It is possible that poor mimicry is an artefact of human perception, and that to the predators that generate the selection on the pattern, these mimics appear to be just as perfect as any other mimic. We have already seen that Dittrich et al. (1993) have suggested that the apparent imperfections of two very common syrphid mimics were only so for human eyes, because they appeared to be categorized as extremely good mimics by pigeons. Against this interpretation is the overall pattern of mimicry obtained from the pigeon experiment, which matched the ordering of mimetic quality of the human eye. It may well be true that certain species have managed to exploit some feature of bird perception in order to appear more perfect than they do to us, but in general this is not the case, and hence it is not a solution to the problem of imperfect mimicry. I have already shown above that the fact that birds can see UV does not seem to be an important feature of syrphid predation, and therefore is an invalid interpretation of the two anomalous patterns (cf. Cuthill and Bennett, 1993; Church et al., 2004).
Alternatively, the fact that bumblebee mimics are on average substantially bigger than wasp mimics might cause them to have different predators, and this might underlie the observed differences in mimetic quality and relative abundance (Edmunds, 2000). According to current thinking, however, this requirement for a visually selective agent restricts the potential candidates basically to birds, because it seems to rule out virtually all invertebrate predators. But is this true?
Kassarov (2003) tries to argue that not even birds are suitable candidates because even their visual abilities are inadequate for perceiving adequately the largest insect patterns (butterflies). He thinks that birds only see the flight movements of potential prey, relying on these to make decisions about whether to attack or not. He rejects the idea that birds can be the selective agents generating mimetic colour patterns. If true, then the patterns must have arisen by magic!
Dragonflies are perhaps the most obvious of insect predators that hunt visually, although their eyes seem to be adapted to detecting potential prey moving against the sky, rather than forming an image that would include the colour pattern (Corbet, 1999: 341). Most insect eyes appear to be primarily movement detectors, and the conventional wisdom is that they probably do not form an image sufficiently detailed to be able to generate selection for high quality mimicry. Some large aeshnids do take many bees (Corbet, 1999: 354, 379), but no-one has recorded them taking bee mimics. While they may sometimes avoid certain prey types, such as wasps (Alonso-Meija and Marquez, 1994; O'Donnell, 1996; Howarth, 1998: 16), unless this avoidance is visually based it is hard to imagine this contributing to the evolution of mimicry. Recent work (Kauppinen and Mappes, 2003; T. Sherratt, personal communication), however, shows that dragonflies are able to select between wasps and flies, and that this discrimination is largely visual: black-and-yellow stripes alone reduce rates of attack. Thus dragonflies may well have contributed to the evolution of mimicry in insect colour patterns.
The beewolves of the genus Philanthus (Sphecidae) are well-known bee predators (Osten-Sacken, 1894: 11). The European P. triangulum takes almost exclusively honeybees, mostly from flowers but also from the hive entrance; they never take any mimetic Diptera (Iwata, 1976: 150). Fabre (1913) tried to deceive one by offering it an Eristalis tenax, which it 'rejected with supreme contempt'! In the USA, P. bicinctus preys on bumblebees in Yellowstone
National Park, but is apparently never deceived by mimetic flies; it makes an interesting contrast with the sphecid Bembix pruinosa, a fly predator that feeds very often on E. tenax (Evans, 1966: 131). In neither case is there any evidence of any protection gained from the resemblance. This is not surprising, since Philanthus is well known to respond visually first to motion, and then when close odour becomes the crucial cue, stimulating the final pounce. This odour-directed prey capture explains why honeybee mimics are never captured. Similarly predators specializing on flies, such as Bembix, often take mimetic syrphids but do not take wasps, for the same reason (Evans and Eberhard, 1970: 52).
Robberflies (Diptera: Asilidae) are also voracious predators, but there is only a single study that suggests they can generate selection based on vision: a study of tiger beetles by Shelly and Pearson (1978) suggested that a robberfly may have been responsible for the evolution of both chemical and aposematic defences. However, the red pattern involved is just a block of colour, very crude in comparison to hoverflies. Brues (1946) suggested that asilids had a 'fondness' for worker honeybees, but this seems unlikely, given the evidence of all the prey records (listed at www.geller-grimm.de/catalog/lavigne.htm). He thought that because they frequently catch E. tenax, this meant that 'to the insect eye Eristalis really looks like a bee'. However, the appropriate null hypothesis of no ability to discriminate, that asilids merely catch both because they co-occur in the same habitats, has never been tested.
Sphecidae (Hymenoptera) are well-known insect predators, but members of only two of the subfamilies (Nyssoninae and Crabroninae) take syrphids (Bohart and Menke 1976; Iwata, 1976). Bembix (Nyssoninae) are large, very fast-flying wasps that deliberately target flies on flowers, or swarming males, and hence often take a large number of muscids, tabanids and syrphids. In Tsuneki's (1956) Japanese study, for example, B. nipponica took prey belonging to 11 families of flies overall, but in some sites and years syrphids formed almost half the prey (47%), mainly Eristalis cerealis. A large number of the hoverflies caught were mimetic, such as the wasp-like Takaomyia and Chrysotoxum, and the beelike eristalines. Large flies took disproportionately longer to find, capture, subdue and bring back to the nest: although it took on average 0.7 s longer to catch a syrphid rather than a non-syrphid, this difference was not significant and there was no evidence that syrphids were harder to catch than other flies. The other subfamily, the Crabroninae, are virtually all dipteran specialists, and the Crabronini are especially significant as hoverfly predators. A number of genera are important, especially Ectemnius. At least one species, E. cavifrons, seems to be a syrphid specialist; it is the commonest species in the UK (Pickard, 1975). The great majority of the prey in Pickard's study consisted of poor mimics (Syrphus spp. and Episyrphus balteatus). There was some selectivity involved because small dark species were greatly underrepresented, and no Eristalis were taken at all: Pickard thought their resemblance to honeybees might have protected them. Thus discrimination on the basis of colour patterns is possible, with a preference for yellow-and-black species. Some other studies have suggested that visual mimicry may protect against solitary wasps (e.g. predation on salticid spiders: Edmunds, 1993).
Social wasps (Spradberry, 1973: 141) take a huge variety of prey, but they concentrate on adult Diptera: some colonies have been recorded as taking up to 84% flies. Hornets can certainly take a lot of honeybees. It is usually thought very unlikely that these predators identify their prey visually, but recently (Tibbetts, 2002) Polistes wasps have been shown to identify individual colony members via variation in facial markings. Such a sophisticated ability indicates the capability for social wasps to generate natural selection for visual mimicry: this needs testing.
It is also possible that spiders make the sort of visual mistakes that would select for mimetic colour patterns, even though Bristowe (1941, vol. 2: 319) states that 'the yellow and black wasp-like appearance of certain syrphids is of no avail against spiders'. In fact spiders treat social wasps and bees with great caution, and only the largest species can tackle them successfully. Pocock (cited in Osten-Sacken, 1894: 11) noticed that Agelena labyrinthica used special precautions before overpowering a honeybee enmeshed in the web, whilst they pounced immediately on normal prey: when offered Eristalis, spiders approached and finally killed them, but used the same precautions as for honeybees.
Spider webs are in general not very good at catching hoverflies (Nentwig, 1982) since these flies are too large, strong and active. Insects with kinetic energies of about 150 ^J are able to fly straight through a web, and those with energies greater than about 500 ^J always do. The weight and flight speeds of syrphids indicate kinetic energies between 25 and 500 J and therefore the smaller species should generally get caught, while the larger species (Eristalis spp., for example) should be able to ignore webs altogether. After becoming entangled in a web, insects differ considerably in their behaviour, and these differences determine whether they escape. Insects such as syrphids that react to web entanglement by continuous vigorous activity are able to escape in the few seconds available before the spider attacks, and syrphids weighing as little as 9 mg escape rather easily: most syrphids are larger and more powerful than this. Orb-web spiders (Araneus diadematus) studied by Myers (1935) seized non-mimetic Diptera such as Calliphora immediately with no precautions, and never wrapped them in silk. When wasps were the victims, the spiders would carefully rotate the prey, showing great skill and alacrity in avoiding both the mouthparts and the apex of the abdomen, and swathe it in silk until completely helpless before biting near the centre of the dorsal surface - the safest position. Honeybees and Eristalis were, if tackled at all, treated with great caution and were nearly always swathed in silk. Thus spiders treated their victims differently, not according to size and vigour, but according to the perceived risk; Eristalis was treated like a bee rather than a fly of the same size.
Flower spiders (Thomisidae), especially Misumena, have evolved to catch flower-visiting insects, and most of their prey are syrphids, honeybees and bumblebees (see Morse, 1986). Only a single study has addressed whether colour patterns might be protective. About half the individuals studied by Tyshchenko (1961) would avoid both wasp models and their mimics, and the other half would eat them; the reluctance of the former group to attack mimics was in proportion to their visual similarity to the model. Thus visually based predation by spiders needs systematic study, since it too seems perfectly capable of generating selection for mimicry.
Which birds might be candidates for agents of selection for mimetic colour patterns? Reviewing Palaearctic bird diets (using Cramp, 1977-1994) does not get us very far since really we need information about whether birds undergo the process of learning to avoid models, and whether they confuse models with mimics, rather than data about the endpoint of the learning process where the birds never take either (i.e. the adult and nestling diets that are normally reported). Bee-eaters (Merops apiaster) feed on both models and mimics, but they hawk in the open savanna on hot days; this is quite different from the habitat of most hoverfly mimics, which are overwhelmingly forest dwellers (Speight et al., 1975; Maier, 1978; Speight, 1983) and avoid the hot midday (Gilbert, 1985). No protective effect occurs with this bird, since it specializes on wasps and bees, preferring them to all other prey. Hirundines such as the swallow (Hirundo rustica) also take syrphids and honeybees, but with their high-speed aerial scooping feeding method it is unlikely that they perceive the colour patterns before or after capture. Spotted flycatchers also take both wasps and hoverflies, but they are not deceived by the resemblance (Davies, 1977), and have an effective method of dealing with the venom of wasps. Thus for these birds, wasps and bees may not be noxious but instead may form part of their normal diet; it is possible that syrphids have longer handling times and are therefore unprofitable.
More likely candidates are birds such as Phylloscopus, Sylvia and Hippolais warblers, and others such as stonechats (Saxicola torquata). All these feed on syrphids, but we know virtually nothing about their selectivity among syrphid species. What we would be looking for would be evidence that: (i) birds had contact with noxious models; (ii) they also took syrphids; and (iii) the spectrum of syrphids upon which they fed was biased towards non-mimetic species. This sort of evidence is amazingly sparse in the literature. For example, Greig-Smith and Quicke (1983) noted that stonechats fed many warningly coloured ichneumonids and large numbers of syrphids to their nestlings, but we do not know what kinds of syrphids these were, and hence whether they might have been mimics. Similarly, we know that wheatears (Oenanthe oenanthe) feed bees and 'large Diptera' to their older nestlings (Cramp, vol. 5: 779), but were these large Diptera bee mimics? Then there are other birds such as Ficedula flycatchers, Acrocephalus warblers, and small passerines such as wagtails (Motacillidae), redstarts, robins and titmice (Paridae). The diet evidence suggests that these birds are minor or insignificant as hoverfly predators, but this is based only on samples of prey caught by experienced adults. Their reluctance to feed on syrphids may be entirely learned behaviour, taught to them by disastrous experiences they had when fledglings.
It is thus certainly possible that spiders and wasps are the main agents of selection for the smaller wasp mimics, whilst birds are the selective agents for bumblebee mimics. Superficially this is an attractive explanation because the much cruder visual abilities of the invertebrates would mean that imperfection of the colour pattern would not matter. However, until we know more about visual aspects of their predation, current knowledge really rules them out. Thus at the moment we can only conclude that inexperienced fledgling birds must be the selective agents responsible for the evolution and maintenance of mimicry in syrphids across the mimicry spectrum, but very little is known about their foraging behaviour. Birds that swoop down on flower-visiting insects from perches are probably the major candidates (see below). Since very high mortalities occur between fledging and recruitment into the adult population, the numbers of such young birds are probably very high relative to those of breeding pairs of adults (the usual density estimates), and hence their selective impact on syrphids might be very large.
Was this article helpful?