Search

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., Author summary Rapid and accurate identification of mosquitoes that transmit human pathogens is an essential part of mosquito-borne disease surveillance. Such identification can be difficult for mosquitoes that transmit malaria, as many are morphologically indistinguishable, including those in the Anopheles gambiae species complex. We photographed 1, 709 individual mosquitoes from 16 laboratory colonies housed at the Centers for Disease Control and Prevention to create a database of whole-body mosquito images. We present a methodology for image processing, data augmentation, and training and validation of a convolutional neural network (CNN). We applied this method to our mosquito image database, finding a 96.96% prediction accuracy for class identification and 98.48% for sex. Further, our best model accurately predicted images between 2 strains of a single species and between 2 storage methods of mosquitoes from the same colony. These results demonstrate that image classification with deep learning can be a useful method for malaria mosquito identification, even among species with cryptic morphological variation. We discuss the application of deep learning to mosquito identification in malaria mosquito surveillance.

Background Establishing the sources of reinfestation after residual insecticide spraying is crucial for vector elimination programs. Triatoma infestans, traditionally considered to be limited to domestic or peridomestic (abbreviated as D/PD) habitats throughout most of its range, is the target of an elimination program that has achieved limited success in the Gran Chaco region in South America. Methodology/Principal Findings During a two-year period we conducted semi-annual searches for triatomine bugs in every D/PD site and surrounding sylvatic habitats after full-coverage spraying of pyrethroid insecticides of all houses in a well-defined rural area in northwestern Argentina. We found six low-density sylvatic foci with 24 T. infestans in fallen or standing trees located 110–2,300 m from the nearest house or infested D/PD site detected after insecticide spraying, when house infestations were rare. Analysis of two mitochondrial gene fragments of 20 sylvatic specimens confirmed their species identity as T. infestans and showed that their composite haplotypes were the same as or closely related to D/PD haplotypes. Population studies with 10 polymorphic microsatellite loci and wing geometric morphometry consistently indicated the occurrence of unrestricted gene flow between local D/PD and sylvatic populations. Mitochondrial DNA and microsatellite sibship analyses in the most abundant sylvatic colony revealed descendents from five different females. Spatial analysis showed a significant association between two sylvatic foci and the nearest D/PD bug population found before insecticide spraying. Conclusions Our study shows that, despite of its high degree of domesticity, T. infestans has sylvatic colonies with normal chromatic characters (not melanic morphs) highly connected to D/PD conspecifics in the Argentinean Chaco. Sylvatic habitats may provide a transient or permanent refuge after control interventions, and function as sources for D/PD reinfestation. The occurrence of sylvatic foci of T. infestans in the Gran Chaco may pose additional threats to ongoing vector elimination efforts., Author Summary Triatoma infestans, a highly domesticated species and historically the main vector of Trypanosoma cruzi, is the target of an insecticide-based elimination program in the southern cone countries of South America since 1991. Only limited success has been achieved in the Gran Chaco region due to repeated reinfestations. We conducted full-coverage spraying of pyrethroid insecticides of all houses in a well-defined rural area in northwestern Argentina, followed by intense monitoring of house reinfestation and searches for triatomine bugs in sylvatic habitats during the next two years, to establish the putative sources of new bug colonies. We found low-density sylvatic foci of T. infestans in trees located within the species' flight range from the nearest infested house detected before control interventions. Using multiple methods (fine-resolution satellite imagery, geographic information systems, spatial statistics, genetic markers and wing geometric morphometry), we corroborated the species identity of the sylvatic bugs as T. infestans and found they were indistinguishable from or closely related to local domestic or peridomestic bug populations. Two sylvatic foci were spatially associated to the nearest peridomestic bug populations found before interventions. Sylvatic habitats harbor hidden foci of T. infestans that may represent a threat to vector suppression attempts.

Technologies based on RNA interference may be used for insect control. Sustainable strategies are needed to control vectors of Chagas disease such as Rhodnius prolixus. The insect microbiota can be modified to deliver molecules to the gut. Here, Escherichia coli HT115(DE3) expressing dsRNA for the Rhodnius heme-binding protein (RHBP) and for catalase (CAT) were fed to nymphs and adult triatomine stages. RHBP is an egg protein and CAT is an antioxidant enzyme expressed in all tissues by all developmental stages. The RNA interference effect was systemic and temporal. Concentrations of E. coli HT115(DE3) above 3.35 × 107 CFU/mL produced a significant RHBP and CAT gene knockdown in nymphs and adults. RHBP expression in the fat body was reduced by 99% three days after feeding, returning to normal levels 10 days after feeding. CAT expression was reduced by 99% and 96% in the ovary and the posterior midgut, respectively, five days after ingestion. Mortality rates increased by 24-30% in first instars fed RHBP and CAT bacteria. Molting rates were reduced by 100% in first instars and 80% in third instars fed bacteria producing RHBP or CAT dsRNA. Oviposition was reduced by 43% (RHBP) and 84% (CAT). Embryogenesis was arrested in 16% (RHBP) and 20% (CAT) of laid eggs. Feeding females 105 CFU/mL of the natural symbiont, Rhodococcus rhodnii, transformed to express RHBP-specific hairpin RNA reduced RHBP expression by 89% and reduced oviposition. Modifying the insect microbiota to induce systemic RNAi in R. prolixus may result in a paratransgenic strategy for sustainable vector control., Author Summary Rhodnius prolixus is an important vector of Chagas disease. The development of insecticide resistance in triatomines has raised the need for new control methods. We propose, as a proof-of-concept, the use of symbiotic bacteria expressing dsRNA in a paratransgenic approach to control vector-borne disease. We first show that ingestion of E. coli, producing long dsRNA specific for R. prolixus genes, can produce systemic RNAi in this insect. By targeting genes with antioxidant function (RHBP and catalase), we show that RNAi effects on nymphs and adult females are systemic and temporal, affecting development and fecundity. Finally, we show that the natural vector symbiont, R. rhodnii, also can be modified to induce systemic RNA interference. The E. coli system can serve to screen potential targets for development of a symbiont-based vector control product that then can be transferred to R. rhodnii.