Horsefly object-directed polarotaxis is mediated by a stochastically distributed ommatidial subtype in the ventral retina

Significance

Insect compound eyes are a random array of 2 or more subtypes of optical units, the ommatidia. Some ommatidia may contain photoreceptors sensitive to polarized light, but their functional integration into the visual system has not been explained. Here, we report that horsefly retina contains 2 ommatidial subtypes that separately analyze polarization of light and color. Horseflies seek their prey by detecting polarized reflections from animal fur. We explain why horseflies are attracted to shiny and to blue objects. Understanding this mechanism will help in controlling fly disease vectors. This study gives an explanation for ommatidial subtype specialization that goes beyond color vision and which is likely a common feature of many insect eyes.

Abstract

The ventral compound eye of many insects contains polarization-sensitive photoreceptors, but little is known about how they are integrated into visual functions. In female horseflies, polarized reflections from animal fur are a key stimulus for host detection. To understand how polarization vision is mediated by the ventral compound eye, we investigated the band-eyed brown horsefly Tabanus bromius using anatomical, physiological, and behavioral approaches. Serial electron microscopic sectioning of the retina and single-cell recordings were used to determine the spectral and polarization sensitivity (PS) of photoreceptors. We found 2 stochastically distributed subtypes of ommatidia, analogous to pale and yellow of other flies. Importantly, the pale analog contains an orthogonal analyzer receptor pair with high PS, formed by an ultraviolet (UV)-sensitive R7 and a UV- and blue-sensitive R8, while the UV-sensitive R7 and green-sensitive R8 in the yellow analog always have low PS. We tested horsefly polarotaxis in the field, using lures with controlled spectral and polarization composition. Polarized reflections without UV and blue components rendered the lures unattractive, while reflections without the green component increased their attractiveness. This is consistent with polarotaxis being guided by a differential signal from polarization analyzers in the pale analogs, and with an inhibitory role of the yellow analogs. Our results reveal how stochastically distributed sensory units with modality-specific division of labor serve as separate and opposing input channels for visual guidance.