AUTOMATED SEXING OF TSETSE PUPAE USING NEAR-INFRARED SPECTROSCOPY.

DETERMINATION DU SEXE DES MOUCHES TSETSE AU STADE PUPAL PAR SPECTROSCOPIE AU PROCHE INFRAROUGE AUTOMATISEE

A.G. Parker1 and F.E. Dowell2

1 International Atomic Energy Agency, Entomology Unit, FAO/IAEA Agriculture and Biotechnology Laboratories, Seibersdorf.
2 USDA-ARS, Grain Marketing and Production Research Center, Manhattan, Kansas, USA.

Résumé

            L’application de la technique du mâle stérile requiert le lâcher d’un grand nombre d’insectes stérilisés. De manière générale, il est préférable de ne lâcher que des mâles, et pour les mouches Tsé-tsé en particulier, il est essentiel de préserver les femelles pour assurer le maintien de la colonie. Aucune caractéristique liée au sexe n’étant connue au stade pupal, la séparation des sexes n’a été possible jusqu'à présent que par tri manuel des mouches Tsé-tsé au stade adulte. Nous montrons ici qu’il est possible de déterminer par spectroscopie au proche infrarouge (SPI) le sexe des mouches Tsé-tsé à l’état de pupe à un stade avancé du développement. Des différences sexe-spécifiques sont détectables environs dix jours avant émergence et un changement significatif du spectre apparaît juste avant l’émergence de l’adulte. Ce changement apparaît 5 jours avant émergence chez les femelles et environ 3 jours avant émergence chez les mâles. L’emploi d’un spectromètre proche infrarouge automatisé, initialement conçus pour la caractérisation de grains de blé, permet la séparation en fonction du sexe de pupes de Glossina pallidipes Austen dans les 5 jours précédant l’émergence à une cadence d’environ 1 pupe par seconde avec une précision supérieure à 80%. 

Summary

            The application of the Sterile Insect Technique requires the release of large numbers of sterilised insects. In general it is preferable to release only males, and for tsetse this is essential as the females are required for colony maintenance. In the past separation of the sexes could only be achieved by hand sorting in the adult stage as no sex specific character was known in the pupal stage of tsetse. We show that the sex of tsetse pupae may be determined by near-infrared (NIR) spectroscopy in the later stages of development. Sex specific differences are detectable from about 10 days before emergence and a marked change occurs in the spectra shortly before adult emergence. This change occurs at about 5 days before adult emergence in females and about 3 days before emergence in males. Using an automated NIR spectrometer designed for the characterisation of wheat kernels, pupae of Glossina pallidipes Austen may be separated according to sex from 5 days before emergence at about 1 per second with better than 80% accuracy.

Introduction

            For the sterile insect technique, it is necessary to sex the tsetse flies before sterilisation so that the females can be kept for the colony. In addition, it is preferable to release only sterile males as sterile females live longer than sterile males, so increasing the risk of flies becoming infected with trypanosomes and increasing disease transmission.

            Currently tsetse are sexed either by chilling and sorting individually, a process which is both labour intensive and potentially detrimental to the flies, or by self-stocking of production cages. This latter relies on the natural difference in pupal period between males and females, with females emerging somewhat earlier than males. To enhance the separation, pupae are transferred to a higher temperature at the point of first emergence to stimulate female emergence. Whilst this technique works, both timing and temperature are critical and it has proved difficult to implement in practice.

            Sexing in the pupal stage would have several advantages, principal of which is that pupae are much easier to handle than adults. Pupal sorting also means that the male pupae are available before emergence for various treatments, such as chilling to control emergence so as to synchronize emergence with the release schedule, irradiation and shipment to a release centre away from the rearing centre.

            Near infrared spectroscopy measures light absorption in the range 700 – 1700 nm, which includes the vibrational modes of various organic bonds. It is used for the identification of organic molecules, and quantitatively to measure the properties of various mixtures, including such things as protein content of grain, milk fat content and pharmaceutical products. During work on characterizing wheat grains, it was observed that the spectrum could identify individual grains infested with beetle larvae, and even under some circumstances identify the species and stage. It was therefore tested on a number of insect species to see what information could be obtained, and while it proved disappointing for many species it showed promise for separating tsetse pupae

Methodology

            The system we have been using is an automated single kernel spectrometer for analysing wheat grains. Tsetse pupae are close in size to wheat grains, and the system can be used with minimum modification for most tsetse species. Individual pupae are picked up by a vacuum system and deposited in the reading head. The spectrum is read in about 100 msec, and the pupa is then sorted into one of four bins based on the calibration criteria. Pupae can be fed at a rate of about one per second, but this could be increased to about four per second with improved mechanical handling.

            The single kernel near-infrared spectrometer operates in the range 950 – 1650 nm. The detector is a 256 diode array, so that each diode represents a wavelength range of about 3 nm. For the calibration, pupae are scanned one at a time and collected in individual emergence containers and the sex scored after emergence. Using this scoring, a calibration is derived using principal component analysis.

Results

            An interesting observation was that the spectrum changes appreciably as the pupae mature, and this change is different for males and females (Fig. 1.). Up to about 6 days before emergence the spectra are very similar, and sorting efficiency is correspondingly low. Over the following 2 days the females spectrum in the short wavelength region (950 – 1200 nm) rises whilst the male spectrum remains fairly constant. By 3 days before emergence the male spectrum changes in the same manner, making them once more similar until emergence, although sexing remains possible.

            The principal component analysis yields a set of beta coefficients for each of the 256 readings from the diode array (Fig. 2). The absolute value of the coefficients indicate the relative significance of areas of the spectrum for the sex separation, the sign not being important. It can be seen that the beta coefficients change markedly with pupal age, and a separate calibration is required for different age pupae.

Figure 1. Change in the male and female G. pallidipes pupa near-infrared absorption spectra over the last 6 days before emergence.

Figure 2. Beta coefficients of the principal component analysis of the near-infrared absorption spectra for G. pallidipes pupae at 5 and 1 day before emergence.

            The SKNIR system offers the advantages of sexing pupae up to 5 days before emergence (Fig. 3), at a rate of up to 80,000 pupae per day with an accuracy of 90 - 95%, sufficient for a colony of about 1 million producing females. With improvements in the mechanical handling, this could be improved to about 250,000 pupae per day, sufficient for a colony of 3 – 4 million producing females. Further, whilst most of the work so far has been on Glossina pallidipes tests so far indicate that multiple species can be sorted with a single calibration, simplifying the set up of the system.

Fig. 3. Efficiency of sex separation by near-infrared spectroscopy in G. pallidipes over the last several days before emergence.

            I would like to acknowledge the contribution to the development of this system of J. Throne, J. Baker and E. Maghirang, USDA-ARS, Grain Marketing and Production Res. Center, 1515 College Av., Manhattan, KS 66502, USA, R. Wirtz and M. Benedict, Centers for Disease Control and Prevention, Atlanta, GA, USA, H. Bossin and A. Robinson, Entomology Unit, FAO/IAEA Agriculture and Biotechnology Laboratory, Seibersdorf, Austria, A. Broce, Kansas State Univ., Dept. Entomology, Manhattan, KS, USA, and J. Perez-Mendoza, Montana State Univ., Dept. Entomology, Bozeman, MT, USA.