Dengue has re-emerged as one of the most important infectious arboviral diseases across the globe and it is exerting a toll not only on human health but also on the economic development of affected countries (1). The disease is caused by four distinct dengue virus (DENV) serotypes namely Dengue 1, 2, 3 and 4, which are transmitted to humans primarily through the bite of an infected Aedes aegypti mosquito (2, 3). According to the WHO, about half of the world’s population is currently at risk of dengue infection, especially in tropical and subtropical areas. Globally, dengue cases have increased >10-fold between 2000 and 2019 (500,000 to 5.2 million), indicating expanding epidemiology of the disease. Reported deaths increased four-fold during the same period (~1000 in 2000 to ~4000 in 2015) (1). The global trend is unparalleled in the eastern African region where frequent dengue outbreaks continue to occur in Kenya and in neighboring countries like Tanzania, Somalia, Djibouti, Eritrea, Sudan and Ethiopia (1, 4–8). For instance, in Kenya the public health impact of dengue has continued to rise sharply, with successive recent outbreaks (between 2011-2022) resulting in hundreds of thousands of human cases and multiple fatalities (9). These frequent outbreaks and the co-circulation of all four DENV serotypes (10) are an indication of the significant burden posed by dengue.

The geographic spread of dengue occurs in part against a backdrop of limited understanding of the transmission dynamics, and ecology, including the bionomic role of geographic populations of the known vectors. Convergence in the basic components of the disease cycle: the virus, vector, and susceptible human hosts; in a permissive environment, provides the necessary ingredients that facilitate local arboviral pathogen emergence (11, 12). Yet, amidst the geographic spread, dengue risk patterns at local scales can be substantially heterogeneous (1, 13, 14). A better understanding of the complex interactions among these factors is required to predict the risk of dengue spread (15), but knowledge on vector populations and adaptation under changing habitat conditions are critical to achieving this.

Epidemic dengue cycles occurring in urban/peri-urban environments are increasing in eastern Africa and are driven primarily by the adaptation and expansion of vector range (e.g., Ae. aegypti) (16). Tracking changes in relevant biological parameters that define vectorial capacity can improve the prediction and response to infections or outbreaks in a timely manner (17, 18). Key among these is human blood-feeding preference (including multiple feeding) and survival (longevity) of the vector which controls opportunities for infection and transmission intensity (19–21). Besides density, measuring these vector traits in natural mosquito populations could be useful in risk assessment and to guide the implementation of vector control strategies. Mosquito longevity is a limiting factor for the transmission of arboviral pathogens, as pathogens must undergo a period of development (extrinsic incubation period - EIP) within the vector before it becomes transmissible via the salivary glands. EIP is temperature-sensitive with mean values of 15 and 6.6 days estimated for DENV at 25°C and 30°C, respectively (22). Few studies especially in Africa have described components of vectorial capacity (VC) for Ae. aegypti mostly relating to human blood meal patterns (13, 21, 23). Also, survivorship of natural populations of Ae. aegypti is seldom estimated (20, 24). Additionally, little information currently exists regarding possible differences in vector survival, as well as human blood-feeding among the two known ecotypes: the sylvan Ae. aegypti formosus and the domestic Ae. aegypti aegypti. Variations in virus vectoring abilities and geographic distribution between the forms have been described (21, 25).

Weather parameters (e.g., temperature, rainfall, humidity) modulate various biological and behavioural processes of mosquito vectors. These and other non-climatic variables such as human mobility and urbanization have been implicated as important drivers of dengue spread and outbreak occurrence (15). How these factors interact to influence bionomic traits and local transmission has received little attention.

To improve our understanding of local-scale variation in dengue occurrence, this study investigated the survival and blood-feeding patterns of wild Ae. aegypti in two ecosystems that vary in dengue outbreak occurrences: peri-urban (coastal Rabai) and rural (Rift Valley Kerio Valley) in Kenya. Also, we examined the relationship between Ae. aegypti age structure and abundance and assessed whether the genetic variability in the vector corelates with the extent of human blood-feeding.

Like other mosquito-borne pathogens, the transmission efficiency of DENV is dependent on critical components of vectorial capacity such as mosquito density, survival and human feeding preference. Here, we document how these factors could interact with each other to influence dengue dynamics influenced by weather factors. We found variation in vector parameters as a function of trapping period, a pattern that appears to mirror the observed seasonal variation in transmission risk in the Kenyan coast (40, 41).

Mosquito abundance increases the likelihood of vector-human contact and is an important parameter in climate-based models to predict dengue dynamics (42). While vector presence is a prerequisite for DENV transmission, increasingly, vector density could be a poor predictor of DENV infection risk in humans. For instance, an area with the lowest Ae. aegypti density in Vietnam displayed the highest incidence of dengue while the converse effect was true in another province (43). A similar inverse relationship between Ae. aegypti abundance and dengue endemicity was observed in Mexico (44). Also, higher abundance of Ae. aegypti in the Kenyan city of Kisumu than Mombasa, could not explain the lack of outbreaks in the former, yet recurrent in the latter (14). In our study, high vector survival rate was associated with reduced trap catches, a proxy for mosquito abundance. The strong association between mosquito age and dengue endemicity (44) or onset and offset of DENV transmission season (24), could indicate that vector demographic structure is a better predictor of dengue risk than density.

The estimated survival rates for Ae. aegypti from this study are consistent with the range of data for this species reported in the literature (parous rates: 27-92%) (24, 35, 45). Parity assumes age-independent survival (46), although Hugo et al. (24), found no difference employing age dependent and independent models in the estimation of mosquito age. The samples analyzed in our study were representative of captures over several days within a spatial scale for each trapping period potentially increasing the reliability of the estimates as previously suggested (46). Of note, parity estimates could be affected by the mosquito population recruitment whether growing, stable, or declining (47) which we did not assess at the different study sites.

The blood meal profiles for Ae. aegypti included humans but also a diverse range of animals. We observed a moderate HBI (0.51) in Rabai, a value which is comparable or even higher than the previously reported estimate (of 0.4) in coastal Kenya (13). In this study, Ae. aegypti is observed to exhibit zoophilic characteristics, an observation that contradicts the primarily anthropophilic behavior (HBI > 0.87-1) of this species in other dengue endemic areas out of Africa (21, 48, 49). The difference in Ae. aegypti HBI could be related to the origin of the samples (indoor vs outdoor) and habitat (urban vs rural/sylvatic). Lower blood fed specimens were encountered in KV (n=3), none of which had fed on humans. Thus, the difference in Ae. aegypti blood feeding between Rabai and KV could relate to the degree of urbanization i.e., human population density, which is an important risk parameter of dengue risk (2, 14). The observation of a more anthropophilic population in Rabai compared to KV may also point to fundamental differences in the behavior of Ae. aegypti populations in areas/regions of differing dengue endemicity in Kenya which, requires further elucidation. While the blood meal data provides insights into the trophic habit of this species, this inference could be affected by the relatively low number of engorged specimens analyzed (n=53). Our trapping methodology mainly targeted host-seeking females generally biased against blood-feeding cohorts whose catches can be improved by incorporating resting collections (50).

Genetic differences underlie the preference for human over other vertebrate hosts between Aaa and Aaf (51). We asked whether this can be verified through correlational studies of the genetics and host meal sources of individually analysed wild blood-fed specimens. Our data show that the mtCOI lineage that grouped with the domestic form Aaa was more associated with a higher human feeding rate (0.53 (16/30)) compared to the lineage having the feral form (0.43 (3/7) ( Figure 2B ). We suspect that the lack of a significant difference in the human feeding rates between the lineages may have been masked by tropic shifts between the trapping periods observed at the coast. Our findings point to sympatric occurrence of Ae. aegypti ecotypes, however, with possible higher abundance of the domestic than feral form in coastal Kenya as previously suggested (13, 52). The potential consequence of seasonality in their occurrence, biting patterns and effect on transmission risk merit additional studies. Feeding on diverse hosts including humans by both lineages representing the forms increases the risk of transmission to humans of diverse vector-borne pathogens (e.g., dengue, chikungunya, and Zika viruses) that are known to have animal reservoirs (53). The contribution of multiple human feeding, a common trait in Ae. aegypti (19, 54) could not be determined based on the single marker used in our present study. PCR followed by amplicon sequencing analysis underestimates the resolution of multiple host meals from individually engorged mosquitoes compared to next-generation sequencing approaches (55, 56).

Coastal Kenya has been particularly prone to multiple dengue outbreaks in the last decade (3, 57) with annual upsurges experienced after the short and long rains typically between February and June (40, 41). While rainfall may increase mosquito abundance and perhaps biting rates, the reasons for this outbreak occurrence patterns are not fully understood. Focusing on our coastal data, we analyzed trends in vector indices and weather-related variables during the two sampling periods of August-September 2019 and February 2020. The data showed that higher vector survival coincided with higher human feeding rates (i.e., HBI) in February 2020. Both values were consistently lower in August-September 2019 (Parity=77% vs 67% and HBI=69% vs 29% for February 2020 and September 2019, respectively). High human feeding rates have been shown to correlate with heightened period of dengue risk (33,54]. February 2020 are generally dry/cool periods which have been found to correlate positively with enhanced vector survival rates and dengue transmission as observed in Vietnam (24). Higher vector survival was associated with both increases in mean temperature and relative humidity, during this period compared to August-September 2019, suggesting them as influential factors affecting mosquito age (or longevity) consistent with previous findings (58). It is noteworthy that temperature increases including a 1°C rise can accelerate a reduction in the extrinsic incubation period (EIP) and thereby, increase pathogen transmission potential (59). Our results suggest that intense DENV transmission and perhaps onset of outbreaks at specific times of the year may be punctuated by enhanced Ae. aegypti vectorial capacity linked to increased vector survival and human feeding in concert with suitable weather conditions. A convergence of these biotic and abiotic attributes could be critical in reinitiating the seasonal pattern of dengue virus transmission in coastal Kenya. The low temperature conditions, low human biting may have dampened the estimated dengue risk in KV despite evidence of high survival rates. It is also noteworthy that we assumed a uniform gonotrophic cycle (3 days) in our estimates of daily survival and longevity and this value has been observed to vary as a function of temperature and seasons (35, 60). Thus, further studies should consider this likely discrepancy which could influence interpretation of the findings.

We conclude that, Ae. aegypti exhibits local variation in important components of vectorial capacity (VC)- abundance, survival, and human blood feeding. In coastal Rabai, vector density was inversely related to mosquito age which together with human feeding are the most sensitive VC parameters (20, 26). Convergence of high survival and propensity to feed on humans as noted in coastal Rabai, could drive intense DENV transmission or onset of outbreaks at certain times of the year; however, in concert with suitable weather conditions. Genetic factors likely underpin the degree of anthropophagy in Ae. aegypti but may be influenced by potential seasonal shift in human feeding rates. Despite high representation of humans in the trophic habit, Ae. aegypti exhibited opportunistic feeding, a trend which increases human exposure to risk of zoonotic pathogens. Modeling transmission is increasingly seen as an important component of disease risk prediction. For dengue, quantifying transmission risk needs to integrate vector bionomic factors beyond abundance together with climatic parameters in simulation studies for reliability of any developed models (15, 61).

The original contributions presented in the study are included in the article/supplementary material. Further inquiries can be directed to the corresponding author.

The study received approval from the Scientific Ethics Review Unit (SERU) of the Kenya 116 Medical Research Institute (Protocol NO. SSC 2787). Additionally, consent was sought from household heads to set up traps around their homesteads.

Conceptualization, RS and DT. Investigation and Methodology, WK, RS, GR, DT. Data analyses, WK, DT. Validation, RS, DT. Supervision, RS, NM, DT. Funding acquisition, RS, BT, DT. Writing—original draft, DT. Writing—review and editing, WK, RS, GR, SA, NM, BT, DT. All authors contributed to the article and approved the submitted version.