,;.

.:,ra: 0,.:,e, taY! .or 8Dl "Evolution Brain Behav Evol 1997;50:8-12 -

The Evolution of Flight: , Department of Zoology, ;. ~I i , University of Washington, Seattle, Wash., USA Implications for the , j , " Evolution of the Nervous System

...... ' . Key Words Abstract phylogeny The Insecta encompasses a prodigiously diverse group as measured at the spe­ Insect flight . cies, family and ordinal levels, but the nervous system bears evidence of conser­ Giant interneurons vatism. The early acquisition of flight must have been a major factor in the Abdominal cerci diversification ofbody form. Arguments are presented that predator evasion was a primary factor in the origin of flight and that a conserved set of giant interneu­ rons played a key element in the transition.

Introduction: The Scale of Insect Diversity histories, and the vast majority of them disperse on the wing. The evolutionary history of the is long - at least What can explain the evolution of such diversity? Cer­ five times that of their vertebrate aerial counterparts, the tainly size, generation time and life history, allowing great birds - and the products of that long history are spectacu­ subtlety in niche exploitation, are disposing·factors ,[May, 1arlydiverse. Their diversity, with nearly a million living 19'78]. But I believe that there are other factors, applicable species described and perhaps five times that yet to be cat­ to all the though most conspicuously manifested alogued, is well known [Wilson, 1992]. But that diversity'is in the Insecta, which stem from the fundamental bauplan of not merely a matter of species richness; it extends to higher arthropods and the developmental processes that generate taxonomic levels too, for 45 living and extinct insect orders it. are genera:ny recognized [Carpenter, 1992], while the com­ First, there is the modular metameric segmental orga­ parable figure for mammals is 26 [Young, 1981], and at the nization; serial segments as a keyboard on which diverse family level insect diversity has eXCeeded that ofpreserved variations can be played, with their functional groups or tetrapods throughout 90% of their evolutionary history tagmata (head, thorax, abdomen) forming the leitmotifs [Labandeira and Sepkoski, 1993]. This extraordinary diver­ shaped by the homeotic genes. sity is the product of an exuberant set of variations on a Second, the insect integument is based on a monolayer structural theme that was established in the early Palaeo­ epithelium, the epidermis, which determines the outer form zoic and exploded first in the with the acqui­ of the directly as a function ofthe local area of cuti­ sition of flight. Later, in the Mesozoic, the advent of meta­ cle laid down [Wigglesworth, 1972]. That integument is, to

morphosis, a sequential polymorphism, i that enabled the an extent, detached from the internal organs which lie sus­ separation of trophic and reproductive/dispersive phases, pended in the cavity ofthe hemocoel. Is it then possible that , , engendered immense diversification, fOr more than 85% of this relative degree of detachment (which means that the described modern insect species have: holometabolan life ,morphogenetic surface of the insect can be thought of as

© 1997 S. Karger AG, Basel John S. Edwards KAR.GER. Department of Zoology E-Mail [email protected] University of Washington Fax+4161306l234 This article is also accessible online at: Seattle, WA 98195 (USA) htlp:/Iwww.karger.ch , http://BioMedNet.comlkarger ' a somewhat independent two-dimensional sheet) confers aerodynamic grounds, the proto-wing was more likely to sur~ o(Treater freedom for evolutionary tinkering than the 'too. too have been an active structure than a passive planing solid flesh' of a vertebrate? In other words, has it been eas- face. If that is so, it becomes necessary to consider the role ier for arthropods to vary the mold for their external form, of the' nervous system from the outset. The precise neure- . by which we .measure diversity, than for the vertebrates, ethological context for the origin of flight in the Paleozoic and has this been a key factor in the diverse niche exploita­ ; period must, ofcourse, always remain conjectural, but there tion by the insects? And how, in the context of this sympo'­ are some clues in the nervous systems ofliving representa­ sium,'is it that their diversity of external form and function tives of insect groups that span the innovation of flight. My contrasts with an apparent conservatism in their nervous thesis, in essence, is that predation was the selective force .systems [Edwards and Palka, 1991; Boyan, 1993]? driving the transition from earth-bound to aerial locomo­ tion, and the evidence is tied to the cercus-giant interneuron mediated evasion behavior common to primitively wingless New Perspectives of Arthropod'Phylogeny and and to Winged insects. The notion that predation played a Their Implications for the Nervous System part in the origin of insect flight is not new; Bristowe ., [1958], in his writings on natural history proposed the The broad picture is clear: The primitively segmental se­ idea, and Hinton [1963] elaborated op it, but a neuroetho­ ries ofganglia have undergone increasing condensation and logical approach was outside their purview. Recent ad­ fusion during the evolutionary passage from insect proto­ vances in palaeontology serve to emphasize the types as represented by the (the bristletails, diversity of arachnid predators in the Palaeozoic landscape e.g., Petrobius) to the most advanced, as represented by the where the pioneer hexapods took to the terrestrial environ­ Diptera (e.g., Drosophila). that much becomes obvious ment [Shear and Kukalova-Peck, 1990]. It must have been from a cursory comparison. But how much has changed in a very risky place, where predator evasion behavior was a neural connectivity, and is there 'evidence for theintroduc:­ sine qua non. tion of novelty? This can best be addressed by first noting the changing views concerning the phylogeny oithe arthro­ pods as a whole. Recent decades have been dominated by Predator Evasion Behavior and the an orthodoxy based on Manton's insistence that the vari­ Origin of Flight ous classes arose through the parallel arthropodisation of several different stem lines [Manton, 1977], but mounting Given the prevalence ofpredation in the Palaeozoic, it is . evidence from molecular biology [Ballard et al., 1992], scarcely surprising that structural and behavioral defenses palaeontology aJ:.ld comparative anatomy [Kukalova-Peck, such as armor, greater adult size and capacity for regen­ 1992] now,points strongly to a monophyletic relationship '. eration, should have arisen as counter-measures. Another ' among the arthropod classes. This change in perspective widespread measure was startle behavior mediated by giant has implications for the interpretation ofthe ground:-plan of interneurones [Bullock, 1984]. For terrestrial arthropods, the arthropod nervous system and makes a closer compari:- this took the form of amechanosensory early-warning sys­ son between insects and crustaceans [Liese, 1991] cogent. tern based onthe fine sensitivity of delicately poised hairs Reinterpretation ofthe segmental ground plan ofthe arthro- to minute bulk air movements caused by an advancing pod limb [Kukalova-Peck,1992] whichcalls in question the predator. The highest elaboration of this system among in­ long-standing distinction between the so-called biramous sects is found on the abdominal cerci of diverse hexapods. and uniramous limbs of Crustacea and Insecta, respectively Their sensory axons terminate on a set of giant interneu­ [Kukalova-Peck, 1983], has implications also for hypoth-, rons, which in turn project to thoracic and higher ganglia eses concerning the origin of wings in insects. Given the Where they elicit dire9ted evasive movements (fig. 1). This range of exites and endites ( to the ) ' system is widespread in modem orthopteroid insects (e.g., on early arthropod limbs the derivation of wings from the , , crickets, mantids), where its exite of a basal leg segment is certainly plausible. . anatomy and physiology are well known [Boyan and Ball, In the next section, I shall address the question of the " 1990]. The neurons ofthe system in some species occupy as origin of flight from the standpojnt of the nervous systelIl. much as 30% of the volume of the abdominal ventral nerve Much, ifnot all of the theory concerning the origin of flight cord in a nervous system for which parsimony is abasic de­ has considered the exoskeleton of the insect in various sign criterion. The interplay of predatory and evasive tac­ aerodynamic contexts. As Dickinson et al. [1997] argue 9n tics has refined the system, as Roeder [1963] noted: '(This)

Origin of Insect Flight Brain Behav Evol 1997;50:8-1~ 1

10 Brain Behav' Evol 1997;50:8-12 Edwards reciprocal selection pressure to shorten the attack and eva­ dination have their homologs in anterior abdominal ganglia sion responses must have pushed the evolution of the rele­ of segments that may have had winglets [Robertson et al., vant neural mechanisms to a level of efficiency and sim­ 1982; Dumont and Robertson; 1986]. The earliest form of plicity limited only by other biological needs and by the aerodynamic experiment, the evasive leap, may have been' basic limitations of neural mechanisms.' There is good evi­ augmented with a glide assisted by lateral planing surfaces. dence that some, 'at least, of the constituent cercal system .Unlikely as this may appear, the tactic is said to be mani- . giant interneurons are homologous throughout the orthop­ fested today by a small Australian salticid spider, Saitis teroid .insects [Boyan and Ball, 1990; Edwards and palki, volans, which is reported to attain glides of 35 body lengths 1991], and it is highly probable that the extends by means of lateral folds on the abdomen [Main, 1984]. at least 100 million years from the first appearance in Gills of modern larval Ephemeroptera are the homolog the fossil record of winged insects back to their wingless of those ancient apterygote protowings. Already equipped ancestors, for counterparts are found in modern apterygote with musculature and rhythmic motor input, they are plau­ survivors, as exemplified by () and sible candidates for the transition to active flight. Hutchin:­ bristletails (Archaeognatha) [Edwcu-ds and Palka, 1991; Ed­ son [1993] discusses the pros and cons of such an origin. wards, 1992]. These identified giant interneurons span the We do not yet know how many new neurons were transition from flightless to flying insects, apparently with required to make the transition from flightless to flying little modification, and have persisted for several hundred insects. The thoracic ganglia of winged insects such as million years to the present. The giant interneurones project orthopterans have more neurones than those of primitively to thoracic ganglia and to the brain. Their input to motor wingless insects (l.W. Truman, pers. comm.), but it it is not circuitry is best known from the Periplanetci known how many of the addeq neurons are specifically americana [Ritzmann, 1984], where a subset of giant inter- ·· concerned with input to the flight :circuitry, and it could be neurones orient the potential prey away from the predator that much of the transition to active flight was achieved by while a second set initiate rapid evasive running. In other rewiring existing neural connections in thoracic and ab- . orthopterans (e.g., grasshoppers, crickets), the evasive re­ dominal ganglia. If that proves to be the case, it would sponse is a jump, which C

My workdiscussed above was supported in part by N.I.H. grant NB07778. I thank my colleague John Palka for his critique of a dtaft Fig. 1. A comparison of abdominal cerci and giant fiber systems of this essay. in cercus-bearing inseets. The cerci are marked by black arrows. Open arrows mark the paracerci, or median filaments ofapterygote insects. AFossil thysanutan Ramsdelepidion schusteri from UpperCarbonif­ erous (neurons unknown). Body length 6 em. B Modem thysanuran Thermobia domestica; body length 1 em. C Cockroach Periplaneta americana (Blattoidea). 0 Locust Locusta migratoria (), with wings extended and cut to reveal small cerci. E Acheta domesticus (Orthoptera). Left hand column: Giant interneuron configurations in terminal ganglion of Thermobia (F), Periplaneta (I), Locusta (L), and Acheta (0). For Thermobia the full complement ofneurons is shown with one member of each parr sh,own on alternating sides. The neuron of seg­ ment 9 is a putative homolog ofthe single neurons shown below for the other species. G shows cross sections at two levels within the ter­ minal ganglion of Thermobia, midganglion to the left, and near con­ nective to the right. H shows cross section ofabdominal connective of Petrobius (Archaeognatha, the most primitive ofliving insect orders). Cross sections of ganglia (J, M, P) and connectives (K, N, Q) are Shown for the ganglia at left. From Edwards and Palka (1991).

Origin ofInsect Flight Brain Beli.av Evol 1997;50:8-12 11 •.•...•...... •...... •.•...... •. ~ ~ ....•••.•...... •...... •...... References . '"

Ballard, J.WO., GJ. Olsen, D.P. Faith, WA. Od­ Dumont, J.P.e., and R.M. Robertson (1986) Neu­ Main, B.Y. (1984) . Collins, Sydney. gers, D.M. Rowell, and P.W Atkinson (\992) ronal circuits: an evolutionary perspective. Manton, S.M. (1977) The Arthropoda. Clarendon Evidence from 12S ribosomal RNA sequences Science, 233: 849-853. Press, Oxford. that onychophorans are modified arthropods. Edwards, J.S . (1992) Giant interneurons and the May, R.M. (1978) The dynamics and diversity of Science, 258: 1345-1348. origin of insect flight. In Nervous Systems, insect faunas. Symp. R. Ent. Soc. Lond., 9: Boyan, G.S. (\993) Another look at insect audi­ Principles 0fDesign and Function (ed. by R.N. 188-204. . tion: the tympanic receptors as an evolutionary Singh), Wiley Eastern, Bombay, pp. 485--495. Ritzmann, R.E. (\984) The cockroach escape specialization ofthe chordotonal system. J. In­ Edwards, J.S., and 1. Palka (1991) Insect neural response. In Neural Mechanisms of Startle sect Physiol., 39: 187-200. evolution - a fugue or an opera? Seminars in Behavior (ed. by R.e. Eaton), Plenum Press Boyan, G.S., and E.E. Ball (1990) Neuronal orga­ the Neurosciences, 3: 391-398. New York, pp. 93-131. ' nization and information processing in the Hinton, H.E. (1963) The origin of flight in insects. Robertson, R.M., K.G. Pearson, and H. Reichert wind-sensitive cercal receptor/giant interneu­ Proc. R. Ent. Soc. Lond. C, 28: 24--25. (1982) Flight interneurons in the locust and the rone system of the locust and other orthopter­ Hutchinson, G.E. (1993) A Treatise on Limnology, origin ofinsect wings. Science, 217: 177-179. oid insects. Prog. Neurobiol., 35: 217-243. Vol 4. Wiley, New York. Roeder, K.D. (1963) Nerve Cells and Insect Be­ Bristowe, WS. (\958) The world of spiders. Col­ Kukalova-Peck, J. (1983) Origin of havior. Harvard University Press, Cambridge, lins, London. and wing articulation from the arthropodan leg. MA. Bullock, T.H. (1984) Comparative neuroethology Can. J. Zool., 61: 1618-1669. Shear, W.A., and J. Kukalova-Peck (\990) The of startle, rapid escape, and giant fibre me­ Kukalova-Peck, J. (1992) The 'Unirarnia' do not ecology ofPaleozoic terrestrial arthropods: the diated responses. In Neural Mechanisms of exist: the groundplan of the as re­ fossil evidence. Can. 1. Zool., 68: 1807-1834. Startle Behavior (ed. by R.C. Eaton). Plenum vealed by from Wigglesworth, v'B. (\972) The principles oflnsect Press, New York, pp. 1-14. Russia (Insecta: Palaeodictyopteroidea). Can. Physiology, 7th ed. Chapman and Hall, Lon­ Carpenter, EM. (1992) Superclass . In J. Zool., 70: 236-255. don. Arthropoda, Vols 3 and 4 (ed. by R.L. Kaesler). Labandeira, e.e., and lJ. Sepkoski (1993) Insect Wilson, E.O. (\992) The Diversity ofLife. Harvard Treatise on Invertebrate Palaeontology. Geol. diversity in the fossil record. Science, 261: University Press, Cambridge MA. Soc. Amer. 310-315. . Young, J.Z. (\981) The Life ofVertebrates. Claren­ Dickinson, M.H., S. Hannaford, and J. Palka Liese, E.M. (1991 ) Evolutionary trends in inverte­ don Press, Oxford. (\997) The evolution of insect wings and their brate ganglion structure. Seminars in the Neu­ sensory apparatus. Brain Behav. Evol., 50: rosciences, 3: 369-377. 13-24.

12 Brain Behav Evol 1997;50:8-12 Edwards