Residual malaria transmission and the role of Anopheles arabiensis and Anopheles melas in central Senegal

Abstract Understanding the behavior and ecology of local malaria vectors is essential for the effectiveness of the commonly used vector-targeted malaria control tools in areas of low malaria transmission. This study was conducted to determine species composition, biting behavior and infectivity of the major Anopheles vectors of Plasmodium falciparum in low transmission settings in central Senegal. Adult mosquitoes were collected using human landing catches during 2 consecutive nights and Pyrethrum Spray Catches in 30–40 randomly selected rooms, from July 2017 to December 2018 in 3 villages. Anopheline mosquitoes were morphologically identified using conventional keys; their reproductive status assessed by ovary dissections, and a sub-sample of Anopheles gambiae s.l. were identified to species level using polymerase chain reaction (PCR). Plasmodium sporozoite infections were detected using real-time quantitative PCR. During this study 3684 Anopheles were collected of which 97% were An. gambiae s.l., 0.6% were Anopheles funestus, and 2.4% were Anopheles pharoensis. Molecular identification of 1,877 An. gambiae s.l. revealed a predominance of Anopheles arabiensis (68.7%), followed by Anopheles melas (28.8%), and Anopheles coluzzii (2.1%). The overall human-biting rate of An. gambiae s.l. was highest in the inland site of Keur Martin with 4.92 bites per person per night, while it was similar in the deltaic site, Diofior (0.51) and the coastal site, Mbine Coly (0.67). Parity rates were similar in An. arabiensis (45%) and An. melas (42%). Sporozoite infections were detected in both An. arabiensis and An. melas with the respective infection rates of 1.39% (N = 8) and 0.41% (N = 1). Results suggest that low residual malaria in central Senegal is transmitted by An. arabiensis and An. melas. Consequently, both vectors will need to be targeted as part of malaria elimination efforts in this area of Senegal.


Introduction
Malaria remains one of the most important parasitic diseases worldwide. According to WHO, 241 million malaria cases and 627,000 deaths were reported worldwide in 2020, representing increases of 6% and 12%, respectively, compared to 2019 (WHO 2021, OMS 2019. Service disruptions during the COVID-19 pandemic are likely to have contributed to these increases, but a stagnation in malaria Malaria transmission is typically more intense where the anopheline species have marked preference for humans and live long enough to allow completion of parasite sporogony development. The lifespan of African anopheline species and their highly anthropophilic behavior are among the reasons that the continent bears the highest malaria burden worldwide (Fontenille and Lochouarn 1999). Across Sub-Saharan Africa, the primary malaria vectors belong to the Anopheles gambiae sensu lato (s.l.) complex and Anopheles funestus s.l. group, targeting of which by vector control interventions is considered crucial for malaria control and elimination. In Senegal, 22 anopheline species have been reported , Coetzee et al. 2000Niang et al. 2014), of which 4 are An. gambiae s.l. (Anopheles arabiensis, Anopheles melas, An. gambiae, and An. coluzzi), which along with An. funestus s.s. are reported as the primary malaria vectors (Vercruysse and Jancloes 1981;Gazin and Robert 1987;Robert et al. 1998Lemasson et al. 1997;Coetzee et al. 2000;Niang et al. 2016;Sy et al. 2018).
In the central-western part of Senegal, the implementation of several control interventions, including seasonal malaria chemoprevention (SMC) from 2008 to 2011 followed by community-based indoor residual spraying with pirimiphos-methyl between 2013 and 2014, and universal LLIN distributions, have contributed to substantially reduce malaria burden in the area (Cissé et al. 2016, WHO/GMP 2011, Sy et al. 2019. However, despite the success recorded at the regional administrative level, a few residual transmission hotspot areas persist where especially suitable micro-environmental conditions for vector development occur (Ndiaye et al. 2020). Recent investigations in central Senegal reported the presence of An. arabiensis, Anopheles coluzzii, and An. melas in the hotspots surveyed (Sy et al. 2018), with local populations displaying resistance to several public health insecticides (Sy et al. 2021).
This longitudinal entomological study took place between July 2017 and December 2018 in 3 mains ecological (coastal, deltaic, and inland) settings in central Senegal to clarify the potential role and relative contributions of local populations of An. arabiensis and An. melas in maintaining malaria transmission in local hotspots.

Study area
This study was conducted in 3 sites located, respectively, in the coastal, deltaic, and inland (continental) areas in the center-west of Senegal spanning the administrative departments of Mbour and Fatick. The village of Mbind Coly (14°17ʹ10″N; 16°54ʹ30″W) is located in the coastal estuary of Mbour department and is surrounded by a mangrove swamp fed by rainwater and tides. The village of Diofior (13°58ʹ41″N; 16°45ʹ45″W) is located in the deltaic area where the Saloum River flows into the Atlantic Ocean. The village of Keur Martin (14°24ʹ29″N; 16°34ʹ29″W) is located in the western mainland in the department of Fatick and is characterized by the presence of saline soils (Fig. 1). Detailed description of the 3 study sites and areas are reported in Ndiaye et al. (2020) and in Sy et al. (2018).

Mosquito sampling and processing
The dynamics of malaria vector populations were monitored in each of the study villages using overnight human landing catches (HLC) and Pyrethrum Spray Catches (PSC) in the morning. HLC were performed between 8 PM and 6 AM for 2 consecutive nights in 3 randomly selected houses, where 2 collectors catch landing host seeking, 1 indoor, and another outdoor distant at least 10 m apart from each another. PSC were performed early in the morning, after the second overnight HLC, at least 30 up to 40 randomly selected rooms per village with 1 room per house. Collections were carried out in July, September, October, and December 2017; then in April, May, July, August, September, October, and December 2018. The use of long-lasting insecticide-treated nets (LLINs) by people in the PSC-selected rooms was also reported in a survey form to determine LLINs coverage among the sampled rooms.

Field and laboratory processing
Collected mosquitoes were morphologically identified to genus level, and Anopheles were subsequently identified to species level using morphological keys (Gillies and Coetzee 1987;Gillies and De Meillon 1968). All mosquito specimens were individually stored in numbered 1.5-ml Eppendorf tubes containing silicagel until laboratory processing. For each collection, 30 randomly sampled female specimens of An. gambiae s.l. caught using HLC were dissected to determine the parity rate. The heads and thoraxes of host-seeking females were screened to detect Plasmodium falciparum infection using the qPCR TaqMan Assay Method described by Bass et al. (2008). Anopheles gambiae s.l. sibling species were discriminated by the PCR method described by Wilkins et al. (2006).

Measured parameters
The human-biting rate (HBR) was calculated as the ratio of the total number of females of each vector species collected by HLC to the total person-nights for a given collection period. The parity rate was estimated as the proportion parous from the total number of specimens dissected. The indoor resting density was defined as the number of mosquitoes per room collected by PSC. The sporozoite rate was calculated as the proportion of the total number of mosquitoes infected with P. falciparum. The entomological inoculation rate (EIR) was calculated as the product of the HBR and the sporozoite rate. LLIN coverage was calculated as the ratio of the number of rooms with nets used to the number of rooms visited during the PSC collection.

Statistical analysis
All parameters measured were computed and analyzed using the RStudio and SPSS 26 Statistics software. Data were compared with the Pearson chi-square or Fisher exact tests, Spearman correlation, Kruskal-Wallis test, Friedman ANOVA (nonparametric), or a binomial generalized linear model as applicable, with the statistical significance threshold in all cases set at P value ≤0.05.

Resting and biting behaviors of An. gambiae s.l. populations
Overall, the resting densities (from PSC) of An. gambiae s.l. were at least 2 times higher in the inland area of Keur Martin than in both the deltaic area of Diofior and the coastal area of Mbine Coly. Indoor Resting Density (IRD) of An. gambiae s.l. was significantly lower in Diofior than in Mbine Coly and Keur Martin (Friedman chisq = 11.29, P = 0.004). IRD varied seasonally and was the highest in September and October, which coincided with the end of the rainy season in all study sites over the 2 years the study was conducted. Thus, in Keur Martin a peak of 5.6 and 7.87 females/room were recorded in September 2017 and September 2018, respectively. In Diofior and Mbine Coly the peaks of IRD were observed in September and October in both years ( Table 2).
The endophagic rates of An. gambiae s.l. females were comparable in all the studied sites with 49% in both Diofior and Keur Martin and 45% in Mbine Coly (Table 3). The human-biting rate of An. gambiae s.l. was the highest in September and October during each year of study. During these 2 months (September and October) no significant difference was noted in HBR among sites, with 4.92 bites per person per night (b/p/n) in the mainland area of Keur Martin, 0.51 b/p/n in Diofior and 0.67 b/p/n in Mbine coly (Friedman chisq = 3.93; P = 0.14). Similarly, no significant difference was found in indoor biting proportion (Friedman chisq = 0.80; P = 0.67) (Table 3). Overall, during the 2 years of survey, the endophagic rates of both An. arabiensis and An. melas varied in the same way in the 3 study sites with peaks observed in September and October. From January to June, corresponding to the dry season, only An. arabiensis showed endophagic rates of 100% (April 2018) (Fig. 2).    Among the study sites, significant difference was observed in LLIN coverage with 70% in Mbine Coly, 91% in Keur Martin, and 95% in Diofior (chi-squared test, X-squared = 6.24, 2 degrees of freedom, P = 0.044) ( Table 4).
Parity rates of An. gambiae s.l. females The parity rate of An. gambiae s.l. females were 38.9% in the inland area of Keur martin, 56.6% in the coastal zone of Mbine Coly, and 59.6% in the deltaic zone of Diofior. Using a generalized linear binomial model initially including collection month, location, species (An. arabiensis or An. melas), and a location × species interaction term, only location was significant in the minimal model. The parity rate was significantly lower in Keur Martin than in the reference location Mbine Coly (P = 0.023), but Diofior was not significantly different (P = 0.30) ( Table 5). In all the study sites, An. arabiensis displayed the highest parity rate compared to An. melas except in the coastal site of Mbine coly where the GLM-species model revealed no significant difference in the parity rate between the 2 species (chisquared = 5.54; df = 2, P = 0.063) ( Table 6).

Sporozoite infection and EIR
The screening of Plasmodium infection suggests that An. arabiensis and An. melas are involved in the residual malaria transmission, with the respective infection rates of 1.39% (N = 8) and 0.41% (N = 1). No significant difference was observed between species (Fisher exact P = 0.29) nor between study sites with species pooled (Fisher exact P = 0.87). Infected An. arabiensis females were found in Keur Martin (inland area) and Diofior (deltaic area) with the respective      Total  26  69  23  48  134  42  71  37  41  24  19  74  201  42  113  45 infection rates of 1.52% and 1.85%. Anopheles melas was found infected only in the inland area (0.54%) ( Table 7). No infection was found in the coastal area for both species. Overall, in the study area, despite the presence of An. coluzzii and An. gambiae, only An. arabiensis and An. melas seem to play a role in malaria transmission, being the only species found infected with respectively EIR of 0.018 and 0.002 ib/p/n, corresponding to an annual infected bites of 6.57 ib/p/year for An. arabiensis and 0.73 ib/p/ year for An. melas. Both species were found infected in the same locality at the same time only in the inland area of Keur Martin (Table  7). No Plasmodium infection was found in the coastal area of Mbine Coly, whereas in the deltaic area of Diofior, only An. arabiensis was found infected with an EIR of 0.0055 ib/p/n (or 2 ib/p/year [ Fig. 3]).

Discussion
This study was conducted over 2 consecutive years in 2017 and 2018 and confirmed the predominance of An. gambiae s.l. representing 97% of the anopeline fauna with minor presence of other species, An. funestus and An. pharoensis. Indeed, the An. gambiae s.l. complex contains the major malaria vectors among the 22 species described nationwide , Coetzee et al. 2000. The use of molecular approaches on morphologically identified An. gambiae s.l. revealed the presence of 4 members of the gambiae complex: An. arabiensis, An. melas, An. gambiae, and An. coluzzi, and the presence of gambiae/coluzzii hybrids. In the study area, An. arabiensis and An. melas were the 2 most abundant species of the gambiae complex. Anopheles arabiensis is known as a predominant malaria vector over all the country specially in the dried and arid environments, except the south and south-east part of Senegal (Dukeen and Omer 1986, Faye et al. 1995, Lemasson et al. 1997. The presence of An. melas was reported in brackish water resulting from the mixture of sea water with the freshwater of the Saloum River in the deltaic area where it flows into the ocean, known to be provide preferred breeding sites (Bryan et al. 1987). The IRD of An. gambiae s.l. was relatively more important in the inland area of Keur Martin than in the 2 other areas (deltaic and coastal areas). In contrast to the other study sites, Keur Martin village is characterized by the presence of hydromorphic and halomorphic soils, which in addition to their capacity to retain water over a long period of time, are characterized by the presence of surface brackish water bodies, suitable for the development of An. arabiensis and An. melas depending on the level of salinity (Ndiaye et al. 2020). In all the sites, the exophagy rate of An. gambiae s.l females have been relatively important, this situation could be explained by the predominance of An. arabiensis, accounting for 68.72% of the anopheline fauna in the area and known for its behavioral plasticity, being more exophilic in some areas compared to its sibling (Mahande et al. 2007). Furthermore, the high coverage of LLINs, following the LLIN mass campaign implemented by the PNLP in 2016 across the study area could suggest additional pressure and which could lead to an induced exophilic behavior due to the repellent effect of pyrethroid, the class of insecticide used for net impregnation, as previously shown in an experimental study (Darriet et al. 2002, PNLP 2016. The parity rate of An. gambiae s.l. was the lowest in the inland site of Keur martin (38.9%) and was comparable with the rates (25-33%) previously reported from the mangrove area in the Gambia (Bryan et al. 1987) and from Casamance (47%) . The high level of pyrethroid-impregnated LLINs coverage in the studied houses of Keur Martin (>90%) could explain the reduced longevity of the studied populations of An. gambiae s.l.  Only An. arabiensis and An. melas were found infected in the study area, this finding is in line with recent studies carried out in the central Senegal (Sy et al. 2018).
Generally, human populations are at high risk of being infected by malaria local vector species, especially during the second half of the night when they were not effectively using the LLINs. LLINs mass distribution campaign and IRS are the 2 main malaria vector control strategies in Senegal and sub-Saharan Africa generally (Dione 2014, Sy et al. 2019, PMI 2020 with the main goals to reduced host-vector contact and vector population longevity. Several authors have reported the possible involvement of An. arabiensis, An. melas, and An. coluzzii species, possibly playing a main role in malaria transmission in the study area here (Hamon et al. 1963, Mahande et al. 2007, Sy et al. 2018. The absence of infected females of An. coluzzii during the current study could be explained by the low sample representativity of this species during the study period. The average EIR for An. arabiensis and An. melas almost 7 time higher for the first species than the latter without significant difference due to the low numbers of An. melas, being nevertheless lower than reported previously in same the area with 13.14 infective bites per person per year (Sy et al. 2018). This low level of transmission as observed here is probably related to a variety of malaria control interventions implemented by the NMCP in this area. Indeed, a SMC in children under 10 years of age was implemented in the area between 2008 and 2011, followed by 2 targeted indoor residual spraying campaigns with pirimiphos-methyl in 2013and 2014(Cissé et al. 2016, Sy et al. 2019.

Conclusion
This study demonstrated the predominance and the involvement of both An. arabienis and An. melas in maintaining malaria in hotspot found in an overall low malaria transmission in the central Senegal. The findings were generated on infection using a sensitive qPCR approach, which allows the detection of low plasmodial infection in mosquitoes, suitable for areas of low malaria transmission. The data generated will be shared with the PNLP and will certainly allow to better tailor the monitoring of malaria in this area, eligible for malaria elimination. Subsequent investigations on insecticide phenotypic resistance and intensity for these 2 species in the area, are needed to better complement evidence to better select and target vector control strategies to drive toward the goal of malaria elimination as aimed by the NMCP in eligible areas.