We report the results of cutting experiments on embryos of the crinoid Florometra serradísima, which produce, in the doliolaria stage, a striped pattern of ciliary bands. Embryos at gastrula and post-hatching elongation stages were divided into anterior and posterior fragments. Complementary fragments express parts of the total pattern without adding extra pattern elements, i.e. the pattern is a mosaic. Some fragments elongate which, from an examination of internal structures, we interpret as due to the elongation and displacement of the mesenteric sac. The number of pattern elements expressed correlates with degree of elongation and internal landmarks correlate with certain external pattern features. This suggests that the pattern mosaic may reside in the internal tissues, i.e. in the mesentoderm, but we are as yet unable to prove this. The results are discussed with reference to the roles of tissues of different germ layer origin in related embryos, including vertebrates, in which the mesentoderm has a significant instructive role.

The development of the doliolaria stage in crinoids involves the conversion of a uniformly ciliated larva to one with regularly spaced ciliary bands. In terms of pattern, the larval epithelium differentiates into what is essentially an array of stripes over about a 20 h period. The first paper in this series (Lacalli & West, 1986) dealt with normal development of the ciliary bands in the crinoid Florometra serradísima. We report here the results of simple cutting experiments designed to demonstrate the regulatory capabilities of this pattern-forming system, i.e. whether missing pattern elements are replaced in embryonic fragments or whether there is any particular pattern feature, band spacing for example, that regulates. On the evidence, the pattern is a mosaic from the earliest stage we have been able to test, the midgastrula, and there is circumstantial evidence for the involvement of underlying mesentodermal tissues in some aspects of morphogenesis, and possibly in pattern formation as well. The results are discussed with reference to the patterning capabilities of the mesentoderm in other deuterostomes, notably recent evidence for selforganization in vertebrate mesoderm (Cooke & Webber, 1985a,b). The crinoid embryo may provide a simple model system for studying early patterning events of this type.

Larval rearing and preparation of specimens for electron microscopy were carried out at the Bamfield Marine Station following methods given in Lacalli & West (1986). For scanning (SEM) and transmission electron microscopy (TEM), fixation was by the semisimultaneous method as described using 2 – 3 % glutaraldehyde as a primary fixative for 1 – 4min before addition of osmium. Variation within these ranges has no obvious effect on fixation quality. Specimens for TEM were block stained in 2 % aqueous uranyl acetate overnight at 60°C, embedded in Spurr’s resin, sectioned and examined without further staining. In most instances individual larvae were selected, oriented and sectioned in their entirety; series of sections were collected on slotted grids at 7 μ m intervals through the whole larva. It was then possible to generate reconstructions (Figs 19-22), count mitotic figures and search for points of contact between the various tissue layers in the same larva.

The cutting experiments were done using a dissecting microscope and fine glass needles. For each developmental stage tested, between 20 and 50 larvae were cut into fragments. The larvae are opaque and SEM is the only means we have so far discovered of clearly demonstrating band patterns. For this reason, and because of the practical difficulties involved in processing large numbers of specimens for SEM, we could not follow fragments individually through development. Instead, using the relative size of anterior and posterior fragments in freshly cut larvae as a measure of the approximate level of transection, equivalent anterior-posterior fragment pairs were pooled in groups of 5 – 10 and processed together. Consistent results were obtained within each such group, and these correlate with the relative position of transection when groups are compared. We cannot be sure of the precise position of transection and so discuss the results in relation to a ‘cutting zone’ with sufficiently generous boundaries to include all cuts intended to produce half-embryos. We have not yet found a way to remove the fertilization membrane from eggs or cleavage stages without affecting development, so gastrulae had to be cut within their membrane. Clean separation of the halves is then difficult and there is frequently loss of tissue. The resulting fragments show more malformations and are consequently more difficult to interpret than post-hatching stages, which give a clear-cut result.

Development of transverse half-embryos

We have performed a variety of cutting experiments on stages from the midgastrula through the period of band formation, but the general nature of the results, illustrating the mosaic nature of the pattern, is best seen in experiments in which embryos were cut transversely into anterior and posterior fragments. We report here on series of experiments of this type, mainly on the development of half-embryos cut at midgastrula (22 h), late gastrula (28 h) and during and following the post-hatching elongation stage (42 and 46h respectively). Band formation begins at about 56h. A 140 h control larva, with a full complement of four bands, apical pit and vestibule, is shown in Fig. 2. Bands are numbered, by convention, from the anterior end. The apical cilia do not form a distinct tuft at this stage, but are instead scattered diffusely over the apical surface. They are therefore easily distinguished from bands and band fragments even when the latter are, as in some experimental fragments, apically positioned.

The type of result obtained from experiments on the post-hatching stages is illustrated in Fig. 1, and typical examples are shown in Figs 3 – 9. It is easy to make an exactly transverse cut, but difficult to be sure of its exact position along the body axis. Cuts within the zone indicated in Fig. 1A, i.e. cuts within ±10% of body length from the midpoint, gave what we refer to as ‘half-embryos’ and two classes of results.

Fig. 1.

Typical results from transection experiments at 42 and 46h. Half-embryos produced by cuts in the zone indicated (A) develop, by 140 h, as shown in B and C. Each pair has an apical pit (p), vestibule (v) and four ciliary bands (numbered 1–4).

Fig. 1.

Typical results from transection experiments at 42 and 46h. Half-embryos produced by cuts in the zone indicated (A) develop, by 140 h, as shown in B and C. Each pair has an apical pit (p), vestibule (v) and four ciliary bands (numbered 1–4).

Fig. 2.

Normal control larva at 140h, ventral view, labels as in Fig. 1, × 400.

Fig 3 – 9. Results of transection experiments on 42 h embryos (except Fig. 8, done at 46 h), larval fragments at 140 h, labels as in Fig. 1. All are to the same scale (x400) except Fig. 7 (x530). Figs 3-5, 8 and 9 show anterior (A) and posterior (B) fragments from larvae transected at a progressively more posterior level. Cuts well forward of the cutting zone in Fig. 1 give (Fig. 3) anterior fragments with scattered tuft cilia and an occasional band fragment; posterior fragments have a pit, no tuft, some defects in band 1, but are otherwise normal. Cuts further back but still forward of the cutting zone give (Fig. 4) anterior fragments with tuft cilia, pit and a single circumferential band (the last two are usually reduced or defective); posterior fragments lack a tuft and pit, have a much reduced band 1 (Fig. 7C gives a better view of the apex of a similar fragment), but are otherwise normal. Cuts within the cutting zone near its anterior margin give (Fig. 5) anterior fragments with a well-developed pit and circumferential band, usually with a medial cross band (visible at arrow in Fig. 6, which shows the underside of the fragment in Fig. 5A); posterior fragments generally lack band 1 or have only small pieces of it. Fig. 7 shows the range of variation in such fragments, from naked dome (Fig. 7A) to rectangular patch (Fig. 7B, arrow) to tonsure-like partial band (Fig. 7C). Figs 8 and 9 show results from cuts near the posterior margin of the cutting zone. In Fig. 8, the posterior fragment has the vestibule and a reduced band 2; the anterior fragment is usually tapered at its posterior end so that band 2, which may also be defective, is shorter than normal, as if it has been split between the two fragments. In anterior fragments expressing the vestibule (Fig. 9), band 2 is of normal size; posterior fragments have only bands 3 and 4.

Fig. 10. Anterior (A) and posterior (B) fragments from a transected 28 h late gastrula, larval fragments at 140 h, × 400, labels as in Fig. 1. In terms of the bands represented, the results are roughly comparable to those in Fig. 5, but corresponding fragments are reduced in overall size, due to tissue loss, and the bands are reduced or fragmentary.

Fig. 2.

Normal control larva at 140h, ventral view, labels as in Fig. 1, × 400.

Fig 3 – 9. Results of transection experiments on 42 h embryos (except Fig. 8, done at 46 h), larval fragments at 140 h, labels as in Fig. 1. All are to the same scale (x400) except Fig. 7 (x530). Figs 3-5, 8 and 9 show anterior (A) and posterior (B) fragments from larvae transected at a progressively more posterior level. Cuts well forward of the cutting zone in Fig. 1 give (Fig. 3) anterior fragments with scattered tuft cilia and an occasional band fragment; posterior fragments have a pit, no tuft, some defects in band 1, but are otherwise normal. Cuts further back but still forward of the cutting zone give (Fig. 4) anterior fragments with tuft cilia, pit and a single circumferential band (the last two are usually reduced or defective); posterior fragments lack a tuft and pit, have a much reduced band 1 (Fig. 7C gives a better view of the apex of a similar fragment), but are otherwise normal. Cuts within the cutting zone near its anterior margin give (Fig. 5) anterior fragments with a well-developed pit and circumferential band, usually with a medial cross band (visible at arrow in Fig. 6, which shows the underside of the fragment in Fig. 5A); posterior fragments generally lack band 1 or have only small pieces of it. Fig. 7 shows the range of variation in such fragments, from naked dome (Fig. 7A) to rectangular patch (Fig. 7B, arrow) to tonsure-like partial band (Fig. 7C). Figs 8 and 9 show results from cuts near the posterior margin of the cutting zone. In Fig. 8, the posterior fragment has the vestibule and a reduced band 2; the anterior fragment is usually tapered at its posterior end so that band 2, which may also be defective, is shorter than normal, as if it has been split between the two fragments. In anterior fragments expressing the vestibule (Fig. 9), band 2 is of normal size; posterior fragments have only bands 3 and 4.

Fig. 10. Anterior (A) and posterior (B) fragments from a transected 28 h late gastrula, larval fragments at 140 h, × 400, labels as in Fig. 1. In terms of the bands represented, the results are roughly comparable to those in Fig. 5, but corresponding fragments are reduced in overall size, due to tissue loss, and the bands are reduced or fragmentary.

(1) Cuts near the anterior margin of the zone gave anterior halves with two bands and a vestibule, posterior halves with two bands only (Figs 1B, 9). (2) Cuts near the posterior margin gave anterior halves with one band, posterior halves with three bands and a vestibule (Figs 1C, 5). Variants of (1) occur, with a reduced second band and the vestibule in the posterior half (Fig. 8). It is typical of anterior halves in (2) for there to be a medial patch or cross band connecting front and back of the principal band to form a figure-eight pattern (Fig. 6). We interpret this as a remnant of one of the axial elements of the initial pattern (see Fig. 12), which would normally disappear in intact larvae. With this exception, fragments never express together more than the normal complement of four bands, though a band may be split between fragments (Fig. 8), and there is never more than a single tuft, apical pit and vestibule between them. Progressively smaller anterior fragments, from cuts forward of the cutting zone, express progressively less in terms of pattern. They form either tuft cilia plus a partial single band without medial elements (Fig. 4A) or, in the smallest fragments, tuft cilia only (Fig. 3A). Complementary posterior fragments form more of the normal pattern (Figs 3B, 4B) in a smooth progression depending on fragment size. Fig. 7 shows a series of progressively more complete versions of band 1 at the anterior end of posterior fragments of increasing size. The posterior end of anterior fragments shows a similar progression in the expression of all or parts of bands 3 and 4. We can identify no specific structure or pattern element that always forms at the new surface established by a cut.

In general then, the pattern at 42 and 46 h appears to be a mosaic. Pairs of fragments together express no more than the normal complement of pattern elements and there is no sign of replacement of missing parts or adjustment to new boundaries. The two fragments of a given pair do, however, differ in morphogenetic capability and in the proportion of the normal pattern each expresses, depending on which of them receives the vestibule. Normal larvae elongate after hatching, and increase by about 40 % in length. By 42 h they are 9/10 of their final length. Half-embryos generated at this stage can elongate further by up to 35 %, but only the fragment with the vestibule does so. Fragments without a vestibule remain spherical. Based on internal anatomy (see below), our interpretation is that only the fragment receiving the enterohydrocoelic vesicle, which is closely associated with the development of the vestibule, will elongate. And, the more elongate a fragment, the more pattern elements it produces. Band spacing is not constant in fragments (see Fig. 9A,B), but differs less from the normal as a consequence of elongation than would otherwise be the case. We observe spacings in the range of 55 – 75 μ m in our various fragments, but in well-formed, elongate posterior fragments like those shown in Figs 3 – 5, spacing is typically closer to the normal value of 65 μ m.

The experiments on gastrulae give similar results to those on post-hatching stages, though there are malformations, frequent size reduction and often less than a full complement of four bands. An example is shown in Fig. 10. The posterior half in this case is similar to that in Fig. 8B, but the anterior half expresses much less than the corresponding anterior fragment from 42 h (Fig. 8A). It instead has only a single rather irregular band, much like that in Fig. 5A, but without the medial cross-band. We attribute both decrease in larval size and incompleteness of pattern to the loss of tissue that occurs when gastrulae are cut. This is consistent with the notion that pattern is already a mosaic. Band spacing is reduced in these smaller larvae as well, to 50 μ m in some instances.

A surprising result of all of the above experiments is the apparent unequal division of the potential to form pattern elements between anterior and posterior halves. Each can produce two bands in some instances, but it is more common to get three bands in posterior halves and a single one in anterior halves, while the converse is never encountered. In normal development, at the time of pattern expression, the pattern elements responsible for each of the four bands are spaced more-or-less uniformly over the whole surface of the ectoderm (Fig. 12A). From our experiments, a fate map for 42h and earlier would displace the pattern posteriorly so as to place the region lying between the rudiments of bands 1 and 2 at least as far back as the front of the cutting zone (Fig. 12B). Note that anterior halves that fail to elongate have surface areas about equal to half the surface of a normal larva (Fig. 11) and form only one band. In an intact larva the same epithelium would have formed two bands.

Fig 11.

Size, as surface area, of complementary anterior and posterior fragments as a function the ratio in length of anterior/posterior fragments. Arrows indicate, by figure number, where along the horizontal axis the examples illustrated in Figs 5, 8 and 9 would fall, which roughly defines the range of results obtained for cuts within the cutting zone. Over this range, because the experimental fragments elongate, pairs of fragments are together larger (i.e. generate more total surface) than a normal intact larva. The straight line marks half the surface area of the normal control.

Fig 11.

Size, as surface area, of complementary anterior and posterior fragments as a function the ratio in length of anterior/posterior fragments. Arrows indicate, by figure number, where along the horizontal axis the examples illustrated in Figs 5, 8 and 9 would fall, which roughly defines the range of results obtained for cuts within the cutting zone. Over this range, because the experimental fragments elongate, pairs of fragments are together larger (i.e. generate more total surface) than a normal intact larva. The straight line marks half the surface area of the normal control.

The potential to form particular pattern elements, if intrinsic to the ectoderm, would appear therefore to be shifted forward comparatively late in development. This is not simply a consequence of cells being displaced from a localized proliferative zone in the posterior end, since this would occur too slowly to be of any significance. Based on cell counts from our previous work on the dorsal half of the larva (Lacalli & West, 1986), we estimate that a typical dorsal half, including half of both the anterior and posterior cap, comprises 1800 cells at 42 h and 2250 cells at the beginning of band formation. If proliferation on the dorsal surface were, as an extreme, restricted to its posterior half, 900 cells would have to generate 1850 cells in 14h, i.e. about a doubling of posterior cell numbers would be needed. The half-way mark on the dorsal surface of the 42 h stage would move, at most, to the two-thirds mark, which is nearly sufficient to account for the pattern shift observed. We do not, however, find a substantial difference in the number of mitotic figures in anterior versus posterior halves in sections through whole larvae. Their distribution is instead roughly uniform over the length of the larva (Fig. 13). As an alternative explanation, it may be that pattern is generated at an early stage but resides elsewhere than in the ectoderm, presumably in some underlying tissue, so that shifts in ectodermal expression of pattern would reflect changes in the arrangement or position of internal structures as they undergo normal morphogenesis. This possibility has led us to examine the development of the mesentoderm in more detail.

Fig. 12.

(A) The initial band pattern expressed in the larval ectoderm based on Lacalli & West (1986), a view of the dorsal surface. Schematic, anterior end up, stippled regions are developing band, unstippled regions are non-ciliated presumptive interband. Horizontal elements in the pattern are retained as bands (numbered), while the axial connections between these progressively disappear. (B) From experiment, the apparent distribution of potential to form the pattern elements in (A) at 42 h. The pattern is displaced towards the posterior end and cuts at the midpoint (line) would transect the axial connection between bands 1 and 2.

Fig. 12.

(A) The initial band pattern expressed in the larval ectoderm based on Lacalli & West (1986), a view of the dorsal surface. Schematic, anterior end up, stippled regions are developing band, unstippled regions are non-ciliated presumptive interband. Horizontal elements in the pattern are retained as bands (numbered), while the axial connections between these progressively disappear. (B) From experiment, the apparent distribution of potential to form the pattern elements in (A) at 42 h. The pattern is displaced towards the posterior end and cuts at the midpoint (line) would transect the axial connection between bands 1 and 2.

Fig. 13.

Total numbers of mitotic figures in longitudinal sections take at 7 pm intervals through entire specimens, for gastrulae at 22 and 28 h and hatched larvae at 39 and 50h. Counts are by zone along the anterior-posterior axis as shown in the diagram below each pair of graphs.

Fig. 13.

Total numbers of mitotic figures in longitudinal sections take at 7 pm intervals through entire specimens, for gastrulae at 22 and 28 h and hatched larvae at 39 and 50h. Counts are by zone along the anterior-posterior axis as shown in the diagram below each pair of graphs.

Development of the mesentoderm

Florometra embryos are too opaque for direct observation of internal morphogenetic events. A timetable of the main early events is given in Fig. 14 based on TEM sections and reconstructions (Figs 1524) of nine representative stages between 22 and 64 h. Crinoids gastrulate by invagination (Figs 15,16) as in other echinoderms, but subsequent events differ from what is seen in the other classes. As the blastopore closes, near hatching, a substantial mass of mesenchyme is released from the tip of the archenteron (Fig. 17) while the remaining, epithelial portion of the archenteron forms a closed chamber, the mesenteric sac, at the posterior end of the larva (Fig. 18). The mesenteric sac is at first flattened, but is progressively converted, over a period of about 10h, to a more elongate, hourglass shape as a constriction forms about its middle (Figs 19-21). The anterior portion of the sac becomes the enterohydrocoel, which gives rise to two coelomic derivatives, the axocoel and hydrocoel, and the endodermal part of the larval gut. The posterior portion of the sac expands laterally (Figs 22, 23) and divides to give the right and left body coeloms, the somatocoels.

Fig. 14.

Timetable for the main early events in the development of the mesentoderm. It shows early and late gastrulae (the mesentoderm and its epithelial derivative, the mesenteric sac, have cell boundaries sketched in), formation of the mesenchyme (stippled), and three stages in the morphogenesis of the mesenteric sac.

Fig. 14.

Timetable for the main early events in the development of the mesentoderm. It shows early and late gastrulae (the mesentoderm and its epithelial derivative, the mesenteric sac, have cell boundaries sketched in), formation of the mesenchyme (stippled), and three stages in the morphogenesis of the mesenteric sac.

Fig 15–18.

Longitudinal sections through developing larvae at 22, 28, 36 and 39h, all x690. The figures show gastrulation (Figs 15, 16), with the gastrulae still within the fertilization membrane, mesenchyme formation (Fig. 17), and the larva soon after hatching (Fig. 18) with its anterior mass of mesenchyme (meh) and mesenteric sac (me).

Fig 15–18.

Longitudinal sections through developing larvae at 22, 28, 36 and 39h, all x690. The figures show gastrulation (Figs 15, 16), with the gastrulae still within the fertilization membrane, mesenchyme formation (Fig. 17), and the larva soon after hatching (Fig. 18) with its anterior mass of mesenchyme (meh) and mesenteric sac (me).

Fig 19–22.

Larvae at 39, 45, 50 and 62 h, reconstructions from sections showing the external surface (dashed lines), and the outer (solid lines) and inner, luminal surface (dotted lines) of the mesenteric sac. All × 310.

Fig 19–22.

Larvae at 39, 45, 50 and 62 h, reconstructions from sections showing the external surface (dashed lines), and the outer (solid lines) and inner, luminal surface (dotted lines) of the mesenteric sac. All × 310.

Overall, the mesenteric sac first lengthens, initially by about 50 μ m or 65%, and then shortens as the somatocoels constrict, but it is shifted forward at the same time so that its anterior end stays well forward of the midpoint of the larva (Fig. 25). These changes are associated with a posterior shift in the position of the mesenchyme. From its initial, anterior position it envelops the enterohydrocoel, fills in the zone of constriction and eventually piles up behind the somatocoels. We cannot rule out the possibility that some of this mesenchyme arises from the mesenteric sac secondarily, during its morphogenesis, but we have not seen this in sections. It seems more likely that the posterior cells originate in the anterior mesenchyme and move posteriorly over the surface of the sac as morphogenesis proceeds. When first formed, the mesenteric sac is in direct contact with ectoderm over all but its anterior surface. As the enterohydrocoel forms, it becomes surrounded by mesenchyme, which separates it completely from the ectoderm (Fig. 24). The somatocoels retain contact laterally, but not dorsoventrally, where an intervening layer of mesenchyme develops just as the posterior mass of mesenchyme begins to make its appearance. The elongation and displacement of the mesenteric sac, whether intrinsic to the sac itself or driven by mesenchyme, provides an explanation for the elongation seen in experimental fragments. Fragments with a vestibule, which we interpret as those containing a significant piece of enterohydrocoel, and which are usually posterior fragments, elongate. At 42 and 46 h, cuts at the front and back of the cutting zone would segregate most of the developing enterohydrocoel into, respectively, either the anterior or posterior fragment. Elongation of the sac in normal larvae is apparently accommodated by the displacement of anterior mesenchyme. In posterior halves there would be nothing to displace, so the fragment as a whole would have to stretch. Anterior fragments simply form spheres. Mesenchyme alone, which is what such fragments must contain, presumably cannot drive elongation.

Fig. 24.

Frontal section at 62 h showing the separation of the enterohydrocoel (en) from the somatocoel rudiment (sm), also in the process of subdividing. Shows both the anterior (meh) and posterior (*) mesenchyme. × 690.

Detail of Fig. 23 showing the layer of mesenchyme (between arrows) separating the wall of the enterohydrocoel (en) from the ectoderm (ec). × 2117.

Fig. 24.

Frontal section at 62 h showing the separation of the enterohydrocoel (en) from the somatocoel rudiment (sm), also in the process of subdividing. Shows both the anterior (meh) and posterior (*) mesenchyme. × 690.

Detail of Fig. 23 showing the layer of mesenchyme (between arrows) separating the wall of the enterohydrocoel (en) from the ectoderm (ec). × 2117.

Fig. 25.

The proportion of the body axis (position in μm from the posterior end, at 0) occupied by various structures during the period of mesentodermal morphogenesis. The region occupied by the mesenteric sac is stippled. The solid line, for reference, marks 50% of body length.

Fig. 25.

The proportion of the body axis (position in μm from the posterior end, at 0) occupied by various structures during the period of mesentodermal morphogenesis. The region occupied by the mesenteric sac is stippled. The solid line, for reference, marks 50% of body length.

The position of the mesenteric sac at the time of band formation is such as to bring it into register with pattern elements then appearing in the ectoderm (Fig. 26). The initial pattern in Florometra consists of anterior and posterior nonciliated caps and three tiers of interband domains, each offset relative to the next (Fig. 26B). The middle and posterior interband elements line up roughly with the centres of the enterohydrocoel and the undivided somatocoel rudiment (Fig. 26A). It may be significant that beneath the posterior pair of interband domains the somatocoels are in direct contact with ectoderm, but there is no similarly simple relation between the remaining pattern elements and underlying tissues. The enterohydrocoel is covered with a layer of mesenchyme and there is nothing but a solid mass of mesenchyme adjacent to the anterior interband. These juxtapositions may have a role in specifying pattern. It is not entirely clear, however, whether there is sufficient displacement of the mesenteric sac to account for the whole of the shift in pattern we observe experimentally.

Fig. 26.

Comparison of the position of internal structures along the body axis (A) with that of the main pattern elements expressed in the ectoderm (B). Numbers are position along the axis from the posterior end as a fraction of total body length. (A) Position of the anterior end and midpoint of the enterohydrocoel, and the midpoint of the somatocoe) rudiment, range reflecting variability between stage for larvae 50 to 62 h old.

(B) Midpoints for each of the three tiers of interband domains, from SEM preparations of 62 and 64 h larvae, range reflecting variability between specimens.

Fig. 26.

Comparison of the position of internal structures along the body axis (A) with that of the main pattern elements expressed in the ectoderm (B). Numbers are position along the axis from the posterior end as a fraction of total body length. (A) Position of the anterior end and midpoint of the enterohydrocoel, and the midpoint of the somatocoe) rudiment, range reflecting variability between stage for larvae 50 to 62 h old.

(B) Midpoints for each of the three tiers of interband domains, from SEM preparations of 62 and 64 h larvae, range reflecting variability between specimens.

Among echinoderms, the sea urchin embryo has been the organism of choice for most experimental investigations. Previous work on crinoids is limited to one study on banded, late-stage larvae by Runnstrom (1925), but this provides no information on how bands form. Transection experiments similar to those reported here have been done on asteroid and holothurian gastrulae and larvae by Hórstadius (1928, 1973), and in general show these to be developmental mosaics. Missing parts are replaced in some instances, but by a secondary regenerative mechanism. The literature on sea urchin embryos, in contrast, emphasizes their ability to regulate (e.g. Hórstadius, 1973), which may simply reflect the timing of the experiments, generally done on cleavage stages in the case of the sea urchin, and on gastrulae or older stages in the other groups. Observations on the development of isolated blastomeres in asteroids show, as in sea urchins, that they have considerable regulatory capacity (Dan-Sohkawa & Satoh, 1978).

Our results on crinoid pattern conforms with the asteroid and holothurian results in that the pattern behaves as a mosaic as early as the midgastrula. There is a degree of ‘regulation’ with regard to size, i.e. some fragments elongate to more nearly normal proportions, but this can be explained simply as a result of elongation and displacement of the mesenteric sac. It is more difficult to explain the correlation between degree of elongation in fragments and the number of pattern elements expressed. Longer fragments make more bands, but we have no evidence for the maintenance of constant spacing per se as a pattern feature. It would seem instead that whatever is responsible for elongation may also be involved in specifying pattern.

Knowing the ultimate site of pattern formation, i.e. the tissue responsible, then becomes important. Ectoderm has been shown to contain pattern information in the sea urchin embryo. The site of the stomodeum and larval arms is established first in the ectoderm, and while mesenchyme is required for continued, normal morphogenesis, it is probably guided by an ectodermal template (Gustafson & Wolpert, 1967). Further, patterns of enzyme distribution correlating with later pattern events can be demonstrated in the ectoderm quite early in development (Czihak, 1962, 1963). The mesentoderm is, however, also capable of organizing its own morphogenesis to a degree. It first produces masses of mesenchyme from quite specific locations and then subdivides by constriction, which can occur even in exogastrulae in the complete absence of normal tissue relationships. Once formed, the mesenchyme also has some capacity for selforganization. It can, for example, produce complex, species-specific skeletal structures in isolation in culture (Okazaki, 1975). The main role usually ascribed to the mesenchyme, however, is to act mechanically under the influence of information apparently generated and maintained in the ectodermal and endodermal epithelia to coordinate their morphogenesis so as to shape and join them in an appropriate fashion. Mesenchyme may play a similar role in the crinoid embryo, but it is difficult to observe this directly in the solid, rather nondescript mass of mesenchyme that is formed. The morphogenesis of the mesenteric sac is associated with a migration towards the posterior of mesenchyme as a thin layer maintaining contact both with ectoderm and with the sac itself. Pattern information in either of these two epithelia could provide the necessary guidance. The constriction of the mesenteric sac precedes any sign of ectodermal pattern by a considerable period (approx. 20 h). This process could nevertheless be controlled by ectodermal cues translated into morphogenetic forces by the mesenchyme, but could also be a consequence of the organized activities of mesenchyme alone or be an intrinsic capability of the mesenteric sac. Our experiments do not allow a clear choice between these alternatives. The position of the mesenteric sac can be shown to correlate with external pattern elements and shift in the potential to form these (Fig. 26). The flow of instructional information could still, however, be in either direction: the ectoderm could instruct mesentoderm and its derivatives or vice versa, or there could be a more complex sequence of exchanges involving both.

It is useful in this regard to refer to recent work on vertebrate pattern (Cooke & Webber, 1985a,b) demonstrating the capacity for selforganization in mesoderm. Crinoid embryos could conceivably represent a parallel but more primitive situation in which, unlike the sea urchin, mesodermal derivatives including mesenchyme generate all or most of the pattern and ectoderm responds passively to this. It may be significant that what we are concerned with in the crinoid is a ciliary band pattern that is related to the undulatory single band proposed as an evolutionary precursor of the vertebrate neural tube (Garstang, 1928), and that ciliary bands are a primary source of nerves in primitive larvae (Lacalli & West, 1985; Lacalli, 1982; Burke, 1978). In vertebrates, ectoderm is induced to form neural structures under instruction from an essentially mesodermal derivative, the notochord. It would not be surprising to find that band pattern in echinoderm larvae, and particularly the band rearrangements that occur during development, are under similar control by mesodermal derivatives of some type. We have previously dealt with crinoid development in relation to other stripe-forming systems, notably insects (Lacalli & West, 1986). With regard to the overall positioning of the initial pattern elements, however, it is the possibility of a link with primary events in vertebrate morphogenesis that is of special interest. The crinoid may provide a useful model for such events.

This work was supported by NSERC Canada. We thank the director and staff of the Bamfield Marine Station for their assistance.

Burke
,
R. D.
(
1978
).
The structure of the nervous system of the pluteus larva of Stronglylocentrotus purpuratus
.
Cell Tiss. Res
.
191
,
233
247
.
Cooke
,
J.
&
Webber
,
J. A.
(
1985a
).
Dynamics of the control of body pattern-in the development of Xenopus laevis. I. Timing and pattern in the development of dorsoanterior and posterior blastomere pairs isolated at the 4-cell stage
.
J. Embryol. exp. Morph
.
88
,
85
112
.
Cooke
,
J.
&
Webber
,
J. A.
(
1985b
).
Dynamics of the control of body pattern in the development of Xenopus laevis. II. Timing and pattern in the development of single blastomeres isolated at the 2-cell stage
.
J. Embryol. exp. Morph
.
88
,
113
133
.
Czihak
,
G.
(
1962
).
Entwicklungsphysiologie der Echinodermen
.
Fortschr. Zool
.
14
,
238
267
.
Czihak
,
G.
(
1963
).
Entwicklungsphysiologische Untersuchungen an Echiniden
.
Wilhelm Roux’ Arch. EntwMech. Org
.
154
,
272
292
.
Dan-Sohkawa
,
M.
&
Satoh
,
N.
(
1978
).
Studies on dwarf larvae developed from isolated blastomeres of the starfish Asterina pectinifera
.
J. Embryol. exp. Morph
.
46
,
171
185
. .
Garstang
,
W.
(
1928
).
The morphology of the Tunicata, and its bearing on the phylogeny of the Chordata
.
Q. JI. Microsc. Sci
.
72
,
51
187
.
Gustafson
,
T.
&
Wolpert
,
L.
(
1967
).
Cellular movement and contact in sea urchin morphogenesis
.
Biol. Rev
.
42
,
442
498
.
Hôrstadius
,
S.
(
1928
).
Uber die Determination des Keimes bei Echinodermen
.
Acta Zool
.
9
,
1
191
.
Hôrstadius
,
S.
(
1973
).
Experimental Embryology of Echinoderms
.
Oxford
:
Clarendon Press
.
Lacalli
,
T. C.
(
1982
).
The nervous system and ciliary band of Miiller’s larva
.
Proc. R. Soc. Lond. B
217
,
37
58
.
Lacalli
,
T. C.
&
West
,
J. E.
(
1985
).
The nervous system of a pilidium larva: evidence from electron microscope reconstructions
.
Can. J. Zool
.
63
,
1909
1916
.
Lacalli
,
T. C.
&
West
,
J. E.
(
1986
).
Ciliary band formation in the doliolaria larva of Florometra. I. The development of normal epithelial pattern
.
J. Embryol. exp. Morph
.
96
,
303
323
.
Okazaki
,
K.
(
1975
).
Spicule formation by isolated micromeres of the sea urchin embryo
.
Amer. Zool
.
15
,
567
581
.
Runnstrôm
,
J.
(
1925
).
Zur experimentellen Analyse der Entwicklung von Antedon
.
Arch, mikrosk. Anat. EntwMech
.
105
,
63
113
.