Managing Disease Disease Has Become a Classic Example of the Impact of Introduced Diseases on Naive DENNIS A

CHAPTER SEVENTEEN The extinction and decline of Hawaiian avifauna due to introduced mosquito-borne Managing Disease disease has become a classic example of the impact of introduced diseases on naive DENNIS A. LAPOINTE, wildlife populations (Warner 1968; this CARTER T. ATKINSON, volume, Chc'lPter9). Along with rinderpest AND SUSAN 1. JARVI virus and rabies virus, avian malaria (Plas- modiumrelictum)and avian pox virus (Avipox- virus sp.) stood for decades as rare cases of invasive pathogens in wildlife species (Guliand 1995). Today,however, with ever- increasing globalization, emergent and invasive human and wildlife diseases are on the rise (Daszak et al. 2000, Gubler 2001, Friend et al. 2004). Once an academic curiosity to conservation biologists outside of Hawaii, avian malaria and pox and their impact on Hawaii's native forest birds now have continental relevance with West Nile virus's sweep across North Amer- ica and its potential threat to endangered species populations (Marra et al. 2004, Kilpatrick et al. 2007). The Hawaiian Islands, like other isolated oceanic islands, experienced limited natural colonization by terrestrial biota. These founders brought few, if any, parasites and pathogens with them (Torchin et al. 2003; this volume, Chapter 1). For parasites with complex life cycles, alter- nate hosts were absent, as were the insect vectors of pathogens, such as mosquitoes, black flies, biting midges, and others. This all changed with the arrival of Westerners to the Hawaiian Islands in 1778 and the subsequent introduction of mosquitoes and mosquito-borne avian disease. Intro- duced mosquito-borne avian disease is considered a major factor limiting Hawaiian forest bird populations and an obstacle to the restoration of the islands' avifauna (U.S. Fish and Wildlife Service 2006). Future efforts to protect remaining Hawaiian forest bird species and to restore their populations

405 406 APPLYING RESEARCH TO MANAGEMENT will rely on the development of disease native forest bird habitats, and only C.quin- management strategies that can be applied quefasciatusis presently common above 900 at the landscape level. The long history doc- m in elevation (Goff and van Riper 1980, umenting the control of vector-borne hu- LaPointe 2000). Culex quinquefasciatusis a man disease (Harrison 1978) suggests that known vector of avian malaria in Hawai'i there will be no simple solutions and that (LaPointe et al. 2005) and is the most likely an integrative and diligent approach will vector of avian pox virus (van Riper et al. be necessary (Rose 200 1) . 2002). Only a few individuals of Aalbopic- In this chapter we (1) provide a brief tus and W mitchellii were found to support overview of the mosquitoes in Ha«rai'i sporogany of Plasmodiumrelictum in the lab- and the pathogens they vector to birds (see oratory (LaPointe et al. 2005). Preliminary Chapter 9 for a more detailed account of trials with the newly established A japonicus disease biology); (2) outline possible man- suggest that this species does not support agement practices and evaluate them in the sporogony in the laboratory (LaPointe, context of endemic Hawaiian bird conser- unpubl. data). Avian pox virus does not vation, from single-species captive popu- require a specific vector and may be trans- lations to intact communities across broad mitted by any mosquito or biting arthro- landscapes; and (3), in synthesis, suggest pod. Although these three mosquito species management strategies to minimize the im- may be more or less opportunistic in host pact of vector-borne disease in Hawaiian selection (Tempelis et al. 1970, Edman forest birds. and Haeger 1977, Tanaka et al. 1979) and abundant in some lowland Hawaiian forests, their marginal susceptibility to avian KNOW THINE ENEMY malaria and limited altitudinal distribution make them unlikely to be important Mosquito Vectors of Avian vectors of either avian malaria or pox. Their Pathogens in Hawai'i possible role as minor, incidental vectors Six biting species of mosquito have become of avian malaria or pox is still unknown. established in the Hawaiian Islands since Thus, due to its high level of vector com- the nineteenth century: Culex quinquefasciatus petence and altitudinal distribution, C.quin- (established by ca.18 2 6), Aedes aegypti (ca. quefasciatuscan alone account for the cur- 1892), Aedes albopictus(ca. 1902), Aedes vex- rent distribution and prevalence of avian ans nocturnus (ca. 196 2), Wyeomyia mitchellii malaria and pox in forest bird communities. (ca. 1981), and Aedes japonicus (ca. 2003) (Hardy 1960, Joyce and Nakagawa 1963, Hawaiian Landscapes, Feral Pigs, Shroyer 1981, Larish and Savage 2005) (Fig. and Mosquito Abundance 17. 1). The first three species are known vectors of human pathogens and have been Larval mosquitoes are aquatic and can be widely distributed throughout the tropic found in a wide range of temporary and and subtropic regions by Western com- permanent waters. C. quinquefasciatuslarvae merce (LaPointe 2007). By contrast, A v. occur in natural and artificial containers, nocturnus and A japonicusare competent lab- ditches, puddles, irrigation channels, cess- oratory hosts of encephalitis viruses but pools, and the margins of ponds and are not documented vectors of human or flowing streams. This species is adapted wildlife pathogens (Turell et al. 2001, to eutrophic waters heavily enriched Sardelis et al. 2003). W mitchellii is not with organic matter. Although generally known to transmit any vertebrate patho- not found in forest ground pools or open gens (Shroyer 198 1) . bogs in Hawai'i (Fig. 17.2), C. quinquefas- Only C.quinquefasciatus,A. albopictus,A[cpon- ciatus larvae have been recovered from icus, and W mitchellii have been found in ground pools and wallows where fecal MANAGING DISEASE 407

Figure17.1. Common mosquito species found in forest ests above 1,500 m in elevation on Hawai'i bird habitats on Hawai'i Island. Left side, top to bottom: Island (Goff and van Riper 1980, LaPointe Egg rafts, larvae, and adult female Culex quinquefasciatus,the 2000). Intermittent and ephemeral streams, main vector of avian disease in the Hawaiian Islands. Right side, top to bottom: Aedes albopictus,Aedes japonicus, however, may be important larval mosquito and Wyeomyia mitchellii.Source:Photos by Dennis LaPointe, habitats in some Hawaiian landscapes. U.S. Geological Survey. Surveys in Kipahulu Valley on Maui documented C. quinquefnsciutus larvae in rock matter from livestock or feral ungulates pools of intermittent streambeds (Aruch may have enhanced the microhabitat (D. et al. 2007). LaPointe, pers. 0bs.) . In stark contrast, the younger volcanic Mosquitoes do not typically occur in landscapes of the east flank of Mauna many natural areas in the Hawaiian Islands Loa, Hawai'i Island, are all but devoid of because of temperature constraints on their permanent surface water. C. quinquefasciatus, development or the absence of suitable however, are abundant in these wet forests, larval habitat. Adult and larval mosquitoes where their larvae rely primarily on rain- were rarely encountered in windward for- water-filled cavities in the native tree fern, 408 APPLYING RESEARCH TO MANAGEMENT

large tracts of forest undermines this nat-

60 ural protection (LaPointe 2000) (see Fig. Q) t1l 50 17.2). 2: ...Jt1l 40 Feral pigs, however, are not the only cul- £; prits. The numbers of C. quinquefasciatus are .3: 30 much greater in suburban and agricultural c 20 70 Q) areas than in natural areas. Conservation ~ 10 Q) 0.. areas in Hawai'i often abut residential and 0 Hapu'u Cavities Ground Pools agricultural communities that can produce high densities of mosquitoes through the Figure17.2. Larval Culex quinquefasciatusoccupancy of avail- creation of larval habitat and the presence able aquatic habitats in windward Mauna Loa forests, of abundant hosts for blood meals (Mian Hawai"i Island. The numbers over the columns rep- et al. 1990, Reisen et al. 199 0, Reisen et al. resent the total number of individual habitats of the habitat type sampled for mosquito larvae that were en- 1992). The rural community of Volcano countered in 4.5 ha of forest. Although prevalent in Village, located just outside the bound- these forests, ground pools do not appear to support aries of Hawai'I Volcanoes National Park, mosquito larvae (LaPointe 2000). serves as an excellent example. Mosquito capture rates in the village are nearly three times greater than capture rates within the hapu 'u (Cibotium spp.) (Fig. 17.3). These nearby forest (Reiter and LaPointe 2007). cavities are formed by the feral descen- Household water storage in residential dants of domestic pigs (Sus scrofa), which areas contributes to local mosquito abun- feed on the starchy core of the tree fern. dance, but the impact of artificial containers After-extracting the starch, a cup- or bowl- and impoundments may be several times shaped cavity remains that will collect rainwater and leaflitter, thereby providing Figure 17.3. Water-filled cavity in a hapu'u tree fern (Ci- a favorable habitat for larval mosquitoes. botiumglaucum) trunk created by feral pig feeding. Such Although the geological and hydrological cavities are the main breeding sites for mosquitoes in some native forests.Removal of feral pigs would reduce nature of the Mauna Loa landscape pre- the number of breeding sites, which in turn would cludes the production of mosquitoes, the potentially reduce disease transmission among birds. occurrence of hapu'u cavities throughout Source:Photo by Daniel Lease, U.S.Geological Survey. MANAGING DISEASE 409 greater on agricultural lands. Cattle opera- trol of infectious disease in humans and tions, in particular, create favorable habi- domestic animals are well documented. tats for C.quinquefascietus in the form of stock However, these approaches have not been ponds, troughs, cisterns, settlement ponds, as successful in wild animal populations. and the old tires and tarps commonly used Chemotherapy of wildlife populations is to cover feed (Reiter and LaPointe 2007). generally limited to very local situations, Because the native tree ohi'a-lehua (Met- and immunizations of wildlife are depen- rosideros polymorpha) still dominates many of dent on novel methods of vaccine delivery these residential and agricultural land- (Wobeser 2002). Most wildlife vaccines scapes, birds such as 'Apapane (Himatione are directed at mammalian reservoirs of s.sanguinea) and T'iwi (Vestiaria coccinea) move human disease (Slate et al. 2005), livestock readily between the forests and suburban disease (Wilkinson et al. 2004), or criti- or agricultural areas where the likelihood cally endangered populations (Hastings of exposure to malaria and pox may be et al. 1991). There are no examples of vac- greater. Mosquitoes from these areas may cine use in wild birds. also be dispersing into forests, thereby aug- menting forest populations. Female C. quin- ANTIMALARIAL AGENTS quefasciatus were recaptured at the 3 km trap boundary during mark-release-recapture Chemotherapy of birds has been used experiments in a closed-canopy Hawaiian only in captive or closely managed flocks. forest. After accounting for trap density, Common antimalarial agents used to treat 1.6 km would be a conservative estimate human malaria, such as chloroquine and of the average dispersal of C. quinquefasciatus primaquine, were first evaluated in birds in 10 days (LaPointe 2008). Mainland stud- and may work as a prophylaxis, as sup- ies in urban environments showed that portive treatment, or as a radical cure for C. quinquefasciatus can disperse up to 12.6 km malaria in captive birds (Hewitt 1940, (Reisen et al. 1992). The dispersal ability Stoskopf and Beier 1979). For example, of this species has serious implications for chloroquine and primaquine were used disease management in fragmented areas. to treat captive-reared, endangered 'Alala On Mauna Loa's east flank, forest (Hawaiian Crows, Corvus hawaiiensis) that be- mosquito populations disappear during came infected with malaria while accli- droughts but return after sufficient rain, mating in outdoor aviaries prior to release. suggesting that residential or lowland areas The crows recovered completely and were may serve as the source of these rebound- ultimately released (Massey et al. 1996). ing populations (LaPointe 2000). Low- The value of antimalarial agents in the elevation mosquito populations outside of treatment of wild birds is limited by the forest bird habitats might hypothetically difficulties of delivery. Dosing wild birds contribute to avian malaria and pox trans- via artificial nectar feeders is impractical mission if carried by wind to higher ele- and likely to fail at the landscape level be- vations (Scott et al. 1986). cause dosages cannot be regulated, and low doses could lead to rapid selection for drug- resistant parasites (White 2004). Similarly, POTENTIAL DISEASE deliberate exposure of birds to malaria MANAGEMENT STRATEGIES and chemotherapeutic control of the acute phases of infection might allow individ- Manipulation of Susceptible Hosts uals to mount an immune response that by Chemotherapy and Vaccination would ultimately be able to regulate and The successes in vaccine development and control chronic infections. However, this the use of antimicrobial agents in the con- approach will also ultimately select for drug 410 APPLYING RESEARCH TO MANAGEMENT

resistance in the parasite population. In 1940s and are commonly used in high- addition, chronic infections may relapse risk areas where birds and/or mosquitoes or recrudesce when birds are immuno- occur in high densities (Ritchie 1995). logically stressed (Valkiiinas 2005). The vaccine is typically delivered by injection into the wing web, but reimmuniza- tion is often necessary, limiting its useful- AVIAN MALARIA VACCINE ness in wild birds. Another major concern Vaccines against protozoan pathogens is recombination of the attenuated virus have been notoriously difficult to develop, in the vaccine with other viral strains or but they can avoid some of the problems pox viruses, resulting in more virulent associated with chemotherapy and drug strains. Vaccine contaminants are also of resistance (Jones and Hoffman 1994). At- concern. Reticuloendotheliosis virus causes tempts to develop subunit vaccines for immunosuppressive disease in domestic human malaria have been unsuccessful fowl and is a common contaminant of because of the complex life cycle of the commercial fowlpox vaccine (Garcia et al. parasite, coinfection of multiple strains, and 2003) . Finally, immune responses to avian the ability of Plasmodiumto change its anti- pox can be very host-species specific, and genic, surface proteins (Desowitz 2000). additional information on the diversity However, recently there has been renewed and pathogenicity of avian pox in Hawai'i interest in using irradiated sporozoites as is needed before an effective vaccine can a live, attenuated human vaccine. In labo- be developed (Tripathy et al. 2000; Jarvi, ratory trials with rodents and human sub- Triglia, et al. 2008). jects, this approach provided protective Even in the event of the development of immunity for at least 10 months (Nussen- a successful malaria or pox vaccine, diffi- zweig et al. 1967,Clyde et al. 1975, Hoff- culties in the method of delivery remain a man et al. 2002). Its potential for use in formidable obstacle to the immunization Hawaiian birds has yet to be evaluated. oflarge wild populations. But technolog- A technically more sophisticated ap- ical advances offer some hope. It is now proach is the use of DNA vaccines based on theoretically feasible to have a ubiquitous short sequences of specific genes (Doolan and potentially benign virus, such as fowl et al. 1996). Recently, a DNA vaccine using pox, vector recombinant DNA for the anti- the circumsporozoite protein (CSP) of P.re- genic proteins of any number of potential lictum afforded moderate but short-lived pathogens (Paoletti 1996). Because pox protection in canaries (McCutchan et al. viruses are easily transmitted by mosquito, 2004). In a similar CSP vaccine trial, the a nonpathogenic pox virus or viral strain prevalence of malaria in a captive flock of might serve as a vector for a Plasmodium or penguins was reduced from 50% to 17% arboviral gene. A fitting irony would be use following vaccination. The vaccination of of the mosquito as the ultimate delivery part of the captive flock also protected un- system of this engineered vaccine. vaccinated individuals by decreasing the proportion of susceptible individuals (Grim Naturally Malaria-Tolerant Birds et al. 2004). DNA vaccines have clear ben- efits for captive populations, but the short There is evidence that some lowland pop- duration of the immunity afforded limits ulations of Hawai'i 'Amakihi (Hemignathus their effectiveness in wild populations. virens)may be evolving an innate tolerance to avian malaria (Woodworth et al. 2005). Translocation of tolerant birds to new sites AVIAN pox VIRUS VACCINE may introduce this presumably heritable Attenuated live pox virus vaccines have trait to currently vulnerable populations. been available for domestic birds since the However, translocated birds may introduce MANAGING DISEASE 411 pathogenetically different strains of malaria of a highly competent vector, a favorable or new pathogens into the recipient pop- climate, and an ample reservoir of hosts ulation and should be properly screened make for efficient transmission even at low and/ or treated before release (Atkinson and densities of C.qumquefosctetus relative to main- van Riper 1991, Cunningham 1996). It has land areas. Third, native birds with chronic also been suggested that tolerant individ- malarial infections are reservoirs for life uals be brought into captivity as a breed- (Jarvi et al. 2002), and although there are ing population for propagation and future no estimates of species longevity, there release into the wild. are records of individually banded native Whether translocated or captive-bred passerines' living in excess of 10 years birds are used, the presumed heritable (Lindsey et al. 1998). Finally, the land need- traits of disease tolerance may not become ing control is enormous; approximately fixed in the wild populations if similar 25% of the state's land area (405,000 ha) selective pressures-especially perennial is under some form of natural resource malaria transmission-do not prevail at protection (Loope and ]uvik 1998). These the release site or if gene flow from distant limitations suggest that management of populations overcomes the introduced avian disease in Hawai'i will require con- resistant genes. The latter is most likely to trol methods that are exceptionally effec- occur with highly vagile species such as tive, cost-efficient, environmentally safe, 'I'Iwi and 'Apapane (Fancy and Ralph and indefinitely maintained. 1997). For these reasons, the translocation of disease-tolerant individuals may have its SOURCEREDUCTION OF LARVAL greatest value in facilitating the lowland MOSQUITO HABITAT expansion of naturally evolving, disease- tolerant populations. The most effective way to limit mosquito numbers is to remove larval mosquito habitats, a practice referred to as source Manipulation of the Vector reduction. From the time of the Roman Reduction of vector abundance and lon- Empire to recent times, great efforts were gevity remains the central control strat- made to drain swamps and lowlands where egy for mosquito-borne disease today. human malaria was common (Harrison Whether the pathogen is dependent on 1978). These efforts were successful, and the vector to complete transmission (as source reduction was practiced widely in the case of Plasmodium and arboviruses) throughout the world until the environ- or the vector merely facilitates mechanical mental movement of the 197 as raised the transmission (as with avian pox), the mos- alarm that wetlands were invaluable eco- quito is the crucial link between infected systems of great productivity and bio- and susceptible host. Remove the link, and diversity (Mitsch and Gosselink 1993). transmission ceases. The early successes in Although swamps and wetlands are no human malaria and yellow fever control longer drained in the name of mosquito were both brought about through the con- control, source reduction of artificial trol of the mosquito vector. habitats or altered natural habitats remains Many characteristics of the Hawaiian a valuable component of most mosquito avian disease system would seem to make abatement programs. control through vector reduction an un- Perhaps the greatest potential use of attainable goal (Chapter 9). First, avian source reduction in Hawai 'i lies in the malaria and avian pox are pervasive in the residential and agricultural communities Hawaiian Islands, and the prevailing cli- that encroach on natural areas. Catchment mate favors transmission throughout most systems for household water are com- of the year. Second, the combined effects mon in rural Hawai'i and are often poorly 412 APPLYINGRESEARCHTO MANAGEMENT maintained. Artificial containers, particu- construction project. These examples em- larly cemetery vases, refuse containers, dis- phasize that for source reduction to suc- carded tires, and construction materials, ceed, mosquito control must be a priority may harbor hundreds of mosquito larvae whenever alterations to refuge lands are in a relatively small volume of water. Un- considered. fortunately, control of mosquito produc- Source reduction need not be entirely tion by means of source reduction requires limited to human infrastructure. As men- considerable cooperation and vigilance tioned earlier, feral pigs create larval mos- from the public, and without strong in- quito habitats throughout the wet forests centives (or penalties) community-based of Hawai'i. Source reduction of hapu'u campaigns are seldom successful. Source cavities by elimination of the feral pig is reduction campaigns are more likely to suc- feasible but not without great cost and ceed on government-owned lands where controversy. The costs of pig eradication environmental management practices can in Hawai'i Volcanoes National Park during be mandated. On a small scale, such as in the 1980s were estimated to be $24,000 an atoll refuge, mandated source reduc- per km for fencing, $25,000 per year tion has the potential to eliminate vectors for fence maintenance and inspection on entirely. 71 km offence line, and an average of$95 In Midway Atoll National Wildlife per pig for hunting (Hone and Stone Refuge, removal of refuse containers and 1989, Katahira et al. 1993). However, the tarpaulins associated with environmental cost of feral pig control is not the only mitigation work and the decommission- obstacle.Feral pigs are a favored game ing of military infrastructure was followed species, and proposals to control or elim- by a decrease in the occurrence of avian inate them from public lands have been pox in nestling seabirds (J. Hale, U.S. Fish met with opposition from some hunters. and Wildlife Service, pers. comm.). Unfor- After years of dialogue between conser- tunately,C.quinquefcsciorus and A. albopictusare vationists and hunters, the controversy over still present on Midway Atoll, and in 2005 the control of feral pigs on some public an avian pox outbreak occurred in nestling lands continues (Tummons 1997). Laysan Albatrosses (Phoebastriaimmutabilis) 0. Klavitter, pers. comm.).The creation CHEMICAL CONTROL of artificial wetlands for the restoration of OF MOSQUITO VECTORS Laysan Ducks (Anas laysanensis) on Midway (U.S. Fish and Wildlife Service 2004b) may The use of insecticides to control mos- have inadvertently increased mosquito quitoes has saved countless lives world- populations. wide from the ravages of vector-borne Similarly, in Hawaii Volcanoes National disease. Chemical approaches to mosquito Park, the feasibility of breaking avian control also have had a negative impact malaria transmission through the source on the environment. Over the past century, reduction of larval mosquito habitat was the chemical control of mosquitoes has tested at a 100 ha former cattle ranch shifted from treating for larvae (larvicid- where the larval mosquito habitat was as- ing) with coal oil (kerosene) (Van Dine sociated with ranch infrastructure. Though 1904) and the chlorinated hydrocarbon unpublished, the study demonstrated that DDT to targeting adults (adulticiding) with the removal and treatment of troughs and organophosphates and pyrethroids (Naka- cisterns virtually eliminated mosquitoes gawa 1964, Mulla 1994, Rose 2001). In (D. LaPointe, unpubl. data). Unfortunately, the past three decades, concern over the the mosquito population rebounded when impact of pesticides on human health and a new larval habitat was created by a nearby the environment has brought about a more MANAGING DISEASE 413 integrated approach to mosquito control, mosquito numbers often rebound quickly combining source reduction and biological after adulticiding. In Hawaii, C. quinquefas- control with novel, environmentally sound ciatushas multiple cohorts throughout the larvicides and ultralow-volume (ULV) year and would require continuous appli- adulticiding (Mulla 1994, Rose 2001). cations to suppress adults for any length When source reduction is inappropriate of time.Although chemical control may be and larval mosquito habitats are accessible, effective in large open wetlands or in sub- larviciding is often the preferred method urban or agricultural environments, it is of control. Since the 197 Os,artificial insect difficult to imagine a cost-efficient and ef- growth regulators and surfactants have been fective landscape-level broadcast of insec- used in natural and man-made waters for ticides over native Hawaiian forests. control of mosquito larvae (Mull a 1994). Aside from the economic and logistical Insect growth regulators such as metho- constraints, there are greater concerns prene (Altosid®) interfere with pupal molt- regarding the use of chemical pesticides. ing by mimicking a hormone common to Evolving insecticide resistance makes chem- all insects and crustaceans, whereas sur- ical control subject to complete failure. At- factant monomolecular films alter the tempts to manage resistance may postpone surface tension of natural waters and affect or even prevent failure of chemical con- surface- respirating invertebrates. Although trol, but such management would require some adverse nontarget effects have been exceptional coordination with the agricul- reported, these chemicals are essentially tural community (Brogdon and McAllister nontoxic to vertebrates and most inver- 1998). The final factor weighing against tebrates when applied at recommended chemical control is the actual and per- rates (Nayar and Ali 2003, Pinkney et al. ceived adverse impact of insecticides on 2005). These larvicides can be effectively human health and the environment. Main- used in suburban and agricultural settings, land monitoring for nontarget effects as- but larval mosquito habitats in Hawaiian sociated with ULV adulticiding has been forests are too widely dispersed and in- largely limited to a few indicator species accessible to allow employment of con- (Mount 1998). Little is known about the ventionallarviciding. long-term effects of adulticiding on bio- When environmental conditions are not diversity. Hawai'i has more than 5,000 en- suitable for larviciding or when faced with demic arthropod species, including more a public health emergency, mosquito con- than 20 species proposed for federal listing trol professionals rely on adulticiding. as endangered (Howarth and Mull 1992). Modern ULV adulticiding applies a small The regulatory hurdles required just to ini- amount of organophosphate or pyrethroid tiate adulticiding in Hawaiian natural areas insecticides per hectare and, using aerial would be daunting and likely insurmount- applications, can treat vast areas in a short able. Given the nonselective toxicity of time. Under ideal conditions, ULV adulti- these chemicals and the already fragile sta- ciding can achieve 90% control without tus of Hawaii's native fauna, it would be detectable nontarget effects (Mount 1998, extremely problematic to consider chem- Jensen et al. 1999, Zhong et al. 2003). ical control in Hawaiian forest bird habitats. Still, ULV adulticiding is no more likely to aid in the control of avian disease in CLASSICAL BIOLOGICAL CONTROL Hawai'i than is conventional larviciding. OF MOSQUITO VECTORS ULV applications are not as effective in dense vegetation (Mount 1998) and there- Classical biological control can be de- fore are unlikely to penetrate a closed- fined as the control of a pest species by canopy Hawaiian forest. Furthermore, adult introduced natural enemies. The earliest 414 APPLYING RESEARCH TO MANAGEMENT attempts to control mosquitoes in Hawai'i the populations of A. albopictusor any other included the establishment of the wrinkled container-inhabiting species (Nakagawa frog (Rana rugosa) and the dart-poison frog 1963). The failure of Toxorhynchitesas a self- (Dendrobatesauratus).Both species failed to sustaining biocontrol agent in Hawai'i is control mosquito numbers (Oliver and consistent with other attempts made world- Shaw 1953). wide; only through inundative releases A number of poeciliid fish have been of this predator has control been achieved identified as predators of mosquito larvae (LaceyandOrr 1994). and have been used extensively in Hawai'i Copepods in the genus Mesocyclopshave (Nakagawa and Ikeda 1969). The first spe- shown some promise for the control of cies imported for mosquito control were container-inhabiting mosquitoes (Riviere the western mosquito fish (Gambusiaaffinis), and Thirel 198 1). Marten (1 984) observed sailfin molly (Poecilialatipinna), guppy (Poe- that in Hawai'i A. albopictuslarvae were read- cilia reticulata), green swordtail (Xiphophorus ily fed upon by a naturally occurring cope- hellerii), and southern platyfish (Xiphophorus pod, M.leuckarti pilosa. Under experimental maculatus) (Van Dine 1907, Brock 196 0). conditions, these copepods provided com- The western mosquito fish is extremely plete control of the mosquito. Similar suc- adaptable and effectively reduces mos- cess using various Mesocyclopsspp. has been quito numbers. Unfortunately, it also dis- reported for other container-inhabiting Aedes places native fish species (Meisch 1985). spp. (Gorrochotegui-Escalante et al. 1998, The guppy is more effective in polluted Kay et al. 2002, Nam et al. 2005). Mesocy- waters than mosquito fish and has been clopsare less effective at controlling C. quin- used in Hawai'i to control C.quinquefasciatus quefasciatus, although some species show in poultry waste runoff (Bay 1985). Both promise. A few copepod species have been species were released in natural areas as adventitiously established in the Hawaiian well as artificial impoundments, so today Islands, including a species commonly used these fish are ubiquitous in lowland waters in control, Mesocyclopsaspericornis (Nishida of the state. Introduced poeciliid fish have 2002). Copepods are frequently encoun- been implicated in the population loss tered in streambed rock pools even at high of native Megalagriondamselflies in Hawai'i elevations (1,800 m), but their signifi- (Englund 1999). Due to their proven neg- cance in limiting mosquito populations is ative impact on native aquatic biodiversity, unknown (D. LaPointe, unpubl. data). The introduced fish should not be considered value of Mesocyclopsfor control of mosqui- for mosquito control in natural areas. toes in natural areas is greatly minimized Over the past century, a number of in- by its susceptibility to desiccation and lim- vertebrate predators have been considered ited dispersal ability. As in the case of other for mosquito control (Lacey and Orr invertebrate predators, periodic augmen- 1994). Toxorhynchiteiss a genus of tree hole- tation of copepod populations would be inhabiting mosquitoes characterized by necessary to achieve lasting control (Lacey large predacious larvae that prey on smaller and Orr 1994). mosquito larvae.The iridescent blue-green Microbial pathogens of mosquitoes in- adults are notably larger than other mos- clude protozoa, fungi, bacteria, and viruses. quitoes and do not feed on blood (Steffan Many of these agents do not recycle or 1968). Two species, T.brevipalpisand T.am- reproduce and must be reapplied in the boinensis,were released between the years manner of a chemical agent. They are com- 1950 and 1959 (Bonnet and Hu 1951, monly referred to as biopesticides. The most Nakagawa 1963, Steffan 1968) . Although extensively used agent is the bacterium both species are probably established on Bacillusthuringiensisisraelensis(B.t.i.) (Margalit all major islands, they have not reduced and Dean 1985). It produces a toxin very MANAGING DISEASE 415

selective for mosquitoes and other nema- most compelling evidence comes from the toceran flies such as midges, blackflies, and Hawaiian Islands, where early agricultural biting midges. No significant adverse non- researchers introduced hundreds of gen- target effects have been reported from eralist predators and parasitoids, to the numerous laboratory and field trials (Lacey detriment of native insects (Howarth 1991 , and Mulla 1990). However, there is evi- Brenner et al. 2002). Certainly, the early dence that the long-term use of B.t.i. may introductions of fish for mosquito control alter aquatic invertebrate communities have had an impact on the native aquatic (Hershey et al. 1998). In Hawaii, there biota of the islands (Englund 1999). It is great concern that microbial agents will would appear that neither vertebrate nor impact already threatened native species invertebrate predators are suitable for such as Megalagrion damselflies (Howarth control of mosquitoes in Hawaiian natural 199 I), but there is little evidence of direct areas, because these generalists might de- or indirect toxicity of B.t.i. to odonates predate an already naturally depauperate (Painter et al. 1996).The major disadvan- aquatic fauna. Short-lived, selective bio- rages of B.t.i. use are its reduced efficacy in pesticides pose the least environmental polluted water, its nonresidual nature, and hazard and have their greatest value in ar- difficulties associated with its application tificial impoundments, although they may in dense vegetation. prove useful in streambeds or constructed Bacillussphaericus came into commercial wetlands. Both B.t.i. and B. sphaericus are use later than B.t.i. and is particularly ef- currently used by Hawai'i Department of :ective against Culexspecies (Singer 1985). Health (HDOH) Vector Control for mos- Unlike B.t.i., B.sphaericusprovides good resid- quito control in artificial impoundments. ual control, in part due to some natural B. sphaericushas been approved for use in recycling in the environment (Lacey et al. natural wetlands. 1987). Some C.pipienscomplex populations have developed resistance to B.sphaericus,but THESTERILE MALE TECHNIQUE there is no evidence of cross-resistance with AND CYTOPLASMIC INCOMPATIBILITY B.t.i.Thus, simple rotation of control agents in operational use should delay or prevent Perhaps the most efficient and elegant resistance development (Zahiri and Mulla form of mosquito control would make the 2003). No acute toxicity to or toxic trophic mosquito the agent of its own demise. effects on nontarget species have been re- Autocidal control methods have been con- ported (AlyandMulla 1987). sidered since the 1940s and employ radio- The most recently discovered microbe or chemosterilization of males, genetic showing some promise for controlling translocations, or bacterial symbiotes to C. quinquefasciatuspopulations is a baculo- interfere with insect reproduction. virus found naturally infecting C.nigripalpus The sterile male technique (SMT) has (Becnel et al. 200 1) . This baculovirus has been employed numerous times against undergone limited testing to date and ap- both agricultural pests and vectors ofhu- pears to infect only Culex species. It is a man or livestock disease (Asman et al. likely candidate for further development 1981,Klassen 2003).However, most at- as a biopesticide (Andreadis et al. 2003). tempts have had limited success in sup- In recent decades, the practice ofbiolog- pressing or eliminating populations, and ical control has been reevaluated as a grow- many of those that have been successful ing body of evidence has clearly demon- have targeted island or incipient popula- strated the loss of native biota following tions (Vreysen et al. 2000, Koyama et al. the introduction of biological control agents 2004). Perhaps the most successful at- (Simberloff and Stiling 1996). Some of the tempt to suppress a mosquito population 416 APPLYING RESEARCHTO MANAGEMENT with SMTwas conducted on Seahorse Key must be in place (Asman et al. 1981, in Florida in 1969. The daily release of as Townson 2002). many as 18,000 chemosterilized males C. quinquefnscictuspopulations have been for 10 weeks effectively eliminated C.quin- suppressed in isolated villages and islets quefasciatusfrom the key (Patterson et al. using SMT and CI, but is either approach 1970). Later attempts by Patterson and co- applicable to the control of avian disease workers (1977) were not as successful, in the Hawaiian Islands? U.S. Department and their failure was attributed to reduced of Agriculture facilities and programs for sexual competitiveness in radiosterilized the mass rearing and sterilization of teph- males and immigration into the target area ritid fruit flies have been in Hawai'i for by inseminated females. some time, and many of the logistical con- Cytoplasmic incompatibility (CI) was siderations about the rearing and release first observed in the 195Oswhen researchers of these insects could be adapted for reported that some crosses of mosquito mosquitoes. An archipelago-wide survey populations in the C. pipienscomplex re- for Wolbachiainfection in C.quinquefasciatus sulted in aborted embryonic development needs to be conducted to determine if (Laven 1959). Originally thought to be a multiple crossing types are present or if genetic effect, this phenomenon was later an incompatible synthetic strain could be discovered to be caused by a maternally developed. inherited bacterium, Wolbachiapipientis(Yen The size and landscape heterogeneity of and Barr 1971). CI can occur when an in- the larger Hawaiian Islands would be infected male mates with an uninfected fe- surmountable obstacles to mosquito erad- male or when infected males and females ication by SMTor C1.These techniques are of different crossing types mate (Bourtzis more appropriate to smaller islands and and O'Neill 1998). The potential use of CI atolls. Midway Atoll is geographically iso- for the control of mosquitoes was demon- lated, small, and accessible by air transport, strated by Laven (1967), who was able to making it an excellent location for an SMT eradicate a wild population of C. quinque- or CIproject. Complete eradication of mos- fasciatusin Burma through the repeated re- quitoes on Midway Atoll would open the leases of incompatible males. Despite the way for the translocation of Northwestern apparent success of this early trial, CI has Island endemic passerines such as Nihoa been largely ignored as a control strategy. Finches (Telespizaultima) or Laysan Finches Successful suppression or eradication of (T.cantans) . mosquitoes using SMT or CI requires that certain biological and logistical criteria TRANSGENIC OR GENETICALLY be met: (1) target species should display MODIFIED MOSQUITOES (GMMS) female monogamy and male polygyny, (2) target populations should be limited SMT and CI approaches to mosquito in number and isolated from immigra- suppression have been largely eclipsed in tion from native inseminated females or recent years by genetic engineering, and mosquitoes of different CI crossing types, the theoretical replacement of vector pop- (3) sterile or Wolbachia-infected males ulations with a refractory transgenic mos- should exhibit competitive mating behav- quito (Box 17.1) has become the new ior, (4) the technique must achieve a rate Holy Grail of malaria control (Crampton of sterility >95 %, (5) there must be ade- et al. 1990, Morel et al. 2002). The main quate facilities and proven techniques for thrust of this research has been the devel- the mass production of sterile or Wolbachia- opment of molecular tools for the stable infected males and the 100% exclusion of genetic transformation of mosquitoes (Cat- females, and (6) a reliable delivery system teruccia et al. 2000, Allen et al. 2001), MANAGING DISEASE 417 identification of refractory effector genes genic mosquito should be done in such a that will block parasite development (Niare way that any negative consequences would et al. 2002), and the development of be limited in both geographical extent and mechanisms to drive the refractory geno- impact on the target population. The avian type into wild mosquito populations (Ras- malaria system in the Hawaiian Islands has gon and Gould 2005). Much of this work many of the recognized criteria of a re- has focused on the Anopheline vectors of sponsible (from an anthropocentric per- human malaria, most notably Anopheles garn- spective) test location (Clarke 2002), and biae.The recently completed genome map the first step in developing a transgenic C. of A.garnbiaemay aid the development of a quinquefasciatushasalready been taken (Allen refractory transgenic mosquito (Holt et al. et al. 2001). It remains to be seen if fur- 2002), but finding an efficient genetic ther consideration of the Hawaiian avian drive mechanism remains the major tech- malaria system in transgenic research will nical obstacle and may take several more materialize. Transgenic techniques might years to resolve (Morel et al. 2002). have greater immediate value in the devel- There are many concerns associated with opment of markers or conditional lethal the transgenic approach to malaria con- genes for mosquitoes used in CLor SMT trol, ranging from the ethical to the eco- programs (Benedict and Robinson 2003). logical. General criticism focuses on the Even if this long-shot technique were fea- redirection of available malaria research sible, changing the vector competence of dollars away from integrated control strate- a mosquito population for malaria would gies and toward the transgenic approach. have no impact on the mosquito's ability to Some biological concerns include the out- serve as a mechanical vector of avian pox. right failure of this approach to reduce transmission, the variability in phenotypic expression of the trans gene, the trans- PROTECTING ENDANGERED gene's impact on fitness, inadvertent en- BIRD POPULATIONS hanced virulence, and a breakdown in efficacy over time, followed by a pandemic A number of high-elevation refugia with resulting from a loss of population immu- intact endemic forest bird communities nity (Spielman 1994, Scott et al. 2002, were identified on the islands of Kaua'I, Tabachnick 2003). Maui, and Hawai'i (Scott et al. 1986). Only What is unanimously agreed upon is some of this land was already under federal, that any responsible trial release of a trans- state, or private management (Chapter 16).

Box 17. 1. Transgenic Mosquitoes 101

In genetic engineering, genes from one DNA that moves genes about-the mo- organism are incorporated into the ge- lecular equivalent of cutting and past- nome of another organism. The resulting ing. Refractory effector genes encode for offspring that have successfully received mechanisms that prevent the develop- the transgene are said to be transformed ment of the parasite or pathogen in a or transgenic. Initial transformation is mosquito, and they are the target genes achieved by direct injection of the target for malaria vector transformation. A gene and a transposable element into genetic drive mechanism is a method the recipient eggs with a microsyringe. by which the refractory genes are spread The transposable element is a segment of throughout a vector population. 418 APPLYING RESEARCH TO MANAGEMENT

In 1985, land was set aside for the Hakalau impoundment on transmission could be Forest National Wildlife Refuge (NWR) , Significant. Vector densities would increase and in 1986 the Hanawi Natural Area Re- as available larval habitat became more serve was established on Maui. Preserva- abundant, and temperature constraints on tion of high -elevation forest has continued parasite development might be circum- with the acquisition of the South Kona vented by infrastructure micro climates subunit of Hakalau Forest NWR (1998) (Garnham 1948). Artificial structures ab- and the new Kahuku unit of Hawai'i Vol- sorb radiant energy, creating micro climates canoes National Park (2003).Studies of that foster parasite development. Careful malaria prevalence in birds at high eleva- consideration must be given to the con- tions (above ~ 1,500 m) suggest thatlocal struction of water impoundments and transmission is at most a very rare event infrastructure so as not to create an envi- (Feldman et al. 1995; Atkinson et al. 2005; ronment conducive to high-elevation trans- this volume, Chapter 9) constrained by low- mission of disease. temperature effects on vector and parasite Even some techniques employed for development (LaPointe 2000, Ahumada game bird management, conservation, and et al. 2004). habitat restoration work can backfire to support pathogen transmission. The use of black plastic tubs, barrels, or pond liners Maintaining Transmission-Free Refugia to provide water for game birds, outplant- Acquisition and management of transmis- ings, or bait for feral ungulates can sup- sion-free high-elevation habitat is crucial port mosquito populations in areas where to the preservation and restoration of na- natural water sources or favorable water tive Hawaiian forest birds but is valuable temperatures (through absorbed solar ra- only if it can be maintained as such well diation) do not exist (D. LaPointe, pers. into the future. Barring the possible effects obs.). The construction of artificial wet- of global warming or a land use change lands for native waterfowl, such as Koloa that would create a warmer microclimate, (Hawaiian Ducks, Anas wyvilliana), is partic- these refugia should remain avian malaria ularly risky and should be weighed against transmission-free zones even in the pres- the potential harm to the overall avian ence of increasing mosquito numbers. community of the area. Approval of con- Global warming could increase the alti- structed wetlands should be contingent on tudinal distribution of avian disease and routine monitoring for mosquito produc- would particularly threaten present-day tion and control when necessary. refugia where the geology or land use, such as cattle grazing, might impede the Creating Small Disease- Free Refugia upslope expansion of suitable forest habitat (Benning et al. 2002). Securing de- In the Hawaiian avian disease system, very forested and pasture land adjacent to pro- low vector numbers support high rates of tected refugia and managing it for forest transmission. Therefore, a mere reduction growth is the best long-term contingency in vector numbers may not be sufficient to plan against a warming scenario. prevent disease. Whenever feasible, control Of more immediate concern is the na- measures should strive for local eradica- ture ofland use change after cattle ranch- tion.Although SMT and CI strategies ing operations are phased out. Should land might be viable options for eradication of use in these adjacent areas instead shift to mosquitoes on smaller islands (Niihau, a more intensive agricultural or residential Lana'i, and Kaho'olawe) and atolls (Mid- use, the effect of infrastructure and water way), source reduction is the best strategy MANAGING DISEASE 419

N o 5 Forest and A KM Woodland

Figure 17.4. A model of mosquito ingression at the for the main islands. Source reduction will Hakalau Forest National Wildlife Refuge. The stippling most likely succeed in habitats that are al- indicates the zone of mosquito penetration from un- ready marginal for vectors. managed lands outside the refuge and is based on the conservative estimate of Culexquinquefasciatusdispersal at Unlike wet forests, where natural and 1.6 km mean dispersal distance. In this scenario, mos- feral pig-associated larval mosquito habi- quitoes have the potential to inundate the entire area of tats are likely to be abundant, xeric and the detached northern Maulua Unit and reduce the mesic forests may produce mosquitoes only effective transmission-free area of the main refuge by 60%. This figure emphasizes the importance of refuge through human activities that accidentally configuration to avoid unmanaged inholdings and to or purposely impound water. Source re- realize the significance of unmanaged adjacent lands in duction and/or treatment of impound- disease transmission. Source: Adapted from LaPointe 2008: ments can be accomplished with far less 607, Fig. 5, with permission from The Entomological effort and environmental harm than use of Society of America. such strategies in more natural habitats. Eliminating local production of mosquitoes will not be sufficient, however, if there are substantial nearby populations that can 420 APPLYINGRESEARCHTO MANAGEMENT disperse into the refuge. Klpuka-islands (Hemignathus spp.) , appears to have evolved of forest surrounded by lava flows-would tolerance. The year-round transmission that make ideal sites for future refuges. Their is characteristic of these coastal forests natural isolation by younger lava flows could be driving evolved tolerance. Low- would provide a maintenance-free buffer elevation, disease-ridden forests may not against dispersing vectors. Contiguous habi- be the endemic bird wasteland once sup- tat would require the creation of a buffer posed; instead they may serve as the prin- to protect the core habitat from dispersing cipal habitat for coevolution of introduced vectors. The extent of this buffer should be pathogens and native birds (Woodworth no less than 1.6 km, preferably 2-3 km, et al. 2005). and refuge design should strive for a low What seems clear is that the restoration ratio of perimeter to volume to minimize of native Hawaiian forest birds will de- edge effects (Fig. 17.4) . As a final consid- pend on the preservation of the full extent eration' planned refuges should not be in of native forest habitat, from the high- alignment with or adjacent to residential elevation refugia through the epizootic zone areas, roadways, or unmanaged wet forests at middle elevations down to the coast, along the path of prevailing winds. These where coevolutionary adaptation toward areas might provide an abundance of vec- disease tolerance can occur. Ideal conser- tors or corridors for wind-aided dispersal. vation landscapes would emulate the traditional Hawaiian unit of land division, the ahupua'a, extending from mountains DISEASE MANAGEMENT to seashore, and efforts to control disease IN LANDSCAPE-SCALE should focus on midelevation forests. CONSERVATION UNITS Most of the large, midelevation forests remaining in the Hawaiian Islands are wet The prevalence of avian malaria and its im- to mesic forests where C. quinquefasciatus and pact on host populations vary greatly over malaria are prevalent. The windward side the altitudinal range of forest bird habitat, of the main Hawaiian islands receives heavy so native Hawaiian forest birds are con- rainfall (2,500-11,000 mm annual pre- spicuously missing from suitable habitat. cipitation)(Giambelluca and Schroeder On the high islands in particular, large tracts 1998). However, precipitation is not the of native forest still extend the breadth of sole predictor of a landscape's potential for their elevational range, from coast to tim- production of vectors or transmission of berline, yet native bird densities and di- disease. Climate, geology, surface hydrology, versity are severely depleted at lower ele- and land use are strong determinants of vations (Scott et al. 1986). disease prevalence. Currently, eradication of mosquitoes is unlikely in wet forests, but some strategies may reduce their over- The Landscape of Mosquito-Borne all abundance and limit their populations Avian Disease during years of favorable conditions. Although pathogen transmission is virtually absent in high elevation forests Mosquito Control over Broad Landscapes (> 1,500 m elevation), it is common sea- sonally in native birds in midelevation Although high-tech solutions such as vac- forests, often resulting in fatal epizootics. cines and GMMs should not be ruled out, Most intriguing, however, is the situation there are proven traditional approaches to in low coastal forests, where disease preva- mosquito control that are applicable and lence is extremely high and yet at least immediately available. Elimination of feral one group of honeycreepers, the' amakihi pigs and source reduction of artificial lar- MANAGING DISEASE 421

val mosquito habitats should be consid- few realistic options for control. It remains ered the first step in eliminating or reduc- to be determined if streambeds can be suc- ing vector numbers. On windward Mauna cessfully treated by aerial application of Loa, for example, mosquito populations in biopesticides, and the cost is expected to large tracts of wet forests could be virtu- be great. Aerial applications may be lim- ally eradicated through indirect source ited to years oflow rainfall, when stream- reduction in the form of feral pig removal. beds make significant contributions to Though this management strategy is ex- mosquito numbers. Additionally, C.quinque- pensive and politically challenging, its po- fasciatusproduction in riparian landscapes tential overall benefit to the restoration of may be reduced by protecting streams forest bird populations is unparalleled. from fecal contamination by feral and do- Due to the dispersal capability of C.quin- mestic ungulates. quefasciatus,however, the success of feral pig removal in reducing transmission will be dependent on the size and shape of the LESSONS FROM 100 YEARS OF HUMAN managed land and on adjacent land use. MALARIA CONTROL EFFORTS Conservation units are better protected from immigrating mosquitoes when they Since the early 1900s, many nations have are shaped to minimize edge effects and been engaged in malaria control. The early large enough to include a minimum dis- successes of Gorgas in reducing yellow persal buffer. The creation of narrow con- fever in Panama, Soper's eradication of servation units along altitudinal contours is A. gambiaein Brazil, and Italy's success in to be avoided, because such units are more the Pontina and Sardinia fueled an opti- vulnerable to upslope, wind-enhanced mism that with enough financial resources mosquito dispersal. and manpower, malaria could be eradi- For residential and agricultural areas cated (Harrison 1978). With the introduc- adjacent to conservation units, integrated tion and use of DDT, a formal worldwide methods may be the best strategy. Initial malaria eradication campaign was launched survey and control measures could be con- by the World Health Organization (WHO) ducted by the HDOH Vector Control staff in 1956 (Desowitz 1991, Spielman et al. and community volunteers while training 1993). landowners to identify and eliminate mos- The key strategy of the campaign cen- quito production on their personal prop- tered on the use of a residual pesticide, erty. Simple yard sanitation will suffice in DDT, to kill infected females and in turn many cases. However, domestic and agri- break transmission long enough for exist- cultural impoundments that cannot be re- ing cases of malaria to self-cure, a period moved or covered may have to be treated estimated at five years. Additionally, active with fish or a biopesticide such as B.t.i. infections were treated with chloroquine, Care should be taken when using fish to the most effective chemotherapeutic agent ensure that accidental introduction into at the time. By 1967, a handful of Euro- natural waterways does not occur, and the pean and Caribbean nations claimed erad- continued use of biopesticides should in- ication, and India, Pakistan, and Sri Lanka clude some rotation of agents to avoid re- had greatly reduced malaria transmission. sistance.The success of such efforts will However, resistance to DDT and chloro- depend on annual outreach, assistance from quine quickly turned the tables. Five years HDOHVector Control, and the strength of later, WHO formally abandoned the goal neighborhood organizations. of global eradication and advocated in- In landscapes where intermittent streams stead for malaria control, first by new anti- produce the majority of vectors, there are malarial drugs and later by an elusive 422 APPLYING RESEARCH TO MANAGEMENT malarial vaccine that has never material- geographically isolated areas where mos- ized (Desowitz 1991).By the time of the quitoes cannot reinvade and where source malaria eradication campaign's demise, reduction can completely eliminate larval malaria had undergone a global resurgence habitat. For larger, more heterogeneous and the campaign had cost the United landscapes, an adaptive and sustainable States alone $ 790 million (Desowitz 1991). management approach using integrative The failure of the Global Eradication of control strategies coupled with an active re- Malaria Project was due to many factors. search program will ensure the selection of Poor communication from the field to locally effective methods. Technologically policy makers led to an illusion of success, sophisticated approaches such as transgenic and a strict and unrealistic time frame refractory mosquitoes may be a risky in- derailed project funding before success vestment but offer some potential. Finally, could be achieved (Desowitz 1991, Spiel- it is important to recognize that control man et al. 1993). For some regions, such efforts may require years of investment be- as sub-Saharan Africa, the project failed fore Significant results become evident and because planners grossly underestimated bird populations begin to recover. the variability and plasticity of the vector and pathogen (Spielman et al. 1993). As pesticide and drug resistance spread around SUMMARY the world, the fallacy of reliance on a single control strategy was made clear. One of the In the past century, Hawaiian forest birds most obvious errors of the program was have undergone extinctions and steep pop- the diversion of financial resources to con- ulation declines largely due to introduced, trol efforts with an almost complete aban- mosquito-borne avian pox and malaria. donment of research and training (Spiel- Today, while a few bird species appear to man et al. 1993). When the old techniques be coevolving to exist with these patho- failed, there were few alternative control gens, most populations of endemic species methods to try, and few individuals were continue to be severely limited by them. trained to take over the fight. Disease is most prevalent in wet forests Although world health agencies are below 1,500 m, where the climate is fa- still lured by modern technology, pursu- vorable to both the introduced mosquito ing elusive vaccines and the ultimate en- vector and avian malaria. Feral pigs have a gineered mosquito, it is unlikely that a key role in this disease system, for they of- panacea for malaria will ever be found. ten create the only or most favored larval Low-tech solutions, such as the use of habitat. As we begin to understand the insecticide-treated bed nets, have resulted dynamics and impacts of these diseases, it in modest reductions in transmission, but is clear that control of avian disease will when combined with traditional mosquito be crucial to the recovery and restoration control and pharmaceutical treatment they of native Hawaiian birds. could have a greater and lasting impact Due to the difficulties of mass adminis- (Vogel 2002). Local, conservatively sized tration or immunization, the use of anti- projects using integrated strategies are more malarial drugs and vaccines is limited to likely to be sustainable and effective in the protection of captive birds.Resistant malaria control (Spielman et al. 1993). bird populations may seem a ready source Though carried out at a much smaller of individuals for restoration, but these scale, management actions to control populations and the epidemiological con- mosquito-borne avian disease in Hawali ditions that maintain them are relatively (Table 17.1) would do well to heed these rare. Therefore, the options for avian dis- lessons. Eradication is realistic only for ease control in wild birds must target vec- MANAGING DISEASE 423

Table 17.1. A summary of management strategies for the control of avian disease through vector manipulation and potential obstacles to their use

Research or Technological Technology Development Politically Environmentally Management Recommendation Available? Needed? Sensitive? Sensitive?

Improve communication with HDOH Yes No Maybe No Vector Control Develop and disseminate outreach Yes Yes No No materials Mandate source reduction and mosquito- Yes No No No proof design of refuge infrastructure Consider vector movements during refuge Yes No No No acquisition and design Eliminate feral pigs from conservation Yes No Yes No lands Attempt eradication of mosquitoes on Yes Yes No No Midway Atoll using SMT or CI Monitor and control mosquitoes in refuge Yes Yes Maybe Yes wetlands Evaluate efficacy and nontarget impacts of Yes Yes Maybe Yes B.t.Lcontrol in natural streambeds Support disinfection of aircraft and No Yes Yes No container cargo Develop and release a GM refractory No Yes Yes Maybe mosquito

tor abatement. Traditional pesticides and modification of artificial containers and growth regulators offer little applicability impoundments and indirect source re- in conservation areas where larval habitats duction through the elimination of feral are difficult to locate and treat and where pigs can successfully reduce and poten- native wildlife and invertebrates would be tially eliminate vector populations. SMT affected by nonspecific toxicity. Histori- may work to eradicate mosquitoes from cally, classical biological control of mos- small islands and atolls, but the transgenic quitoes in the Hawaiian Islands has failed mosquito approach is too immature for to be self-sustaining or, worse, has led to immediate consideration. the decline of native species. Even long- It is unlikely that there will be any magic term augmentation and inundative releases bullets for abating avian malaria and pox of most biocontrol agents are unlikely to in Hawaii. The most productive conser- succeed given the vast area to be covered vation strategy would be to keep high- and the low density of the target prey. Only elevation refugia disease free, while the the crossover biopesticides are sufficiently main management emphasis would focus biologically selective and cost-effective to on vector suppression at the middle eleva- be considered for some riparian and wet- tions. We recommend investing resources land areas in Hawai'i. into integrative mosquito control and the Source reduction and mosquito-proof development oflong-term disease manage- design of refuge infrastructure is the first ment strategies. The successes and failures step in avian disease management. Direct of the Global Malaria Control Program source reduction through removal and demonstrate the need to adopt and adapt 424 APPLYING RESEARCH TO MANAGEMENT the old, proven technologies, such as ural Resource Protection Program, and source reduction, while keeping a hand the National Science Foundation (Grant in developing technologies. Avian disease DEB0083 944) for the financial support management is essential to the preserva- that made this research possible. The col- tion of the Hawaiian avifauna, and with lection of data for the various studies de- sound research and a commitment from scribed here would not have been possible land management agencies, the impact of without the assistance and support of disease can be minimized. numerous technical staff, co-workers, and student interns, who spent many long days in the field and laboratory under often ACKNOWLEDGMENTS difficult living and working conditions. We thank]. McAllister, WReisen, and two We thank the U.S.Geological Survey (USGS) anonymous reviewers for their comments Invasive Species program, the USGS Nat- on earlier drafts of this chapter.