FAU Institutional Repository

http://purl.fcla.edu/fau/fauir

This paper was submitted by the faculty of FAU’s Harbor Branch Oceanographic Institute.

Notice: ©2000 Marine Biological Laboratory. The final published version of this manuscript is available at http://www.biolbull.org/. This article may be cited as: Young, C. M., & George, S. B. (2000). Larval development of the tropical deep‐sea echinoid Aspidodiadema jacobyi: phylogenetic implications. The Biological Bulletin, 198(3), 387‐395. Reference: Biol. Bull. 198: 387-395. (June 2000)

Larval Development of the Tropical Deep-Sea Echinoid Aspidodiademajacobyi: Phylogenetic Implications

CRAIG M. YOUNG* AND SOPHIE B. GEORGEt Division of Marine Science, Harbor Branch Oceanographic Institution, 5600 U.S. Hwy. 1 N., Ft. Pierce, Florida 34946

Abstract. The complete larval development of an echi- Introduction noid in the family Aspidodiadematidaeis described for the first time from in vitro cultures of Aspidodiademajacobyi, Larval developmental mode has been inferredfrom egg a bathyal species from the Bahamian Slope. Over a period size for a large numberof echinodermspecies from the deep of 5 months, embryos grew from small (98-,um) eggs to sea, but only a few of these have been culturedinto the early very large (3071-pum)and complex planktotrophicechino- larval stages (Prouho, 1888; Mortensen, 1921; Young and pluteus larvae. The fully developed larva has five pairs of Cameron, 1989; Young et al., 1989), and no complete red-pigmented arms (preoral, anterolateral,postoral, pos- ontogenetic sequence of larval development has been pub- lished for invertebrate.One of the terodorsal,and posterolateral);fenestrated triangular plates any deep-sea species whose have been described et at the bases of fenestratedpostoral and posterodorsalarms; early stages (Young al., 1989) is a small-bodied sea urchin with a complex dorsal arch; posterodorsalvibratile lobes; a ring Aspidodiademajacobyi, flexible that lives at in the of cilia around the region of the preoral and anterolateral long spines bathyal depths eastern Atlantic 1). The of arms; and a long, unpaired posterior process containing a tropical (Fig. early development A. jacobyi is unusual in that its small (94-100 ,um) fenestrated rod. The presence of a posterior process and eggs contain sufficient yolk reserves to support a posterodorsalarms makes the larva of Aspidodiademaja- prefeeding developmentalperiod that extends well into the larval stage cobyi much more similarto larvaeof irregularurchins in the (Young et al., 1989), a feature that might permit larvae to order Spatangoideathan to other families of the order Dia- move long distances,either verticallyor horizontally,before dematoida, to which the family is normally assigned. This feeding becomes necessary. In this paper, we describe the unexpected larval form lends supportto a recommendation complete larval development of A. jacobyi, document un- that the Aspidodiadematidaeshould be either elevated to expected morphological traits in the late-stage larvae, and ordinalstatus as a sister group of the orderDiadematoida, or discuss how these unexpected traits could help justify a split off as a sister of the other families within the group realignment of the currently accepted ordinal-level taxon- order. In either if we the case, accept parsimonioushypoth- omy of the Euechinoidea. esis that the aboral and arms were process posterodorsal The echinoid family Aspidodiadematidaecontains only derived once in the of euechi- only evolutionary history two genera, Aspidodiademaand Plesiodiadema, all species noids, then the larval data that the suggest Aspidodiadema- of which are found at bathyal and abyssal depths (Hyman, tidae be near the node where may very the irregularand 1955), many on island slopes in tropical seas. On the basis regular euechinoids first diverged. of adult characters, notably aulodont teeth and hollow spines, Mortensen (1927) placed the family in the order Diadematoida,suborder Aulodonta, in the company of the Received 8 December 1999; accepted 1 February2000. * the and the Author for correspondence.E-mail: [email protected] Echinothuriidae, Pedinidae, Diadematidae.Re- t Present address: Biology Department,Georgia Southern University, cent authors (e.g., Smith, 1984) have elevated the echi- Statesboro, GA 30460. nothuriidsand pedinids to ordinal level, placing three fam- 387 388 C. M. YOUNG AND S. B. GEORGE

the present paper, we confirm Emlet's prediction concern- ing the presence of posterodorsalarms in the Aspidodiade- matidaeand documentother larval characterssupporting the idea that this family of urchinsmay have originatednear the node where the regular and irregularurchins diverged.

Materials and Methods

Adult Aspidodiademajacobyi were collected by suction using Johnson-Sea-Link submersibles between depths of 500 and 750 m at various sites in the Northern Bahamas Figure 1. Adult Aspidodiademajacobyi from the Bahamian Slope, (Young, 1992). They were transportedto the surface in flexible L. F. showing long spines. (Photo by Braithwaite). closed acrylic containers of seawater and transferredto 14?C incubatorswithin minutes of arriving at the surface. ilies in the order Diadematoida: the Aspidodiadematidae, Although we collected this species duringthe springmonths the Diadematidae,and the Lissodiadematidae.Larvae of the of every year from 1985 through 1999, we were able to Lissodiadematidaeremain unknown (Emlet, 1988a; Pearse induce spawning only twice, once in 1987 (Young et al,, and Cameron, 1991). Emlet (1988a) has arguedon the basis 1989) and once on 11 April 1994. Spawning was inducedby of larval form that the Micropygidae,which have also been intracoelomicinjection of 0.55 M KC1.In the 1994 spawn- raised to ordinal status by Jensen (1981), should also be ing, we injected 28 individuals; of these, 13 males and 2 includedas a family in the Diadematoida.Although it seems females released gametes. Cultures were established by clear that the aspidodiadematidsare closely related to the mixing the eggs of a single female with active sperm from diadematoids,the relationships among orders and families 6 males in a small bowl of seawater. The other female was near the base of the euechinoid tree remain obscure. This not used because she released very few oocytes, many of portion of the tree remains inadequatelyresolved because a which appearedirregular. Zygotes were split among five 1-1 thorough cladistic analysis of extant species has not been jars of 0.45-,am filteredseawater, at an initial density of 800 undertaken(R. Mooi, CaliforniaAcademy of Sciences, San embryos per liter. After larvae had developed complete Francisco, pers. comm.), and also because fossil material guts, they were fed with a mixture of the algae Isochrysis from the relevant time periods contains too few meaningful galbana, Thalassiosira weissflogii, and Dunaliella terti- charactersfor construction of a robust tree (A. B. Smith, olecta. At the outset, we used an algal concentration of The NaturalHistory Museum, London, pers. comm.). Fossil 1000-2000 cells/ml, but this was increased graduallyas the aspidodiadematidsclosely resembling moder urchins in larvae grew, up to a maximum of 5000 cells/ml, 49 days the genus Plesiodiadema have been described from the after fertilizationand thereafter.These algal concentrations upper Jurassic clays of England and France (Smith, 1995), were chosen because similar concentrationspromote opti- indicating that the family differentiatedmore than 200 mil- mal larval growth in some shallow-waterechinoderm larvae lion years ago. (Fenaux et al., 1985; George, 1990; George et al., 1990). By analyzingthe crystallineaxes of echinoid apical plates Larvae were reared in dark incubatorsexcept during water (Raup, 1965), Emlet (1985, 1988a) was able to infer certain changes, when they were exposed briefly to the fluorescent details of larval form for a number of rare species with overhead lights. To simulate temperature changes that unknown larvae. On the basis of this indirect evidence, he bathyal larvae would encounter if they undertook ontoge- predicted that all members of the genera Aspidodiadema netic vertical migration(Young et al., 1996a), we increased and Plesiodiadema should have larvae with both preoraland the culture temperaturefrom 14?C to 16?C 14 days after posterodorsal arms (Emlet, 1988a). This prediction is in fertilization, reduced it to 14?C 66 days after fertilization, stark contrastto the 2-arm larvae with widely spread arms then finally to 13?C (the temperaturenear the bottom of the (the "Echinopluteus transversus" of Mortensen, 1921) adult depth range) 77 days after fertilization. Initially, sea- found in all members of the closely related families Diade- water was changed and the larvae were fed every 2 days. matidae and Micropygidae. By assuming that the 2-arm This interval was increased to 4 days beginning 28 days larval form with widely spread arms is synapomorphic, after fertilization. Emlet (1988a) argues parsimoniously that the Aspidodia- Each time water was changed, we examined the larvaefor demidae must be a sister group to the Diadematoida,having the appearance of any new arms or other structures and branchedoff the major clade of regulareuechinoids earlier measured the lengths of the postoral, somatic, posterodor- than the diadematids or micropygids. Indeed, he recom- sal, and posterolateralrods. Because there were 2-day or mends that the Aspidodiadematidae should be removed 4-day intervalsbetween observations,the times for reported from the Diadematoidsand elevated to the rank of order.In morphologicalchanges (Table 1) should be regardedas only DEEP-SEA ECHINOID LARVAE 389

Table 1

Developmental timetablefor Aspidodiademajacobyi, showing the presence (x) and absence (-) of larval arms and other importantstructures on days when cultures were examined

Larval arms Other structures

Days after Posterior Ciliated Posterodorsal Echinus fertilization Postoral Anterolateral Posterodorsal Preoral Posterolateral Mouth process ring lobes rudiment Podia

6 x - 9 x x - 11 x x - x 15 x x - - x 25 x x x - x 28 x x x -x -- - 32 x x x x x 34 x x x x -x x 35 x x x x x x 39 x x x x x x 42 x x x x x x x 46 x x x x x x x 49 x x x x x x 53 - x x x x x x 56 x x x x x x x 62 x x x x x x x 75 x x x x x x x x 116 -x x x x x x x x x 120 - -x x x x x x x x x

Larvae were reared at 14-16?C (see text for details). Because cultures were not examined every day, the times do not accurately representthe first occurrences of larval stages.

approximatemarkers for the initiationof larval stages. Dur- ing the early larval stages, total larval length was defined as the linear measurementfrom the tip of the postoral rod to 3500 the point where the two somatic rods meet mid-sagittallyon the end of the larva. In the later total larval posterior stages, 3000 length was defined as the linear measurementfrom the tip of the postoral rod to the posterior end of the unpairedaboral process. During the early stages, we generally made all 2500 measurements on at least 10 larvae. After the 46th day, E when relatively few larvae remained in the cultures, we I 2000 were concerned about stressing the remaininglarvae, so we -c measured only the most advanced larva at each sampling 0) C 1500 period. Terminology used for larval arms and spicules is i from Mortensen (1921). 1000 Results Larval growth and form 500 The early developmental stages of Aspidodiademaja- cobyi have been described by Young et al. (1989), so the 0 present study characterizes only the echinopluteus larval 0 10 20 30 40 50 60 stages. The developmental timetable is summarizedin Ta- Daysafter Fertilization ble 1 and linear growth is plotted in Figure 2. Mortality, Figure 2. Mean and maximum observed of size, and stage variability increased over time, but body lengths Aspidodia- living dema jacobyi larvae the first 63 after fertilization. Errorbars larvae from all cultures during days generally appearedhealthy, regard- are standarddeviations. Sample size is reporteddirectly below each mean less of stage. Decreasing the temperaturefrom 16?Cto 13?C value. 390 C. M. YOUNG AND S. B. GEORGE after the 77th day appeared to have a positive effect on in some individuals (Fig. 3H, I). Forty-six days after fertil- larval development;larvae transferredto the lower temper- ization, however, as many as 80% of the larvae were still at ature continued to increase in length and width, whereas the 8-arm stage. During the 10-armstage, the posterodorsal larvae that remained at the higher temperatureincreased in and posterolateralarms continued to lengthen and widen, length only. Throughout development, larvae swam with while the anterolateraland preoral arms were generally their anterior ends upwards, remaining near the surface resorbed (Figs. 3H, I, 4A, D). during the early larval stages (2- to 6-arm stages) and near Forty-nine days after fertilization, two small ciliated the bottom duringlater larval stages (8- and 10-armstages). lobes appearedin the basal region of the posterodorsalarms The 2-arm stage began 6 days after fertilizationwith the (Fig. 4A). By 53 days, these vibratilelobes had become very appearanceof postoral arms (Fig. 3A). Fenestratedarm rods prominent,and it was apparentthat they were not supported were visible by cross-polarizedlight, but opacity of the gut by skeletal elements (Fig. 4B). At 53 days, more than 50% region obscured skeletal details in the posterior end, as of the larvae in all cultureswere at the 10-armstage, having previously reported by Young et al. (1989). The postoral posterolateralarms, posterodorsal vibratile lobes, and a pos- arms were widely spread, and red pigment spots were scat- terior process. No additional arms developed after the 10- tered throughoutthe larval arms. Nine days after fertiliza- arm stage, though larvae continued to increase in size until tion, the anterolateralarms appeared, forming the 4-arm the largest larvahad attaineda maximumlength of 3071 ,um echinopluteus. The gut region became progressively less 56 days after fertilization. opaque duringthis period, and the digestive system could be Duringthe 10-armstage, a ciliated ridge began to develop clearly seen (Fig. 3B). Eleven days after fertilization,larvae near each posterodorsallobe. These two ridges ran in an had grown to more than twice the original length of the anterior-posteriordirection, parallel to the anterolateralrods early embryos (Fig. 2). At this stage, the anterolateralarms on opposite sides of the larvae (Fig. 3H) and eventually were very small, but the mouth opened and the larvae began joined to form a distinctive loop surroundingthe region of to feed. Fifteen days after fertilization, 4-arm larvae with the resorbing preoral and anterolateralarms (Fig. 4C). By well-developed postoral and anterolateralarms (Fig. 3C) the 75th day, the ciliated ridges were thick and heavily were present in all cultures, and by the 25th day, these pigmented along their entire length (Fig. 4C) and were larvae were more than 700 Aumin length (Fig. 2). reflexed at the anteriorend to form a loop that surrounded The posterodorsalarms characteristicof the 6-arm stage the region where the anterolateraland postoral arms had appeared25 days after fertilization (Table 1, Fig. 3D) and been (Fig. 4C). Also at this time, large brownish orange oil were fully developed (Fig. 3E) by the 28th day. Some larvae droplets were observed near the bases of the postoral, pos- remained in the 6-arm stage for up to 15 days, but small terodorsal,and posterolateralarms. buds of preoral arms were apparentin some individuals as Sixty-two days after fertilization,the posteriorprocess of early as the 34th day (Fig. 3E). Red pigment spots initially the largest larva attainedits maximum length of 1224 /um; present throughout the larval arms became increasingly the posterolateral arm was now 459 /um in length. The concentratedat the tips after this stage. A gradualchange in posterodorsallobes continuedto elongate until they attained larval shape became apparentafter the 28th day, and 4 days a maximum size of 237.5 jum, 127 days after fertilization. later an unpaired process supported by a fenestrated rod Larvae began to decrease in overall length after the 75th appeared at the posterior end of the larva (Fig. 3F). This day, but the postoral arms began to widen near the bases, posterior process continued to grow throughoutthe 6-arm until on the 134th day they had attaineda maximum basal stage and subsequentstages until it was as long as the rest diameterof 384 /xm (Fig. 4D). of the larval body. At 16?C, the larvae remained in the A few larvae survived in culture for up to 151 days, 6-arm stage for about 10 days (Table 1). though none metamorphosed. An echinus rudiment ap- By the 39th day, there was considerable variability peared in a single larva before the 77th day. This larva was among cultures in the relative proportionsof individuals in isolated and exposed to deep-sea sedimentsin an unsuccess- different larval stages. Some culture jars contained only ful attempt to induce metamorphosis.Many more individ- 6-arm larvae with posteriorprocesses, whereas up to 60% of uals developed echinus rudiments between the 77th and larvae in other cultures had attainedan 8-arm stage. During 134th days. In most animals, anatomical details of the the 8-arm stage and early 10-arm stage (Fig. 3G), the rudiment were difficult to visualize because the rudiment posterodorsalarms and the posteriorprocess were similar in was opaque and was partially obscured by an equally length, about one-third less than the length of the postoral opaque gut region (Fig. 4D, E). Nevertheless, podia were arms. The base of the postoral arms broadenedconsiderably observed in the rudimentsof some animalsbeginning on the at this stage and measured 122.4 g,m in width. 116th day, and adult spines were observed between 120 and Between the 39th and 42nd day, the buds of the postero- 127 days after fertilization. As in many other species lateral arms formed, thereby initiating the 10-arm stage (Mortensen, 1921), the arms of late-stage larvae became (Fig. 3G). By the 42nd day these arms were fully developed smaller as the rudimentgrew. DEEP-SEA ECHINOID LARVAE 391

Larval skeleton rods (Fig. 4F) and posterolateralrods (Figs. 31, 4D) were The postoral, posterodorsal,and posterior rods were all simple, not fenestrated.The distal portions of the anterolat- fenestrated and thorny (Fig. 4F), whereas the anterolateral eral and preoralrods were thorny and the proximalportions

: . ., < * ;. i ' S'::','t , ,;:.:.:: :., :: !. : '' ::i''' l :ji.- .S,i:-.,,.t.,.,.i ..i>i4;:~ ;'::-'ii:::"~" ;?. ,I:..,?- :, .:.:.?:. ,,. ..:... . ?.,:, -,,, ' o ...... : '-' ??>;, -.!.; +, ? +.sw?e , *+, vv,@,5,}.-;,c.* ?- * t ? ? + ???? ????-?- ,.:e.-;-ir,.S . i

:: .,(?,(; ?o?M 3??:~?::; -: ( V: ;::;:nril,): s n:

Figure 3. (A) Two-arm echinopluteuslarva of Aspidodiademajacobyi at 6 days, priorto the opening of the mouth, showing well-developed postoralarms (PO) and opaque gut region (OG). (B) Eleven-day-oldlarva in the early four-arm stage, viewed dorsally with newly formed anterolateralarms (AL). The stomach region (S) is becoming more visible. (C) Ventralview of a 15-day, 4-arm pluteus larva. Note that the anterolateralarms (AL) are supportedby simple, unfenestratedrods. (D) Early 6-arm echinopluteus,25 days old, viewed from the dorsal side with small posterodorsal arms (PD) developing. (E) Intermediate 6-arm echinopluteus stage viewed ventrally 28 days after fertilization,with well-developed posterodorsalarms (PD) and presumptivebuds where the preoralarms (PR) will eventually develop. (F) Advanced 6-arm echinopluteus (ventral view), 32 days old, shortly after the first appearanceof the unpairedposterior process (UPP). Preoralarms (PR) are still present only as tiny buds. (G) Ventralview of 39-day, early 10-armlarva showing well-developed unpairedposterior process (UPP) and preoralarms (PR). Posterolateralarms (PL) are just beginning to grow in this individual, but many in the same cultures were still at a late 8-arm stage. (H) Forty-two-day-old, 10-arm echinopluteus viewed ventrallyjust after the appearanceof lateral ciliary ridges (CR). Posterolateralarms (PL) are growing asym- metrically and anterolateralarms have been resorbed. (I) Ten-arm larval stage (42 days old) after preoral and anterolateralarms have been resorbed. The posterior ciliary band (PCB) is visible as a dark region partially obscured by the transparentpreoral lobe. 392 C. M. YOUNG AND S. B. GEORGE

W.

1

Figure 4. (A) Forty-nine-day-old echinopluteus larva of Aspidodiadema jacobyi in dorsal view. The posterodorsallobes (PDL) have just appeared,and the posterior ciliary band is visible. PCB, posterior ciliary band; PO, postoral arm; PD, posterodorsal arm; UPP, unpaired posterior process. (B) Posterior view of a 53-day-old larva showing one of the large posterodorsallobes (PDL). (C) Mid-anteriorregion of a 75-day-old echinopluteus,showing the lateralciliated ridges (CR) now expandedinto a large loop of cilia that surroundsthe entire region of the resorbed pre-oral and anterolateralarms. (D) Ventral view of a 134-day-old larva with echinus rudiment (R) and very broad bases of the postoral arms (PO). (E) Echinopluteuslarva 134 days old, showing echinus rudiment (R). (F) Posterior region of a late echinopluteus larva digested with sodium hypochlorite to reveal skeletal structures.Note that the postoral rods (POR) and posterodorsalrods (PDR) are fenestratedand tightly coupled with a basket-likesterom near their bases. The anterolateralrod (ALR) is simple. (G) Dorsal arch (DA) of an echinopluteuswith preoralrods (PRR) originatingon the two ends and a large central process (CP) in the middle. The central process is simple at the proximal end and fenestrateddistally (barely visible at the far left-handside of the micrograph).(H) Bases of the two postoralrods with their fenestratedplates of stereom (FP) fused together to create part of the posteriorbasket. were smooth (Fig. 4G). The posterolateral rods were postoral rods and the posterodorsalrods formed a truncated branchedand extremely thorny. The dorsal arch was com- pyramidal structure.The posterolateralrod and the fenes- plex and slightly asymmetrical,having two large holes and trated thorny rod (Fig. 3H) of the posterior process devel- one much smallerhole in the centralregion of the arch (Fig. oped from the posterior transverserod, which was situated 4G). A long central process, half of which was thorny and dorsally. solid and the otherhalf of which was thornyand fenestrated, projected from the base of the dorsal arch (Fig. 4G). Be- Discussion tween the 42nd and 49th day after fertilization,the base of each postoralrod expandedinto a dense networkof stereom Early stages of embryology and larval development have that eventually took the form of a highly fenestratedtrian- been reportedfor a numberof bathyal and abyssal inverte- gular plate (Fig. 4F). This plate fused with the correspond- brates (Prouho, 1888; Mortensen, 1921; Young and Cam- ing plate at the base of the other postoral rod (Fig. 4F) to eron, 1989; Young and Tyler, 1993; Young et al., 1996b, c), form a basket-like structurethat filled the broadbases of the but the complete larval development has been describedfor postoral arms (Fig. 4H, D). A slightly smaller fenestrated only a single eurybathicechinoid, Cidaris cidaris (Prouho, plate developed at the base of each posterodorsalrod (Fig. 1888). Thus, the present work is not only the first complete 4F). Together, the fenestrated plates originating from the descriptionof larval development in the family Aspidodia- DEEP-SEA ECHINOID LARVAE 393 dematidae, but also the only complete developmental se- Culture conditions might also explain why the preoral quence known for any exclusively deep-sea invertebrate. and anterolateralarms (and occasionally other arms: Fig. For the first time, we also have a direct estimate of the 4A) were resorbed in most individuals shortly after the length of the pelagic larval period for a deep-sea animal. On onset of the 10-armlarval stage. On the other hand, many of the basis of these laboratoryrearings, we may conclude that, these larvae survived for an additional 2 months or more given adequatefood, larvae are capable of dispersing for at and some developed rudiments, suggesting that they were least 5 months. In the Tongue of the Ocean, Bahamas, we not unhealthy. Resorption of some arms as others develop have measured current speeds as high as 18 cm/s at the appears to be normal in certain species of echinoids depths where Aspidodiademajacobyi lives (Young et al., (Mortensen, 1921; Pearse and Cameron, 1991). 1993). The current speed in this region averages about 5 Mortensen (1921, 1931) described the morphology of cm/s and flows mostly in a single direction (Young et al., echinodermlarvae in minute detail with the expectationthat 1993). Using these flow values for a rough eulerian calcu- larval characterswould be keys to the unravelingof phylo- lation of dispersaldistance, a larva could drift about 4.3 km genetic relationships within the phylum. Since that time, per day, or 645 km in 5 months. The real dispersaldistance, larval characters have been widely used for interpreting of course, will be influenced by the local currentregimes, life-history evolution (reviewed by Havenhand, 1995; topographicalfeatures, and settlementopportunities that the Wray, 1995), and for resolving phylogenetic issues at higher larva encounters during the course of its drift. taxonomic levels (e.g., Hadzi, 1963; Strathmann, 1988; Young et al. (1996a) showed that larvae of A. jacobyi Nielsen, 1995). However, recent analyses that combine ge- cannot tolerate the high temperaturesfound in the upper netic data with adult and larval charactersets have shown 200 m of the water column, suggesting that dispersaloccurs that larval charactersoften evolved independentlyof adult below the permanentthermocline and that larvae must feed body form (Smith and Littlewood, 1997) and that larval on items sinking out of the euphotic zone. In an earlierstudy charactersare more prone to homoplasy than adult charac- (Young et al., 1989), it was noted that the larvae of A. ters (Smith et al., 1995, 1996). Wray (1992) pointed out jacobyi have an unusually long prefeeding period, and that many similaritiesamong larvae from groups as unrelatedas they can survive on maternalreserves for up to 2 months. It ophiuroids and diadematoids,and as spatangoidsand phy- was hypothesized that this attributemay confer nutritional mostomatoids. These commonalities are problematic for flexibility on these larvae, enabling them to disperse and interpretingphylogeny; indeed, both Strathmann(1988) and survive in deep oligotrophic waters where food items are Raff et al. (1988) have recognized that any phylogeny based scarce and patchy (Young et al., 1989). In the present study, on larval charactersmust accept a number of remarkable we detected no additionalmorphological attributes that can convergences. Despite the problems of interpretinglarval be interpretedas special adaptationsfor deep-sea dispersal. charactersfrom a phylogenetic standpoint,parsimony de- Complex ciliation and long arms presumably keep these mands that we be very careful before concluding that an large larvae suspended in the water column, but similar attributehas evolved independentlyin multiple clades. complexity is found in late-stage larvae of many shallow- The order Diadematoidais currentlycomposed of three water echinoids (Mortensen, 1921; Emlet, 1988b; Pearse families: Diadematidae,Lissodiadematidae, and Aspidodia- and Cameron, 1991). The large posterior process on the dematidae (Smith, 1984). The fully developed larva of A. larva of A. jacobyi lowers the center of mass, presumably jacobyi differs significantly from the known diadematid facilitating uprightorientation (Pennington and Strathmann, larvae in its extraordinarycomplexity. Whereaslarvae of A. 1990; Young, 1995), but this is a feature also found in jacobyi have five pairs of larval arms and a posterior pro- shallow-water spatangoid larvae (Mortensen, 1921). cess, larvae in the family Diadematidaehave only one pair The developmental rate of A. jacobyi to the 2-arm stage of larval arms (the postorals) and no posterior process (6 days) was similar to that reportedby Young et al. (1989), (Mortensen, 1921, 1931). Although the family Aspidodia- but after the 2-arm stage, differences were observed be- dematidaeis found exclusively in deep water and the other tween the two studies. The mouth opened 11 days after two families are restricted to shallow waters, the known fertilizationin the presentstudy and between 18 and 21 days larvae of diadematoids do share a number of common after fertilizationin the study by Young et al. (1989). Fully features.Like Aspidodiadema,all known larvae in the Fam- developed 4-armA. jacobyi larvaewere observed in cultures ily Diadematidae (e.g., Diadema antillarum, Astropyga 15 days after fertilizationin the present study, but not until pulvinata, and Diadema setosum; Mortensen, 1921, 1931) 24 days after fertilizationby Young et al. (1989). Diet could have very long, thorny, and fenestrated postoral rods; a explain differences in development after the onset of feed- body skeleton in the shape of a 4-sided truncatedpryramid; ing, but not differences in the rate at which the mouth and a posteriortransverse rod, which in some species takes developed. The differences between these two studies could the form of a fenestratedtriangular plate during advanced be attributedto genetics, egg quality (Emlet et al., 1987; stages. Most significantly, Mortensen (1921) reportedthat George et al., 1990; George, 1996) or culture conditions. the larva of "Echinopluteus transversus" e and f (later 394 C. M. YOUNG AND S. B. GEORGE shown to be Diadema setosum) has a very short,fenestrated, * 10-12 arms, posteriorprocess and unpairedossicle immediatelybelow the posteriortrans- * 2 no verse rod. He expressed the opinion (Mortensen, 1921, page arms, posteriorprocess 90) that although this structurebears a superficial resem- blance to the posterior process of spatangoid larvae, the j Pedinoida& structures are not homologous. Whereas the spatangoid Irregularia posteriorprocess is an outgrowthof the posteriortransverse rod, that of the diadematidlarva is not connected to the rest * Aspidodiadematidae of the skeleton. 0 Diadematidae Larvaeof Aspidodiademajacobyihave several featuresin common with larvae of the phymostomatoidsArbacia lixula * Lissodiadematidae (Fenaux, 1969) and A. punctulata (Garman and Colton, 1883). Like A. jacobyi, these species possess very long * Micropygidae postoral and posterodorsalarms, swept-back posterolateral Figure 5. Phylogenetic tree suggested by and redrawn from Emlet arms, a of vibratile lobes, and fenestrated of pair triangles (1988a). This tree organizes the Diadematidaand related euechinoids in stereom that form at the base of the postoraland posterodor- such a way as to preserve the importantrelationships of adult characters sal rods before metamorphosis.The Arbacia species, like A. (Jensen, 1981) while requiringthat the 2-arm echinopluteus evolve only jacobyi, also develop parallelciliated ridges that originatein once. The larval form of Aspidodiadema fits perfectly into this scheme, with no revisions to the demands of the region of the posterodorsalarms and run parallel to the required satisfy parsimony. anterolateralrods along the midsection of the larva (Fenaux, 1969). However, these terminatewith the formation ridges dation. The larval form of Aspidodiadema is interpreted of the sixth of larval arms in the pair (anterodorsalarms) parsimoniouslyonly if the Aspidodiadematidaeare classi- not with a aroundthe base of phymostomatoids, ciliary ring fied as a separate order outside the Diadematoida or as a the and anterolateralarm as in A. preoral region jacobyi. sister group to the other families within the order. Indeed, Because of its arms and posterolateral unpairedposterior the remarkablesimilarities between A. jacobyi larvae and larvae of A. bear a resemblanceto process, jacobyi striking those of the spatangoidsand clypeasteroidssuggest that the the characteristic larval form of urchins in the irregular Aspidodiadematidaemay be near the stem group that sep- order 1921; Rees, The fe- Spatangoida(Mortensen, 1953). aratedregular and irregulareuechinoids. nestratedbasket structureof Aspidodiadema larvae is also similar to the basket that characterizesthe very Clypeaster- Acknowledgments oids, another major clade of irregularurchins (Pearse and Cameron, 1991). Larvaewith posteriorprocesses have been We thank Mary Clark, who supplied us with algae; Paul hithertounknown outside the spatangoids,and all authorsto Tyler, who assisted at sea and helped spawn the urchins;and date have considered this characterto be absolutely diag- Tracy Griffin, who assisted in the preparationof figures. nostic of this order (Mortensen, 1921; Pearse and Cameron, Andrew Smith and RichardMooi advised us in matters of 1991). Other features shared by these two groups include phylogeny, though the opinions we express herein are en- long postoral and posterodorsal arms and well-developed tirely ours. This work was funded by the National Science anterolateralarms. However, spatangoid larvae have six Foundation (OCE-9116560 and OCE-9633784) and by a pairs of arms, whereas Aspidodiadema larvae have only Harbor Branch postdoctoral fellowship to Sophie George. five. The loss of arms is easy to explain in the evolution of HBOI contributionnumber 1339. larval forms, but the independentacquisition of a structure as distinctive as a fenestrated,unpaired posterior process in Literature Cited two different clades seems improbable. Emlet, R. B. 1985. Crystal axes in recent and fossil adult echinoids Emlet (1988a) used crystal orientations in the apical indicate trophic mode in larval development. Science 230: 937-940. plates to predict that all larvae in the family Aspidodiade- Emlet, R. B. 1988a. Crystallographicaxes of echinoid genital plates reflect larval matidae should have at least two pairs of larval arms form: some phylogenetic implications. Pp. 299-310 in (pos- Echinoderm toral and and less larval Phylogeny and EvolutionaryBiology, C. R. C. Paul and posterodorsalarms) widely spread A. B. Smith, eds. ClarendonPress, Oxford. arms than the Diadematidaeand both of Lissodiadematidae, Emlet, R. B. 1988b. Larvalform and metamorphosisof a "primitive"sea which should have only widely spreadpostoral arms. On the urchin, Eucidaris thouarsi (Echinodermata:Echinoidea: Cidaroida), basis of this prediction, Emlet (1988a) suggested that the with implications for developmental and phylogenetic studies. Biol. Bull. 174: 4-19. family Aspidodiadematidaemust be a sister group to the R. L. B. and R. R. Strathmann. 1987. Echi- families Diadematidae, and Emlet, B., McEdward, Lissodiadematidae, Micropygi- nodermlarval viewed from the 55-136 in Echinoderm all of which have a ecology egg. Pp. dae, only single pair of larval arms (Fig. Studies Vol. 2, M. Jangoux and J. M. Lawrence, eds. A. A. Balkema, 5). Our findings lend strong supportto Emlet's recommen- Rotterdam. DEEP-SEA ECHINOID LARVAE 395

Fenaux, L. 1969. Les echinopluteus de la Mediterranee. Bull. Inst. Smith, A. B. 1984. Echinoid Paleobiology. George Allen and Unwin, Oceanogr. (Monaco) 68: 1-28. London. 190 pp. Fenaux, L., C. Cellario, and M. Etienne. 1985. Variations in the Smith, A. B. 1995. Echinoids from the JurassicOxford clay of England. ingestion rate of algal cells with morphologicaldevelopment of larvae Palaeontology 38: 743-755. of Paracentrotus lividus (Echinodermata: Echinoidea) Mar. Ecol. Smith, A. B., and D. T. J. Littlewood. 1997. Molecular and morpho- Prog. Ser. 24: 161-165. logical evolution during the post-Palaeozoic diversification of echi- Garman, H., and B. Colton. 1883. Some notes on the development of noids. Pp. 559-583 in Molecular Evolution and Adaptive Radiation, Arbacia punctulata Lam. Studies Biol. Lab. Johns Hopkins University T. J. Givnish and K. J. Sytsma, eds. Cambridge University Press, 2: 247-255. Cambridge. George, S. B. 1990. Populationand seasonal differences in egg quality of Smith, A. B., D. T. J. Littlewood, and G. A. Wray. 1995. Comparing Arbacia lixula (Echinodermata:Echinoidea) Invertebr. Reprod. Dev. patternsof evolution: larval and adult life history stages and ribosomal 17: 111-121. RNA of post-Palaeozoicechinoids. Philos. Trans.R. Soc. Lond. B 349: George, S. B. 1996. Echinodermegg and larval quality as a function of 11-18. adult nutritionalstate. Oceanol. Acta 19: 297-308. Smith, A. B., D. T. J. Littlewood, and G. A. Wray. 1996. Comparative George, S. B., C. Cellario, and L. Fenaux. 1990. Populationdifferences evolution of larval and adult life-history stages and small subunit in egg quality of Arbacia lixula (Echinodermata:Echinoidea): proxi- ribosomal RNA amongst post-Palaeozoic echinoids. Pp. 234-254 in mate composition of eggs and larval development. J. Exp. Mar. Biol. New Uses for New Phylogenies, P. H. Harvey, A. J. L. Brown, J. M. Ecol. 141: 107-118. Smith, and S. Nee, eds. Oxford University Press, Oxford. Hadzi, J. 1963. The Evolution of the Metazoa. PergamonPress, Oxford. Strathmann, R. R. 1988. Larvae, phylogeny, and von Baer's law. Pp. 499 pp. 53-68 in EchinodermPhylogeny and EvolutionaryBiology, C. R. C. Havenhand, J. N. 1995. Evolutionaryecology of larval types. Pp. 79- Paul and A. B. Smith, eds. ClarendonPress, Oxford. 122 in Ecology of Marine InvertebrateLarvae, L. McEdward,ed. CRC Wray, G. 1992. The evolution of larval morphology during the post- Press, Boca Raton, FL. Paleozoic radiationof echinoids. Paleobiology 18: 258-287. Hyman, L. H. 1955. The Invertebrates:Echinodermata. McGraw-Hill, Wray, G. 1995. Evolution of larvae and developmental modes. Pp. New York. 763 pp. 413-447 in Ecology of Marine InvertebrateLarvae. L. McEdward,ed. Jensen, M. 1981. Morphology and classification of the Euechinoidea CRC Press, Boca Raton, FL. Bronn, 1860-a cladistic analysis. Vidensk. Medd. Dan. Naturhist. Young, C. M. 1992. Episodic recruitment and cohort dominance in Foren. 143: 7-99. echinoid populations at bathyal depths. Pp. 239-246 in Marine Eu- Mortensen, T. A. 1921. Studies of the Developmentand Larval Forms of trophicationand Population Dynamics, G. Colombo, I. Ferrari,V. U. Echinoderms.G. E. C. Gad, Copenhagen. 260 pp. Ceccherelli, and R. Rossi, eds. Olsen & Olsen, Fredensborg,Denmark. Mortensen, T. 1927. Handbook of the Echinodermsof the British Isles. Young, C. M. 1995. Behavior and locomotion duringthe dispersalphase Oxford University Press, London. 471 pp. of larval life. Pp. 249-277 in Ecology of Marine InvertebrateLarvae, Mortensen, T. A. 1931. Contributionsto the study of the development L. McEdward,ed. CRC Press, Boca Raton, FL. and larval forms of echinoderms. I-II. K. Dan. Vidensk. Selsk., Skr. Young, C. M., and J. L. Cameron. 1989. Developmental rate as a Naturvid. Math. Afd. 9, 4:(1) 1-65. function of depth in the bathyalechinoid Linopneusteslongispinus. Pp. Nielsen, C. 1995. Animal Evolution. Interrelationships of the Living 225-231 in Reproduction,Genetics and Distributions of Marine Or- Phyla. Oxford University Press, Oxford. 467 pp. ganisms, J. S. Ryland and P. A. Tyler, eds. Olsen & Olsen, Fredens- Pearse, J. S., and R. A. Cameron. 1991. Echinodermata:Echinoidea. borg, Denmark. Pp. 513-662 in Reproductionof Marine Invertebrates, Vol. 6, A. C. Young, C. M., and P. A. Tyler. 1993. Embryosof the deep-sea echinoid Giese, J. S. Pearse, and V. B. Pearse, eds. Boxwood Press, Pacific Echinus affinus requirehigh pressurefor fertilization.Limnol. Ocean- Grove, CA. ogr. 38: 178-181. Pennington, J. T., and R. R. Strathmann. 1990. Consequences of the Young, C. M., J. L. Cameron, and K. J. Eckelbarger. 1989. Extended calcite skeletons of planktonic echinoderm larvae for orientation, pre-feeding period in planktotrophiclarvae of the bathyal echinoid swimming, and shape. Biol. Bull. 179: 121-133. Aspidodiademajacobyi. J. Mar. Biol. Assoc. UK 69: 695-702. Prouho, H. 1888. Recherches sur le Dorocidaris papillata et quelques Young, C. M., P. A. Tyler, R. H. Emson, and J. D. Gage. 1993. autres 6chinides de la Mediterran6e.Arch. Zool. Exp. Gen. (2) 213- Perception and selection of macrophyte detrital falls by the bathyal 380. echinoid Stylocidaris lineata. Deep-Sea Res. 40: 1475-1486. Raff, R. A., K. G. Field, M. T. Ghiselin, D. L. Lane, G. J. Olsen, N. R. Young, C. M., M. G. Devin, W. B. Jaeckle, S. U. K. Ekaratne, and S. B. Pace, A. L. Parks, B. A. Parr, and E. Raff. 1988. Molecular George. 1996a. The potential for ontogenetic vertical migration by analysis of distant phylogenetic relationships in echinoderms. Pp. larvae of bathyal echinoderms. Oceanol. Acta 19: 263-271. 30-41 in EchinodermPhylogeny and EvolutionaryBiology, C. R. C. Young, C. M., P. A. Tyler, and J. D. Gage. 1996b. Vertical distribution Paul and A. B. Smith, eds. ClarendonPress, Oxford. correlateswith pressuretolerances of cleaving embryos in the deep-sea Raup, D. M. 1965. Crystal orientationsin the echinoid apical system. J. asteroid Plutonaster bifrons. J. Mar. Biol. Assoc. UK 76: 749-757. Paleontol. 39: 934-951. Young, C. M., E. Vazquez, A. Metaxas, and P. A. Tyler. 1996c. Rees, C. B. 1953. The larvae of the Spatangidae.J. Mar. Biol. Assoc. UK Embryology of vestimentiferantube worms from deep-sea methane/ 32: 477-490. sulfide seeps. Nature 314: 514-516.