Dengue-mediated changes in the vectorial capacity of Aedes aegypti (Diptera: Culicidae): manipulation of transmission or infection by-product?

Abstract

An arthropod’s vectorial capacity summarizes its disease transmission potential. Life-history traits, such as fecundity or survival, and behavioral traits, such as locomotor activity, host-seeking and feeding behavior, are important components of vectorial capacity. Studies have shown that mosquito-borne pathogens may alter important vectorial capacity traits of their mosquito vectors, thus directly impacting their transmission and epidemic potential. Here, we compile and discuss the evidence supporting dengue-mediated changes in the yellow fever mosquito Aedes aegypti (L.), its primary vector, and evaluate whether the observed effects represent an evolved trait manipulation with epidemiological implications. Dengue infection appears to manipulate essential traits that facilitate vector–host contact, such as locomotor activity, host-seeking, and feeding behavior, but the underlying mechanisms are still not understood. Conversely, life-history traits relevant to vector population dynamics, such as survival, oviposition, and fecundity, appear to be negatively impacted by dengue virus. Overall, any detrimental effects on life-history traits may be a negligible cost derived from the virulence that dengue has evolved to facilitate its transmission by manipulating Ae. aegypti behavior and feeding performance. However, methodological disparities among studies render comparisons difficult and limit the ability to reach well-supported conclusions. This highlights the need for more standardized methods for the research into changes in virus-mediated traits. Eventually, we argue that the effects on life-history traits and behavior outlined here must be considered when assessing the epidemiological impact of dengue or other arbovirus–vector–host interactions.

Aedes aegypti, dengue, trait manipulation, life-history trait, feeding behavior, vectorial capacity

Introduction

Mosquito-borne arboviruses within the Orthoflavivirus genus are important viral pathogens causing severe human diseases (Pierson and Diamond 2020). Dengue viruses (DENV serotypes 1–4), etiological agents of dengue fever, are the most prominent pathogens of the genus: in 2023, WHO estimated around 100–400 million DENV infections occurring yearly (WHO 2024). Although the disease is endemic to (sub)tropical regions, about half of the world’s population is now at risk of contracting the pathogen due to the expanding range of its primary vector, the yellow fever mosquito Aedes aegypti (L.) (Bhatt et al. 2013). Viral dispersal is facilitated by globalization, trade, changes in public health policies, and climate change, which established the role of humans as both reservoir and amplifying host (Murray et al. 2013, Weaver 2013, Yu and Cheng 2022). The global spread of DENV can be traced to a change from occasional emergences via spillovers from sylvatic cycles (Hanley et al. 2013) to endemicity within the human population (Islam et al. 2021). In comparison to other orthoflaviviruses, such as the West Nile or Japanese encephalitis viruses, DENV has a relatively limited host range, and its epidemic transmission cycle primarily involves human hosts and Ae. aegypti (Weaver and Barrett 2004, Yu and Cheng 2022). The anthropophilic nature of Ae. aegypti favors this cycle because they feed preferentially on human blood to complete their gonotrophic cycles (Pruszynski et al. 2020), and human odors induce a unique neural code that facilitates host-seeking behaviors (Zhao et al. 2022).

Given Ae. aegypti human feeding preference, DENV may have evolved to alter epidemiologically relevant traits of its vector to favor its spread. These traits are summarized as vectorial capacity, an important component of mosquito-borne disease epidemiology, defined as the mosquito population’s pathogen transmission potential (Macdonald 1957). Vectorial capacity is central in assessing disease risk and a core aspect of the modeling of vector-borne diseases in epidemiological studies (Smith et al. 2012). It is dynamically influenced by multiple factors, including viral genetic diversity (Fontaine et al. 2018), mosquito genetic background and its interaction with viral genetic diversity (Lambrechts et al. 2009), and the mosquito microbiome (Cansado-Utrilla et al. 2021), as well as changes in blood feeding-related behaviors and key parameters of the vector population (e.g., density) related to life-history traits (Maciel-de-Freitas et al. 2011), and abiotic parameters (e.g., temperature) (Liu-Helmersson et al. 2014). Virus-induced alterations in mosquito behavior and life-history traits may impact vectorial capacity and therefore be highly relevant in epidemiological risk assessment.

Such associations have been extensively researched in arthropod-borne diseases, such as malaria caused by parasites within the Plasmodium genus (Rich et al. 2009). In some empirical studies, Plasmodium-infected mosquitoes displayed enhanced feeding behavior during the sporozoite developmental stage of the parasite, when it can be inoculated into the human host (for a review: Martinez et al. 2021). In addition, symptomatic malaria in the host itself seemingly enhances their attractiveness for mosquitoes by producing a signature “malaria odor” (Ellwanger et al. 2021). Similar observations have also been made for hosts infected with Rift Valley fever virus (Phlebovirus) (Turell et al. 1984, Rossignol et al. 1985). Observed effects of Plasmodium infection in both vector and host appear to enhance mosquito vectorial capacity, yet an actual quantifiable impact on transmission has yet to be demonstrated (Cator et al. 2012). Moreover, although it is appealing to interpret these effects as a direct manipulation exerted by the parasite, they may simply be by-products of the infection response displayed by the host (Stanczyk et al. 2017).

The potential impact that DENV, similar to Plasmodium, may have on mosquito behaviors and life-history traits to promote Ae. aegypti vectorial capacity has attracted some attention from the scientific community in the last 15 yr. In this review, we compile the known evidence for DENV-mediated alterations in Ae. aegypti vectorial capacity-related traits (Table 1; Fig. 1) and evaluate the role and potential epidemiological implications of these changes as well as current limitations.

Table 1.

Compiled evidence for DENV-mediated changes in Aedes aegypti behavior (red) and life-history traits (blue)

References	Serotype	Infection	Trait	Effect	Days post-infection
Putnam and Scott (1995)	DENV-2	Intrathoracic	Feeding behavior	No effect	14
Platt et al. (1997)	DENV-3	Intrathoracic	Feeding behavior	Increased feeding and probing time	5, 8, 11
Lima-Camara et al. (2011)	DENV-2	Intrathoracic	Locomotor activity and host-seeking	Increased activity	2-6
Maciel-de-Freitas et al. (2011)	DENV-2	Oral	Survival	Increased mortality	>12
			Oviposition and fecundity	Reduced fecundity	late stage
Richards et al. (2012)	DENV-1	Oral	Oviposition and fecundity	No effect	14
Sim et al. (2012)	DENV-2	Oral	Feeding behavior	No effect	14
Maciel-de-Freitas et al. (2013)	DENV-2	Oral	Feeding behavior	Reduced feeding motivation, increased avidity	7 and 14
Sylvestre et al. (2013)	DENV-2	Oral	Feeding behavior	Increased feeding time; longer blood ingestion time; no effect in probing time	2 week
			Oviposition and fecundity	Reduced fecundity; smaller clutches	18
			Survival	Increased mortality	14
			Locomotor activity and host-seeking	No effect	unspecified
Hill et al. (2014)	DENV-2	Oral	Survival	Increased mortality	unspecified
			Oviposition and fecundity	No effect	3
Gaburro et al. (2018b)	DENV-2	Oral	Oviposition and fecundity	Altered site-fidelity; reduced fecundity	3–10
Tallon et al. (2020)	DENV-1	Oral	Locomotor activity and host-seeking	Early-stage increased locomotor activity, followed by a late-stage decrease; late-stage increased host-seeking	4–6 (early stage); 14–16 (late stage)
Feitosa-Suntheimer et al. (2022)	DENV-2	Oral	Oviposition and fecundity	Reduced fecundity; smaller clutches; no effect on oviposition	Early stage
Wei Xiang et al. (2022)	DENV-2	Oral	Feeding behavior	Increased biting frequency; increased probing frequency	Unspecified
Maraschin et al. (2023)	DENV-4	Oral	Survival	No effect	N/A
Petersen et al. (2023)	DENV-1	Oral	Oviposition and fecundity	No effect	14
Javed et al. (2024)	DENV-2	Oral	Locomotor activity and host-seeking	Increased flight durations and flight count; no effect on locomotor speed	12
Keirsebelik et al. (2024)	DENV-1	Oral	Survival	Increased mortality in 3 out of 4 sampled populations	7 dpi
			Oviposition and fecundity	Reduced fecundity in one population	Unspecified

Fig. 1.

Summary of the evidence of DENV-mediated changes in Aedes aegypti behavior and life-history traits.

Dengue-mediated Trait Alterations

Behavior

The efficiency of DENV transmission is dependent on factors related to mosquito–host contact. Behavioral manipulation in mosquitoes that could increase successful mosquito–host contact may occur during the host-seeking and feeding stages (Vogels et al. 2017). In the host-seeking stage, pathogens could induce a stronger mosquito response to kairomones, which are host-specific odor cues (Martinez et al. 2021). Aedes aegypti previously has shown a preference to feed on hosts affected by flavivirus infection (DENV and ZIKV), possibly due to the increased release of acetophenone, a skin microbiota compound moderating mosquito attraction (Zhang et al. 2022). In the feeding stage, infections could impede the normal phenotype by lowering feeding performance in the mosquito. Typically, mosquitoes must locate suitable hosts and obtain a sufficient blood meal while avoiding host defenses (Sylvestre et al. 2013). In the context of a lowered mosquito feeding efficiency post-infection, viral infection must increase host probing, biting frequency, and ability to detect multiple suitable hosts to obtain the same resources as it would pre-infection (Vogels et al. 2017). Pathogens have been shown to alter these behaviors in Ae. aegypti: for example, infection with Plasmodium gallinaceum (Brumpt) increased host-seeking duration (Rossignol et al. 1984) whereas infection with Brugia pahangi (Buckley and Edeson) affected mosquito flight capacity (Rowland and Lindsay 1986). This section will review DENV-mediated behavioral changes that could enhance Ae. aegypti–host contact and therefore virus transmission.

Locomotor activity and host-seeking

DENV infection in Ae. aegypti appears to upregulate genes involving locomotor activity and host-seeking, effects which change with the progression of the infection. A study conducting intrathoracic DENV-2 infection in Ae. aegypti females reported an increase in total locomotor activity 2 to 6 days post-infection (dpi), most prominently in the light-on/light-off transition (Lima-Camara et al. 2011). The authors compared this observation to a naturally occurring activity peak, likely under circadian control (Gentile et al. 2009). Similarly, a mosquito orthologue of the CLOCK gene, which is an important circadian regulator in Drosophila, previously was found to be strongly upregulated after DENV-2 infection (Xi et al. 2008); this supports the idea of a DENV-mediated behavioral change, enhancing the naturally occurring activity peak. These findings were complemented by the known neurotropism of DENV-2, as the virus can replicate in Ae. aegypti primary neuron culture (Gaburro et al. 2018a) and infected mosquitoes show detectable levels of viral antigens in the nervous tissue as early as 5 dpi (Salazar et al. 2007). In fact, the heads and salivary glands of Ae. aegypti were shown to be the only regions where antigens continue to accumulate until 21 dpi, whereas other tissues showed a decrease in infection markers (Salazar et al. 2007). A recent study also reported increased Ae. aegypti flight count and duration at 12 dpi following DENV-2 infection (Javed et al. 2024). In contrast, another study mentioned no difference observed in DENV-mediated host-seeking behavior, following Ae. aegypti oral infection with DENV-2 (Sylvestre et al. 2013).

In parallel, Tallon et al. (2020) observed an increase in Ae. aegypti locomotor activity 4 to 6 dpi with DENV-1. However, the authors pointed toward serotype-specific effects: although there was an increase in locomotor activity early after infection, the odor-driven host-seeking behavior was not altered. In contrast, a decrease in locomotor activity was reported at 14 to 16 dpi, accompanied by an enhanced sensitivity to kairomones. This improved sensitivity of the peripheral olfactory system was correlated with antennal-specific gene upregulation and neural signaling transcripts, and the effect seemed to be dependent on the completion of the extrinsic incubation period of the virus at around 14 dpi (Tallon et al. 2020). The authors speculated that DENV-1 may selectively increase Ae. aegypti mobility early during infection to increase spatial exploration and predation avoidance while mobilizing host-seeking resources later (Tallon et al. 2020). This agrees with the idea of circadian upregulation early post-infection, as hypothesized by Lima-Camara et al. (2011).

Feeding behavior

The literature reports mixed results regarding changes in the feeding behavior of DENV-infected Ae. aegypti. Intrathoracic infection with DENV-3 increased probing and feeding time in female Ae. aegypti (Platt et al. 1997). The difference between infected individuals and controls was significant at specific time points associated with viral spread in tissue essential for feeding behaviors, such as the salivary glands, the brain and compound eyes. Platt et al. (1997) hypothesized that longer feeding and probing times may explain the efficiency of Ae. aegypti as a DENV vector because multiple sources are required for a sufficient blood meal due to interruptions by reactive hosts. In contrast, Putnam and Scott (1995) reported no difference in feeding duration at 14 dpi between intrathoracic DENV-2 infected Ae. aegypti and controls, a result which was later replicated with oral infections (Sim et al. 2012). The authors speculated that an increased probing time would be detrimental to the fitness of the vector, because feeding efficiency is prioritized over avoiding host defensive behavior. By controlling for host defensive behavior with the use of anesthetized mice, Sylvestre et al. (2013) demonstrated an age-related effect of infection with DENV-2: starting 2 week post-infection, infected Ae. aegypti generally showed increased feeding time and required a longer time to ingest a blood meal compared to age-matched controls. However, there were no observed differences in probing time, meaning that infected mosquitoes did not start feeding faster. This lack of a probing effect was associated with no significant difference in host-seeking behavior between groups, yet the authors’ findings on this parameter were not supported by other studies (see the section “Locomotor activity and host-seeking”).

Furthermore, Maciel-de-Freitas et al. (2013) extended the assessment of mosquito feeding behavior by including 2 additional parameters in their study: motivation to feed (the rate of initiating feeding) and avidity (the rate of re-feeding after interrupted feeding). In this study, Ae. aegypti orally infected with DENV-2 were evaluated for these parameters at 7 or 14 dpi and showed a reduced motivation to feed. The authors measured avidity as the time required to start probing when again offered a host, 2 h after the initial feeding round. Infected Ae. aegypti were more likely to show avidity despite having a larger blood meal size at the first feed. In comparison, uninfected mosquitoes with a larger blood meal at the first feed did not show increased avidity, and infected mosquitoes that did not show feeding motivation at the first feed did show increased avidity. These findings indicated that DENV-2 infection promoted avidity irrespective of initial feeding performance, which could affect transmission as it enhanced mosquito–host contact. In fact, Luz et al. (2011) simulated the epidemiological impact of a theoretical increase in the biting activity of DENV-infected Ae. aegypti. Their model showed that an assumed 50% increase in the biting rate of infected mosquitoes would increase the number of dengue infections. This theoretical increase also would amplify the current pattern of biennial epidemics in endemic areas; an increase in transmission would result in a relative rise in secondary infections, which are more severe due to antibody-dependent enhancement (Guzmán and Kourí 2002, Luz et al. 2011).

A recent study on DENV-2-mediated changes in Ae. aegypti confirmed the previously reported increase in host-seeking behavior and biting frequency in infected individuals (Wei Xiang et al. 2022). Due to a lowered biting efficiency, infected mosquitoes were observed engaging in successive short probes (~20 s) to compensate for unsuccessful feeding. Wei Xiang et al. (2022) generated a mathematical model based on the integrated behavioral data and noted how the basic reproduction rate of the virus (R₀, number of infected hosts per infected vector) tripled under these apparent DENV-2-mediated manipulations. Together with the theoretical simulation of Luz et al. (2011), Wei Xiang et al. (2022) were the first to quantify the impact that such behavioral changes in the vector might have on the epidemiology of DENV.

Life-history Traits

Because DENV viruses rely on Ae. aegypti to facilitate their transmission, it is important to consider how virulence affects the life-history traits indicative of vector population dynamics, such as mosquito survival and reproductive success (i.e., oviposition and fecundity). Any costs to vector fitness exerted by infection with DENV could impede transmission, whereas developing a benign, if not beneficial, relationship could enhance transmission (Lambrechts and Scott 2009). This section evaluates how the vector–virus interaction affects life-history traits relevant to vectorial capacity.

Survival

Infection with DENV appears to negatively affect Ae. aegypti lifespan (Hill et al. 2014), with one study reporting halved longevity at 14 dpi in experimentally infected mosquitoes compared to the control group (Sylvestre et al. 2013). This observation was supported by the study of Maciel-de-Freitas et al. (2011), where survival rates remained similar between groups up to 12 dpi, followed by a steep increase in mortality in infected individuals. This effect on longevity was associated with an increased DENV-2 titer, as quantified by the number of viral RNA copies (Maciel-de-Freitas et al. 2011). However, results from field studies indicated that survival rates past the third gonotrophic cycle (> 20 dpi) were generally low, which questions the impact of these age-specific findings in nature (Maciel-de-Freitas et al. 2007).

A meta-analysis by Lambrechts and Scott (2009) investigating cumulative arboviral virulence in mosquito vectors further rejected the apparent impact of DENV on mosquito survival: Aedes mosquitoes were the only genus considered to display no statistically significant increase in mortality after arboviral infection. This meta-analysis, however, did not specifically consider the Aedes-DENV interaction. Although more research has since been conducted on this specific interaction, the relevance of the apparent reduction in vector longevity is inconclusive. Only recently have research groups investigated how DENV serotypes other than DENV-2 impact mosquito survival: DENV-4 appears not to affect Ae. aegypti survival (Maraschin et al. 2023), whereas DENV-1 generally increased mortality across multiple mosquito populations (Keirsebelik et al. 2024).

Nonetheless, an interesting observation related to DENV-exposed, but not infected, mosquitoes should be investigated. Maciel-de-Freitas et al. (2011) found that mosquitoes fed with a DENV-2-infected blood meal, but negative for subsequent infection by qRT-PCR assays, had the lowest survival rates compared to both controls and infected mosquitoes; this increase in mortality was especially noticeable in the first days after exposure to the infected blood meal. These findings hint toward possible fitness costs associated with an immune response, as a strong immune activation in response to DENV infection could act as a population bottleneck before virus dissemination (Schmid-Hempel 2005, Maciel-de-Freitas et al. 2011). However, any moderating immune effects have been overlooked so far: it is unclear whether the Ae. aegypti immune response to DENV could impact longevity after contracting the virus (Maciel-de-Freitas et al. 2011).

Oviposition and fecundity

DENV infection was shown to overexpress genes associated with learning and synaptic plasticity, which could alter female mosquito selection of oviposition sites, which are thought to be influenced by the semiochemical conditions during the individual’s own larval development (Gaburro et al. 2018b). This phenomenon is called “site-fidelity,” meaning that females are more likely to locate oviposition sites based on early-stage learning (McCall et al. 2001). Previous literature has shown that mosquitoes are capable of associative learning, for example, in the context of odor cues (Menda et al. 2013). Associative learning can influence vector behavior, as adapting to odor cues is essential for navigating the environment and preserving fitness (Vinauger et al. 2014). Alterations in odor cue learning could change the preference for oviposition sites, an important fitness parameter. Taken together, experiments with Ae. aegypti support the idea of site-fidelity (Kaur et al. 2003), and the study of Gaburro et al. (2018b) provided evidence for DENV-2-mediated changes in this trait. In a y-tube olfactometer behavioral assay, uninfected females preferred oviposition within the segment containing the same odor cue as their rearing environment (water or skatole—a very attractive substance for mosquitoes), thus demonstrating olfactory conditioning and associative learning. In contrast, DENV-2-infected individuals preferred water oviposition, irrespective of their rearing environment, and displayed reduced fecundity compared to controls. This observation was associated with a significant overexpression of genes involved in learning and synaptic plasticity at 7 dpi, with a significant decrease occurring at 10 dpi. Furthermore, the viral RNA load was increased within the heads of DENV-2-infected mosquito between 4 dpi and 10 dpi when results showed a 100% infection rate. This is congruent with previous findings showing that head tissues present DENV viral antigens by 5 dpi (Salazar et al. 2007).

Regarding fecundity, DENV-mediated effects on Ae. aegypti show an age-specific outcome of infection, with studies outlining contrasting patterns. In the work of Sylvestre et al. (2013), Ae. aegypti females orally infected with a DENV-2 strain laid eggs less frequently than controls, with a smaller clutch size; this difference was only significant from the third clutch post-infection (approx. 18 dpi), with comparable outcomes between groups in the first 2 clutches (< 11 dpi). Although a general age-related decline can be observed in the oviposition success rate and clutch size, infection with DENV-2 exacerbated this effect. These findings were supported by Maciel-de-Freitas et al. (2011): Ae. aegypti infected with DENV-2 showed lower fecundity, and the difference only became significant with later gonotrophic cycles. In addition, the overall number of eggs per cycle showed an age-specific halving of fecundity by the fifth clutch, irrespective of infection status (Maciel-de-Freitas et al. 2011). In contrast, a study by Feitosa-Suntheimer et al. (2022) observed a significant decrease in fecundity in DENV-2-infected Ae. aegypti during the first gonotrophic cycle. Oviposition time remained unaffected, occurring at 3 days post-blood meal on average. Other studies reported a similar pattern of fecundity alterations in the first gonotrophic cycle of Ae. aegypti mosquitoes infected with other arboviruses, such as chikungunya virus (Sirisena et al. 2018) and Zika virus (Petersen et al. 2018). In addition, infection with DENV-1 appeared to generally yield no significant impact on fecundity in Ae. aegypti (Richards et al. 2012), although a slight decline with mosquito age was still observed (Petersen et al. 2023), and some mosquito populations seemed to be more susceptible than others (Keirsebelik et al. 2024). Similarly, both Hill et al. (2014) and Gaburro et al. (2018b) found no significant effects of DENV-2 infection on fecundity and fertility. Feitosa-Suntheimer et al. (2022) discussed the possibility of methodological differences that could yield these contrasting results, such as the specific mosquito strain and virus strain/serotype, but additional factors could be considered, such as the origin and composition of the blood meal. Nevertheless, the clear moderating effect of mosquito lifespan was consistent throughout the literature.

Considering possible molecular mechanisms, Sylvestre et al. (2013) associated their observed late-lifespan effects on fecundity with the progression of the infection, as ovarian tissues presented infection markers at around 15 dpi, when the extrinsic incubation period has been completed (Salazar et al. 2007). An immune response elicited by DENV particles in the ovaries was hypothesized to impact oviposition and the gonotrophic cycle (Sylvestre et al. 2013). Feitosa-Suntheimer et al. (2022) offer an alternative explanation, as their study showed an early-lifespan effect: they conducted a transcriptomic analysis to investigate which differentially expressed genes in the ovaries correlated with observed patterns in fecundity. Two upregulated genes in DENV-2-infected mosquitoes were shown to play an antiviral role against DENV-2, which may interfere with the earlier gonotrophic cycles. Other studies have shown that multiple mosquito immune pathways were upregulated in response to a DENV-2 immune challenge at the cost of reducing lifespan and fecundity (Jupatanakul et al. 2017, Li et al. 2020). Overall, it is unclear whether these fecundity patterns were due to an increased viral load in the ovaries, as expected for a late-lifespan progression of the infection (Sylvestre et al. 2013) and illustrated by an increase of vertical transmission (Lequime et al. 2016), or if the early-lifespan explanation is more plausible, due to an initial immune activation (Feitosa-Suntheimer et al. 2022). Both effects are likely interconnected, yet no study has studied this relationship.

Manipulation or Infection By-product?

In the previous section, we summarized the known evidence for DENV-mediated changes in Ae. aegypti traits related to vectorial capacity. Whether these observed alterations were due to direct manipulation by the virus or just a by-product of a complex interaction remains inconclusive.

In terms of behavior, DENV-infected Ae. aegypti showed increased locomotor activity in the early stages of the infection, with amplified host-seeking behavior in the later stages, when infectious (Lima-Camara et al. 2011, Tallon et al. 2020). It could be of evolutionary importance to switch from spatial exploration to host-seeking once the infection has been successfully disseminated and the mosquito has overcome the burden of the infection, when the vector is able to transmit the virus without risking predation (Tallon et al. 2020). Yet, it is still unclear what viral mechanisms could lead to the direct manipulation and upregulation of the necessary genes, beyond the confounding effects of the Ae. aegypti normal immune response to blood meals (Giraldo-Calderón et al. 2020) and the prioritization of oviposition site exploration immediately following a blood meal. Observed behavioral alterations in mosquitoes should be correlated with candidate molecular mechanisms that could explain the effects, while controlling for potential confounding factors such as the mosquito’s immune response (Feitosa-Suntheimer et al. 2022). As there is no evidence for pathology at the tissue level for Ae. aegypti infected with dengue viruses, contrary to findings in other arbovirus-mosquito pairs (e.g., Mims et al. 1966, Weaver et al. 1988, Bowers et al. 2003), there is currently little support for the idea that these observed effects are due to “sickness behavior.”

Furthermore, one study showed no effect of infection on host-seeking (Sylvestre et al. 2013), however, host-seeking was quantified as the time elapsed between locating the host and introducing the proboscis into the host, which is an ambiguous criterion describing multiple behaviors. Due to observer subjectivity when applying protocols, it can become difficult to differentiate among the effects of interconnected traits. Moreover, Sylvestre et al. (2013) conducted their behavioral assay on anesthetized mice, representing another limitation. In comparison, other studies used synthetic human odors to simulate mosquito attractiveness to the host (Tallon et al. 2020), which plays more naturally into Ae. aegypti’s anthropophilic preferences (Pruszynski et al. 2020) and odor-cue learning (Menda et al. 2013, Vinauger et al. 2014). To better evaluate the relationship among all factors involved in transmission, DENV’s impact on the vertebrate host’s attractiveness to uninfected Ae. aegypti should also be considered.

The lack of methodological cohesion across studies also impacts assessment of the feeding behavior of infected mosquitoes: it is difficult to quantify the observed results into one concrete phenotype due to the large variability of the parameters measured and the limited reliability of behavioral assay measurements (Maire et al. 2024). Overall, DENV-infected Ae. aegypti were reported to show increased feeding (Platt et al. 1997, Sylvestre et al. 2013) and probing time, and increased probing frequency (Platt et al. 1997, Wei Xiang et al. 2022). Some studies found no effects of infection on feeding or probing time (Putnam and Scott 1995, Sylvestre et al. 2013), whereas other authors found reduced feeding motivation, increased avidity after interrupted feeding (Maciel-de-Freitas et al. 2013), and increased biting frequency (Wei Xiang et al. 2022). However, whether these parameters are different or convey the same observation is largely unclear. These behaviors generally are not individually quantified in experiments, but rather are condensed into the ‘biting rate’ variable, the classic measurement used in epidemiological modeling (Maire et al. 2024). Wei Xiang et al. (2022) generated a high-throughput behavioral assay after highlighting methodological inconsistencies in the literature, such as host choice, infection method, and a limited number of well-defined parameters. Reducing methodological variability among laboratories would help compare the behaviors seemingly altered by infection with DENV. This could be achieved by employing deep learning algorithms over manual video annotations, and using novel modeling techniques that can integrate mosquito body part tracking to quantify feeding behaviors and host-seeking (Hol et al. 2020, Maire et al. 2024), 3-dimensional object detection algorithms for locomotor activity monitoring (Javed et al. 2024), or modifications of systems initially developed to assess Drosophila behavior (Gaviraghi and Oliveira 2020, Henriques-Santos et al. 2023). Automating data collection would reduce observer subjectivity in reporting mosquito traits and behavioral patterns, which are already naturally variable, even in the absence of infection (Zahid et al. 2023).

For example, mosquito feeding behaviors depend on many external factors, such as host type and biting area: Ae. aegypti was shown to feed faster on humans than mice, and probe for a shorter time on highly vascularized areas on the host (Martin-Martin et al. 2023). Using anesthetized mice as the host of choice also could reduce the impact of important evolutionary constraints for the observed behaviors, in the form of host defensive behavior. A reactive host can strongly influence the evolutionary pressure on DENV to affect Ae. aegypti feeding performance, either by hindering performance to generate multiple feeding events (Wei Xiang et al. 2022) or by improving performance to ensure vector fitness and avoid host defenses (Putnam and Scott 1995). The intrathoracic inoculation method used by some studies (Putnam and Scott 1995, Platt et al. 1997, Lima-Camara et al. 2011), together with the widely used membrane feeders for oral infections, might further reduce the ecological validity of the findings. These methodologies could lead to confounding effects (e.g., injury after intrathoracic inoculation), remove essential host-dependent effects (e.g., blood clotting, diuresis and peritrophic membranes development), and underestimate early DENV-induced immune responses in the mosquito (Martin-Martin et al. 2023) that may exert evolutionary pressure on feeding behavior. Blood meal composition (e.g., origin, presence/absence of clotting factors, dilution with cell culture medium, different blood-feeding systems) should also be standardized among laboratories, as variability in the choice of blood meal has been shown to impact the fecundity and blood-feeding rate of Ae. aegypti (Suresh et al. 2024).

When considering life-history traits, DENV infection displayed a general age-dependent effect on Ae. aegypti, yet the specific patterns of these effects differed in the literature. Studies have shown that DENV reduced infected mosquito fecundity and clutch size. It is unclear, however, whether this takes place in the first gonotrophic cycle due to an initial immune activation in response to the virus (Feitosa-Suntheimer et al. 2022) or in later cycles, when viral particles have disseminated to the ovaries (Maciel-de-Freitas et al. 2011, Sylvestre et al. 2013). Interestingly, although these contrasting findings were found with DENV-2, infection with DENV-1 appears to not impact fecundity in most mosquito populations (Keirsebelik et al. 2024). Therefore, it is likely that the neutral-to-disadvantageous effects observed on fecundity (Gaburro et al. 2018b) and survival (Maciel-de-Freitas et al. 2011) represent a by-product of the interaction between different Ae. aegypti populations and different DENV strains and serotypes (Sylvestre et al. 2013). There is a significant bias in the literature for DENV-2, with studies only recently being conducted with DENV-4 (Maraschin et al. 2023). Although the incubation period of all 4 serotypes is similar (Chan and Johansson 2012), they are phylogenetically and epidemiologically distinct viruses (Yu and Cheng 2022). Therefore, studies should compare DENV-mediated trait changes among serotypes, while standardizing the viral and mosquito strains used as controls across research groups. On a larger scale, it is also possible that variations in the altered phenotype of life-history traits and feeding behavior may be attributed to the specific interaction between natural strain variations within the DENV serotypes themselves and Ae. aegypti populations. These gene–gene modulations of the observed phenotypes, when also considering the variance present at a specific time-points across a mosquito population, over a time interval within an individual, or in response to environmental triggers (Cator et al. 2020), should be further considered when generating epidemiological models.

Given the low levels of vertical transmission observed in DENV, as compared to the established efficiency of horizontal transmission (Adams and Boots 2010, Grunnill and Boots 2016), and the relatively low prevalence of DENV infection in endemic areas (Vikram et al. 2015, Peña-García et al. 2016, Pérez-Castro et al. 2016, Chetry et al. 2020), there appears to be no strong evolutionary pressure to protect vector fitness. Reduced fecundity could explain the overall lower success rates of vertical transmission in DENV, if virus dissemination indeed impacts later gonotrophic cycles (Lequime et al. 2016). It has been previously suggested that DENV may establish “stabilized” infections, resulting in continuous vertical transmission across generations (Joshi et al. 2002). In this case, there could be different selective pressures that maintain vector fitness. However, the low levels of vertical transmission observed in the field do not support the frequent establishment of such stabilized infections within mosquito populations. One interesting finding not entirely accounted for is that infection with DENV appeared to alter Ae. aegypti’s odor-cued oviposition site selection (Gaburro et al. 2018b). This could indeed be a manipulation by the virus to enhance environmental exploration and feeding behaviors dependent on odor cues, even though it comes at the cost to fecundity (Gaburro et al. 2018b).

Disentangling virus-mediated manipulations from infection by-products, e.g., mediated by the innate immune response, is challenging. Comparing mosquitoes infected with DENV or insect-specific flaviviruses (ISFs) might provide additional elements to explore this question. ISFs are phylogenetically related to mosquito-borne flaviviruses such as dengue but lack replication capability in vertebrates (Blitvich and Firth 2015). They trigger similar immune responses in the vector (Öhlund et al. 2021), but have different evolutionary pressures linked to alternative modes of transmission, including stabilized infections maintained across generations by vertical transmission (Blitvich and Firth 2015). In a recent study, the ISF cell fusing agent virus, for example, has shown no significant impact on Ae. aegypti fitness and blood-feeding behavior, except for a limited increase in human attraction (Suzuki et al. 2024). Additional studies with dual-host affiliated insect-specific flaviviruses that are phylogenetically closer to dengue and other major arboviruses would be particularly interesting.

Conclusion

Overall, there is evidence for DENV-mediated changes that influence Ae. aegypti vectorial capacity at essential points during infection: at 5 dpi, when viral particles are present in neural tissue (early-stage), and after 14 dpi, when transmission to a new host is possible (late-stage). On the one hand, infection appears to manipulate mosquito locomotor activity, host-seeking, and feeding behavior, with epidemiological models estimating an increase in DENV transmission as a direct consequence of this manipulation (Luz et al. 2011, Wei Xiang et al. 2022). On the other hand, life-history traits such as survival, oviposition, and fecundity seem to be negatively impacted by infection. Based on the expanding range of Ae. aegypti (Bhatt et al. 2013) and the high number of yearly DENV cases (WHO 2024), the exerted costs on life-history traits seem negligible for vectorial capacity. Thus, any detrimental effects on life-history traits may be just by-products of the virulence DENV evolved to facilitate its horizontal transmission by manipulating mosquito feeding performance.

Concurrent with potential dengue-mediated changes, vectorial capacity is influenced by many biotic and abiotic factors (Kramer and Ciota 2015). Although the full exploration of all these factors and their interactions is pragmatically out of reach for experimental assessment, epidemiological modeling in an adapted framework could integrate them all. This would allow for an evaluation of the considered factors’ relative effects on whole epidemiological processes and lead to improved risk assessments for public health authorities. For example, an agent-based epidemiological modeling framework (Lequime et al. 2020a) was previously used to leverage empirical evidence for arthropod-borne diseases (Lequime et al. 2020b, Aubry et al. 2021) and could easily be parameterized to incorporate all these multiple factors, including dengue-mediated changes in mosquito behavior and life-history traits.

Although the findings presented in this review are key to understanding DENV epidemiology and developing efficient vector containment strategies, many pieces of this complex interaction puzzle are still missing. It is also important to investigate whether the effects outlined in this review are specific to the interaction between DENV and Ae. aegypti or could be generalized across arboviruses and mosquito populations or species. Some studies have found that infection with Zika virus reduced Ae. aegypti locomotor activity, survival, oviposition success, and fecundity, while increasing blood meal size (Padilha et al. 2018, Petersen et al. 2018). In comparison, chikungunya virus (Alphavirus) appeared to increase the number of eggs laid by Ae. aegypti (Mulatier et al. 2023), while reducing oviposition success and egg hatching (Resck et al. 2020). More than this relatively small sample of available studies is needed to draw a reliable conclusion. Still, it highlights the need for additional research efforts, coupled with appropriate epidemiological models accounting for specific virus–vector interactions and their outcome on vectorial capacity, as arboviruses represent a growing epidemiological burden on the global healthcare system.

Author contributions

Ioana Mateescu (Conceptualization [equal], Data curation [lead], Visualization [equal], Writing—original draft [lead], Writing—review & editing [equal]), and Sebastian Lequime (Conceptualization [equal], Project administration [lead], Supervision [lead], Visualization [equal], Writing—original draft [supporting], Writing—review & editing [equal])