Avian malaria (Plasmodium spp.) has been implicated in the decline of avian populations in the Hawaiian Islands and it is generally agreed that geographically isolated and immunologically naïve bird populations are particularly vulnerable to the pathogenic effects of invasive malaria parasites. In order to assess the potential disease risk of malaria to the avifauna of Socorro Island, México, we surveyed for Plasmodium isolates from 1,300 resident field-caught mosquitoes. Most of them were identified as Aedes (Ochlerotatus) taeniorhynchus (Wiedemann, 1821), which were abundant in the salt marshes. We also collected Culex quinquefasciatus Say, 1823 close to human dwellings. Mitochondrial ND5 and COII gene sequences of Ae. taeniorhynchus were analyzed and compared to corresponding sequences of mosquitoes of the Galápagos Islands, Latin America, and the North American mainland. Aedes lineages from Socorro Island clustered most closely with a lineage from the continental U.S. Plasmodium spp. DNA was isolated from both species of mosquitoes. From 38 positive pools, we isolated 11 distinct mitochondrial Cytb lineages of Plasmodium spp. Seven of the Plasmodium lineages represent previously documented avian infective strains while four were new lineages. Our results confirm a potential risk for the spread of avian malaria and underscore the need to monitor both the mosquito and avian populations as a necessary conservation measure to protect endangered bird species on Socorro Island.
Avian malaria is a vector-borne parasitic disease caused by various species of Plasmodium and is known to have adversely affected bird populations (Valkiūnas 2005). All currently known avian Plasmodium spp. rely on mosquitoes for the completion of their natural life cycles and transmission. It has been suggested that mosquitoes play a major role in limiting geographic ranges of bird populations and in the spread of avian malaria worldwide (Valkiūnas 2005). While most studies on the prevalence and diversity of avian malaria have focused on isolates from vertebrates, some more recent studies have examined strains and lineages in field-caught mosquitoes (Ejiri et al. 2008, Gager et al. 2008, Ishtiaq et al. 2008, Kim et al. 2009, Kimura et al. 2006, Njabo et al. 2009, Tompkins and Gleeson 2006).
We collected mosquitoes on Socorro Island to assess the disease risk to its endemic wildlife. The island is located approximately 700 km west of the port city of Manzanillo, Colima, México. With three other volcanic islands, it is part of the Revillagigedo Archipelago Biosphere Reserve (SEDESOL 1994). Wildlife is depauperate on Socorro Island, with terrestrial birds predominating among a few reptiles and only introduced mammals (Brattstrom and Howell 1956). Although Ae. taeniorhynchus was reported earlier in coastal habitats (Vázquez 1960), there is no recent documentation on the distribution and diversity of the mosquito fauna on this island. Because of its location, Socorro Island shares several biogeographical attributes with the Galápagos Islands and has a high degree of biotic endemicity (Brattstrom 1990, Levin and Moran 1989).
We hypothesized that due to high levels of human-aided transport through boats and airplanes and a comparable location in the Eastern Pacific Ocean, Socorro Island would similarly harbor mosquito species found in the Galápagos Islands. By employing a phylogenetic approach using mitochondrial DNA (mtDNA) sequence analyses we explored the possible origins of the mosquitoes on the island and the diversity of their haemosporidian parasites. We examine for the first time avian parasite infectivity in mosquitoes found on this remote island and discuss its conservation implications.
MATERIALS AND METHODS
Mosquitoes were collected in July, 2009 from five sites on the island (Figure 1): Las Grutas (Site 1; N 18° 44’ 4. 8”, W 110° 56’ 48. 3”), Los Cedros (Site 2; N 18° 45’ 28. 9”, W 110° 56’ 44. 4”), Playa Blanca (Site 3; N 18° 48’ 51”, W 111° 02’ 422. 5”), Llano de Borregos (Site 4; N 18° 45’ 57. 5”, W 110° 56 ‘58. 5”), and the pier (Site 5; N 18° 44’ 41. 6”, W 110° 56’ 30. 7”) using eight Center for Disease Control (CDC) Miniature Light Traps baited with CO2 emitting packets (John W. Hock, Gainesville, FL). Traps were hung from trees approximately 1 m above ground and 6 m apart from one another and operated from dusk until dawn (approximately 12 h). Early in the mornings, collection bags were removed from the traps and the mosquitoes were immobilized by smoke fumigation and sorted by sex. Mosquitoes were then pooled into groups of 20 and held in vials containing silica beads as a desiccant to prevent DNA degradation until identification.
Voucher female and all male mosquitoes were pinned and identified using a stereomicroscope and morphological keys (Belkin, 1962). Genitalia were removed from the few male mosquitoes collected and slide mounted according to Belkin (1962).
Mosquito and parasite screening
The head/thorax of each female mosquito from each pool was severed from the abdomen following the methods described in Ishtiaq et al. (2008). DNA was extracted from both the abdomen and head/thorax pools using the Wizard SV Genomic DNA Purification kits (Promega Corporation, Madison, WI) with the modification of adding an additional 20 μl of Proteinase K prior to incubation and homogenization with a heat-sealed pipette tip (Njabo et al. 2009). All polymerase chain reactions (PCR) were carried out in a 25 μl reaction mixture containing 10–100 ng of genomic DNA (2 μg of template DNA), 10 mMTris-HCl (pH8.3), 50 mMKCl, 3.0 mMMgCl, 0.4 mM of each dNTP, 0.4 mM of each primer, 5 μl of CL buffer (Qiagen, Valencia, CA), and 0.5 units Taq (Qiagen, Valencia, CA).
For mosquito mtDNA sequence analysis, we amplified from four pools of abdomens ∼654 bp of the mitochondrial cytochrome oxidase subunit 2 gene (COII) by using the thermocycler protocols and the forward primer SCTL2-J3037 from Bataille et al. (2009b) and Cook et al. (2009), along with a modified reverse primer (5’-GATTTAAGAGATCATTACTTGC-3’; Bataille et al. 2009b). We also amplified ∼610 bp of the NADH dehydrogenase subunit 5 gene (ND5) using the ND5 and tRNAPhe primers and thermocycler protocols from Krzywinski et al. (2001).
The DNA extracted from head/thorax and abdomen mosquito pools was then screened for both Plasmodium and Haemoproteus parasite DNA using PCR. We amplified ∼ 750 bp of the parasite mitochondrial cytochrome oxidase subunit-b gene (cytb) using two sets of primers. The first set of primers used were L15183 and H15730 (Chasar et al. 2009, Fallon et al. 2003). The second primer set consisted of the nested primers HaemNF, HaemNR2, HaemF and HaemR2 as described in (Waldenström et al. 2004). All PCR products were screened and viewed on 1.8% agarose gels stained with ethidium bromide for examination of amplification products.
Positive PCR products were bi-directionally sequenced by Elim Biopharmaceuticals Inc., Hayward, CA. From four pools of Ae. taeniorhynchus mosquitoes, we sequenced 654 bp of the COII gene and 588 bp of the ND5 gene. All amplicons sequenced where aligned and edited using Sequencher 4.8 (GeneCodes, Ann Arbor, MI). Because both amplicons had low levels of variation, they were combined for a total of 1,242 bp to increase the resolution power for data analysis.
From mosquito abdomen and head/thorax pools that were positive for parasite DNA, we then sequenced 750 bp of the Plasmodium spp. cytb gene and edited using Sequencher 4.8 (GeneCodes, Ann Arbor, MI). Sequences were identified as Plasmodium spp. using the NCBI nucleotide Blast search and retrieved matches ranging from 97–99% in GenBank. Parasite sequences that differed by only 1 to 3 bp were considered distinct lineages (Ricklefs and Fallon 2002). Distinct lineages were then verified by repeating an independent PCR and sequencing analysis. Sequence chromatograms were visually inspected for presence of double peaks to ensure that no pools contained multiple distinct lineages of Plasmodium spp.
All final sequences were deposited to GenBank with accession numbers HQ853668-HQ853680. And lastly, to estimate the proportion of infected individual mosquitoes from the pooled samples, the maximum likelihood estimation (MLE) was calculated using the equation MLE= 1 – (1 – number of positive pools/number of total pools)1/pool size (Gu et al. 2009). This number represents the expected proportion of infected individuals in the sample incorporating a correction for the pooling process.
Phylogenetic relationships were analyzed using MrBayes version 3.1.2 (Huelsenbeck and Ronquist 2001) using the GTR+G model for both the mosquito and parasite phylogenetic hypotheses obtained from MrModeltest (Nylander 2004). In addition to Ae. taeniorhynchus COII and ND5 sequences obtained from Socorro Island (HQ853679- HQ853680), sequences from 32 mosquitoes originating from the Galápagos Islands, Latin America, and North America (Bataille et al. 2009b) were obtained from Genbank (accession numbers FM992116-FM992321). The COII and ND5 genes were considered a single dataset since they were characterized by the same nucleotide frequencies and the best-fit nucleotide substitution model. Due to the lack of an appropriate outgroup, we constructed an unrooted tree. Such an approach was not possible for Cx. quinquefasciatus because we did not have sufficient sequence data for this species.
Plasmodium spp. mitochondrial cytb sequences were analyzed from 11 distinct lineages from Socorro Island and 34 sequences that were downloaded from Genbank. Genbank Plasmodium spp. sequences were from American mainland (Kimura et al. 2006), South Pacific (Ishtiaq et al. 2008), Japanese (Ejiri et al. 2008), and African parasite samples (Beadell et al. 2009, Chasar et al. 2009). Their respective accession numbers are provided in Figure 3. Plasmodium gallinaceum was used as an outgroup lineage. Two Markov Chain Monte Carlo (MCMC) simulations were run simultaneously for ten million generations with sampling every 200 generations generating 100,000 trees. The first 25,000 trees were discarded from the sample as the “burn-in” period that accounted for 25% of the trees. The remaining trees were used to construct a majority rule consensus tree and to calculate posterior probabilities of the individual clades (Labarthe et al. 1998).
Mosquitoes were observed and collected at only two of the five sampling sites, Playa Blanca (Site 3) and the pier (Site 5; Figure 1). From these two sites a total of 1,300 mosquitoes were either pinned as voucher specimens (Bohart Museum, UC Davis) or pooled for DNA isolation. Two species were identified; Culex quinquefasciatus Say, 1823 and the black salt marsh mosquito Aedes (Ochlerotatus) taeniorhynchus (Wiedemann, 1821). Identification of Cx. quinquefasciatus as a member of the Culex pipiens complex was confirmed by morphology of male genitalia (DV/D index) as described by Cornel et al. (2003). Cx. quinquefasciatus was only collected at the pier (site 5) and consisted of 13 specimens, two of which were males. Ae. taeniorhynchus was by far the most abundant species and consisted of the remaining 1,287 specimens pooled or pinned, which in fact represented only a fraction of the total numbers collected of this species. We are unable to confirm potential breeding sites since no larvae were collected, but based on the location where the two mosquito species were collected (Figure 1), we can speculate where they might occur. Culex quinquefasciatus was only collected near the Naval base where some water cisterns are not properly covered, thus providing a potential breeding site for this mosquito species (Belkin et al. 1970). The pier also harbored Ae. taeniorhynchus, a mosquito known for breeding in transient pools of salt and brackish water in low-laying coastal areas (Belkin et al. 1970). Aedes taeniorhynchus was also collected at Playa Blanca, a cove surrounded by Conocarpus erectus mangrove and temporary pools of brackish water. The volcanic nature of Socorro Island results in uneven terrain that provides ample opportunity for pools to form during rains and after high tides, which Ae. taeniorhynchus may use as breeding sites.
Female mosquitoes were pooled to create 61 separate head/thorax and abdomen pools of Ae. taeniorhynchus (20 in each) and one pool each of 11 Cx. quinquefasciatus head/thorax and abdomens for DNA analysis. Just over half of the Aedes abdomen pools (37/61) tested positive for parasite DNA, yielding a maximum likelihood estimation (MLE) of 4.8%, while the single Culex abdomen pool tested positive for parasite DNA yielding an MLE of 100%. Only one of 61 Aedes head/thorax pools tested positive for parasite DNA yielding an MLE of 0.08%.
Phylogenetic relationships among the Ae. taeniorhynchus mosquitoes from Socorro Island, Galápagos Islands, and the American mainland (Figure 2) reveal five distinct clusters of Aedes lineages: I. the Atlantic coast and Gulf of México, II. the Galápagos Islands, III. the Pacific coast, IV. Latin America, and V. Socorro Island. The Aedes lineage found on Socorro Island interestingly clustered with a lineage from the Gulf of México (Lou6 from Louisiana, U.S.A.). We should point out that the sequences used for the Socorro Island Ae. taeniorhychus were derived from four pools of 20 mosquitoes in each. Each pool produced amplicons of identical sequence for both ND5 and CO II genes.
In total, 11 distinct Plasmodium Cytb lineages were isolated. No Haemoproteus spp. were detected in the mosquitoes. Phylogenetic relationships among Plasmodium spp. found in mosquitoes from Socorro Island and those found in the American mainland, South Pacific, Japan, and Africa are depicted in Figure 3. The tree shows that Aedes isolates SocP1 and SocP8 fell in clade IV and SocP7 fell into clade III. SocP11, which was found in the only Culex pool, fell in clade II. Lineage SocP1 was the most abundant parasite in mosquitoes sampled; it was found in 25 out of 61 Aedes pools and it included the single positive thorax/head Aedes pool. The remaining seven lineages were all found in a monophyletic clade (II in Figure 3) with Plasmodium spp. found in birds of Africa.
Our study provides significant documentation of the mosquito fauna on Socorro Island. The species collected were Cx. quinquefasciatus and Ae. taeniorhynchus, both of which occur on the Galápagos Islands (Bataille et al. 2009b). Cx. quinquefasciatus, reported for the first time on Socorro Island, is a known vector for avian pox, West Nile virus (WNV), and avian malaria in Hawaii (Atkinson et al. 1995, Van Riper et al. 1986, Warner 1968). Ae. taeniorhynchus, on the other hand, has not been implicated as a competent vector for avian malaria, but is a known vector of dog heartworm (Dirofilaria immitis) in Latin America (Labarthe et al. 1998) and a competent vector for St. Louis encephalitis (Nayar et al. 1986) and West Nile viruses (Nayar et al. 1986, Turell et al. 2001). We cannot absolutely confirm that these are the only two species on Socorro Island, as we did not place traps in a systematic manner to cover every habitat on the island. Furthermore, most trapping nights were windy and not entirely conducive to optimal mosquito capture.
Distribution patterns of these two mosquito species on Socorro Island differed. Cx. quinquefasciatus were collected only at the human-inhabited site at the pier near the Mexican Naval Station. This is consistent with findings on the Galápagos Islands where Cx. quinquefasciatus was also found only near human settlements, and that they are recent introductions. As in the Galápagos Islands (Bataille et al. 2009a, Bataille et al. 2009b), rainfall on Socorro Island is low, resulting in a scarcity of fresh water sources especially on lower altitude vegetation types (e.g., Flores-Palacios et al. 2009). Ae. taeniorhynchus was widely distributed along the coast of Socorro Island and was not caught inland nor above 200 m. This differs from the findings on the Galápagos Islands where Ae. taeniorhynchus were found up to 20 km from the coast and up to an altitude of 700 m. This unusual mosquito distribution led Bataille et al. (2009b) to suggest that Galápagos Ae. taeniorhynchus have undergone adaptations in breeding and radiated out of solely salt marshes into other breeding sites. This occurred over a period of 200,000 years, an estimate for the introduction of this species into the Galápagos Islands (Bataille et al. 2009b).
On Socorro Island, Ae. taeniorhynchus shares a more similar biology to mainland mosquitoes where it is rare to be found >6 km from the coast line, and if they are seen inland, it is only to breed (Nielsen and Nielsen 1953, Provost 1951). Our phylogenetic analysis of the ND5/COII genes (Figure 2) suggests that Ae. taeniorhynchus colonized Socorro Island more recently than the Galápagos Islands, as their sequences aligned most closely with a lineage found in the Gulf of México. These mosquitoes were most likely introduced by human activity, as was surmised with Culex mosquitoes from the Galápagos (Bataille et al. 2009b), but there is no documentation of when this may have occurred.
Our results provide evidence that several Plasmodium species occur on Socorro Island, but whether Cx. quinquefasciatus and Ae. taeniorhynchus are the transmitting vectors of all the lineages is not clear. Most Plasmodium spp. isolates were obtained from pools of mosquito abdomens, which does not verify their competence as vectors. To confirm vector competence, infective sporozoites must at least be detected in mosquito salivary glands (Njabo et al. 2009). One Plasmodium isolate was obtained from a head/thorax of an Aedes pool that lacked a recent blood meal, which suggests that Ae. taeniorhynchus could transmit Plasmodium spp. on the island. This is the first study providing evidence that Ae. taeniorhynchus might be a competent vector for avian malaria.
The phylogenetic tree of the parasite lineages (Figure 3) found in the mosquitoes illustrates the diversity of the Plasmodium spp. on Socorro Island. All distinct lineages found on this island did not match any other Plasmodium spp. lineages found on GenBank, suggesting that the Socorro Plasmodium spp. may be unique to this island. Clade I was composed of seven distinct parasite lineages found on Socorro Island along with three Plasmodium spp. that have been described from avian populations of Africa. Although there is one endemic lizard present on the island (Urosaurus auriculatus) (Brattstrom 1990), none of the 11 Plasmodium distinct lineages fell into the clade with two known Plasmodium spp. to infect reptiles; Plasmodium mexicanum and Plasmodium chiricahuae. This indicates that the trapped mosquitoes relied on avian populations as their main source of blood meals. However, we were unable to analyze blood meals for host identities due to the lack of recently blood-fed trapped mosquitoes during our collection. Moreover, due to the low support for the SocP11 placement in Clade II (Figure 3), which is the parasite lineage isolated in the only Culex pool, we are unable to make any assumptions of how it is related to the other Plasmodium lineages.
In order to protect the endangered avifauna of Socorro Island, and to prevent further extinctions, mosquito populations should be monitored continuously and control measures should be incorporated in the management plans of this Biosphere Reserve. Although it is unclear whether Ae. taeniorhynchus is a competent vector for avian malaria, it should be considered a serious threat to the avifauna of Socorro Island. If Ae. taeniorhynchus were to move inland, threatened birds that currently reside away from the coastline might be exposed to Plasmodium spp. Dispersal patterns of native birds require further study to help design epidemiological models to estimate disease risk and spread across the island. At this time, with a combination of timely research and sound management practices, there exists an opportunity to avert the scenario presented to the Hawaiian Islands: the collapse of an insular endemic avifauna.
We offer our sincere gratitude and recognition to the Secretaría de Marina – Armada de México, in particular, to Admiral Ezequiel Ramiro Bordonave, Captain Pedro Villaciz León and the military personnel stationed at Socorro Island for their hospitality and logistical support, without which this study would not have been possible. We thank Dr. Chindi Peavey at the San Mateo County Mosquito and Vector Control District for offering technical advice and for allowing us to use mosquito traps. Dr. Víctor Rico-Gray and Dr. Vicente Hernández at the Instituto de Ecología in Xalapa (INECOL) provided valuable insights. Funding was provided by the College of Science and Engineering, San Francisco State University, the American Society of Parasitologists Willis A. Reid Student Research Grant, the San Francisco State University Instructionally Related Activities Grant, the Island Endemics Foundation, and INECOL. All research was carried out under permit (SGPA/DGVS/04005/08) issued by the Secretaría de Medio Ambiente y Recursos Naturales. Research protocols in this study comply with IACUC permit #A9–002 SFSU.