Survival rate, blood feeding habits and sibling species composition of Aedes simpsoni complex: Implications for arbovirus transmission risk in East Africa

Aedes simpsoni complex has a wide distribution in Africa and comprises at least three described sub-species including the yellow fever virus (YFV) vector Ae. bromeliae. To date, the distribution and relative contributions of the sub-species and/or subpopulations including bionomic characteristics in relation to YF transmission dynamics remain poorly studied. In this study conducted in two areas with divergent ecosystems: peri-urban (coastal Rabai) and rural (Rift Valley Kerio Valley) in Kenya, survival rate was estimated by parity in Ae. simpsoni s.l. mosquitoes sampled using CO2-baited BG Sentinel traps. We then applied PCR targeting the nuclear internal transcribed spacer 2 (ITS2), region followed by sequencing and phylogenetic analytics to identify the sibling species in the Ae. simpsoni complex among parous and blood fed cohorts. Our results show that Ae. bromeliae was the most dominant sub-species in both areas, exhibiting high survival rates, human blood-feeding, and potentially, high vectorial capacity for pathogen transmission. We document for the first time the presence of Ae. lilii in Kenya and potentially yet-to-be described species in the complex displaying human feeding tendencies. We also infer a wide host feeding range on rodents, reptile, and domestic livestock besides humans especially for Ae. bromeliae. This feeding trend could likely expose humans to various zoonotic pathogens. Taken together, we highlight the utility of genotype-based analyses to generate precision surveillance data of vector populations for enhanced disease risk prediction and to guide cost-effective interventions (e.g. YF vaccinations).

Introduction Yellow fever (YF) is a re-emerging arboviral threat in Africa and South America despite availability of an effective vaccine to protect humans. This is unparalleled in eastern Africa exemplified by increased frequency and magnitude of outbreaks recorded recently in Sudan (2012), South Sudan (2020), Uganda (2011,2016,2019,2020), and Ethiopia (2012)(2013)(2014)2018; 2020) [1,2]. Imported cases into Kenya following recent outbreaks in Angola and the Democratic Republic of Congo (DRC) (2015-2016) [2] demonstrate the potential for continued YF spread in the region. Improved understanding of the drivers of YF virus (YFV) transmission are urgently required through risk assessment to inform cost-effective preventive strategies.
Of the recognized YFV transmission cycles (sylvatic, rural, and urban cycles), outbreaks in the eastern African region have been described as sylvatic [3]. This is inclusive of the last documented YF outbreak in Kenya (1992-95) that implicated sylvatic Aedes vectors including Aedes simpsoni s.l. [4]. However, YFV transmission trends appear to be changing with demographic and environmental changes. For instance, the 2015/16 outbreaks in Angola and DRC were urban [2] and whether the East Africa region may face same fate in the future remains uncertain. The abundance of Ae. simpsoni s.l. vectors in peri-domestic human environments and associated Stegomyia indices have raised concerns about potential urban/rural risk of YFV transmission in Kenya [5]. Aedes simpsoni s.l. thus, has the potential to serve as bridge vector, moving the YFV from the sylvatic/rural to the urban transmission cycle. Aedes simpsoni complex comprises at least three described species including; Ae. lilii, Ae. simpsoni s.s., and the YF vector Ae. bromeliae [6]. To date, the distribution and relative contributions of the sub-species and/or subpopulations including bionomic characteristics in relation to YFV transmission dynamics remain poorly studied.
Demographic and societal changes including unplanned urbanization as occurring in Africa [7], are among important drivers of vector-borne disease spread. The consequent changes in urban architecture, via effects on environment, human abundance, vector density, and biting behavior have the potential to alter the vectorial capacity of mosquitoes and diseases transmission risk. We here, report data on survival and host feeding patterns of Ae. simpsoni mosquitoes in two areas of contrasting ecosystems: peri-urban, coastal Rabai, and rural, Rift Valley Kerio Valley, in Kenya. These parameters provide opportunities for vector infection and efficient transmission and are the two most sensitive indicators of disease transmission potential (i.e. vectorial capacity) [8,9]. Because, human vector feeding must precede outbreaks, knowledge of vector host-feeding behavior is essential to our understanding of the inter-epidemic maintenance patterns of YF. Furthermore, we assessed how these bionomic attributes vary among subspecies of the Ae. simpsoni complex.

Ethics statement
The study received approval from the Scientific Ethics Review Unit (SERU) of the Kenya Medical Research Institute (Protocol NO. SSC 2787). Additionally, consent was sought verbally from household heads to set up traps around their homesteads.

Study site
Adult female Aedes simpsoni senso lato (s.l.) (hereafter as Ae. simpsoni) mosquitoes that had been collected from Rabai (peri-urban environment; human density~600/km 2 ) and from rural/sylvatic Kerio Valley (KV) (human density~45/km 2 ) as part of an arbovirus surveillance project were used in this study. Rabai (Kilifi County) is located northeast and~25 km from Mombasa City in coastal Kenya while Kerio Valley with a history of YF outbreak [4] in the Rift Valley (Fig 1). The mosquito breeding habitats differ with Kerio Valley providing rural woodland setting with numerous tree holes for mosquito breeding as opposed to Rabai, with the peri-urban setting providing typical water storage containers for mosquitoes to breed and also outdoors breeding in plant axils and water receptables. Kerio Valley is an arid to semi-arid ecology that is sparsely populated. The main economic activities of the inhabitants include farming of crops like maize and cotton, and livestock keeping especially goats and cattle. Rabai is one of the seven administrative sub-counties of Kilifi county. The main economic activities in the area include subsistence agriculture, casual labor, crafts and petty trading.

Mosquito survey
Adult host seeking female mosquitoes were surveyed using CO 2 -baited BG Sentinel traps at three trapping periods with trapping exercise beginning about three weeks after the area first received rainfall continuously for at least a week. This timing was critical to maximize assessment of vector abundance, considering the predominant tree hole and plant axils breeding ecology of Ae. simpsoni [10]. The trapping periods were August-September 2019 (mean daily temperature = 25.4˚C; mean daily rainfall = 0.58mm; relative humidity = 75.9%) and February 2020 (mean daily temperature = 27.4˚C; mean daily rainfall = 0.30mm; relative humidity = 80.3%) in Rabai and November 2019 (mean daily temperature = 19.9˚C; mean daily rainfall = 3.76mm; relative humidity = 72.4%) in KV. This trap type suitably targets Stegomyia mosquitoes which are active during the day [11]. Traps were set from morning to evening (06:30-18:00) on the same day, for at least 8 consecutive days per trapping period in peridomestic areas around homesteads. Sampling was designed to cover a large spatial area within each site. Traps were placed outdoors in the vegetation around human habitations and moved every other day to a new locality at least 500 m away. After retrieval, the mosquitoes were immobilised using triethylamine, sorted and transported in liquid nitrogen from the field to the Emerging Infectious Diseases laboratory at ICIPE, Nairobi. The mosquitoes were later identified morphologically to species level using published taxonomic keys [12] before storing in -80˚C freezers for later use.

Parity rate, daily survival and longevity estimation
Mosquitoes after retrieval from -80˚C were allowed to thaw on ice (4˚C) and then dissected for parity after observing the degree of dilation of the tracheolar skeins of the avarioles [13]. After dissection, the remaining portion of each mosquito was preserved and processed for DNA extraction and molecular speciation (described below). Daily survival rates were derived from estimated parity rates for each sampling period as described previously [14] based on the formula: P n = M, where P is the daily survival rate, M the parity rate and n the gonotrophic cycle (the number of days between emergence of adult females and first oviposition). A value of 3 days was assumed for this species [9]. Finally, the longevity (days) was estimated using the formula: 1/-lnL where L is the estimated survival [15].

Analysis of blood meals
Blood-fed specimens were individually dissected by separating the abdomen containing engorged blood from the head/thorax. The head/thorax were preserved for each mosquito and subjected to DNA extraction and molecular speciation (described below described previously [9]. PCRs in a 10 μl reaction volume comprised 2 μl 2x MyTaq Mix (Bioline, Meridian Bioscience, Germany), 10 μM of each primer, 0.2 U of Mytaq DNA polymerase and 1 μl of the template DNA (~20 ng). Thermal cycling conditions were 95˚C for 3 min followed by 40 cycles at 95˚C for 20 s, 59˚C for 30 s and 72˚C for 30 s and 72˚C for 7 min. Amplicons were resolved on 1.2% agarose gel electrophoresis against a 100bp DNA HyperLadder (Bioline, Meridian Bioscience, Tennessee, USA). The PCR products were purified using the SureClean Plus kit (Bioline, Meridian Bioscience) and outsourced to Microgen (Netherlands) for Sanger sequencing using the forward primer. DNA sequences were compared using the BLAST algorithm and the GenBank database (http://blast.ncbi.nlm.nih.gov/ Blast.cgi). Species level identification was determined when sequences exhibited � 98% identity spanning at least 300 bp as described previously [9].

Species typing of mosquitoes
Genomic DNA was extracted from the head/thorax of each blood fed specimen and remaining portion of selected parous mosquitoes as described previously. PCR was performed to identify the sibling species of Ae. simpsoni targeting the nuclear internal transcribed spacer 2 (ITS2) region using the primers ITS2A (5'-TGTGAACTGCAGGACACAT-3') and ITS2B (5'-TATGCTTAAATTCAGGGGGT-3') [17]. ITS2 is fast evolving as a marker of choice widely used for species-level discrimination [18][19][20] with vast sequence representation in public databases (e.g. GenBank). PCRs were performed in 20μl reaction volumes including 3μl 5x HOT FIREPol Blend Master Mix Kit (Solis Biodyne, Tartu, Estonia), 0.5μl of 10 μM forward and reverse primers, of 5x Mytaq HS mix polymerase and 20 ng DNA template in a ProFlex PCR systems thermocycler (Applied Biosystems, Foster City, CA, USA). The thermal cycling conditions were 95˚C for 15 min followed by 40 cycles at 95˚C for 30 s, 60˚C for 30 s and 72˚C for 45 s and 72˚C for 7 min. Amplicons were confirmed by gel electrophoresis as described previously. Similarly, the PCR products were purified using SureClean Plus kit and outsourced for Sanger sequencing, in both the forward and reverse direction. Sequences were viewed and edited in Chromas, embedded in MEGA v.6 [21] prior to querying the GenBank using BLASTn (www.ncbi.nlm.nih.gov/blast). Multiple sequence alignments of the resulting contiguous sequences were performed using ClustalW in MEGA v.6 with default parameters. Maximum likelihood (ML) trees were constructed with nodal support for the different groupings evaluated through 1000 bootstrap replications utilizing the Jukes and Cantor as best-fit model of sequence evolution.

Statistical analyses
The parity rate calculated as the percentage of parous mosquitoes to the total number dissected was established for each trapping period and comparisons made by Pearson chi-squared tests. The 95% confidence intervals (CIs) for the parity rates were estimated using binom.confint function. The human blood index is expressed as the proportion of blood-feeding on humans of the total number of blood-fed mosquitoes examined and established for subspecies and tested for significant differences using Pearson chi-squared tests. All analyses were performed at P = 0.05 using R v. 4.0.4 software [22].

Genetic composition of parous Ae. simpsoni s.l.
A subset of parous specimens (n = 43; Rabai = 37; KV = 6) randomly selected were profiled for DNA sequencing and phylogenetic analysis of the ITS2 region. Our findings show that most of the samples clustered with reference sequences of Ae. bromeliae with a strong bootstrap support (73%) (Fig 2). One specimen from Rabai grouped together with Ae. lilii in the Genbank and Ae. simpsoni samples included from Kajulu, in Kisumu County, western Kenya (Fig 2).

Blood meal feeding patterns
A total of 27 blood fed Ae. simpsoni s.l. mosquitoes were analysed from Kerio Valley (n = 9) and Rabai (n = 18), of which blood meal sources were successfully identified from 24 specimens. These represented 9 specimens from Kerio Valley and 15 from Rabai. The data revealed a total of 10 different hosts belonging to fairly large mammals (humans, goat, cow, domestic cat), rodents (grass rat, squirrel, mastomys mouse, African giant pouched mouse, mongoose) and reptile (lizard) (Fig 3). The host range was more diverse in Rabai than Kerio Valley (Fig 3). In both areas, Ae. simpsoni had fed more on humans, then squirrel (Rabai) followed by a lower representation of other host types (Fig 3). The overall human-blood-index (HBI) was 0.33 (8/24).

Association of Ae. simpsoni subspecies and human blood feeding
Next, we examined the relationship between subspecies of Ae. simpsoni s.l. and their influence on human blood feeding. A total of 27 Ae. simpsoni sl blood-fed specimens were profiled for phylogenetic analysis of sequenced ITS-2 region, with most of samples obtained from Rabai (n = 18), then Kerio Valley (n = 9). ITS2 sequences of the three blood-fed samples with unsuccessful blood meal data were included. The Ae. simpsoni sl samples resolved into 3 clades, with well supported bootstrap values (Fig 4). One of these clustered with Aedes bromeliae (GenBank No: KF135509) and had most of the samples from Rabai (n = 11) and Kerio Valley (n = 6). Clade II contained samples exclusively from Rabai (n = 4), while clade III had samples solely from Kerio Valley (n = 3). Thus, our findings show overwhelming representation of Ae. bromeliae among the blood-fed samples. Each of the clades exhibited human feeding tendency with no variation in the estimated human blood index (HBI) between the clades (P>0.05); clade I which clustered with Ae. bromeliae: (4/18); clade II: 2/3; clade III: 2/3 (Fig 4).

Discussion
Here, we provide contemporary estimates of survival and blood feeding patterns for the important arboviral vector Ae. simpsoni s.l. collected from two areas with distinct ecosystems. High parity rates and hence survival rates/longevity were evident for this mosquito across the three trapping periods. Aedes simpsoni s.l. from Kenya was previously reported to be susceptible to YFV [23], and because infectious mosquitoes reflect the age structure of adult female mosquitoes [24], our findings suggest a high vectorial capacity for pathogen transmission by this mosquito species despite moderate human feeding rates.  Ae. bromeliae, which is the only known YFV vector among Ae. simpsoni mosquitoes in East Africa [6,25], has equally been suggested as the only human biting sibling species occurring in Kenya [6]. Ae. bromeliae appears dominant in the study areas based on our parity data (randomly selected) and blood fed specimens. While our result agrees with data reported in the literature [6,25], this trend might have been affected by its domestic/peri-domestic habits where sampling was conducted. The high survival/longevity found in the current study may contribute to the vectorial capacity of Ae. bromeliae. Age is the most sensitive vectorial capacity (a measure of pathogen transmission potential) parameter [8,9]. Higher YFV transmission risk relating to this subspecies is therefore, suggested in Kenya, although validation of its competence is still an important knowledge gap. Ae. simpsoni mosquitoes commonly breed in tree holes/plant leaf axils [5]. Its adaptation to breeding in water holding containers around human habitations [5] may contribute to seasonal abundance to enhance human risk as a result of increased vector human contact.
Our sequencing and phylogenetic analyses document for the first time the presence of Ae. lilii in Rabai, Kenya, thereby confirming previous speculations on the existence of this subspecies in the area [26]. The presence of Ae. lilii in Kenya was further supported by data from Kajulu, Kisumu in western Kenya (Fig 2). The results suggest geographic differences in the distribution of the sibling species in Kenya, warranting further studies to investigate this hypothesis. The very low numbers and absence of blood feeding for Ae. lilii, suggest minimal importance of this non-human biter [6,25], in disease transmission to humans in the study areas. Notably, we uncover potentially undescribed species (clades II and III) within the complex based on phylogenetic resolutions which displayed human feeding tendencies (Fig 4). The diversity of species in this complex could be much higher than previously thought which would require additional confirmatory studies. These studies could incorporate additional markers including analysis at the genomic level. Additionally, studies on the distribution and detailed ecology that could facilitate assessment of arboviral disease risk associated with distinct subspecies are warranted. Furthermore, a PCR-based method to distinguish the vector Ae. bromeliae from the non-vector Ae. lilii based on the ITS region was recently described [25]. Application of this protocol to our samples gave mixed and inconclusive results because several specimens amplified with both specific primer sets targeting these species.
Blood meal data showed feeding on diverse hosts for Ae. simpsoni with a corresponding moderate human blood index, a trend largely driven by Ae. bromeliae. The overall engorged specimens analyzed was relatively small (n = 24) reflecting the difficulty to trap blood-fed cohorts. This could as well have been affected by our trapping approach possibly biased towards host-seeking females. However, we attempted the newly developed Aedes gravid traps to enhance gravid and engorged mosquito cohorts [27] but had no catches. Thus, we suggest that further studies should incorporate resting collections to enhance the number of freshly engorged mosquitoes [28]. While a large sample size could allow for better inference on the trophic habit of these species, nonetheless, our study provides useful baseline data regarding host utilization by this heterogenous species group inferring opportunistic feeding in these arboviral disease foci. Feeding on humans by Ae. bromeliae and the potential uncharacterised species, increases the risk of transmission to humans of diverse vector-borne pathogens (e.g., yellow fever, dengue, chikungunya, and Zika viruses) including zoonotic ones circulating in livestock or rodent hosts.
YF outbreak preparedness and the Eliminate Yellow Fever Epidemics (EYE) strategy largely centers on vaccination scale-up, a costly venture that must be guided by justified risk. This focus must expand to include mosquito-based surveillance of critical parameters that define vectorial capacity in specific ecological contexts [8,9]. Combining laboratory experiments and field ecology while incorporating genotype-based analysis of vector populations should generate integrated data that could be modeled to predict potential YFV spread and re-emergence risk.
We conclude that blood-fed and parous specimens of Ae. simpsoni s.l. mosquitoes in the study areas were mainly Ae. bromeliae (the primary YFV vector in East Africa), and it appears to be the most dominant species in this complex. High parity rates and hence survival rates/longevity were evident for this mosquito across the three trapping periods, suggesting a high vectorial capacity for pathogen transmission by this mosquito. We report the presence of Ae. lilii in Kenya with potentially yet-to-be described species in the complex (clades II and III, Fig 4) displaying human feeding tendencies. We also infer a wide host feeding range on rodents, reptile, domestic livestock besides humans especially for Ae. bromeliae, a feeding trend that could likely expose humans to various zoonotic pathogens. Overall, we highlight the importance of precision surveillance data of vector populations through genotype-based analyses for enhanced disease risk prediction and to guide cost-effective interventions (e.g. YF vaccinations).