Somitogenesis in Amphibian Embryos

/. Embryol. exp. Morph. Vol. 58, pp. 107-118, 1980 107 Printed in Great Britain © Company of Biologists Limited 1980

Somitogenesis in amphibian embryos

III. Effects of ambient temperature and of developmental stage upon pattern abnormalities that follow short temperature shocks

By JONATHAN COOKE1 AND TOM ELSDALE2 From the Division of Developmental Biology, National Institute for Medical Research, Mill Hill, London

SUMMARY Temperature shocks of a few minutes duration at 37 °C to tail-bud embryos of Rana induce zones of abnormal segmentation along the somite files subsequently produced. The immediate result of a temperature shock is a temporary arrest of development as a whole, following which the schedule of somite determination and formation is resumed at the normal rate. It is during the period immediately following this that the zone of abnormal somite pattern is determined. Thus the length of the abnormal zone reflects the total time taken by the morphogenetic system to recover from the disturbance, and might depend upon variables affecting both the duration of the initial arrest and the duration of the recovery period itself. Observations are presented demonstrating how the lengths of abnormal zones, caused by a temperature shock of any particular severity, are affected by three variables: (1) the ambient temperature to which the embryos were adapted before shock, (2) the ambient temperature of post-shock development, (3) the stage in somitogenesis, i.e. the number of somites already formed at the time of shock. The data (in this and previous papers of the series) support models postulating that the spatial periodicity in cell behaviour, that is somite morphogenesis, reflects a normal interaction between two hidden aspects of development, one a wavefront of cellular activation passing down the body axis, and the other having the character of a temporal periodicity throughout the tissue. Temperature shock, as well as halting the wavefront (i.e. stopping development) temporarily, leads to a subsequent period during which there is only gradual recovery of normal co-ordination between the periodicity within cells of the tissue and the wavefront progress. It is the relative rate of this recovery, alone, that is responsible for varia- tion in the length of the abnormal zone.

INTRODUCTION In recent papers (Elsdale, Pearson & Whitehead, 1976; Pearson & Elsdale, 1979; Cooke, 1978) the occurrence has been described of localized disruptions in the normally regular pattern of somites and the fissures between them, following the exposure of anuran amphibian embryos to high temperature (37 °C) shocks of a few minutes duration. The embryos as a whole show no signs of 1 Author's address; Division of Developmental Biology, National Institute for Medical Research, Mill Hill, London NW7 1AA, U.K. a Author's address; M.R.C. Population and Cytogenetics Unit, Western General Hospital, Crewe Road, Edinburgh, U.K. 8 EMB 58 108 J. COOKE AND T. ELSDALE damage beyond a few hours initial delay in the developmental schedule and somite cells in the disturbed region are able to complete normal histodifferentia- tion. The precise spatio-temporal sequence of cellular activities in forming somites has, however, been disrupted over regions of paraxial tissue that would normally have formed some three to ten segments. It should be emphasized that whereas the high temperature is experienced for minutes only, the determination and then later execution of the cellular activities which are found to have been disorganized extends over hours or days (depending upon species, ambient temperature and the length of the abnormal region seen). It is this feature which makes the results of interest. The cell movements of somite formation in Xenopus, in their spatial and temporal aspects, have been described by Hamilton (1969) and by Cooke & Zeeman (1976). There are two separate developmental time periods during which local abnormalities may be induced in the somite series by high temperature, in either Rana temporaria or Xenopus laevis. Shocks to late blastulae or gastrulae induce areas of disruption, on a random basis, approximately evenly distributed through the tissue due to form about the anterior 25 somites of the body pattern, or clustered towards the more posterior of these after shocks delivered near the end of this period. Similar shocks to stages from early neurula onwards induce, with great reliability, a single, bilateral zone of disrupted pattern precisely in that tissue which was due to begin the visible sequence of somite morphogenesis at a particular time after the shock (2-6 h at lab. ambient temp., dependent upon species). Temperature sensitivity in tissue during this second sensitive period thus moves back through the body pattern in a regular way, just ahead of the steadily advancing wavefront of morphogenesis itself (Elsdale et al. 1976; Pearson & Elsdale, 1979). These two developmental periods are separated by a recognizable interval during which any temperature shock that allows development to continue at all will leave the somite pattern unaffected. One possible meaning of this refractory period and of the two spatial patterns of effect on either side of it has been discussed by Cooke (1978), but the early sensitive period may be understandable in quite other ways than the one sug- gested in that paper (see Elsdale & Pearson, 1979). In the present paper we describe the effects of ambient temperatures, prior to shock and during development after the shock, and the precise stage of morphogenesis at shock itself, as variables influencing the amount of pattern made abnormal by a given number of minutes spent at 37-5 °C during the second of the sensitive periods. For this, as for all our more recent work we have used only Rana temporaria eggs collected from the wild in central Scotland, having established that the effect of temperature shocks on morphogenesis are essentially identical, in relation to the developmental scale, with those seen in Xenopus. The tight natural synchrony and very homogeneous developmental rate within egg clutches, the low incidence of spontaneously abnormal morphogenesis, the slower tempo of pattern formation and the wide temperature range allowing Somitogenesis in amphibian embryos, III 109 normal development in this species, all permit a more precise temporal analysis of events in somite formation. We establish in this paper that, during the second sensitive period, only the precise stage of morphogenesis at shock - operationally, the number of somites then already formed - determines the position of the anterior edge of the resulting patch of pattern disruption. Since somite boundaries are determined and then form in a strict head-to-tail time sequence, the length of the abnormal zone, i.e. 'amount of pattern made abnormal' in terms of somites-worth of tissue, reflects the amount of post-shock development that occurs before normal relationships are re-established between processes which must interact to organize the pattern. Normal development has of course its temperature/rate coefficient. This differs apparently from that of some of the processes of recovery from acute periods of high temperature, because development at different ambient temperatures (after a given shock) changes the amount of pattern liable to be made abnormal. This and other findings are discussed in terms of a class of models that is currently being considered for the mechanism underlying development of repeated structures.

MATERIALS AND METHODS Spawn of Rana temporaria (2- to 4-cell stage) was collected in season from ponds in central Scotland. Embryos were allowed to develop at particular temperatures between 6 and 20 °C in aerated pond-water after separation into small groups within the jelly to ensure uniform conditions. Development is normal over this temperature range and survival unimpaired. Embryos are manually decapsulated (but vitelline membranes left intact) before dropping in batches of 15-40 from the submerged mouth of a pipette into a 150 ml bottle of 5 % Dulbecco saline in pond water at 37-5 °C. After the required number of minutes (8-12) they were gently pipetted into an excess of medium at 15 °C. Five minutes later they were set to develop in shallow dishes of pond water at a chosen ambient temperature within the aforementioned range. There was no death or detectable cell loss following such shocks in these experiments. Somites were examined by stripping the integument from larvae about 5 min after the onset of fixation in 1 % glacial acetic acid in 0-5 % K2Cr2O7 solution. Larvae were stripped at stages that postdated the return to normal segmentation in the forming tail somites. Abnormally shaped or absent fissures were scored bilaterally after lining up larvae beside controls. Selected embryos were examined in this way at earlier stages after shock, or fixed in half-strength Karnovsky's fluid at pH 7-4 for 1 jam plastic sectioning in exact horizontal section and stain- ing with toluidine blue.

8-2 110 J. COOKE AND T. ELSDALE •S

Fig. 1. Four typical abnormal zones, caused by lOmin. at 37-5 °C experienced around the 4-somite stage, drawn in camera lucida. Left-hand sides of embryos with epidermis stripped. P = outline of pronephros. S = outline of somitic mesoderm. Lines of normal intersomitic fissures and equivalent separations between cell populations, drawn in. For further details of abnormalities see Pearson & Elsdale (1979). Fissures are initially formed as lines of de-adhesion separating populations of cells, each such population assuming a rosette configuration. Later, due to cell re- alignment, spindle-shaped cells span the distance between adjacent fissures.

RESULTS Temperature shocks of the order of 100 sec often cause minute but definite disturbances in the shape and position of one or two fissures, at a point in the somite pattern which would have been the front edge of the abnormal zone had the shock been somewhat longer (Cooke, 1978; Cooke & Elsdale, unpublished). In the present material a shock length of 8-12 min was chosen (see Fig. 1), so that the typical result was a region of disrupted pattern embracing the tissue that would have formed some three to ten successive somites, beginning very abruptly anteriorly, and returning much more gradually, posteriorly, to a normal sequence of cell populations separated by regular fissures. It is this appearance which has led to the postulate (Pearson & Elsdale, 1979) that the initial, sudden disruption among pattern-co-ordinating processes, engendered Somitogenesis in amphibian embryos. Ill 111

Table 1. Ultimate developmental retardation, in relation to initial delay or arrest of morphogenesis and normal rate of morphogenesis (h/somite) at each of two post-shock temperatures, caused by 10 min at 37-5 °C in a synchronous population

No. of .". Estimated no. of hours and embryos ex- somites-worth of initial arrest amined at Duration (A) and ultimate retardation Temperature each sampling post-shock Mean no. of (i?) in relation to (C) control of post-shock timepoint examined somites formed development rate development post-shock (h) since shock 20 °C 6 Experimental 0 Control 5 8 Experimental 0-75 Control 6-5 10 Experimental 2-5 A and R = 7 h approx or 5-75 Control 8-25 somites-worth. C = l-2h/ somite 6°C 24 Experimental 0 Control 2-33 48 Experimental 0 Control 4-33 72 Experimental 2-17 A and R = 50 h approx or 4-5 Control 6-67 somites-worth. C = 11 h/ somite This investigation was made on eggs from one clutch, adapted to a lab. ambient temp, of 15 °C before shock, and forming somite 22 (+ or — £) at the time of shock. Somite formation in each freshly dissected embryo was counted to the newest £ somite on the right-hand side, as each somite forms in rapid dorso- ventral sequence. Ultimate retardation due to shock, in somites, was measured at control 35-somite stage.

by the shock, is subject to gradual recovery after the resumption of development as a whole. Shocks of this length result also in a developmental delay of some hours relative to synchronous control siblings, representing a deficit of some five formed somites when the populations are later compared at any stage prior to completion of the normal body complement of somites (40+) in the controls. This delay is incurred as an arrest immediately following shock, with subsequent resumption of development at normal rate (see Elsdale et al. 1976; Pearson & Elsdale, 1979). Visible morphogenesis recommences after a delay similar to the delay observed in developmental schedule much later on (Table 1). We conclude from this that, after the initial arrest at shock, the wavefront representing morphogenesis recommences its passage down the body at the same rate as in control embryos even though there is to occur subsequently a zone in the body (thus, a time period) within which this wavefront fails to achieve normal expression as somite pattern. Temperature shock thus results in a time delay in the schedule of pattern determination that is occurring tailwards of visible morphogenesis, and also causes an abnormality in the pattern region 112 J. COOKE AND T. ELSDALE

Table 2. The effect of ambient temperatures of development before and after the shock, as variables affecting the amount of pattern made abnormal by 8 mins at 37-5 °C

Mean length of abnormal zone Experiment Temperature of Temperature of in somites-worth no. N pre-shock devel. post-shock devel. of tissue 1 15 15 °C 20 °C 6-7 (7th somite forming) 6°C 4-3 20 °C 20 °C 2-8 (9th somite forming) 6°C 2-6 2 15 12 °C 20 °C 7-3 (14th somite forming) 6°C 5-9 20 °C 20 °C 5-7 (15th somite forming) 6°C 41 3 15 15 °C 18 °C 10-5 (20th somite forming) 6°C 60 4 20 6°C 20 °C 9-5 (5th somite forming) 20 °C 20 °C 2-9 (5th somite forming) ' ]V' refers to the number of synchronous sibling larvae in each sample within an experiment. The exact stage of pattern formation at each temperature of pre-shock development, in each experiment, is given. By t testing, only the post-shock temperature comparison at 20 °C pre-shock development in Exp. 1 was not significant at the 0-05 level. Significance of difference according to post-shock temperature at 15 °C pre-shock development in Exps. 1 and 3, and according to pre-shock temperature in Exps. 1 and 4, exceeded the P = 001 level. which is first to be laid down when development, including pattern determination, recommences. Delay in development and pattern disruption are separable consequences of temperature shock. After a 10 min, 35-5 °C shock, 2 °C lower than the normal one, no abnormal pattern was seen in 20 larvae, despite a rather homogeneous final delay in morphogenetic schedule of 4| h (two formed somites). Table 2 shows the results of experiments where two conditions were varied between samples from a sibling population receiving the same shock at closely comparable points in development (as monitored by visible somite formation). These were: ambient temperature of development from early blastula stages until shock, and ambient temperature of development in the period after shock. Larvae were scored at an advanced but incomplete stage of pattern development, so that the developmental delay incurred by shock could be computed in terms of number of somites formed at the post-shock temperature of development, and time taken to form each somite at that temperature. It can be seen that low temperature of development after shock has a marked effect in decreasing the Somitogenesis in amphibian embryos. Ill 113

Table 3

Mean no. of somites' Temperature of Mean length of delay in morpho- pre-shock and abnormal zones genesis, relative to post-shock No. of somites in somites-worth sibling controls at N development formed at shock of tissue 15 °C, caused by shock 20 15 °C 4 2-9 70 20 13 41 7-5 By / testing, the comparison of abnormal zone lengths as between 4 and 13 formed somites at shock was significant at the P = 001 level. The difference in mean final delay caused by the shocks at the two stages did not approach significance.

amount of pattern made abnormal, provided that intrinsic sensitivity of the egg clutch is such as to produce long zones of abnormality at normal laboratory ambient temperatures. Ambient temperature up to the time of a particular shock, however, is related in the opposite way to the amount of abnormal pattern caused by that shock. The greater the difference between temperature of prior development and that of the shock itself, the greater the pattern disruption effect tends to be. A large batch of embryos was reared at 6 °C from the 2-cell stage and subjected to a 28 °C shock for 12 min (i.e. an equivalent temperature jump to a 37-5 °C shock experienced by embryos having developed at normal ambients in these experiments). They showed no signs of abnormal somite pattern. Much work on Xenopus and Rana indeed suggests a very steep relationship between absolute temperature of a shock, around 36-40 °C, and the onset of developmental disruption (see, for example, the 35-5 °C shock with delay but no pattern disruption, reported here). We cannot therefore suppose that the effect of pre- shock ambient temperatures upon abnormalities is due entirely to the difference between the temperature to which embryos have become adapted and the high one to which they are then acutely subjected. In Table 3 and Fig. 2 results are presented of experiments where the critical variable is the precise stage reached in pattern determination of the somite series at the time of shock. For Table 3, samples were withdrawn from a synchronous population, developing at constant ambient temperature, to receive similar shocks (8J min) at two developmental stages 24 h apart. For Fig. 2 a spread population of embryos was built up from one egg clutch by releasing small groups of eggs from their very slow development through cleavage stages at 6 °C, at a succession of times. They were then pooled to develop together to tail-bud stages at 15 °C, when a single shock was experienced by them all at a time corresponding to between 4 and 13 formed somites (or 48-72 h of adaptation to 15 °C). These procedures produced the expected range of first abnormal somites, around 7-16, and in the experiment of Fig. 2 the 114 J. COOKE AND T. ELSDALE 11 O

0 o / o o °/ (9 o O// o / / o / / o o

o 4 -

1st abnormal somite 16 Fig. 2. The correlations between developmental stage at shock and the length of abnormal zones produced by 8 min at 37-5 °C, under two ambient temperatures of post-shock development. Open circles, 18 °C post-shock ambient. Correlation coefficient r = 0-88, significant at the P = 001 level. Filled circles, 6 °C post-shock ambient. Correlation coefficient r = 0-60, significant at the P — 0-05 level. effects of the shock were also examined after development at each of two post- shock ambients. Within these experiments all embryos therefore received the same shock, and had adapted over long periods to the same pre-shock ambient temperature. In what way is the amount of pattern made abnormal related to the position in the body pattern where abnormality commences? Does the number of pattern units or somites-worth of tissue affected remain constant, for instance, meaning that a shorter tract of tissue is made abnormal by shocks experienced later where each somite is made of fewer cells? The results show clearly that this is not so, but that a given shock experienced later in the genesis of the body pattern incurs a longer tract of disrupted pattern in terms of somite numbers. Although the wavefront of morphogenesis slows down in terms of tissue (cells) traversed per time as it progresses down the body, it proceeds at a uniform rate in terms of somites formed (Cooke & Zeeman, 1976). This applies also to the earlier, hidden wavefront, namely somite determination which occurs a set time before actual somite formation at each body level (Elsdale et al. 1976; Cooke, 1977; Pearson & Elsdale, 1979). The last results therefore mean that when shock is experienced later in development, relatively more subsequent morphogenesis is Somitogenesis in amphibian embryos. Ill 115 able to occur before the co-ordination among dynamic processes underlying pattern determination returns to normal, as reflected in the regularity of the pattern determined.

DISCUSSION Experiments employing temperature shock have demonstrated the existence of a wavefront of somite determination, advancing down the axis a set time in advance of a further, visible wavefront, that of somite formation. The spatial separation between the two wavefronts is given by the difference between the number of somites already visible at the time of shock, and the ordinal number of the first somite of the abnormal zone resulting from that shock (Elsdale et al 1976). In Rana, the separation is normally three somites-worth of material, meaning that during most of morphogenesis, the boundaries on three more somites than those visible have been determined, and this value is constant except at the very beginning of the somite series (Elsdale & Pearson, 1979). Observations in the present work indicate that it is also constant for a wide range of ambient temperatures, although the absolute rates of passage of the wavefronts are steeply temperature-dependent. The most economical way of understanding these two successive wavefronts, of determination followed by morphogenesis, is as expressions of two time points in one intracellular process of development, which itself occurs according to a timing 'gradient' along the body pattern to give the appearance of waves, even though no truly wave-like propagation is involved at the time of expression (a 'kinematic' wave). The spatially periodic somite pattern, whereby successive similar-sized groups of cells participate in a similar sequence of activities to form each somite, might then be due to regular punctuation of the progress of this (otherwise smooth) developmental wavefront by its interaction with some other, synchronized cellular activity within the tissue, having the character of a periodicity in effect if not in underlying mechanism (Cooke & Zeeman, 1976; Pearson & Elsdale, 1979). Previous work on post-neurula temperature shocks (Elsdale & Pearson 1979) argues that the lengths of abnormal zones are a function of these two processes recovering their co-ordination slowly after shock. The present effects, of post- shock ambient temperature and of age at shock upon the lengths of abnormal zones, offer further evidence for two component processes underlying the genesis of normal somite regularity. They are normally co-ordinated, but experimentally separable because their recovery after a temperature shock responds differentially to ambient temperature of development and to stage of development at shock. During recovery from shock, the travelling wavefront appears to be resumed normally when development as a whole is resumed whereas the resumption and intercellular co-ordination of the periodicity responds somewhat independently of development as a whole. The detailed argument is as follows. 116 J. COOKE AND T. ELSDALE On the general model for determination of somite pattern outlined above, rate of recovery of normal co-ordination between periodicity and wavefront after the shocks is less temperature-dependent than rate of development itself. At lower post-shock ambients therefore, relatively less development takes place with imperfect co-ordination; hence the shorter zone. Recovery of co-ordination also appears to be relatively slower, in relation to the constant rate of pattern determination (Elsdale et al. 1976) after shocks to later stages. Data in Tables 1 and 3 show that, after shocks of a particular severity, resumption of development at widely different temperatures or at different stages of development requires very similar amounts of developmental time in terms of the incurred morphogenetic delay (i.e. somite number x laboratory time taken to form each somite at the temperature and stage concerned). If we did not know this we might postulate that, at low post-shock temperatures (Table 2, Figure 2), all aspects of development (wavefront and co-ordination) have recovered fully before so much new post-shock pattern has been set up as is the case in siblings at higher post-shock ambients. In fact, wavefront recovery requires a similar amount of developmental time (in terms of number of somites determined and made) at low temperature, but co-ordination recovery has been relatively faster (i.e. less temperature dependent) thus exposing the embryo to less actual morphogenesis while in an abnormal state. Similarly, we might have postulated that the effect whereby more pattern is disrupted by shocks to older embryos, affecting posterior somites, is due to actually shorter delay in progress of the wavefront despite a similar period for re-coordination, to produce the converse effect of low ambient temperature. In fact, it must be the recovery period for co-ordination of periodicity which is longer, since the delay in wavefront is essentially unchanged. The spontaneous incidence of local abnormalities in control populations, corresponding to minimal versions of the disruptions due to temperature shocks, increases sharply with somite number posterior to a certain position in the body. This suggests that the periodic component of pattern formation, or the cells' response to it, is less well organized at later stages even in undisturbed development. This may ultimately help to explain the fact that following shock in later development, after more repetitions of the periodicity have elapsed, the developmental time taken to approach normal co-ordination is relatively greater. The effect of low post-shock temperatures in reducing the length of abnormal zones can be understood by an example. Suppose that embryos are shocked when, say, somite 10 is forming. Then when somite 17 has formed at the end of the experiment in all embryos developed at low post-shock temperature, we note that in the high post-shock temperature embryos, the tissue that would have been incorporated into this somite is usually within the abnormal zone where small groups of cells are isolated by chaotic fissures, or form a continuous tissue. This is despite the fact that in the low-temperature embryos, there would have Somitogenesis in amphibian embryos, III 117 been an equivalent amount of total developmental time (including arrest) elapsed when the wavefront crossed the position of this somite. The return to full function and intercellular co-ordination for the periodic process, leading to spatially regular fissure formation, must therefore have a markedly flatter temperature coefficient than does development as a whole. The particular model presented elsewhere (Cooke & Zeeman, 1976; Cooke, 1975, 1977; Zeeman, 1975) proposes that this second, periodic component process in somite pattern formation may be a smooth biochemical oscillator, phase-linked among cells of the embryo, and acting to advance and retard the development of cell behaviour expressed in the wavefront. The present results, and indeed all the results from heat shocks to date, are by no means proposed as evidence for such a theory, but only for the more general model of inter- acting wavefront and periodicity. The actual period length of any rhythmical process, measured in terms of the progress of the determination wavefront, must remain highly constant throughout development. This requires a very similar temperature response by the two processes. Only in this way could relative constancy of somite numbers across developmental temperatures be achieved, meaning that a closely similar number of cycles of the periodicity are incorporated into the passage-time for the determination wavefront down the whole embryo whatever the temperature. In this context, it is interesting to note that vertebrates such as Rana, adapted to develop in small bodies of fresh water, show strong constancy of somite numbers over wide developmental temperature ranges, whereas certain marine fish on the one hand and birds on the other (Fowler, 1970; Lindsay & Moodie, 1967), adapted to more temperature-buffered development, show poor canaliza- tion of the total number of somites against artificial manipulation of temperature. The number of hours disturbance to development in eggs of a given clutch, after a given shock, can depend on the ambient temperature to which the embryos were physiologically accustomed prior to shock (Table 2), as if some adaptation of cellular machinery to low temperature rendered it more disruptible by brief exposure to a high one. The literature on acute high temperature effects would suggest that there tends to be a pronounced threshold for onset of damage within a certain temperature range for each species, but with this proviso, the dynamics of the rhythmic process itself must be more disruptible in some way after pre-adaptation to the cold. There are observations (Fraenkel & Hopf, 1940; Hoar & Cottle, 1952) including some on developing systems, that suggest that cellular adaptation to a range of temperatures occurs partly via systematic alteration in the lipid composition, through altered balance in the metabolic pathways that construct such lipids. If membranes are involved in rhythmical aspects of cellular activity, as at least one general model for biological clocks would suggest (Njus, Sulzman & Hastings, 1974), we might expect temperature contrasts between ambient and shock to play a role. 118 J. COOKE AND T. ELSDALE The work is supported by the Medical Research Council.

REFERENCES COOKE, J. (1975). Experimental analysis and a theory of the control of somite number during amphibian morphogenesis. In Squaw Valley Winter Conference: Developmental Biology (ed. D. McMahon), pp. 205-226. New York: Benjamin. COOKE, J. (1977). The control of somite number during amphibian development: Models and experiments. In Vertebrate Limb and Somite Morphogenesis. B.S.D.B., Symp. 3 (ed. D. A. Ede, J. R. HinchlifTe & M. Balls), pp. 433-448. Cambridge University Press. COOKE, J. (1978). Somite abnormalities caused by short heat shocks to preneurula stages of Xenopus laevis. J. Embryol. exp. Morph., 45, 283-294. COOKE, J. & ZEEMAN, E. C. (1976). A clock and wavefront model for the control of the numbers of repeated structures during animal development. /. theoret. Biol. 58, 455-476. ELSDALE, T. & PEARSON, M. (1979). Somitogenesis in amphibian embryos: II. /. Embryol. exp. Morph. 53, 245-267. ELSDALE, T., PEARSON, M. & WHITEHEAD, M. (1976). Abnormalities in somite segmentation following heat shock to Xenopus embryos. /. Embryol. exp. Morph. 35, 625-635. FOWLER, J. A. (1970). Control of vertebral number in teleosts: an embryological problem. A. Rev. Biol. 45, 148-167. FRAENKEL, G. & HOPF, H. S. (1940). Temperature adaptation and saturation of phospha- tides, blowfly larvae. Biochem. J. 34, 1085-1092. HAMILTON, L. (1969). The formation of somites in Xenopus laevis. J. Embryol. exp. Morph. 22, 253-260. HOAR, W. S. & COTTLE, M. K. (1952). Relation of dietary fat and tissue fat to temperature tolerance in goldfish. I and II. Can. J. Zool. 30, 41-48, 49-54. LINDSAY, C. C. & MOODIE, G. E. E. (1967). The effect of incubation temperature on vertebral count in the chicken. Can. J. Zool. 45, 891-892. NJUS, D., SULZMAN, F. M. & HASTINGS, J. W. (1974). A membrane model for biological clocks. Nature, Lond. 248, 116-119. PEARSON, M. & ELSDALE, T. (1979). Somitogenesis in amphibian embryos. I. Experimental evidence for an interaction between two temporal factors in the specification of somite pattern. /. Embryol. exp. Morph. 51, 27-50. ZEEMAN, E. C. (1975). Appendix to J. Cooke. A. Rev. Biophys. Bioeng. 4 (ed. L. J. Mullins et al), pp. 210-213. Palo Alto, California: Annual Reviews Inc.

{Received 25 September 1979, revised 4 February 1980)