Study areas
This study was conducted in six villages of Kilombero and Ulanga districts in south-eastern Tanzania (Fig. 1). Five of these villages were located at altitudes less than 300 m above sea level, while the sixth was at an altitude greater than 400m. In Kilombero district, the study villages were Ikwambi (-7.97927°S, 36.81630°E), Kisawasawa (-7.89657°S, 36.88058°E) and Mpofu (-8.17220°S, 36.21651°E), while in Ulanga district, the villages were Kilisa (-8.37544°S, 36.57355°E), Ruaha (-8.9063°S, 36.7194°E) and Tulizamoyo (-8.35447°S, 36.70546°E). The study villages were selected based on the high abundance of adult An. funestus mosquitoes based on previous surveillance work done by Ifakara Health Institute (unpublished data). The annual rainfall and temperature ranges in these villages were 1200 - 1800 mm and 20 - 32.6 °C respectively. The main economic activities are crop farming (mostly rice and maize farming) and livestock keeping.
Larvae collection and rearing
This study was done between January and September 2018, and repeated between October and December 2019. The study villages were surveyed for the presence of aquatic habitats along transects of 1000 metres, each radiating from an approximated village centroid. All identified water bodies were marked, geo-referenced, physically characterized and examined for the presence of Anopheles larvae. Standard 350 ml dippers or 10 L plastic buckets were used to sample water from the pools (Fig. 2). When the water bodies consisted of rivers and streams, larval sampling was done along the river length over distances not exceeding 1000 m, so as to match the 1000 m transects in the main survey. Parts of the rivers with or without Anopheles larvae were similarly characterized and geo-referenced. The buckets were used in sites where it was impractical to use the dippers (e.g. habitats with depths greater than 50 cm), and also to collect the larvae for further rearing and identification. The larvae collected from different aquatic sites were transported to the insectary at Ifakara Health Institute for rearing to adults.
Once in the insectary, the larvae were kept in rearing pans (32 cm diameter and 5 L holding capacity) labelled with information on the dates and place of larvae collection. The temperature in the insectary was kept at 26°C ± 2°C and relative humidity at 82% ± 10%. The larvae were fed with Tetramin® fish food until they developed into pupae and emerged into adult mosquitoes. Emerging adult mosquitoes were collected using mouth aspirator, killed by freezing and all Anopheles were identified using morphology-based identification keys developed by Gilles and Coetzee [9,24]. All identified An. funestus mosquitoes were then packed individually in 1.5 ml Eppendorf tubes with silica gel and submitted to molecular laboratory for sibling species identification by polymerase chain reaction (PCR) assays as described by Koekemoer et al. [25]. Habitats positive for An. funestus were then identified among all the surveyed habitats.
Characteristics of aquatic habitats
Characteristics of all the aquatic habitats as well as the surrounding environments were recorded. For the habitats, information collected included water movement (stagnant or slow), water colour (clear, coloured, or polluted), a tree canopy (shade) over habitat (none, partial, heavy), habitat size in circumference (less than 10 m, between 10-100 m, more than 100 m), vegetation type (none, submerged, floating, emergent), vegetation quantity (none, scarce, moderate, abundant), algae quantity (none, scarce, moderate), water depth (less than 10 cm, between 10-50 cm, more than 50 cm), distance from the nearest homes (less than 100m, between 100-500 m, more than 500 m) and water type (semi-permanent, permanent). The habitats were considered temporary, semi-permanent or permanent if retained water for less than 3 months, 3-9 months and throughout the year respectively.
Additionally, the physicochemical characteristics of water in the larval habitats were assessed in four of the six villages, namely Tulizamoyo, Ikwambi, Kisawasawa and Kilisa. Parameters assessed included: water temperature (degree celsius), pH (scale of 0-14), conductivity (Siemens/meter), total dissolved solids (mg/L) and turbidity (nephelometric turbidity units, using 2100Q portable turbidity meter). Assessments of these parameters were conducted in the field sites immediately after the collection of larvae from the habitats. Lastly, nitrate levels (milligrams per litre) were also analysed by spectrophotometric method. To do this, one litre of water samples from each habitat in the study sites was collected, stored in a cooler box and sent to the laboratory at Ifakara Health Institute for analysis within 24 hours post collection.
Data analysis
Analysis was done using open source software, R programming language [26]. A total of 16 environmental variables were used to identify the main predictors for the presence of An. funestus larvae in the study villages. At first, all main predictors were initially assessed individually using univariate logistic regression and assess its impact on the presence of An. funestus larvae. Secondly, all the variables were included in the final model and assess their effect on the presence of An. funestus larvae. Odds ratios and their 95% confidence intervals are reported, and the statistical differences were considered significant when p-values<0.05.