Embryonic, Larval, and Juvenile Development of the Sea Biscuit Clypeaster subdepressus (Echinodermata: Clypeasteroida)

Sea biscuits and sand dollars diverged from other irregular echinoids approximately 55 million years ago and rapidly dispersed to oceans worldwide. A series of morphological changes were associated with the occupation of sand beds such as flattening of the body, shortening of primary spines, multiplication of podia, and retention of the lantern of Aristotle into adulthood. To investigate the developmental basis of such morphological changes we documented the ontogeny of Clypeaster subdepressus. We obtained gametes from adult specimens by KCl injection and raised the embryos at 26C. Ciliated blastulae hatched 7.5 h after sperm entry. During gastrulation the archenteron elongated continuously while ectodermal red-pigmented cells migrated synchronously to the apical plate. Pluteus larvae began to feed in 3 d and were 20 d old at metamorphosis; starved larvae died 17 d after fertilization. Postlarval juveniles had neither mouth nor anus nor plates on the aboral side, except for the remnants of larval spicules, but their bilateral symmetry became evident after the resorption of larval tissues. Ossicles of the lantern were present and organized in 5 groups. Each group had 1 tooth, 2 demipyramids, and 2 epiphyses with a rotula in between. Early appendages consisted of 15 spines, 15 podia (2 types), and 5 sphaeridia. Podial types were distributed in accordance to Lovén's rule and the first podium of each ambulacrum was not encircled by the skeleton. Seven days after metamorphosis juveniles began to feed by rasping sand grains with the lantern. Juveniles survived in laboratory cultures for 9 months and died with wide, a single open sphaeridium per ambulacrum, aboral anus, and no differentiated food grooves or petaloids. Tracking the morphogenesis of early juveniles is a necessary step to elucidate the developmental mechanisms of echinoid growth and important groundwork to clarify homologies between irregular urchins.


Introduction
Sea biscuits and sand dollars are representatives of the order Clypeasteroida, the latest major branch of irregular echinoids to evolve. Clypeasteroids diverged from a cassiduloid-like ancestor around the Paleocene [1][2][3] and by Middle Eocene (49-37 million years ago) the suborders of this group were already established [2] and present in oceans worldwide [1]. Morphological changes during the early evolution of clypeasteroids were tightly associated with the occupation of shifting sand beds and involved the retention of juvenile traits into adulthood [2]. Strong evidence of this phenomenon is, for example, the presence of a lantern of Aristotle in adult clypeasteroids, while in the remaining extant and fossil irregular echinoids, including in the ancestral lineage of Clypeasteroida (a branch of Cassiduloida -see [3] for an analysis of cassiduloid-clypeasteroid relationships), the lantern atrophies before adult life [2]. Besides being the only irregular echinoids with the lantern of Aristotle and perignathic girdles in adult individuals, clypeasteroids exhibit more than one podium per ambulacral plate. This feature accounts for the drastic increase on the number of nonrespiratory tube feet and allows greater efficiency in the collection of food [2]. In addition, a recent study established the relatively large anus, shortening of primary spines, and flattening of the body as synapomorphies of Clypeasteroida [4].
The sea biscuit Clypeaster subdepressus (Gray, 1825) belongs to the suborder Clypeasterina, sister group of Scutellina, and to the Clypeasteridae, a single-genus family of Clypeasteroida and sister group of the Arachnoididae. C. subdepressus is found in the Caribbean and East Coast of America, from North Carolina to Rio de Janeiro [5] and São Paulo [6,7]. Adult individuals live semi-buried in coarse biogenic sand of coastal waters and feed on organic matter present in the sediment [5,8]. Sand particles are collected by accessory podia of the oral surface, taken to the mouth region, and then crushed by the lantern teeth [9].
The larval development and early juvenile growth of C. subdepressus was first described in Caribbean specimens [10]. The study compared the development of C. subdepressus and C. rosaceus showing that the former has an obligatory planktotrophic larva in contrast to the facultative feeding larva of the latter. Since this study focused on the characterization of C. rosaceus development, the timing of early development and a detailed ontogeny of morphological traits of C. subdepressus were not provided.
Given the unique morphological evolution, a complete fossil record [11], and readily feasible developmental studies [12], clypeasteroids can be a valuable group for evolutionary analyses. Furthermore, C. subdepressus has been used in recent studies related to development [13][14][15][16], posing the need for a detailed developmental table in order to comprehend the evolution of echinoid morphology. The present work provides a complete description of C. subdepressus development including stage timing, cleavage and gastrulation dynamics, and post-metamorphic development of juveniles.

Results
The morphological changes occurring during the embryonic, larval, and juvenile development of C. subdepressus are depicted in Table 1.

Spawning
Adult males and females of C. subdepressus had 5 gonopores at the apex. A long (2-3 mm) genital papilla (Fig. 1A,B) and up to 3 shorter (v1 mm) accessory papillae were present at the distal end of each gonoduct. During spawning induced by KCl injection one or two gonopores released gametes continuously for several minutes (maximum of 30 min with specimen underwater); the flux of gametes through a genital papilla varied between 7 and 120 eggs s (Video S1). Accessory papillae of males released small quantities of sperm under low intensity flux ( Fig. 1A; Video S2). A spontaneous spawning event was observed in the laboratory tanks on March 8, 2007 at 17 h 30 min. Five individuals moved the sand away from the apex and initiated spawning through the five gonopores simultaneously (Fig. 1C). Spawning ceased after 10 min and the apex was again covered with sand grains by action of surrounding spines. After a few minutes it was uncovered and gamete release re-initiated; this sequence was repeated at least 3 times for each specimen.

Fertilization
After sperm entry, the male pro-nucleus was pushed by microtubules towards the center of the egg ( Fig. 2; Video S3). It moved at a constant speed of 0:1 mm s . When touched by microtubules, the female pro-nucleus was rapidly pulled towards the male pro-nucleus. The fusion took place approximately 12 min 30 s after sperm entry. The cytoplasmic movements increased and the cell surface acquired an irregular aspect. Video footage accelerated 625| revealed that the deformations were dynamic and that the membrane was vibrating (Video S3). Immediately before the first cleavage the membrane ceased the vibration, the cell surface became regular, and the hialine layer thickened.

Cleavages
Cell divisions initiated 80 min after fertilization (Table 2) and were holoblastic. First and second cleavages were meridional, dividing the embryo in 4 equal blastomeres ( Fig. 3A-C). The third division was equatorial, separating animal and vegetal blastomeres at 120 min; the latter were 30% smaller in diameter (Fig. 3D). Micromeres originated unequally from vegetal blastomeres while 8 mesomeres were formed by a meridional cleavage of animal blastomeres ( Fig. 3E-H). Equatorial division of mesomeres, meridional division of macromeres, and unequal micromere division formed embryos with 32 cells 180 min after fertilization (Fig. 3I,J). Sixty-cell embryos were formed when blastomeres went through an equatorial division while the 4 micromeres experienced a meridional division. The seventh cleavage occurred without micromere division resulting in embryos with 108 cells (Fig. 3L).

Blastulae
Cells acquired a polygonal shape during the consolidation of the epithelium between 3.5 and 6.5 h post fertilization (hpf) (Fig. 4). The vegetal plate thickened and cilia were formed 7.5 hpf, immediately before hatching.

Gastrulae
Primary mesenchyme cells (PMC) detached from the vegetal pole, became spherical, and aggregated in an unipolar manner on the vegetal pole; ingression to the blastocoel took place approximately 10 hpf ( Fig. 5A-C). PMC migrated through the blastocoel forming a ring connected by thin filopodial pseudopodia on the posterior end (Fig. 6A).
Red-pigmented cells differentiated on the vegetal pole 12 hpf and migrated through the epithelium, simultaneously with PMC, towards the apical plate 13 hpf (Fig. 5D). Secondary mesenchyme cells (SMC) originated on the vegetal pole, extending cytoplasm projections towards the blastocoel during archenteron invagination. The archenteron invagination began at 14 hpf (Fig. 5E).  PMC formed two aggregates 15 hpf (Fig. 5F) and initiated the secretion of a calcareous triradiate spicule (Fig. 6B). SMC were composed of different cell types including cells with filopodial and lobopodial pseudopodia, ameboid cells without cytoplasm projections, and cells with red-pigmented granulae. SMC on the archenteron reached the anterior pole 16,5 hpf (Fig. 5G) while red-pigmented epithelial cells reached the anterior pole 19 hpf, when the blastocoel was occupied by SMC (Fig. 5H). Epithelial red-pigmented cells were not present on the ventral (oral) region of the embryo at prism stage. The surface of the embryo at prism stage was covered by cilia with an apical tuft on the anterior pole and a ciliated ring around the anus (Fig. 6C).
The embryo height decreased during gastrulation from 257+9 mm at 12 hpf to approximately 225 mm from 16.5 hpf to 20 hpf; the width had a slight increase of 20 mm (Fig. 7). The archenteron elongated continuously on a linear manner from 13.5 hpf (Fig. 7) and reached 84+10% of the blastocoel height (Fig. 7B).

Pluteus
In 48 hpf two coelomic pouches were present next to the esophagus (Fig. 8A). At this stage the mouth opened, but the larvae were unable to feed; microalgae captured by the larval arms were carried towards the mouth, but were deflected away possibly by an opposing current (Video S4). The gut was not functional, but already had three portions identified as esophagus, stomach, and intestine ( Fig. 8A). Muscles of the esophagus began to contract 70 hpf (Video S5); the stomach grew in diameter while its epithelium became thinner. The gut became functional and the larvae began to feed 3 d post-fertilization (dpf) (Fig. 8B).
The skeleton of postoral and posterodorsal arms was fenestrated while the anterolateral and preoral pairs were non-fenestrated (see Fig. S1 for the arrangement of larval arms). The hydropore formed above the cardiac sphincter approximately 3 dpf by an extension of the left coelomic pouch and opened on the surface 5 dpf (Fig. 8A,B detail). The vestibule appeared between 4 and 5 dpf on the left side of larvae between the postoral and postero-dorsal arms (Fig. 9A). The vestibule fused with the left coelom (Fig. 9B,C) forming the rudiment (Fig. 9D) between 6 and 10 dpf.
The rudiment developed podia and spines, which became active still inside the larval body (Fig. 9D). No pedicellariae were formed on the surface of the larval body, as commonly observed in competent larvae of regular echinoids. Some rudiments were oriented 45 0 , and not perpendicular, to the antero posterior axis of the larva.
Starved. Compared to fed larvae, starved ones formed the same structures at similar developing times during the first week post-fertilization. Differences were noticed only during vestibule invagination. The vestibule of starved larvae began to invaginate, but did not reach the left coelom; larvae died 17 dpf. Metamorphosis Competent larvae exhibited a typical substrate-test behaviour which consisted of swimming near the bottom and exposing the vestibule pore with protruding podia (type I, see below) by moving the left arms ( Fig. 10A; Video S6). Larvae were able to fully evert the rudiment by opening the arms 180 0 posteriorly, attach to the substrate, return to the larval conformation and resume swimming. Metamorphosis occurred when larvae attached firmly to the bottom with the protruding podia and the larval tissues began to regress and accumulate on the aboral surface of the rudiment (Fig. 10B). During this process larval spicules became exposed and broke off; fragments of tissue were lost (Fig. 11A). Larval tissues accumulated on the aboral surface of the rudiment forming a globoid structure (Fig.  11B). Metamorphosis took approximately 1 h 30 min from attachment to the complete regression of the larval tissues (Video S7).

Juvenile
Metamorphosis is followed by the resorption of larval tissues and the development of juvenile structures. The body of metamorphosing and early postlarval juveniles did not exhibit clear evidence of bilateral symmetry when seen from the aboral surface (Fig. 11B). However, a close look at the oral surface under polarized light revealed hints of bilateral symmetry identifiable by the shape of the body and position of the rudiments of the lantern of Aristotle (Fig. 12A). The bilateral symmetry passing through the III-5 plane (Lovén's axis) became evident after the resorption of larval tissues. The bilaterality of postlarval juveniles was identified by the ovoid body shape, the longer interambulacrum-5 pair of posterior spines, the anus opening posteriorly, and the ambulacrum-III always pointing in the moving direction of the juvenile.
Appendages of 1-day-old postlarval juveniles were disposed in two rows along the ambitus, an infracoronal (oral) and a supracoronal (aboral) row. The infracoronal row had one spine for each interambulacral zone and one sphaeridium (a modified spine of uncertain function) between a pair of accessory podia for each ambulacral zone, totalling 5 sphaeridia, 5 spines, and 10 podia on the oral surface ( Fig. 13). There were two types of suckered podia in postlarval juveniles; no buccal podia were observed. Type I had a narrow tip without an internal spicule (Fig. 11F, left) while type II had an expanded tip and a ring-shaped spicule inside (Fig. 11F, right). The former originated in the outer margin of the peristome (inner margin of the skeleton of the oral surface), but not embraced by the skeleton at ambulacral positions Ib, IIb, IIIa, IVb, and Va (bbaba; see Fig. 13 and 14A); podia was positioned in a concavity of the skeleton (Fig. 14B,C). The latter, podial type II, emerged from a skeleton pore at the complementary ambulacral columns Ia, IIa, IIIb, IVa, and Vb (aabab). Podial type I occupied an inner (closer to the mouth) ambulacral position than podial type II on the infracoronal row of juveniles (Fig. 14B,C). At this stage, sphaeridia were equal in size bearing a birefringent oval tip 20 mm in diameter (Fig. 11E), but they were smaller during metamorphosis (width between 3 and 17 mm; see Fig. 11A). The supracoronal row had a pair of spines for each interambulacral area and a type I podium for each ambulacral zone. The postlarval juvenile had 15 spines, 15 podia, and 5 sphaeridia in total; miliary spines were occasionally present on ambulacral zones and no pedicellariae were observed at this stage. Primary spines and podia of C. subdepressus were permanent and did not regress during the observed period.
Early postlarval juveniles had no skeleton on the aboral surface, except for the remnants of larval rods (Fig. 11D). The gut was not yet formed and neither mouth nor anus were present. During the resorption of larval tissues the rudiments of the lantern of Aristotle were visible in the oral region under polarized light. Teeth were positioned on the interambulacral zones between a pair of demipyramids and a pair of epiphyses (Fig. 12A). Each set of 5 independent ossicles was intercalated by a single ossicle, a rudiment rotula (Fig. 13). Except for the teeth, the skeleton had the trabecular structure typical of stereom.
Two days after metamorphosis the demipyramids were larger and still separated by the tooth slide (Fig. 12B). Tips of the teeth were narrow without ornamentation (Fig. 12B) and became robust with a complex surface 4 dpm. The digestive tract became visible through the aboral surface two days after metamorphosis; the gut tube exits the lantern of Aristotle onto the right side of the cavity and towards the anterior region; it turned counterclockwise towards the left side of the cavity, into the posterior region, and then again reaching the right side. The tube then bent 180 0 left, pointing posteriorly, and opened on the posterior portion of the aboral surface below an anal plate (Fig. 15E).
The aboral surface was occupied by calcareous plates 2 dpm (Fig. 15A). Eight miliary spines (with a crown-like tip) grew on the supracoronal row and 4 appeared on the aboral surface between 2 and 4 dpm (Fig. 15B,C); older juveniles also had miliary spines between other appendages, including ambulacral zones. Each pair of demipyramids were sutured together into one pyramid 7 dpm (Fig. 12C). No other skeletal element was formed on the oral surface of the juvenile during this period.
The mouth opened on the centre of the oral surface 7 dpm and juveniles began to feed (Fig. 15F,G). At this stage the peristomial membrane and the lantern of Aristotle were actively moving while the esophagus exhibited peristaltic contractions (Video S8). Exotrophic juveniles frequently had food in their gut and eventually in the rectum; fecal pellets were observed (Fig. 15E). Type II podia significantly increased in number and seemed to play a major role in locomotion. Juveniles moved quickly among sand grains, extending the anterior type II podia forward more than a body length ahead ( Fig. 15H; see Video S9 for locomotory examples).
One-month-old juveniles exhibited a digestive tract with three distinct parts besides the esophagus. The stomach and intestine were wider, opaque, and brownish while the rectum was covered by red-pigmented cells (Fig. 15E). Four ophicephalous pedicellariae appeared on the ambitus of the posterior region 14 dpm (Fig. 15D). Tridentate pedicellariae appeared at the posterior region 30 dpm; they were approximately one quarter of the size of the posterior spines with a stem two thirds of the length. This kind of pedicellaria was also seen on the middle region of the juvenile, but not on the anterior portion.
Juveniles did not eat free living benthic organisms, but apparently rasped organic material off larger sand grains. Smaller particles were manipulated by podia and held in position by the peristomial membrane while the lantern rubbed the surface of the grain (Video S10). They survived in laboratory cultures for 8 months and 20 days after fertilization. Post-metamorphic juveniles were initially 200-270 mm wide and 300-330 mm in length and reached *480 mm in diameter and *570 mm in length. Juveniles were active and showed no sign of food privation, except for the last 3 weeks when they became less active and more pale. A single sphaeridium per ambulacrum was still visible (not internalized) by the time of juvenile death.

Discussion
Direct observations of the behavior of C. subdepressus during spawning revealed that specimens covered by a layer of sand (not completely buried) repeatedly exposed the apex, released gametes, and covered the apex by moving the grains adapically. Apex exposure before spawning was observed previously in burrowed individuals of Arachnoides placenta [17], but intermittent  gamete release during spontaneous spawning was previously undescribed for clypeasteroids. Until there are field observations to better comprehend the external fertilization dynamics of clypeasteroids, their reproductive behavior will remain largely unknown.
Although it was suggested that long papillae, swollen during spawning, could disperse gametes across a thin layer of sand [17], our observations showed that C. subdepressus avoid contact of gametes with sand grains by clearing the apex during spawning. Thus, the genital papillae seem to play a role in allowing gametes to be released well above the spines, consequently decreasing the number of eggs and sperm trapped within the skin mucus [17] and avoiding mechanical stress caused by the movement of spines and pedicellariae.
Data on the gastrulation of C. subdepressus fits the correlation between the type of gastrulation and the pattern of migration of  red-pigmented cells in echinoids [21]; red-pigmented cells originate on the vegetal pole and migrate through the ectoderm to the apical plate while the archenteron elongation is continuous. Also, the behaviour of red-pigmented cells matches the pattern described for C. japonicus where these cells migrate through the ectoderm, but are also seen at the archenteron tip [21]. Redpigmented cells can have a regulatory role and are known to trigger gastrulation in Echinometra mathaei [22]. These cells might participate on the morphological changes during prism formation and early axis specification of plutei, because they were absent from the ventral ectoderm of C. subdepressus; a region that remained flat until the formation of the larval mouth.
Starved larvae of C. subdepressus neither develop a complete vestibule invagination nor a rudiment, but the eggs provided sufficient energy for the survival of a functional pluteus larva for 17 dpf. Based on larval skeleton measurements, the facultative feeding period of larvae was estimated in 5 clypeasteroid species [13]. The authors determined when fed and starved larvae differed in size from fertilization, indicative of the exhaustion of maternal provisioning. In C. subdepressus larvae diverged between 60 and 72 hpf. This range matches with the necessary timespan for the differentiation of the larval gut, ciliated bands and beginning of larval feeding, described in the present study. Thus, we directly observed that the pre-feeding stage of C. subdepressus larvae ends 3 dpf. The appearance of size differences between larvae might not solely indicate when starved larvae stopped growing, but also when fed larvae increased their growth rate because of food ingestion. Since starved larvae reached the 8-arm pluteus stage without qualitative morphological differences, the exhaustion of maternal resources in C. subdepressus apparently does not occur as early as suggested; even though the absence of food immediately after the pre-feeding stage could directly affect early growth rates.
Competent larvae of C. subdepressus exhibited a substrate-test behaviour similar to other echinoid species [23][24][25][26]. Although early postlarval juveniles resemble regular urchins with a spherical body, bilateral symmetry could be identified soon after the resorption of larval tissues and was probably determined during rudiment formation. Several morphological features mark the early bilateral symmetry of C. subdepressus such as elongated body, longer posterior spines, and anus positioned posteriorly. In other irregular urchins bilateral symmetry is recognized by different characters. In the spatangoid Echinocardium cordatum, the subanal fasciole is visible on the rudiment still inside the larvae [26], the peristome is positioned anteriorly and the posterior region is more developed [27]. The bilateral symmetry of the sand dollar Echinarachnius parma is identified by the size and number of plates in the ambulacra and by differences in size of the sphaeridia [28].
The bbaba distribution at ambulacral columns and the inner position of infracoronal type I podia suggests strict adherence to Lovén's rule of echinoid peristome development [29]. These patterns indicate that C. subdepressus podial type I preceded type II during ontogeny. Since we did not track the formation of skeletal plates, the coordinance between the appearance of podia and growth of skeletal plates has not been elucidated. We observed that the first podium (type I) was not encircled by the skeleton and occupied the edge of the peristome, while the following podium on the ambulacrum (type II) protruded from a pore in the calcareous mesh. At this early stage there were no signs of disruption in the alternating pattern of podial organization, which in other echinoids is a by-product of the alternating deposition of plates known as the ''ocular plate rule'' [30]. Therefore additional data on juvenile skeletal growth remains crucial to trace the plate formation and ontogenetic fate of these podia. In addition, precise tracking of ambulacral growth can identify the developmental basis behind a wellknown innovation of Clypeasteroida, the breakage of the onepodium-per-plate rule and further increase in the number of nonrespiratory podia [2].
Following metamorphosis C. subdepressus juveniles had 3 spines per interambulacrum (15 total), 3 podia and one spheridium per ambulacrum (15 podia and 5 sphaeridia total), and no pedicellariae. In contrast, competent larvae of the regular urchins Paracentrotus lividus and Strongylocentrotus franciscanus already have pedicellariae during the late larval period and after metamorphosis [25,31], while pedicellariae of S. purpuratus appear some time after metamorphosis. Competent larvae of E. cordatum do not exhibit spines or pedicellariae [26]. Differently from C. subdepressus, newly metamorphosed juveniles of these regular sea urchins have 4 primary spines per interambulacrum (20 total) [25,31]. The irregular echinoid E. cordatum has a greater number of primary spines per interambulacrum after metamorphosis and also differs from C. subdepressus by the presence of secondary spines and a subanal fasciole with clavulae and 4 primary spines [26]. By contrast, C. subdepressus uniquely displays 3 podia per ambulacrum after metamorphosis while S. franciscanus and S. purpuratus [31], P. lividus [25] and E. cordatum [26] have 1 podium per ambulacrum. The two podial types identified in postlarval juveniles of C. subdepressus did not clearly fit into any specific category previously described for adult clypeasteroids [32]. Further podial specialization should occur associated with body growth and differentiation of food grooves. The appendages of C. subdepressus did not regress after metamorphosis as observed for E. cordatum, but the postlarval period of C. subdepressus (7 d to exotrophy) was considerably longer than observed for E. cordatum (3.5 d) [26]; regular echinoids have a longer postlarval period (8-10 d) [25,31]. Differences in nutrient storage and latent effects of larval life [33] might account for variations in the postlarval period. Finally, the mouth of E. cordatum opens before the anus while the mouth of C. subdepressus opened after the anus.
The initial formation of the lantern of Aristotle of C. rosaceus was identified between 24 and 48 h; although it became functional 7 d later, its development was completed in approximately 12 d, when the mouth opened [10]. In contrast, the rudiments of the lantern of C. subdepressus were already present during metamorphosis, and feeding began soon after (7 dpf). At this early stage, size obviously poses limitations to the feeding mechanisms of juveniles. While adult C. subdepressus systematically collects and grinds sand grains [9], juveniles actively harvested individual grains using the lantern of Aristotle as a rasping tool. The strength to crush sand grains and the ability to collect many particles simultaneously depend on a larger lantern of Aristotle and a greater number of podia available, respectively; both are expected to be acquired with body growth. In other words, size can directly shape the feeding habits of developing sea biscuits.
Adult specimens of Cassiduloida and Oligopygoida, ancestral lineages of the Clypeasteroida, have multiple open sphaeridia per ambulacrum [2]. The Clypeasterina has 2 closed sphaeridia per ambulacrum while the remaining clypeasteroids (Lagani-formes and Scutelliformes) have only 1 closed sphaeridium. Although a complete ontogeny of sphaeridia is known only for isolated species, it is known that all irregular echinoids have a single open sphaeridium per ambulacrum after metamorphosis [2]. The spatangoid E. cordatum bears 5 sphaeridia on the ambulacral zones after metamorphosis [26,34] like E. parma [28] and C. subdepressus (present work). A subtle difference between spatangoids and clypeasteroids is that sphaeridia of the latter are not equal in size during the resorption of larval tissues after metamorphosis. This might indicate that the rudiments of spatangoids are more developed at metamorphosis. Regular urchins exhibit calcified sphaeridia only 9 d after metamorphosis in both S. franciscanus and S. purpuratus [31] and 6 d after metamorphosis in P. lividus [25], even though a sphaeridial bud grows after 4 d in the former. The sphaeridia remained open for nearly 9 months showing that enclosure does not occur shortly after metamorphosis in C. subdepressus as previously suggested for Clypeasteroida [2]. We have no further evidence to suggest that enclosure of sphaeridia occurs before or after the appearance of the second sphaeridium, which is present in adult specimens of Clypeasterina.
We found only 3 small-sized specimens and no individuals less than 8 cm in diameter at sampling sites (personal observation). The occasional nature of recruitment events of C. subdepressus might affect their size distribution in the field. Also, the cryptic habit, high mortality [35] and the minute size of newly metamorphosed juveniles greatly decreases the probability of finding them in their natural habitat.
Exotrophic juveniles of C. subdepressus raised in laboratory cultures for nearly 9 months did not grow more than 500 mm across. Accordingly, post-metamorphic juveniles of C. subdepressus and C. rosaceus from the Caribbean increased in diameter for 8 d and then ceased growth at approximately 420 mm [10]. Although C. subdepressus juveniles from São Sebastiã o survived considerably longer than the 30 d observed for the Caribbean specimens, the conditions of death were similar; juveniles progressively became more transparent with slower movements. Since juveniles frequently had food in the gut and released fecal pellets, they could have suffered a progressive nutritional deficiency because of the limited conditions of laboratory cultures. The absence of essential nutrients and/or appropriate diet might have inhibited their further development. To explain the minimal growth, one could suggest that Clypeaster species have inherently slow growth rates and a long life span. However, this hypothesis still needs to be properly tested once we learn more about the feeding habits and ideal habitat conditions for the growth and survival of juveniles. Until then, a series of morphological changes occurring between the juvenile and adult life, such as the anus migration to the oral surface, the appearance of the second sphaeridium, the enclosure timing of sphaeridia within the test, and associated shape changes, will remain undocumented in these large clypeasteroids.

Cultures
Embryos and larvae. We induced spawning by puncturing the medial portion of one food groove (ambulacrum III) and injecting 2-3 mL of 0,53 M KCl solution into the body cavity. Normally, a second injection of 2 mL was required to trigger gamete release. We poured the eggs through a 300 mm nylon mesh and suspended them in a 500 mL beaker with filtered sea-water (FSW). Eggs were fertilized with 1 mL solution of diluted sperm and moved to a 4.5 L bowl with FSW. After fertilization we kept cultures at 26 0 C and natural light. Approximately 10 h post fertilization (hpf) we transferred swimming larvae to vessels of 100 mL with FSW and no aeration. We kept a maximum of 1 larva mL and inspected the cultures daily. During these observations we removed dead larvae and changed 80% of the sea water with a plastic pipette and a 60 mm nylon mesh, or transferred healthy larvae to a new vessel. When larvae began to feed we added Dunaliella tertiolecta and Rhodomonas sp. to the cultures daily at a final concentration between 8|10 5 and 1|10 6 cells mL . Ten cultures were deprived of food, but received a normal maintenance routine (i.e. water changes).
Metamorphosis. We transferred advanced pluteus larvae to petri dishes with FSW and added a few sand grains from the sampled sites. Successfully metamorphosed larvae were kept in the same container overnight and then moved to a different culture, shown below.
Juveniles. We kept juveniles in the same 100 mL vessels used for larvae, but with a monolayer of sand on the bottom. We changed the water weekly and sealed the cultures with PVC film to minimize water evaporation and subsequent rise in salinity. During water changes we added 3 mL of a plankton sample or biofilm of aquaria, in an attempt to keep food (organic content) available for the juveniles, once we did not change the substrate of these cultures.

Kinetics of cleavages
We homogenized the suspension of developing embryos and collected 2 samples of 3 mL in 10% sea-water formalin every 20 min during a period of 180 min. Posteriorly, we counted the total number of embryos (sum of samples) and estimated the relative frequency of each stage, according to the number of cells (1, 2, 4, 8, 16 and 32 cells), for each timespan.

Archenteron elongation during gastrulation
To verify if the gastrulation of C. subdepressus was continuous or truncated we fixed samples of developing embryos every 30 min between 12 and 20 hpf (except 12.5 hpf) in 1% paraformaldehyde. For each period we transferred the embryos to glass slides, photographed, and measured their width, height, blastocoel height and archenteron length (n~99) with the image processing software ImageJ [36]. The final degree of invagination was calculated with the ratio between the archenteron length and blastocoel height.

Light microscopy
We documented the development under differential interference contrast (DIC) with a Nikon Coolpix 4500 for photomicrographs and Sony DCR HC1000 digital video camera for videos. Cameras were attached to a Zeiss Axioplan2 compound scope and a Zeiss Stemi SV11 APO stereomicroscope. We visualized the calcareous elements of larvae and juveniles under polarized light.
Pluteus larva reconstruction. After manually capturing 118 sequential focal planes of a pluteus larva under DIC we imported the images into ImageJ [36]. The stack was converted to 8-bit, aligned, inverted, and the voxel depth set to 2 mm. We corrected the contrast/brightness so that only areas in focus of each image remained visible. Finally, we created a maximum intensity z-projection (Fig. S1A) and a red-cyan anaglyph of the larva (Fig. S1B).

Scanning electron microscopy
We fixed samples in 2% glutaraldehyde, post-fixed in 1% osmium tetroxide, and dehydrated in ethanol series until 100% ethanol. Standard procedures of critical point and metal coating were done at Instituto de Biociências da Universidade de São Paulo.