Assessing Near-Infrared Spectroscopy (NIRS) for Evaluation of Aedes aegypti Population Age Structure

Abstract

Given that older Aedes aegypti (L.) mosquitoes typically pose the greatest risk of pathogen transmission, the capacity to age grade wild Ae. aegypti mosquito populations would be a valuable tool in monitoring the potential risk of arboviral transmission. Here, we compared the effectiveness of near-infrared spectroscopy (NIRS) to age grade field-collected Ae. aegypti with two alternative techniques-parity analysis and transcript abundance of the age-associated gene SCP1. Using lab-reared mosquitoes of known ages from three distinct populations maintained as adults under laboratory or semi-field conditions, we developed and validated four NIRS models for predicting the age of field-collected Ae. aegypti. To assess the accuracy of these models, female Ae. aegypti mosquitoes were collected from Maricopa County, AZ, during the 2017 and 2018 monsoon season, and a subset were age graded using the three different age-grading techniques. For both years, each of the four NIRS models consistently graded parous mosquitoes as significantly older than nulliparous mosquitoes. Furthermore, a significant positive linear association occurred between SCP1 and NIRS age predictions in seven of the eight year/model combinations, although considerable variation in the predicted age of individual mosquitoes was observed. Our results suggest that although the NIRS models were not adequate in determining the age of individual field-collected mosquitoes, they have the potential to quickly and cost effectively track changes in the age structure of Ae. aegypti populations across locations and over time.

Keywords: NIRS; SCP1; Sonoran; aging; mosquito; parity.

Figure 1

NIRS scanning and typical spectra. (A). Examples of mosquito positioning under the NIRS probe. The NIRS probe was positioned 3 mm above the Spectralon plate and ~1.5 mm above the mosquito thorax. (B). Average log(1/R) spectra from 1- (green line; n = 53), 9- (red line; n = 53) and 18-day-old (blue line; n = 60) mosquitoes. Wavelength across the visible and near-infrared spectrum is indicated on the X-axis.

Figure 2

Location of field collections of Ae. aegypti mosquitoes in Maricopa County, Arizona. Aedes aegypti mosquitoes were collected during the monsoon seasons (approximately July through October) in southeastern Maricopa County and included the cities of Chandler, AZ, and Gilbert, AZ (A). Mosquitoes were collected from 20 paired collection sites (B) using BG sentinel traps (3 traps distributed per site) for one night each week throughout the collection period. A total of 201 female Ae. aegypti in 2017 and 172 in 2018, representing ~15% of the total number of females collected, were used for parity analysis and SCP1/NIRS age predictions.

Figure 3

Development of four models for NIRS age grading of Ae. aegypti mosquitoes. Predictive NIRS models for the age of female Ae. aegypti were generated for four different treatment groups of laboratory-reared mosquitoes. The UGAL lab model (A) was generated from 770 laboratory-reared UGAL female Ae. aegypti maintained at 27 °C and 70% RH. The UGAL semi-field model (B) also utilized the UGAL line but maintained 748 adult mosquitoes in shaded field cages in Tucson, AZ. The Tucson model (C) was generated from 540 female Ae. aegypti from an F4 line established from eggs collected in Tucson, AZ, and maintained under laboratory conditions. Finally, the Maricopa model (D) was generated from 475 female Ae. aegypti using an F3 line established from Ae. aegypti eggs collected at our actual collection site in southeastern Maricopa County and maintained under laboratory conditions. For all models, open circles represent the calibration set and show mean predicted age at each timepoint of all mosquito spectra used in the model calibration. Solid triangles represent the test set and show the mean age prediction from mosquito spectra withheld from the model calibration and used to validate the model. The number of mosquitoes utilized for the prediction and test set, the spectral range and bend regions selected from the model, the number of factors and other parameters describing each model are shown in Table 1.

Figure 4

Parity status of field-collected mosquitoes relative to NIRS age predictions. Female Ae. aegypti from our field-collected samples in 2017 (A) and 2018 (B) were separated based on parity status. NIRS spectra for field-collected nulliparous and parous mosquitoes were acquired, and NIRS age predictions for the four models were determined. As expected, field-collected parous mosquitoes were scored significantly older than nulliparous mosquitoes by each of the NIRS models. Significance for each pairing was determined using the Kruskal–Wallis ANOVA followed by a Dunn’s post hoc test (significance is only shown between parous and nulliparous comparisons * = p < 0.05; ** = p < 0.01, *** p < 0.0001).

Figure 5

Scatterplots comparing the age of individual field-collected female Ae. aegypti predicted with SCP1 or the four NIRS models. For 2017 and 2018, the scatterplots show the relationship of individual age predictions based on SCP1 (X-axis) and each of the NIRS models (Y-axis). (A). UGAL lab, (B). UGAL semi-field, (C). Tucson and (D). Maricopa models.