A Hydrodynamic Interpretation of Sand Dollar Morphology

BULLETIN OF MARINE SCIENCE, 31(3): 605-622, 1981

A HYDRODYNAMIC INTERPRETATION OF SAND DOLLAR MORPHOLOGY

Malcolm Telford

ABSTRACT Sand dollars are shallow-domed echinoids which act as lifting bodies in water currents,

Coefficients of lift (CI) were determined experimentally in air for a lunulate species, Mel/ita qllinqlliesperforllta (0,040) and for a nonlunulate species, Echinarachllius parma (0.077). The difference in CI., which was shown to be statistically significant (P = 0.001), can be attributed partly to differences in camber and partly to the presence of lunules. Observations of water flow at 10 cm' sec-1 confirmed that flow was attached and of a lift- generating pattern. Flow around E. parma was hindered by a relatively large standing vortex at the anterior margin. For M. sexiesperforata, which is thinner edged and less steeply cambered, the flow was more perfectly attached; flow from oral to aboral surfaces via the lunules was distinct. Similarly, Encope emarginata had a closely attached flow with some passage through the anal lunule and ambital notches. Calculations from weight-length relationships and coefficients of lift indicated that M. qllillqlliesperfora({l should be able to maintain its position in more severe current regimes than could E. parma. Excess pressure on the oral surface of lunulate sand dollars is relieved by flow along pressure drainage channels which lead from the central region of the disc into the lunules and ambital notches. Scanning electron micrographs show that these lack the podia char- acteristic of food grooves and that podia in adjacent food-gathering areas sweep parallel to or away from the pressure drainage channels. Several earlier theories of lunule function appear to be either insufficiently general or untenable. That lunules shorten the food-path is not fully supported by observation; that they aid in righting movements is contrary to observations of Mel/ita species and that they might strengthen the test is shown by both observation and theoretical considerations to be incorrect.

In the preface to her monograph on the Echinodermata, the late Libbie Hyman (1955) said: "I also here salute the echinoderms as a noble group especially designed to puzzle the zoologist." This is perhaps most true of clypeasteroids which include the familiar sand dollars, sea biscuits and key-hole urchins. Many of these are of bizarre form: the African rotulids have numerous slender indentations of the posterior margin and some also have anterior perforations or lunules; the American genus Mellita (Fig. 1, left) has five or six slit-like lunules piercing the test and the related genus Encope (Fig. 1, right) has a single posterior (anal) lunule and five notches around the margin; the Indo-Pacific genus Echi- nodiscus has two posterior slit-like lunules whilst the fossil genus Amphiope (Miocene, France) had a round port hole in each of the two posterior ambulacra. In Durham's (1966) classification the Clypeasteroida includes twenty-four Recent genera and of these, eight (Mellita, Encope, Leodia, Mellitella, Astriclypeus, Echinodiscus, Rotu/a and Heliophora) have lunules and/or ambital indentations. In addition, at least three fossil genera (Monophoraster, Amphiope and Scutas- ter) had lunules and a further three dubious genera appeared to have notches or broad indentations of the margin. Seilacher (1979) has suggested that lunules have appeared independently in six different groups. The role of these assorted indentations and perforations is speculative and no satisfactory, unifying explanation has yet been offered. However, their repeated occurrence suggests that they

605 606 BULLETIN OF MARINE SCIENCE, VOL. 31, NO.3, 1981

Figure I. (Left) Mel/ita sexiesperjorata: Aboral surface showing pressure drainage channels (black arrow) and food grooves (white arrow). (Right) Encope emarginata: has pressure drainage channels (black arrow) and food grooves (white arrow) very similar to M. sexiesperjorata.

confer significant adaptive advantages on these organisms, suiting them to their special environments. Sand dollars are generally shallow burrowers or exposed, epibenthic organisms. Most species occur in shallow water, some are intertidal, some inhabit deep water and a few, such as Echinarachnius parma, extend from the intertidal zone to considerable depths (Stanley and James, 1971). For burrowing their thin-edged discoidal form is ideal. Equally important, as echinoids living on an unstable sediment surface swept by waves and currents, their form minimizes drag and facilitates maintenance of position. A consequence of their low, domed profile is that they act as hydrofoils and generate lift. Chia (1973) reported the accumulation of high density sand grains which appear to act as a weight belt in Dendraster exentricus. O'Neill (1978) has provided an explanation of the peculiar upright feeding position of D. excentricus which exploits its lift characteristics in a group facilitation of feeding. Also, in the more typical horizontal posture, sand dollars generate lift, in a similar manner to flounders (Arnold and Weihs, 1978) which are able to regulate the amount of lift they generate by adjustments of their fins, thus altering their curvature. I propose the hypothesis that notches and lunules evolved as adaptive modifications to reduce lift, without significantly increasing drag nor disrupting water flow over the surface of horizontally oriented sand dollars exposed on the surface of the sediment.

MATERIALS AND METHODS

Sand Dol/ars.-Echinarachnius parma was collected intertidally and in shallow water at 81. An- drews, N.B.; Mel/ita (Leodia) sexiesperjorata in Barbados; M. quinqlliesperjorata at Beaufort, N.C.; D. excentricus near Victoria, B.C. and E. emarginata was purchased in Florida. For most of this study clean, dry tests were used, prepared in commercial bleach (sodium hypochlorite) and washed in distilled water. Measurements.-Lengths,.widths and thicknesses were obtained to the nearest 0.1 mm with vernier calipers; weights were recorded to the nearest 0.1 g,"Densities of whole specimens were determined TELFORD: SAND DOLLAR MORPHOLOGY 607

Figure 2. (Left) Installation of pressure ports in Echinarachnius parma. a. Longitudinal section: capillaries installed to monitor pressure on aboral (upper) surface. b. Arrangement of pressure ports along anterior-posterior diameter.

Figure 3. (Right) Diagram of lift balance. Specimens were poised in the center of air stream (shown by arrow) in a horizontal position (zero angle of attack) with lift indicator needle centered. As air stream velocity increased and the specimen lifted, weights were added to the balance pan returning lift indicator to the center position.

by displacement. Plan areas were estimated by exposing the sand dollars on photographic paper, then cutting out and weighing the silhouettes. Determination of Lift.- Two techniques of measuring lift, using the Philip Harris Ltd. (Birmingham, England) Air Stream Generator P10600, were employed.

In the first method, the specimens were mounted horizontally in the center of the air stream with the balance arm immobilized so that the sand dollars were unable to lift. Pressures over the surface were measured by means of individual pressure ports. These consisted of lengths of 1.5 mm glass tubing connected to small manometers filled with Krebs' modification of Brodie's solution (Dawson et aI., 1959). The sand dollars were perforated by a dental drill and the capillary tubes glued into place under a microscope, with their open ends minutely recessed below the surface (Fig. 2). By combining data from three individuals of E. parma and two of M. quinquiesperforata, with pressure ports arranged in different positions, composite isobars were prepared for both species. Areas within isobars were estimated by superimposing the diagrams on squared paper. These determinations were made with the specimens poised in the center of the air stream. A simple disc of Plexiglas, flat on each side and a Plexiglas model of E. pamUl domed on the upper surface, were made. Both were tested in the free air stream and again, glued to a flat board which simulated a substrate. For the second method, the Harris Drag and Lift Ba]ance No. P]0602 was used. The specimens were again mounted horizontally in the center of the air stream but with the arm counter-balanced and free to lift. As air velocity was increased and the specimens lifted the starting horizontal position was restored by adding weights to the balance pan (Fig. 3). Balance measurements of lift were obtained for twenty specimens each of E. parma and M. quinquiesperforata with the air flow passing exactly anterior to posterior and again with the flow reversed, posterior to anterior. For M. quin- quiesperforatll both of these measurements were repeated with the ]unules smoothly filled with paraffin wax. Water Tunnel Experiments.-Specimens of E. parnUl, M. sexiesperforata and E. emarginata were placed flat on the floor of the observation chamber, simulating their natural position on the substrate. Depending on size, the specimens were 10-]5 em from the chamber walls (a full description of the facility is given by Dobrodzicki, 1972). Using a fine jet of fluorescein for visualization, observations were made at a flow rate of ]0 cm·sec-t• Sand dollars were sprayed with black paint to enhance contrast. 608 BULLETIN OF MARINE SCIENCE. VOL. 31. NO.3. 1981

Figure 4. Isobars on aboral surface of Mellita quinquiesperforata (left) and Echinaraclmius parma (right) in air stream of 1,437 em' sec-I. Pressures, indicated in mm Kreb's solution, are relative to stagnation pressure of oral surface. Data combined from three specimens of E. parma and two of M. quinquiesperforata.

Scanning Electron Microscopy.-Specimens of both M. sexiesperforata and M. quinquiesperforara were cleaned and washed as described, air dried, mounted on stubs and sputter coated with gold in a SEMPREP 2 (Nanotech (Thin Films) Ltd., Cambridge, England) before scanning with a Cambridge SI80 SEM. Righting Experiments.-Laboratory and field experiments using M. sexiesperforata were conducted in Barbados. The sphaeridia (Fig. 10) were removed under the microscope by means of No. 0 steel insect pins. All field data were collected whilst using SCUBA gear. Feeding .-Observations of M. sexiesperforata feeding were made in the laboratory using a suspension of yeast stained with neutral red.

RESULTS Coefficient of Lift, CL.-The distribution of pressure over the surface of E. parma and M. quinquiesperj'orata is shown in Figure 4. Calculated areas within the isobars are given in Table 1. To calculate the coefficient of lift, the lift equation (Prandtl and Tietjens, 1934) was used:

2 L = 0.5pv ACL (1)

2 or CL = 2L1pv A (2) where lift (L) is given in dynes; density of the fluid medium (p) in g' cm-3; velocity of flow (v) in cm· sec-1 and A is the projected (plan) area in cm2• The coefficient of lift, CL, is dimensionless. The estimated value for E. parma was C" = 0.079 and for M. quinquiesperforata CL = 0.044. Experiments with the lift balance gave an average coefficient of lift for 20 E. parma CL = 0.077 (SE ± 0.004) and for 20. M. quinquiesperforata CL = 0.040 (SE ± 0.002). The difference between the two means was highly significant (t = 46.082, P = 0.001). For both species specimens were chosen to cover as large a size range as possible (2.88 cm to 6.59 cm for E. parma and 3.02 cm to 5.85 cm for M. quinquiesperforata). Neither of these species is perfectly symmetrical. The anterior edge of E. TELFORD: SAND DOLLAR MORPHOLOGY 609

Table I. Percent areas within isobars of sand dollars in open wind tunnel with air stream velocity 1,437 cm' sec-I (combined data for three specimens of E. parma and two of M. quinquiesperforata)

Pressure % Total Contribution Species mm Krebs' Area to Lift Dynes

E. parma -5 1.6 388.0 -4 3.9 756.5 -3 8.4 1,222.1 -2 15.4 1,493.7 -I 36.1 1,750.7 0 28.1 +1 4.9 -237.6 +2 3.6 -349.2 total: 5,024.2 M. quillquiesperforata -5 1.3 178.1 -4 2.8 306.9 -3 7.8 641.1 -2 14.2 778.1 -I 32.6 893.2 0 39.1 +1 4.1 -112.3 +2 2.2 -120.6 total: 1,583.3

parma is approximately twice the thickness of the posterior edge and in some specimens the camber is slightly steeper in the anterior part of the test. When its position was reversed in the wind tunnel, with the air stream approaching posteriorly, the average lift generated decreased by 15%. When M. quinquiesperforata was reversed in the air stream, lift was decreased by 33%. The specimens of M. quinquiesperforata were tested again with the air stream from the anterior end and the lunules filled with paraffin wax, at which time the

average coefficient of lift was CL = 0.060 (SE ± 0.003). When reversed with the lunules occluded no increase in lift was observed, i.e., the average value was within 5% of lift in the reversed position with open lunules. Pressure and balance determinations of lift were made in the free air stream. Under natural conditions sand dollars live on the surface of the substrate or burrow to a shallow depth. An attempt to simulate the position of a sand dollar on the substrate was made with the Plexiglas model of E. parma. The model was imperfect in three respects: it was completely symmetrical; it was smoother than a real sand dollar and, it did not have such a prominent "anterior" edge. Poised in the center of the air stream, the pressure differences over the curved surface were similar to those of E. parma, but no area of positive pressure was observed and the greatest lifting pressure was slightly lower. On the plane under-surface no difference from stationary air pressure was found, neither was it found using real tests. No differences were detected when the Plexiglas model was mounted on a board so that only the upper surface was traversed by moving air. A simple flat disc was also used and, as anticipated, showed no lifting characteristics at all. The stereom of the sand dollar test is porous and it seemed possible that this porosity might reduce the amount of lift generated. A number of specimens were varnished to seal the pores and re-tested without discernible effect. Water Tunnel Observations.-EcHlNARACHN/US PARMA(SPINES ATTACHED). A distinct, stationary boundary zone was apparent around the anterior edges. At the 610 BULLETIN OF MARINE SCIENCE, VOL. 31, NO.3, 1981

Figure 5. (Upper left) Flow around Echinarachnius parma in water tunnel at 10 cm· sec'. To left of specimen (anterior) there is a standing vortex; back-flow into the area of separation can be seen posteriorly. (Upper right) Lateral view of E. parma in flume. From the anterior (left) flow was attached over the apex to about the middle of the posterior slope. (Lower left) Water tunnel flow around Me/lita sexiesperforata. Passage of water from oral (under) to aboral surface via the lunules is distinct. (Lower right) Encope emarginata. showing some flow through the anal lunule and ambital notches.

most anterior point a pronounced standing vortex was visible (Fig. 5). Above this the flow approached the specimen smoothly and became attached just posterior to the margin and remained attached to about the middle of the posterior slope, at which point it separated. Some back-flow into the area of separation was apparent (Fig. 5). When the flow was briefly accelerated to 20 cm' sec-1 it remained attached almost to the trailing margin.

MELLITA SEXIESPERFORATA (SPINESATTACHED).No additional specimens of M. quinquiesperforata were available, hence the choice of this species (M. sexiesperforata). Unlike M. quinquiesperforata, M. sexiesperforata is even thinner- edged, has a lower camber and the anterior edge does not touch the substrate when lying flat. The boundary layer and anterior vortex were greatly reduced and some flow underneath was discernible. Flow over the upper sutface was smoothly attached from the anterior edge to the middle of the anal lunule, at which point it became separated. A pronounced flow through all of the lunules, except the most anterior, was readily visible (Fig. 5).

EN COPE EMARGINATA (TEST ONLY). Despite the relatively thick edge of this species, water flow at the anterior edge showed no formation of a vortex and the boundary layer around the anterior was small. In fact, water flow appeared to enter the anterior ambital notch in many trials. Over the entire anterior sutface flow was closely attached and separated at the anterior lip of the anal lunule. A TELFORD: SAND DOLLAR MORPHOLOGY 611

E u :i >- (!) E Z u W ...J cr: w >- 10 w :;:

10 50 WEIGHT, g.

10 100 VELOCI TY, C m.sec-1

Figure 6. (Left) Weight-length relationships for sand dollar tests fit a power curve, L = aWb. (Upper) Non-Iunulate species: Del/draster excel/triClis (broken line), L = 3.62 WU.'6 (r' = 0.95, n = 30); Echi- I/lirachl/ius parma (solid line), L = 3.69 wo.,S (r' = 0.86, n = 60). (Lower) Lunulate species: EI/cope emargillata (broken line), L = 4.46 wo.,o (r' = 0.72, n = 30); solid lines are for two species of Mel/ita. M. sexiesperjorara, (upper), L = 4.23 WO.'6 (r' = 0.85, n = 30) and M. quillquiesperjorata, (lower), where L = 3.62 WO.27(r2 = 0.92, n = 50). Figure 7. (Right) Water velocity theoretically required to lift sand dollars of different diameters. Data for Dem/raster excelllricus based partly on O'Neill's (1978) determination of coefficient of lift. a: D. excelltricus; b: E. parma; c: M quillquiesperjorata with lunules closed by wax; d: M. quill- quiesperjorala with lunules open (normal).

mirror beneath the chamber floor and under the specimen showed rapid penetra- tion of the dye, which started to stream from the notches and anal lunule (Fig. 5). Back-flow into the area of separation was apparent. Wei[?ht-Length Relationships .-For five species, the length (anterior-posterior diameter) and weight of the dry test fitted a power curve (Fig. 6):

Length (y) = a Weight (X)b (3) The values of r2 (coefficient of determination) indicate a good fit for all species with the possible exception of E. emarginata for which only large specimens were available (7.80 to 9.50 cm). Wet weight and length of 30 E. parma also provided a very close fit (L = 2.48 WO.31,r2 = 0.99). Density was determined for these specimens (1.45, SE ± 0.01) and for 19 M. quinquiesperjorata (1.37, SE ± 0.03). From the densities and the relationship between wet weight and dry weights, it is possible to calculate the weight in water for living sand doIlars of various sizes. From this the water velocities (VIJ theoreticaIly necessary to lift the sand doIlars under ideal conditions can be determined. Figure 7 shows the calculated velocities necessary to lift E. parma and M. quinquiesperjorata with lunules both

open (normal) and closed (with wax). Diameter of the sand doIlar and VL fit a 612 BULLETIN OF MARINE SCIENCE. VOL. 31. NO.3, 1981

2 power curve. For E. parma L = 0.008 VL1.56 (r = 0.99); for M. quinquiesper- 2 forata (normal) L = 0.0002 VL2.33 (r = 0.98) and with the lunules closed, L = 2 0.0003 VL2.33 (r = 0.98). These coefficients of determination (r ) indicate little scatter of the data. They were used to derive t-values (Stanley, ]960) which showed a significant correlation between V L and diameter, with probabilities in each case of 0.0]. The line for D. excentricus was calculated from O'Neill's (1978) coefficient of lift, assuming the same density as E. parma. Scanning Electron Microscopy.-The test of M. sexiesperforata was examined to map the locations of accessory podia. Figure 8 (upper) shows a branch of the main food groove with two smaller branches joining it. The food grooves are characteristically pierced by relatively large pores for the podia which move food towards the mouth. The pores are elongated in the line of podial sweep, slightly narrower at the end of the effective stroke (i.e., towards the mouth). Surrounding the bases of the spines are more numerous, smaller pores for podia which move food towards the food grooves. Figure 8 (center) shows one of the shallow depressions, which Goodbody (1960) has called "food tracts," entering a lunule. There are no podia in these depressions but the podial pores are conspicuous in adjacent areas. The lower plate in Figure 8 shows the vertical wall of the lunule itself: it can be seen that podial pores appear up to the lip of the lunule but not within it. The main food grooves converge and join near the mouth (Fig. 9). The shallow depressions which lack podia do not extend as far as the mouth, but terminate before the point of juncture of the food grooves. Figure ]0 shows the edge of the mouth of M. sexiesperforata which has larger pores for the feeding podia close to the mouth and two large apertures for the circum-oral podia beside each of the five sphaeridial chambers. Feeding podia do not extend onto the aboral surface (Fig. 10). The edge of the test (ambitus) is pierced by larger pores for sensory podia around the bases of the frill spines (Fig. 10). Mel/ita quinquiesperforata has very similar features: Figure 9 shows both a small branch of a food groove at the edge of a lunule and also one of the shallow depressions at the point where it enters the lunule. The micrographs of Figure 9 show that accessory podia occur up to the lip of the lunule but not within it. Feeding of Mellita sexiesperforata.-Qualitative observations were made by feeding M. sexiesperforata a yeast suspension and observing the movement of cells under the microscope. On the aboral surface ciliary currents sweep the yeast cells centrifugally to the ambitus, whilst on the oral surface food particles move centripetally towards the mouth. This movement appears to be due to the small accessory podia and the particles are rapidly incorporated into mucus-bound strands before entering the food grooves proper. Some of the yeast cells passed through the lunules en route to the oral surface and, although this was not measured quantitatively, it appeared to be insignificant at less than ]0% of the total food flow. Mucus entrapment of food particles seems to occur in the areas where the small podia sweep the food towards the food grooves, as indicated by Good- body (1960). Burrowing and Righting Behavior of M. sexiesperforata.-In still water, in the laboratory, M. sexiesperforata is unable to right itself from an inverted position. In over 30 trials with different sand dollars none managed to right themselves although they moved around for several hours on their aboral surfaces. In natural locations sand dollars are turned over by several agencies, including people, fish and waves (personal observations) but one seldom finds an inverted sand dollar: evidently they are able to recover their proper positions. Two trials were made TELFORD: SAND DOLLAR MORPHOLOGY 613

Figure 8. Scanning electron micrographs of Melli/a sexiesperfora/ll. (Upper) Oral surface (x75): convergence of small branches of food groove system (arrow). The grooves are pierced by elongate pores for the podia which move food. They are flanked by non-ambulatory spine bases and smaller pores for food sweeping podia whose pores are also elongated in the direction of sweep. (Center) Oral surface (x70): pressure drainage channel (arrow) ("food tract" of Goodbody, ]960), devoid of podial pores. (Lower) Oral surface (x60): lip of lunule. Food sweeping podia do not extend into lunule; direction of sweep is parallel to or away from pressure drainage channel (arrow). 614 BULLETIN OF MARINE SCIENCE. VOL. 31. NO.3. 1981

Figure 9. Scanning electron micrographs of M: sexiesperjorafa (upper) and M. qllillqlliesperjorafa (center and lower). (Upper) Oral surface (x50): convergence of main branches of food grooves. Pressure drainage channels, without podia, do not extend to mouth but terminate short of food groove convergences. (Center) Oral surface (x50): food groove (black arrow) with pores for podia and pressure drainage channel (white arrow) devoid of podia, close to lip of lunule. (Lower) Edge of lunule (x50): pressure drainage channel (black arrow) reaches the edge of the lunule as do the areas of food sweeping podia but these do not extend into it. White arrow indicates the vertical wall of the lunule. TELFORD: SAND DOLLAR MORPHOLOGY 615

Figure 10. Scanning electron micrographs of M. sexiesperforata. (Top left) Edge of mouth (x50): close to the mouth, pores for the feeding podia (arrow) become larger. The sphaeridial chambers are located in the food grooves, bordering the mouth. Beside each of the five chambers there is a pair of very large apertures for the circum-oral podia. (Top right) Ambitus (x65): between the bases of the fringing spines there are pores for large sensory podia (arrow). (Bottom left) Aboral surface (x65): accessory podia do not extend onto the aboral surface; food particles are here moved by cilia between the spine bases. (Bottom right) Aboral surface (x65): spines ill situ; each of the shoe spines is encircled by smaller spines which allow fine particles to fall onto body surface. in shallow water (3 m) in each of which 30 M. sexiesperforata were inverted and watched for over 60 min by a SCUBA diver. The specimens were eventually flipped over by wave action. "Flip-over" time (Fig. II) and subsequent behavior were recorded. Projected time for recovery of position by 99% of test population was 83 min. After flipping over the sand dollars buried themselves: time taken to 616 BULLETIN OF MARINE SCIENCE. VOL. 31. NO.3. 1981

100 " """ ", "...... · J' "-- ... ,," ~" iii'." ,, ,, "" . 10 " / "". ,,/ " " " . "" ,." " ,"

1 +---'T'""-r--'T'""'T'""_..,....,"T'""---r--or-_.__._ .•...•...•..••.• 1 10 100 MINUTES Figure II. Righting or turnover times for Mellita sexiesperforata inverted on the sea floor in area with low wave activity.

become fully buried (invisible) ranged from 2 to 10 min with a mean of 5.9 (SE ± 0.30). In some cases a small amount of sand passed through the lunules but most of the burrowing activity resulted from forward locomotion and the use of the anterior fringing spines to lift sand onto the upper surface. Another aspect of righting behavior is the capacity to correct orientation when buried in the inverted position, with the oral surface uppermost. In laboratory and field trials five normal M. sexiesperforata, buried inverted, emerged fully from the sand in less than 1 h. They first appeared after 25 to 35 min at an angle of approximately 30° from horizontal and by the time that they were half emerged, the angle increased to as much as 60°. Following this, they emerged fully and always fell, oral surface uppermost, into a horizontal position. Sphaeridia were removed from five sand dollars and in five control animals comparable wounds were made adjacent to the sphaeridial chambers (Fig. 10) but not actually touching them. Four days after operation the experimental animals behaved normally but burrowed slowly; the mock operated control animals appeared to be perfectly normal in all respects. In both groups the wounds were healthy (injured areas of M. sexiesperforata tend to go black then to lose spines when infected). When buried in inverted positions the operated animals failed to emerge in 24 h. After 24 h rest, the same animals behaved in exactly the same way and again failed to emerge. Mock-operated animals always emerged from the sand and although the sample size is small, appeared to do so in the same time as normal ones. TELFORD: SAND DOLLAR MORPHOLOGY 617

DISCUSSION

Hydrodynamic Role of Lunules.-Sand dollars are discoidal echinoids with the aboral surface slightly domed and the oral surface nearly flat or very slightly concave. In a current of water such shapes generate lift as a consequence of acceleration of fluid over the domed surface. Prandtl and Tietjens (1934) showed that the extent to which lift is generated is dependent upon the camber of the upper surface. The two species investigated here, E. parma and M. quinquiesperforata, differ in general geometry and in the presence of lunules in Mel/ita. Both features contribute to differences in their coefficients of lift. Compared to Echinarachnius, species of Mellita are thinner-edged, less domed, with greater camber anterior to the apex and less behind, whereas Echinarachnius is more nearly symmetrical. Finally, where Echinarachnius is almost perfectly flat on the oral surface, Mellita is distinctly, but slightly, concave. When the lunules were

filled with wax M. quinquiesperforata had an average CL of 0.060 which generates only 78% of the lift of E. parma: this is the difference due simply to geometry. When the lunules were open the lift produced by M. quinquiesperforata was only 52% of that of E. parma. The presence of lunules effectively reduces the lift produced by 33%. For an airfoil or hydrofoil to generate lift, fluid flow over its surface must remain attached without turbulence. The presence of the lunules could reduce lift either by interfering with attachment of flow or by bleeding off pressure from the under surface. Observations in the water tunnel showed that flow was attached over an extensive part of the surface and became more so as velocity was increased. Further, flow was attached over a greater part of the surface of Mellita than it was on Echinarachnius. This results from the formation of a relatively large standing vortex at the anterior margin of Echinarachnius which apparently inhibited early attachment of flow. The less steeply cambered posterior surface of Mellita encouraged the maintenance of attached flow, whereas separation occurred relatively early on Echinarachnius. The lunules of Mel/ita reduce lift by providing channels to bleed off excess pressure from the oral surface. This is shown by the clear flow of water, marked with fluorescein, through the lunules. The lunulate sand dollars, such as Mellita, Encope (Fig. 1) and, to a less extent, Astriclypeus, have shallow depressions on the oral surface leading directly to the lunules or ambital notches: these are clearly evident in the illustrations of Sei- lacher (1979), Durham (1966) and very striking in that shown by Goodbody (1960). Goodbody is the only investigator to propose a function for these depressions in M. sexiesperforata. where he suggests that they are "food tracts." Bell and Frey (1969) observed that they played no role in feeding of M. quinquiesperforata. The depressions facilitate the flow of water to the lunules or notches when excess pressure occurs on the oral surface. Examination of the scanning electron micrographs (Figs. 8, 9, 10) show that the narrow, well-defined food grooves are equipped with podia which move food streams towards the mouth. The podia occur throughout the entire length of the food groove systems in Echinarachnius, Dendraster, Encope and Mellita (Phe- lan, 1977). Their apertures are oval and elongated in the direction in which the podia swing whilst moving food. The food grooves are flanked by fields of non- ambulatory spines which, according to Seilacher (1979), serve to turn over surface material in the search for food. Between the spines there are numerous small podia with apertures surrounding the spine bases. These podia sweep food particles towards the food grooves (Fig. 8) in the direction of elongation of their pores. The shallow depressions, which from here on will be called pressure drain- 618 BULLETIN OF MARINE SCIENCE. VOL. 31. NO.3. 1981

age channels, have no associated podia (Fig. 8), as previously noted by Bell and Frey (1969) and illustrated by Durham (1966). Further, there are no food collecting or sweeping podia immediately adjacent to them. The feeding podia, where they occur, sweep away from, or parallel to, the pressure drainage channels, but not into them. Goodbody's (1960) figure is particularly striking: the specimen of M. sexiesperforata was stained in toluidine blue which is taken up by the podia and mucus secreting glands. The pressure drainage channels are devoid of stain and, contrary to Goodbody's observation, the channels do not extend to the mouth, but terminate close to the bifurcation of the food grooves (Fig. 9). The podia are confined to the oral surface and do not emerge from the walls of the lunules (Fig. 8). Encope emarginata has well developed pressure drainage channels (Fig. I) leading to the ambital notches and a relatively broad anal lunule. This is very similar to the pressure drainage system and podia distribution shown by Durham (1966) in E. grandis. Flume observations show that a flow emerges from the notches and lunule and that the main flow is strongly attached over the anterior surface up to the region of the anal lunule. With its long, shallow sloping anterior region and the truncate, steeply sloped posterior, this species seems to be stream- lined especially for unidirectional flow. Some Encope species are intertidal in their distribution and it is possible that their shape is specially adapted to this habitat. Dexter (1977) and Ebert and Dexter (1975) have reported on intertidal populations of some Encope species but, unfortunately, they give no data on the prevailing orientation of the animals vis a vis current flow. That sand dollars might be subject to hydrodynamic lift was first suggested by the observations of Chia (1973) concerning the "weight belts" of D. excentricus. Recently O'Neill (1978) has shown how Dendraster uses its lift characteristics in feeding. Her calculated coefficient of lift for D. excentricus was very high (0.494) because in its upright posture the "reflection" phenomenon due to the presence of the ground effectively doubles lift. Accordingly, in the horizontal position used in this study, the expected coefficient of lift for D. excentricus would be 0.247. In comparison with the measurements made here, this still seems to be high. It is possibly due to the greater camber of the test of Dendraster and perhaps to O'Neill's theoretical idealization in calculating the zero angle of attack. The ground effect has not been taken into account in the current investigation beyond observing that in the horizontal position oral surface pressure did not differ from ambient pressure outside the wind stream, and that this was also true for a simulated sand dollar on a flat board. This suggests that the observed pressure differences between the upper and lower surfaces would probably occur under natural conditions. The form of the standing vortex at the anterior margin of E. parma might well be due to its proximity to the ground. Another factor which would affect flow and lift is the presence of spines. According to Prandtl and Tietjens (1934) the presence of spines should serve to turbulate the boundary layer, reducing its thickness and steepening its velocity profile. This is analogous to the practice of model aeroplane makers who roughen the leading edges of the wings in order to increase lift or to that of golf ball manufacturers who dimple the surface to decrease separation and hence drag, during flight. The Plexiglas model of E. parma was completely smooth and produced lower pressure differ- entials than those generated by real sand dollar tests. The calculation of current velocities required to lift a sand dollar from the substrate shows that under experimental conditions small individuals are at greater risk than large ones and that Echinarachnius is at greater risk than Mellita (Fig. 7). TELFORD: SAND DOLLAR MORPHOLOGY 619

Do sand dollars ever experience lifting velocities on the sea floor? A 1 cm M. quinquiesperforata would be lifted in a current of about 40 cm' sec-I and a 7 cm individual at 90 cm· sec-I: for similar sized specimens of E. parma lifting velocities would be 22 and 77 cm' sec-I, respectively. Data on the distribution of different sized sand dollars and current velocities are scarce. Salsman and Tolbert (1965) have reported observations on M. quinquiesperforata in the Gulf of Mexico where " ... currents are predominantly long shore, seldom exceeding 0.5 knot (17 cm' sec-I)." It is implied by their discussion of sand ripples that this is the sea floor current. This flow would be sufficient to lift a newly settled juvenile M. quinquiesperforata during the period before development of the lunules. Palmer and Wilson (1975) have reported currents of 0-85 em' sec-Ion dune crests, with flow exceeding 21 em' sec-I for 58% of the time; Caston (1976) cited values up to 100 em' sec-I "near" the sea floor and Murray (1970) found fluctuating bottom flows with peaks of 160 cm' sec-I and possibly higher during a hurricane. The fluctuating flow associated with wave action is difficult to ascertain but is probably significant in shallow water environments. Bottom currents may often be strong enough to lift moderately sized or even quite large sand dollars. This might be of interest in interpreting the fossil record. The camber of D. excentricus varies with the flow regine in which they live (O'Neill, 1978) and Stanton et al. (1979) have used morphometric differences of fossil Dendraster for the analysis of possible current regimes in ancient marine deposits. A knowledge of the hydrodynamic tolerances of sand dollar species could assist interpretation of other fossil deposits. The results shown in Figure 7 suggest that lunulate sand dollars should be able to tolerate higher energy environments than those without. The very high coefficient of lift calculated by O'Neill (1978) for Dendraster would make it extremely vulnerable at relatively low rates of flow. Seilacher (1979), suggested that the modern, lightly domed discoidal form of sand dollars has evolved as an adaptation to active burrowing and was derived from a more globose ancestral type which lived within a tunnel. The shape of the modern sand dollar minimizes locomotory effort and hydrodynamic drag when exposed on the surface. According to this view the phenomenon of lift should be regarded as a negative, undesirable factor. O'Neill's (1978) analysis suggests that evolution of this shape might have had much more positive impact: enhancing feeding activity by streamlining particulate matter over the surface and facilitating ventilation of the respiratory podia. For the three species examined in the water tunnel (E. parma, M. sexiesperforata and E. emarginata), the flow was clearly attached over the petalloid areas which are themselves slightly raised and have steeper cambers than the remainder of the test. Similarly, in Clypeaster, the petalloids are distinctly elevated. However, the generation of hydrodynamic lift will also have detrimental effects. Under severe wave action M. sexiesperforata has been seen to invert (personal observation) and D. exentricus occasionally gets swept away (O'Neill, pers. comm.). Lunules and notches allow sand dollars to survive in potentially lift-generating patterns of flow with less than the otherwise expected hazard of lift. Sand dollars without these features must resort to behavioral modifications, for example, the accumulation of weight belts (Chia, 1973; Seilacher, 1979); deeper burrowing and orienting towards waves and possibly also by the adjustment of their angle of attack to decrease lift. Some species are restricted in range for at least part of their life time and this may also be a response to lift-related problems of survival. For example, individuals of E. parma only occur subtidally as juveniles and enter shallower water as they grow, whereas M. sexiesperforata are found in very shallow water as juveniles and move into deeper water as they age. In this species, the juveniles (5 mm) which 620 BULLETIN OF MARINE SCIENCE. VOL. 31. NO.3, 198\ only have a single anal lunule burrow much more rapidly and steeply than the adults and remain buried all of the time. Alternative Explanations of Lunule Function.-FEEDING. Reference has already been made to Goodbody's (1960) suggestion that the lunules shorten the pathway of food particles moving towards the mouth. His explanation of feeding partly relies on the role of ciliary currents in moving food particles into the food grooves. As Hyman (1958) suspected, the podia playa more important role in moving mucus-bound food (see also Phelan, 1977; O'Neill, 1978; Timko, 1975) and, further, she thought that ciliary currents might not be strong enough to move the food unassisted. Bell and Frey (1969) noted that the pressure drainage channels in M. quinquiesperforata are not involved in feeding and they make no reference to the lunules in this connection. Only a small fraction of the food actually passes through the lunules en route to the mouth and the food grooves do not enter the lunules proper. It is reasonable to suppose that the lunules assist food movement although this is probably not their primary function. Seilacher (1979) raised an alternative feeding hypothesis: that the lunules rep- resent an allometric increase in edge structure and hence of frill spines. He be- lieves that the frill spines are used in sieving or stirring sand when held upright. Seilacher comments on the location of these extra frill spines "Also significant is the concentration of notches and lunules near the trailing posterior edge, where their sieving interferes least with burrowing." This is not entirely clear: resistance to burrowing would depend largely on the area of spines presented, be they anterior or posterior. The advantage of sieving would be to increase the rain of small particles onto the aboral surface where they could be further sorted by the shoe spines (Fig. 10). Placement of the extra spines at the posterior margin would tend to deposit this filtrate in the wake of the animal. If this hypothesis is correct, the additional spines should be located more towards the anterior, as they apparently were in the fossil Scutaster. BURROWING.Several authors (Berrill and Berrill, 1957) have suggested that lunules assist burrowing by allowing the upward movement of sand. There are various opinions concerning the passage of sand through the lunules. Telford (1978) found that this occurred occasionally in M. sexiesperforata and it has also been noted in M. quinquiesperforata (Bell and Frey, 1969) and Astriclypeus (Ikeda, 1941). However, Weihe and Gray (1968) and Hyman (1958) say that M. quinquiesperforata never passes sand through the lunules during burrowing. Per- sonal observation indicates that M. sexiesperforata can burrow either with or without passing sand through the lunules: it appears to be of very little significance. RIGHTING.Ikeda (1941) was of the opinion that the function of the lunules in Astriclypeus manni was to assist righting by permitting passage of sand from aboral to oral surface whilst moving to an upright position. Species of Mellita which have this kind of righting ability, such as M. quinquiesperforata (Weihe and Gray, 1968) and M. lata (Kenk, 1944), apparently do not use the lunules in this way. The present investigation indicates that M. sexiesperforata lacks in- dependent righting ability and must wait passively for wave action to turn it over. If Astriclypeus does derive some assistance from the lunules during righting it still fails to explain the presence of lunules in other species or genera which do not. The study shows that the sphaeridia of M. sexiesperforata contribute to righting from the buried position: presumably, as suggested by Parker (1927), they are balance organs. TELFORD: SAND DOLLAR MORPHOLOGY 621

STRUCTURAL REINFORCEMENT OF THE TEST. Berrill and Berrill (1957) suggested as another possibility that the lunules might be viewed from inside the animal as hollow columns supporting the upper, domed surface and increasing the strength/weight relationship of lunulate species. However, this seems unlike- Iy, as Hyman (1958) noted, because clypeasteroids are well provided with internal supports and further, the theory of crack propagation (Gordon, 1978) suggests that such apertures would weaken, not strengthen the structure. When exposed to a bending stress, lunulate sand dollars almost always break through the lunules, which appear to weaken the test as predicted. The hydrodynamic interpretation of sand dollar lunules and ambital notches provides a general explanation of their role, whereas previous suggestions have been either narrowly applicable to a single species or were otherwise untenable. There is a paucity of information on the precise distribution, orientation and behavior of various species in different current regimes. It is hoped that the hypothesis presented here will stimulate researchers to collect such data accu- rately for sand dollars and other types of animals living under comparable conditions where hydrodynamic lift might be an important environmental factor.

ACKNOWLEDGEMENTS

This work was supported by the Natural Sciences and Engineering Research Council of Canada through Operating Grant #A 4696. Thanks are extended to Mr. G. Dobrodzicki of the National Aeronautical Establishment, Ottawa, for his assistance in making the water tunnel observations and to Mr. E. Lin, Department of Zoology, University of Toronto, for his technical assistance with scanning electron microscopy. The author is indebted to Dr. J. DeLaurier of the Institute for Aero- space Studies, University of Toronto, for his helpful and critical comments as the work progressed. Use of the facilities of Bellairs Research Institute, Barbados and of the Huntsman Marine Laboratory, St. Andrews, N .B. is gratefully acknowledged.

LITERATURE CITED

Arnold, G. P., and D. Weihs. 1978. The hydrodynamics of rheotaxis in the plaice (Pleurollectes pill/essa L.). J. Exp. BioI. 74: ]47-]69. Bell, B. M., and R. W. Frey. 1969. Observations on ecology and the feeding and burrowing mech- anisms of Mel/i/ll qllillquiesperfora/a (Leske). J. Paleontol. 43: 553-560. Berrill, N. J., and J. Berrill. 1957. 1001 questions answered about the sea. Dodd, Mead and Co., N.Y. 305 pp. Caston, V. N. D. 1976. A wind driven near-bottom current in the southern North Sea. Estu. Coastal Mar. Sci. 4: 23-32. Chia, Fu-Shiang. 1973. Sand dollar: a weight belt for the juvenile. Science 181: 73-74. Dawson, R. M. C., D. C. Elliott, W. H. Elliott, and K. M. Jones. 1959. Data for biochemical research. Oxford University Press. 299 pp. Dexter, D. M. 1977. A natural history of the sand dollar Ellcope stokesi L. Agassiz in Panama. Bull. Mar. Sci. 27: 544-551. Dobrodzicki, G. A. 1972. Flow visualization in the National Aeronautical Establishment's water tunnel. National Research Council of Canada Aeronautical Report LR-557, Ottawa. Durham, J. W. 1966. Clypeasteroids. Pages 450-491 ill R. C. Moore, ed. Treatise on invertebrate paleontology, Part U. Echinodermata 3. Geol. Soc. Am. and University of Kansas Press, Law- rence, Kansas. Ebert, T. A., and D. M. Dexter. 1975. A natural history study of Encope gralldi.l· and Mel/ita grall/i. two sand dollars in the northern Gulf of California, Mexico. Mar. BioI. 32: 397-407. Goodbody, I. 1960. The feeding mechanism in the sand dollar Mel/i/ll .I'exie.l'perforata (Leske). BioI. Bull. 119: 80-86. Gordon, J. E. 1978. Structures: or why things don't fall down. Penguin Books Ltd., Harmondsworth, Middx., England. 395 pp. Hyman, L. 1955. The invertebrates 4: Echinodermata. McGraw-Hili Co., N.Y. 763 pp. --. ]958. Notes on the biology of the five-Iunuled sand dollar. BioI. Bull. 1]4: 54-56. Ikeda, H. 1941. Functions of the lunules of A.I'tric/ypelis as observed in the righting movement (Echinoidea). Annot. Zoo I. Japon. 20: 79-83. 622 BULLETINOFMARINESCIENCE.VOL.31,NO.3. 198\

Kenk, R. 1944. Ecological observations on two Puerto-Rican echinoderms, Mellita lata and Astro- pecten marginatus. BioI. Bull. 87: 177-187. Murray, S. P. 1970. Bottom currents near the coast during hurricane Camille. J. Geophys. Res. 75: 4579-4582, O'Neill, P. 1978. Hydrodynamic analysis of feeding in sand dollars. Oecologia 34: 157-174. Palmer, H. D. and D. G. Wilson. 1975. Nearshore current regimes in a linear shoal field, Middle Atlantic Bight, U.S.A. Congres Int. de Sedimentologie 9: 137-141. Parker, G. H. 1927. Locomotion and righting movements in echinoderms, especially in Echinar- achnius. Am. J. Psych. 39: 167-180. Phelan, T. F. 1977. Comments on the water vascular system, food grooves and ancestry of the clypeasteroid echinoids. Bull. Mar. Sci. 27: 400-422. Prandtl, T. F., and O. G. Tietjens. 1934. Applied hydro- and aeromechanics. Dover Publications Inc., N.Y. (1957). 311 pp. Salsman, G. G., and W. H. Tolbert. 1965. Observations on the sand dollar Mel/ita quinquiesperforata. Limnol. Oceanogr. 10: 152-155. Seilacher, A. 1979. Constructional morphology of sand dollars. Paleobiol. 5: 191-221. Stanley, D. J., and N. P. James. 1971. Distribution of Echinllrachnius parma (Lamarck) and associated fauna on Sable Island Bank, Southeast Canada. Smithsonian Contributions Earth Sciences 6: 1-24. Stanley, J. 1960. Lucid biometrics. McGill University Press, Montreal. 107 pp. Stanton, R. J., J. R. Dodd, and R. R. Alexander. 1979. Eccentricity in the clypeasteroid echinoid Dendraster: environmental significance and application in Pliocene paleoecology. Lethaia 12: 75-87. Telford, M. 1978. Distribution of two species of DissodactyJus (Brachyura: Pinnotheridae) among their echinoid host populations in Barbados. Bull. Mar. Sci. 28: 65]-658. Timko, P. 1975. Sand dollars as suspension feeders: a new description of feeding in Dendraster excentricus. Biol. Bull. 151: 247-259. Weihe, S. C. and J. E. Gray. 1968. Observations on the biology of the sand dollar Mel/ita quin- qlliesperforatll (Leske). J. E]isha Mitchell Sci. Soc. 84: 315-327.

DATEACCEPTED: March 16, 1981.

ADDRESS: Department of Zoology. University of Toronto, Ontario, Canada. M5S IAI