Larval habitats of the Anopheles farauti and Anopheles lungae complexes in the Solomon Islands

The study was conducted in Central and Western Provinces in the Solomon Islands where small villages are mainly found in coastal areas. The climate of the region is hot and wet; the median annual rainfall in Central Province is 2837 mm (based on 43 years of data collected from Tulagi; [29]). Annual rainfall estimates for Western Province are 2667 mm from Gizo (based on 28 years of data up to 1952; [29]) and 3725 mm from Munda (based on 11 years of data from 1999–2009; Solomon Islands Bureau of Meteorology, Unpublished data). Although rain falls throughout the year, relatively less rain falls between May to September. The mean annual temperature on the coast is 26 °C and is reasonably constant throughout the year. Mean daily temperatures range between 24 and 30 °C. Malaria is primarily transmitted by An. farauti s.s.

The distribution of anophelines was investigated by larval presence-absence surveys. The surveys were focused within 2 km of the coast line, as this is where villages are located. Larval habitats were categorized as one of six classes and locations recorded by GPS. Surveys were conducted from August to September 2011 in Central Province (9°0′S, 159°45′E) and in February and May 2013 in Western Province (8°0′S, 157°0′E). In Central Province, 14 villages were surveyed on the islands of Ngella Sule and Tulagi Islands. In Western Province, 54 villages were surveyed on the islands of Vella Lavella, Ranonnga, Gizo, Kolombangara, Vonavona, Kohinggo and New Georgia Islands. Potential larval habitats were sampled with 250 ml dippers and larval samples were stored in 70 % ethanol in micro-centrifuge tubes for subsequent identification by molecular analysis of the internal transcribed spacer region II (ITS2) of the ribosomal DNA [30].

The micro-distribution of larval instars at different sites within a large permanent river-mouth lagoon in Haleta village, Central Province, Solomon Islands was evaluated [7]. The lagoon was a known larval habitat that forms when surface water run-off accumulates behind a sand bar that prevents drainage to the sea. After periods of heavy rain, the sandbar breaks and releases the accumulated water to the ocean and the area of the larval habitat is reduced. Hereafter, this larval habitat will be referred to as the lagoon. The numbers of the different anopheline larval instars were monitored daily at five stations (1 × 5 m; Fig. 1) along the northern edge of the lagoon over 10 consecutive days from the 2nd–11th December 2012. Each station was sampled once daily by 10 dips with a standard 250 ml dipper. The number of larvae by instar was recorded per dip. The water temperature and salinity was concurrently measured at each station with a thermometer and a hand held refractometer (Atago Co. Ltd, Japan), respectively.

Density-dependent regulation of mosquito larvae was investigated by seeding first instar An. farauti larval at different densities into floating larval cages encompassing a surface are of 380 cm² (diameter 22 cm; Fig. 2). The larval cages permitted exchange of water and microfora with the habitat in which they were placed but excluded the entry of predators or other fauna. The larval cages tracked the temperature, nutrition and other environmental conditions in the permanent larval habitat in which they were placed.

Wild-caught female An. farauti captured by human landing catches (HLC) (see [1, 7, 31]) were used to generate the F1 first instars used here. Ethical approvals for conducting HLC were obtained from the National Health Research and Ethics Committee, Solomon Islands (02-05-2011), the James Cook University Human Research Ethics Committee, Australia (H4122) and the University Hospitals Case Medical Centre Institutional Review Board for Human Investigation, USA (05-11-11). Blood-fed specimens from HLCs were isolated and placed into 70 ml plastic specimen jars holding a piece of damp cotton-wool covered with filter paper on the bottom as an oviposition substrate. The top of each oviposition chamber was covered with mosquito netting overlaid with damp cotton-wool to ensure high humidity. Filter papers with eggs wer transferred into a petri dish containing rain water for hatching. F1 first instars from wild-caught females were mixed and allocated to larval cages at densities of 10, 25, 50 and 100 per larval cage (i.e. 1 larva per 38, 15.2, 7.6 and 3.8 cm², respectively) with five replicates of each density. For each treatment group the survival of larvae was monitored daily over 10 consecutive days from the 2nd–11th December 2012. Larval survival was defined as total larval present at a time point divided by the total released into the cage.

For the larval distribution study, three related data sources were created: (1) the geo position of each site surveyed (shapefile); (2) the field survey conducted during each site survey (tabular); and (3) the molecular data containing the species identifications of the larval samples (tabular) [32]. The data for each source was linked with a unique identifier that was allocated to each site during the PDA-based survey.

For lagoon micro-productivity, a dataset was constructed that detailed the total number of anophelines per dip, and scored for presence as 0 (negative) or 1 (positive for larvae) for each dip [32]. The influence of location (sampling stations) on both the presence and density of larvae was analysed with generalized linear models (GLMs). The influence of the two environmental factors, water temperature and salinity, with both the presence and density of larvae was analysed with generalized linear mixed models (GLMMs) with location as a subject variable to account for repeated sampling. The distributions for the models were: (1) binary data (presence or absence) fitted to a binomial distribution with a logit link function and (2) count data (density) fitted to a negative binomial distribution with a log link function because data were not normally distributed.

The density-dependent regulation of mosquito larvae was analysed using a Cox regression to compare the survival of mosquito larvae held at different densities [32]. The Cox regression determined the relative risk of dying (hazard ratios) for each density group compared with the lowest density tested (10 larvae per cage [1 larva per 38 cm²]). The model was weighted by the replicate number to account for longitudinal sampling. All analyses were conducted using the R package V3.1.2 [33].

Anopheline larvae were collected from 108 larval habitats (58 sites in Central Province and 50 sites in Western Province) (Figs. 3, 4). Overall 391 specimens of five species were identified by PCR: An. farauti s.s., An. hinesorum, An. lungae, An. nataliae and An. solomonis. Anopheles farauti s.s. and An. hinesorum were the most abundant and widespread species found in both provinces. Anopheles lungae and An. nataliae were present but less common in both provinces. Anopheles solomonis was only found in Central Province. The anophelines were found across a range of larval habitats: coastal lagoons and swamps, drains, transient pools, man-hade holes, riverine and spring wells (Table 1). Both An. farauti s.s. and An. hinesorum were found in all habitat classes, with the highest prevalence habitat being coastal lagoons and swamps. The most commonly used habitat for An. lungae was riverine areas.

A total of 408 anopheline larvae were collected; 56 % (n = 227) were early instars (I–II) and 44 % (n = 181) were late instars (III–IV). The location of the sampling station along the length of the lagoon influenced both the presence (χ² = 13.26, df = 4, p = 0.001) and density (χ² = 22.92, df = 4, p < 0.001) of larvae per dip. The larval density was highest at the monitoring stations most proximal and distal from the sandbar behind which the lagoon formed (stations 1 and 5; Table 2). The water temperature was fairly uniform across the sampling stations, ranging from 30.8 to 31.6 °C, with salinity diminishing from 1 ppt at the station closet to the sandbar to 0 ppt at the site most distal from the sandbar (Table 2). Evidence for an impact of either presence or density of larvae by either water temperature (β = 0.37, df = 43, p = 0.359; β = 0.001, df = 43, p = 0.995, respectively) or salinity (β = −0.11, df = 43, p = 0.904; β = 0.109, df = 43, p = 0.301, respectively) was not found.

The survival of An. farauti larvae was more than twofold lower when larvae were held at the highest experimental density (100 per cage or 1 larva per 3.8 cm²) when compared with the lowest density (10 per cage or 1 larva per 38 cm²; Hazard ratio [HR] = 2.11, se = 0.21, p = 0.0003; Fig. 5). The survival of larvae at densities of 25 (1 larva per 15.2 cm²) and 50 (1 larva per 7.6 cm²) per cage was not significantly different from 10 per cage (25 larvae: HR = 1.24, se = 0.23, p = 0.351; 50 larvae: HR = 1.45, se = 0.21, p = 0.082).

For the lowest experimental density (10 per cage) the surviving larvae (n = 31) all pupated with a cumulative 14 adults emerging by day 9 following larval release into the cages. The development rate was delayed at higher densities. Adults emergence was not seen from any of the other densities with the exception of two undersized (by visual inspection) adults from the highest density cage (holding 100 larvae). Only 8 of the surviving 68 (11.7 %) larvae pupated by day 10 in the 25 larvae per cage experimental density while only 10 of the surviving 131 larvae (7.6 %) pupated in the 100/cage experimental density. Larval and pupal sizes diminished with increasing density based on a visual inspection.