Portal de Periódicos da CAPES

Larval habitat for the avian malaria vector Culex quinquefasciatus (Diptera: Culicidae) in altered mid-elevation mesic-dry forests in Hawai'i

2009; Wiley; Volume: 34; Issue: 2 Linguagem: Inglês

10.1111/j.1948-7134.2009.00028.x

1948-7134

Matthew E. Reiter, Dennis A. LaPointe,

Viral Infections and Vectors

Effective management of avian malaria (Plasmodium relictum) in Hawai'i's endemic honeycreepers (Drepanidinae) requires the identification and subsequent reduction or treatment of larval habitat for the mosquito vector, Culex quinquefasciatus (Diptera: Culicidae). We conducted ground surveys, treehole surveys, and helicopter aerial surveys from 2001–2003 to identify all potential larval mosquito habitat within two 100+ ha mesic-dry forest study sites in Hawai'i Volcanoes National Park, Hawai'i; 'Ainahou Ranch and Mauna Loa Strip Road. At 'Ainahou Ranch, anthropogenic sites (43%) were more likely to contain mosquitoes than naturally occurring (8%) sites. Larvae of Cx. quinquefasciatus were predominately found in anthropogenic sites while Aedes albopictus larvae occurred less frequently in both anthropogenic sites and naturally-occurring sites. Additionally, moderate-size (≈20–22,000 liters) anthropogenic potential larval habitat had >50% probability of mosquito presence compared to larger- and smaller-volume habitat (<50%). Less than 20% of trees surveyed at 'Ainahou Ranch had treeholes and few mosquito larvae were detected. Aerial surveys at 'Ainahou Ranch detected 56% (95% CI: 42–68%) of the potential larval habitat identified in ground surveys. At Mauna Loa Strip Road, Cx. quinquefasciatus larvae were only found in the rock holes of small intermittent stream drainages that made up 20% (5 of 25) of the total potential larval habitat. The volume of the potential larval habitat did not influence the probability of mosquito occurrence at Mauna Loa Strip Road. Our results suggest that Cx. quinquefasciatus abundance, and subsequently avian malaria, may be controlled by larval habitat reduction in the mesic-dry landscapes of Hawai'i where anthropogenic sources predominate. The widespread decline of endemic honeycreepers (subfamily: Drepanidinae) in the Hawaiian archipelago is of concern to conservation biologists and land managers (Jacobi and Atkinson 1995, Scott et al. 2002). Introduced mosquito-borne diseases (avian malaria [Plasmodium relictum] and avian pox [Poxvirus avium]) are recognized as major factors in the continued decline of Hawaiian birds (Warner 1968, van Riper et al. 1986, Atkinson et al. 1995, Jacobi and Atkinson 1995, Scott et al. 2002); as P. relictum significantly reduces survival in the honeycreepers (Atkinson et al. 1995). The spread of this disease to Hawai'i required the introduction of both the vector and parasite. Of the six introduced biting mosquitoes found in the Hawaiian Islands, the southern house mosquito, Culex quinquefasciatus, has been identified as the primary vector of avian malaria (Warner 1968, van Riper et al. 1986, LaPointe et al. 2005) and, most likely, avian pox. Effective disease management strategies are necessary for the successful restoration and long-term conservation of Hawai'i's honeycreepers (LaPointe et al. 2009). The elimination or reduction of vector populations as a management strategy to reduce disease transmission requires basic knowledge of the target species' larval ecology. In wet forests, where avian malaria is most prevalent, Cx. quinquefasciatus larvae are commonly found in water-filled tree fern cavities created by feral pigs (Sus scrofa; Goff and van Riper 1980) and in isolated rock pools along intermittent stream beds (Aruch et al. 2007). However, little is known about the larval ecology of Cx. quinquefasciatus in the mesic and dry forests of Hawai'i. Recent studies modeling mosquito distribution as a function of climate influences (Ahumada et al. 2004) and landscape factors (Kolivras 2006, Reiter and LaPointe 2007), including the potential range expansion of Cx. quinquefasciatus in Hawai'i as the result of global climate change (Benning et al. 2002), identify broad regions that promote increased mosquito abundance but do not provide recommendations for local-scale vector management. Systematic ground surveys to identify larval habitat can provide the data necessary for the development of local-scale management strategies. Low-level aerial surveys could also be employed to survey larger areas for potential larval habitat. As part of a larger study on avian malaria in native Hawaiian birds inhabiting mesic-dry forests, we conducted larval habitat surveys at two forest study sites in Hawai'i Volcanoes National Park (hereafter, HAVO), located on the island of Hawai'i. From 2001–2003, we conducted ground surveys, treehole surveys, and helicopter aerial surveys to identify larval habitats. The specific objectives of these surveys were to (1) locate all potential larval habitat, (2) assess each larval habitat for immature mosquito presence, (3) evaluate factors influencing the probability that a habitat supported larval mosquitoes, (4) evaluate the efficacy of helicopter aerial surveys to identify possible larval habitat across a broad landscape, and (5) identify associated aquatic invertebrates that might be impacted by vector management. Our study area was within the boundaries of HAVO on the eastern flank of Mauna Loa and Kilauea Volcanoes (Figure 1). The substrate consists of geologically young and porous lava flows (Wolfe and Morris 1996). Avian malaria and pox has been previously documented in this area (van Riper et al. 1986, 2002). The 'Ainahou Ranch site (hereafter, 'Ainahou; Figure 1) was purchased by HAVO in 1971, prior to which it was a cattle ranch. Much of the ranching infrastructure (e.g., ranch house, out buildings, cisterns, etc.) remains intact and is protected under the National Historic Preservation Act of 1966. 'Ainahou is located within the mid-elevation (1,000 m above sea level) woodland zone and can be described as mesic-dry forest (Stone and Pratt 1994). Between 1994 and 2003, 'Ainahou averaged 84 cm of rain per year. During 2001 and 2002, when primary surveying and monitoring occurred, 84.92 cm and 81.89 cm of rain per year were recorded, respectively. Location of 'Ainahou Ranch and Mauna Loa Strip Road Study sites in Hawai'i Volcanoes National National Park, Hawai'i, USA. The 'Ainahou area was divided into the 100 ha (small square in center) and 4,900 ha (entire hatched area) units. Inset map shows study area in relation to the main Hawaiian Islands. The 'Ainahou site was dominated by medium stature native trees, mainly 'ōhi'a (Metrosideros polymorpha) and māmane (Sophora chrysophylla), native shrubs including pūkiawe (Leptecophylla tameiameiae), 'a'ali'i (Dodonaea viscosa), 'ōhai (Sesbania tomentosa), and a selection of exotic species, most prominently the highly invasive fire tree (Morella faya) and molasses grass (Melinis minutiflora). The endangered 'akepa (Loxops coccineus) and 'i'iwi (Vestiaria coccinea) were two endemic bird species once common to this area. Currently, only the endemic honeycreeper, 'amakihi (Hemignathus virens), is abundant at the site, while the only other native passerine occurring at the site, 'apapane (Himatione sanguinea), is uncommon. 'Ainahou is an important breeding area for the endangered Hawaiian goose, nēnē (Branta sandvicensis). Non-native avian species occurring at 'Ainahou include Japanese white-eye (Zosterops japonicus), northern cardinal (Cardinalis cardinalis), house finch (Carpodacus mexicanus), nutmeg manikin (Lochura punctulata), kalij pheasant (Lophura leucomelana), and Erckel's francolin (Francolinus erckelii). Our second study site, Mauna Loa Strip Road (hereafter, MLS), was approximately 1,500 m above sea level and also a mesic-dry forest (Figure 1). It is generally classified as upland forest and woodlands (Stone and Pratt 1994). The eastern edge of the study area was the HAVO border and active cattle ranching occurred on adjacent lands. Between 1993 and 2003 average rainfall at the nearest weather station, Mauna Loa Keamoku (1,800 m), was 99.25 cm per year. Annual rainfall in 2001 and 2002 was 101 cm and 151 cm, respectively. Tree species at MLS included mainly koa (Acacia koa) and māmane with some 'ōhi'a on recent lava flows while native shrubs, such as pūkiawe and 'a'ali'i, dominated much of the understory. The open-grassland understory was composed of native grasses, alpine hairgrass (Deschampsia nubigena) and mountain pili (Panicum tenuifolium), and alien species including common velvet grass (Holcus lanatus), sweet vernal grass (Anthoxanthum odoratum), and Florida blackberry (Rubus penetrans). Native forest birds such as 'apapane, 'amakihi, and 'elepaio (Chasiempis sandwichensis ridgwayi) were common, while 'i'iwi were present in low abundance. These koa forests were once habitat for endangered 'akepa, 'akiapōlā'au (Hemignathus munroi), and Hawai'i creeper (Oreomystis mana), however they have not been observed in these forests since the late 1970s (Conant 1980, Banko 1984). In addition to those non-native birds found at 'Ainhaou, the red-billed leothrix (Leothrix lutea) occurred at MLS. We conducted systematic ground surveys for mosquito larval habitat (hereafter, larval habitat) on a 100 ha study site at 'Ainahou and MLS from June to September 2001. At 'Ainahou, an additional 82 ha were surveyed adjacent to the 100 ha study site and we surveyed and sampled potential larval habitat visible on private ranchland adjacent to our MLS study site. The survey team consisted of 3–4 observers who were spread out perpendicular to the direction traversed. We spaced observers so that we were able see potential larval habitat up to half the distance to the adjacent observer(s) and we adjusted inter-observer distances based on density of vegetation and visibility. We assumed the probability of detection to be 1. Observers at each end of the survey line carried Global Positioning System (GPS) units (GPS 12×L, Garmin Corp., Olathe, KS) and recorded track lines. We downloaded track lines into ArcView 3.3 (Environmental Systems Research Institute © 2000) and mapped daily search areas to ensure 100% coverage. We classified natural cavities (e.g., treeholes, rock holes, ground pools, and stream drainages) and anthropogenic containers (e.g., cisterns, troughs, water tanks, and artificial containers) holding or that could hold water, as potential larval habitat. At each potential larval habitat site, we recorded the type of habitat, status (wet or dry), volume of habitat (length × width × height [cm]; reported in liters), mosquito presence or absence, mosquito developmental stage(s), mosquito species, and geospatial coordinates in Universal Transverse Mercator (UTM). If the habitat held water, we sampled water using one of two methods: (1) a turkey baster and 25–60 mesh miniature sieves (Mini-Sieve™, Forestry Suppliers, Jackson, MS) if water volume was 2,000 ml), ten dips with a 500 ml mosquito dipper (Clarke Mosquito Control Inc., Roselle, IL) and miniature sieves. We rinsed sieves from each larval habitat sample into separate plastic bags and indentified all invertebrates. All instars of mosquito larvae were retained by sieves and identified to species. When we encountered stream drainages, we used a GPS unit to record the path of the drainage through the study area. We resurveyed the entire length of each drainage on the study area during regular visits in 2002 and 2003. Similarly, mapped potential larval habitat was rechecked opportunistically from 2001 to 2003 to reassess status as active larval habitat. We classified "potential" larval habitat as "active" if mosquito presence, of any stage, was recorded during ≥ one survey event. During ground surveys, we identified treeholes containing water that were located 0.5. We considered a model with a 95% CI for AUC that did not overlap 0.5 to predict better than random, however only a model with an AUC > 0.7 was considered a good fit. We calculated 95% CI for the AUC of each model, the proportion of trees with potential larval habitat, and the probability of detecting potential larval habitat from aerial surveys using 200-bootstrap resamples of the data and the percentile method (Manly 2001). We identified 91 potential larval habitats during ground surveys at 'Ainahou (Table 1). Fifty-two were identified within the 100 ha study site while an additional 39 were discovered in systematic ground searches of 82 ha adjacent to the 100 ha site. Forty-three (47%) potential habitats were anthropogenic and ranged from a large, open-topped cistern to a short section of exposed pipe. Forty-eight (57%) were naturally-occurring tree holes (predominately in 'ōhi'a). No stream drainages were identified at 'Ainahou. Median volume of anthropogenic sites (198.45 liters, range: 0.40 – 1 194 094, n= 29) was significantly (χ2= 40.63, df= 1, P < 0.0001) larger than the naturally occurring sites (1.38 liters, range: 0.01 – 10.50, n= 39). Eighteen (42%) of 43 anthropogenic sites contained Cx. quinquefasciatus larvae, pupae, or emerging adults at least once during our surveys. In four (9%) of 43 anthropogenic sites, we found Aedes albopictus co-occurring with Cx. quinquefasciatus. Only four (8%) of 48 naturally-occurring, potential larval habitats were active during the surveys. Two of the natural sites were 'ōhi'a tree holes and contained the larvae and pupae of Ae. albopictus. We also found Ae. albopictus larvae in a cavity in a native tree fern (Cibotium glaucum). We found no evidence of Cx. quinquefasciatus larvae inhabiting cavities in native trees but did find Cx. quinquefasciatus larvae in the water-filled root buttress of an exotic fig (Ficus sp.). The fig buttress collected leaf litter and fruit producing an organically rich habitat. There was a statistically significant association of anthropogenic sites with the probability of Cx. quinquefasciatus presence compared with naturally occurring sites (χ2= 17.64, df= 1, P < 0.001). However, there was no significant difference in the proportion of sites harboring Ae. albopictus between naturally-occurring and anthropogenic sites (χ2= 0.12, df= 1, P= 0.73). AUC supported a model with the quadratic form of the natural-log of habitat volume (VOL) as a better fit to the mosquito data from anthropogenic sites at 'Ainahou than a linear model form (Table 2). Coefficient estimates and predicted values suggested that small ( 22,000 liters) volume containers have a lower probability (<50%) of mosquito presence than mid-sized volumes (Figure 2; Table 2). Fitted logistic regression model of the probability of mosquito presence (MOSQUITO) in potential larval mosquito habitat at the 'Ainahou study area in Hawai'i Volcanoes National Park, Hawai'i. The model included an intercept term (β0), the natural-log of volume (VOL), and the quadratic effect of the natural log of volume (VOL*VOL). We identified 25 potential larval habitat sites at MLS (Table 1). Only three (12%) of 25 were anthropogenic. Median volume for natural sites (3.01 liters, Range: 0.07 – 280.57, n= 22) was smaller but not significantly different (χ2= 0.17, df= 1, P= 0.68) than the median volume of anthropogenic sites (4.96 liters, Range: 0.27 – 118.58, n= 3). While no anthropogenic sites contained evidence of mosquitoes, five (23%) of 22 naturally-occurring sites contained Cx. quinquefasciatus. Ae. albopictus was not identified in any larval habitat at MLS. We located 11 tree holes containing water, all in koa, but there was no evidence of mosquito presence. All naturally-occurring sites that contained Cx. quinquefasciatus were in stream drainages discovered during ground surveys. We conducted four surveys of two stream drainages discovered during initial ground surveys within MLS from September 2002 to July 2003. During each survey we walked the entire length of the drainage (0.9 and 1.3 km) and recorded the number of rock holes holding water and sampled the holes for mosquito presence. Evidence of mosquito reproduction was detected in two of five (40%) rock holes in the September survey, two of seven (29%) rock holes in the January survey, and one of four (25%) rock holes in the May survey. No mosquitoes were detected in a survey of five rock holes in July. We also identified seven anthropogenic potential larval habitat sites <3 km east of MLS on an adjoining active cattle ranch. These sites were surveyed once (March 2003) and, at that time, only three contained water. No mosquitoes were present. We surveyed 142 trees among 30 plots between June and November 2002 at 'Ainahou. One hundred-fourteen (80%) of the 142 trees were 'ōhi'a and 28 (20%) were exotic species (fire tree, Russian olive [Elaeagnus angustifolia], lemon [Citrus limon], slash pine [Pinus elliottii], common persimmon [Diospyros virginia], and ti [Cordyline fruticosa]). DBH measurements averaged 6.1 (Range: 2.9 – 8.9, n= 30) cm for class A, 16.4 (Range: 10.2 – 24.2, n= 30) cm for class B, 38.9 (Range: 30.3 – 60.0, n= 30) cm for class C, 78.4 (Range: 60.5 – 112.7, n= 24) cm for class D, and 17.7 (Range: 1.8 – 74.8, n= 28) cm for the exotics. Only 28 (19.7%) of all surveyed trees contained potential larval habitat. The highest proportion of trees with potential larval habitat, 0.38 (95% CI: 0.20 – 0.58) and 0.33 (95% CI: 0.16 – 0.50), occurred in the two largest DBH size classes of 'ōhi'a, respectively. The smallest size class of 'ōhi'a trees and the exotics had the lowest occurrence of potential larval habitat, 0.07 (95% CI: 0.00 – 0.17) and 0.07 (95% CI: 0.00 – 0.18), respectively. None of the water-filled tree holes sampled contained evidence of mosquitoes at any time during the survey. We identified 38 potential larval habitats during aerial surveys of 4900 ha at 'Ainahou. Overall, 29 of 52 (55.8%) potential larval habitats that were identified during systematic ground surveys of the 100 ha study site were subsequently observed during aerial surveys. We estimated the probability of detecting potential larval habitat during aerial surveys in this landscape to be 0.56 (95% CI: 0.42 – 0.69). Nine potential larval habitats identified in aerial surveys occurred in the surrounding area. Structures and objects observed from the helicopter were variable in size and included small pipes with a concrete base, tires at a gun firing range, water troughs, cisterns, and outhouses. Along with Cx. quinquefasciatus and Ae. albopictus, we encountered a limited diversity of macro- and micro-invertebrate fauna (Table 1). The few predatory species (Microvelia vagans [Veliidae], Mesovalia mulsanti [Mesoveliidae], Rhantus pacificus [Dytiscidae], and Buenoa pallipes [Notonectidae]) were restricted to large surface area cisterns. Chironomid midges (Chironomus hawaiiensis [Chironomidae]), ostracods, and pionid water mites (Piona lapointei) were most common in shallow, temporary habitat like rock holes and various small water containers. The only endemic species identified were the commonly-occurring Rhantus pacificus and Chironomus hawaiiensis. The identification of larval habitat is critical for the management of vector-borne diseases. Our data indicated that naturally-occurring larval habitat was limited in the mesic-dry forests of HAVO. In addition, at both 'Ainahou and MLS, none of the potential larval habitat was created by feral pigs. This was in contrast with wet forests in Hawai'i where feral pigs create numerous water-filled tree fern cavities which produce mosquitoes (Goff and van Riper 1980). Rock holes in the bed of intermittent stream drainages may be the most significant natural larval habitat for Cx. quinquefasciatus in mesic-dry Hawaiian forests. Cx. quinquefasciatus larvae have been observed in the rock holes and pools of intermittent streams elsewhere in the Hawaiian Islands (Aruch et al. 2007; personal observation, DAL). Our repeated surveys of the stream drainages suggested that a single survey may yield biased results as the proportion of rock pools harboring mosquitoes varied from 0–75% depending on the season the survey was conducted. Despite repeated observations throughout the year, we did not find larval Cx quinquefasciatus in the treeholes that were common in the larger size classes of 'ōhi'a and koa. This finding was in contrast to the observations of Goff and van Riper (1980) who found Cx. quinquefasciatus larvae inhabiting koa treeholes within a few km of our MLS site. Our observations of Ae. albopictus larvae in 'ōhi'a treeholes and identification of other invertebrate species occupying potential larval habitat in koa suggested that neither limited rainfall nor invertebrate predation accounts for the absence of Cx. quinquefasciatus larvae in treeholes. Overall, the density of potential larval habitat at 'Ainahou (52 per 100 ha) was twice that found at MLS (25 per 100 ha). At 'Ainahou, the proportion of anthropogenic sites with mosquito presence was much higher than in natural sites, whereas at the MLS study area, naturally occurring sites were more abundant than anthropogenic sites and the only habitat type used by Cx. quinquefasciatus. These differences in the habitat types have implications for vector management. At 'Ainahou, our study suggests the removal or selective larvicidal treatment (e.g., Bacillus thurgensis israelis) of anthropogenic larval habitats between 20 and 22,000 liters would reduce the mosquito population. While a limited number of treeholes did support larval Ae. albopictus, we found no evidence of immature Cx. quinquefasciatus in treeholes at 'Ainahou. Thus, vector management of anthropogenic sites would not be confounded by contributions from unmanaged natural sites. Vector management is further facilitated at an isolated location such as 'Ainahou as it is less likely influenced by unmanaged landscapes outside of HAVO boundaries which likely produce mosquitoes (Reiter and LaPointe 2007) and much of the area surrounding the immediate ranch-house area is grassland and barren lava. In addition, habitat removal or larvicidal treatment would have little impact on natural populations of endemic invertebrate species at 'Ainahou. Most of the non-target invertebrates collected during our surveys were non-native, adventive species (Nishida 2002) and the two endemic species, the non-biting midge Chironomus hawaiiensis and the diving beetle Rhantus pacificus, are common and widespread species in the Hawaiian Islands (Williams 1936, 1944). These aspects make 'Ainahou a likely site for successful vector management and potentially mid-elevation reintroductions of endangered native bird species. Unlike 'Ainahou, Cx. quinquefasciatus at MLS were found in naturally-occurring habitats, thereby confounding vector management. Physical alteration or larvicidal treatment of natural drainages on National Park Service property may require identifying potentially harmful effects to endemic aquatic invertebrates, and the completion of an Environmental Impact Assessment to comply with the National Environmental Policy Act of 1969 (Truett et al. 2005). However, the treatment of anthropogenic sites occurring on adjacent ranch lands may reduce the probability that mosquitoes persist in MLS rock holes, as rock holes dry up during periods of drought. In July 2003, there were few rock pools remaining in the stream drainages and none contained mosquito larvae or exuvia. The duration over which mosquitoes are absent from the stream drainage likely depends on the proximity of more permanent larval habitats. We discovered seven anthropogenic potential larval habitat sites just east of the MLS site, which could provide a source for mosquitoes. Monthly stream drainage dip surveys combined with the management of anthropogenic larval habitat sites outside of the park could provide valuable insight as to the potential for long-term vector management at this location. Due to the highly mobile nature of both birds and mosquitoes, large areas must be surveyed for larval mosquito habitat to reduce transmission of disease over a small area. To eliminate or effectively reduce mosquitoes in our 100 ha study area, an area as large as 4,900 ha may need to be surveyed to account for the dispersal ability of Cx. quinquefasciatus (LaPointe 2008). Broad-based techniques such as aerial surveys and remote sensing, followed by ground visits and sampling could greatly reduce survey time and effort in mesic-dry forests of Hawai'i. We achieved a 56% probability of detection when using helicopter aerial surveys to identify potential larval habitat. We are unaware of other attempts to use low-level aerial surveys for potential larval habitat identification and to estimate probability of detection, but we feel that aerial surveys can be an effective tool to reduce the area searched on the ground and increase the focus of these searches. New capabilities in remote sensing and improved resolution of both aerial and satellite imagery may provide another method to survey large landscapes (Kitron et al. 1996, Kitron 1998, Troyo et al. 2008) or at least increase the efficiency of ground searches. Although limited to two study sites, our study provides valuable information on the larval ecology of Cx. quinquefasciatus in mesic-dry forests of Hawai'i. Guided by our data, HAVO resource managers should be able to reduce mosquito abundance through larval habitat elimination, modification, and/or treatment with larvicides. Furthermore, this study emphasizes that, as local land use shifts away from ranching and agriculture in Hawai'i, removal of abandoned infrastructure is critical to the control of invasive mosquitoes and mosquito-borne avian disease in mesic-dry landscapes. The difficulty of conducting effective surveys (ground or aerial) is compounded when vegetation returns to an area reducing visibility and the probability of detection. We encourage land managers to conduct larval habitat surveys in the early stages of land restoration. Further efforts, like our aerial surveys, are essential to evaluate possible alternatives to ground surveys for broad, landscape-level larval habitat surveys. Additionally, larval habitat surveys of other mesic-dry forest habitat throughout the Hawaiian Islands are needed. We would like to thank J. Lease for her dedicated work on the project, C. Atkinson, C. Henneman, E. Tweed, and D. Okita for helping with aerial surveys, and the many interns that provided technical assistance throughout the project. We acknowledge the assistance of Hawai'i Volcanoes National Park resources management staff, including D. Hu, J. Chase, and K. Misajon. We thank S. Hess, D. Hu, and two anonymous reviewers for comments made on earlier drafts of this manuscript. This research was funded by the National Park Service through the Natural Resources Protection Program and the U.S. Geological Survey Invasive Species Program. Any use of trade, product, or firm names in this publication is for descriptive purposes only and does not imply endorsement by the U.S. Government.