Introduction

The southern house mosquito, Culex quinquefasciatus Say (Diptera: Culicidae), is a worldwide distributed vector of several pathogens, such as Wuchereria bancrofti Cobbold (Kasili et al., 2009), Dirofilaria immitis Leidy (Labarthe et al., 1998) and avian malaria (Farajollahi et al., 2011). Besides, this mosquito is also one of the main species responsible for the transmission of the arboviruses West Nile virus (WNV) and Saint Louis encephalitis virus (SLEV), among others (Farajollahi et al., 2011). Because of this, this species is of critical concern due to its impact on both public and veterinary health.

Since arboviruses are transmitted by the bite of infected mosquitoes between vertebrate hosts, activity of these pathogens is determined, at least in part, by the blood feeding habits of these vectors. Host preference of Culex mosquitoes is variable and it can be classified in four categories: (1) mammophilic (Muturi et al., 2008), (2) ornithophilic (Jansen et al., 2009), (3) herpetophilic (Janssen et al., 2015) and (4) generalist (Molaei et al., 2007). Additionally, some species may experience a host switch in its feeding habits, from avian to mammal hosts (Hancock & Camp, 2022). These species are of greater concern from an epidemiological standpoint because they can act as bridge vectors in the transmission of some arboviruses, such as SLEV and WNV (Kilpatrick et al., 2005).

Culex quinquefasciatus is generally considered an ornithophilic species in several areas around the globe (Takken & Verhulst, 2013). Despite its ornithophilic nature, this and other Culex species sometimes feed primarily on mammal hosts, including humans, in certain situations. Various studies have pointed out that in autumn there is an increased activity of SLEV and WNV in human populations (Kilpatrick et al., 2006; Spinsanti et al., 2008), suggesting that viral activity spills over to mammals from birds as a result of the mosquito host switch (Edman & Taylor, 1968; Kilpatrick et al., 2006). A few hypotheses have attempted to explain why primarily ornithophilic mosquitoes also feed upon mammals in some contexts. One of the most widely accepted hypothesis states that the shift on feeding behaviour of Culex mosquitoes is a consequence of the autumn migration of American robin (Turdus migratorius), the principal host in the northeastern US, generating an opportunistic change into mammal hosts (Kilpatrick et al., 2006). In South America in general and in Argentina in particular, studies regarding host switch are scarce. Nevertheless, a few limited studies have indicated a seasonal variation throughout the seasons (Stein et al., 2013; Beranek, 2018). Despite that, Argentina has acknowledged seasonal variation in the activity of SLEV in humans, which can be attributed to variations in mosquito host-feeding pattern (Spinsanti et al., 2008). In Southern latitudes, species such as T. migratorius are lacking, hence the “migration hypothesis” is very unlikely for many austral areas, being restricted only for some regions of the US.

A second hypothesis proposed by Burkett-Cadena et al. (2011) suggests that host breeding cycles drive the host shift of mosquitoes. According to this hypothesis, in hosts with parental care, during reproductive seasons there is a greater investment of energy in assuring the offspring survivorship, consequently producing an increase in susceptibility of being bitten by mosquitoes as a result of a decrease in defensive behaviours. This event leads to the detection of peaks of host use during periods of reproductive investment, in summer for birds and autumn for mammals (Burkett-Cadena et al., 2011).

While some other minor hypotheses, unsupported by field data, have been proposed, none of them have been specifically focused on vector biology. Numerous biological factors, including stress, metabolic rate, and the source of a blood meal, as well as environmental variables such as temperature and photoperiod, exert an influence on mosquito physiology that might affect various reproductive traits such as fecundity, development rate or survivorship. This set of factors gives rise to new nutritional requirements that, ultimately, may lead to seasonal variation in host choice by mosquitoes (Ciota et al., 2014; Costanzo et al., 2015; Gervasi et al., 2016; Yan et al. 2017, 2018). In the literature, several studies have evaluated the impact of these biological and environmental variables on mosquito reproduction, as mentioned before. However, due to the complex nature of mosquito biology, there are likely multiple interactions among variables that give rise to new responses that may not be observed individually. Therefore, the aim of the present study was to assess whether exists an interaction effect between the source of the host blood meal and seasonality (in terms of temperature and photoperiod) on three reproductive traits of Culex quinquefasciatus mosquitoes, fecundity, fertility, and hatchability. Our hypothesis states that the interaction between these two variables influence the reproductive outcomes, potentially leading to a seasonal shift in host selection, driven by a reproductive advantage. Considering the reported host preference changes in Cx. quinquefasciatus, in autumn we expect a greater number of eggs (fecundity) and larvae (fertility) in mosquitoes after feeding on a mammal host compared to an avian host, and the opposite relationship in summer.

Materials and methods

Establishment and maintenance of mosquitoes

Egg rafts of Culex quinquefasciatus were collected from a drainage ditch at Universidad Nacional de Córdoba Campus, Córdoba city, in February 2021. Each raft was individually maintained in plastic containers with one liter of distilled water. The hatched larvae were fed with 100 mg of liver powder three times per week until pupation. Pupae were then transferred to plastic emerging cages (21 cm x 12 cm) covered with a tulle-like fabric, containing distilled water but no food. Adults emerging from each raft were identified morphologically (Darsie, 1985) and molecularly (Smith & Fonseca, 2004) to ensure they corresponded to Cx. quinquefasciatus, since Cx. pipiens and its hybrids are also present in Córdoba (Branda et al., 2021). All adults were reared in a cardboard cage of 22.5 liters (28 cm x 36.5 cm) and provided ad libitum with a 10% sugar solution soaked in cotton pads placed on plastic cups. For long-term maintenance of the colony, 24-hour-starved mosquitoes were offered a blood meal from a restrained chicken twice a month. Four days after feeding, a plastic container with distilled water was placed inside the cage to allow engorged females to lay egg rafts. Batches of egg were collected and transferred to plastic containers (30 cm x 25 cm x 7 cm) in a proportion of 3 rafts per container, filled with three liters of distilled water. The hatched larvae were also fed with liver powder at the same proportion described above. Pupae were transferred to emerging cages, and adult mosquitoes were placed in the final cardboard cage. The colony has been maintained for over twenty generations in the Insectary of the Instituto de Virología “Dr. J.M. Vanella” (InViV). Room conditions were controlled at 28°C, with a 12L:12D photoperiod and 70% relative humidity.

Experimental design

Blood source

For experimental trials, avian hosts (live chicks of the species Gallus gallus) and mammalian hosts (live mice of the species Mus musculus, strain C57BL/6) were used to evaluate the effect of blood meal source on fecundity, fertility, and hatchability. Chicks were generously donated by the Bartolucci poultry farm (Córdoba, Argentina), while mice were commercially obtained from the Instituto de Investigación Médica Mercedes y Martín Ferreyra (CONICET - Universidad Nacional de Córdoba). In each trial, 24-hour-starved adult female mosquitoes were provided with a blood meal from restrained chicks or mice. Vertebrate hosts were offered to mosquitoes one hour before the lights were turned off and were kept for 3 hours during 2 consecutive gonotrophic cycles.

The experimental use of animals (G. gallus and M. musculus) was approved by the ethical committee at Facultad de Ciencias Médicas, Universidad Nacional de Córdoba (FCM-UNC) in compliance with the legislation regarding the use of animals for experimental and other scientific purposes (accession code: CE-2022-00518476-UNC-SCT#FCM).

Seasonality

To assess the effect of seasonality (photoperiod + temperature) on fecundity, fertility, and hatchability, a “typical” summer and autumn day from Córdoba city was simulated in an incubator, where mosquitoes were housed throughout the entire experiment. Summer conditions were as follows: T°_min = 22°C, T°_max = 28°C, photoperiod = 14L:10D. Autumn conditions were characterized by: T°_min = 16°C, T°_max = 22°C, and photoperiod = 10L:14D. The humidity levels were maintained at 60-70% for both conditions. Data for simulated conditions were obtained from Climate Data website (https://es.climate-data.org/).

Feeding trial design

The interaction effect of blood source (bird or mammal) and seasonality (autumn and summer) was evaluated during two consecutive gonotrophic cycles (I and II). Eight experimental colonies were established using egg rafts collected from the maintenance colony: mammal-autumn-I (MA-I), mammal-summer-I (MS-I), bird-autumn-I (BA-I), bird-summer-I (BS-I), mammal-autumn-II (MA-II), mammal-summer-II (MS-II), bird-autumn-II (BA-II), and bird-summer-II (BS-II).

The experimental colonies were maintained under controlled conditions to ensure that all adults were of the same size (Kauffman et al., 2017). Adult mosquitoes were placed in cardboard cages of 8.3 liters (21 cm x 24 cm) and were provided with ad libitum access to a 10% sugar solution, which was soaked in cotton pads placed on plastic cups.

Feeding trials were conducted at two time points: 5 days post-emergence (first cycle) and 14 days post-emergence (second cycle). Following each blood meal, female mosquitoes were anesthetized using CO₂ and classified as fully engorged (VI Sella’s stage), partially fed (I-V Sella’s stage), or unfed (I Sella’s stage), following the classification by Silva Santos et al. (2019). Fully engorged females were counted and separated into a separate cage to complete their gonotrophic cycle. Four days after feeding, a cup containing distilled water was placed inside the cage to facilitate oviposition. After each oviposition cycle, egg rafts were collected, counted, and transferred to a 12-well plate containing 3 mL of distilled water. They were then photographed to subsequently determine the number of eggs per raft. Rafts were maintained in each well until they hatched into L1 larvae, at which point they were preserved with 1 mL of 96% ethanol, and the number of L1 larvae per raft was counted.

Statistical analysis

The reproductive outputs measured for each experimental colony were fecundity, fertility, and hatchability. All analyses were performed using R Studio statistic software, v.4.2.1 (R Core Team, 2022).

Fecundity was defined as the number of eggs per raft and fertility as the number of L1 larvae hatched per raft. To evaluate the effect of blood meal source and seasonality on these two variables, a Generalized Linear Model (GLM; package MASS) with negative binomial error distribution and logarithmic link function was adjusted (Venables & Ripley, 2002). Fecundity and fertility served as the response variables, while the explanatory variables included blood source, seasonality, and gonotrophic cycle, all in a three-way interaction.

Hatchability (= hatching rate) was defined as the ratio of the number of larvae to the number of eggs per raft. A GLM was applied using a quasipoisson error distribution and a logarithmic link function. The response variable was the number of larvae, while the explanatory variables included blood source, seasonality and gonotrophic cycle as interaction effects. Additionally, the logarithm of the number of eggs was incorporated as an offset in the model.

Multiple comparisons among treatments were conducted using the Tukey’s honestly significant difference test (HSD Tukey), incorporating the Kramer (1956) correction for unbalanced data, also known as the Tukey-Kramer method. This approach allows to set the family-wise error rate to 0.05. These contrasts were performed using the multcomp package (Hothorn et al., 2008).

Model validation was assessed using a simulation-based approach to generate randomized quantile residuals to test goodness of fit (Q-Q plot), homoscedasticity, and outliers and influential points (Cook’s distance and Leverage plots). These analyses were conducted with the DHARMa package (Hartig, 2022).

Model accuracy via simulations for the fecundity and fertility models

To assess the accuracy of the results, a simulation approach was conducted using a Monte Carlo algorithm to simulate data and model parameters according to the methodology proposed by Carsey & Harden (2013). For this purpose, parameters including coefficients, treatment means and dispersion parameter (referred to as theta, θ), were extracted from the fecundity and fertility models. Initially, a new dataset consisting of 1000 values (sample size of 125 per treatment) was generated. This was achieved using the rnbinom function from the MASS package, specifying the treatment means (computed from model coefficients) and the theta parameter. Subsequently, a negative binomial GLM was fitted to the simulated data, and the expected means for each of the eight treatments were obtained from the model. This simulation process was repeated 1000 times. The ratio between the number of iterations in which the simulated means fell within the observed 95% confidence interval divided by the total number of simulations was computed as an approximation of the robustness of results.

Results

Effect of blood source and seasonality on fecundity and fertility

Two interaction effects were found to be statistically significant in both the fecundity and fertility models (Tables 1-4). The first interaction effect was found between blood source and seasonality (fecundity: LRT X² = 25.0, p = 7.72 x 10^-7; fertility: LRT X² = 13.82, p = 0.002), meaning that the effect of blood meal source on the number of eggs and larvae per raft varied with season (Tables 2, 3). In general, mosquitoes that fed on avian blood sources exhibited greater fecundity and fertility than those that fed on mammalian blood source, regardless of the season. When treatments were examined by pairwise comparisons, fecundity and fertility were 13% and 18% greater in autumn compared to summer for bird-fed mosquitoes, while they were 17% and 19% lower for mammal-fed mosquitoes (Fig. 1 A, C, Table 1). The second statistically significant interaction was between blood meal source and the gonotrophic cycle (fecundity: LRT X² = 9.29, p = 0.0023; fertility: LRT X² = 14.4, p = 0.00073). This indicated that the effect of blood type on the number of eggs and larvae per raft differed between the first and second feeding trial (Tables 2, 3). For mosquitoes that fed on mammalian blood, fecundity and fertility were 29% lower in the second cycle compared to the first. Although the difference was not statistically significant, fecundity and fertility showed a slight increase in the second cycle for bird-fed mosquitoes. During the second cycle, these outcomes were 41% and 40% higher, respectively, for bird-fed mosquitoes compared to mammal-fed mosquitoes (Fig. 1 B, D, Table 1).

Figure 1.

Table 1.

Table 2.

Table 3.

Table 4.

Effect of blood source and seasonality on hatchability

The overall hatchability reached 86%, with a range spanning from 76% for MA-II to a peak of 96% for MA-I, as outlined in Table 1 and depicted in Figure 2. The model found a statistically significant effect of blood meal source (LRT X² = 73.06, p = 0.0012) and an interaction effect between seasonality and gonotrophic cycle (LRT X² = 41.02, p = 0.0015) on hatching rates (Fig. 3, Table 4). Pairwise comparisons revealed a statistically significant 15% decrease in hatchability for mosquitoes from autumn from the first to the second gonotrophic cycle. Additionally, mosquitoes fed on mammals had greater hatchability than those fed on birds. In summer, hatchability during the first cycle was also higher than during the second cycle. These two trends were not found to be statistically significant.

Figure 2.

Figure 3.

Model accuracy via simulations for the fecundity and fertility models

When comparing the simulated means for each of the eight treatments (combinations of blood meal source, seasonality, and gonotrophic cycle) with the means calculated from the observed data a high level of accuracy exceeding 90% was observed. Most of the simulated means fitted within the original 95% confidence interval, with an accuracy rate greater than 98%. The greatest mismatching among simulated and original data were found in the treatment MS-I, with an accuracy of 93% and 95% for the fecundity and fertility model, respectively. Accuracy remained high despite variations in sample sizes and data dispersion (Figs. 4, 5).

Figure 4.

Figure 5.

Discussion

Our results revealed that there exists an interaction effect of blood meal source and seasonality on reproductive outputs of Cx. quinquefasciatus. Although we expected a higher fecundity and fertility in summer for bird-fed mosquitoes and higher outputs in autumn for mammal-fed mosquitoes, we found the opposite pattern.

Most studies have reported a greater fecundity and fertility in Cx. quinquefasciatus fed on different bird species (Akoh et al.,1993; Richards et al., 2012; Telang & Skinner,2019). This trend is also present in other species of Culex (Shroyer & Silvery, 1972) and Aedes mosquitoes (Harrison et al., 2021). The widely accepted hypothesis that has emerged to explain why avian blood produces a greater fecundity and fertility even in mosquitoes with a known mammal-preferred host choice focuses on the nutritional quality of host blood. Usually, it is addresses that because avian blood possesses nucleated and larger erythrocytes this offers greater quantities of proteins and energy than mammalian blood (Alto et al., 2014). However, this assumption lacks solid evidence due to studies regarding nutritional quality of avian blood and its impact in mosquito fitness are scarce. Actually, some studies have shown that mammalian blood are nutritionally richer than avian blood in terms of total protein content (Pearce 1977, Sant’Anna et al., 2010). Pre-diuresis, a physiological process where red blood cells and proteins are concentrated in the midgut by excreting large amounts of water, has been postulated as a mechanism for explaining the higher fecundity and fertility observed in bird-fed mosquitoes (Erram et al., 2022). Furthermore, avian blood has higher concentrations of isoleucine than mammalian blood, a crucial limiting amino acid necessary for egg production (Greenberg, 1951, Harrison et al., 2021).

The effect of temperature on fitness follows a unimodal relationship with a narrow optimal range. For mosquitoes of genus Culex and Aedes, this optimal band ranges from 20°C to 30°C, with an optimal at 28°C for species such as Cx. quinquefasciatus (Mordecai et al., 2019). Several studies have corroborated this trend, with higher fecundity and fertility towards the optimal of 28°C (Abouzied, 2017; Mogi, 1992), regardless of the blood source (bird or mammal). Regarding photoperiod, the effects are not as evident as temperature since data are somewhat contradictory. While Mogi (1992) found that mid and long day lengths induced greater fecundity, Costanzo et al. (2015) did not find differences of day length on fecundity. Because temperature and photoperiod are always coupled, in general their effects follow the same unimodal trend. Although our experiment was conducted within this optimal range, the highest fecundity and fertility in optimal conditions (summer) was present only in mammal-fed mosquitoes, while higher fitness in bird-fed mosquitoes was detected in autumn. This difference may indicate an interaction of seasonality with blood source.

While our findings may have certain limitations due to moderate sample sizes in some treatments and their reliance solely on laboratory experiments, they verify an interaction effect between blood meal source and seasonality, which had not been previously investigated or documented in any other study. Nevertheless, the precise reasons behind the observed pattern remain inconclusive. The highest levels of fecundity and fertility observed in bird-fed mosquitoes during autumn, when conditions are more challenging, may confer a reproductive advantage, leading to increased offspring production as an adaptive response to these adverse conditions. In contrast, during the more favorable conditions of summer, this pattern may not be as readily discernible. This proposed hypothesis needs further studies to be confirmed.

A second interaction was observed between blood meal source and gonotrophic cycle. We found that fecundity and fertility during the first cycle were similar between mosquitoes fed on chickens and those fed on mammals, and this was also similar with bird-fed mosquitoes from the second gonotrophic cycle, regardless of the season. The only difference we observed was between the fitness of mammal-fed mosquitoes from the second gonotrophic cycle and the earlier mentioned treatments. This pattern is not different from those found in other studies, as the first and second cycles are generally the most productive (Awahmukalah & Brooks, 1985; Christiansen-Jucht et al., 2015). Although we observed a slight increase of fecundity and fertility in bird-fed mosquitoes between cycles, it was statistically non-significant. To our knowledge, only two studies have reported this increasing pattern: an older one by Bennett (1970) and unpublished data from Obholz (personal communication), both conducted with Ae. aegypti. On the contrary, a decrease in fitness between cycles has been reported in other studies for Cx. quinquefasciatus (Richards et al., 2012; Telang & Skinner, 2019) fed upon birds or mammals. This is consistent with our findings regarding mouse-fed mosquitoes; however, because the fecundity and fertility of mouse-fed mosquitoes during the second cycle yielded the fewest egg rafts, this decreasing pattern might also be an effect due to the insufficient sample size.

In addition to assessing fecundity, we also investigated whether hatchability could account for the variations in fertility observed among different treatments. The only statistically significant trend we identified was a decrease in the percentage of hatched eggs from the first to the second gonotrophic cycle, but this was observed only in autumn mosquitoes. This decline in hatching rate may be attributed to a constant exposure to cooling conditions, due to the challenging environmental conditions during this season. This temperature-related effect aligns with findings from other studies (van der Linde et al., 1990). However, although our data suggest that changes in fertility may be linked to variations in hatching rate, it becomes apparent that fecundity has a more pronounced impact on larval production due to the effects of blood meal source and seasonality than hatching rate. Hence, the influence of hatchability on the number of larvae per raft is rather limited.

In summary, our study investigates the critical issue of host shifting patterns in Culex mosquitoes, which holds significant importance due to their role as bridge vectors between bird and human hosts. Our model suggests that changes in photoperiod and environmental temperature induce a different physiological state that requires the mosquito to feed on a blood meal source that results in higher fitness (increased fecundity and fertility). We predicted that mosquitoes fed on mammals in autumn will perform higher fecundity and fertility than those fed on birds in the same season. Unexpectedly, we observed the opposite pattern. Despite this, our study confirmed a significant interaction between blood meal source and seasonality, suggesting a complex interplay of factors affecting mosquito reproduction and its relationship with host use. Due to the observed opposite trend, we can consider three plausible explaining scenarios. Firstly, it is possible that additional factors, such as genetic mechanisms or variations in midgut microbiota composition, play a role in influencing host use patterns. Secondly, host switching might be explained by factors related to vector biology other than fitness. Lastly, it remains a possibility that Argentinian Cx. quinquefasciatus populations do not exhibit the same host-switching behaviour observed in US populations, and the observed interaction effects may be linked to different mosquito behaviours. Further investigations are warranted to corroborate our findings and gain a deeper understanding of the interaction between mosquito feeding pattern and seasonality. Ultimately, these investigations will provide valuable insights into mosquito-borne disease transmission and forecasting its emergence.

Acknowledgements

We would like to thank Dr. Agustín Quaglia for providing valuable help in statistical analysis. In addition, we greatly appreciate very helpful comments on the manuscript by anonymous reviewers. This study was supported by grant PICT 2018-1172 (FONCYT) and Consolidar Secyt-UNC (2018-2023).