that are the most
disruptive to ecosystems for biological control. Rather efforts are directed at species
that are of immediate concern in a specific area. This attitude is not surprising. It is the
result of management mandates, financial constraints and program needs. In part it
derives from the agricultural origin of biological control where economic impact was the
determining factor in establishing control priorities. It is also important that potential
impact be considered when establishing priorities for weed control in natural
ecosystems, but in this case it is the ecological rather than the economic impact that
should be of foremost concern. To date It has proved difficult to reach consensus about
control priorities. In Hawaii, priority target weeds are likely to be different for each island.
The decision to target strawberry guava was based on the experience and
recommendations of national park managers and of scientists working with other land
managers. No practical manual or chemical control strategy was available. Even in this
case, however, only the National Park Service paid the bill. A consensus to target
Mconia celveacens took several years even thoqKth~3itaation
m T-atritiwaswolr
known. Scientists and managers on Maui took the initiative and effectively publicized
the threat in the local press. Oahu and Kauai joined in the effort somewhat later but
managers on Big Island did not become involved in a serious way for several years even
though Miconia populationswere large and readily visible. They believed that the
species could be eradicated by manual and chemical means. The difficulty in
maintaining the necessary funding and managing the program island-wide ultimately led
to their joining the search for a biological control alternative.
Discussants agreed that biocontrol had a place in the management of natural
ecosystems in Hawaii. Done correctly it can offer managers effective, long-term, lowcost solutions to major weed control problems. Managers should understand that
biological control generally results in reduction of the target population, often by
restricting its range or its density; eradication is rarely achieved. There is a need for
more effective communication between biocontrol scientists and managers to develop
priorities for biocontrol research and to plan for the restoration of ecosystems following
successful weed reduction. Research opportunities in studies of post-release behavior of
populations are considerable. They include their epidemiology, factors contributing to
efficacy of biocontrol, and the nature of population interactions with pests, pathogens,
and non-target species.
EDUCATION AND PUBLIC INFORMATION
Perhaps the biggest weakness in Hawaiian biological control programs has been the
lack of documentation of all stages of development of biocontrol agents. Priority has
been placed on rapid evaluation and release of agents to the exculsion of post-release
follow-up and publication of results. Failure to document results has resulted in a poor
understanding of the biological control efforts in the state to the detriment of the
program. Lack of information undermines public support in the political process of
securing funding and administrative support. Misinformation abounds and the absence
of the discipline of peer review can contribute both to the perpetuation of Inappropriate
methods and lack of public confidence in the results. Focus on successful establishment
of an agent rather than on its impact on the target organism has meant that many
significant issues, including effective release strategies, a better understanding of the
risk to non-target species, and interactions with local pests and pathogens have gone
unaddressed. In the absence of publication, valuable information effectively is lost in
files or in memories as researchers relocate or retire. Critical analysis and thinking
required in the process of publication improves the quality of the research process and
forges interactions with other scientists that will enhance the effectiveness of Hawaiian
scientists working in isolation.
Hawaii's biological control program has lost its deserved international reputation in
the absence of publication. A reputation for careful, high-quality science in the
development of biological control agents will also foster greater public understanding
and acceptance of biological control as a safe, effective approach to the management of
invasive weeds, bringing with it better support in the form of needed facilities and
funding. There was consensus among forum participants that a greater effort to inform
the public about the objectives and methods of biological control as well as about its
successes and failures likely would be met by better public support and understanding.
POST-RELEASE MONITORING
Many-paaidpants-emphasized_theirnportance~ff
allo~~upinformafiD~~onihe
establishment, effectiveness, and impacts of agents released into the environment.
Perhaps one of the most important oversights of biological control research in Hawai'i
has been its failure to carry out post release monitoring of new releases. A large
number of species have been released but failed to establish. Yet we have no
understanding of factors contributing to this failure. Were release techniques
inadequate? Were agents released into inappropriate habitat? Were local predator or
pest loads too high? Failure to follow through with post-release evaluations risks loss of
potentially successful agents as well as the time and effort spent in exploration and
evaluation of the agent leading up to the release.
Many successes are claimed but not substantiated in any way. As noted earlier,
success frequently is defined as establishment of the agent, not in effectiveness in
limiting the population growth, abundance or distribution of the target organism.
Unpublished anecdotal information exists in a few instances, but that information as well
will disappear as researchers retire.
Delfosse (pers. comm.) noted that federal agencies involved in biological control
now require substantial post-release monitoring studies. It is important that Hawaiian
biological control programs have the same requirement. Without such studies, we are
unable to evaluate factors associated with the release of a potential control agent which
may contribute to population declines, including climatic variation, competition, or
existing pest or pathogen pressures. Coincidence of population decline following the
release of a biological control agent is not always strong evidence of effectiveness. For
example, substantial disagreement still surrounds the success of two pathogens
released against Hamakua pamakani (Agemtina riparia Asterales, Asteraceae) and
banana poka (Passiflorn mollissima (Kunth.) L. H. Bailey - Violales, Passifloraceae). It
is impossible to distinguish among alternative hypotheses about causes for their decline
because of a lack of scientific data. The disagreement in turn undermines faith of
managers in the effectiveness of biological control as a weed management strategy,
reducing support for further development.
Forum participants were unanimous in their support for the publicationof information
on insects and pathogens using the target organism In Hawai'l as well as in its original
and other parts of their naturalized ranges; of life history studies of all potential agents
and especially for species actually released; and of the post-release status of the agent
an4 its impact on target and non-target species.
-
HOST RANGE EVALUATION
Several contributors recommended revisiting the commonly used zero-tolerance release
criterion that a potential control agent shows no propensity for feeding on native species.
If researchers can demonstrate that impacts on native species are likely to be minor,
then low levels of feeding on non-target species might be tolerated for promising control
agents. To the extent possible, research objectives should focus on the potential
population level impacts of the control agent on both target and non-target species. We
should seek to identify agents that have a high likelihood of strongly affecting population
processes of target species while inflicting minor effects at most on non-target species.
This is a formidable challenge because the behavior of a potential agent in a new
environment may be predictable only weakly by its behavior in quarantine or in natural
communities in countries of origin. A better understanding of post-release
consequences of releases should improve our understanding of these processes.
Research is needed also on the magnitude of biotic intelference in Hawaii. For
example, several participants suggested that lepidopteran leaf-feeders may not be
effectivebiocontrda~entsiatlasaraUbe~~u~~~theyaresovu~able0
introduced
parasitoids. Others suggested that diseases were responsible. However, we lack data
bearing on these hypotheses or on the impacts of predators and other negative
influences.
THE ROLE OF WORK IN COUNTRIES OF ORIGIN
Early in the 20Ihcentury, evaluation of biological control agents was necessarily
conducted in the country where they were to be used because there were few
institutions where such studies could be conducted in the native environments of the
target weed. More stringent host-specificity requirements, expansion of target
organisms to include weeds of consenration areas, financial and political constraints on
agencies, quarantine facility bottlenecks, and demands for more demonstrable impacts
by potential agents on target species have mandated a better understanding of the
biology of potential control agents while at the same time imposing stringent cost
controls. At the same time institutions in countries of origin are increasingly well
qualified to conduct the necessary research. Consequently, ever more investigative
work is being done in the natural range of the host.
There are four distinct advantages to this approach. First, impact analyses of
potential agents can be conducted in a natural environment not, as in most previous
studies, within the confines of quarantine. The most deleterious agents are identified,
eliminating unnecessary research on species that will not be useful. Second, host
specificity can be evaluated in the wild first by examining the activity of the potential
agent on all associated species (a very broad analysis of host specificity), then by
experimental tests in the field against common agricultural/horticultural species. The
danger of false positive tests is minimized. Host specificity testing in Hawaiian
quarantine facilities may then be restricted to high priority species designated by
oversight agencies. Third, detailed studies of the life-history and rearing techniques can
provide a more complete ecological understanding of the potential agents. Manipulation
of the species is quarantine is facilitated and post-release monitoring programs better
organized because the biology of the agents is understood. Fourth, development of
scientists and institutions in the country of origin is supported through funding,
professional collaborations, student training, publications, museum collections, and the
like, further strengthening the capacity for research.
WHICH SPECIES SHOULD BE TARGETS FOR BIOLOGICAL CONTROL?
-- -
-
Pemberton (this volume) recommends that weeds with native congeners should be
a low priority as targets of biological control. This recommendation is of particular
importance in Hawai'i where almost all of our endemic flora has evolved in the absence
of the normal array of predators. Unfortunately, some particularly problematic weeds of
Hawaiian natural areas are in this category. For example, alien species in the genus
Rubus, (R. argutus Link, R. ellipticus Sm., and R. niveus Thunb - Rosales, Rosaceae),
are spreading so rapidly that they will be controlled only if suitable biological control
agents can be found. Hawai'i has two native species of Rubus (R. hawaiiensis Gray and
R. macmei Gray) . Pemberton's caution is particularly applicable in Hawai'i because
insular endemic species often lose many of their defenses against consumers. Data
from host range expansions elsewhere also indicate that allopatric congeners can be
susceptible to host-shifting insects when the previously allopatric congeners come to
occupy the same habitat.
Pemberton's recommendation should be extended to confamilial species during
--host-range evalu~onsi~Hawai'i,Hight-(this-volume),f~r-examplef-hasshown that-the
sawfly of the Brazilian pepper (Schinus terebinthifolius Raddi - Sapindales,
Anacardiaceae) attacks the native Hawaiian Rhus sandwicensis Gray -Saphales,
Anacardiaceae), but not Rhus species native to Florida. The sawfly thus may be an
acceptable agent in Florida but not in Hawai'i. In contrast, Wikler and Smith (this
volume) note that a 10-year study in Brazil was able to identify five gall-forming insects
that attack Psidium cattleianum Sabine (Myrtales, Myrtaceae) but not P. guajava L. In
this case, appropriate and highly selective agents may be identified to target one of a
pair of congeners. Development of biological control agents today demands challenging
the insects with native species related at the familial level, a practice that has led to the
exclusion of a number of potential agents. However, several Hawaiian conservationists
take exception to such a cautious approach noting that where the impact of alien species
is so substantial that threatened and endangered species face extinction we may
consider accepting higher levels of non-target impacts. Although this concern is
particularly acute in Hawai'i because of its high numbers of threatened and endangered
species, the interaction of biological control and endangered species will become a point
of conflict between the law and the needs of managers and consetvationists.
Disagreements among the various stakeholders in Hawaiian biological control are
not new. Howarth has raised concern about techniques to develop biological control
agents targeting insect pests. In contrast, the agricultural industry and political urgency
create substantial pressure for rapid evaluation of potential agents. Resolution of these
conflicts is not easy. Though much of the information is anecdotal, the Leucaena
leucocephala (Lam.) de Wit - Fabales, FabaceadHetempsylla cubana Crawford(Homoptera, Thripidae) story is illustrative of these disagreements. The arrival of the
psyllid where Leucaena provided fodder in environmentally marglnal areas for ranching
was met with considerable concern by the cattle industry. In contrast conservationists
initially opposed the HDOA search for biological control agents against the psyllid
because Leucaena had been a serious weed of lowland xeric and mesic habitats. An
insect that partially controlled this weed therefore was welcomed by one segment of the
community. However, as Leucaena stands opened up, broomsedge (Andmpogon
virginicus L. Cyperales, Poaceae) invaded the undarstov, providing a continuous fuel
for wildfires. As fires in mesic lowland areas became hotter and more extensive than in
the past, pockets of remnant natural vegetation were threatened. Opposition to
development of the psyllid agent weakened among the conservation community and a
coccinellid eventually was released.
Biological control projects profit from community oversight, preferably from projectspecific "Action Committees". Such committees, representing a wide array of people
and organizations affected by the weed, provide guidance to the project and help to
raise financial support, where they may compete with island lnvasive Species
Committees. Statewide coordination and facilitation of biological control programs would
be considerably strengthened by appropriate leadership.
Biological control remains the strategy of last resort in resolving previously
intractable weed management problems in agriculture, forestry and natural areas in the
Islands. Development of effective and safe agents demands a long-term commitment to
the development, evaluation, and post-release monitoring of each agent on the part of
the conservation, scientific, management and political communities. Prevention, early
eradication and limitation of spread are more cost effective strategies for the control of
potentially invasive species. Until such are implemented, however, the need for new
biological agents will continue to grow and to challenge the skills, patience and wisdom
of all concerned with the health of natural ecosystems.
-
J. Denslow
C. Smlth
S. Hight