INTRODUCTION

The early stages of development in amniotes are particularlyimportant subjects to study because it is at this time that theearly body plan is laid down. However, there are no detailedfate maps available for these stages in any amniote. In the chickembryo, which is probably the best studied, one reason was theabsence of a suitable staging system for the preprimitive streakembryo until 1976, when Eyal-Giladi and Kochav producedtheir detailed stage table. Most of the chick fate maps that havebeen published were produced before this staging systembecame available.

Now that both an accurate staging system and a bettermethod for mapping (carbocyanine dyes) exist, we have usedthem to construct a detailed fate map for the preprimitive streakand early primitive streak chick embryo. The results reveal anorderly pattern of cell movements, consistent with previousobservations of gross morphogenetic movements in the earlyblastoderm. The present results also reveal considerableoverlap between different prospective areas, as has recentlybeen described for the mouse (Lawson et al., 1991), rather thansharp dividing lines between different territories, as suggestedby some older publications. Less expected was the finding thatcertain presumptive areas appear to move independently ofothers, even against the predominant gross pattern ofmovements.

MATERIALS AND METHODS

Embryo techniques Fertile hens’ eggs (obtained from Coppocks Poultry Farm, Carterton)were incubated at 38°C for 0-8 hours to obtain embryos between Eyal-

Giladi and Kochav (1976; in Roman numbers for preprimitive streakstages) stage-X and Hamburger and Hamilton (1951; in Arabicnumerals for later stages) stage-3. Embryos were explanted in Pannettand Compton (1924) saline by the technique of New (1955), withmodifications (Stern and Ireland, 1981). Following marking with DiIand/or DiO (see below), the embryos were incubated for 24-36 hoursat 38°C in a humid atmosphere.

Fate mapping with carbocyanine dyes, DiI and DiOThe carbocyanine dyes 1,1

′-dioctadecyl-3,3,3′,3′-tetramethyl indocar-bocyanine perchlorate (Molecular Probes, Inc.) (DiI) and 3,3′-dioc-tadecyloxacarbocyanine perchlorate (Molecular Probes Inc. Di-O-C18-(3)) (DiO) were used for labelling, following methods previouslydescribed (see Stern, 1990; Selleck and Stern, 1991). These intenselyfluorescent dyes are lipophilic, become incorporated into cellmembranes and are not transferred between cells (Honig and Hume,1989; Serbedzija et al., 1990; Wetts and Fraser, 1989).

Briefly, DiI or DiO was first dissolved at 0.5% (w/v) in absoluteethanol and this diluted 1:9 with 0.3 M sucrose in distilled water at40°C (see Serbedzija et al., 1990; Selleck and Stern, 1991). The dyewas applied to the desired region using gentle air pressure, usingmicropipettes pulled from 50 µl capillaires (Sigma) with an Ealingvertical electrode puller. A small group of cells (5-30) could thus belabelled.

A total of 1326 injections were made into 726 embryos (600embryos were labelled with both dyes), applied to different positions(see Table 1). Embryos were not used for subsequent analysis if afterincubation: (i) they had developed abnormally; (ii) most of theembryo contained label; or (iii) dye was seen as fine granular materialover the surface of the tissues. 358 of the embryos developed normallyto stages 11-13. A total of 332 injections, in 200 embryos, was usedfor the final analysis.

Staging of embryos and standardization of position oflabelled cellsTo use a more accurate staging system than those published, a quan-

2879Development 120, 2879-2889 (1994)Printed in Great Britain © The Company of Biologists Limited 1994

We have used carbocyanine dyes (DiI and DiO) to generatefate maps for the epiblast layer of the chick embryobetween stage X and the early primitive streak stage (stages2-3).

The overall distribution of presumptive cell types inthese maps is similar to that described for other laboratoryspecies (zebrafish, frog, mouse). Our maps also revealcertain patterns of movement for these presumptive areas.Most areas converge towards the midline and then moveanteriorly along it. Interestingly, however, some presump-

tive tissue types do not take part in these predominantmovements, but behave in a different way, even if enclosedwithin an area that does undergo medial convergence andanterior movement. The apparently independentbehaviour of certain cell populations suggests that at leastsome presumptive cell types within the epiblast are alreadyspecified at preprimitive streak stages.

Key words: chick embryo, primitive streak, epiblast, DiI, DiO,carbocyanine dyes, mesoderm induction, gastrulation

SUMMARY

A fate map of the epiblast of the early chick embryo

Yohko Hatada* and Claudio D. Stern*

Department of Human Anatomy, South Parks Road, Oxford OX1 3QX, UK

*Present address: Department of Genetics and Development, College of Physicians and Surgeons of Columbia University, 701 West 168th Street, New York, NY10032, USA

2880

titative method was used (Fig. 1), in which the position of the anteriorborder of the hypoblast sheet was used as a reference for staging. Aneyepiece graticule divided into 100 units was used. Using a zoom inthe dissecting microscope (Nikon SMZ-2T), the visible size of thearea pellucida was adjusted to fit exactly into the 100 eyepiece units,with position ‘0’ corresponding to the posterior edge of Koller’ssickle. Using our scale, the width of the sickle varies between 5 and12 units. The position of the anterior edge of the hypoblast at themidline was recorded, as was the site of injection. A similar methodwas used to record the position of the site of injection relative to thewidth of the area pellucida. In this way, a grid was constructed fromwhich the positions of injection sites were mapped.

According to this method, in stage XI embryos, the anterior borderof the hypoblast sheet had reached a position of less than 30; stageXII was ascribed to a hypoblast sheet extending to position 30-69;stage XIII corresponds to a position in the range 70-100. Stage XIVembryos were those in which a cellular bridge visible posterior toKoller’s sickle (see Eyal-Giladi and Kochav, 1976) and/or the sickleitself has become V-shaped. Embryos with a primitive streak wereused up to the stage at which the tip of the streak had reached nofurther than the 55 position.

This quantitative method also allowed us to assess retrospectivelythe original position of marked cells in those cases where the orien-tation of the embryo had not been assessed correctly at the time ofmarking. In total, the orientation had been assessed correctly in323/358 (90%) embryos. It is interesting that even stage 2 embryos,which already have developed a primitive streak, sometimes (3/111)changed the orientation of the embryonic axis during subsequentculture. Because of this, we analyzed particularly carefully each ofthe 35 embryos in which the orientation had shifted from the original:in none of these did the fate of the labelled cells fall outside the bound-aries of that presumptive region as determined from the remaining323 embryos.

Fixation and histologyAfter incubation, the embryos were fixed in PBS (pH 7.0) containing0.25% glutaraldehyde and 4% formaldehyde for at least 1 hour andstored in this fixative. All embryos were mounted in cavity slides andexamined in toto with an epifluorescence microscope (OlympusVanox-T, with 200 W high-pressure mercury lamp). DiI was visual-ized using 547 nm (green) peak excitation, where it emits at 571 nm(red); DiO was excited at 484 nm (blue) and observed by its emissionat 507 nm (green). In a few cases, labelled embryos were also observedusing a BioRad MRC500 confocal laser scanning microscope.

In addition to examination of the fixed embryos as whole mounts,some embryos were processed histologically to confirm the localiza-tion of the labelled cells. For this, the fluorescence had first to be pho-toconverted to an insoluble product by photooxidation of 3,3′-diaminobenzidine (DAB) exactly as described previously (Stern,1990; Selleck and Stern, 1991). After this, embryos were dehydratedin a series of alcohols, embedded in fibrowax, sectioned at 10 µm,mounted on glass slides and dewaxed in xylene before being mountedin DePeX for bright-field observation.

RESULTS

Spatial and temporal resolution of the mapsThe results obtained (based on a pattern of 873 contributions

to different tissues from 332 injections analysed), are shown inFigs 2-3, and specific examples shown in Figs 4-5.

We decided to standardize the positions of the injectionsusing a scale of orthogonal coordinates, adjusted to 100 unitsalong each the anteroposterior axis and the left-right axis, asdescribed in Materials and Methods. To standardize thetemporal aspect of the maps, we used the position of theanterior border of the hypoblast at stages X-XIII and theanterior tip of the primitive streak at stages XIV-3 (Fig. 1). Inthis way, a continuous scale was obtained, spanning thesestages. For the summary maps shown in Figs 2 and 6, however,embryos were grouped by the stage criteria of Eyal-Giladi andKochav (1976).

The following description summarises the main resultsemerging from detailed analysis of the data, classifiedaccording to each cell type containing labelled cells.

Gut endoderm111 injections contributed labelled cells to the gut (Figs 2A,4D,I). The gut endoderm differs from most other presumptiveareas in that it is quite compact at all stages studied. At stageX, no presumptive gut cells are found in the midline except fora very small component at the posterior end of the embryo;they form a wing shape, hinged at the posterior midline (Fig.2A). In general, this presumptive area converges towards themidline and moves anteriorly at the stages studied. The anteriormovement seems to occur in two distinct stages (Fig. 3): a firstone, at stages X-XI, when the anterior border of the prospec-tive gut tissue at the midline moves at the same rate as theanterior border of the hypoblast. The second stage occurs atstages XIII-XIV, with the midline component of the prospec-

Y. Hatada and C. Stern

Table 1. Number of embryos used and number ofinjections performed

Total number of embryos marked 726Number developing normally to stages 11-13 358Number of embryos used for analysis 200Total number of injection sites used to construct fate maps 332

Fig. 1. Diagram summarising the method used to determine theposition of marked regions in preprimitive streak stage embryos. Thediameters of the area pellucida along the midline and at right anglesto this (left lateral to right lateral) were each divided into 100 units.The position of the site of dye injection was recorded. The samescale was used to measure, in each embryo, the positions of theanterior border of the hypoblast sheet (at stages X+ to XIV) and ofthe anterior tip of the primitive streak (stages 2-3). Thesemeasurements provide an additional, quantitative assessment ofdevelopmental stage.

2881Fate map of chick epiblast

tive gut region reaching the centre of the blastoderm at stageXIV. Between stages XII-XIII, there is no apparent anteriormovement at the midline. The most lateral cells contributingto the gut converge towards the midline up to stage XIV. Ingeneral, our impression is that injections into the midline ofthe presumptive gut region produced labelled cells in only thedorsal part of the gut at stages 11-13; more lateral regions ofthe blastoderm contribute to other, lateral/ventral regions of thegut (data not shown).

Prechordal plateOnly a few (20) injections contributed cells to this region (Fig.5A). For this reason, it is difficult to draw general conclusionsabout its movement. From the data obtained, it can be seen thatthis presumptive population tends to be restricted to themidline of the embryo at all stages studied (Fig. 2B). Whencomparing the three stages that contributed more than one datapoint (XI, XII, 2-3), a tendency for this area to move anteri-orly during development can be seen.

Chordamesoderm (notochord/head process)93 injections contributed labelled cells to the chordamesoderm(Figs 2C, 4I,K). Like the presumptive gut, this region is wing-shaped at stage X, hinged at the posterior midline; themovements of the two regions are also similar and occur in twodistinct stages (Fig. 3). At stages 2-3, the presumptive chor-damesodermal territory is wider than that of the gut at the samestage (Figs 2C,6).

Somite mesoderm49 injections contributed to the somitic mesoderm (Figs 2D,4G,H,K, 5C): 40 to the medial halves (Figs 2D, 4K, 5C) and18 to the lateral halves (Fig. 2D). Nine of the injections (twoat stage X, three at stage XI, two at stage XIII and two at stage2-3) contributed labelled cells to both halves (Fig. 4H). Atstage X, the lateral somite territory is contained completelywithin that of the medial somite, but the overlap graduallydiminishes. The separation appears to be brought about mainlyby the medially directed movement of the medial-half somitepopulation, whilst the lateral cells remain at their originalposition. The presumptive medial-half-somite cells, inaddition, show a tendency to move anteriorly at these stages.

Intermediate mesoderm10 injections gave rise to labelled cells in the intermediatemesoderm (Figs 2E, 4G,H). Their positions are sparsely dis-tributed over many regions of the blastoderm and it is thereforedifficult to draw definite conclusions about their movement.

Heart38 injections contributed to the heart (Figs 2F, 4D,H). Fromstage XII to stages 2-3, there is relatively little change in theposition or size of the presumptive heart territory.

Lateral plate mesodermCells contributing to the lateral plates (88 injections) arewidely spread around the surface of the blastoderm, but stillforming wing-shaped territories hinged about the posteriormidline (Figs 2G, 4G,J,L, 5D). There is little indication ofmovement before formation of the primitive streak.

Neuroectoderm211 injections contributed cells to neural tissues (Figs 2I,J,4A-C,E, 5B), of which 65 were restricted to the forebrain,anterior tip of the neuroepithelium (prospective olfactoryregion) or optic lobes (Figs 2J, 4A-C, 5B). The neuroectodermterritory, like the prospective lateral plate mesoderm, is widelydistributed over the surface of the blastoderm at all stagesstudied and shows no particular movement, except for anteriormovement of its posterior border from stage XIV, leading toabsence of this presumptive cell type from the posterior-lateralregion. The presumptive olfactory and optic regions form adiscrete sub-region of the neuroectodermal territory, whichgradually becomes localized at the midline. showing anteri-orly directed movement from stage XI (Fig. 3). Several injec-tions produced descendants in neural tissues and in lateralplate mesoderm, and the two labelled regions were continu-ous at the posterior end of the embryo, level with the regress-ing Hensen’s node at stage 11-13. A similar phenomenon hasbeen observed in the urodele

Triturus pyrrhogaster (Hama,1978).

Surface ectoderm and extraembryonic tissuesThese tissue types are considered together for three reasons:(1) because at early stages and in posterior regions theboundary between amnion and surface ectoderm is poorlydefined, (2) because it was sometimes difficult to distinguishunambiguously the precise layer containing the labelled cellsand (3) because many of the injections contributed to bothtissues. 244 injections contributed to these regions (Figs 2H,4A,C). In general, cells contributing to surface ectoderm andto extraembryonic tissues (amnion, yolk sac and its stalk, etc.)are very widely distributed over the entire surface of theprestreak blastoderm. However, at stage XIV, there is a largelateral/posterior region devoid of cells with these fates (Fig.2H).

Shifting patterns in the mesodermSix injections (one at stage XI, one at stage XII, four at stages2-3; in Fig. 4G) gave rise to labelled progeny. In these cases,labelled cells were found laterally in anterior regions, and moremedially in more posterior parts of the embryo (as describedby Hama, 1978 for Triturus pyrrhogaster). In the majority ofcases, the most anterior cells were located in the lateralmesoderm and gradually shifted via the intermediatemesoderm to the lateral halves of the somites at more posteriorpositions.

Extent of contribution to different levels of theanteroposterior axisIn general, many injections contributed progeny to largeregions of the anteroposterior axis of the embryo. We aretherefore unable to ascribe particular regions of the anteropos-terior axis to specific areas in the early stages of development.This is consistent with the findings of Selleck and Stern (1991)who reported that even progeny derived from some singlelabelled cells spans extended regions of the embryonic axis.However, one exception is the sensory areas associated withthe forebrain, where some injections only contributed progenyto this region.

2882

Incidence of progeny crossing themidlineWe also analyzed the distribution of descen-dants of labelled cells in terms of whether ornot they crossed the midline. The results of thisanalysis are shown in Table 2. In general,labelled progeny tend to be restricted to oneside of the embryo. When this analysis isextended by comparing the distribution indifferent tissue types, it is seen that the descen-dants of presumptive somite cells do not followthis rule: labelled progeny tend frequently tocross the midline (χ2, 1 d.f. = 7.06; P<0.01).We also compared the frequency with whichprogeny crossed the midline in terms of thestage at which the injection had been done. Nosignificant differences were found.

DISCUSSION

Assessment of the technique usedFate maps of the early embryo have been con-structed by many authors, both before theformation of the primitive streak (Kopsch,1926; Gräper, 1929; Wetzel, 1929; Kopsch,1934; Pasteels, 1937; Malan, 1953; Vakaet,1970, 1984) and after (e.g. Rosenquist 1966;Nicolet, 1971). These maps were producedeither by local killing, or marking using spotsof water-soluble dyes applied through thevitelline membrane in ovo, or using carbonparticles. However, these studies all sufferfrom technical defects. Transplantation ofmarked cells might have disturbed the spatial,and perhaps temporal, organization of thetissues to be mapped, while carbon or carmineparticles may not always follow the cells intheir movements. In addition to the disadvan-tages of these methods, the classical mapssuffer from the problem that they wereproduced before a reliable staging system wasavailable for early stages of development.

Recently, both some new marking methods(the carbocyanine dyes, DiI and DiO; seeHonig and Hume, 1989; Wetts and Fraser,1989; Serbedzija et al., 1990; Stern, 1990;Selleck and Stern, 1991) and a new stagingtable (Eyal-Giladi and Kochav, 1976) havebeen described, allowing the labelling of smallgroups of cells at precisely controlled stagesof the very early chick blastoderm withoutusing transplantation or particulate markers.

We opted for a quantitative method forassessing the position of sites of injection andfor staging the embryos, so that data obtainedfrom different embryos can be compared. Thepositions of injection sites were measured withrespect to a pair of orthogonal axes (antero-posterior and mediolateral, each divided into

Y. Hatada and C. Stern

2883Fate map of chick epiblast

100). By this method, size variation along each of these axesof the area pellucida was eliminated by scaling.

To avoid complications introduced by possible left/rightasymmetry of the embryo (c.f. Strehlow and Gilbert, 1993), we

decided to collect data from the left half of the blastoderm; thedegree of left/right asymmetry remains to be assessed in futurestudies. It is worth mentioning, however, that injections to theleft side often gave rise to labelled cells that extended to the

contralateral side of the embryo, as found byGallera and Nicolet (1969) and Rosenquist(1966) in chick and Lawson et al. (1991) inmouse.

In some cases, the embryos were observedunder fluorescence immediately after injection,to confirm the position and size of the label.After incubation, embryos were againobserved by fluorescence and the germ layerscontaining labelled cells discriminated byfocussing through a high numerical apertureobjective (10×, NA=0.40 and 20×, NA=0.65).If there was doubt concerning which tissuescontained labelled progeny, the embryos werephotooxidised and examined in histologicalsections.

It is worth pointing out that these fate mapswere constructed by labelling groups of cellsrather than single cells. Therefore, when aninjection produces progeny in more than onetissue, we cannot formally distinguish betweentwo possibilities: (a) each presumptive areacontains a mixture of cells with different pre-sumptive fates and (b) each area containscommon progenitors for these fates.

Comparison with other vertebratespeciesIn general, the distribution of differentprospective cell types in the early chickembryo agrees broadly with findings made inother vertebrate species (zebrafish, amphibian,mouse). There are a few minor differences,however. Among them, in the chick, the regioncontributing to mesodermal cell types appearsbroader than in the mouse (Lawson et al.,1991). Also, compared to the mouse andurodele fate maps (see Lawson et al., 1991 forsummaries of both), the region of cells con-tributing to the notochord in the chick is alsobroader. In contrast, the region contributing to

gut endoderm appears marginally broader in the mouse than inthe chick.

Convergence and anterior movement start longbefore primitive streak formationOur maps show that specific prospective areas change shapesand positions during development. From these changes, thegross pattern of movements can be extrapolated. Overall, thepredominant movements for most areas between stages X and3 are convergence towards the midline and anteriormovement along the midline, in the posterior half of the areapellucida. Different presumptive tissues undergo thesemovements at different times. For example, ‘axial’ tissues suchas the chordal mesendoderm undergo convergence at stages X-XI and appear to move anteriorly in two steps, at stages X-XIand XIII-XIV (Fig. 3). The presumptive prechordal plate is

Fig. 2. Diagrams showing the data obtained from cell markingexperiments. Each point represents one injection in one embryo,made at the stage shown (X-2/3), which contributed to a given celltype (see below). Thus, each group of 6 diagrams (A, B, etc.)represents the contributions to one tissue type according to the stageat which labelling was performed; intermediate stages betweenlabelling and fixation were not analyzed. Each diagram representsthe right half of the area pellucida (seen from the ventral side of theembryo). (A) Contributions to the definitive (gut) endoderm. (B) Contributions to the prechordal plate. (C) Positions of cellscontributing to the chordamesoderm (notochord and head process).(D) Locations of somite progenitors;

v, medial halves of thesomites; ¶, lateral halves. (E) Intermediate mesoderm (prospectivepro- and mesonephros). (F) Heart. (G) Lateral plate mesoderm. (H) Surface ectoderm and extraembryonic membranes. (I) Neuroectoderm (excluding sensory placodes and optic lobes). (J) Sensory organs: olfactory and optic evaginations.

2884

already at the midline at stage X-XI. The posterior-medial partsof the prospective neuroectoderm and of the notochord terri-tories appear to move anteriorly together with the tip of theprimitive streak.

This overall pattern, convergence towards the midline andanterior migration, has been described previously for earlystages of the chick (Patterson, 1909; Graeper, 1929; Wetzel,1929; Pasteels, 1937; Spratt, 1946; Vakaet 1960, 1970, 1985)and other vertebrates, such as teleosts (Oppenheimer, 1936;Pasteels, 1936; Ballard, 1973, 1981, 1982, Trinkaus and Fink,1992), amphibians (Nakamura, 1938; Pasteels, 1940;Nieuwkoop and Sutasurya, 1979) and reptiles (Pasteels, 1937).In the mouse, Lawson et al. (1991) describe a similar conver-gence (“anisotropic” spread between members of a clone, ori-entated towards the forming primitive streak). Thus, the“Polonnaise” movements that have been described in the lit-erature can be seen as an overall trend in the epiblast. However,when individual prospective areas are followed, it is found thatmany depart from this general pattern.

The pattern of movements explains the shapes ofprospective regionsAfter stage XIII, but before the appearance of the primitivestreak, the anterior border of the presumptive olfactory placodeand optic lobe, notochord, medial somite and prechordal plateareas reach the centre of the area pellucida. This observationagrees with Spratt’s (1946) careful study of epiblastmovements. He describes a convergence of the whole epiblastto the posteromedial part, which could be likened to a fan,hinged at the centre of the area pellucida and closing beforethe primitive streak appears. The centre of the area pellucidadoes not appear to move until later (stage 3), during primitivestreak elongation.

After stage XIV areas such as the heart and neuroectoderm(Figs 2F,I, 6) appear ‘T’-shaped, but are not all coincident.Vakaet’s (1984) study of cell movement seems best to explainhow this shape arises from the patterns seen at earlier stages.He showed not only medial convergence towards the posteriorstreak, but also anteromedially directed movement just in frontof the elongating streak at stage 2. Our study suggests that thismovement may start even before stage XIV.

Cell mixing is not random Classical fate maps of the chick embryo were drawn with thedifferent presumptive territories separated by sharp borders(Patterson 1909; Graeper, 1929; Wetzel, 1929; Pasteels, 1937;Rosenquist 1966; Vakaet, 1984, 1985; Nicolet, 1971).However, more refined techniques have now become available.These are starting to indicate that, in general, such sharpborders do not exist. For example, in the frog, Wetts and Fraser

Y. Hatada and C. Stern

Fig. 3. Graphs showing the change in the position (Y-axis), withincreasing developmental stage (X-axis), of different presumptivecell populations that had been labelled at the midline of the areapellucida. The position of the anterior border of the hypoblast sheetis shown as an interrupted line on the left of each graph, and theanterior tip of the primitive streak as a dashed line on the right.These graphs are derived from data included in Fig. 2.

Fig. 4. Some examples of the results obtained. (A) After labellingcells at position [x=10, y=50] in a stage 2 embryo with DiI (red),labelled cells are seen in the surface ectoderm and amnion, seen herefrom the ventral side of the embryo. (B) DiO (green) was applied tocells at position [10,40] of a stage XI embryo; their descendants arefound in the forebrain, including the olfactory region. (C) Opticlobes and diencephalon. At a different focal plane, labelled cellswere also found in the ectodermal covering of the diencephalic andmesencephalic regions. (D) DiO-labelled cells derived from aninjection at position [25,30] of a stage XII embryo contribute to themidline of the foregut and in the heart. The cells in the heart have thecharacteristic spindle-like shape of cardiac myocytes. (E) Neuraltube and migrating neural crest cells derived from an injection atposition [15,15] at stage XIV. (F) The endothelial lining of theembryonic blood vessels contain DiI-labelled cells, descended fromprogenitors labelled at position [30,70] at stage XII. (G) DiI-labelleddescendants (from an injection at position [20,40] at stage XIII) inthe somites and DiO-labelled cells (from dye applied to position[40,40]) in the intermediate and lateral plate mesoderm. In the lattertwo tissues, the labelled cells are found at a more lateral positionanteriorly and more medially in more posterior regions. (H) Thesomites, intermediate mesoderm and heart contain DiI-labelled cells,derived from progenitors at position [25,30] of a stage XII embryo.(I) Notochord and midline of the gut endoderm, arising from aninjection at stage XI in position [10,50]. (J) Unilaterally distributedcells in the lateral plate, from cells labelled at position [20,50] in astage XIII embryo. (K) DiI (red)-labelled cells are found in thenotochord and DiO (green)-labelled cells in the medial halves of thesomites. (L) Bilateral distribution of labelled cells in the lateralplates, after injection into position [30,60] of a stage 2 embryo. Scalebars, 500 µm in I,J; 200 µm in A,B,L; 100 µm for the remainingphotographs.

2885Fate map of chick epiblast

(1989) report that the descendants of labelled ectoderm cellsbecome mixed, albeit slowly, a process that continues through-out early development. During gastrulation, some of thismixing is driven by cell intercalation (Shih and Keller,1992a,b; Keller et al. 1992), as is also seen in teleosts (Trinkausand Fink 1992). In the French frog, Pleurodeles waltl, Delarueand Boucaut (1992) have demonstrated finer differences in theextent of cell mixing between deep and superficial circum-blastoporal cells: dorsally, deep and superficial cells intermixmore extensively than ventrally. In the zebrafish (Kimmel andLaw, 1985a,b; Kimmel and Warga, 1988; Warga and Kimmel,1990; Ho, 1992) and preimplantation (Winkel and Pedersen,

1988) and postimplantation mouse (Lawson et al., 1991);however, there is much more extensive intermixing of cells,resulting in some indeterminacy in the patterns of descendantsderived from single, identified cells. Lawson et al. (1991) sum-marised this with particular clarity, by concluding that “mor-phogenetic movements occur in the presence of extensive,although not indiscriminate, cell mixing in the epiblast, andthat descendants of a single progenitor may be spread widely,and also be present in different germ layers” (p. 905).

The present experiments reveal two, apparently contradic-tory, trends. First, the finding that certain presumptive celltypes in the epiblast undergo rather different movements to

2886

those of other presumptive regions suggests that there is indeedextensive cell mixing. Second, the finding that descendants ofa small group of cells, labelled by a single dye injection, tendto contribute only to a few, more or less adjacent tissue types,suggests that this mixing is not as extensive. This is the casein most regions except the posterior midline, where many pre-sumptive cell populations come together and overlap. One wayto reconcile these observations is to suggest that cell mixing isfairly widespread, affecting large areas of the epiblast, but thatit is not random and that cells do not wander to very distantsites (c.f. Lawson et al., 1991: “not indiscriminate”; seeabove).

Some cell types may become specified early duringchick developmentInterestingly, certain presumptive cell types move in charac-teristic ways, sometimes against the prevailing currents. Forexample, the prospective lateral plate mesoderm territory does

not appear to change position between stages X and 3, eventhough this territory overlaps with others, which do move asdescribed above from stage X. Likewise, the presumptive opticlobe and olfactory areas seem to move differently from othersurrounding cell fates. This could be taken to indicate thatsome prospective cell types are already specified at very earlystages of development, perhaps as early as stage X, althoughwe cannot formally rule out the possibility that morphogeneticcell movements are not well coordinated at these early stagesof development. The exact degree of such specification fordifferent cell types may vary between different vertebrateclasses, giving rise to the apparent discrepancies in the findingsof different authors in terms of the degree of cell mixing fordifferent species.

Induction may be involved in the specification ofprospective gut cellsSlack (1991) has argued that a comparison between fate mapsand specification maps is of great value in identifying thoseregions that require cell interactions (‘induction’) to define thefates of the cells contained in them. In addition to the presentstudy, fate maps of the early embryo before the formation ofthe primitive streak were published by Kopsch (1926), Gräper(1929), Wetzel (1929), Kopsch (1934), Pasteels (1937), Malan(1953) and Vakaet (1970, 1984, 1985). ‘Specification maps’for the unincubated blastoderm were produced by Hoadley(1926a,b,c,d, 1927), Olivo (1928a,b), Murray and Selby(1930), Waddington (1933, 1935), Butler (1935), Dalton(1935), Rudnick (1932, 1935, 1938a,b, 1944, 1948, 1961),Hunt (1937), Spratt (1940, 1942, 1947) and Rawles (1943).These were generated by culturing pieces of blastoderm on thechorioallantoic membrane of a host embryo, in a plasma clotor on the surface of another early blastoderm. Unfortunately,these specification maps are very crude because no goodstaging system was available at the time, because the pieces

Y. Hatada and C. Stern

Fig. 5. Transverse sections through specimens like those shown in Figs 4-5, after photo-oxidation of DiI-labelled cells. (A) Section at the levelof the prechordal plate. (B) Section through the optic region, showing labelled cells in one of the optic evaginations. (C) Labelled cells locatedin the medial and dorsal part of a somite. (D) The anterior tip of the segmental plate of a stage 11 embryo, showing labelled cells restricted tothe lateral plate. Scale bars, 25 µm in A,B; 50 µm in C,D.

Table 2. Laterality of distribution of labelled descendantsUnilateral Bilateral

Tissue type distribution distribution Total

Neural 15 (79) 4 (21) 19Olfactory region, optic lobes 10 (83) 2 (17) 12Somites 13 (54) 11 (46) 24Intermediate mesoderm 4 0 4Lateral plate 16 (70) 7 (30) 23Extraembryonic tissues and 38 (84) 7 (16) 45surface ectoderm

Total 96 (76) 31 (14) 127

Number of labelled cells found unilaterally or bilaterally after a singlelateral injection in the area pellucida, classified according to variouspresumptive tissue types. These tissue types exclude the notochord, heart andgut, where this type of analysis is not meaningful. The table includes allembryos in which the required details had been recorded (n=127). Thenumbers in brackets are % for the tissue type being considered.

2887Fate map of chick epiblast

isolated are often very large and often comprised more thanone germ layer.

In addition, many reviewers (e.g. Pasteels, 1945; Romanoff,1960; Nicolet, 1971; Balinsky, 1975) have failed to distinguishbetween fate maps and specification maps and have amalga-mated results from the literature to construct composites thatprobably have little value. The maps produced by Butler(1935) and by Rudnick (1948) stand out from othersbecause they appear to have been constructed morecarefully by attempts to separate the layers. In the formercase the description of the early embryos is clear enoughto allow us to conclude that it refers to embryos at aroundstage XII.

We have therefore compared our results with those ofButler (1935) and Rudnick (1948) (Fig. 6) for stages XIIand 2-3, respectively. Several areas, like neuroectoderm,eye, notochord and heart are in equivalent positions in ourfate maps and in the published specification maps. Theonly territory that does differ is the presumptive gut. Thisis smaller in our fate map than in Butler’s specificationmap. When parts of the epiblast (including area opaca)are isolated and cultured, they become gut irrespective oftheir original positions. However, in normal develop-ment, only the posterior one third of the epiblast becomesgut. This indicates that cell interactions are required afterstage XII for restricting the gut territory, implying thatone or more inductive interactions are involved in theearly development of the gut.

Cell- or region-specific markers and the originof early mesendodermal cellsStaining with monoclonal antibody HNK-1 reveals amosaic, salt-and-pepper pattern in chick embryos atstages XII-XIII (Canning and Stern, 1988). At stages XII-2, staining is graded in the posterior-to-anterior andmedial-to-lateral directions, with the primitive streak con-taining more immunoreactive cells than more remoteregions of epiblast. By immunogold labelling andablation experiments, Stern and Canning (1990) showedthat the HNK-1-positive cells at stages XII-XIII are pre-cursors of the mesendoderm of the early primitive streak.In agreement with these patterns, our fate maps, as wellas those of previous authors, show that the posteriorepiblast and that surrounding the streak contribute morecells to the mesendoderm than anterior and lateralregions.

The finding that the epiblast contains a mixture ofHNK-1-positive and -negative cells, together with theresults of immunogold lineage analysis, led to the sug-gestion (Stern and Canning, 1990) that precursors of theearly mesendoderm are mixed with other cells in thistissue before appearance of the primitive streak. Our fatemaps cannot provide an independent test of this becausethe method used labels groups of cells rather than singleones. Future analysis of the descendants of single epiblastcells will be required to provide further insights into thisquestion.

Several authors have described restricted expression ofother markers at early stages of chick development (stageX-3). For example the carbohydrates FC10.2 (Loveless etal. 1990) and (NAc-lac)n (Thorpe et.al. 1988), the

cytoskeletal proteins vimentin and cytokeratin (Page, 1989)and the homeobox gene goosecoid (Izpisúa-Belmonte et al.1993) are expressed in certain subsets of cells in the earlyembryo. In general the investigators concluded, however, thatthese are markers specific for cell states such as ingression,movement or ‘organizing’ properties rather than for particularfates.

stage XI-XIIIstage XFATE SPECIFICATIONFATE

stage 2-3stage XIVFATE SPECIFICATIONFATE

Gut

Neuroectoderm

Olfactory Region, Optic Lobe

Medial Somite

Notochord

Prechordal Plate

Heart

Lateral Somite

Lateral Plate

Fig. 6. Summary fate maps at different stages of development: X, XI-XIII,XIV and 2-3. Each line is made to enclose all of the positions contributing toeach prospective cell type. In cases where there were too few data to allowus to draw a line with confidence, dashed lines are used. For stages XI-XIIIand 2-3, the right half of the diagram shows the specification maps producedby other authors for these stages, for comparison. Only those cell types forwhich there is specification information available (Butler, 1935; Rudnick,1948) are shown. The preprimitive streak specification map is based onButler (1935); the map for primitive streak stage embryos is based onRudnick (1948). In these specification maps, the frequency of each cell typeis represented as the density of symbols.

2888

Origin of Hensen’s node and of the ‘organizer’propertyHensen’s node, the ‘organizer’ of the amniote embryo, hasbeen shown by many studies to contain several distinct celltypes: gut endoderm, prechordal plate, notochord/headprocess, the medial halves of the somite mesoderm and floorplate of the neural tube (e.g. Rosenquist, 1966; Selleck andStern, 1991; Schoenwolf, 1992; Schoenwolf et al., 1992).Which cell populations of the early embryo give rise to thenode? Although we have not analyzed the movements of eachregion in detail at intermediate stages, our data are consistentwith the view that regions containing various presumptive celltypes later found in the node may come together at the posteriormidline very early, at about stage XII. After this, but stillbefore primitive streak formation (about stage XIV), theymove together to a position close to the centre of the blasto-derm, where Hensen’s node will eventually form (Spratt,1946).

In a recent study, Izpisúa-Belmonte et al. (1993) showed thatthe homeobox gene goosecoid, a marker for organizer cells, isfirst expressed in a small population of cells in the middle layerof the prestreak stage embryo, associated with Koller’s sickle.Fate mapping with DiI reveals them to be precursors of someof the cells of Hensen’s node, which also express this gene.However, the same study demonstrated that these earlygoosecoid-expressing cells have the ability to induce othersalso to express the gene, and the node itself contains more cellsthan can be accounted for from the early expressing popula-tion. Therefore, the node appears to be derived from at leasttwo distinct populations of cells: one group, found in themiddle layer around Koller’s sickle from stage X, and others,in the epiblast.

Our fate maps show the locations of the territories occupiedby cell types derived from the node in the epiblast betweenstages X and XIV. In the future, it will be interesting to inves-tigate the onset of inducing ability: is it associated with anyparticular prospective cell type, or does it require a particularcombination of cell types?

This study was funded by a Wellcome Trust Prize Studentship toYH and by grants from the Human Frontier Science Program (heldjointly with Drs E. M. De Robertis and P. Gruss) and the WellcomeTrust to C. D. S. We are grateful to Dr A. Stoker for allowing accessto his computer, to Dr T. Cunnane for his generous help with printingof computer graphics in colour, and to Mr. G.J. Carlson for skilfulltechnical assistance. We are also grateful to Drs Kirstie Lawson,Julian Lewis and Jonathan Slack for their helpful comments on themanuscript.

REFERENCES

Balinsky, B. I. (1975). An Introduction to Embyology. W. B. Saunders,London.

Ballard, W. W. (1973). A new fate map for Salmo gairdneri. J. Exp. Zool. 184,49-74

Ballard, W. W. (1981). Morphogenetic movements and fate maps ofvertebrates. Am. Zool. 21, 391-399.

Ballard, W. W. (1982). Morphogenetic movements and fate maps of thecrypriniform teleost, Catostomus commersoni (Lacepède). J. Exp. Zool. 219,301-321.

Butler, E. (1935). The developmental capacity of regions of the unincubatedchick blastoderm as tested in chorio-allantoic grafts. J. Exp. Zool. 70, 387-338.

Canning, D. R. and Stern, C. D. (1988). Changes in the expression of thecarbohydrate epitope HNK-1 associated with mesoderm induction in thechick embryo. Development 104, 643-655.

Dalton, A. J. (1935). The potencies of portions of young chick blastderms astested in chorio-allantoic grafts. J. Exp. Zool. 71, 17-51.

Delarue, M. and Boucaut, J. C. (1992). A fate map of superficial and deepcircumblastoporal cells in the early gastula of Pleurodeles waltl.Development 114, 135-146.

Eyal-Giladi, H. and Kochav, S. (1976). From cleavage to primitive streakformation: a complementary normal table and a new look at the first stages ofthe development of the chick. Dev. Biol. 49, 321-337.

Gallera, J. (1971b). Primary induction in birds. Adv. Morphogen. 9, 149-180. Gallera, J. and Nicolet, G. (1969). Le pouvoir inducteur de l’endoblaste

presomptif contenu dans la ligne primitive jeune de poulet. J. Embryol. Exp.Morph. 21, 105-118.

Gräper, L. (1929). Die Primitiventwicklung des Hühnchens nachstereokinematischen Untersuchungen, kontrolliert durch vitaleFarbmarkierung und verglichen mit der Entwicklung anderer Wirbeltiere.Wilhelm Roux. Arch. Entwmech. Org. 116, 382-429.

Hamburger, V. and Hamilton, H. L. (1951). A series of normal stages in thedevelopment of the chick embryo. J. Morph. 88, 49-92.

Hama, T. (1978). Dynamics of the organizer. B. New findings on theregionality and morphogenetic movements of the organizer. In Organizer - aMilestone of a Half-century since Spemann. (ed. Nakamura, O. andToivonen, S.) pp. 71-90. Elsevier/North-Holland Biomedical Press.

Ho, R. K. (1992). Axis formation in the embryo of the zebrafish Brachydaniorerio. Development 1992 Supplement, 65-73.

Hoadley, L. (1926a). The in situ development of sectioned chick blastderms.Arch. Biol. 36, 225-309.

Hoadley, L. (1926b). Developmental potencies of parts of the early blastodermof the chick. I. The first appearance of the eye. J. Exp. Zool. 43, 151-178.

Hoadley, L. (1926c). Developmental potencies of parts of the early blastodermof the chick. II. The epidermis and the feather primordia. J. Exp. Zool. 43,179-196.

Hoadley, L. (1926d). Developmental potencies of parts of the early blastodermof the chick. III. The nephros, with especial reference to the pro- andmesonephric portions. J. Exp. Zool. 43, 197-223.

Hoadley, L. (1927). Concerning the organization of potential areas in the chickblastoderm. J. Exp. Zool. 48, 459-473.

Honig, M. G. and Hume, R. I. (1989). DiI and DiO: versatile fluorescent dyesfor neuronal labeling and pathway tracing. Trends Neurosci. 12, 333-336.

Hunt, T. E. (1937). The development of gut and its derivatives from themesoderm of early chick blastoderms. Anat. Rec. 68, 349-370.

Izpisúa-Belmonte, J. C., De Robertis, E. M., Storey, K. G. and Stern, C. D.(1993). The homeobox gene goosecoid and the origin of organizer cells in theearly chick blastderm. Cell 74, 645-659.

Keller, R., Shih, J. and Domingo, C. (1992). The patterning and functioningof protrusive activity during convergence and extension of the Xenopusorganiser. Development 1992 Supplement, 81-91.

Kimmel, C. B. and Warga, R. M. (1988). Tissue-specific cell lineagesoriginate in the gastrula of the zebrafish. Science 231, 365-368.

Kimmel, C. B. and Law, R. D. (1985a). Cell lineage of zebrafish blastomeresI. Cleavage pattern and cytoplasmic bridges between cells. Dev. Biol. 108,78-85.

Kimmel, C. B. and Law, R. D. (1985b). Cell lineage of zebrafish blastomeresII. Formation of the yolk syncytial layer. Dev. Biol. 108, 86-93.

Kopsch, F. (1926). Primitivstreifen und organbildende Keimbezirke beimHühnchen untersucht mittels elektrolytischer Marken am vital gefarbtenKeim. Z. MikrAnat. Forsch. 35, 512-560.

Kopsch, F. (1934). Die Lage des Materials für Kopf, Primitivstreifen, undGefässhof in der Keimscheibe des unbebruteten Hühnereies und seineEntwicklung während der ersten beiden Tage der Bebrütung. Z. MikrAnat.Forsch. 35, 254-330.

Lawson, K. A., Meneses, J. J. and Pedersen, R. A. (1991). Clonal analysis ofepiblast fate during germ layer formation in the mouse embryo. Development113, 891-991.

Loveless, W., Bellairs, R., Thorpe, S. J., Page, M. and Feizi, T. (1990).Developmental patterning of the carbohydrate antigen FC10. 2 during earlyembryogenesis in the chick. Development 108, 97-106.

Malan, M. (1953). The elongation of the primitive streak and the localization ofthe presumptive chorda-mesoderm on the early chick blastderm studied bymeans of coloured marks with Nile blue sulphate. Arch. Biol. 64, 149-182.

Murray, P. D. F. and Selby, D. (1930). Chorio-allantoic grafts of entire andfragmented blastoderms of the chick. J. Exp. Biol. 7, 404-415.

Y. Hatada and C. Stern

2889Fate map of chick epiblast

Nakamura, O. (1938). Tail formation in the urodele. Zool. Mag. (Tokyo) 50,442-446

New, D. A. T. (1955). A new technique for the cultivation of the chick embryoin vitro. J. Embryol. Exp. Morph. 3, 326-331.

Nicolet, G. (1971). Avian gastrulation. Adv. Morphogen. 9, 231-262. Nieuwkoop, P. D. and Sutasurya, L. A. (1979). Primodial Germ cells in the

Chordates. Cambridge University Press, Cambridge. Olivo, O. M. (1928a). Über die frühzeitig Determinierung der Herzanlage beim

Hühnerembryo und der histologische und physiologische Differenzierung invitro. Anat. Anz. 66, 108-118.

Olivo, O. M. (1928b). Précoce détermination de l’ébauche du coeur dansl’embryon de poulet et sa différenciation histologique et physiologique invitro. C. R. Assoc. Anat. 23, 357-374.

Oppenheimer, J. M. (1936). Processes of localization in developing Fundulus.J. Exp. Zool. 73, 405-444.

Page, M. (1989). Changing patterns of cytokeratins and vimentin in the earkychick embryo. Development 105, 97-107.

Pannett, C. A. and Compton, A. (1924). The cultivation of tissues in salineembryonic juice. Lancet 206, 381-384.

Pasteels, J. (1936). Études sur la gastrulation des vertebrés méroblastiques. I.Teleostéens. Arch. Biol. 47, 205-308.

Pasteels, J. (1937). Études sur la gastrulation des Vertébrés méroblastiques II.Reptiles. III. Oiseaux. IV. Conclusions générales. Arch. Biol. 48, 105-488.

Pasteels, J. (1940). Un aperçu comparatif de la gastrulation chez les Chordés.Biol. Rev. 15, 59-106.

Pasteels, J. (1945). On the formation of the primary entoderm of the duck(Anas domestica) and on the significance of the bilaminar embryo in birds.Anat. Rec. 93, 5-21.

Patterson, J. T. (1909). Gastrulation in the pigeon’s egg. A morphological andexperimental study. J. Morph. 20, 65-124.

Rawles, M. E. (1940). The pigment-forming potency of early chickblastoderm. Proc. Natn. Acad. Sci. USA 26, 86-94.

Rawles, M. E. (1943). The heart forming areas of the chick blastoderm.Physiol. Zool. 16, 22-41.

Romanoff, A. L. (1960). The Avian Embryo. Structural and FunctionalDevelopment. MacMillan, New York.

Rosenquist, G. C. (1966). A radioautographic study of labeled grafts in thechick blastoderm. Development from primitive streak stages to stage 12.Contrib. Embryol. Carnegie Inst. Washington 38, 71-110.

Rudnick, D. (1932). Thyroid-forming potencies of the early ckick blastoderm.I. J. Exp. Zool. 62, 287-317.

Rudnick, D. (1935). Regional restriction of potencies in the chick duringembryogenesis. J. Exp. Zool. 71, 83-99.

Rudnick, D. (1938a). Thyroid-forming potencies of the early chick blastoderm.II. J. Exp. Zool. 79, 399-427.

Rudnick, D. (1938b). Differentiation in culture of pieces of the early chickblastoderm. II. Short primitive streak stages. J. Exp. Zool. 89, 399-427.

Rudnick, D. (1944). Early history and mechanics of the chick blastoderm. Areview. Quart. Rev. Biol. 19, 187-212.

Rudnick, D. (1948). Prospective areas and differentiation potencies in thechick blastoderm. Ann. NY Acad. Sci. 49, 761-772.

Rudnick, D. (1961). Teleosts and Birds. In Analysis of Development (ed.Wilier, P. A., Weiss, P. and Hamburger, V.). pp. 297-314. Saunders,Philadelphia.

Schoenwolf, G. C. (1992). Morphological and mapping studies of theparanodal and postnodal levels of the neural plate during chick neurulation.Anat. Rec. 233, 281-290.

Schoenwolf, G. C., García-Martínez, V. and Días, M. S. (1992). Mesodermmovement and fate during avian gastrulation and neurulation. Devl.Dynamics 193, 235-248.

Selleck, M. A. J. and Stern, C. D. (1991). Fate mapping and cell lineageanalysis of Hensen’s node in the chick embryo. Development 112, 615-626.

Serbedzija, G. N., Fraser, S. E. and Bronner-Fraser, M. (1990). Pathways ofneural crest cell migration in the mouse as revealed by vital dye labelling.Development 108, 605-612.

Shih, J. and Keller, R. (1992a). Cell motility driving mediolateral intercalationin explants of Xenopus laevis. Development 116, 901-914.

Shih, J. and Keller, R. (1992b). Patterns of cell motility in the organizer anddorsal mesoderm of Xenopus laevis. Development 116, 915-930.

Slack, J. (1991). From Tadpole to Frog (2nd edition). Cambridge UniversityPress, Cambridge.

Spratt, N. T. (1940). An in vitro analysis of the organization of the eye-formingarea in the early chick blastderm. J. Exp. Zool. 85, 171-209.

Spratt, N. T. (1942). Location of organ-specific regions and their relationshipto the development of the primitive streak in the early chick blastoderm. J.Exp. Zool. 89, 69-101.

Spratt, N. T. (1946). Formation of the primitive streak in the explanted chickblastoderm marked with carbon particles. J. Exp. Zool. 103, 259-304.

Spratt, N. T. (1947). Localization of the prospective neural plate in theprimitive streak blastoderm of the chick. Anat. Rec. 99, 654.

Stern, C. D. (1990). The marginal zone and its contribution to the hypoblastand primitive streak of the chick embryo. Development 109, 667-682.

Stern, C. D. and Canning, D. R. (1990). Origin of cells giving rise tomesoderm and endoderm in chick embryo. Nature 343, 273-275.

Stern, C. D. and Ireland, G. W. (1981). An integrated experimental study ofendoderm formation in avian embryos. Anat. Embryol. 163, 245-263.

Strehlow, D. and Gilbert, W. (1993). A fate map for the first cleavages of thezebrafish. Nature 361, 451-453.

Thorpe, S. J., Bellairs, R. and Feizi, T. (1988). Developmental patterning ofcarbohydrate antigens during early embryogenesis of the chick: expressionof antigens of the poly-N-acetyllactosamine series. Development 102, 193-210.

Trinkaus, J. P. and Fink, R. D. (1992). On the convergent cell movements ofgastrulation in Fundulus. J. Exp. Zool. 261, 40-61.

Vakaet, L. (1960). Quelques précisions sur la cinématique de la ligne primitivechez le poulet. J. Embryol. Exp. Morph. 8, 321-326.

Vakaet, L. (1970). Cinephotomicrographic investigations of gastrulation in thechick blastoderm. Arch. Biol. (Liège). 81, 387-426.

Vakaet, L. (1984). Early development of birds. In Chimeras in DevelopmentalBiology (eds. Le Duarin, N. M. and McLaren, A.). pp. 71-88. AcademicPress, London.

Vakaet, L. (1985). Morphogenetic movements and fate maps in the avianblastoderm. In Molecular Determinants of Animal Form (ed. Edelman, G.M.). pp. 99-109. Alan R. Liss, New York.

Waddington, C. H. (1933). Induction by the endoderm in birds. Roux Arch.EntwMech. Org. 128, 502-521.

Waddington, C. H. (1935). The development of isolated parts of the chickblastoderm. J. Exp. Zool. 71, 273-288.

Warga, R. M. and Kimmel, C. B. (1990). Cell movement during epiboly andgastrulation in zebrafish. Development 108, 569-580.

Wetts, R. and Fraser, S. (1989). Slow intermixing of cells during Xenopusembryogenesis contributes to the consistency of the blastomere fate map.Development 105, 9-15.

Wetzel, R. (1929). Untersuchungen am Hühnchen. Die Entwicklung des Keimswährend der erste beiden Bruttage. Arch. EntwMech. Org. 119, 188-321.

Winkel, G. K. and Pedersen, R. A. (1988). Fate of the inner cell mass inmouse embryos as studied by microinjection of lineage tracers. Dev. Biol.127, 143-156.

(Accepted 15 July 1994)