Anopheline bionomics, insecticide resistance and transnational dispersion in the context of controlling a possible recurrence of malaria transmission in Jaffna city in northern Sri Lanka

Background Malaria was eliminated from Sri Lanka in 2013. However, the influx of infected travelers and the presence of potent anopheline vectors can re-initiate transmission in Jaffna city, which is separated by a narrow strait from the malaria-endemic Indian state of Tamil Nadu. Methods Anopheline larvae were collected from different habitats in Jaffna city and the susceptibility of emergent adults to DDT, malathion and deltamethrin investigated. Results Anopheline larvae were found in wells, surface-exposed drains, ponds, water puddles and water storage tanks, with many containing polluted, alkaline and brackish water. Anopheles culicifacies, An. subpictus, An. stephensi and An. varuna were identified in the collections. Adults of the four anopheline species were resistant to DDT. Anopheles subpictus and An. stephensi were resistant while An. culicifacies and An. varuna were possibly resistant to deltamethrin. Anopheles stephensi was resistant, An. subpictus possibly resistant while An. varuna and An. culicifacies were susceptible to malathion. DNA sequencing showed a L1014F (TTA to TTC) mutation in the IIS6 transmembrane segment of the voltage-gated sodium channel protein in deltamethrin-resistant An. subpictus—a mutation previously observed in India but not Sri Lanka. Conclusion Anopheles subpictus in Jaffna, like An. stephensi, may have recently originated in coastal Tamil Nadu. Besides infected overseas travelers, wind- and boat-borne carriage of Plasmodium-infected anophelines across the Palk Strait can potentially reintroduce malaria transmission to Jaffna city. Adaptation to diverse larval habitats and resistance to common insecticides in anophelines are identified as potential problems for vector control should this happen.


Background
Malaria remains a global health problem with an estimated 228 million cases and 405,000 deaths worldwide in 2018 [1]. Although the global malaria burden has shown a decreasing trend, and a number of countries have eliminated malaria or are approaching elimination, increasing resistance of vector populations to insecticides in many countries is a cause for concern [1]. Malaria, which had been endemic in Sri Lanka for centuries, was eliminated from the island in 2013. However, the approximately 100 travelers with malaria who arrive every year from endemic countries, and the prevalence of potent anopheline vectors throughout the island, pose the risk of reinitiating local malaria transmission [2]. Indeed, an instance of local malaria transmission from a malaria-infected Indian national to a Sri Lankan resident was documented in late 2018 [3].
The principal malaria vector Anopheles culicifacies exists as a species complex in Sri Lanka [4]. Among the two sibling species B and E of the Culicifacies complex, species E was previously considered to be the major vector of malaria in the island [5]. Molecular characterization of An. subpictus complex revealed the presence of species A and the morphologically similar An. sundaicus (previously classified as An. subpictus species B) in Sri Lanka [6]. Anopheles subpictus species A and An. sundaicus along with An. annularis functioned as secondary vectors in Sri Lanka together with several minor vectors including An. varuna [7][8][9][10][11][12]. Anopheles stephensi (sensu stricto) (type-form), the potent urban vector of malaria in India, was first detected in Sri Lanka in 2017 [13,14]. The expansion of its range to Sri Lanka occurred after its spread to South India as well as Djibouti, Ethiopia and the Sudan in Northeast Africa [15,16]. This range expansion was attributed to human migration [15] as well as the adaptation of An. stephensi to develop in cement wells and water storage tanks associated with human dwellings, a process exemplifying what was previously termed as the anthropogenically-induced adaptation to invade in mosquito vectors [17].
The coastal city of Jaffna (population ~ 100,000, area 10.5 km 2 ) is located in the Jaffna Peninsula (population ~ 600,000, area 1130 km 2 ) and is the largest city in the Northern Province of Sri Lanka. The Jaffna Peninsula is separated by the 64-137 km wide Palk Strait from Tamil Nadu state in India (Fig. 1a, b). Numerous small inhabited and uninhabited islands lie between the peninsula and Tamil Nadu. Malaria remains endemic in proximal coastal areas of Tamil Nadu [18,19]. The Jaffna Peninsula is located within the dry zone of Sri Lanka, has a limestone geology and is undergoing rapid salinization of its ground water resources [20,21]. The salinization is caused by unsustainable rates of water extraction from freshwater aquifers and is exacerbated by rising sea levels. Adaptation of freshwater disease vectors to salinity in coastal areas heightens the potential for transmitting vector-borne diseases as previously described [22]. Malaria control efforts were severely hampered in the Jaffna Peninsula during the civil war of 1983-2009 and malaria vector surveys have not been performed lately in Jaffna city. Anopheles species identified recently in the Jaffna peninsula and associated islands include An. varuna [23,24], An. culicifacies [23,24], An. barbirostris [23,25], An. subpictus [6,23,26], An. sundaicus [6,12,15,26,27] as well as An. stephensi [14,15]. Anopheles stephensi, the dominant urban malaria vector in India, was first detected in Mannar island in the Northern Province of Sri Lanka in 2017 ( Fig. 1a) [13][14][15] and subsequently in domestic wells and water storage tanks in Jaffna city in 2018, with a genotype consistent with an origin in Tamil Nadu [15]. Indigenous malaria transmission re-emerging in Jaffna city and the Jaffna Peninsula remains possible because of the proximity to malaria-endemic areas of Tamil Nadu. There is presently no form of authorized sea travel between Jaffna and India but the movement of combatants, refugees and fishermen between Jaffna and Tamil Nadu was common during the 1983-2009 civil war. Characterization of anopheline larvae and their habitats in the Jaffna city and the susceptibility of adults emerging from the collected larvae to common insecticides was undertaken to assess the potential for malaria transmission re-establishing itself in the city.

Larval collection and rearing
Larval sampling was done and water quality in larval habitats determined fortnightly, at four locations in Jaffna city (Fig. 1c) during the period April 2018 to August 2019. Samples were collected from Kurunagar (9°39′17.5″N, 80°127.05″E), Navanthurai (9°40′16.3″N, 80°0′29.6″E), Aryalai (9°39′39.9″N, 80°3'1.5″E) and Nallur (9°39′59.99″N, 80°01′60.00″E). Anopheline larvae were collected from deep wells using string-connected conical drop nets with 15 cm diameter and 10 cm depth. Standard 350 ml dippers were used to collect larvae from puddles, ponds, open surface drains, water storage tanks and shallow wells ( Fig. 2) with an average of 10 dips per habitat. The pH, salinity, dissolved oxygen (DO), total dissolved solids (TDS), and electrical conductivity in water collections with anopheline larvae were measured with a Multi Parameter Probe (Hanna Instruments, Leighton Buzzard, UK). Larval density for each species was calculated as the number of larvae per liter for every collection site. Habitats where anopheline larvae were not found were not analyzed for water characteristics or statistical studies.
Collected larvae were reared as described previously in the insectary of the Department of Zoology, University of Jaffna, Jaffna, Sri Lanka, to reach adulthood [26] in water from their respective habitats. Emergent adults were identified morphologically using published keys [28,29] and used for insecticide bioassays when 3-5 days-old.

Insecticide susceptibility tests
Insecticide susceptibility tests were performed according to the standard WHO protocol for adult mosquitoes [30] as previously described [31]. Adults of each species from all collection sites were pooled and exposed to insecticide impregnated papers of 0.05% (w/v) deltamethrin, 5% (w/v) malathion and 4% (w/v) DDT using WHO bioassay kits and procedure [30]. Mosquitoes were exposed for 1 h and the mortalities determined after a 24 h recovery period. Mosquitoes were exposed to each insecticide in two or three replicates in two separate experiments. Control mosquitoes were exposed to papers impregnated with carrier oil alone in each experiment. Mortality was determined and adjusted using Abbot's formula if the control mortalities were between 5-20% [30,32]. The WHO criteria were used to define a population as susceptible (mortality of > 98%), possibly resistant (mortality of 90-97%) and resistant (mortality of < 90%) [30]. Mosquitoes that survived exposure to deltamethrin in the bioassay were frozen individually at − 20 °C to screen for knock-down resistance (kdr) mutations in the voltagegated sodium channel protein (VGSC).

DNA sequencing to differentiate An. subpictus and An. sundaicus
One hundred and fifty adult An. subpictus comprising approximately 50 mosquitoes that were tested for each insecticide were randomly selected for differentiation from the morphologically similar An. sundaicus by the previously described diagnostic allele-specific PCR [6].

DNA sequencing of the voltage-gated sodium channel protein gene
Twenty each of deltamethrin-resistant An. subpictus and An. stephensi were used to PCR amplify the IIS6 transmembrane coding segment of the VGSC gene using the primers KdrF (5′-GGA CCA YGA TTT GCC AAG ATG-3′) and KdrR (5′-CGA AAT TGG ACA AAA GCA AAG-3′ [33]. The PCR products were purified using QIAquick ® PCR Purification Kit (Qiagen, California, USA) and sequenced in both directions at Macrogen, Seoul, South Korea. The sequences were edited in Finch TV (Geospiza, Seattle, USA) and aligned with Clustal W in MEGA 5.0 software [34].

Statistical analysis
A one-way ANOVA was performed to determine associations between the different types of habitats where anopheline larvae were found and each of the measured water parameters separately. Since the sample size was small for each species type of habitat data, and similar habitats where larvae were absent were excluded from the study, the species-type of habitat interaction was not statistically analysed. The influence of water characteristics on the density of each species of anopheline larvae in different habitats where the larvae were present was analysed using generalised linear mixed model (GLMM) with habitats as a unique identifier. The larval densities were fitted to a negative binomial distribution. The total model was represented as, physical parameters / larvae density = species/solution (effect of water parameters). The logistic regression model was used in the analysis. From the effect of water parameters on overall species obtained from the total model, least square means were compared to understand the effect on each species. The value of P < 0.05 was considered statistically significant. Statistical analyses were performed using the SAS University Edition (SAS Institute Inc., NC, USA).

Anopheline species and water habitat characteristics in preimaginal habitats
A total of 2369 larvae of four Anopheles species were collected from five different types of habitats ( water storage tanks (n = 10), open surface drains (n = 9), puddles (n = 3) and ponds (n = 9), in all four locations but An. stephensi was only collected in wells (n = 14) and water storage tanks (n = 2), as shown in Table 1 in three out of four locations: Gurunagar (wells and water storage tank); Navanthurai (wells); and Nallur (wells and water storage tank). Larvae of An. culicifacies, An. subpictus and An. varuna were often found to occur together in the same habitat. However, other anopheline species were not found in the limited number of water tanks and wells where An. stephensi larvae were present. Water parameters in the different habitats where larvae were collected were quite variable ( Table 2, Fig. 2). ANOVA revealed no significant differences in the mean pH of the habitats where anopheline larvae were collected (F ( It was apparent that all four anopheline species could oviposit and undergo preimaginal development to adults in relatively alkaline, brackish and polluted water in Jaffna city. The GLMM analysis showed that larval densities of each of the four anopheline species were significantly (t ranging from -4.38 to -7.67; P = 0.001 to 0.008) and negatively associated with salinity but not significantly (P > 0.05) associated with pH, TDS, DO or conductivity of the habitats where the respective species of anopheline larvae were present (Additional file 1: Table S1)

Adult insecticide susceptibility bioassays
The results of the insecticide susceptibility bioassays according to the WHO criteria are presented in Table 3 (see also Additional file 2:

Sequence analysis of the VGSC gene
We sequenced the IIS6 coding domain of the VGSC gene in An. subpictus and An. stephensi that were resistant to deltamethrin. Fifteen sequences, of approximately 250 bp in size, of each species were analyzed. This revealed the presence of the same A to C transversion leading to Table 2 Characteristics of water habitats associated with the different species of anopheline larvae Notes: The results show the range, mean and standard deviation of the mean (sd) of measured parameters. Freshwater and brackish water are defined as containing < 0.5 and 0.5-30 ppt salinity respectively [38,39]. Water with ≥ 400 ppm TDS, ≤ 4.5 ppm DO and ≥ 800 μS/cm conductivity are indicative of pollution [58] Abbreviations: TDS, total dissolved solids; DO, dissolved oxygen; ppt, parts per thousand; ppm, parts per million; µS/cm, micro Siemens per centimetre; SD, standard deviation

Discussion
Anopheles culicifacies had been widely held to only undergo preimaginal development in unpolluted freshwater habitats in rural Sri Lanka [9,35]. However, recent studies have shown that it is able to develop in brackish water of 2-4 ppt salinity in the Jaffna Peninsula and its offshore Island Delft [23] and in the Eastern Province [36]. Anopheles culicifacies larvae have also been found in polluted drains with low DO content in the Eastern Province [37]. Our study suggests that it can undergo preimaginal development to adulthood in polluted water with high TDS (> 400 ppm) and low DO (≤ 4.5 ppm) and relatively alkaline water in Jaffna city. The ability to tolerate brackish, polluted and alkaline water in Jaffna city was also observed with An. subpictus (identified here as being mostly if not exclusively as An. subpictus species A and distinct from the closely-related, salinity-tolerant An. sundaicus) which had also been previously regarded as developing only in fresh and unpolluted water [38]. Although we had previously observed An. stephensi to tolerate brackish water of up to 3.5 ppt salinity in Jaffna city [15] which is consistent with the present findings, we now show that it can also develop in polluted and alkaline water. Because of its likely recent arrival from coastal Tamil Nadu, it is possible that tolerance to brackish and polluted water in An. stephensi are characteristics developed in Tamil Nadu or elsewhere in India that have helped it expand its range to Jaffna in an example of the anthropogenically-induced adaptation to invade in mosquito vectors [15,17]. We also found that An. varuna, previously regarded as another freshwater mosquito [35], is capable of developing in brackish water of up to 1.4 ppt salinity as well as polluted and alkaline water in Jaffna city. Ground water is relatively alkaline throughout the Jaffna Peninsula because of its limestone geology [21]. Ground water is becoming increasingly brackish in the Jaffna Peninsula, due to an unsustainable depletion of its freshwater limestone aquifers and influx of sea water [21,39]. These factors may have promoted adaptation to alkalinity and salinity in the four anophelines in Jaffna city. Rising sea levels exacerbating salinization of ground water in the peninsula [21,22] can further increase salinity tolerance in mosquito vectors in the future.
However, the statistical analysis of the relationship between larval density and habitat salinity also suggest that the four anopheline species in Jaffna city have still retained a preference for habitats of lower salinity for laying eggs and undergoing preimaginal development. Preimaginal development in brackish water habitats in Jaffna city and elsewhere in the Jaffna Peninsula was also recently documented in the typical freshwater arboviral vectors Aedes aegypti and Ae. albopictus [23,40]. Salinity adaptation in Ae. aegypti has been shown to involve heritable and non-heritable biological changes [41,42]. The ability of Ae. aegypti to develop in polluted water in surface drains in Jaffna city is associated with greater resistance to pyrethroids [43]. Elucidating biological changes in anophelines associated with similar adaptation to salinity and pollution in Jaffna city can further help understand the adaptability of different mosquitoes to anthropogenic environmental changes. Furthermore, our results suggest that mosquito larval control measures in the city that are presently directed almost exclusively towards fresh water habitats, need to be extended to brackish and polluted water habitats. This is presently important for Ae. aegypti and Ae. albopictus because dengue and chikungunya are endemic in Jaffna [44] but can become essential also for anopheline vector control should malaria transmission recur in Jaffna.
The four anopheline species collected in Jaffna city were resistant to DDT despite the cessation of DDT use for indoor residual spraying (IRS) in Jaffna in early 1970s. Resistance to DDT declined slowly after cessation of use, but increased again after 1983 due to a glutathione-S transferase-based mechanism, postulated to be initially caused by DDT and subsequently by exposure to the malathion used for IRS and other organophosphates used for controlling agricultural pests [2,27,45,46]. The results confirm that An. stephensi in Jaffna is resistant to all three insecticides tested consistent with our recent previous report [15]. Anopheles culicifacies and An. varuna were susceptible to malathion but both, together with An. subpictus, showed varying levels of resistance to deltamethrin. Resistance to deltamethrin in Ae. aegypti developing in polluted drain water in Jaffna city was suggested to be caused by adaptation to cope with organic pollutants in drain water and possibly the common use of pyrethroids to control adult mosquitoes and other insect pests in Jaffna [43]. The same factors may also be responsible for deltamethrin resistance in Jaffna city anophelines.
Pyrethroids and DDT target the VGSC to alter ion transport in arthropod nerve cell membranes [47,48]. This leads to paralysis and eventual death of the insect. Mutations in VGSC, prominently L1014F, confer knockdown resistance (kdr) against pyrethroids and DDT in many anopheline species worldwide [47]. In An. subpictus, an L1014F mutation is reported to arise from an A → T transversion in the North Central Province of Sri Lanka [49] and India [50] and an A → C transversion only in India [50]. Only one of 25 etofenprox resistant An. subpictus carried the A → T transversion and none carried the A → C transversion in Sri Lanka [49]. In contrast the frequencies of A → T and A → C transversions were 0.19 and 0.67 respectively in deltamethrin-resistant Indian An. subpictus [50]. The A→C transversion that leads to the L1014F was found in Indian An. subpictus that had 100% genetic similarity with Sri Lankan An. subpictus species A in the mitochondrial cytochrome oxidase I gene [50]. Our results show that an A → C transversion that leads to the L1014F amino acid substitution is found in all 15 An. subpictus sequenced in Jaffna city. To our knowledge, this is the first report of an A → C transversion leading to an L1014F mutation in An. subpictus sibling species A in Sri Lanka, and it raises the possibility that deltamethrin-resistant An. subpictus in Jaffna city, like An. stephensi [15], could have originated in India rather than Sri Lanka.
Although the recently arrived An. stephensi in Jaffna was resistant to DDT and deltamethrin [15], we now show that there are no mutations in the sequenced IIS6 segment of the VGSC protein in this mosquito. Resistance may, therefore, be caused by other mechanisms such as increased activities of detoxifying enzymes [27,45,46] and this merits further investigation.
Mosquito vectors have been regarded as migrating at most a few kilometers after taking a blood meal [51,52] and therefore Plasmodium-infected mosquitoes were thought incapable of traversing the Palk Strait separating the Jaffna Peninsula from coastal Tamil Nadu (Fig. 1). However, a recent study in the Sahel region of Africa provided evidence that bloodfed anopheline vectors can disperse > 100 km aided by winds of < 10 m/s [53]. The wind action over the Palk Bay, Bay of Bengal, India and Sri Lanka is mainly influenced by two monsoons, the Southwest (May to July) and Northeast (October to January). The average wind speed during the Southwest monsoon is ~5-7 m/s and during this monsoon the wind has a strong component blowing from coastal Tamil Nadu towards the Jaffna Peninsula (arrowed in Fig. 1a) [54]. Data from the Conformal Cubic Atmospheric Model (CCAM) simulation over Sri Lanka [55] predicts that some wind during the Southwest monsoon period may be deflected in a more easterly direction towards Jaffna because of the Western Ghats mountain range (Fig 1a). We therefore propose that wind-borne migration of anophelines is possible from coastal areas of Tamil Nadu to the Jaffna Peninsula and Mannar Island, and this might explain the genetic characteristics of An. subpictus and An. stephensi in Jaffna city. However, genetic characterization at multiple genetic loci in An. subpictus from Jaffna city and many different locations in Tamil Nadu and elsewhere in Sri Lanka are needed to establish this hypothesis for An. subpictus. We had previously suggested that An. stephensi from India may have arrived as eggs/larvae in water collections in boats used by fishermen and combatants during the 1983-2009 civil war [15]. Arrival as eggs/larvae in boats is a possibility for both An. subpictus and An. stephensi, but wind-borne and boat-borne migration of adult female anophelines are important new possibilities for consideration. The wind-aided arrival of female anophelines that have taken a Plasmodium-infected blood meal in Tamil Nadu could readily re-initiate malaria transmission in locations such as Jaffna city because of the high population density and the presence of potent local vectors. Wind-borne movement of Plasmodiuminfected anophelines across national borders is therefore an additional factor for consideration in malaria control and elimination efforts in other parts of the world [2,15] including islands [39].
Currently, the regional Anti-malaria Campaign of the Jaffna District applies temophos to unused wells and introduces larvivorous fish to domestic wells to control preimaginal development of anophelines in Jaffna city. We have previously shown that the larvivorous fish Oreochromis mossambicus readily feeds on An. subpictus [56], and the introduction of Poecilia reticulata to domestic wells and water tanks is effective in controlling An. stephensi in Jaffna [15]. In the face of widespread insecticide resistance, the introduction of larvivorous fish into wells and water storage tanks may contribute to controlling anopheline populations in Jaffna city.
A limitation of our study is that the An. culicifacies collected in Jaffna city was not identified as being species B or species E due to the difficulty in karyotyping and the lack of differentiating DNA markers [57]. More extensive sampling of habitats where larvae are present or absent in Jaffna city is also needed for firmly associating the types of habitats and water characteristics with species abundance. Both entailed a larger project than was possible in the present study.

Conclusions
Vector control plays an important role in limiting the spread of malaria. Our findings highlight resistance to common insecticides and the adaptation of anopheline