Speaker: Tom Sherratt
Title: The Evolution of Mimicry in Hoverflies
Affiliation: Carleton University, Ottawa, Canada.
Date: Friday, February 9, 2007.
Place and Time: Room 101 - Love Building, 4:00-4:55 pm.
Refreshments: Room 204 - Love Building, 3:30 pm.

Abstract. Hoverflies (Diptera: Syrphidae) - sometimes referred to as flower flies - appear to gain protection from predators by resembling a variety of noxious models, including wasps, bumblebees and honeybees. However, beyond these basic assertions we know very little about mimicry in this group and why it has evolved.

In this seminar, I begin by reviewing the evidence that invertebrate, rather than vertebrate predators are the major selective force maintaining mimicry in this group. I then ask whether mimicry might confer additional benefits not mediated by predation, such as reducing interference competition when foraging on flowers.

While some mimetic hoverflies resemble their hymenopteran models closely, other hoverfly species bear only a crude likeness, and there has been continuing debate as to why the degree of mimetic similarity is not further improved by natural selection. In an earlier experiment, pigeons were trained to discriminate between images of wasps and non-mimetic flies and then presented with images of a range of hoverfly species to assess their response. Here, I describe the development of a neural network-based model to identify a candidate set of biometrical features that these pigeons may have employed when assessing the similarity of the hoverfly species to wasps. In highlighting the importance of certain features and the irrelevance of others, the approach may help explain why certain hoverfly species that appear to be poor mimics to humans are judged to be good mimics by birds. Finally, hoverflies exhibit behavioural mimicry, and there have been repeated suggestions that hoverflies "squawk" or "buzz" like their hymenopteran models when attacked. We have put this to a test, by quantitatively comparing the sound produced by a variety of non-mimetic flies, hoverflies, wasps and bees on attack, and have come up with some surprising results.