Discrimination of Hover Fly Species and Sexes by Wing Interference Signals

Abstract Remote automated surveillance of insect abundance and diversity is poised to revolutionize insect decline studies. The study reveals spectral analysis of thin‐film wing interference signals (WISs) can discriminate free‐flying insects beyond what can be accomplished by machine vision. Detectable by photonic sensors, WISs are robust indicators enabling species and sex identification. The first quantitative survey of insect wing thickness and modulation through shortwave‐infrared hyperspectral imaging of 600 wings from 30 hover fly species is presented. Fringy spectral reflectance of WIS can be explained by four optical parameters, including membrane thickness. Using a Naïve Bayes Classifier with five parameters that can be retrieved remotely, 91% is achieved accuracy in identification of species and sexes. WIS‐based surveillance is therefore a potent tool for remote insect identification and surveillance.


Supplementary Tables:
Table S1: The 30 studied species of insect pollinators from the family Syrphdae.Species that display weak sexual dimorphism according to the literature are highlighted in gray.M.mellinum

Figure S2 :
Figure S2: Membrane thickness distribution for each species and sex of 30 species of hover flies.The solid line in each plot represents the median, and the colored shading represents the variance.

Figure S3 :
Figure S3: Scatter plot of the weighted medians and IQRs of the membrane thickness dwing vs. the fringe heterogeneity λ0.The species list is sorted and color-coded based on genetic similarity.

Figure S4 :
Figure S4: Scatter plot of the weighted medians and IQRs of the membrane thickness dwing vs. the WIS bias β.The species list is sorted and color-coded based on genetic similarity.

Figure S5 :
Figure S5: Scatter plot of the weighted medians and IQRs of the membrane thickness dwing vs. the fringe amplitude α.The species list is sorted and color-coded based on genetic similarity.

Figure S6 :
Figure S6: Scatter plot of the membrane thickness dwing vs. the wing area Awing vs. the modulation depth M.

Figure
Figure S8: (A) Relationship between WIS features and ecological traits.(B) Species that display weak sexual dimorphism in body shape and size, according toTable 1, are indicated by gray circles.

Figure S11 :
Figure S11:Wing thickness variation for male and females across the hover fly family.The thickness parameter folds twice across the family, this somewhat explains the moderate (linear) correlation between genetic distance and wing thickness differences.

Figure S12 :
Figure S12: Confusion matrix and estimation of possible overlap between species based on estimated wing beat frequency f ̂.

Figure S13 :
Figure S13: Confusion matrix and estimation of possible overlap between species based on the estimated membrane thickness dwing.

Figure S14 :
Figure S14: Confusion matrix and estimation of possible overlap between species based on the estimated membrane thickness dwing and modulation depth M.

Figure S15 :
Figure S15: Confusion matrix and estimation of possible overlap between species based on the estimated membrane thickness dwing and fringe heterogeneity λ0.

Figure S16 :
Figure S16: Confusion matrix and estimation of possible overlap between species based on the estimated membrane thickness dwing and estimated wingbeat frequency f ̂.

Figure S17 :
Figure S17: Confusion matrix and estimation of possible overlap between species based on the estimated wing beat frequency f ̂ , estimated membrane thickness dwing and fringe heterogeneity λ0.

Figure S18 :
Figure S18: Confusion matrix and estimation of possible overlap between species based on the estimated membrane thickness dwing, modulation depth M and estimated wing beat frequency f ̂.

Figure S19 :
Figure S19: Confusion matrix and estimation of possible overlap between species based on the estimated membrane thickness dwing, modulation depth M, wing area A, and estimated wing beat frequency f ̂.

Figure S20 :
Figure S20: Confusion matrix and estimation of possible overlap between species based on the estimated membrane thickness dwing, modulation depth M, wing area A, and estimated wing beat frequency f ̂.

Table 1
, are indicated by gray circles.