Delimiting cryptic morphological variation among human malaria vector species using convolutional neural networks

Deep learning is a powerful approach for distinguishing classes of images, and there is a growing interest in applying these methods to delimit species, particularly in the identification of mosquito vectors. Visual identification of mosquito species is the foundation of mosquito-borne disease surveillance and management, but can be hindered by cryptic morphological variation in mosquito vector species complexes such as the malaria-transmitting Anopheles gambiae complex. We sought to apply Convolutional Neural Networks (CNNs) to images of mosquitoes as a proof-of-concept to determine the feasibility of automatic classification of mosquito sex, genus, species, and strains using whole-body, 2D images of mosquitoes. We introduce a library of 1, 709 images of adult mosquitoes collected from 16 colonies of mosquito vector species and strains originating from five geographic regions, with 4 cryptic species not readily distinguishable morphologically even by trained medical entomologists. We present a methodology for image processing, data augmentation, and training and validation of a CNN. Our best CNN configuration achieved high prediction accuracies of 96.96% for species identification and 98.48% for sex. Our results demonstrate that CNNs can delimit species with cryptic morphological variation, 2 strains of a single species, and specimens from a single colony stored using two different methods. We present visualizations of the CNN feature space and predictions for interpretation of our results, and we further discuss applications of our findings for future applications in malaria mosquito surveillance.     

A New Method for the Analysis of Soft Tissues with Data Acquired under Field Conditions

Analyzing soft-tissue structures is particularly challenging due to the lack of homologous landmarks that can be reliably identified across time and specimens. This is particularly true when data are to be collected under field conditions. Here, we present a method that combines photogrammetric techniques and geometric morphometrics methods (GMM) to quantify soft tissues for their subsequent volumetric analysis. We combine previously developed methods for landmark data acquisition and processing with a custom program for volumetric computations. Photogrammetric methods are a particularly powerful tool for field studies as they allow for image acquisition with minimal equipment requirements and for the acquisition of the spatial coordinates of points (anatomical landmarks or others) from these images. For our method, a limited number of homologous landmarks, i.e., points that can be found on any specimen independent of space and time, and further distinctive points, which may vary over time, space and subject, are identified on two-dimensional photographs and their three-dimensional coordinates estimated using photogrammetric methods. The three-dimensional configurations are oriented by the spatial principal components (PCs) of the homologous points. Crucially, this last step orients the configuration such that x and y-information (PC1 and PC2 coordinates) constitute an anatomically-defined plane with the z-values (PC3 coordinate) in the direction of interest for volume computation. The z-coordinates are then used to estimate the volume of the tissue. We validate our method using a physical, geometric model of known dimensions and physical (wax) models designed to approximate perineal swellings in female macaques. To demonstrate the usefulness and potential of our method, we use it to estimate the volumes of Barbary macaque sexual swellings recorded in the field with video images. By analyzing both the artificial data and real monkey swellings, we validate our method''s accuracy and illustrate its potential for application in important areas of biological research.     

Modelling Allee effects in a transgenic mosquito population during range expansion

ABSTRACT

Mosquitoes are vectors for many diseases that cause significant mortality and morbidity. As mosquito populations expand their range, they may undergo mate-finding Allee effects such that their ability to successfully reproduce becomes difficult at low population density. With new technology, creating target specific gene modification may be a viable method for mosquito population control. We develop a mathematical model to investigate the effects of releasing transgenic mosquitoes into newly established, low-density mosquito populations. Our model consists of two life stages (aquatic and adults), which are divided into three genetically distinct groups: heterogeneous and homogeneous transgenic that cause female infertility and a homogeneous wild type. We perform analytical and numerical analyses on the equilibria to determine the level of saturation needed to eliminate mosquitoes in a given area. This model demonstrates the potential for a gene drive system to reduce the spread of invading mosquito populations.     

Mosquitoes and other aquatic insects in fallow field biotopes and rice paddy fields

Fallow field biotopes that develop from abandoned rice fields are man‐made wetlands that provide new habitats for various aquatic animals. Although consideration of such biotopes generally focuses on their positive aspects, this study evaluated the negative aspects of establishing fallow field biotopes with regard to mosquito breeding sites. To determine whether fallow field biotopes become breeding habitats for vector mosquitoes, we evaluated mosquito fauna in fallow field biotopes and adjacent rice fields. We found larvae of Anopheles lesteri, Anopheles sinensis and Culex tritaeniorhynchus (all: Diptera: Culicidae) in the biotopes. Although abundances of mosquito larvae in the biotopes and rice fields were statistically similar, mosquito abundances in rice fields increased dramatically in August when the water level reduced after the rainy season. The abundance and variety of the mosquitoes' natural predators were greater in biotopes than in rice fields because the former are a permanent and stable aquatic environment. A generalized linear mixed model showed a negative effect of predator diversity on mosquito larvae abundance in both habitats. Although fallow field biotopes become breeding habitats for vector mosquitoes, establishing biotopes from fallow fields in order to protect various aquatic animals, including mosquito insect predators, may help to control mosquito breeding.     

Modeling the effects of Aedes aegypti’s larval environment on adult body mass at emergence

Mosquitoes vector harmful pathogens that infect millions of people every year, and developing approaches to effectively control mosquitoes is a topic of great interest. However, the success of many control measures is highly dependent upon ecological, physiological, and life history traits of mosquito species. The behavior of mosquitoes and their potential to vector pathogens can also be impacted by these traits. One trait of interest is mosquito body mass, which depends upon many factors associated with the environment in which juvenile mosquitoes develop. Our experiments examined the impact of larval density on the body mass of Aedes aegypti mosquitoes, which are important vectors of dengue, Zika, yellow fever, and other pathogens. To investigate the interactions between the larval environment and mosquito body mass, we built a discrete time mathematical model that incorporates body mass, larval density, and food availability and fit the model to our experimental data. We considered three categories of model complexity informed by data, and selected the best model within each category using Akaike’s Information Criterion. We found that the larval environment is an important determinant of the body mass of mosquitoes upon emergence. Furthermore, we found that larval density has greater impact on body mass of adults at emergence than on development time, and that inclusion of density dependence in the survival of female aquatic stages in models is important. We discuss the implications of our results for the control of Aedes mosquitoes and on their potential to spread disease.     

Wolbachia infection does not alter attraction of the mosquito Aedes (Stegomyia) aegypti to human odours

The insect endosymbiont Wolbachia pipientis (Rickettsiales: Rickettsiaceae) is undergoing field trials around the world to determine if it can reduce transmission of dengue virus from the mosquito Stegomyia aegypti to humans. Two different Wolbachia strains have been released to date. The primary effect of the wMel strain is pathogen protection whereby infection with the symbiont limits replication of dengue virus inside the mosquito. A second strain, wMelPop, induces pathogen protection, reduces the adult mosquito lifespan and decreases blood feeding success in mosquitoes after 15 days of age. Here we test whether Wolbachia infection affects mosquito attraction to host odours in adults aged 5 and 15 days. We found no evidence of reduced odour attraction of mosquitoes, even for those infected with the more virulent wMelPop. This bodes well for fitness and competitiveness in the field given that the mosquitoes must find hosts to reproduce for the biocontrol method to succeed.     

Plasmodium yoelii-Infected A. stephensi Inefficiently Transmit Malaria Compared to Intravenous Route

It was recently reported that when mosquitoes infected with P. berghei sporozoites feed on mice, they deposit approximately 100–300 sporozoites in the dermis. When we inoculate P. yoelii (Py) sporozoites intravenously (IV) into BALB/c mice, the 50% infectious dose (ID₅₀) is often less than 3 sporozoites, indicating that essentially all Py sporozoites in salivary glands are infectious. Thus, it should only take the bite of one infected mosquito to infect 100% of mice. In human subjects, it takes the bite of at least 5 P. falciparum-infected mosquitoes to achieve 100% blood stage infection. Exposure to 1–2 infected mosquitoes only leads to blood stage infection in approximately 50% of subjects. If mosquitoes carrying Py sporozoites inoculate 100–300 sporozoites per bite, and 1 to 2 mosquito bites achieve 50% blood stage infection rates, then this would suggest that the majority of sporozoites inoculated by mosquitoes into the dermis are not responsible for a productive infection, or that a significant number of sporozoite-infected mosquitoes do not inoculate any sporozoites. The objective of this study was to determine if this is the case. We therefore studied the infectivity to mice of the bites of 1, 2, 4, or 5–8 Py-infected mosquitoes. The bite of one Py sporozoite-infected mosquito caused blood stage infection in 41.4% (12/29) of mice, two bites infected 66.7% (22/33), four bites infected 75% (18/24), and five to eight bites infected 100% (21/21). These findings demonstrate that inoculation of sporozoites by mosquito bite is much less efficient than IV inoculation of Py sporozoites by needle and syringe. Such data may have implications for determining the best route and dose of administration to humans of our attenuated P. falciparum sporozoite vaccine, the scientific basis of which is immunity by bites from irradiated infected mosquitoes, and suggest that the challenge is to develop a method of administration that approximates IV inoculation, not one that mimics mosquito bite.     

In a warmer Arctic,mosquitoes avoid increased mortality from predators by growing faster

Climate change is altering environmental temperature, a factor that influences ectothermic organisms by controlling rates of physiological processes. Demographic effects of warming, however, are determined by the expression of these physiological effects through predator–prey and other species interactions. Using field observations and controlled experiments, we measured how increasing temperatures in the Arctic affected development rates and mortality rates (from predation) of immature Arctic mosquitoes in western Greenland. We then developed and parametrized a demographic model to evaluate how temperature affects survival of mosquitoes from the immature to the adult stage. Our studies showed that warming increased development rate of immature mosquitoes (Q₁₀ = 2.8) but also increased daily mortality from increased predation rates by a dytiscid beetle (Q₁₀ = 1.2–1.5). Despite increased daily mortality, the model indicated that faster development and fewer days exposed to predators resulted in an increased probability of mosquito survival to the adult stage. Warming also advanced mosquito phenology, bringing mosquitoes into phenological synchrony with caribou. Increases in biting pests will have negative consequences for caribou and their role as a subsistence resource for local communities. Generalizable frameworks that account for multiple effects of temperature are needed to understand how climate change impacts coupled human–natural systems.     

A Bayesian Hierarchical Model for Estimation of Abundance and Spatial Density of Aedes aegypti

Strategies to minimize dengue transmission commonly rely on vector control, which aims to maintain Ae. aegypti density below a theoretical threshold. Mosquito abundance is traditionally estimated from mark-release-recapture (MRR) experiments, which lack proper analysis regarding accurate vector spatial distribution and population density. Recently proposed strategies to control vector-borne diseases involve replacing the susceptible wild population by genetically modified individuals’ refractory to the infection by the pathogen. Accurate measurements of mosquito abundance in time and space are required to optimize the success of such interventions. In this paper, we present a hierarchical probabilistic model for the estimation of population abundance and spatial distribution from typical mosquito MRR experiments, with direct application to the planning of these new control strategies. We perform a Bayesian analysis using the model and data from two MRR experiments performed in a neighborhood of Rio de Janeiro, Brazil, during both low- and high-dengue transmission seasons. The hierarchical model indicates that mosquito spatial distribution is clustered during the winter (0.99 mosquitoes/premise 95% CI: 0.80–1.23) and more homogeneous during the high abundance period (5.2 mosquitoes/premise 95% CI: 4.3–5.9). The hierarchical model also performed better than the commonly used Fisher-Ford’s method, when using simulated data. The proposed model provides a formal treatment of the sources of uncertainty associated with the estimation of mosquito abundance imposed by the sampling design. Our approach is useful in strategies such as population suppression or the displacement of wild vector populations by refractory Wolbachia-infected mosquitoes, since the invasion dynamics have been shown to follow threshold conditions dictated by mosquito abundance. The presence of spatially distributed abundance hotspots is also formally addressed under this modeling framework and its knowledge deemed crucial to predict the fate of transmission control strategies based on the replacement of vector populations.     

Seasonal dynamics and habitat specificity of mosquitoes in an English wetland: implications for UK wetland management and restoration

We engaged in field studies of native mosquitoes in a Cambridgeshire Fen, investigating a) the habitat specificity and seasonal dynamics of our native fauna in an intensively managed wetland, b) the impact of water‐level and ditch management, and c) their colonization of an arable reversion to flooded grassland wetland expansion project. Studies from April to October, 2010 collected 14,000 adult mosquitoes (15 species) over 292 trap‐nights and ～4,000 pre‐imaginal mosquitoes (11 species). Open floodwater species (Aedes caspius and Aedes cinereus, 43.3%) and wet woodland species (Aedes cantans/annulipes and Aedes rusticus, 32.4%) dominated, highlighting the major impact of seasonal water‐level management on mosquito populations in an intensively managed wetland. In permanent habitats, managing marginal ditch vegetation and ditch drying significantly affect densities of pre‐imaginal anophelines and culicines, respectively. This study presents the first UK field evidence of the implications of wetland expansion through arable reversion on mosquito colonization. Understanding the heterogeneity of mosquito diversity, phenology, and abundance in intensively managed UK wetlands will be crucial to mitigating nuisance and vector species through habitat management and biocidal control.     

Construction of a physical model of the human carotid artery based upon in vivo magnetic resonance images

A method is described for construction of an in vitro flow model based on in vivo measurements of the lumen geometry of the human carotid bifurcation. A large-scale physical model of the vessel lumen was constructed using fused deposition modeling (a rapid prototyping technique) based on magnetic resonance (MR) images of the carotid bifurcation acquired in a healthy volunteer. The lumen negative was then used to construct a flow model for experimental studies that examined the hemodynamic environment of subject-specific geometry and flow conditions. The physical model also supplements physician insight into the three-dimensional geometry of the arterial segment, complementing the two-dimensional images obtained by MR. Study of the specific geometry and flow conditions in patients with vascular disease may contribute to our understanding of the relationship between their hemodvnamic environment and conditions that lead to the development and progression of arterial disease.     

Swarming mechanisms in the yellow fever mosquito: aggregation pheromones are involved in the mating behavior of Aedes aegypti

Mosquitoes of various species mate in swarms comprised of tens of thousands of flying males. In this study, we examined Aedes aegypti swarming behavior and identified associated chemical cues. Novel evidence is provided that Ae. aegypti females aggregate by means of olfactory cues, such as aggregation pheromones. Isolation of Ae. aegypti aggregation pheromones was achieved by aeration of confined mosquitoes and collection of associated volatiles by glass filters. The collected volatiles were identified through gas chromatography mass spectrometry (GCMS). Three aggregation pheromones were collected and identified as 2,6,6‐trimethylcyclohex‐2‐ene‐1,4‐dione (ketoisophorone) (CAS# 1125–21–9, t_R = 18.75), 2,2,6‐trimethylcyclohexane‐1,4‐dione (the saturated analog of ketoisophorone) (CAS# 20547–99–3, t_R = 20.05), and 1‐(4‐ethylphenyl) ethanone (CAS# 937–30–4, t_R = 24.22). Our biological studies revealed that the identified compounds stimulated mosquito behavior under laboratory conditions. The mechanism of mosquito swarm formation is discussed in light of our behavioral study findings. A preliminary field trial demonstrated the potential application of the isolated aggregation pheromones in controlling Ae. aegypti.     

Evolutionary potential of the extrinsic incubation period of dengue virus in Aedes aegypti

Dengue fever is the most common arboviral disease worldwide. It is caused by dengue viruses (DENV) and the mosquito Aedes aegypti is its primary vector. One of the most powerful determinants of a mosquito's ability to transmit DENV is the length of the extrinsic incubation period (EIP), the time it takes for a virus to be transmitted by a mosquito after consuming an infected blood meal. Here, we repeatedly measured DENV load in the saliva of individual mosquitoes over their lifetime and used this in combination with a breeding design to determine the extent to which EIP might respond to the evolutionary forces of drift and selection. We demonstrated that genetic variation among mosquitoes contributes significantly to transmission potential and length of EIP. We reveal that shorter EIP is genetically correlated with reduced mosquito lifespan, highlighting negative life‐history consequences for virus‐infected mosquitoes. This work highlights the capacity for local genetic variation in mosquito populations to evolve and to dramatically affect the nature of human outbreaks. It also provides the impetus for isolating mosquito genes that determine EIP. More broadly, our dual experimental approach offers new opportunities for studying the evolutionary potential of transmission traits in other vector/pathogen systems.     

Is the Mbita trap a reliable tool for evaluating the density of anopheline vectors in the highlands of Madagascar?

BACKGROUND: One method of collecting mosquitoes is to use human beings as bait. This is called human landing collection and is a reference method for evaluating mosquito density per person. The Mbita trap, described by Mathenge et al in the literature, consists of an entry-no return device whereby humans are used as bait but cannot be bitten. We compared the Mbita trap and human landing collection in field conditions to estimate mosquito density and malaria transmission. METHODS: Our study was carried out in the highlands of Madagascar in three traditional villages, for 28 nights distributed over six months, with a final comparison between 448 men-nights for human landing and 84 men-nights for Mbita trap, resulting in 6,881 and 85 collected mosquitoes, respectively. RESULTS: The number of mosquitoes collected was 15.4 per human-night and 1.0 per trap-night, i.e. an efficiency of 0.066 for Mbita trap vs. human landing. The number of anophelines was 10.30 per human-night and 0.55 per trap-night, i.e. an efficiency of 0.053. This efficiency was 0.10 for indoor Anopheles funestus, 0.24 for outdoor An. funestus, and 0.03 for Anopheles arabiensis. Large and unexplained variations in efficiency were observed between villages and months. CONCLUSION: In the highlands of Madagascar with its unique, highly zoophilic malaria vectors, Mbita trap collection was poor and unreliable compared to human landing collections, which remains the reference method for evaluating mosquito density and malaria transmission. This conclusion, however, should not be extrapolated directly to other areas such as tropical Africa, where malaria vectors are consistently endophilic.     

Wetlands and mosquitoes: a review

This review brings together information on mosquitoes, the diseases they transmit and the wetlands that provide habitats for the immature stages (eggs and larvae). Wetland values are mentioned, though the main literature on this does not generally overlap the mosquito issue. Mosquito management is overviewed to include: the use of larvicides, source reduction in intertidal wetlands and management in freshwater systems. There is not a great deal of information on mosquitoes and freshwater systems, except for constructed wetlands and they are considered separately. We then consider restoration mainly in the context of wetlands that have been the subject of habitat modification for mosquito control. Land use and climate change, as they affect mosquitoes and the diseases they transmit, are also reviewed, as this will affect wetlands via management activities. Finally the review addresses the critical issue of balancing health, both human and environmental, in an adaptive framework. It concludes that there is a need to ensure that both mosquito and wetland management communicate and integrate to sustain wetland and human health.     

Multiple blood feeding in mosquitoes shortens the Plasmodium falciparum incubation period and increases malaria transmission potential

Many mosquito species, including the major malaria vector Anopheles gambiae, naturally undergo multiple reproductive cycles of blood feeding, egg development and egg laying in their lifespan. Such complex mosquito behavior is regularly overlooked when mosquitoes are experimentally infected with malaria parasites, limiting our ability to accurately describe potential effects on transmission. Here, we examine how Plasmodium falciparum development and transmission potential is impacted when infected mosquitoes feed an additional time. We measured P. falciparum oocyst size and performed sporozoite time course analyses to determine the parasite’s extrinsic incubation period (EIP), i.e. the time required by parasites to reach infectious sporozoite stages, in An. gambiae females blood fed either once or twice. An additional blood feed at 3 days post infection drastically accelerates oocyst growth rates, causing earlier sporozoite accumulation in the salivary glands, thereby shortening the EIP (reduction of 2.3 ± 0.4 days). Moreover, parasite growth is further accelerated in transgenic mosquitoes with reduced reproductive capacity, which mimic genetic modifications currently proposed in population suppression gene drives. We incorporate our shortened EIP values into a measure of transmission potential, the basic reproduction number R₀, and find the average R₀ is higher (range: 10.1%–12.1% increase) across sub-Saharan Africa than when using traditional EIP measurements. These data suggest that malaria elimination may be substantially more challenging and that younger mosquitoes or those with reduced reproductive ability may provide a larger contribution to infection than currently believed. Our findings have profound implications for current and future mosquito control interventions.