Compositional and Taphonomic Variations in Modern Crinoid-Rich Sediments from the Deep- Water Margin of a Carbonate Bank Ghislaine Llewellyn Harvard University

Nova Southeastern University NSUWorks

Marine & Environmental Sciences Faculty Articles Department of Marine and Environmental Sciences

12-1-1993 Compositional and Taphonomic Variations in Modern Crinoid-Rich Sediments from the Deep- Water Margin of a Carbonate Bank Ghislaine Llewellyn Harvard University

Charles G. Messing Nova Southeastern University, [email protected] Find out more information about Nova Southeastern University and the Halmos College of Natural Sciences and Oceanography.

Follow this and additional works at: https://nsuworks.nova.edu/occ_facarticles Part of the Marine Biology Commons, and the Oceanography and Atmospheric Sciences and Meteorology Commons

Recommended Citation Llewellyn, Ghislaine, and Charles G. Messing. "Compositional and taphonomic variations in modern crinoid-rich sediments from the deep-water margin of a carbonate bank." Palaios (1993): 554-573.

This Article is brought to you for free and open access by the Department of Marine and Environmental Sciences at NSUWorks. It has been accepted for inclusion in Marine & Environmental Sciences Faculty Articles by an authorized administrator of NSUWorks. For more information, please contact [email protected]. 554 RESEARCH REPORTS

Compositionaland TaphonomicVariations in Modern Crinoid-RichSediments from the Deep-Water Marginof a CarbonateBank

GHISLAINE LLEWELLYN

Departmentof Earth and Planetary Sciences,Harvard University,Cambridge, MA 02138 CHARLES G. MESSING

Nova OceanographicCenter, 8000 North Ocean Drive,Dania, FL 33004

PALAIOS, 1993, V. 8, p. 554-573 be identifiedin deep-watercarbonate bank-margin sediments.Such changesare largelya responseto differences Multivariateanalyses ofthe coarse-grainedfraction (>2 in benthicflow regime associated withsmall-scale topo- mm) of sedimentsaccumulating in deep water (419-434 graphicirregularities and may providean importantdi- m) along the westernmargin of the Little Bahama Bank agnostictool for the interpretationof fossil assemblages. reveal identifiable,small-scale compositional and taphonomic variationsamong local subhabitats (ridge crest, slope, foreslope,base of slope, pavementsand scour pit) INTRODUCTION separated by metersto tens of meters.Bulk composition variesbetween planktic- (crest and slope) and lithic-dom- Subfossilskeletal remainsof benthicmarine inverte- inated (pavements,scour pit) sediments.Local macro- bratesare widelyused to diagnosetaphonomic signatures benthicskeletal components also varysignificantly among and revealchanges in communitystructure over a variety subhabitats,but are commonlydominated by echinoid of spatial and temporalscales. Such workprovides impor- and crinoidmaterial; crinoid columnals contribute 9-52% tantinformation about the transitionbetween living com- of the coarse skeletal componentof 17 sedimentsamples munitiesand fossil assemblages (Cumminset al., 1986; considered. Distributional and taphonomic analyses Hendersonand Frey,1986; Firsich and Flessa, 1987;Kid- (abrasion, encrustation,breakage) indicate that colum- well,1988; Miller,1988, 1989; Davies et al., 1989;Parsons, nals produced in dense ridge-crestassemblages of Chla- 1989; Staffand Powell, 1990a; Milleret al., 1992). Similar docrinusdecorus are transporteddown and accumulate studiesusing fossil material contribute to the reconstruc- along an adjacent slope. Sediments fromhardgrounds tion of ancientcommunities and vastlyimprove our un- supportingscattered living assemblages show columnals derstandingof paleoenvironments (Brett and Baird, 1986; with the highestlevels of abrasion,implying prolonged Norris,1986; Speyer and Brett,1986, 1988; Parsonset al., local reworking.Elevated contributionsof Endoxocrinus 1988;Meyer et al., 1989;Kidwell and Behrensmeyer,1988). parrae columnalsto the fewsubhabitats where this spe- Althoughexceptions exist [e.g.,Mullins et al., (1981) on cies dominatesthe livingassemblage suggest limited lat- the recognitionof deep-waterfossil coral bioherms],the eral transportin the absence of steep gradients.High greatmajority of thisliterature treats material from shal- levelsof biologicalencrustation in areas ofthin sediment low-waterenvironments. coversuggest control by lengthof exposureof grains at Stalked crinoidswere important components of shallow thesediment-water interface. Lack ofany correlationbe- marineenvironments during much of the Paleozoic and tweenfrequency of broken columnals in samples and any Mesozoic and have successfullybeen used as taphonomic observedsedimentary or environmentalparameters sug- indicatorsin ancientcrinoid-rich limestones (Blyth Cain, gests the action ofpredators or scavengersin this deep- 1968;Ruhrmann, 1971a, b; Meyeret al., 1989,Meyer, 1990; watersetting. Donovan, 1991; Baumillerand Ausich,1992). As a result Small-scale variationsin sedimentcomposition, ben- ofthe Mesozoic marine revolution, however (Vermeij, 1977), thic skeletal assemblages,and taphonomiccharacteris- stalked crinoidsmoved into deep water,leaving shallow ticsare notunique to shallow-watersettings, but can also marine habitats to unstalked comatulidcrinoids which

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions MODERN CRINOID-RICHSEDIMENTS 555

x . I : : I ' have since diversifiedsubstantially (Meyer and Macurda, A o80- 79" .78' B North Pavement Stalked crinoidshave remained Little Bahama Bank 1977). importantcompo- 27'N . nents of a varietyof deep-waterhabitats (Roux, 1980; S traits of Conan et et Florida | al., 1981; Messing,1985; Messing al., 1990), STUDY SITE* .. but because are moststudies .. ahama they relativelyinaccessible, Florida Island of moderncrinoidal skeletal materialhave concentrated NW Providence on shallow-watercomatulids. Although these unstalked '"' Channel 26" .... Great crinoidsare sometimes abundantand have suc- Miami extremely Bahama been used in biostratinomicand Bimini.. Bank Base of Slope cessfully taphonomic - studies (Meyer, 1971; Liddell, 1975; Lewis and Peebles, 2- 428 1988, 1989), theymake onlyminor contributions to local _ Slope- sediments(Meyer and Meyer,1986; Lewis et al., 1990). ) Limitedtaphonomic and compositionalinformation on l~?'';-T--7~,~6~e modernstalked crinoid skeletal material does exist.In two ""¥o__,'! /: Ii I1 '* Foreslope / sedimentsamples collectedvia submersiblein the north- I ' \ ' e, .PavementI - western (1985) notes that ------Bahamas, Messing (chiefly ,~ , \ -I'wrScourPit crinoidossicles contribute count stalked) 11.6% by grain \ lee _ to a sand fromlithoherm rubble - (7.8% byweight) gravelly _ / / SoutJhwest Margin , (550 m) whereliving crinoids are fewin number,and 8.8% I byweight of a chieflyforaminifer/thecosome sand adjacent to a pavementwith scattered isocrinids (300-400 m). Ame- 7 ziane-Cominardiand Roux (1987) discuss bacterialand lpv fungalbiocorrosion of a varietyof stalkedcrinoid ossicles recoveredfrom cores taken offNew Caledonia (735-1285 FIGURE1 -Map ofstudy area. A-Geographiclocation of study site. m). Althoughcrinoid ossicles make up fewerthan 10% of B-Detailed map of studysite. Dashed linesare isobathsin meters bioclasts,they distinguish autochthonous and allochtho- estimatedfrom individual depth measurements taken during sub- nous components.Ameziane-Cominardi's (1991) studyof mersibledives. Area covered is approximately100 m across. substantialnumbers of ossicles also takenfrom New Cale- donian cores includes morphologicaland distributional analyses,a comparisonof expected and actual proportions theLittle Bahama Bank in an area characterizedby locally of differentskeletal components, and a discussionof hy- dense (to 12 individualsm-2) populationsof the isocrinids drodynamicand biocorrosionalprocesses as possibletaph- Endoxocrinusparrae (Gervais) and Chladocrinus (for- onomicagents. No directcomparisons are made between merlyNeocrinus) decorus (Carpenter).The results re- sedimentsand local livingcommunities, however, and the ported herein are part of a largerproject investigating relativecontributions of crinoidaland othersedimentary stalkedcrinoid biology and ecology,as wellas taphonomy, componentsare not discussed. and the possible use of modernstalked crinoidsas ana- This paperprovides the first detailed compositional and logues forfossil forms. taphonomicaccount of a modern,deep-water, carbonate sedimentrich in stalkedcrinoid material. It also provides PHYSICAL SETTING the firstcorrelation of a livingstalked crinoid assemblage withpenecontemporaneous crinoidal sediment where co- The studyarea is located on the southwesternslope of lumnalsalone contributea mean of 23% by graincount the Little Bahama Bank, 7.41 km south of Settlement to the coarse fraction(>2 mm) of benthicbioclasts. As in Point and 4.55 km offthe beach at the westernend of shallow-waterenvironments, both composition and tapho- Grand Bahama Island (26038'N lat., 78°59' W long.;Fig. nomic characterof accumulatingsediments reflect local 1A). The substrate consists of extensive,gently slop- topographicand habitatvariations, permitting recognition ing, submarine-cemented,carbonate pavements (hard- of environmentaldistinctions and taphofaciescharacter- grounds),interspersed with areas coveredby unconsoli- isticsin a relativelyhomogeneous, deep-water, carbonate dated sediment,chiefly pelagic foram/thecosome material. environment.Criteria are proposedfor recognizing similar Moderate-relief(to about 8 m) topographicirregularities environmentsin the fossilrecord. A deep-watermodel for include locally steeper slopes, ridges,escarpments and the formationof coarse-grained, crinoid-rich limestones is boulders.Larger features, perhaps to 30 m high,were de- also discussed. tectedin pre-divesonar transects,but werenot observed This studygrew out of trawlingsurveys (Meyer et al., fromthe submersible.Regional studies indicatethat un- 1978) and submersibleobservations (Neumann et al., 1977; consolidatedsediments are subjected to winnowingand Messing,1985; Messing et al., 1990) that revealeddense redistributionby contourcurrents (Mullins et al., 1980), assemblagesof stalked crinoids in 300-600m in the north- and are largelyunderlain by carbonate pavements (Wilbur, easternStraits of Florida under the influence of the Florida 1976).Comprehensive descriptions of regional geology, de- Current.In February1991, the authors established a study positionalenvironments and sedimentationin the north- site at about 430 m depth on the southwesternmargin of ernBahamas can be foundin Neumann(1974), Neumann

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions 556 LLEWELLYN& MESSING

TABLE1-Summary of subhabitat characteristics (see also Figure1B).

Ridgecrest (CRS) Easternportion of northern edge of hardground;419-420 m (about8 m abovesurrounding pavements);rippled, medium- to coarse-grained,well-sorted sand alongthe crestedge; re- mainingarea barrenof sediment; dense living crinoids almost exclusively Chladocrinus decorus. Foreslope(FSL) Gentlysloping, smooth hardground immediately south of ridge crest; 420-423 m; thinsedi- mentveneer; scattered crinoids including numerous Endoxocrinus parrae. Slope (SLP) Northflank of ridge crest; 421-428 m; thicksediment cover (>5 cm); sample2 takenfrom midslopearea (424 m), sample3 takenfrom closer to slopebase, sample 4 fromupper slope (423 m); virtuallyno crinoids. Base ofslope (BSL) Pittedhardground at base ofnorth slope; 429 m; sedimentaccumulation variable, thicker in pits;coarse thecosome lag visible;scattered crinoids. Southwestmargin (SWM) Gentlysloping northern edge of hardground at west-southwesternend ofstudy area (contin- uous withridge crest); 434 m; thinsediment veneer on hardground,thicker sediment north ofvisible edge of pavement; scattered crinoids. East pavement(EPV) Brokenhardground southeast of ridge crest and westof scour pit; 428-431 m; thinsediment veneerwith deeper pockets; scattered crinoids. Northpavement (NPV) Pittedhardgrounds north of slope base; 429 m; thickersediment than on East Pavement site;scattered crinoids. Scourpit (SCP) Crescenticdepression at footof a largeboulder southeast of ridge crest; 431 m; muchrubble and coarse-grainedmaterial apparently from break-up of adjacent hardground; dense E. parrae.

et al. (1977), Mullinsand Neumann(1979), Mullinset al. by local topography,reached perhaps 50 cm sec-1 within (1980, 1981) and Hine and Mullins (1983). a meter of the substrate.Time-lapse camera exposures The studysite spans an area approximately100 m (east/ (Messing and Llewellyn,unpubl.) reveal a completecur- west) by 150 m (north/south)centered on a sediment- rentreversal within 48 hr of the initialobservations. Dis- veneered,low-relief pavement that slopes downwardin all location of deployedequipment between successive dive directionsfrom its northeasternapex (419-420 m) to 434 seriessuggests that muchstronger currents may existin- m in the southwestand west (and deeperbeyond the area termittently.Such flowvariations probably contribute to studied),and to 429-431m to thenorth, east and southeast variationsin sedimentcomposition described below. (Fig. 1B). The easternportion of the northermargin of this hardgroundforms a distinctridge with a rounded METHODS northeasterncorner and a steep sediment-coveredslope to thenorth. Removal of sediment between successive dive Sedimentsamples were collectedin February,August operationshas exposed a near verticalscarp up to 0.5 m and November1991 using a suctiondevice mountedon highin someplaces alongthe pavement margin. The ridge/ the submersible Johnson-Sea Link (Harbor Branch pavementmargin slopes down to the west and becomes OceanographicInstitute, Fort Pierce,Florida). Relatively even withthe sediment.Approximately 30 m southeastof thin sedimentlayers underlainby cementedpavements the pavement/ridgeapex, lies a large boulder (approxi- precludedcoring. Substantial amounts of fine material were mately2 m across and 1.5 m high).A crescenticscour pit lost duringthe collectionprocess althoughlarge enough runsaround its southern half while its northern flank slopes volumes(sometimes several liters) of sedimentwere usu- intoa flattenedshoulder-apparently a lithifiedsediment allyretained for analyses. Sampling was precludedon some shadow.Pavement east and southeastof the boulderex- ridgecrest and foreslopeareas because of insufficientsed- hibitsconsiderable erosion into loose slabs and rubble. imentcover. For the followinganalyses, subhabitats are Currentflow at the studysite is generallynorthbound distinguishedon the basis of location,depth, topography, and underthe influenceof the Florida Current.However, substrateand macroepibenthosvisible from the submers- duringthe course of this study importantvariations in ible, and are describedin Table 1. velocityand directionwere observed, perhaps due to tidal Samples were fixedin 70% ethanolimmediately upon influences.Velocities (measured approximately 2 m above surfacing(to preserveany living organisms),and split, the substrate)typically range from less than 5 to about 25 driedand sievedthrough a standard2 mmmesh screen in cm sec-', but, duringinitial operations in February1991, the laboratory.For compositionalanalyses, all grains>2 unexpectedstrong flow from the northbent crinoidsover mmand recognizableas bioclastsare identifiedat least to and produceda narrowband of active asymmetriclinear major taxonomicand ecologic (forexample, planktic vs. rippleson the ridgecrest. This flow,probably intensified benthic)group, with non-biogenic clasts treatedas lithic

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions MODERN CRINOID-RICHSEDIMENTS 557

TABLE 2-Abrasion scale forcrinoid columnals.

Abra- sion scale Diagnosticfeatures

I Virtuallypristine: very minor abrasion 2 Minorabrasion of crenelationsand edges 3 Crenelationsreduced in height:abrasion of edges 4 Crenelationsindistinct: extensive edge rounding 5 Crenelationscompletely removed: edges rounded

grains.Samples mustinclude an adequate coarse-grained fraction(at least 50 grains)and a preciselocation within the studyarea. A total of sixteensamples meet both of these criteriaand are included in subsequenttests. The onlyexception is a singlesample of fine-grained,actively rippledsand fromthe ridge crest (sample la) thatcontains too fewnon-crinoid grains >2 mm to permitinclusion in compositionalanalyses. It does, however,have sufficient crinoidgrains and is included in the taphonomicstudy describedbelow. We use principalcomponents analysis and clusteranalysis to examine relationshipsamong sub-habitats,gross sedimentcomposition (planktic/benthic/lithic) and benthic skeletal assemblages(css: Statistica,Complete Sta- tisticalSystem, 1986-1991 Statsoft, Inc.). Clusteranalysis organizesdata into clustersaccording to degreeof simi- larity;principal components analysis creates new axes that bestexplain variation within the dataset. These techniques have been successfullyused witha varietyof biological, FIGURE2-Scanning electronmicrographs (SEMs) ofcolumnals from paleontologicaland sedimentologicaldata sets (Reyment, Chladocrinusdecorus illustrating range of abrasionvalues. A-Pris- Wilkinsonand et tine;taken from a preservedspecimen (all organictissue removed), 1963; Cheshire,1989; Meyer al., 1989; x22.7. ranked1 on abrasionscale. Brokensur- Valentineand and B-Virtuallypristine, Miller,1988; Peddicord,1967; Colby faceon frontmargin results from disarticulation oftwo partially fused Boardman, 1989; Doyle and Feldhausen, 1981). In our columnals,not produced by abrasion, x 19.8. C-Ranked 2 on abra- analyses,samples are percent-transformedto account for sion scale, slightabrasion of crenelations,x19.9. D-Ranked 3 on differencesin sample size. In analysesof grosssediment abrasionscale, moderateabrasion of bothcrenelations and margin, composition,only variables present in at least 75 % ofsam- x22.0. E-Ranked 4 on abrasionscale, extensiveabrasion of both columnalface and x18.8. F-Ranked 5 on abrasion ples are includedin orderto producea reduceddata set margin, scale, extremelyabraded columnal, surface features barely visible, x 19.7. containingonly the more common variables. All taxonomic categoriesare includedin analysesof benthicskeletal assemblages.Cluster analysis uses the UnweightedPair- forChladocrinus decorus with an ossicledissociated from GroupMethod with Arithmetic Averaging (UPGMA; Im- a livingspecimen for comparison. Between-sample com- brie and Van Andel,1964; Harbaughand Merriam,1968; parisons(described below) use mean values derivedfrom Sneath and Sokal, 1973) withan averageEuclidean dis- all grainsmeasured in a sample. Biological encrustation tance similaritymeasure and singlelinkage joining algo- refersto any depositon a columnalproduced by another rithm(see, forexample, Martin and Wright,1988). Prin- organism(that is, not includingmineral cements); tapho- cipal componentsanalysis uses an average Euclidean nomicbreakage includes any damage to a columnalsurface distancemeasure with a raw varimaxrotation on the ap- notobviously produced by abrasivewear (Fig. 3). Encrust- propriatenumber of factors(Scree test: Cattell,1966) of ation and breakage are assessed on a presence/absence the correlationmatrix. basis per grainand expressedas the percentageof grains In this paper, analysisof taphonomicalteration treats exhibitingany encrustationor breakageper sample. abrasion,biological encrustation and breakageof stalked Ossicle coloralso exhibitsconsiderable variation. Fresh crinoidcolumnals. Abrasion refers to the amountof deg- columnalsare pure whitewhereas many apparently older, radation of articularfacet sculpturingand roundingof typicallymore abraded ossiclesdisplay a rangeof discol- ossicleedges measured on an arbitraryscale of1 to 5 (Table orationfrom yellow through beige to various shades of 2). Figure2 illustratesan exampleof each levelof abrasion brown,with occasional iron oxide staining.Color may also

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions 558 LLEWELLYN& MESSING

ples are combinedas "Slope"; northand east pavement, scour-pitand southwestmargin samples are combinedas "Hardground."The foreslopesample has some aspects in commonwith both "Slope" and "Hardground"categories. It possessesa typicalhardground aspect, supporting a typical hardgroundfauna yet is tilted up towardsthe ridge and is located only a few metersaway fromthe crest. Analysesare thereforerepeated including foreslope in both categoriesto detectany differences.The two ridgecrest/ rippledsand samples are sufficientlydistinct in taphonomic and sedimentarycharacteristics to merittheir own categoryin both habitatand sedimentclassifications. Differencesin crenelationpatterns on columnalarticular facespermit easy distinctionof the two local species.Fig- ure 3A illustratesa columnaldissociated from a livingE. parrae forcomparison with the pristineChladocrinus decorus shownin Figure2A. C. decoruscolumnals are suf- ficientlyabundant to be analyzed as a separate group, allowingcomparison between taphonomic characteristics of a bulk columnalsample withthose of a singlespecies. Recoveryof too fewEndoxocrinus parrae columnalspre- cluded a similarcomparison. The relationshipbetween living and death assemblages of crinoidsis examinedby comparing1) local population densitiesof livingcrinoids with abundances of crinoidal materialin sedimentsamples, and 2) the relativecontri- FIGURE3-SEM ofA-pristine Endoxocrinus parrae columnal from butionsthe two species make to local livingpopulations a preservedspecimen (all organic tissue removed), x 16.0. B-Chla- and sedimentsamples. Densities and species composition docrinusdecorus encrusted by benthic foraminiferan. C-Endoxocri- ofliving crinoid assemblages are assessedfrom video taken nusparrae columnal showing breakage, x 21.0. D-Chladocrinusde- in the ofa theabundance and coruscolumnal showing taphonomic breakage, x 18.4 vicinity sample; composition of crinoidalmaterial in the benthicskeletal component of sedimentsamples characterizesthe death assemblages.A varyon a singleossicle. These variationshave provendif- univariateANOVA is used to analyzethe relationshipbe- ficultto quantifyobjectively and are, therefore,not in- tweenliving crinoid population density and abundanceof cluded in taphonomicanalyses. crinoidalmaterial in the sediment.Living crinoid density We analyzetaphonomic characteristics using Multi- and is theindependent variable; samples are classifiedas dense UnivariateAnalyses of Variance (MANOVA and ANO- (at least severalindividuals m-2), scattered (one or a few VA). These techniquestest the validityof groupingsby individualsper severalm2) or none. Relative abundances comparingdifferences in means, and are widelyused in of the two crinoidspecies in livingpopulations and sedi- analysesof biologicaland taphonomicdata (forexample, mentfrom the differenthabitats are comparedto inves- Wilkinsonand Evans, 1989;Wilkinson and Cheshire,1989; tigatewhether any correspondenceexists between species Davies et al., 1989; Staffand Powell, 1990b). Multi-Be- compositionin livingand death assemblages. tween-GroupsANOVA's and MANOVA's are used to de- termineif any significantinteractions exists in the data RESULTS set. Multi-wayinteractions revealed by theseanalyses are tested Wilks' Lambda multivariatecriteria. Three- using SedimentComposition Analysis way univariateANOVA's are used to investigatespecific effectsof individualindependent variables on dependent Figure 1 illustratesthe locationsof subhabitatsat the variables,testing individual interactions for significance studysite at whichsediment samples were collected. Table withNewman-Keuls post hoc tests (Winer,1971). Inde- 1 summarizesthe general features of sampled subhabitats. pendentvariables are habitatand sedimenttype; depen- Compositionalanalysis of >2 mmgrains (Table 3) reveals dent variables are abrasion,breakage and encrustation. threemain sourcesof coarse sediment:1) lithicgrains; 2) Sedimenttype is classifiedas thin(<2 cm surfaceveneer, testsof calcareous planktic/pelagic organisms (chiefly the- < 5 cm in pockets),thick (> 2 cm surfacelayer) or rippled. cosome "pteropods");and 3) calcareousskeletal material For taphonomicanalyses, several subhabitatsare com- producedby benthicinvertebrates. Crinoid and echinoid binedon the basis ofgeographic proximity and visualsim- grainsdominate the lattercomponent, and are accompa- ilarity,to obtaina moremanageable design (following, for nied by clasts produced by alcyonarians,corals, gastro- example,Meyer et al., 1990; Davies et al., 1989) and more pods, bivalves,benthic foraminiferans,decapod crusta- statisticallyrigorous results. Slope and base of slope sam- ceans and other minor components [stylasterids,

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions MODERN CRINOID-RICHSEDIMENTS 559

TABLE 3-Grain counts fromsediment samples taken at 419-434 m on the southwesternslope of the Little Bahama Bank. Samples labeled according to sample number(No.), monthof collectionand subhabitat(Subh.). See Table 1 forsubhabitat abbreviations. Grain composition abbreviations: PLG, plankticgastropod; PLF, planktic foraminiferan;CRI, crinoid;ECH, echinoid; OPH, ophiuroid;GST, benthicgastropod; BIV, bivalve; BNF, benthic foraminiferan;COR, coral; ALC, alcyonarian;STY, stylasterid;BCH, brachiopod;DEC, decapod crustacean; OST, ostracod; SPO, sponge; BRY, bryozoan;WOR, wormtube; LTH, lithic;UNI, unidentifiable; TOT, total.All sediment >2 mm.

No. MonthSubh. PLG PLF CRI ECH OPH GST BIV BNF COR ALC STY BCH DEC OST SPO BRY WOR UNI LTH TOT

1 Aug CRS 425 0 32 86 1 8 12 3 14 0 1 0 25 0 6 4 6 1 17 641 2 Aug SLP 172 0 275 121 4 103 24 91 26 45 17 7 43 0 10 3 0 19 364 1324 3 Oct SLP 378 0 134 94 1 33 13 13 4 6 2 4 44 0 27 0 3 0 36 792 4 Oct SLP 244 0 123 109 4 28 20 2 0 2 3 3 29 0 7 0 2 3 22 601 5 Oct FSL 205 1 75 124 2 44 18 10 22 46 12 13 24 4 44 4 5 0 141 794 6 Feb BSL 38 0 18 20 0 16 0 13 7 7 1 1 6 2 3 2 0 0 41 175 7 Oct BSL 258 0 112 183 7 39 33 5 8 20 0 3 64 7 29 0 3 6 99 876 8 Oct SWM 161 0 124 134 2 54 23 14 3 18 3 10 32 3 6 0 4 15 252 858 9 Oct SWM 126 0 63 114 0 42 26 22 0 9 1 7 14 2 5 1 1 1 196 630 10 Feb EPV 280 0 62 47 0 47 20 22 7 40 0 2 27 6 17 6 0 17 397 997 11 Aug EPV 443 0 51 112 6 65 15 71 16 96 7 12 60 0 11 1 1 11 1259 2237 12 Feb NPV 234 0 25 34 0 24 3 24 1 16 1 2 9 3 3 1 6 0 374 760 13 Feb NPV 645 0 121 105 0 88 13 71 11 77 0 16 79 7 7 4 0 48 1413 2705 14 Feb NPV 146 0 37 47 0 20 8 30 2 24 0 1 19 8 3 0 0 1 222 568 15 Oct NPV 158 0 116 44 0 33 9 69 8 1 3 8 33 11 14 3 2 1 449 962 16 Feb SCP 747 0 49 53 0 59 6 179 14 139 5 6 14 6 1 0 0 55 3281 4614

brachiopods,sponges (siliceous), bryozoans, brachiopods, spectively,and are used to carryout a varimaxrotation. serpulid polychaetesand ostracods].We findno coarse Considerationof eigenvaluesfor constituent grains shows grainsderived from shallow water. Dead sea grass (Thal- that axis 1 is governedprincipally by abundanceof lithic assia testudinum)blades occasionallyobserved on the grainswhile axis 2 is controlledmainly by amountof the- substrateindicate that off-bank transport may bring fine- cosome (planktic)material. Sample distributionwith re- grainedmaterial to the study site (Heath and Mullins, spect to the firsttwo rotatedaxes (Fig. 6) confirmsthe 1984; Hine et al., 1981), but the extentof such off-bank groupingsproduced by clusteranalysis. Samples, labeled transportcould not be assessed because muchof the finer accordingto clusteranalysis assignments, fall along a broad grain-sizefractions was not recovered. gradientfrom those rich in planktic(upper rightof Fig. Abundancesof the threemain sedimentcategories (> 2 6) to thoserich in lithicgrains (lower left). Group I, at top mm) varyamong subhabitats and are assessed usingmul- right,comprises the ridge crest and two slope samples; tivariatetechniques. A Q-mode clusteranalysis dendro- highin plankticand lowin lithicgrains. Group III, at lower gram distinguishesthree major sample clusters(Fig. 4). left,consists of hardground and scourpit samples,all with Samples fromindividual subhabitats group together with- abundant lithic,but few plankticgrains. The scour pit in clusterswith only one exception.Examination of abun- sample (16), at the extremelower left, contains the largest dance data,ordered according to resultsof two-way cluster lithic componentand smallest skeletal component.The analysis(Q and R mode),reveals the relationshipbetween remainingsamples (Group II) occupythe centralportion constituentgrains and clusterdesignations (Fig. 5). Rel- ofthe diagonaland displaycompositions intermediate be- ativeabundances of planktic (thecosome) and lithicgrains tweenthe othergroups. appear to controlthe threemajor groupings; benthic skeletal are ClusterI in- components relativelyunimportant. BenthicSkeletal AssemblageAnalyses cludes the crestand two slope samples,with at least 40% plankticgrains; Cluster III comprisesnorth and east pave- Crinoid and echinoid clasts dominatethe coarse (>2 ment (hardground)and scour pit samples,with at least mm) benthicskeletal component of most of the 16 sedi- 39% lithicgrains; and ClusterII containsthe foreslope, mentsamples (Table 4). Crinoidossicles (columnals) con- slope base, southwestmargin and one slope samples,with tribute9.2 to 37.1% bygrain count. Four samples are more intermediateabundances of plankticand lithicgrains. than30% crinoidal,five samples 20-29%, fivesamples 15- To testthe validityof these groupings, a principalcom- 19% and two samples 8-10%. Echinoidmaterial contrib- ponentsanalysis was performedon the same data set. The utes 10.0 to 43.4% and consistschiefly of plate fragments firsttwo axes explain 75% and 20% of the variance,re- of spatangoidechinoids. Secondary contributors include

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions 560 LLEWELLYN& MESSING

LinkageDistance (% max. dist.) Variables:R-Mode tM,Arlp, 0 20 40 60 80 100 ,(-iV'iV,1 PLG CRI ECH GST BIV COR BCH SPO DEC ALC BNF LTH 1 CRS 3 SLP I 3 @ . 4 - .. *.... 4 SLP 7 .0...... 5 * * * * * 5 FSL 7 BSL 2 SLP 8 0 6 BSL 8 SWM 1014 ...... · 9 SWM 12 O...... °' 10 EPV * 11 * . * ...... 0 II · ·...... - 14 NPV 13 * . . 15 * ...... 12 NPV 16 ...... * . 11 EPV % Abundance in sample 13 NPV 15 NPV 16 SCP 1-5 5-20 20-40 40-60 60-80 FIGURE5-Percent abundance forsediment composition. FIGURE4-Dendrogram for Q-mode cluster analysis of sediment diagram reduceddata set method Samplesand variablesgrouped according to Q- and R-modecluster compositionusing (unweightedpair-group Abbreviationsas inTable 2. witharithmetic averaging; percent transformation ofsamples; average analyses. Euclideandistance measure; single linkage). Large lettersindicate clusterdesignations discussed in text; numbers and labelsdesignate samplesand samplingenvironment. also lowerthan forbulk sediment.The R-mode dendrogram for benthic skeletal variables (Fig. 8) shows that crinoidsand echinoids,the two most abundant coarse- gastropodshells and fragments(8.0-16.7%), benthicfo- graintypes, are distributedlargely independently of each raminiferans(0.6-33.7%), decapod crustaceanfragments other.They link at 80% of the maximumdistance. The (2.6-13.2%) and alcyonarianfragments (0-26.2%). Ben- remainingtaxonomic groups cluster together with a high thic foraminiferansand alcyonariansdominate a single degreeof similarityrelative to echinoidand crinoidcom- sample (scour pit). Alcyonarianfragments consist chiefly ponents.Examination of linkagedistances, however, re- ofslender cylindrical axis segmentsproduced by the isidid veals that degreeof similarityis largelya functionof rel- Lepidisis caryophyllia.Other groups rarelycontribute ative abundance in samples. (CompareFig. 8 withTable morethan 7 % to any givensample. 4.) Crinoidsand echinoids,the most abundant,are least Multivariateanalysis of the benthicskeletal component similar.Decapod crustaceans,gastropods, benthic forami- of sedimentsamples revealsadditional variations associ- niferansand alcyonarians,the next most abundant groups, ated withenvironmental setting. A Q-mode dendrogram exhibitintermediate linkage distances. Finally, the least of all majortaxonomic groups (Fig. 7) producestwo broad abundant groups such as ophiuroids,stylasterids, bryo- clustersand an outlyingsample, labeled A, B, and C, re- zoans appear mostsimilar. spectively.Cluster A representscrest, slope, foreslope, base A principalcomponents analysis of the same data set of slope and southwestmargin subhabitats and one hard- confirmsthe patternof two broad subhabitatclusters and groundsample (15). ClusterB containsall remaininghard- an outlyingsample seen in the clusteranalysis. Inspection groundsamples and one slope base sample. The outlier, of eigenvaluesshows that the firsttwo axes explain 77% clusterC, is the scourpit sample.These groupingsare not and 12% of the variation,respectively. The firstaxis is as unequivocalas those obtainedin analysesof bulk sed- most highlycorrelated with abundance of echinoidand iment(Fig. 4); samplesfrom each subhabitatdo notcluster crinoidmaterial; the secondaxis dependsmore on gastro- togetheruniformly (for example,the two base of slope pod, benthicforaminiferan and alcyonarianfragments. A samplesfall into separate clusters). In addition,the greater plot of the firsttwo principalcomponents (Fig. 9) distin- relativelinkage distances between samples within clusters guishesthe followinggroups: group A at top rightcontains indicate that the degree of similarityamong samples is ridge crest,foreslope, southwest margin and most slope

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions MODERN CRINOID-RICHSEDIMENTS 561

common scale LinkageDistance (% max. dist.) 3.073 0 20 40 60 80 100 I I I I I I 2 SLP 15 NPV 5 FSL 1 CRS 7 BSL 9 SWM Axis 1 8 SWM 4 SLP 3 SLP III 6 BSL 12 NPV 14 NPV 10 EPV 108R7 -0.344 Axis 2 0.980 13 NPV 11 EPV FIGURE6-Plot ofaxes 1 and 2 forPrincipal Components Analysis of sedimentsamples using reduced data set (rawvarimax rotation 16 SCP and normalizedfactor loadings computed from 3 factorsof a correlationmatrix). Large letters refer to groupdesignations from cluster C analysis;small numbers refer to samplenumbers. See textfor discussionof meaningof axes. FIGURE7-Dendrogram for Q-mode cluster analysis of benthic skeletalmaterial (unweighted pair-group method with arithmetic averaging; transformationof Euclideandistance mea- and is in percent samples;average samples high echinoidand crinoid,low in gas- sure; singlelinkage). Large letters indicate cluster designations dis- tropod,alcyonarian and benthicforaminiferan clasts; group cussed intext; numbers and labelsdesignate samples and sampling C at bottomleft includes only the scourpit samplewhich environment. is dominatedby alcyonarianand benthicforaminiferan material;group B towardsthe center consists of most pave- and scatteredC. decorusaccompanied by isididalcyonari- mentsamples and has a compositionintermediate between ans on a sediment-veneeredpavement (Fig. 11B). The ridge the othertwo. Again, the groupingsare not perfect,as the crest and the isolated boulder adjacent to the scour pit pavementgroup (B) also containsa slope and base ofslope exhibitthe highestdensities, but varydramatically in spe- sample. Because the scourpit sample plottedso farfrom cies composition.Along the ridgecrest, in a narrowband the restof the data points,principal components analysis 1-2 m wide and about 30 m in length,the crinoidfauna was repeatedwithout this sample. The resultingplot (Fig. is almostentirely Chladocrinus decorus and reachesden- 10) showsthe same two groups,but withthe ridgecrest sitiesof about 12 individualsm-2 witha mean of 4.8 m-2 sample isolatedat a greaterdistance. In thisanalysis, the (Fig. 11A). In contrast,thirty-nine Endoxocrinus parrae firsttwo axes explain81% and 8% of the total variation, and one C. decoruscling to an area of about 0.5 by 2.0 m respectively.Axis 1 is mosthighly correlated with crinoid across the crestof the boulderadjacent to the scour pit. grains,but also containsmajor componentsfrom gastro- Much ofthe slope subhabitatis barrenwith a fewscattered pod, benthicforam, and alcyonarianmaterial; axis 2 rep- individualsat the base and just belowthe ridgecrest. The resentsan echinoidaxis. Plottingat lowerleft, the pave- slope appears to be shieldedfrom northbound flow by the mentgroup (plus a slopeand base ofslope sample) contains ridgecrest and is likelysubject to reducedand unfavorable littlecrinoid and echinoidmaterial. By contrast,echinoid watermovement for colonization by crinoids. Thicker sed- materialdominates the ridgecrest sample at upperright. imenthere also limitsthe area of substratesuitable for attachment.All remainingareas: northand east pave- southwest and a scat- Livingand Death Assemblages ments, margin foreslope,support tered crinoidfauna (Fig. 11B). Densities based on timed Densityand speciescomposition of living crinoid assem- 30-160 m transectsrange from about 0.1 to 0.8 crinoids blagesvary considerably throughout the study area (Table m-2 althoughtwo or three individualsof eitheror both 5). Figure11 illustratestwo distinctassemblages: a dense species occasionallyoccur withina single square meter. standof Chladocrinus decorus on theridge crest (Fig. 11A) Substantialbarren stretches also occur.This scatteredfau-

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions 562 LLEWELLYN& MESSING

TABLE 4-Percentage abundances of benthicskeletal grainsby majortaxonomic group and station.Percentages omitconsideration of unidentifiablegrains. See Table 3 forexplanation of abbreviations.

No. CRI ECH OPH GST BIV BNF COR ALC STY BCH DEC OST SPO BRY WOR

1 16.2 43.4 0.5 4.0 6.0 1.5 7.1 0 0.5 0 12.6 0 3.0 2.0 3.0 2 35.8 15.7 0.5 13.4 3.1 11.8 3.4 5.9 2.2 0.9 5.6 0 1.3 0.4 0 3 35.5 24.9 0.3 8.7 3.4 3.4 1.1 1.6 0.5 1.1 11.6 0 7.1 0 0.8 4 37.1 32.8 1.2 8.4 6.0 0.6 0 0.6 0.9 0.9 8.7 0 2.1 0 0.6 5 16.8 27.7 0.5 9.8 4.0 2.2 4.9 10.3 2.7 2.9 5.8 0.9 9.8 0.9 1.1 6 18.8 20.8 0 16.7 0 13.5 7.3 7.3 1.0 1.0 6.3 2.1 3.1 2.1 0 7 21.8 35.7 1.4 7.6 6.4 1.0 1.6 3.9 0 0.6 12.5 1.4 5.7 0 0.6 8 28.8 31.2 0.5 12.6 5.4 3.3 0.7 4.2 0.7 2.3 7.4 0.7 1.4 0 0.9 9 20.5 37.1 0 13.7 8.5 7.2 0 2.9 0.3 1.1 4.6 0.7 1.6 0.3 0.3 10 20.5 15.5 0 15.5 6.6 7.3 2.3 13.2 0 0.7 8.9 2.0 5.6 2.0 0 11 9.7 21.4 1.2 12.4 2.9 13.6 3.1 18.3 1.3 2.3 11.5 0 2.1 0.2 0.2 12 16.5 22.4 0 15.8 2.0 15.8 0.7 10.5 0.7 1.3 5.9 2.0 2.0 0.7 4.0 13 20.2 17.5 0 14.7 2.2 11.9 1.8 12.8 0 2.7 13.2 1.2 1.2 0.7 0 14 18.6 23.6 0 10.1 4.0 15.1 1.0 12.1 0 0.5 9.6 4.0 1.5 0 0 15 32.8 12.4 0 9.3 2.5 19.5 2.3 0.3 0.9 2.3 9.3 3.1 4.0 0.9 0.6 16 9.2 10.0 0 11.1 1.1 33.7 2.6 26.2 0.9 1.1 2.6 1.1 0.2 0 0

R - mode LinkageDistance (% max. dist.) adjusted scale 0.962 0 25 50 75 100 8 47 I I I I I I I I I I I I I 3 4 Echinoid A 5/ Crinoid B 2./ * Ophiuroid /1410 6 / Stylasterid 1213 15 Bryozoan

Brachiopod Axis 1 Serpulid Ostracod Coral Sponge Bivalve Crustacean C Gastropod -0.1107 (16 Foraminifer -0..944 Axis 2 -0.1035 Alvconarian FIGURE9-Plot of axes 1 and 2 forPrincipal Components Analysis FIGURE8-Dendrogram for R-mode cluster analysis of benthic skel- ofbenthic skeletal material (raw varimax rotation and normalized factor etalmaterial (unweighted pair-group method with arithmetic averaging; loadingscomputed from 3 factorsof a correlationmatrix). Large letters percenttransformation ofsamples; average Euclidean distance mea- referto groupdesignations from cluster analysis; small numbers refer sure;single linkage). to samplenumbers. See textfor definitions of axes. * = Anomalous sampleposition. na is dominatedby C. decorus with one exception:the unknown.Although we observedno large crustaceansor foreslopesubhabitat a fewmeters south of the ridgecrest durophagousfishes, several large nonpredatory(on cri- supportsE. parrae in greaternumbers than C. decorus. noids) fishesoccur at the studysite (mistygrouper Epi- Sources of crinoidmortality at the study site remain nephelusmystacus, rough shark Oxynotus caribbeus, and

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions MODERN CRINOID-RICHSEDIMENTS 563

commonscale 0.219

Axis 1

" 0.881 0.295 0.957 FIGURE10-Plot ofaxes 1 and 2 forPrincipal Components Analysis ofbenthic skeletal material removing the scour pit sample (raw varimax rotationand normalizedfactor loadings computed from 2 factorsof a correlationmatrix). Large letters refer to group designations from clusteranalysis; small numbers refer to samplenumbers. See textfor the definitionsofaxes. * = Anomaloussample position.

e large unidentifiedcarcharhinid sharks) that may disrupt or destroycrinoids incidentally, especially while chasing FIGURE11--A. Dense standof Chladocrinusdecorus on the ridge preyor avoidinglarger predators. The largeechinoid Cal- crestwith rippled sand in the background.Foreground is barrenof ocidaris micans has been implicatedin the disruptionof sediment.Current from right foreground. Depth: 420 m. B. Typical isocrinidfiltration fans elsewhere(Messing et al., 1988) hardgroundwith scattered C. decorus(c), isididalcyonarians (i), bas- and occursat the site. We observeda fewcrownless ketstars(b) and spatangoidechinoid test (e). Depth:428 m. Straight study lineacross lowerright was madeby submersible in moderately thick stalks in life positionbut no recentlydead, intactspeci- sedimentveneer. mens. Crinoidsat the studysite also apparentlycontribute to sedimentsvia autotomyof distal stalk segments.We ob- ation,or the small size ofthe sample with abundant crinoid servedseveral detached stalk segmentsconsisting of one material(sample la in the Taphonomysection below). or a fewnoditaxes among dense C. decorusalong the ridge Slope samplesexhibit a roughgradient in crinoidalcon- crest.Sediment samples include some pluricolumnal grains tributionto sediment.Crinoid ossicles compose 18.8-21.8% althoughthe majorityof columnalsoccur as individual ofslope base samples,collected no closerthan about 40 m ossicles.In situ and laboratoryexperiments (unpublished to the dense stands on the ridgecrest, and 35.5-37.1% of data) indicatethat noditaxesof both species may remain slope samples proper,collected 15-27 m fromthe crest. intactfor at least severalmonths. Species ratiosfor crest and slope rangebetween 8.0 and Abundanceand species compositionof crinoidcolum- 17.1.Pavement and southwestmargin samples show a wide nals in the sedimentboth vary throughout the studyarea rangeof relative abundance values (9.7-32.8% ) and species (Table 3). Species composition,measured as the ratio of ratios(7.5-54.0). The foreslopesample has the secondlow- C. decorusto E. parrae columnals,ranges from 2.3 forthe est speciesratio (6.7) and a fairlylow abundance of crinoid scourpit to 54.0 fora southwestmargin sample (Table 5). materialwhile the scour pit sample exhibitsthe smallest The two ridge crest samples are interestingin that one crinoidcomponent (9.2%) and the lowest species ratio (sample la; not includedin compositionalanalyses) shows (2.3). Both of these low ratios reflectthe relativelyhigh the greatestcrinoidal contribution (52.0%) to the coarse contributionof E. parrae to adjacent livingassemblages. benthicskeletal component of any sample, while the other No directoverall correspondence exists between density has a relativelylow proportion of crinoid material (16.2 %). of livingcrinoids and the contributionof crinoidossicles This may be a functionof local patchiness,seasonal vari- to the benthicskeletal component of adjacent sediments

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions 564 LLEWELLYN& MESSING

TABLE 5-Summary of taphonomicdata fromcrinoid columnals in >2 mm sedimentsamples. Subhabitat-See Table 1 for abbreviations.Sediment type-Character of surficialsediment: thick,>2 cm; thin,<2 cm; ripples, linear asymmetricalripples. Local livingcrinoid assemblage-Population densityof livingcrinoid assemblage: dense, >1 m-2; scatterd, <1 m-2; none, no crinoidswithin immediate area. Mean abrasion-Mean level of abrasion forall columnalsper sample. % Encrustation-Percentage of encrustedcolumnals per sample. % Break- age-Percentage ofcolumnals displaying breakage. Species ratio-Ratio of Chladocrinusdecorus to Endoxocrinus parrae columnals and pluricolumnals.

% Crinoid contri- Local living bution % Species Sediment crinoid to benth. Mean Encrus- % ratio No. Month Subhabitat type assemblage skel.sed. abrasion tation Breakage (n:l) 1 Feb CRS Rippled Dense 16.2 1.5 6.9 32.0 8.0 la Aug CRS Rippled Dense 52.0 1.3 8.3 8.3 infinity 2 Aug SLP Thick None 35.8 2.0 17.9 32.8 17.1 3 Oct SLP Thick Scattered 35.5 2.7 17.8 26.3 15.5 4 Oct SLP Thick None 37.1 2.5 8.8 25.8 7.5 5 Oct FSL Thin Scattered 16.8 2.6 32.4 27.0 6.7 6 Feb BSL Thick None 18.8 3.1 41.9 29.7 8.5 7 Oct BSL Thin Scattered 21.8 2.8 33.5 18.1 10.3 8 Oct SWM Thin Scattered 28.8 3.1 26.4 20.2 25.8 9 Oct SWM Thick Scattered 20.5 3.1 20.3 23.7 54.0 10 Feb EPV Thin Scattered 20.5 3.1 23.7 18.4 37.0 11 Aug EPV Thin Scattered 9.7 2.7 41.9 29.7 18.5 12 Feb NPV Thin Scattered 16.5 3.0 56.0 40.0 7.5 13 Feb NPV Thick Scattered 20.2 2.8 34.8 31.3 11.5 14 Feb NPV Thin Scattered 18.6 2.9 31.6 34.2 14.3 15 Oct NPV Thick Scattered 32.8 2.6 15.7 16.9 36.0 16 Feb SCP Thin Dense 9.2 2.8 41.1 27.1 2.3

(Table 5). Sedimentsfrom areas supportingdense living withan additionalsample fromthe ridgecrest subhabitat assemblages (>1 crinoidm-2: ridge crest and scour pit (sample la). Significantinteractions exist between depen- boulder) display the completerange of crinoidalcontri- dent taphonomicvariables (abrasion, encrustationand butions (9.2-52.0%). In areas where livingcrinoids are breakage) and both independentvariables (habitat and scattered(<1 crinoidm-2: pavements,foreslope, base of sedimenttype) (Multi-between-groups ANOVA; Table 6). slope,SW margin),crinoid ossicles contribute 9.7 to 35.5%. UnivariateANOVA's performedon individualtaphonomic The crinoidalcontribution to sedimentsin areas lacking variablesreveal the specificnature of these interactions crinoids(slope and slope base) rangesfrom 18.8 to 37.1%. (Table 7). Much of this patterncan be explainedwith reference to Levels of abrasion differsignificantly among all three local topographicand hydrodynamicfactors (see Discus- groupedsubhabitat categories (Table 7). Abrasionis min- sion). imal in ridge crest samples (range: 1.33-1.52,x = 1.42), Species ratios of livingassemblages and columnalsac- intermediatein slope (includingforeslope and slope base) cumulatingin sedimentsamples correspondto a greater samples(range: 2.01-2.83, x = 2.47),and mostpronounced degree.As mentionedabove, the low species ratios (high in materialtaken from hardgrounds (pavements, scour pit E. parrae contributions)of foreslopeand scour pit sedi- and southwestmargin areas) (range:2.61-3.14, x = 2.90). mentsreflect the relativelydense livingE. parrae assem- Mean frequencyof biologicalencrustation is significantly blagesin thosesubhabitats. Species ratiosfrom other sub- lowerin crestsediment than in hardgroundmaterial. The habitatsall favorC. decorus and reflectthe generaland percentageof damagedcolumnals (Fig. 3C, D) in samples oftenoverwhelming dominance of that species in living does not differsignificantly among any habitatcategories assemblages. (Table 7). Both abrasionand encrustationvary significantly rela- TaphonomicAnalyses tive to sedimenttype, but in both cases this is a conse- quence of the effectsof a singlesample type.Abrasion is Taphonomicanalyses are based on the same set of sed- significantlylower in rippledsand samplescompared with imentsamples as compositionalanalyses described above eitherthick or thinsediments, while frequency of biolog-

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions MODERN CRINOID-RICHSEDIMENTS 565

TABLE 6-Levels of significancefor Multi-Between TABLE 8-Mean values for taphonomic variables Groups ANOVA's to investigateinteractions between grouped according to habitat and sedimenttype cate- abrasion, encrustationand breakage for habitat and gories. sedimenttype. Mean % Sedimenttype Habitat abra- Encrus- % sion tation Breakage Dependent Wilks' Wilks' variables lambda P level lambda P level Sedimenttype Rippled 1.4 7.6 20.2 Thick 2.9 35.8 26.8 Abrasion Thin 2.5 22.5 26.6 Encrustation 0.14 0.0004 0.15 0.0006 Breakage Habitat Crest 1.4 7.6 20.2 Slope 2.5 25.4 26.6 Hardground 2.9 32.4 26.8

TABLE 7-Levels of significancefor 3-way univariate ANOVA tests for variations in attributes: taphonomic TABLE 9-Levels of significancefor Multi-WayANO- abrasion, withhabitat and sed- breakage, encrustation, VA's (MANOVA's) to investigateinteractions between imenttype. Habitatabbreviations: CR, crest; SL, slope environmentand sediment for variables: SLP and BSL type dependent (includes subhabitats; HG, hardground abrasion, breakage, and encrustation.Analyses were NPV, SWM and SCP See (includes EPV, subhabitats). carriedout two categories of environment(hard- Table 1 forsubhabitat abbreviations. Sediment ab- using type groundand slope) and two categories of sedimenttype breviations:RI, thin. ripples;TK, thick;TN, (thickand thin).

Indepen- Dependentvariable Dependent dent Encrus- variable F value P level variable Abrasion Breakage tation Abrasion 0.02 0.88 Habitat F = 31.77 F = 0.61 F = 3.53 Breakage 1.86 0.20 P = 0.000006 P = 0.56 P = 0.006 Encrustation 0.02 0.90 HG > SL > CR HG = SL = CR HG

CR= SL

Sediment F = 21.73 F = 0.61 F = 6.84 non-significanceas those reportedabove, and the results type P = 0.00005 P = 0.56 P = 0.008 are not illustrated. TK RI = TN = TK TK The two ridge crest samples constituteboth the crest habitatcategory and the rippledsand sedimentcategory, TN> RI TN RI TN RI TN > RI resultingin a directcorrespondence between habitat and sedimenttype for those samples.We used MANOVA's to investigatewhether any otherinteractions exist between slope, hardground,thick or thinsediment categories (ex- clusiveof crest/rippledsamples), but none are significant ical encrustationis in significantlyhigher thinsediments (Table 9). MANOVA's werenot conductedusing all cat- relativeto either or thick sedimentareas rippled (uni- egoriestogether because the resultingempty cells in the variate Table ofcolumnal break- ANOVA's; 7). Frequency analysiswould createartificially significant results. age demonstratesno significantbetween-habitat results althoughareas of thick sedimentsshow almost signifi- DISCUSSION cantlyless (P = 0.056) breakagethan areas of thin sediment.Mean values fortaphonomic characteristics by hab- SedimentComposition itat and sedimentare summarizedin Table 8. UnivariateANOVA's repeatedusing Chladocrinusde- Previouscompositional analyses of moderndeep-water corus columnalsalone show the same patternsof signifi- sedimentsin the northernBahamas (Stetsonet al., 1962; cance and non-significancereported above; these results Wilbur,1976; Mullinsand Neumann,1979; Mullins et al., are not shown.Analyses repeated with the foreslopesam- 1980) emphasize regional variations.The featuresde- ple included in the Hardgroundcategory (see Methods scribed,such as winnowedsands, lithohermsand hard- section)again showthe same patternsof significanceand groundhorizons, provide major clues about generalsedi-

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions 566 LLEWELLYN& MESSING

mentary settings, but offerlittle informationabout At this studysite, moderatedepths and temperatures small-scalevariations in depositionalenvironment. Mul- (12-16° C), substantialcurrent flow and a carbonateen- tivariatestatistical techniques, successfully used to distin- vironmentsupport a diverse,highly skeletonized macro- guish sedimentsfrom moderate depths elsewhere(Doyle benthicassemblage. Benthic skeletal material from the >2 and Feldhausen,1981), reveal that samples taken within mm size fractionof sedimentsamples includes clasts pro- 100 m of each otherat the studysite differsignificantly duced by crinoids,ophiuroids, alcyonarians, corals, sty- in proportionsof coarse-fraction(>2 mm) lithic and lasterids,brachiopods, bivalves, bryozoans and sponges plankticgrains. These variationsreflect a combinationof (all chieflysuspension-feeders), echinoids (chiefly irregu- topographicand geomorphiccontrols acting throughout lar,deposit-feeding taxa), gastropods and crustaceans(most the area, and are clearlydisplayed in resultsof cluster likelypredators/scavengers), and benthicforaminiferans. analysis (Fig. 4), two-wayabundance plot (Fig. 5), and Althoughnot the subjectof this paper, it is worthnoting principalcomponents analysis (Fig. 6). Lithicgrains occur that assessmentof noncrinoidliving assemblages via sub- in greatestabundance in areas wherelithified pavements mersibledoes not accuratelyreflect the death assemblage. showsome evidence of breaking up (forexample, scour pit Visual observationsemphasize large,chiefly suspension- and pittedpavements). On the ridgecrest, slope and base feedingtaxa (forexample, crinoids, alcyonarians, sponges) of slope subhabitats,the hardgroundeither appears more and underestimatesmall sessile and vagileforms (for ex- consolidatedor is buried under thickersediment. Here, ample, brachiopods,many gastropodsand crustaceans). sedimentsamples are dominatedby plankticand/or ben- By contrast,analysis of onlythe coarsesediment fraction thic skeletal material.Local topographyintensifies flow certainlyunderestimates the contributionof sponges to over elevated areas such as the ridge crest,and reduces local communitiesdue to theirchiefly smaller spicules. flowin areas shielded fromcurrents by upstreameleva- Multivariateanalyses demonstratethat crinoidsand tions (scour pit, slope, base of slope). In the formerarea, echinoids,the two most abundant grain types, behave very intensifiedflow generates high densities of suspension- differently,clustering at 80% of the maximumlinkage feederssuch as crinoidswith relativelygreater skeletal distance(Fig. 8). Both groupsalso exhibitwide variations production.The latteract as sinks forthe accumulation in relativeabundance among samples. The increasingde- of both planktictests and bedload-transportedbenthic gree of similaritybetween many other benthic skeletal skeletalgrains. componentschiefly reflects their overall decreasing rela- Planktic grainsin the >2 mm fractionare almost ex- tive abundances ratherthan theirconcordance of distri- clusivelyaragonitic tests of pelagic gastropods-thecoso- bution. Thus, groupsthat contributeuniformly little to matous "pteropods" and heteropods-that have settled sediments (that is, ophiuroids,stylasterids, bryozoans, out of the water column (Pilskaln et al., 1987; Berger, brachiopods,serpulid worms, ostracods) exhibit the great- 1978).These thin-walledshells are particularlysusceptible est similarities(Fig. 8). to winnowingand redistribution(Berger, 1978), but are Q-Mode Clusteranalysis (Fig. 7) and principalcompo- not affectedby dissolutionat studysite depths.High con- nentsplots (Figs. 9, 10) illustratehow variationsin echi- centrationsof thecosome tests form conspicuous dark lags noid, crinoid and other aspects of the benthic skeletal in troughsof rippleson the ridgecrest and at the base of assemblagescan distinguishsamples fromdifferent sub- the slope, indicatinglateral transport by bottomcurrents habitats. Crinoid grains contributethe greatestrelative followedby preferentialdeposition in protecteddowncur- amounts to slope samples, most likelydue to downhill rentareas or pocketswithin an otherwisehigh-energy set- transportfrom dense living aggregations on theridge crest. ting. The low proportionof crinoidmaterial in the singleridge crest sample used in the compositionalanalysis may be due to of crinoid BenthicSkeletal Assemblages rapid down-slopetransport grains or, perhaps,an artifactof limitedsampling. Most studiesdocumenting the spatialfidelity of life and Echinoidgrains are proportionallymost abundant in the death assemblagesconcentrate on macrobenthicskeletal crestand southwestmargin samples and in one slope and material from shallow-watersettings (Johnson, 1965; base of slope sample,and are generallyless importantin Warme,1969; Peterson, 1976; Ekdale, 1977;Bosence, 1979, pavementsamples. Most grainsare derivedfrom irregular 1989;Staff et al., 1986;Henderson and Frey,1986; Carthew urchinsthat are typicalof areas ofdeeper sediment. Echi- and Bosence,1986; Staff and Powell,1988,1990a, b; Miller, noid graindistribution provides evidence that sediments 1988). That littleinformation is available on deep-seaset- are subject to substantiallateral transportin this area. tings is not surprising.Pelagic sedimentsdominate and Their abundance on the ridgecrest suggests that testsor skeletonizedmacrobenthos decrease in diversity,relative fragmentsmay be moveduphill with intense, intermittent abundance, and level of mineralizationwith increasing southboundflow as we observedin February1991, and depth.Substantial contributions to deep-watersediments thendeposited there when the currentreverses and weak- by skeletonizedmacrobenthos can onlybe expectedabove ens (Messingand Llewellyn,in prep.). the CaCO3 compensationdepth or wherecurrents impinge Scour pit materialis unique in beingparticularly low in on the bottomand permitnoncalcareous suspension-feed- both echinoid and crinoidand high in alcyonarianand ers such as spongesand antipathariansto survivein num- benthicforaminiferan grains. The minimalcontribution bers. ofcrinoid material may be explained,despite the presence

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions MODERN CRINOID-RICHSEDIMENTS 567

of about 40 Endoxocrinusparrae on the boulderimme- livingassemblages are typicallyscattered (<1 individual diatelyabove the sample site,by the factthat E. parrae m-2). Oji (1989) estimatesthat the stalk of the Pacific has fewercolumnals per crinoidthan Chladocrinusde- isocrinidMetacrinus rotundus grows between 30 and 60 coruswhich dominates in mostother areas. E. parrae also cm y-1. If such growthrates apply to local species, the tends to contributea greaterproportion of its stalk seg- possibilityexists that the composition and densityof living mentsas pluricolumnalsthan C. decorus(personal obser- assemblages,and the autotomizedcolumnal contribution vation).Alcyonarian grains are moreimportant here than to the sediment,could change markedlyover relatively in any othersample perhapsbecause the dominantform shortperiods of time. Such variationsin the absence of is a slendercylindrical axis segmentproduced by the abun- extensivelateral sortingcould producevery different lo- dant isidid,Lepidisis caryophyllia,that may roll along the calized,time-averaged sediments. bottomand collectin depressions.Alternatively, this spe- The scourpit possessesthe lowestproportion of crinoid cies,abundant elsewhere in thestudy area, may have grown skeletal materialof any sample despite the dense popu- here previously. lation of E. parrae on the adjacent boulder.This sample Additionalevidence for substantial lateral transport of site lies in a depressionrelative to surroundinghard- sedimentexists in the discrepanciesin clusteringof sed- groundsand may thus act as a sink forgrains that dilute imentsamples fromthe same subhabitats.Samples that the crinoidcomponent. As an example,the single scour do not clusterwith others from the same subhabitatwere pitsample is uniquelyrich in cylindricalisidid alcyonarian uniformlycollected at differenttimes. Thus, forexample, axis fragmentsthat are perhaps moreeasily transported slope sample 2 whichfalls into ClusterII (bulk sediment along the bottomthan discoidalcrinoid columnals. Alter- Q-mode dendrogramand principalcomponents analysis) natively,E. parrae probablycontributes columnals to the was taken in August 1991 while slope samples 3 and 4 sedimentat a muchslower rate than C. decorus.Stalks of whichgroup closely together in ClusterI were both col- the formerconsist on averageof littlemore than a third lected in October1991 (Figs. 4, 6). Variationsin bottom as manycolumnals as the latter.E. parrae also produces currentvelocity and directionwith time and localityprob- proportionallymore pluricolumnalsediment grains than ably contributeto these within-subhabitatdiscrepancies. C. decorus,further reducing its contribution by grain count. The possibilityexists, however, that significant differences The high proportionof E. parrae grainsin this sample occuron finerspatial scales withinvisually indistinguish- relativeto others(species ratioof C. decorus:E. parrae = able subhabitats.A greaternumber of synopticsamples 2.3:1) certainlyreflects the abundance of adjacent living takenwithin subhabitats, as wellas time-seriesof samples E. parrae. That C. decorus grains still account for the fromprecisely identified sites, are needed to thoroughly majorityof crinoidal clasts, however, may be due eitherto understandvariations in sedimentcomposition in thisarea. the scour pit actingas a sink as mentionedabove, or to time-averagingfrom a periodwhen living C. decorusdom- or both. Livingand Death Assemblages inated the adjacent boulder, In areas dominatedby C. decorus (ridge crest,pave- As mentionedin the resultssection above, no overall ments,southwest margin), both the overallcrinoidal con- directcorrespondence exists between the densityof living tributionto and theratio of C. decorusto E. parrae ossicles crinoidsand the contributionof crinoidossicles to the in local sedimentsvary widely although species ratiosall benthicskeletal componentof adjacent sediments.Lim- favorC. decorus.However, in the twoareas whereE. par- ited patternsof correspondencedo occur,however, and rae dominatesthe livingfauna (foreslopeand scour pit), local geomorphologyexplains some discrepancies between it (E. parrae) contributesproportionately more to thesed- crinoidalabundance in sedimentsamples and densitiesof iment than elsewhere(species ratios of 6.7 and 2.3, re- autochthonousliving crinoids, but sourcesof variation are spectively).This suggeststhat widespreadlateral mixing not completelyunderstood. of sedimentsis not an importantprocess here and that The ridgecrest is an area of highcrinoid density where sedimentsmay reflect at least some aspectsof small-scale crinoidcolumnals are being activelyadded to the sedi- faunal differences.Alternatively, E. parrae grainsare so ment. Still photographsand videotape taken along the rare comparedto those of C. decorus that dramatically crest show several detached pluricolumnalswith intact differentratios may be a resultof minor variations in local cirrion the substrate.Rather than remaining on the crest, abundance of E. parrae grains. however,columnals apparently move downslope, either by gravitationalsettling or active transportvia northbound Taphonomy flow,accumulating in roughlydecreasing relative abundance withincreasing distance from the ridgecrest source. Diagnostic taphonomicsignatures of benthic skeletal Living crinoidsare virtuallyabsent fromthe slope, most materialdistinguish a wide varietyof shallow-wateren- probablybecause thickaccumulations of sediment largely vironments(see reviewsby Kidwell and Bosence,1991 and eliminatehard substrateanchoring sites and because the Donovan,1991). Despite the rangeand varietyof descrip- area is shieldedfrom adequate flowby the usuallyupcur- tions of moderntaphofacies, however, little information rentridge crest. existson the taphonomiccondition of deep-water benthic Pavementand southwestmargin samples show a wide skeletalmaterial. With respectto stalked crinoids,Ame- rangeof crinoidalcontributions to the sedimentalthough ziane-Cominardand Roux (1987) describeintensive, os-

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions 568 LLEWELLYN& MESSING

tensiblymicrobial biocorrosion of ossiclesrecovered from (1991,Fig. 10) limitabrasion to environmentsabove storm corestaken in 785-1285 m offNew Caledonia. They sug- weatherwave base. Althoughthey refer to predominantly gestthat microbial degradation of organic material within molluscanassemblages and we have analyzedonly crinoids the stereomincreases ossicle fragility and permitsround- in detail,we have foundsubstantial abrasive degradation ing without hydraulic abrasion. Ameziane-Cominardi in skeletalgrains from all majortaxonomic groups. (1991) discussestaphonomic modifications of the same and Unlikeshallow-water environments in whichmany dif- additionalmaterial from the same area: connectedossicles ferentbenthic taxa encrustshells partiallyto completely are exceptional;discoloration and degreesof abrasion vary (Ware,1975; Jackson, 1977; Balson and Taylor,1982; Bish- fromnone to strong;and sedimentsare enrichedin colum- op, 1988; Taylor et al., 1989), encrustationat the study nal and brachials,and depleted in cirralsand pinnulars, site is chieflylimited to isolatedserpulid worm tubes and relativeto expectedvalues derivedfrom dissociated whole benthicforaminiferans. Many host columnals are abraded, specimens.Both ofthese studies are limited,however, be- withencrusting organisms usually on articularfaces, in- cause the data wereextracted from core samples; no com- dicatingthat settlementusually occursafter disarticula- parisons could be made betweenlocal livingand death tion. assemblages. Frequency of encrustationis significantlygreater in In the currentstudy, a comparativetaphonomic ex- samplestaken from thin sediment veneers than from either aminationusing ossicle abrasion, encrustation and break- rippled(ridge crest) or thicksediment samples. When an- age offersan effectivemeans of distinguishingsamples on alyzed by habitat,rippled sands fromthe ridgecrest ex- a finespatial scale (metersto tens of meters),as well as hibitsignificantly less encrustationthan hardground sam- providinginformation on bioticinteractions and post-mor- ples. The extentand frequencyof encrustationis usually temhistories of deep-water crinoidal material. Columnals taken to indicate durationof exposureon the sea floor appear to be excellentindicators of taphonomicchange, (Driscoll, 1970; Kidwell and Bosence, 1991). It follows, revealingsignificant variations in levels of abrasion and then,that columnals taken from thin sediments have been encrustationfor the habitatsunder consideration. exposedlongest at the sediment-waterinterface, while co- Columnalabrasion at the studysite increasesprogres- lumnalsfrom the ridge crest have beenexposed least. That sivelyand significantlyfrom rippled-sand, ridge crest sam- ridgecrest samples also demonstratethe least abrasionis ples throughslope to hardgroundsamples. The former additionalevidence that material here is relatively"fresh," exhibitthe least abrasionwhether counted as "crestsam- having sufferedlittle exposureto eitherabrasive or en- ples" in analysisby habitat,or "rippledsamples" in anal- crustingtaphonomic agents. The ridgecrest with its dense ysesby sedimenttype. Downslope migration of columnals livingassemblage of C. decorusis presumablya crinoidal fromdense, ridge crest,living assemblages explains in- sedimentproduction site, but ratherthan remainingon creasedabrasion between ridge crest and slope (Hoskinet the crest,columnals are transporteddownhill and accu- al., 1986; Boss and Liddell, 1987), but does not account mulateon the slope. forincreased abrasion in hardgroundsamples whereco- Unlike degreesof abrasionand frequencyof encrusta- lumnalsare ostensiblybeing producedin situ. Increased tion, the frequencyof damaged or brokencolumnals in abrasionof columnalson hardgroundmay resultfrom in samplesshows no significantoverall or specific relationship situ reworking,small-scale lateral transport, or greater time witheither habitat or sedimentcharacteristics. One anal- of exposureat the sediment-waterinterface (Chave, 1964; ysis approachessignificance (P = 0.056), suggestingthat Driscoll, 1967, 1970; Driscoll and Weltin,1973; Watson thinsediments may contain more damaged columnals than and Flessa, 1979;Hoskin et al., 1983;Cutler, 1987; Fiirsich thicksediments. The lackof correlation between frequency and Flessa, 1987;Meldahl, 1987). No significantdifferences of breakageand habitatimplies a causal agent actingin- in abrasionexist between thick and thinsediment samples, dependentlyof environment. The natureof the damage- however,as mightbe expectedif durationof exposureat irregulargrooves, pits and fractures-rulesout abrasion. the sedimentsurface were a factorcontributing to abra- The mostlikely explanation for such damage is a biological sion.Durations of exposure and transportrequired to pro- agent such as a predatoror scavengerthat causes delib- duce the variousdegrees of observedabrasion remain un- erate or incidental biogenic fragmentation.Several re- known. Local rates of sedimentation,and hardground searchershave observedcrownless isocrinid stalks in up- cementationand erosion are similarlyunknown. We, rightlife position still attachedto the substrateby cirri, therefore,do not knowat presentthe extentof time-av- and typicallylacking the most proximalportion (Conan eragingthat these sediments represent. While the "fresh" et al., 1981; Fujita et al., 1987; pers. obs.), additionalevi- columnalsthat dominatethe ridgecrest certainly repre- dence forpredators or, at least, some importantagent of sent a less time-averagedsediment than the hardground disturbance.Such crownlessstalks can survivein thiscon- sediments,the possibility exists that particularly degraded ditionfor at least severalmonths (Amemiya and Oji, 1992; columnalsmay be reworkedfrom eroding pavements which pers. obs.). would vastlyexpand the periodover whichthe sediment Meyeret al. (1984) and Meyer(1985) recordpredation is time-averaged. on modern,shallow-water comatulid populations, while It is also worthnoting that this study extends the range severalauthors use regeneratedarms as indirectevidence of environmentsin whichabrasion must be recognizedas of sublethal attacks in populationsof both ancient and an importanttaphonomic agent. Kidwell and Bosence modernshallow-water crinoids (Mladenov, 1983; Meyer

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions MODERN CRINOID-RICHSEDIMENTS 569

and Ausich,1983; Arendt,1985; Meyer,1985; Schneider, is also importantto repeatthat "no one criterionalone is 1988). Predationpressure has apparentlybeen an impor- diagnostic,but rathera combinationof criteriashould be tant factorin stalked crinoidevolution and distribution, used wheneverpossible" (Mullins et al., 1981,p. 1008). and is widelyheld at least partlyresponsible for the Me- 1) The presenceof algae or otherorganisms identifiable sozoic retreatof stalkedcrinoids to deep water(Vermeij, as photosyntheticis the mostobvious indicator of a shal- 1977; Meyerand Macurda, 1977; Lane, 1984; Signorand low-waterenvironment, although particularly steep bank- Brett, 1984; Meyer,1985; Aronson,1991). Nevertheless, marginslopes permittransport of shallow-waterskeletal althoughpredation intensity may be lowerthan in shallow materialbelow the euphoticzone. The lattercase may be waterhabitats, the appearanceof upright crownless stalks distinguishedby recognitionof bedformsattributable to and broken columnals in 430 m stronglysuggests that slumping,talus deposits and grainflows (Mullins and Neu- moderndeep-water stalked crinoids are not exemptfrom mann,1979). A lack ofsuch organismsdoes not guarantee potentiallylethal disturbance.The precisenature of such that materialwas deposited in deep water,but the pos- disturbanceevents (that is, whetherrelated to predation sibilityof a deep-waterorigin should be consideredwith or incidentalcollisions with large fish) remains unknown. referenceto othercriteria. Alternatively,upright crownless stalks may representre- 2) Skeletal grainsof planktic/pelagicorigin are rare in centlydead individualssubject to differentialpost-mortem modernshallow-water carbonate deposits whereas they disarticulationrates betweencrown and stalk,regardless typically dominate deep-water carbonate sediments. of cause of death. Plankticgastropod shells (thecosome and heteropod)con- Post-mortemmodification by scavengersor deposit tribute13-66% of the >2 mm fractionof sedimentsam- feedersis widelyrecognized in shallow-watershelly de- ples examinedin the currentstudy. A qualitativeexami- posits and may also explain columnalbreakage (Schiifer, nation indicatesthat plankticforaminiferans contribute 1972;Meldahl and Flessa,1990; Kidwell and Bosence,1991; substantiallyto finerfractions. Tshudy et al., 1989; Walker,1988; Gregoryet al., 1979; 3) Verticaland lateral facies transitionsshould differ Bertnessand Cunningham,1981; Alexander, 1986). Scav- considerablybetween shallow- and deep-watercrinoidal engersact mainlyat the sediment-waterinterface. Re- deposits. Along the marginsof the Bahama Banks, the duced (althoughnot quite significant:0.056) breakageof winnowedsands of the studyarea grade into obviousba- columnalsfrom thick sediments relative to thosefrom thin sinal depositscharacterized by pelagic oozes and turbid- sedimentssuggests that scavengers may play a rolein post- ities. Debris and grain flowdeposits are absent. In the mortemdegradation of columnals. otherdirection, "peri-platform sand facies" (Mullins and Neumann,1979) still retaina substantialpelagic compo- MODEL AND APPLICATIONS nent. By contrast,Oji (1985) recognizesan isocrinidas- semblagein the Miyako Group (Upper Aptian of Japan) Modern deep-waterskeletal sedimentsrich in stalked as shallow-watervia a transitionfrom the crinoid-con- crinoidmaterial offer potential opportunities for better tainingsiltstone to fossilbeachrock. understandingof fossil assemblages. Although deep-water 4) Althoughnumerous potential biases exist (see, for environmentsare traditionallythought of as low-energy example,Schopf, 1980), stable oxygenisotope ratios offer relative to shallow-watersettings, particular configura- a possible means of distinguishingat least warmversus tions of benthic currentactivity and regional geomor- cool paleoenvironments.The mean temperaturein the phology,such as the deep-watercontour currents along studysite area is about 15-16° C (Leaman et al., 1987). the westernmargin of the LittleBahama Bank (Neumann This figureis unusuallyhigh forthis depth because iso- and Ball, 1970; Leaman et al., 1987; Messinget al., 1990), thermsexhibit strong geostrophic tilting across the Straits can producerelatively coarse-grained sediments and ben- of Florida. Similartemperatures occur in 100 m alongthe thic communitiesdominated by suspensionfeeders (but Floridaslope. Isotopeanalyses that suggest similar or low- see Thistle et al., 1985 fora differentenvironmental re- er paleotemperaturesin otherwisetropical settings may sponse to deep near-bottomcurrents). These crinoid-rich indicatea deeper-waterenvironment. depositsalso displaysmall-scale variations in taphonomic The lack of recognitionof ancientdeep-water crinoid- and compositionalcharacter more commonlyassociated richdeposits may reflecteither their actual sparsenessor with shallow-waterenvironments. If such environments theirmisinterpretation in the literaturedue to lack of an were representedin an ancient marinesequence, subtle appropriaterecent example. This study providesa car- diagnosticcriteria would be the only way of confirming bonatebank-margin model for the production of relatively theirdeep-water origin. coarse-grained,crinoid-rich sediments that may permit Mullinset al. (1981) describecriteria by whichancient reinterpretationof some ancientcrinoid-rich limestones. deep-watercoral mounds may be distinguishedfrom shal- Ruhrmann(1971a), forexample, explains the occurrence low-watercoral complexes in therock record. Because they ofDevonian offshore crinoid-rich calcarenites by invoking conductedtheir study in a high-energy,bank-margin, car- uphilltransport of crinoidmaterial onto a seawardshoal bonateenvironment north of the Little Bahama Bank sim- (p. 244-245). Using the model proposedin this paper,in- ilar in manyways to the currentstudy site, their criteria creasedflow along the bankmargin crates a locallyhigher- are worthnoting as theyapply to the recognitionof deep- energyregime with associated winnowed crinoid-rich sed- water,coarse-grained, crinoid-rich paleoenvironments. It imentswithout resort to extensiveand unlikelyupslope

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions 570 LLEWELLYN& MESSING

transport.Similarly, an increasedflow regime rather than augmentedflow over small-scaletopographic irregulari- a loweredsea levelmay explain the developmentof coarse ties,explain much of the variationobserved in livingcri- crinoidfacies on the flanksof a Waulsortianmud mound noid assemblages,and at least some of the compositional complexotherwise recognized as relativelydeep water(220- and taphonomicvariations found in the >2 mmsediment 280 m; Bridgesand Chapman,1988). analyses. Small-scalevariations in taphonomicand compositional 5) This environmentprovides a high-energy,deep-wa- characterproduced, for example, by the downslopemove- ter,basin-margin model for the formation of coarse-grained, mentof crinoid ossicles from dense ridge crest populations, crinoid-richsediments. Criteria for the recognition of such may providea model forunderstanding similarly scaled an environmentin ancientcrinoid-rich limestones include relationshipsbetween flanking crinoidal sediment accu- lack ofphotosynthetic organisms such as algae,abundance mulationsand livingassemblages on a varietyof ancient of planktic/pelagiccomponents, and lateral facies rela- moundsincluding Waulsortian (Wilson, 1975) and smaller tionships. Ft. Payne (Osagean) buildups (Meyer et a., 1989; D.L. Meyer,personal communication,December, 1992). Fur- therexamination of the paleontological literature may pro- ACKNOWLEDGMENTS vide more of ancient crinoid-rich examples deep-water, liketo thankthe officersand crewsof the limestones. We would R/V Edwin Link and Seward Johnson,the crewsof the John- son Sea Link I and II, and the submersibleoperations CONCLUSIONS staffat the HarborBranch Oceanographic Institute, Fort Pierce, FL, especially Tim Askew and Pam Keen. Lou crinoid-richsediments are foundon 1) Coarse-grained, Fisherof the BrowardCounty Office of Natural Resources the southwesternmargin of the Little Bahama Bank in Protectionmade available sedimentsieves and sieveshak- water of 419-434 m. Extensive depths flat,hardground er. David L. (Universityof Cincinnati),Nick Hol- characterizemuch of the withlow Meyer pavements studyarea, land Instituteof Oceanography),and Tomasz relief and isolatedboulders occasionalto- (Scripps ridges affording Baumiller advice, ideas, elevation. (Harvard University)provided pographic and direction,for all of whichwe are thankful.We also The three main sources of coarse sedimentare: 2) a) wishto thankour reviewers for their thoughtful comments. lithic derivedfrom of hard- grains break-up pre-existing This studyis supportedby National Science Foundation testsof ground;b) planktic/pelagicorganisms; and c) ben- EAR-9004232and the HarborBranch Oceanograph- thic skeletalmaterial in situ cri- grant produced by echinoids, ic Institute,Fort Pierce, FL. This is contributionno. 4 benthic noids, gorgonians,corals, gastropods,bivalves, fromthe Deep Ocean Society,Miami, FL. foraminiferans,and other minorcomponents. Crinoidal skeletalmaterial shows considerable fidelity with respect to livingassemblages, and bothsediment composition and REFERENCES benthicskeletal materialdemonstrate small-scale variations correspondingto differencesin local environment. ALEXANDER, R.R., 1986,Resistance to and repairof shellbreakage Grosssediment varies in inducedby durophages in Late Ordovicianbrachiopods: Journal composition mainly proportions v. 273-285. oflithic and in whileben- ofPaleontology, 60,p. plankticgrains present samples, AMEMIYA, S. and OJI,T. 1992,Regeneration in sea lilies:Nature, v. thic skeletalassemblages vary mainly in proportionsand 357, p. 546-547. abundance of echinoid and crinoidmaterial, which to- AMEZIANE-COMINARDI,N., 1991, Distribution bathymetrique des pen- getheraccount for over 45 % of total coarse (>2 cm) ben- tacrinesdu Pacifiqueoccidental. Essai de modelisationet d'ap- thic skeletalgrains. plicationaux faunes du Lias: Documentsdes Laboratoires de Geo- A carriedout us- logieLyon, No. 116,253 p. 3) comparativetaphonomic analysis, et mi- crinoid reveals small-scale variationsin AMEZIANE-COMINARDI,N., and Roux, M., 1987, Biocorrosion ing columnals, critisationdes ossiclesd'echinodermes en milieubathyal au large taphonomiccharacteristics related to differencesin local de la Nouvelle-Caledonie:Comptes Rendu de l'Academiedes Sci- environment.Abrasion levels demonstratesignificant in- ences,Paris, v. 305,s6rie II, p. 701-705. creases fromridge crest to slope and hardgroundhabitat ARENDT. Y.A., 1985,Biotic relationships of crinoids:Paleontology groups,reflecting down-slope transportation and rework- Journal,v. 19,p. 67-72. in theDevo- ing. Encrustationis greatestin areas of thin sediment, ARONSON,R.B., 1991,Escalating predation on crinoids nian:Negative community-level evidence: Lethaia, v. 24,p. 123- apparentlyreflecting length of exposureat the sediment/ 128. waterinterface. Breakage frequency does notcorrelate with BALSON,P.S., and TAYLOR,P.D. 1982, Paleobiology and systematics any observedphysical variables and is presumedto rep- of large cyclostomebryozoans fromthe Pliocene Coralline Crag resentthe action of predatorsand/or scavengers in the of Suffolk:Palaeontology, v. 25, p. 529-554. area. BAUMILLER, T.K., and AUSICH, W.I., 1992,The "Broken Stick" model contourcurrents in the as a null hypothesisfor crinoid stalk taphonomyand as a guide 4) Regionalbank-margin study in fossils: v. area contributeto the of coarsesed- to distributionof connective tissue Paleobiology, 18, production relatively p. 288-298. imentsand favorbenthic communities dominated by sus- BERGER, W.H., 1978,Pteropod distribution and the aragonitecom- pension-feeders,especially where flow exposes hard sub- pensationdepth: Deep-Sea Research,v. 25,p. 447-452. strates.Variations in benthiccurrent regime, notably the BERTNESS,M.D., and CUNNINGHAM,C., 1981, Crab shell-crushing

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions MODERN CRINOID-RICHSEDIMENTS 571

predationand gastropodarchitectural defense: Journal of Exper- FUJITA,T., OHTA,S., and OJI,T., 1987, Photographicobservations of imental Marine Biology and Ecology, v. 50, p. 213-230. the stalked crinoid Metacrinus rotundus (Carpenter) in Suruga BISHOP,J.D.D., 1988, Disarticulated bivalve shells as substrates for Bay, central Japan: Journal of the Oceanographical Society of encrustationby the bryozoan Cribrilinapuncturata in the Plio- Japan, vol. 43, p. 333-343. Pleistocene Red Crag of Eastern England: Palaeontology,v. 31, FURSICH,F.T., and FLESSA,K.W., 1987,Taphonomy of tidal flatmol- p. 237-253. luscs in the northernGulf of California:Paleoenvironmental anal- BLYTH CAIN,J.D., 1968,Aspects of the depositionalenvironment and ysis despite the perils of preservation:PALAIOS, v. 2, p. 543-559. paleoecologyof crinoidallimestones: Scottish Journalof Geology, GREGORY,M.R., BALLANCE,P.F., GIBSON,G.W., and AYLING,A.M., v. 4, p. 191-208. 1979, On how some rays (Elasmobranchia) excavate feedingde- BOSENCE,D.W.J., 1979, Live and dead faunas fromcoralline algal pressions by jetting water: Journal of SedimentaryPetrology, v. gravels,Co. Galway, Eire: Palaeontology,v. 19, p. 365-395. 49, p. 1125-1130. BOSENCE,D.W.J., 1989, Surface sublittoralsediments of Florida Bay: HARBAUGH,J.W., and MERRIAM,D.F., 1968, Computer Applications Bulletin of Marine Science, v. 44, p. 434-453. in StratigraphicAnalysis: Wiley, New York, 282 p. Boss, S.K., and LIDDELL, W. D., 1987, Patterns of sediment com- HEATH, K.C., and MULLINS,H.T., 1984, Open-ocean, off-banktrans- position of Jamaican fringingreef facies: Sedimentology,v. 34, p. portof fine-grained carbonate sediment in the NorthernBahamas, 77-87. in STOW, D.A.B., and PIPER,D.J.W., eds., Fine Grained Sediments: BRETT,C.E., and BAIRD,G.C. 1986, Comparativetaphonomy: A key Deep-water Processes and Facies: Blackwell ScientificPublica- to paleoenvironmentalinterpretation based on fossilpreservation: tion, London, p. 199-208. PALAIOS, v. 1, p. 207-227. HENDERSON, S.W., and FREY, R.W., 1986, Taphonomic redistribution BRIDGES,P.H., and CHAPMAN,A.J., 1988, Anatomyof a deep water of mollusk shells in a tidal inlet channel, Sapelo Island, Georgia: mud-moundcomplex to the southwestof the Dinantian platform PALAIOS, v. 1, p. 3-16. in Derbyshire,U.K.: Sedimentology,v. 35, p. 139-162. HINE, A.C., and MULLINS,H.T., 1983, Modern carbonate shelf-slope CARTHEW,R., and BOSENCE,D.W.J., 1986, Communitypreservation breaks: SEPM Special Publications no. 33, p. 169-188. in Recent shell-gravels,English Channel: Palaeontology,v. 29, p. HINE, A.C., WILBER, R.J.,BANE, J.M., NEUMANN, A.C., and LORENSON, 243-268. K.R., 1981, Off-banktransport of carbonate sands along open CATTELL, R.B., 1966, The scree test forthe numberof factors:Mul- leeward bank margins:Northern Bahamas: Marine Geology,v. 42, tivariateBehavioural Research, v. 1, p. 245-276. p. 327-348. CHAVE, K.E., 1964, Skeletal durabilityand preservation,in IMBRIE, HOSKIN,C.M., GEIER,J.C., and REED, J.K. 1983, Sediment produced J.E., and NEWELL,N.D., eds., Approaches to Paleoecology: Wiley, fromabrasion ofthe branchingstony coral Oculina varicosa: Jour- New York, p. 377-387. nal of SedimentaryPetrology, v. 53, p. 779-786. COLBY,N.D., and BOARDMAN,M.R., 1989, Depositional evolution of HOSKIN,C.M., REED, J.K., and MOOK,D.H., 1986, Production and a windward,high-energy lagoon, Graham's Harbor,San Salvador, off-banktransport of carbonate sediment,Black Rock, southwest Bahamas: Journalof SedimentaryPetrology, v. 59, p. 819-834. Little Bahama Bank: Marine Geology,v. 73, p. 125-144. CONAN,G., Roux, M., and SIBUET,M., 1981, A photographicsurvey IMBRIE,J. and VAN ANDEL,T.J.H., 1964, Vector analysis of heavy of a population of stalked crinoidDiplocrinus (Annacrinus) wy- mineral data: Geological Society of American Bulletin, v. 75, p. villethomsoni(Echinodermata) fromthe bathyal slope of the Bay 1131-1156. of Biscay: Deep-Sea Research, v. 28, 441-453. JACKSON,J.B.C., 1977, Habitat area, colonisation,and development CUMMINS,H., POWELL,E.N., STANTON,R.J., JR., and STAFF,G., 1986, of epibenthiccommunity structure, in KEEGAN,B.F., O'CEIDIGH, The rate of taphonomic loss in modern benthic habitats: How P., and BOADEN,P.J.S., eds., Biology of Benthic Organisms:Per- much of the potentially preservable communityis preserved?: gamon Press, Oxford,p. 349-358. Palaeogeography,Palaeoclimatology, Palaeoecology, v. 52, p. 291- JOHNSON,R.G., 1965, Pelecypod death assemblages in Tomales Bay, 320. California:Journal of Paleontology,v. 39, p. 80-85. CUTLER,A.H., 1987, Surface texturesof shells as taphonomic indi- KIDWELL, S.M., 1988, Taphonomic comparisonof passive and active cators: Paleontology Society Special Publication, no. 2, p. 164- continentalmargins: Neogene shell beds of the Atlantic Coastal 176. Plain and northernGulf of California:Palaeogeography, Palaeo- DAVIES,D.J., POWELL,E.N., and STANTON,R.J., JR. 1989, Tapho- climatology,Palaeoecology, v. 63, p. 201-223. nomic signature as a functionof environmentalprocess: Shells KIDWELL,S.M., and BEHRENSMEYER,A.K., 1988,Overview: Ecological and shell beds in a hurricane-influencedinlet on the Texas coast: and evolutionaryimplications of taphonomic processes: Palaeo- Palaeography, Palaeoclimatology, Palaeoecology, v. 52, p. 317- geography,Palaeoclimatology, Palaeoecology, v. 63, p. 1-13. 356. KIDWELL, S.M., and BOSENCE, D.W., 1991, Taphonomy and time- DONOVAN,S.K., 1991, The taphonomy of echinoderms:Calcareous averagingof marine shelly faunas, in ALLISON,P.A., and BRIGGS, multi-elementskeletons in the marineenvironment, in DONOVAN, D.E., eds., Taphonomy: Releasing the Data Locked in the Fossil S.K., ed., The Processes of Fossilization: Columbia Univ. Press, Record: Topics in Geobiology,v. 9, Plenum, New York, p. 115- New York, p. 241-269. 209. DOYLE, L.J., and FELDHAUSEN,P.H., 1981, Bottom sedimentsof the LANE,N.G., 1984, Predation and survivalamong inadunate crinoids: easternGulf of Mexico examinedwith traditional and multivariate Paleobiology,v. 10, p. 453-458. statisticalmethods: Mathematical Geology,v. 13, p. 93-117. LEAMAN,K.D., MOLINARI,R.L., and VERTES,P.S., 1987, Structureand DRISCOLL,E.G., 1967, Experimental field study of shell abrasion: variabilityof the Florida Currentat 27°N: April 1982-July1984: Journalof SedimentaryPetrology, v. 37, p. 1117-1123. Journalof Physical Oceanography,v. 17, p. 565-583. DRISCOLL, E.G., 1970, Selective bivalve destructionin marine envi- LEWIS, R.D., CHAMBERS, C.R., and PEEBLES, M.W., 1990, Grain mor- ronments,a fieldstudy: Journal of SedimentaryPetrology, v. 40, phologies and surfacetextures of Recent and Pleistocene crinoid p. 898-905. ossicles, San Salvador, Bahamas: PALAIOS, v. 5, p. 570-579. DRISCOLL,E.G., and WELTIN,T.P., 1973, Sedimentaryparameters as LEWIS,R.D., and PEEBLES,M.W., 1988,Surface textures of Nemaster factors in abrasive shell reduction: Palaeogeography,Palaeocli- rubiginosa (Crinoidea; Echinodermata) fromSan Salvador, Ba- matology,Palaeoecology, v. 13, p. 275-288. hamas: Geological Society of America,Abstracts with Programs, EKDALE, A.A., 1977, Quantitative paleoecological aspects of modern v. 20, no. 7, p. A341. marine mollusk distribution,northeast Yucatan coast, Mexico: LEWIS,R.D., and PEEBLES,M.W., 1989,Surface texturesof Nemaster AmericanAssociation of Petroleum Geologists Studies in Geology, rubiginosa (Crinoidea; Echinodermata), San Salvador, Bahamas: v. 4, p. 195-207. Proceedingsof the Fourth Symposiumon the Geology of the Ba-

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions 572 LLEWELLYN& MESSING

hamas,Bahamian Field Station, San Salvador,Bahamas, p. 203- MILLER,A.I., LLEWELLYN,G., PARSONS,K.M., CUMMINS,H., BOARD- 207. MAN, M.R., GREENSTEIN,B.J., and JACOBS,D.K., 1992, Effectof LIDDELL,D.W., 1975,Recent crinoid biostratinomy: Geological So- Hurricane Hugo on molluscan skeletal distributions,Salt River cietyof America, Abstracts with Programs, v. 7, no. 7, p. 1169. Bay, St. Croix, U.S., VirginIslands: Geology,v. 20, p. 23-26. MARTIN,R.E., and WRIGHT,R.C., 1988,Information loss in thetran- MLADENOV,P.V., 1983,Rate ofarm regenerationand potentialcauses sitionfrom life to deathassemblage of foraminifera in back reef of arm loss in the featherstar Florometra serratissima (Echino- environments,Key Largo,Florida: Journal of SedimentaryPe- dermata: Crinoidea): Canadian Journalof Zoology,v. 61, p. 2873- trology,v. 62,p. 399-411. 2879. MELDAHL,K.H., 1987,Origin of shell beds and evolutionof a shelly MULLINS,H.T., and NEUMANN,A.C., 1979,Deep carbonatebank mar- sand spit,Bahia La Choya,Northern Gulf of California:Pale- gin structureand sedimentationin the northernBahamas: SEPM ontologicalSociety Publication. v. 2, p. 189-199. Special Publication, no. 27, p. 165-192. MELDAHL,K.H., and FLESSA,K.W., 1990, Taphonomic pathways and MULLINS, H.T., NEUMANN,A.C., WILBER, R.J., HINE, A.C., and comparativebiofacies and taphofacies in a Recentintertidal/shal- CHINBURG,S.J., 1980, Carbonate sediment drifts in NorthernStraits lowshelf environment: Lethaia, v. 23,p. 43-60. of Florida: Bulletin of the American Association of Petroleum MESSING,C.G., 1985, Submersible observations of deep-water crinoid Geologists,v. 64, p. 1701-1717. assemblagesin thetropical western Atlantic Ocean, in KEEGAN, MULLINS,H.T., NEWTON,C.R., HEATH, K., and VANBUREN,H.M., B.F., and O'CONNOR,B.D.S., eds.,Proceedings of the Fifth Inter- 1981, Modern deep-water coral mounds north of Little Bahama nationalEchinoderm Conference, Galway: A.A. Balkema, Rotter- Bank: Criteriafor recognition of deep-watercoral biohermsin the dam,p. 185-193. rock record:Journal of SedimentaryPetrology, v. 51, p. 999-1013. MESSING,C.G., AND LLEWELLYN, G., (in prep.),Morphological, pos- NEUMANN,A.C., 1974, Cementation,sedimentation, and structureon turaland distributional variations relative to topography and cur- the flanks of a carbonate platform,north-western Bahamas, in rentflow in an assemblageof livingsea lilies (Echinodermata: Recent Advances in Carbonate Studies: FairleighDickinson Uni- Crinoidea). versity,West Indies Laboratory Special Publ., no. 6, p. 26-30. MESSING,C.G., NEUMANN,A.C., and LANG,J.C., 1990, Biozonation NEUMANN,A.C., and BALL,M.M., 1970, Submersible observationsin of deep-waterlithoherms and associatedhardgrounds in the the Straits of Florida: Geology and bottom currents:Geological NortheasternStraits of Florida: PALAIOS, v. 5, p. 15-33. Society of American Bulletin. v. 81, p. 2861-2874. MESSING,C.G., ROSEMYTH,M.C., MAILER,S.R., and MILLER, J.E., NEUMANN,A.C., KOFOED,J.W., and KELLER,G.H., 1977, Lithoherms 1988,Relocation movement in a stalkedcrinoid (Echinodermata: in the Straits of Florida: Geology,v. 5, p. 4-10. Crinoidea):Bulletin of Marine Science, v. 42,p. 480-487. NORRIS,R.D., 1986,Taphonomic gradientsin shelffossil assemblages: MEYER,C.A., 1990,Depositional environment and paleoecologyof Pliocene Purisima Formation,California: PALAIOS, v. 1, p. 91- crinoid-communitiesfrom the Middle JurassicBurgundy-Plat- 103. formof Western Europe, in DE RIDDER,C., DUBOIS,P., LAHAYE, OJI,T., 1985, Early Cretaceous Isocrinus fromnortheast Japan: Pa- M., and JANGOUX,M., eds.,Echinoderm Research: A.A. Balkema, laeontology,v. 28, p. 629-642. Rotterdam,p. 25-30. OJI, T., 1989, Growth rate of stalk of Metacrinus rotundus (Echi- MEYER,D.L., 1971,Post mortemdistarticulation of Recentcrinoids nodermata: Crinoidea) and its functionalsignificance: Journal of and ophiuroidsunder natural conditions: Geological Society of Faculty of Science Universityof Tokyo (II), v. 22, p. 39-51. America,Abstracts with Programs, v. 3, no. 7, p. 645-646. PARSONS, K.M., 1989, taphonomy as an indicator of environment: MEYER,D.L., 1985,Evolutionary implications of predation on recent Smuggler's Cove, St. Croix, US VI, in HUBBARD,D.K., ed., Ter- comatulidcrinoids from the GreatBarrier Reef: Paleobiology, v. restrial and Marine Geology of St. Croix, U.S. Virgin Islands: 11,p. 154-164. Fairleigh Dickinson University,West Indies Laboratory,Special MEYER,D.L., and AUSICH,W.I., 1983, Biotic interactions among Re- Publication, no. 8, p. 135-143. cent and among fossil crinoids,in TEVESZ,M.J.S., and MCCALL, PARSONS,K.M., BRETT, C.E., and MILLER,K.B., 1988, Taphonomy P.L., eds. BioticInteractions in Recentand FossilBenthic Com- and depositional dynamicsof Devonian shell-richmudstone: Pa- munities:Plenum, New York,p. 377-427. laeogeography,Palaeoclimatology, Palaeoecology, v. 63, p. 109- MEYER,D.L., AUSICH,W.I., and TERRY,R.E., 1989,Comparative ta- 139. of echinodermsin carbonatefacies: Fort For- phonomy Payne PETERSON,C.H., 1976, Relative abundance of living and dead mol- of and Tennessee:PA- mation(Lower Mississippian) Kentucky luscs in two Californialagoons: Lethaia, v. 9, p. 137-148. LAIOS, v. 4, p. 533-552. PILSKALN,C.H., NEUMANN,A.C., and BANE,J.M., 1987, Periplatform MEYER, D.L., LAHAYE, C.A., HOLLAND,N.D., ARNESON,A.C., and carbonate flux in the northernBahamas: Deep-Sea Research, v. STRICKLER,J.R., 1984, Time-lapse cinematography offeature stars 36, 1391-1406. (Echinodermata:Crinoidea) on the Great BarrierReef, Australia: p. REYMENT, R.A., 1963, Multivariate analytical treatmentof quanti- Demonstrations of posture changes, locomotion,spawning and tative species associations: An example frompalaeoecology: Jour- possible predation by fish:Marine Biology,v. 78, p. 179-184. nal of Animal Ecology, v. 32, p. 535-547. MEYER, D.L., and MACURDA, D.B., 1977, Adaptive radiation of the Les comatulid crinoids:Paleobiology, v. 3. p. 294-302. Roux, M., 1980, Crinoides pedoncules (Echinodermes) photo- sur les dorsales de et du Paci- MEYER,D.L., MESSING,C.G., and MACURDA,D.B., 1978, Zoogeogra- graphies oc6aniques l'Atlantique Rendu de l'Aca- phy of tropicalwestern Atlantic Crinoidea (Echinodermata): Bul- fique: Implications biog6ographiques:Comptes des v. S6rie 901-904. letin of Marine Science, v. 28, p. 412-441. demie Sciences, Paris, 291, D, p. Sedimentation unterdovonischer MEYER, D.L., and MEYER, K.B., 1986, Biostratinomyof Recent cri- RUHRMANN,G., 1971a, Riff-ferne Neues noids (Echinodermata) at Lizard Island, Great BarrierReef, Aus- Krinoidenkalke im Kantabrischen Gebirge (Spanien): v. tralia: PALAIOS, v. 1, p. 294-302. Jahrbuchfur Geologie und Palaontologie, Monatshefte, 137, p. MILLER,A.I., 1988, Spatial resolutionin subfossilmolluscan remains: 231-248. Kri- Implications forpaleo-biological analyses: Paleobiology,v. 14, p. RUHRMANN,G., 1971b, Riff-naheSedimentation palaozoischer 91-103. noiden-Fragmente:Neues Jahrbuchfur Geologie und Palionto- v. 56-100. MILLER, A.I., 1989,Lateral mixingand spatial resolutionin molluscan logie, Abhandlungen, 138, p. assemblages of Smuggler'sCove, St. Croix,U.S.V.I., in HUBBARD, SCHAFER,W., 1972, Ecology and Paleoecology of Marine Environ- D.K., ed., Terrestrialand Marine Geologyof St. Croix,U.S. Virgin ments: Universityof Chicago Press, 568 p. Islands: Fairleigh Dickinson University,West Indies Laboratory, SCHNEIDER,J.A., 1988, Frequency of arm regenerationof comatulid Special Publication, no. 8, p. 129-134. crinoidsin relationto lifehabit, in BURKE,R.D., MLADENOV,P.V,

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions MODERN CRINOID-RICHSEDIMENTS 573

LAMBERT,P., and PARSLEY,R.L. eds.,Echinoderm Biology: Bal- THISTLE,D., YINGST,J.Y., and FAUCHALD,K. 1985,A deep-sea benthic kema,Rotterdam, p. 531-538. communityexposed to strong near-bottom currents on the Scotian SCHOPF,T.J.M., 1980, Paleoceanography: Harvard University Press, Rise (Western Atlantic): Marine Geology,v. 66, p. 91-112. Cambridge,341 p. TSHUDY,D.T., FELDMANN,R.M., and WARD,P.D., 1989,Cephalopods: SIGNOR,P.W., and BRETT,C.E., 1984,The mid-Paleozoicprecursor Biasing agents in the preservationof lobsters: Journal of Pale- to theMesozoic marine revolution: Paleobiology v. 10,p. 187-204. ontology,v. 63, p. 621-626. SNEATH,P.H.A., and SOKAL,R.R., 1973, Numerical Taxonomy: W.H. VALENTINE,J.W., and PEDDICORD,R.G., 1967, Evaluation of fossil Freeman,San Francisco,573 p. assemblages by clusteranalysis: Journalof Paleontology,v. 41, p. SPEYER,S.E., and BRETT,C.E., 1986,Trilobite taphonomy and Mid- 502-507. dle Devoniantaphofacies: PALAIOS, v. 1, p. 312-327. VERMEIJ,G.J., 1977, The Mesozoic marinerevolution: Evidence from SPEYER,S.E., and BRETT,C.E., 1988,Taphofacies models for eperic snails,predators and grazers:Paleobiology, v. 3, p. 245-258. sea environments:middle Paleozoic examples: Palaeogeography, WALKER, S.E., 1988,Hermit crabs as taphonomicagents: PALAIOS, Palaeoclimatology,Palaeoecology, v. 63, p. 225-262. v. 4, p. 439-452. STAFF,G.M., and POWELL,E.N., 1988,The paleoecologicalsignifi- WARE,S., 1975, BritishLower Greensand Serpulidae: Palaeontology, cance of diversity:The effectof timeaveraging and differential v. 18,p. 93-116. preservationon macroinvertebratespecies richness in deathas- WARME,J.E., 1969, Live and dead molluscsin a coastallagoon: Jour- semblages:Palaeogeography, Palaeoclimatology, Palaeoecology, v. nal ofPaleontology, v. 43, p. 141-150. 63,p. 73-89. WATSON,M.K., and FLESSA,K.W., 1979,Time-averaged molluscan STAFF,G.M., and POWELL,E.N., 1990a,Taphonomic signature and assemblagesfrom Cholla Bay, Gulf of California:Geological So- theimprint of taphonomic history: Discriminating between tapho- cietyof America, Abstracts with Programs, v. 11,p. 134. faciesof the innercontinental shelf and a micro-tidalinlet, in WILBER, R.J.,1976, Petrology of submarine-lithified hardgrounds and MILLER,W., eds., PaleocommunityTemporal Dynamics: Pale- lithohermsfrom the deep flankenvironment of LittleBahamas ontologicalSociety, Special Publication, p. 370-390. Bank(northeastern Straits of Florida): Master's thesis, Duke Uni- STAFF,G.M., and POWELL,E.N., 1990b,Local variabilityof tapho- versity,241 p. nomicattributes in a parautochthonousassemblage: Can tapho- WILKINSON, C.R., and CHESHIRE,A.C., 1989,Patterns in the distri- nomicsignature distinguish a heterogeneous environment?: Jour- butionof spongepopulations across the centralGreat Barrier nal ofPaleontology, v. 64, p. 648-658. Reef:Coral Reefs, v. 8, p. 127-134. STAFF,G.M., STANTON,R.J., JR.,POWELL, E.N., and CUMMINS,H., WILKINSON,C.R., and EVANS,E., 1989,Sponge distribution across 1986, Time averaging,taphonomy and theirimpact on paleocom- Davies Reef,Great Barrier Reef, relative to location,depth and munityreconstruction: Death assemblages in Texas bays: Geo- watermovement: Coral Reefs, v. 8, p. 127-134. logical Society of America Bulletin, v. 97, p. 428-443. WILSON, J.L., 1975,Carbonate Facies in GeologicHistory: Berlin, STETSON,T.R., SQUIRES,D.F., and PRATT,R.M., 1962, Coral banks Springer-Verlag,471 p. occurringin deepwater on theBlake Plateau: American Museum WINER,B.J., 1971, Statistical Principles in ExperimentalDesign, 2nd Novitates,no. 2114.,39 p. edition.New York,McGraw-Hill, 1055 p. TAYLOR,P.D., SCHEMBRI,P.J., and COOK,P.L., 1989,Symbiotic as- sociationbetween hermit crabs and bryozoansfrom the Otago ACCEPTED APRIL 2, 1993 region,southeastern New Zealand: Journal of Natural History, v. 23,p. 1059-1085.

This content downloaded from 137.52.76.29 on Mon, 9 Jun 2014 17:10:16 PM All use subject to JSTOR Terms and Conditions