Article : the garden mimics.

Edmunds, Malcolm Available at http://clok.uclan.ac.uk/1620/

Edmunds, Malcolm (2008) Hoverflies: the garden mimics. Biologist, 55 (4). pp. 202-207. ISSN 0006-3347

It is advisable to refer to the publisher’s version if you intend to cite from the work.

For more information about UCLan’s research in this area go to http://www.uclan.ac.uk/researchgroups/ and search for .

For information about Research generally at UCLan please go to http://www.uclan.ac.uk/research/

All outputs in CLoK are protected by Intellectual Property Rights law, including Copyright law. Copyright, IPR and Moral Rights for the works on this site are retained by the individual authors and/or other copyright owners. Terms and conditions for use of this material are defined in the policies page.

CLoK Central Lancashire online Knowledge www.clok.uclan.ac.uk Hoverflies: the garden mimics

Mimicry offers protection from predators by convincing them that their target is not a juicy morsel after all. it happens in our backgardens too and the is an expert at it.

Malcolm overflies are probably the best the mimic for the model and do not attack Edmunds known members of tbe or- it (Edmunds, 1974). is far more Hder Diptera after houseflies, blue widespread in the tropics than in temperate bottles and mosquitoes, but unlike these lands, but we have some of the most superb they are almost universally liked examples of mimicry in Britain, among the by the general public. They are popular hoverflies. because of their bright colours and darting Apparent Batesian mimics occur in every flight, interspersed with hovering around garden throughout the summer months, garden flowers. Many of these brightly including the droneflies tenax, E. coloured hoverflies look like or pertinax and E. arbustorum all of which - and they use this similarity to protect resemble honeybees (Apis mellifera) (Fig- themselves. ure 1), the black and yellow boverflies Batesian mimicry is where a palatable ribesii and S. vitripennis which Title image: A form of equestris (the mimic) gains protection from resemble social wasps { vulgaris mimicking Bombus resembling a noxious animal (the model) so and related species) (Figure 2), and the pascuorum that predators are deceived into mistaking furry black and yellow and white hover-

202 Biologist I Volume 55 Number 4, November 2008 Hoverflies I IOB

Figure 1. Eristaiine hoverfties, from ieft to right: Eristalis ßeriinax, Eristalis arbustorum (both honeybee mimics), Heiophilus pendulus (a mimic) and Eristaiis intricarius (bumbiebee mimic). These hoverflies are cioseiy reiated in the subfamiiy Eristaiinae and have a iooped vein distally on the wing which is absent from many other hoverfiies. and Eristalis intri- make predictions on the assumption that carius (Figures 3 and 1) which resemble hoverflies are Batesian mimics and then . But do they actually derive test these predictions. protection against birds because of their similarity to hymenopterans or are they Ecology and Batesian mimicry coloured like their respective models for In 1994 Brigitte Howarth surveyed three reasons unrelated to mimicry? areas of semi-natural ancient woodland in Although colour is important in many Lancashire and Cumbria each week from insects for defence against predators, it can April until October, recording all hover- also be important in thermo-regulation, flies and hymenopterans seen on standard male-male competition and courtship. In walks. Some were identified to species, oth- hoverflies the abundance of species that ers to a group of similarly coloured species. appear to mimic hymenopterans suggests The five common and wasp groups were very strongly that in these species colour honeybees, social wasps, black and yellow must have some defensive value. Seventy bumblebees {, B. lucorum years ago Mostler (1935) showed that sev- and B. hortoi'um), black, yellow and rusty- eral species of hirds will eat hoverflies hut red bumblebees (Bombus pratorum) and quickly leam to avoid attacking wasps, brown bumblebees {B. pascuorum). and they then also refuse to attack the Her first prediction was that if hoverflies wasp-like hoverflies Serieomyia and Chry- are Batesian mimics, then they should sotoxum (Figure 2). In the same way he occur at the same season as the models also showed that droneflies are protected and be rarer than the models. If they oc- by their resemblance to honeybees while cur at other times or are commoner than bombylans is protected by its the models, then birds are likely to catch similarity to bumblebees (Figure 3). a mimic and find it tasty before they have These experiments were in an aviary, caught a model and learned to avoid in- though, and obtaining proof that any hov- sects with that colour pattern; in such erfly really does gain protection through cases, mimicry is unlikely to evolve. Batesian mimicry in the field is not at all All of the hoverflies that resembled easy: one would need to show that birds bumblebees (including Volucella bomby- (or otber predators) kill fewer flies that lans, berberina and mimic wasps than similar flies that do not superbiens., Figure 3) were indeed much mimic wasps. rarer than their models, so too were the One way of tackling this problem is to wasp-like Serieomyia silentis and Chiy-

Figure 2. Wasp-like hoverfiies, from ieft to right: , Episyrphus batteatus. the marmaiade (these are both 'poor' mimics), Serieomyia silentis and Chrysotoxum festtvum (two 'good' mimics with more wasp-iike patterns and, in Chrysotoxum, conspicuous antennae).

Volume 55 Number 4, November 2008 I Biologist 203 IOB I Hoverflies

Figure 3. mimics, from top left to bottom rigtit: typical form mimicking Bombus terrestris, Criorhina berberina oxyacantbae mimicking Bombus pascuorum, a form of Merodon equestris mimicking Bombus pascuorum, Che'ilosia iiiustrata mimicking Bombus pratorum. Voiuceiia bombylans typica mimicking Sombus lapidarius and Voiuceita bombylans piumata mimicking Somûus terrestris.

sotoxum spp. (Howarth and Edmunds, mimicry may be breaking down. However, 2000). These are all insects that hear a many years ago Jane Brower (1960), using very close resemhlance to their models so mealworm prey with caged birds, showed can he called 'good mimics'. But the yellow that if a model is sufficiently nasty then and black Hdophilus spp. (Figure 1) and mimics can still gain some protection even Syrphus spp. and the orange and hlack when they outnumber the model by nine Episyrphus baiteatus (Figure 2) were all to one. So do superabundant droneflies and much more numerous than their supposed wasp-like hoverflies indeed benefit from wasp models, as also were droneflies com- Batesian mimicry? pared with their honeybee models. To the In her survey work Brigitte Howarth human eye many of these abundant hov- made a second prediction: that if mimicry erflies (particularly the wasp-like ones) occurs then hoverfly numbers should be are not as similar to their supposed models positively related to numbers of their sup- as are the rarer 'good mimics' so they can posed models. If there is no mimicry then be referred to as 'poor mimics'. there should be no necessary relationship Dittrich et al. (1993) trained pigeons to between dronefly and honeybee numbers. avoid pictures of wasps but to peck at flies, In collaboration with Francis Gilbert this and found that the yellow and black hov- was tested by means of a General Linear erflies that looked like wasps were better Model (GLM) to explore how dronefly num- protected than black flies. But pigeons at- bers were related to a variety of environ- tacked the orange and black Episyrphus mental factors (see Table 1). As one would baiteatus the least, even though to our eyes expect, dronefly numbers were significantly it is a 'poor mimic'. This work suggests that affected by time of year, habitat (woodland, these apparently poor mimics may still be scrub or pasture), temperature and other protected by Batesian mimicry because hoverflies (more droneflies were present on birds perceive them differently from hu- warm days and when other flies were also mans. common). However, there was also a sig- More recently Azmeh et al (1998) found nificant positive relationship to numhers that these very abundant but poorly mi- of honeybees but a negative relationship to metic wasp-like hoverflies are not so com- numhers of other hymenopterans (Howarth mon in ancient forests of Poland and Rus- et al, 2004). This is precisely what one sia as they are in secondary forests, fields would expect if droneflies are indeed mim- and gardens, so it is possible that they have icking honeybees. There were similar posi- increased due to human-induced changes tive relationships between numbers of the to the environment. If true, then the pro- smaller honeybee mimic Eristalis arbusto- tection they receive may be minimal and rum and honeybee numbers, and between

204 Biologist I Volume 55 Number 4, November 2008 Hoverflies I IOB both Syrphus spp. and Episyrphus baitea- Table 1. Results of General Linear Model showing that there are significant relationships tus numbers and social wasp numbers. (déviances) between dronefly numbers (Eristaiis ienax & £. pertinax) and all six variables, All of these hoverflies are often more and that for three of the four covariates the relationship is positive. Thus there are more droneflies when it is warm, when there are more honeybees and when there are more numerous than their models, hut it is diffi- hoverflies present. The six variables listed account for 65% of the variance in dronefly cult to think of a good reason why this cor- numbers, relation should occur unless the hoverflies are indeed gaining protection through Ba- Variable Deviance d.f. P Slope of tesian mimicry. Similar significant posi- covariate tive relationships between hoverfly num- Month 63.6 5 <0.001 bers and model numbers were found for Habitat 177.0 7 <0,001 several of the 'good mimics' (Table 2), hut Temperature 37.8 1 <0,001 + not for all of them. Perhaps the numhers Numtjers of honeybees 7.0 1 <0.01 + of these insects are too small for there to Numbers of other be any additional advantage to be gained hymenopterans 29,4 1 <0,001 - from evolving similar activity patterns to the model. Numbers of other hoverfiies 119.4 1 <0,001 + An exception to this relationship for the

superabundant hoverflies is Heiophiius residuai 460.7 196 spp.: numbers of Heiophiius are not related 1306.0 212 to numbers of social wasps (which they re- total

semble in colour), but when the GLM was % variance accounted for = 100 * (1306 - 460.7)/1306 = 65% run again with the five main hymenop- P - probability; d.f. - degrees of freedom. teran groups separately it was found that Heiophiius numbers are significantly and positively related to numhers of honeybees. Figure 4) (Golding and Edmunds, 2000). One possihle reason for this relationship The conclusion from this work was that is that Heiophiius is closely related to the droneflies have modified their foraging honeybee mimics Eristalis tenax, E. per- behaviour to he more Uke that of their tinax and E. arbustorum, and there may honeybee models than of other hoverfiies, simply not have been enough time or suffi- yet dronefiies are feeding only themselves cient selective advantage to be gained from while honeybees are collecting food for the altering this ancestral pattern after Heio- colony. The most plausible reason why phiius evolved its wasp-like colouration. this should have occurred is because it en- When the GLM for the bumblebee mimic hances their similarity to bees and hence Eristalis intricarius was nm again with their chances of avoiding being eaten by honeyhees and the three bumblebee taxa birds through Batesian mimicry. separately there was a significant positive Have dronefiies altered their flight be- relationship with honeyhee numbers for haviour to resemble honeybees in other males but not for females; this is consistent ways? When honeybees fly away from a with the hypothesis that ancestral drone- flower after collecting , they clean flies were honeybee mimics but some have pollen from their body and push it into the since evolved a morphological resemblance pollen baskets on their hind legs. The large to social wasps or to bumblebees. dronefiy Eristalis tenax dangles its hind legs when in flight very much like a honey- Behaviour and Batesian mimicry bee, but of course it has no pollen baskets, We can also predict that if a hoverfly gains so this behaviour may have survival value protection from mimicry then its foraging by enhancing its resemblance to a honey- hehaviour should be similar to that of its bee. In collaboration with Roland Ennos, supposed model. If there is no mimicry then its behaviour should be more like that of other hoverflies. Yvonne Golding 25- Bramble (June) measured the times dronefiies and other Figure 4, Times insects spent on flowers (either feeding 20 - (seconds) spent resting or resting), and the times they took flying or feeding on brambie 15 - flowers by different from one flower to another. She studied insects: droneflies the feeding and flying times for insects on 10 • and honeybees spend eight plants making 16 comparisons in all; similar lengths of time; of these the times for droneflies did not 5 - bumblebees spend significantly less time differ significantly from those of honeybees O on each flower whiie in 13 instances whereas they very often the hoverfly Voiiucelia Dronefiy Honeybee Bumblebee Voiuceiia differed from the times for bumblebees, pellucens peiiucens spends other hoverflies and muscid flies (Table 3, significantly more time.

Volume 55 Number 4, November 2008 I Biologist 205 IOB I Hoverflies

Tabie 2. Resuits from the Gênerai Linear Model for supposed hoverfiy mimics showing the déviances of the model covariates, their significance and siope. For ten of the species the déviances and siope show that hoverfiy numbers are significantiy and positively related to the numbers of their supposed model, which is consistent with the hypothesis that they are indeed gaining protection through Batesian mimicry.

.Batesian madeL Deviance of madeLi:DvaiÍatei „Slope. Bee mimics Eristalis tenax/pertinax honeybees 7.0« • Eristalis arbustorum honeybees 23.8* *• pagana Lasioglossum bees 0.7 Wasp mimics Syrphus ribesii sociai wasps 14.8*" Episyrphus balteatus sociai wasps 5.3 • pendulus sociai wasps 39.7* silentis sociai wasps 0.2 florea sociai wasps 0.8 Xanthogramma citrofasciatum Nómada bees. 14.4*'* yellow & black solitary wasps 6.7* Bumblebee mimics Cheilosia illustrate ali bumblebees 4.2* Arctophila superb/ens brown bumbiebees 0.3 Criorhina berberina biack, yellow & rusty-red bumblebees 1.7 C.b. var. oxyacantbae brown bumblebees 2.3 Criorbina floccosa brown bumblebees 1.1 Eristalis intncarius 9 black & yellow bumbiebees 5.7** £. intricarius cC black, yeiiow & rusty-red bumblebees 5.1* var. plumata black & yellow bumblebees 3.7*

*-P<0.05;**-P<0.01;*" -P< 0.001 ' All probabilities were halved in accordance with the one-taiied nature of the hypothesis being tested ' Not significant because one-taiied tests were adopted, and this siope is negative

Yvonne Golding analysed flight behaviour Newly released homing pigeons charac- further (Golding ei a¿. 2001). First she ma- teristically fly around for a few minutes nipulated a patch of everlasting daisies so before heading off in the direction of their that the flower heads were at almost the home, presumably they take this time to same level, then she set a video camera orientate themselves in relation to the sun above the patch. The flight paths between or other cues in the sky. Honeybees may flowers of various insects visiting the patch do the same: the looping may give them a were filmed and analysed by stopping the few moments to 'get their bearings' before fllm frame-by-frame and plotting the posi- flying off in the right direction. The loop- tion of the insects on acetate sheets. The ing flight of droneflies may then have no flight paths and speeds ofthe insects were specific function related to orientation or then calculated. obtaining food but simply increase their The results (Figures 5 and 6) showed mimetic resemblance to a honeybee. that muscids fly much faster than honey- Similar experiments showed that Syr- bees and droneflies while the wasp-like phus ribesii has very similar fiight behav- Syrphus ribesii spends more time hover- iour to social wasps - it flies at similar ing in the air. Flying honeybees often exe- speeds and follows similar fiight paths cute loops in their flight and droneflies do - but the other wasp-mimics Sericomyia so as well, but Syrphus and muscids rarely silentis and Helophilus sp. do not (Golding perform loops. But why do bees loop? etal, 2005).

Conclusions Mean times Insects spent flying between flowers The pattern that emerges from all of these studies is that the ecology and behaviour of many bee- and wasp-like hoverfiies is very similar to that of their supposed models Figure 5. Times (seconds) taken for different insects which supports the hypothesis that they do to fly from one everlasting indeed gain protection through Batesian daisy fiower to another: mimicry. Hoverfly mimicry of honeybees, honeybees and dronefiies bumblebees and social wasps involves take simiiar iengths of similarity of colour, morphology, behaviour, time but Syrphus ribesii takes significantiy more ecology and phenology. But each species of time and housefiiestake hoverfly has evolved a different spectrum significantiy less. of these adaptations. So long as predatory

206 Biologist I Volume 55 Number 4, November 2008 Hoverflies i IOB birds are deceived by the resemblance to Figure 6. Percentage of time spent hovering in front of flowers by different insects: Syrpbus hymenopterans, the mimetic patterns, hab- ribesii spends more time hovering than do either droneflies or honeybees, while housefiies do its or behaviour will persist - and give in- not hover at ail. creased fascination to us human observers % of flight time spent iiovering by different insects of these delightful and beautiful insects.

References Azmeh S, Owen J, Sorensen K, Grewcock D and Gil- bert F (1998) Mimicry profiles are affected by hu- man-induced habitat changes. Proceedings ofthe Royal Society of London B, 265, 2285-2290. Brower J V Z ( 1960) Experimental studies of mimic- ry. 4, The reactions of starlings to different propor- tions of models and mimics. American Naturalist, Dronefiy Honeybee Syrphus Housefly 94,271-282 Dittrich W, Gilbert F, Green P, McGregor P and Grewcock D (19931 Imperfect mimicry: a pigeon's Table 3. The mean times of droneflies feeding on flowers and flying between flowers are perspective. Proceedings of the Royal Society of very similar to the times for honeybees but not to those for bumblebees. London B. 251,195-200. + - significant difference Edmunds M (1974) Defence in : a survey of - - no significant difference anti-predator defences. Longman, Harlow. Golding Y C and Edmunds M (2000) Behavioural honeybee (model) bumblebee (con Ero i ) mimicry of honeybees [Apis meliifera ) by dronefiies Flower feeding flying feeding flying {Eristalis spp., Diptera, Syrphidae). Proceedings of the Royal Society of London B, 267, 903-909. Brambie (June) Golding Y C, Ennos A R and Edmunds M (2001) {fiubus fruticosus] Similarity in flight behaviour between the hon- Bramble (Juiy) eybee Apis meliifera (: ) and [ fruticosus) its presumed mimic, the dronefiy Eristalis tenax Snowberry (Diptera: Syrphidae). Journal of Experimental Bi- {Sympboricarpus albus) no control ology. 204, 139-145. Everlasting daisy Golding Y C, Edmunds M and Ennos A R (2005) (Helichrysum bracteum) Flight behaviour during foraging of the social Michaelmas daisy wasp Vespula uutgaris (Hymenoptera: ) {Aster novi-beigii) Goidiiocks and four mimetic hoverflies (Diptera: SyTphidae) (Aster linosyris) Serieomyia silentis, Myathropa florea, Helophilus control Knapweed sp. and Syrphus sp. Journal of Experimental Biol- [Centaurea nigra) ogy, 208, 4523-4527. Rosebay willowherb Howarth B and Edmunds M (2000) The phenology of (Cbamerion angustlfolium) Syrphidae (Diptera): are they Batesian mimics of Hymenoptera? Biological Journal ofthe Linnean Society, 71,437-457. Howarth B, Edmunds M and Gilbert F (2004) Does uleate Hymenoptera. British Journal of Entomol- the abundance of hoverfly (Syrphidae) mimics de- ogy and. Natural History, 13, 1-39. pend on the numbers of their hymenopteran mod- Edmunds M (2000) Why are there good and poor els? Evolution, 58, 367-375. mimics ? Biological Journal of the Linnean Soci- Mostler G (1935) Beohachtungen zur Frage der ety, 70, 459-466. Wespenmimikry. Zeitschrift fr Morphologie und kologie der Tiere, 29, 381-454. Malcolm Edmunds is Emeritus Professor of Zoology at the Further reading University of Central Lancashire. His research interests are Stubbs A E and Falk S J ( 1983) British Hoverflies: an the bioiogy and of opisthobranch molluscs and the Illustrated Identification Guide, British Entomo- antipredator defences of animals. He has published on the logical & Natura! History Society, London. defences of hawkmoth caterpillars, mimetic butterfiies, praying Howarth B, Clee C and Edmunds M (2000) The mim- mantids, sea siugs, ant-mimicking spiders and hoverflies. Email: icry between British Syrphidae (Diptera) and ac- [email protected]

The IOB on the web

Visit the iOB website for up to date news, poiicy and education issues, careers information and details of courses and publications.

INSTITUTE OF BIOLOGY www.iob.org

Volume 55 Number 4, November 2008 | Biologist 207