Community structure of Quaternary coral reefs compared ...marinepalaeoecology.org/.../2011/...Paleobiology.pdf · Paleobiology,27(4), 2001, pp. 669–694 Community structure of Quaternary

q 2001 The Paleontological Society. All rights reserved. 0094-8373/01/2704-0005/$1.00

Paleobiology, 27(4), 2001, pp. 669–694

Community structure of Quaternary coral reefs compared withRecent life and death assemblages

Evan N. Edinger, John M. Pandolfi, and Russell A. Kelley

Abstract.—This paper assesses the reliability with which fossil reefs record the diversity and com-munity structure of adjacent Recent reefs. The diversity and taxonomic composition of Holoceneraised fossil reefs was compared with those of modern reef coral life and death assemblages inadjacent moderate and low-energy shallow reef habitats of Madang Lagoon, Papua New Guinea.Species richness per sample area and Shannon-Wiener diversity (H9) were highest in the fossil reefs,intermediate in the life assemblages, and lowest in the death assemblages. The taxonomic com-position of the fossil reefs was most similar to the combination of the life and death assemblagesfrom the modern reefs adjacent to the two fossil reefs. Depth zonation was recorded accurately inthe fossil reefs. The Madang fossil reefs represent time-averaged composites of the combined lifeand death assemblages as they existed at the time the reef was uplifted.

Because fossil reefs include overlapping cohorts from the life and death assemblages, lagoonalfacies of fossil reefs are dominated by the dominant sediment-producing taxa, which are not nec-essarily the most abundant in the life assemblage. Rare or slow-growing taxa accumulate moreslowly than the encasing sediments and are underrepresented in fossil reef lagoons. Time-aver-aging dilutes the contribution of rare taxa, rather than concentrating their contribution. Conse-quently, fidelity indices developed for mollusks in sediments yield low values in coral reef deathand fossil assemblages. Branching corals dominate lagoonal facies of fossil reefs because they areabundant, they grow and produce sediment rapidly, and most of the sediment they produce is notexported.

Fossil reefs distinguished kilometer-scale variations in community structure more clearly thandid the modern life assemblages. This difference implies that fossil reefs may provide a better long-term record of community structure than modern reefs. This difference also suggests that modernkilometer-scale variation in coral reef community structure may have been reduced by anthropo-genic degradation, even in the relatively unimpacted reefs of Madang Lagoon. Holocene and Pleis-tocene fossil reefs provide a time-integrated historical record of community composition and maybe used as long-term benchmarks for comparison with modern, degraded, nearshore reefs. Com-parisons between fossil reefs and degraded modern reefs display gross changes in communitystructure more effectively than they demonstrate local extinction of rare taxa.

Evan N. Edinger.* Department of Earth Sciences, Laurentian University, Ramsey Lake Road, Sudbury,Ontario P3E 2C6, Canada

John M. Pandolfi. Department of Paleobiology, National Museum of Natural History, Smithsonian Insti-tution, Washington, D.C. 20560-0121

Russell Kelley. Watermark Films, Pty., Ltd., Post Office Box 1859, Townsville, Queensland 4810, Australia*Present address: Department of Geography, Memorial University of Newfoundland, St. John’s, Newfound-

land A1B 3X9, Canada

Accepted: 10 May 2001

Introduction

Paleontologists are constantly faced withthe question of the extent to which a fossil as-semblage resembles the original biologicalcommunity. This question is particularly im-portant for fossil reefs, which generally recordboth lateral and vertical facies zonation, andwhich may record temporal trends in com-munity structure akin to ecological succession(e.g., Walker and Alberstadt 1975; Copper1988). This study sheds light on the quality ofthe coral reef fossil record, in particular the re-liability with which it can be used to recon-

struct community structure for the original bi-ological community. We do this by comparingthe diversity and taxonomic composition ofHolocene fossil reefs with adjacent modernreef coral life and death assemblages.

This study also examines the taphonomicprocesses affecting coral reefs, particularlytime-averaging (Kowalewski 1996; Kidwell1998; Olszewski 1999). Time-averaging is aprimary constraint on the temporal and en-vironmental resolution of the fossil record(Johnson 1965; Kowalewski 1996; Kidwell1998). Time-averaging includes faunal con-

670 EVAN N. EDINGER ET AL.

densation (Fursich 1978), vertical mixing, cy-cling of bioclasts into and out of the taphon-omically active zone (Meldahl et al. 1997), andthe accumulation and mixing of successive co-horts of organisms occupying the same loca-tion at different times. These processes mayproduce different results for reefs than forshelly fossils in sediments (Scoffin 1992; Kid-well 1998; Best and Kidwell 2000; Zuchsin etal. 2000).

The temporal resolution of the fossil recorddepends on the relative rates of addition ofskeletons versus addition of sediment. Whereskeletal input exceeds sediment accumulation,either because of sediment starvation or be-cause of winnowing or erosion, the resultingfossil deposit averages the input of fossils overtime, and from various local environments(Kowalewski 1996). Time-averaging has beenstudied extensively in modern death assem-blages of mollusks (e.g., Kidwell and Bosence1991; Meldahl et al. 1997; Kowalewski et al.1998; see reviews in Kidwell and Flessa 1995,in Kidwell 1998, and in Behrensmeyer et al.2000) and in fossil assemblages of brachio-pods and other shelly fossils (e.g., Boucot1953; Copper 1997; Olszewski and West 1997),and it is pervasive in shelly fossil assemblagesin siliciclastic sedimentary environments.Time-averaging has received much less study,however, in carbonate environments, particu-larly on coral reefs (Scoffin 1992; Pandolfi andMinchin 1995; Greenstein and Pandolfi 1997;Pandolfi and Greenstein 1997a; Greenstein etal. 1998a; Zuchsin et al. 2000). For coral reefs,in which sediment accumulation rates aregenerally rapid and most of the containingsediment is composed of bioclasts derivedfrom the reef itself, faunal condensation bysediment starvation or winnowing is unlikely.Reefs may still exhibit time-averaging due tothe presence of multiple overlapping cohortsof corals and other reef-building organisms(Scoffin 1992). The extent to which time-aver-aging affects the quality of the coral reef fossilrecord is important because of the recent en-thusiasm for using the fossil record of Qua-ternary corals to gauge the effects of environ-mental change on living reefs (Aronson andPrecht 1997; Greenstein et al. 1998c; Pandolfi1999; Pandolfi and Jackson 2001).

How accurately do fossil reefs record the di-versity and taxonomic composition of the livereef communities from which they are de-rived? This study quantitatively examines thatquestion in a high-diversity system. Pandolfiand Minchin (1995) compared the taxonomiccomposition and diversity of modern life anddeath assemblages of Madang Lagoon, PapuaNew Guinea. Here, we extend the Madang La-goon work to the fossil record, by comparingthe life and death assemblages with Holocenefossil reefs in similar facies exposed immedi-ately adjacent to the Recent reefs studied byPandolfi and Minchin (1995). Although life,death, and Pleistocene assemblages on reefshave recently been compared in the Caribbean(Greenstein and Pandolfi 1997; Pandolfi andGreenstein 1997b; Greenstein et al. 1998a,b,c),there have been no previous direct modern-ancient comparisons from the vast Indo-Pacif-ic faunal province, which is far more diversethan the Caribbean (Veron 1993).

We examine the preservation of communitystructure in four ways. We compare speciesrichness, diversity, and evenness among thelife, death, and fossil assemblages. We com-pare the taxonomic composition of the life,death, and fossil assemblages at two numeri-cal scales of resolution: relative abundancesand presence or absence of taxa. We comparethe relative abundance of growth forms in thelife, death, and fossil assemblages, to deter-mine if different functional groups of reef-building corals are differentially preserved(cf. Fagerstrom 1987; Pandolfi and Greenstein1997a). Finally, we apply fidelity indices fromcomparisons of modern molluscan life anddeath assemblages to comparisons betweenthe coral reef life and fossil assemblages (Kid-well and Bosence 1991; Kidwell 2001).

Methods

Study Area and Sampling

Sampling of the modern reef coral life anddeath assemblages in Madang Lagoon, PapuaNew Guinea, was described by Pandolfi andMinchin (1995). The modern localities, Won-gat Island West (WIW) and Jais Aben Resort(JAR) represent low and intermediate waveexposure conditions, respectively. Wongat Is-

671HOLOCENE REEF DIVERSITY AND COMPOSITION

FIGURE 1. Map of Madang lagoon area, showing mod-ern and fossil transects. Modified from Pandolfi andMinchin 1995.

land is a coral cay at the outer edge of the Ma-dang Lagoon. The west side of Wongat Islandis the leeward protected side of one of the is-lands at the outer edge of the lagoon. Thefringing reefs near Jais Aben Resort are ex-posed to wind-driven waves developed with-in the lagoon only, a fetch of approximately 5km (Fig. 1).

The reefs of the Madang Lagoon area, andPapua New Guinea in general, have escapedmuch of the destruction visited on reefs else-where in southeast Asia, and no major ecolog-ical shifts have been observed. Water qualityremains good through most of the MadangLagoon, including over the fringing reefs ad-jacent to Jais Aben Resort. No blast or cyanidefishing has been reported from Madang La-goon. Similarly, there have been no reports ofwidespread coral diseases or dramatic chang-es in coral composition, as recorded in the Ca-ribbean (cf. Aronson and Precht 1997; Green-stein et al. 1998a). We caution, however, that

relatively little coral reef research has beenconducted in Papua New Guinea.

During 1992, the life and death assemblageswere sampled in five 30-m transects at each oftwo depths at these sites; ‘‘shallow’’ transects(1–2.5 m) were in the breaker zone, whereas‘‘deep’’ transects (2.5–4 m) were on a slightlylower shelf and were affected by wave surgemore than breaking waves. We use ‘‘sites’’ torefer to depths at a given locality (e.g., JARshallow) and ‘‘samples’’ to refer to a giventransect (e.g., JAR live shallow transect no. 1).

The fossil assemblage was sampled fromraised fossil reefs approximately 2300–1600years old, which are exposed in beachfrontcliffs 3–5 m high at Jais Aben, and 2–3 m highat Wongat Island (Tudhope et al. 2000). Wesampled the fossil assemblage using transectsplaced between 0.5 and 1 m above the hightide line. This position corresponds to the low-er portion of the lower coastal cliff unit, abovethe disconformity, as described by Tudhope etal. (2000: Fig. 4). These reef units grew duringa period of tectonic subsidence between twomajor co-seismic uplift events at approximate-ly 3000 calendar years BP and .1000 calendaryears BP. Total tectonic subsidence during thisinterval is estimated at approximately 2 m.Co-seismic uplift ranged between 0.5 m and4.5 m per uplift event (Tudhope et al. 2000).Preservation of the corals in the fossil assem-blage is excellent, with little or no diageneticalteration (Tudhope et al. 1997, 2000).

Life, death, and fossil assemblages weresampled with the line-point-intercept transectmethod (Loya 1978). Ten transects were mea-sured on the modern assemblages at each site,divided into five ‘‘shallow’’ and five ‘‘deep’’at each site (Pandolfi and Minchin 1995). Suf-ficient exposure of the fossil reefs allowed formeasurement of five fossil transects at JARand four fossil transects at WIW. All transectswere separated by at least 5 m. Where possi-ble, fossil transects were 20 m long, but sometransects were shorter, because exposure waslimited or discontinuous. Transect numbersand lengths are listed in Table 1. Point sam-ples were recorded every 20 cm along thetransect tape. The abundance of each taxon, orgrowth form, was recorded as the number ofpoints at which that taxon, or growth form, oc-


TABLE 1. Transect numbers, lengths (in m), and sampling intensities.

AssemblageNo. of

transects Average length Max. length Min. length Total m sampledAverage %coral cover

JAR live shallowJAR dead shallowJAR live deepJAR dead deepJAR fossil

55555

28.828.827.227.217.6

3030303020

2626151510

144144136136

88

22.03108.04

32.04230.95

55.64WIW live shallowWIW dead shallowWIW live deepWIW dead deepWIW fossil

55544

26.426.421.423.314.3

3030303020

2222111111

132132107

9357

25.37109.89

19.82127.65

53.87

curred. The death assemblages were mea-sured by recording whole dead corals on themodern transects and by counting coral rub-ble (.16 mm) collected from four grab sam-ples extending down 25 cm into the sediment(Pandolfi and Minchin 1995). The orientationof corals on all transects was recorded.

Corals were identified to the species levelwhere possible, and otherwise to genus, fol-lowing identifications of Veron and Pichon(1976, 1980, 1982), Veron et al. (1977), and Ve-ron and Wallace (1984). Exceptions are somemassive faviid corals that could not be iden-tified to genus, and the small free-living fun-giid (mushroom) corals, which were treated asone taxon. In general, taxon identification wasequally reliable in all three assemblages, withthe exception of rare Acropora species, whichwere collected from the life assemblages anddetermined by C. C. Wallace (Museum ofNorth Queensland).

To determine the adequacy of sampling, sep-arate species-sampling curves were construct-ed for the life, death, and fossil assemblages.These species-sampling curves were then log-transformed to species-log(sampling) lines, theslopes of which were calculated using linear re-gression, and compared using the 95% confi-dence limits on each regression line.

Data Analysis

We compared species richness, Shannon-Wiener diversity, and Pielou’s evenness amonglife, death, and fossil assemblages at each siteusing one-way analysis of variance (ANOVA),after verifying normality and homogeneity ofvariance. Species richness (S) was recorded asthe number of coral species recorded per tran-

sect. Because transects were not all of equiv-alent length, species richness was normalizedto transect length, which we term S9. Shan-non-Wiener diversity (H9) and Pielou’s even-ness (J9) were calculated from relative abun-dance data for each species on each transect.Percent similarity of taxon presence/absencebetween the total life, death, and fossil assem-blages was determined according to the Jac-card similarity index.

We used multivariate analysis techniquesidentical to those of Pandolfi and Minchin(1995) to facilitate comparisons between thefossil assemblage and the life and death as-semblages. Briefly, we compared the compo-sition of taxa and of growth forms in the life,death, and fossil assemblages using the statis-tical test of analysis of similarity (ANOSIM)(Clarke 1993) and visually by ordination usingnon-metric multidimensional scaling (MDS).These methods compare the ranked Bray-Cur-tis similarity (or dissimilarity) of the taxa ineach sample (Bray and Curtis 1957). The AN-OSIM test compares the ranked square-roottransformed abundances of taxa in each sam-ple (i.e., transect), standardized to the maxi-mum abundance obtained by each taxon in thedata set (Clarke 1993). Both ANOSIM andMDS were repeated for relative abundance ofcoral taxa, presence/absence of coral taxa,and relative abundance of coral growth forms(see Rahel 1990). Replicate transects withineach assemblage allowed assessment of vari-ation within assemblages (between transects)and between assemblages (life, death, and fos-sil assemblages from different water depthsand localities). ANOSIM analyses of the fossilreefs versus modern life and death assemblag-


es indicate significance or nonsignificance.Fossil reefs are closest in composition to thesubset of modern assemblages that yield non-significant ANOSIM results for taxon relativeabundance, presence/absence, and growthform relative abundance. Thus nonsignificantANOSIM results imply a potential match,whereas significance rules out a given modernassemblage as the match for the fossil assem-blage.

For further visual comparisons of the rela-tive abundance of growth forms of corals ineach assemblage, the relative abundance ofeach growth form in each assemblage wassummed and normalized to total cover of livecorals or, for death and fossil assemblages, to-tal number of corals recorded. These datawere used to produce a ternary diagram ofrelative abundance of growth forms (see Edin-ger and Risk 1999), plotting one average com-position per assemblage and depth.

The MDS and ANOSIM analyses were fol-lowed by a similarity percentage analysis(SIMPER) (Clarke 1993) to determine whichtaxa were responsible for the greatest similar-ity within assemblages, and which were mostresponsible for dissimilarity among assem-blages. Within an assemblage, those taxa forwhich the ratio of mean similarity to standarddeviation of similarity is .1 typify the samplegroup (Clarke 1993). Among assemblages, wemeasured the average dissimilarity in abun-dance of each taxon between assemblages andthe standard deviation of that dissimilarity,and then we calculated the percent contribu-tion of each taxon to the total dissimilarity be-tween the two assemblages. Those taxa with aratio of mean dissimilarity to standard devi-ation of dissimilarity .1 are the taxa that re-liably discriminate among assemblages(Clarke 1993). Only taxa with a ratio of aver-age dissimilarity/SD dissimilarity .1 are list-ed in comparisons between the fossil reef andthe modern life or death assemblages (see Ap-pendix 1).

We used four fidelity indices to estimate thefidelity of the fossil assemblage with respectto the life assemblage (Kidwell and Bosence1991; Kidwell 2001). First we used the threeindices presented in Kidwell and Bosence(1991) to facilitate comparison with coral life

assemblage-death assemblage fidelity (Pan-dolfi and Minchin 1995): the percentage of livetaxa also found in the fossil assemblage (per-cent live found fossil), the percentage of fossiltaxa found in the life assemblage (percent fos-sil found live), and the percentage of individ-uals of species found in the fossil assemblagethat were also found in the life assemblage(percent fossil individuals found live). Num-bers of individuals encountered in the life as-semblages were corrected for sampling inten-sity by dividing the number of individuals bythe ratio of total live transect length to totalfossil transect length. Numbers of individualswere further corrected for live coral cover bymultiplying the number of live individualsencountered by the ratio of fossil coral abun-dance to live coral cover for the site in ques-tion. Both uncorrected and corrected resultsare presented. The contribution of rare taxa tofidelity was also examined. Rare taxa were de-fined as those contributing less than 1% to thetotal abundance of corals in the life assem-blage or fossil assemblage (see Kidwell 2001).We recalculated all three fidelity indices forcommon taxa only, applying the same correc-tion factors for total length of transect.

Last, a fourth fidelity index was applied, therank-order of taxon abundance (Kidwell2001). Spearman-rank correlations betweentaxon relative abundance in life, death, andfossil assemblages were calculated for all taxa,for common taxa only (taxa contributing morethan 1% of the life assemblage), and for coralgrowth forms.

Results

Sampling

Species-sampling curves for most modernsites and assemblages approached horizontal-ity by the fifth transect, indicating generallyadequate sampling (Fig. 2). Exceptions werethe life assemblage transects at WIW and theshallow life assemblage at JAR. Both fossilsites approached horizontality by the fourthtransect. Substitution of transect lengths fornumber of transects did not alter the shape ofthe curves. The slopes of the fossil assemblagespecies-log(sampling) lines were not signifi-cantly greater than those of the life assem-


FIGURE 2. Species-sampling curves. A, Jais Aben Resort(JAR). B, Wongat Island West (WIW).

blages at JAR or WIW, but they were signifi-cantly greater than those of both shallow anddeep death assemblages at JAR, and of theshallow death assemblage at WIW (Table 2).

A total of 64 coral taxa were recorded in the

pooled life, death, and fossil assemblages. Thegreatest numbers of taxa were recorded in thelife assemblages, (JAR: 39 spp.; WIW: 29; 51total), followed by the fossil assemblages(JAR: 28; WIW: 20; 34 total) and the death as-semblages (JAR: 21; WIW: 19; 34 total). Thefossil assemblage contained 72% of the num-ber of taxa represented in the life assemblage.When sampling intensity is measured in me-ters of transects measured, total fossil reefsampling intensity was 31% of live samplingintensity at JAR, and 23% of live sampling in-tensity at WIW.

Species Richness, Shannon-Wiener Diversity,and Evenness

Species richness normalized for transectlength (S9) was significantly greater in fossilassemblages than in life assemblages or deathassemblages for all samples when localities(i.e., JAR, WIW) are pooled (F(2,45) 5 9.94, p ,0.0005), and for both JAR alone (F(2,22) 5 9.88,p , 0.001, Fig. 3A) and WIW alone (F(2,20) 57.83, p 5 0.003). Average Shannon-Wiener di-versity of fossil assemblages was consistentlyhigher than in life or death assemblages (JAR:F(2,22) 5 39.46, p , 0.0001; WIW: F(2,20) 5 18.76,p , 0.0001; Fig. 3B). Species richness (S9) andShannon-Wiener diversity (H9) were higher inthe fossil assemblages than in the life assem-blages partly because more individual coralswere encountered per unit length on the fossilreefs than on the modern reefs. Average livecoral cover in the life assemblage transectswas 28%, whereas 58% of the length of the fos-sil assemblage transect intercepted corals.There was very little sediment between mostindividual fossil corals. Pielou’s evenness (J9),however, was consistently highest in life as-semblages, closely followed by the fossil as-semblages, and was much lower in the deathassemblages (JAR: F(2,22) 5 49.56, p , 0.0001;WIW: F(2,20) 5 36.36, p , 0.001; Fig. 3C).

The taxon presence/absence Jaccard simi-larity between the total life assemblage andtotal fossil assemblage was 59%. Jaccard sim-ilarity was 62% between the total death as-semblage and total fossil assemblage, and 74%between the life and death assemblages.


TA

BL

E2.

A.

Slop

e,95

%co

nfi

den

celi

mit

s,r2

valu

es,a

nd

sig

nifi

can

ceof

log

-tra

nsf

orm

edsp

ecie

s-sa

mp

lin

gcu

rves

.Hig

her

slop

esre

pre

sen

tm

ore

div

erse

asse

mbl

ages

.

Ass

embl

age

Slop

eL

ower

95%

lim

itU

pp

er95

%li

mit

r2p

JAR

live

shal

low

JAR

dea

dsh

allo

wJA

Rli

ved

eep

JAR

dea

dd

eep

JAR

foss

il

10.1

88.

9918

.23

9.77

14.5

6

5.07

4.40

10.5

66.

527.

63

15.2

913

.58

25.9

013

.01

21.4

9

0.88

0.88

0.92

0.95

0.89

0.00

50.

006

0.00

30.

001

0.00

4W

IWli

vesh

allo

wW

IWd

ead

shal

low

WIW

live

dee

pW

IWd

ead

dee

pW

IWfo

ssil

13.1

18.

0510

.72

9.77

11.1

8

8.99

5.82

6.27

7.14

7.08

17.2

310

.28

15.1

612

.41

15.2

8

0.95

0.96

0.92

0.98

0.96

0.00

090.

0006

0.00

30.

001

0.00

3

B.

Sig

nifi

can

t-d

iffe

ren

cem

atri

xfo

rsl

opes

oflo

g-t

ran

sfor

med

spec

ies-

sam

pli

ng

curv

es.

Liv

esh

allo

w

JAR

WIW

Dea

dsh

allo

w

JAR

WIW

Liv

ed

eep

JAR

WIW

Dea

dd

eep

JAR

WIW

Foss

il

JAR

WIW

Liv

esh

allo

wD

ead

shal

low

Liv

ed

eep

Dea

dd

eep

Foss

il

— ns * ns

ns

— * ns

ns

ns

— * ns *

— ns

ns *

— * ns

— ns

ns

— *— *

——

*p

,0.

05.


FIGURE 3. A, Species richness/area surveyed (S9). B,Shannon-Wiener diversity (H9). C, Evenness (J9) for life,death, and fossil assemblages, separated by site. Means6 95% confidence limits.

Taxonomic Composition of Life, Death, andFossil Assemblages

Relative Abundance of Coral Taxa. The fossilassemblages at each location were not signif-icantly different from their respective life as-semblages when depths were pooled, but theywere significantly different from their respec-tive life assemblages when depths were sep-arated (Table 3A). At JAR, the fossil assem-blage was not significantly different from thecombined deep life and deep death assem-blages, and from the life and death assem-blages with depths pooled (Table 3A). AtWIW, the fossil assemblage was not signifi-cantly different from the shallow death assem-blage, or from the combined life and death as-semblages, with depths either separated orpooled (Table 3A).

The fossil assemblages at the two localitieswere significantly different from each other,but the life assemblages were not (Table 3B).Average similarities between life and deathassemblages, within a reef or between the twolocations, were less than the similarity be-tween the two fossil assemblages (Table 3B).

On the taxon relative abundance MDS plots(Fig. 4A,B), fossil assemblages were much lessvariable than either life or death assemblages.The fossil assemblage at JAR appeared closestto a combination of the deep life and deepdeath assemblages. At JAR, the fossil assem-blage largely overlapped with both the deeplife and deep death assemblages, but not withthe shallow life or shallow death assemblages.The fossil assemblage at WIW appeared dis-tinct but was closest to the shallow life assem-blage and, to a lesser extent, to the highly var-iable shallow death assemblage, again agree-ing with the ANOSIM results. The WIW fossilassemblage did not directly overlap with anyof the deep life or death assemblages.

Presence/Absence of Coral Taxa. The fossilassemblage at JAR was not significantly dif-ferent from the life or death assemblages withdepths pooled, or from the combined life anddeath assemblages with depths separated orpooled (Table 4A). The fossil assemblage atWIW was not significantly different from theshallow life assemblage, the combined shal-low life and shallow death assemblages, the


TABLE 3. Taxon relative abundances.

A. ANOSIM resulsts.

Assemblage

JAR fossil vs.

Shallow Deep Shallow 1 deep

WIW fossil vs.


Life R 5 0.600p 5 0.008*

R 5 0.416p 5 0.008*

R 5 0.052p 5 0.324

R 5 0.362p 5 0.024*

R 5 0.400p 5 0.024*

R 5 0.196p 5 0.093

Death R 5 0.418p 5 0.016*

R 5 0.426p 5 0.008*

R 5 0.095p 5 0.238

R 5 0.219p 5 0.063

R 5 0.906p 5 0.024*

R 5 0.308p 5 0.024*

Life 1 death R 5 0.265p 5 0.036*

R 5 20.008p 5 0.467

R 5 0.125p 5 0.797

R 5 0.006p 5 0.434

R 5 0.299p 5 0.064

R 5 0.061p 5 0.33

B. Average % similarity among sample types, depths pooled, using taxon relative abundance data.

Group JAR live JAR dead JAR fossil WIW live WIW dead WIW fossil

JAR liveJAR deadJAR fossilWIW liveWIW deadWIW fossil

—26.61*32.75ns28.20ns27.33***24.80**

—29.82ns21.88***30.38ns20.71ns

—28.50ns37.15ns41.60*

—29.56*26.09ns

—28.95ns —

* p , 0.05.** p , 0.01.

*** p , 0.001.

life assemblage with depths pooled, or thecombined life and death assemblages withdepths pooled (Table 4A). Neither life nor fos-sil assemblages were significantly differentbetween localities (Table 4B).

The taxon presence/absence MDS plot forJAR showed some separation between life anddeath assemblages, with the fossil assemblag-es falling between the two, but overlappedonly with the deep life assemblage (Fig. 5A).The fossil assemblage at JAR was intermediateamong all the life and death assemblages butonly directly overlapped with the highly var-iable shallow death assemblage (Fig. 5A). Thefossil assemblage at WIW overlapped with allthe life and death assemblages except the deepdeath assemblage and was closest to the shal-low life assemblage and the shallow death as-semblage (Fig. 5B).

Growth Forms

Ternary Diagram. Branching corals, includ-ing both fine and stout branching forms,formed the largest constituent in all assem-blages except the shallow life assemblage atJAR, where branching corals and massive plusplaty corals were equally abundant (Fig. 6).Branching corals were most abundant in thedeath assemblages, intermediate in fossil as-semblages, and least abundant in life assem-

blages. Death assemblages were consistentlydominated by branching corals, and in mostcases, branching corals composed more than95% of the corals encountered in the death as-semblages. In growth form composition, thefossil assemblages were intermediate betweenthe life and death assemblages but more close-ly matched the life assemblages than the deathassemblages. Encrusting, free-living, andfrondose corals were underrepresented in thefossil assemblages, relative to the life assem-blages, probably as a result of the superabun-dance of branching corals.

Ordination and Analysis of Similarity. AN-OSIM showed that in growth form composi-tion the fossil assemblage at JAR was not sig-nificantly different from the deep life assem-blage, from the combined deep life and deepdeath assemblages, or from the life assem-blage with depths pooled (Table 5A). The fos-sil assemblage at WIW was not significantlydifferent in growth form composition fromthe combined shallow life and shallow deathassemblages (Table 5A). The growth formcompositions of the fossil reefs at the two lo-calities were more similar to each other thaneither was to its respective life or death assem-blage with depths pooled (Table 5B).

The growth forms MDS plots showed thatat JAR, the fossil assemblage was most similar


FIGURE 4. 3-D MDS plots of taxon relative abundances,depths separated in life and death assemblages. A, JaisAben Resort (JAR), stress 5 0.11. B, Wongat Island West(WIW), stress 5 0.10. Fossil assemblages show less var-iability than life or death assemblages.

to the death assemblages (Fig. 7A). Likewise,at WIW, the fossil assemblage was closest tothe deep death assemblage, followed by thedeep life assemblage, and plotted intermedi-ate between these two assemblages.

Characteristic and Discriminating Taxa

Characteristic Taxa in Each Assemblage. Atboth JAR and WIW, the characteristic taxa ofthe fossil assemblages included one or both ofthe characteristic taxa of the life and death as-semblages (Table 6, Fig. 8). Analysis at thepresence/absence scale revealed the samecharacteristic taxa as analysis at the relativeabundance scale. More taxa characterized thefossil assemblages than the life or death as-semblages because the variation in taxon rel-ative abundances between transects within as-semblages was lower for the fossil assemblag-es than for the life or death assemblages.

Discriminating Taxa. At JAR, a subset of thetaxa that distinguished the fossil and shallowlife assemblages also distinguished the fossiland shallow death assemblages (Tables A, B ofAppendix 1). Likewise, most of the same taxadistinguished the fossil from deep life anddeep death assemblages, and the discriminat-ing taxa between the fossil and deep death as-semblages were a subset of those discriminat-ing between the fossil and deep life assem-blages (Tables C, D of Appendix 1). The dis-criminating taxa at JAR were mostly the sametaxa that distinguished the fossil assemblageat WIW from the WIW shallow life, deep life,and death assemblages (Tables E, F, G of Ap-pendix 1).

Results of dissimilarity analyses generallyagreed between the relative abundance andpresence/absence scales of analysis. The mostcommon taxa, Acropora spp. (branching) andPorites spp. (massive) usually contributed sig-nificantly to dissimilarity at the relative abun-dance scale, but not at the presence/absencescale. Differences in relative abundances, par-ticularly of branching corals, contributed farmore to dissimilarity between the fossil as-semblage and death assemblage than did taxathat only occurred in one or the other of thesetwo assemblages.

Rare taxa usually occurred on fewer thanfive of the total 48 transects, and did not gen-

erally discriminate between sample groups,particularly live from fossil (Appendix 2, Fig.8). Exceptions are Anacropora sp., Pocilloporasp., Porites lichen, and Galaxea sp. Anacropora sp.was the second most abundant taxon in thedeep death assemblage at JAR, where it oc-


TABLE 4. Taxon presence/absence.

A. ANOSIM results.

Assemblage

JAR fossil vs.


WIW fossil vs.


Life R 5 0.434p 5 0.008*

R 5 0.300p 5 0.032*

R 5 0.065p 5 0.284

R 5 0.269p 5 0.071

R 5 0.581p 5 0.008*

R 5 0.232p 5 0.088

Death R 5 0.308p 5 0.008*

R 5 0.438p 5 0.008*

R 5 0.105p 5 0.214

R 5 0.400p 5 0.024*

R 5 0.531p 5 0.029*

R 5 0.413p 5 0.01*

Life 1 death R 5 0.212p 5 0.075

R 5 0.115p 5 0.181

R 5 0.063p 5 0.638

R 5 0.229p 5 0.086

R 5 0.343p 5 0.038*

R 5 0.185p 5 0.142

B. Average % similarity among sample types, depths pooled, using taxon presence/absence data.



—29.85*37.08ns32.75ns31.74**27.13**

—31.89ns26.24**32.33ns21.69*

—31.98ns39.06ns41.84ns

—34.13ns27.95ns

—28.82** —

* p , 0.05.** p , 0.01.

*** p , 0.001.

curred on three transects. Anacropora contrib-uted 5.1% to the total dissimilarity betweenthe deep death assemblage and fossil assem-blage at JAR, at the presence/absence level,but it was not an effective discriminating tax-on at the relative abundance level. Pocilloporasp. and Porites lichen were discriminating taxaat WIW, and Galaxea sp. was a discriminatingtaxon at both JAR and WIW.

Fidelity Indices

Fewer than half of the coral taxa recorded inthe life assemblages were also recorded in thefossil assemblages, and only 60–71% of taxarecorded in the fossil assemblages were alsorecorded in the life assemblages (Table 7). Wecaution, however, that double the number offossil transects, and roughly triple (JAR) orquadruple (WIW) the length of fossil transectswere sampled in the life assemblages. Thusthe numbers of fossil individuals found in thelife assemblages were usually more than 100%(range 87–122%). When corrected for sam-pling intensity and the frequency with whichcorals were recorded on each type of transect,the percentage of fossil individuals found inlife assemblages was reduced to 50–76%. Fi-delity indices were uniformly higher at JARthan at WIW. The fidelity index results com-paring the life and fossil assemblages for JAR

and WIW combined were roughly comparableto the life assemblage/death assemblage re-sults (Pandolfi and Minchin 1995).

Most of the taxa found in the life assem-blage, but not in the fossil assemblage, wererare, contributing less than 1% to the life as-semblage (Fig. 8). Of 51 species recorded inthe modern life and death assemblages, 25were rare species. Eighteen of 34 species re-corded in the two fossil assemblages contrib-uted less than 1% of the total fossil assem-blage. When rare taxa were excluded from theanalyses, the fidelity indices comparing num-bers of species all rose, but the index compar-ing numbers of individuals declined, for bothsites pooled and separated (Table 7). Rare taxain the life assemblages included nine Acroporaspecies that would not have been distinguish-able in fossil material and that may have beenrecorded in the fossil assemblage as Acroporaspp.

The rank-order of taxon abundances of alllive taxa was significantly correlated with thatin the death assemblage for JAR and WIWboth pooled and separated (Table 8). This cor-relation was weaker, but still significant, forcommon taxa at JAR and for the two localitiespooled, but not for WIW alone. Growth formranked abundances were strongly and signif-icantly correlated between life and death as-


FIGURE 5. 3-D MDS plots of taxon presence/absence,depths separated in life and death assemblages. A, JaisAben Resort (JAR), stress 5 0.13. B, Wongat Island West(WIW), stress 5 0.11. Symbols as in Figure 4. Differenc-es between life and death assemblages are more pro-nounced than differences between depth (shallow vs.deep), in contrast to the pattern observed for taxon rel-ative abundances (Fig. 4).

semblages at both localities, pooled and sep-arated. By contrast, rank-order abundances oflive taxa were significantly correlated withthose in the fossil assemblages at JAR, but notat WIW, for all taxa, common taxa only, andcoral growth forms. Rank order abundancesof taxa in the death assemblage were signifi-cantly correlated with rank-order abundancesin the fossil assemblage at JAR, but not WIW,for all taxa and common taxa, but not for coralgrowth forms. In general, correlation coeffi-

cients between fossil assemblages and life as-semblages were lower than those between fos-sil assemblages and death assemblages forcoral taxa, but higher than between fossil as-semblages and death assemblages for coralgrowth forms.

The life assemblage-fossil assemblage fidel-ity indices from this study were quite differentfrom those typical of life assemblage-death as-semblage comparisons for molluscan assem-blages in sediments. The percentage of livespecies found fossil was much lower than lifeassemblage/death assemblage fidelity frommollusks in sediments (Kidwell and Bosence1991; Kidwell and Flessa 1995; Zuchsin et al.2000), and comparable to the life assemblage/death assemblage fidelity from Madang La-goon (Pandolfi and Minchin 1995). By contrast,the percentage of fossil species found live wasmuch higher than that typical of mollusks insediments, and lower than that the percentageof dead species found live in Madang Lagoon.The percentage of fossil individuals found livewas much lower than the percentage of deadindividuals found live (Pandolfi and Minchin1995), when corrected for sampling intensityand coral cover. When rare taxa are excluded,the life assemblage/fossil assemblage fidelityindices from Madang Lagoon improved, butthey were still much lower than fidelity indicestypical of mollusks in sediments (Kidwell andFlessa 1995; Kidwell 2001). Five of six Spear-man-rank correlations between rank abun-dances of life and death assemblages were sig-nificant, at all three levels, but this pattern wasweaker between life and fossil assemblagesand did not hold between death and fossil as-semblages.

Discussion

We seek to understand the composition ofthe fossil assemblage in terms of depth andrelative contribution of life and death assem-blages. The fossil assemblages most closely re-semble those life or death assemblages forwhich ANOSIM analyses of the growth formsand both levels of taxonomic composition allindicate a lack of significant differences. Theresults for JAR are consistent with the fossilassemblage comprising either the combineddeep life and deep death assemblage, or the


FIGURE 6. Ternary diagram of coral growth form composition on life assemblages, death assemblages, and fossilassemblages, depths separated in life and death assemblages. Each point represents the average of all transectsrecorded in that assemblage. Growth form categories include: fine-branching (mainly Acropora, Stylophora, Pocillo-pora, Seriatopora), stout branching (mainly Acropora, Porites cylindrica), massive, encrusting, platy (e.g., Pachyseris,tabular Acropora), frondose (e.g., Turbinaria, Pavona), columnar, and free-living (mainly Fungia). End-members ofthe ternary diagram are (1) branching corals, (2) massive and platy corals, and (3) encrusting, frondose, free-living,and columnar corals. Most death assemblages are overwhelmingly dominated by branching corals, and fossil as-semblage compositions are intermediate between life and death assemblage compositions.

TABLE 5. Growth forms relative abundance.

A. ANOSIM results.

Assemblage

JAR fossil vs.


WIW fossil vs.


Life R 5 0.608p 5 0.016*

R 5 0.128p 5 0.159

R 5 0.170p 5 0.096

R 5 0.608p 0.008*

R 5 0.506p 5 0.016*

R 5 0.332p 5 0.037*

Death R 5 0.908p 5 0.008*

R 5 0.948p 5 0.008*

R 5 0.974p 5 0.0001*

R 5 0.563p 5 0.016*

R 5 1.000p 5 0.029*

R 5 0.705p 5 0.003*

Life 1 death R 5 0.352p 5 0.013*

R 5 0.194p 5 0.06

R 5 0.166p 5 0.024*

R 5 0.294p 5 0.062

R 5 0.487p 5 0.006*

R 5 0.309p 5 0.045*

B. Average % similarity among sample types, depths pooled, using growth form relative abundance data.



—35.69***49.45ns49.63ns41.73***44.75***

—46.05ns36.13***60.80ns40.90***

—49.65ns50.56***65.26ns

—43.40**45.29*

—44.34* —

* p , 0.05.** p , 0.01.

*** p , 0.001.


FIGURE 7. 3-D MDS plots of growth forms, depths sep-arated in life and death assemblages. A, Jais Aben Re-sort (JAR), stress 5 0.05. B, Wongat Island West (WIW),stress 5 0.05. Symbols as in Figures 4, 5. Note that thefossil assemblage at JAR is most similar to death assem-blages, whereas at WIW, the fossil assemblage was mostsimilar to the deep life and deep death assemblages.

pooled shallow life and deep life assemblage(Table 9). We favor the interpretation that thefossil assemblage at JAR represents the com-bined deep life and deep death assemblages(see depth zonation and paleobathymetry, be-low). The fossil assemblage at WIW is com-posed of the combined shallow life and shal-low death assemblage (Table 9). Our resultshave important implications for depth zona-tion and paleobathymetry in fossil reefs, thenature of time-averaging on fossil reefs, theapplication of fidelity indices to fossil reefs,

and the utility of fossil reefs in reef conser-vation research.

The Holocene fossil reefs at Madang provid-ed an accurate overall reflection of the taxo-nomic composition of the shallow-water coralcommunity as it accumulated in the sediment:a depth-specific combination of the life anddeath assemblages. This pattern is similar tothat observed in Caribbean sites that have notundergone radical shifts in community com-position (e.g., Greenstein et al. 1998b; Pandolfi1999; Pandolfi and Jackson 2001). The patternsof growth form, diversity, and taxonomic com-position observed in the Holocene fossil reefsin Madang imply that on coral reefs, the deathassemblage is not a progressive precursor tothe fossil assemblage. Rather, the fossil assem-blage is formed from the life assemblage, withadmixture of material from the co-occurringand surrounding death assemblage (see Green-stein et al. 1998a). As such, the Holocene raisedreefs in Madang represent short-term compos-ites of the corals on the reef, both live and dead,at the time the reef was preserved. Such com-posites may be a general phenomenon forraised Pleistocene and Holocene reefs, and per-haps for fossil reefs in general (Greenstein et al.1998b).

Depth Zonation and Paleobathymetry

Of the two possible interpretations of theJAR fossil assemblage based on ANOSIM re-sults (Table 9), we favor the interpretation thatthe deep life and deep death assemblages havebeen combined to form the fossil assemblagefor four reasons. (1) Age and height data fromfossil corals at both Jais Aben Resort and Won-gat Island indicate reef growth during tectonicsubsidence, followed by preservation duringrapid co-seismic uplift (Tudhope et al. 2000).(2) The fossil reef cliff at JAR is 1–3 m higherthan that at WIW, and we sampled the lowerpart of the cliff at each location. Vertical zo-nation was evident within the cliff at both sites(Tudhope et al. 2000). The basal portions ofthe lower coastal cliff unit at JAR are domi-nated by branching coral thickets, indicatinggrowth in shallow subtidal conditions, where-as the upper portions include massive coralmicroatolls, indicating growth at sea level(Tudhope et al. 2000). (3) The MDS results are


TABLE 6. Characteristic taxa for each group, taxon abundances. Characteristic taxa are those for which the ratio ofaverage abundance to standard deviation of abundance is .1. Abundances are double square-root transformedcounts.

SiteAssem-blage Taxon

Growthform

Averageabundance Avg./SD

JAR Life Porites sp.Faviidae

MM

4.42.4

1.111.08

Death Acropora spp. B 35.8 1.17

Fossil Acropora spp.Porites sp.Millepora sp.Montipora digitataGoniastrea retiformis

BMBBM

16.48.64.84.62.0

2.644.073.001.021.12

WIW Life Porites sp.Acropora spp.

MB

5.13.9

1.231.14

Death Seriatopora sp.Acropora spp.

BB

64.7815.67

1.711.70

Fossil Acropora spp.Stylophora sp.Porites sp.

BBM

18.753.254.5

5.872.133.6

more consistent with the combined life anddeath assemblage interpretation than with themixed depths interpretation. (4) The alterna-tive interpretation of the ANOSIM results,that the fossil reef represented the combinedshallow life and deep life assemblages, couldonly have occurred if the reef were raisedgradually from the deep zone, through theshallow zone, to its current raised position.Such a prolonged uplift would have left anabrasion and encrustation record on the fossilcorals, which we did not observe. Fossil coralassemblages composed mainly of branchingcoral rubble, showing intense abrasion or highincidence of encrustation and boring, wouldbe indicators of slow regression and pro-longed reworking in the intertidal zone (seeGreenstein and Moffat 1996; Pandolfi andGreenstein 1997a; Greenstein et al. 1998a).

Our data support the widely recognizedpattern that fossil reefs can reliably preservedepth zonation and depth-related differencesin species composition. Vertical zonation offossils within single outcrops of fossil reefs iscommon (e.g., Copper 1988; Perrin et al. 1995).In the Caribbean, it is possible to distinguishdepth zones in fossil reefs using the relativeabundance of certain species such as Acroporapalmata and A. cervicornis; moreover, the dif-

ferences in depths between zones are greaterthan for the zones we studied in the MadangLagoon (Mesolella 1967; Perrin et al. 1995;Pandolfi and Jackson 2001). Likewise, on Pleis-tocene reefs of the Huon Peninsula, depth dif-ferences between zones are greater than thoseobserved in Madang Lagoon (Chappell 1974;Pandolfi 1996).

Time-Averaging on Fossil Reefs

The diversity and species composition ofthe fossil reefs, compared with those of mod-ern assemblages, suggest that the Madang fos-sil reefs are time-averaged. The radiometricages of corals within the lower coastal cliffunit at Jais Aben indicate growth of this reefunit for a minimum duration of 813 years anda maximum of 1584 years, corresponding toaverage rates of reef accretion of 2.2 mm/yrand 1.1 mm/yr, respectively (Tudhope et al.2000). These durations of reef growth repre-sent maximum durations of time-averaging;corals that are buried beneath the depth of thetaphonomically active zone would not be sub-ject to reworking for the full duration of reefgrowth during subsidence, or upon uplift (Ol-szewski 1999).

Actual duration of time-averaging is esti-mated from the generation times of branching


FIG

UR

E8.

His

tog

ram

sof

ind

ivid

ual

spec

ies

aver

age

abu

nd

ance

sin

asse

mbl

ages

.Abu

nd

ance

sd

oubl

esq

uar

e-ro

ottr

ansf

orm

ed.


TABLE 7. Fidelity indices, life assemblage/fossil assemblage, following Kidwell and Bosence 1991. For numbers offossil individuals found in life assemblage, sampling intensity was standardized to cumulative length of transects.Common taxa defined as those contributing more than 1% of the total abundance to the life or fossil assemblages.

Taxa and location(s)sampled

% Live speciesfound in fossil

assemblage

% Fossil speciesfound in liveassemblage

% Fossil individualsfound in lifeassemblage

Corrected % fossilindividuals found

in lifeassemblage*

All taxaJAR 1 WIWJAR onlyWIW only

473935

716460

122107

87

766950

Common taxa onlyJAR 1 WIWJAR onlyWIW only

615532

817575

1039289

645951

Death assemblage(Pandolfi Minchin 1995)

% Live founddead % Dead found live

% Dead individualsfound live

Corrected % fossilindividualsfound live

All sites pooled 54 90 94 n/a

* Number of life assemblage individuals corrected for differences in sampling intensity between life assemblages and fossil assemblages, and for coralcover.

TABLE 8. Spearman-rank correlations of ranked abundances of all taxa, common taxa (.1% of life assemblage),and growth forms between life, death, and fossil assemblages.

Taxa and location(s)sampled life/death life/fossil death/fossil

All taxaJAR 1 WIW (n 5 64)JAR only (n 5 57)WIW only (n 5 48)

0.63***0.65***0.57***

0.29*0.31*0.06ns

0.39**0.43***0.21ns

Common taxa onlyJAR 1 WIW (n 5 24)JAR only (n 5 23)WIW only (n 5 20)

0.52**0.64***0.18ns

0.32ns0.58**0.25ns

0.48*0.62**0.18ns

Growth formsJAR 1 WIW (n 5 9)JAR only (n 5 8)WIW only (n 5 7)

0.87**0.84*0.82*

0.72*0.74*0.14ns

0.52ns0.45ns0.17ns

* p , 0.05.** p , 0.01

*** p , 0.001.

corals and from the size and growth rates ofmassive corals. Each stratigraphic unit prob-ably averages 5–10 generations of branchingcoral thickets, with average generation timesof branching corals ranging from 10 to 20years (Scoffin 1992; Hubbard 1997). Similarly,massive corals encountered in the fossil reefsranged from ,10 cm to ;2 m in diameter, cor-responding to ages of 10–150 years, at averagegrowth rates of 10–15 mm/yr (Buddemeierand Kinzie 1976; Scoffin et al. 1992). The prod-ucts of these ranges yield an estimated dura-tion of time averaging of 50 to 200 years. Du-ration of time averaging is likely to be faciesdependent.

The results of time-averaging of the Ma-dang Lagoon fossil assemblages show bothsimilarities and differences with time-aver-aged molluscan assemblages in sediment (seeScoffin 1992; Kidwell and Flessa 1995; Kowa-lewski 1996; Best and Kidwell 2001). Time-av-eraging affects the diversity and species com-position of death (and fossil) molluscan as-semblages in four ways (Kidwell and Flessa1995): (1) Species richness and diversity arehigher in death or fossil assemblages than inthe life assemblages from which they are de-rived. (2) More rare taxa are recorded in deathor fossil assemblages than in life assemblages.(3) Variation in taxon abundances is lower in


TABLE 9. ANOSIM significant similarity table, at three levels of analysis (see Rahel 1990): G 5 growth form relativeabundance, P 5 taxon presence/absence, A 5 taxon relative abundance. Capital letters indicate that the fossil as-semblage and the given modern assemblage are significantly different in their composition at the specified level;lowercase letters indicate that the fossil assemblage and modern assemblage are not significantly different. Com-parisons for which all three levels of analysis agree are in italics. Fossil assemblages are interpreted as those forwhich ANOSIM yields a nonsignificant difference at all three levels of analysis. Therefore the fossil assemblage atJAR can be interpreted as either the combined deep life and deep death assemblages or as the pooled shallow lifeand deep life assemblages. The fossil assemblage at WIW can be interpreted only as the combined shallow life andshallow death assemblages.

Assemblage

JAR fossil vs.


WIW fossil vs.


Life APG

APg

apg

ApG

APG

apG

Death APG

APG

apG

aPG

APG

APG

Life 1 death ApG

apg

apG

apg

aPG

apG

death or fossil assemblages than in life assem-blages. (4) The rank-order of taxon abundanc-es is generally not changed between the deathor fossil assemblages and the life assemblages.Of these four patterns, (1) and (3) were re-corded in the Madang fossil reefs, relative totheir corresponding life assemblages, (2) wasnot observed, and (4) was weakly observed.

1. The fossil reefs showed greater speciesrichness and Shannon-Wiener diversity thandid their live and dead counterparts, as ob-served for time-averaging of molluscan as-semblages in sediment. Species richness nor-malized to transect length (S9) and Shannon-Wiener diversity (H9) were both greater in thefossil assemblage than in the life or death as-semblage.

2. In contrast to the patterns typically ob-served in sediment, rare coral taxa were un-derrepresented in the fossil assemblage, rela-tive to the life assemblage. There may be tworeasons for this contrast with the pattern ob-served in time-averaged molluscan assem-blages. First, rare species in the Madang fossilreefs may be diluted by coral rubble producedby the common species (Scoffin 1992). Second,some coral taxa recorded in the life assem-blages are very difficult to distinguish to thespecies level in fossil material, such as the rareAcropora species.

3. As observed in sediment molluscan as-semblages, the fossil reef assemblages appar-

ently integrated spatial and temporal fluctu-ations in the abundance of different corals,particularly thicket-forming branching corals.This integration yielded an overall fauna con-taining all the major species and with less var-iation in composition between samples thanobserved in the life assemblages (insignificanttime-averaging; see Kowalewski 1996; Ol-szewski and West 1997). Because the fossil as-semblages had less variability in species rel-ative abundance than did the life assemblages,the fossil assemblages were able to distin-guish kilometer-scale variations in taxonomiccomposition that were not significant in thelife assemblages (see Kidwell 1998).

4. In contrast to the pattern observed formollusks in sediment (Kidwell 2001), therank-order of taxon abundances for the tenmost abundant taxa in the life assemblageswas not reflected in the fossil assemblagesconsidered separately, although this correla-tion was higher for the entire fossil assem-blage pooled. The rank-order of taxa in thefossil assemblages differed from that in thelife assemblages because the relatively fewtaxa of rapidly growing branching corals weremore abundant in the fossil assemblages thanin the life assemblages.

The similarities and discrepancies betweenthe results of time-averaging observed amongmollusks in sediment and the patterns ob-served among corals in the Madang fossil


FIGURE 9. Time-averaging in fossil reefs. The representation of a given coral taxon in a fossil reef reflects both itsabundance and its growth rate. Branching corals in the life assemblage grow, reproduce, and become rubble severaltimes faster than massive corals. Dashed lines within massive corals in the fossil assemblage represent the incre-mental growth of the massive corals during the same time that the branching corals have grown, reproduced, andformed rubble. A snapshot fossil assemblage includes the life assemblage at time of fossilization plus accumulateddeath assemblages in the sediment, which are dominated by branching coral rubble. The fossil assemblage is there-fore composed of multiple cohorts of branching corals and a single cohort of massive corals. In lagoonal facies fromwhich sediment export is unlikely, branching corals dominate sediment production and are overrepresented in thefossil assemblage, relative to their abundance in the life assemblage. This pattern is less likely to hold in higherenergy facies from which most branching coral rubble is exported.

reefs can be explained by examining the rel-ative rates of accumulation of most coral taxaversus the rate of accumulation of the encas-ing sediment. In the fossil reefs, most coraltaxa were diluted within the coral rubble thataccumulated from the rapidly growingbranching corals (Acropora spp., Montipora dig-itifera, Seriatopora spp., Stylophora sp., Pocillo-pora sp., etc.) that produced most of the sedi-ment in which the reef was preserved (Scoffin1992).

Time-averaging between the life and deathassemblages appears to be an inherent featureof fossil reefs. Organism life spans on reefs arehighly variable and include both short- and

long-lived corals. Because coral sediment pro-duction rates are higher than non-bioclasticsedimentation rates, reefs preserve a series ofoverlapping cohorts of corals and other reef-building organisms (see Kowalewski 1996).The composition of the fossil assemblage part-ly reflects the different rates of sediment pro-duction of branching corals versus massivecorals, for example (Fig. 9).

The degree and nature of time-averaging inreefs also depends on exposure. In lagoonalfacies, from which little rubble-sized sedimentis exported, those corals that produce largeamounts of bioclastic sediment are overrep-resented in the fossil assemblage, relative to


the life assemblage (Fig. 9). High energy reeffacies may underrepresent fragile taxa that ac-cumulate as rubble and are mostly exportedduring storms (e.g., Scoffin 1981, 1992; Hub-bard et al. 1990; Hubbard 1997).

Fidelity Indices

The fossil reefs accurately preserved the di-versity and community composition of depth-specific portions of the modern assemblages,yet the fidelity indices suggest that the fossilassemblages were relatively poor recorders ofthe modern life assemblages. The discrepancybetween the accurate preservation of diversityand composition patterns and the low fidelityindices may result from lower sampling inten-sity in the fossil assemblage than in the life ordeath assemblages, reduced ability to distin-guish species in fossil material, time-averag-ing, or a combination of these factors.

Lower sampling intensity in the fossil reefsrelative to the modern reef may be partly re-sponsible for the scarcity of rare coral taxa inthe fossil reefs. Many of the rare taxa recordedin the life assemblages might have been foundby exhaustive collection of the fossil assem-blage. Others might have occurred in the fossilassemblages but were not recognized, andwere recorded at the genus level only (e.g., Ac-ropora spp.).

In the time-averaged fossil reefs, the encas-ing sediment holding the fossil corals is com-posed of numerically dominant corals thatproduce vast quantities of coral rubble. There-fore, rather than the fossils being concentratedin the sediments by accumulating faster thanthe sediments, most coral taxa are dilutedwithin the fossil reef by the rapid sedimentproduction by some corals, particularlybranching Acropora and other rapidly growingcorals (Scoffin 1992). This process reduces thelikelihood of encountering rare coral taxa inthe fossil assemblage, compared with the lifeassemblage. Dilution of some coral taxa bycoralline sediment is probably most intense inreef lagoons and would have less effect inhigher-energy reef facies such as reef crestsand fore-reefs (Scoffin 1992).

Finally, fidelity indices developed for life as-semblage/death assemblage comparisons ofmollusks in sediment may not be appropriate

for reefs. Life assemblage/death assemblagecomparisons of mollusks on Red Sea reefs andother hard substrata yielded much different fi-delity indices than did such comparisons formollusks in sediment, either carbonate or sil-iciclastic (Zuchsin et al. 2000). Hard substratemollusks had a lower percentage of live taxafound dead, and a higher percentage of deadtaxa found live, than did mollusks collectedfrom sediment. Furthermore, the relativeabundances of mollusk species in the life anddeath assemblages were radically differentand largely reflected differences in life-historycharacteristics and the likelihood of being ex-ported from the hard-substrate environmentupon death.

Taphonomic Bias by Growth Forms

The taphonomic processes affecting pres-ervation of the fossil reef acted both on coralgrowth forms and at the level of taxonomiccomposition. That different coral growthforms should be affected differentially bytaphonomic processes is hardly surprising(Pandolfi and Greenstein 1997a). The growth-form bias toward branching corals (primarilyas rubble) in the death and fossil assemblagesof the Madang Lagoon is expected, given thatbranching corals are frequently the dominantsediment producers in lagoonal facies of coralreefs (Stearn et al. 1977; James and Bourque1992).

The death assemblage bias toward branch-ing corals and against massive corals (Pan-dolfi and Minchin 1995), was also observed invarious studies from the Caribbean (Green-stein and Pandolfi 1997; Pandolfi and Green-stein 1997b; Greenstein et al. 1998b). The ab-solute abundance of massive corals in the Ma-dang Holocene reef was similar to that in theMadang Lagoon life assemblage. Combiningthe life and death assemblages to produce thefossil assemblage increased the representationof branching corals in the fossil assemblagesand shifted the relative abundance toward thebranching corals (Fig. 9). Different rates ofsediment production by branching and mas-sive corals, integrated over several genera-tions and accumulated in the sediment, shift-ed the taxonomic composition of the fossilreefs away from the life assemblages and to-


ward the death assemblages (Figs. 6, 9). Oursampling methods also favored branchingcorals in the death assemblage, becausebranching coral fragments are the most likelytypes of coral debris to be counted in an ex-cavated rubble sample (Pandolfi and Green-stein 1997b).

Applications to Marine Conservation

Fossil reefs accurately portray reef com-munity structure integrated over ecologicaltime, providing more information than biolog-ical snapshots measured at yearly or decadaltimescales (see Peterson 1976; Kidwell 1998).Fossil assemblages may be better able to dis-tinguish subtle differences in communitycomposition than life assemblages becausefossil assemblages have higher signal-to-noiseratios in taxon relative abundance variationsthan life assemblages (insignificant time av-eraging, Kowalewski 1996).

Raised Holocene and Pleistocene fossil reefsmay provide an excellent long-term record ofcoral reef taxonomic composition and com-munity structure prior to human interference.The taxonomic composition of these fossilreefs can be compared with degraded near-shore reefs where no historical records exist ofreef condition prior to human interference(Greenstein et al. 1998c; Edinger et al. 2000).Although Pleistocene and Holocene reefsgrew under different sea levels and sea-sur-face temperatures without major changes intheir taxonomic composition (Pandolfi 1996),many modern reefs in nearshore environ-ments are now undergoing dramatic changedue to human activities ranging from destruc-tive fishing practices to construction anddumping of raw sewage (see Birkeland 1997for review). Similarly, comparisons of modernand Pleistocene molluscan reef faunas haveshown the general stability of reef faunasthrough sea-level fluctuations, although softsubstrate and lagoonal facies lost more diver-sity during the last glaciation than did moreexposed reef facies (Taylor 1978; Crame 1996;Paulay 1996). As with the modern, human-in-duced reductions in diversity, these Pleisto-cene mollusk local extinctions were largely re-lated to habitat reduction or elimination.

The Holocene reef at Madang preserved a

time-averaged, low-variance record of near-shore reef taxonomic composition and com-munity structure prior to extensive human in-terference. Our data suggest that raised Ho-locene and Pleistocene reefs can be used as en-vironmental benchmarks for comparinglong-term change in community compositionof modern reefs (Greenstein et al. 1998b; Pan-dolfi and Jackson 2001). Comparison of fossilreefs with degraded modern reefs may recordgross changes in the relative abundance ofcommon taxa (e.g., Aronson and Precht 1997;Greenstein et al. 1998b) but is less likely to re-cord disappearances of rare taxa. The mostappropriate use of fossil reefs as long-term en-vironmental benchmarks may be to examinechanges in community structure, reduced spe-cies diversity, loss of keystone species (Pan-dolfi 1999), and transitions from coral-domi-nated to algal-dominated reefs (Greenstein etal. 1998a).

Conclusions

1. Fossil lagoonal reefs from Madang, Pap-ua New Guinea, preserved depth-specificcombinations of the life and death assemblag-es of the modern reefs from which they werederived. Depth zonation was preserved on thefossil reefs. The Madang fossil reefs are time-averaged composites of the life and death as-semblages.

2. Fossil lagoonal assemblages had higherspecies richness per unit area sampled, andhigher Shannon-Wiener diversity than life ordeath assemblages, but recorded fewer totaltaxa than did the life assemblages. Fossil la-goonal reefs had lower variation in taxon rel-ative abundance than did adjacent Recent lifeand death assemblages.

3. Fossil reefs are time-averaged. In the Ma-dang Lagoon, fast-growing branching coralswere the dominant sediment producers andwere overrepresented in the fossil assemblagerelative to the life assemblages. Rare and slow-growing coral taxa were underrepresented inthe fossil reefs because their abundances werediluted by the large volume of branching coralrubble inherited from the death assemblage.

4. Fidelity indices based on percentage oflive taxa recorded in the fossil assemblage,and vice-versa, are low for lagoonal facies of


fossil reefs because the encasing sediment ac-cumulates faster than many of the coral taxapreserved in the reef. The low fidelity indiceson reefs are a consequence of the nature oftime-averaging on reefs.

5. Taphonomic bias on reefs affects coralgrowth forms differentially. Lagoonal deathassemblages are overwhelmingly dominatedby branching coral rubble. When life anddeath assemblages are combined to producethe fossil assemblage, branching corals remainoverrepresented. Branching corals are lesslikely to be overrepresented in high energyreef facies from which branching coral rubbleis frequently removed.

6. Fossil reefs provide an accurate long-termrecord of reef diversity and community struc-ture. They can be used as a benchmark forcomparison with modern degraded nearshorereefs where no pre-disturbance measure-ments are available.

Acknowledgments

R. Aiello, M. Best, and C. Lovelock providedinvaluable assistance in the field. We thank M.Jebb and the staff of the Christensen ResearchInstitute, Madang, Papua New Guinea, for lo-gistical support. M. Best and J. Wingerath as-sisted with data tabulation and analysis. P.Copper, J. O. Ebbestad, C. Lovelock, J. Rendell,M. Zuchsin, and an anonymous reviewermade helpful comments on the manuscript. E.E. thanks the Geology Department, Universityof California, Davis, for a student travel grant(1993), and thanks the Geology Department,St. Francis Xavier University, for computingand library support. Fieldwork in Papua NewGuinea was supported by the Australian In-stitute of Marine Science, by the ChristensenResearch Institute, and by grants from theAustralian Research Council to J. M. P. Dedi-cated to the memory of Amor Halperin.

Literature Cited

Aronson, R. B., and W. F. Precht. 1997. Stasis, biological distur-bance, and community structure of a Holocene reef. Paleobi-ology 23:326–346.

Behrensmeyer, A. K., S. M. Kidwell, and R. A. Gastaldo. 2001.Taphonomy and paleobiology. In D. H. Erwin and S. L. Wing,eds. Deep time: Paleobiology’s perspective. Paleobiology26(Suppl. to No. 4):103–147.

Best, M. M. R., and S. M. Kidwell. 2000. Bivalve taphonomy in

tropical mixed siliciclastic-carbonate settings. I: Environmen-tal variation in shell condition. Paleobiology 26:80–102.

Birkeland, C. E. 1997. Life and death of coral reefs. Chapmanand Hall, London.

Boucot, A. J. 1953. Life and death assemblages among fossils.American Journal of Science 251:25–40.

Bray, J. R., and J. T. Curtis. 1957. An ordination of the uplandforest communities of southern Wisconsin. Ecological Mono-graphs 27:325–349.

Buddemeier, R. W., and R. A. Kinzie. 1976. Coral growth. Ocean-ography and Marine Biology Annual Review 14:183–225.

Chappell, J. 1974. Geology of coral terraces, Huon Peninsula,New Guinea: a study of Quaternary tectonic movements andsea-level changes. Geological Society of America Bulletin 85:533–570.

Clarke, K. R. 1993. Non-parametric multivariate analyses ofchanges in community structure. Australian Journal of Ecol-ogy 18:117–143.

Copper, P. 1988. Ecological succession in Phanerozoic reef eco-systems: is it real? Palaios 3:136–152.

———. 1997. Articulate brachiopod shellbeds: Silurian exam-ples from Anticosti, Eastern Canada. Geobios 20:133–148.

Crame, J. A. 1996. Late Pleistocene molluscan assemblages fromthe coral reefs of the Kenya coast. Coral Reef 4:183–196.

Edinger, E. N., and M. J. Risk. 1999. Reef classification by coralmorphology predicts coral reef conservation value. BiologicalConservation 92:1–13.

Edinger, E. N., J. Lundberg, and M. J. Risk. 2000. Mid-Holocenefossil reef at Jepara, Central Java, Indonesia: a benchmark ofnearshore reef diversity and composition before human dis-turbance? Ninth International Coral Reef Symposium, Ab-stracts, p. 49.

Fagerstrom, J. A. 1987. The evolution of reef communities. Wi-ley, Toronto.

Fursich, F. T. 1978. The influence of faunal condensation andmixing on the preservation of fossil benthic communities. Le-thaia 11:243–250.

Greenstein, B. J., and H. A. Moffat. 1996. Comparative taphon-omy of Holocene and Pleistocene corals, San Salvador, Ba-hamas. Palaios 11:57–63.

Greenstein, B. J., and J. M. Pandolfi. 1997. Preservation of com-munity structure in modern reef coral life and death assem-blages of the Florida Keys: implications for the Quaternary re-cord of coral reefs. Bulletin of Marine Science 19:39–59.

Greenstein, B. J., H. A. Curran, and J. M. Pandolfi. 1998a. Shift-ing ecological baselines and the demise of Acropora cervicornisin the western North Atlantic and Caribbean province: aPleistocene perspective. Coral Reefs 17:249–261.

Greenstein, B. J., J. M. Pandolfi, and H. A. Curran. 1998b. Thecompleteness of the Pleistocene fossil record: implications forstratigraphic adequacy. Pp. 75–109 in S. K. Donovan, ed., Theadequacy of the fossil record. Wiley, London.

Greenstein, B. J., A. White, and H. A. Curran. 1998c. Compari-son of recent coral life and death assemblages to Pleistocenereef communities: implications for rapid faunal replacementon recent reefs. Carbonates and Evaporites 13:23–31.

Hubbard, D. K. 1997. Reefs as dynamic systems. Pp. 43–67 in C.Birkeland, ed. Life and death of coral reefs. Chapman andHall, New York.

Hubbard, D. K., A. I. Miller, and D. Scaturo. 1990. Productionand cycling of calcium carbonate in a shelf-edge reef system(St. Croix, US Virgin Islands): applications to the nature ofreef systems in the fossil record. Journal of Sedimentary Pe-trology 60:335–360.

James, N. F., and P. A. Bourque. 1992. Reefs and mounds. Pp.323–347 in R. G. Walker and N. P. James, eds. Facies models:response to sea level change. Geological Association of Can-ada, St. John’s, Nfld.


Johnson, R. G. 1965. Pelecypod death assemblages in TomalesBay, California. Journal of Paleontology 39:80–85.

Kidwell, S. M. 1998. Time-averaging in the marine fossil record:overview of strategies and uncertainties. Geobios 30:977–995.

———. 2001. Ecological fidelity of molluscan death assemblag-es, in J. Y. Aller, S. A. Woodin, and R. C. Aller, eds. Organism–sediment interactions. Belle W. Baruch Library in Marine Sci-ence, No. 21. University of South Carolina Press, Columbia (inpress).

Kidwell, S. M., and D. W. J. Bosence. 1991. Taphonomy and time-averaging of marine shelly faunas. Pp. 115–209 in P. Allisonand D. E. G. Briggs, eds. Taphonomy: releasing the datalocked in the fossil record. Plenum, New York.

Kidwell, S. M., and K. Flessa. 1995. The quality of the fossil re-cord. Annual Review of Ecology and Systematics 26:269–299.

Kowalewski, M. 1996. Time-averaging, overcompleteness, andthe quality of the fossil record. Journal of Geology 104:317–326.

Kowalewski, M., G. A. Goodfriend, and K. W. Flessa. 1998.High-resolution estimates of temporal mixing within shellbeds: the evils and virtues of time-averaging. Paleobiology 24:287–304.

Loya, Y. 1978. Plotless and transect methods. Pp. 197–217 in D.R. Stoddart and R. E. Johannes, eds. Coral reefs: researchmethods. UNESCO, Paris.

Meldahl, K. H., K. W. Flessa, and A. H. Cutler. 1997. Time-av-eraging and postmortem skeletal survival in benthic fossil as-semblages: quantitative comparisons among Holocene envi-ronments. Paleobiology 23:207–229.

Mesolella, K. J. 1967. Zonation of uplifted Pleistocene coral reefson Barbados, West Indies. Science 156:638–640.

Olszewski, T. 1999. Taking advantage of time-averaging. Paleo-biology 25:226–238.

Olszewski, T., and R. West. 1997. Influence of transportation andtime-averaging in fossil assemblages from the Pennsylvanianof Oklahoma. Lethaia 30:315–330.

Pandolfi, J. M. 1996. Limited membership in Pleistocene reefcoral assemblages from the Huon Peninsula, Papua NewGuinea: constancy during global change. Paleobiology 22:152–176.

———. 1999. Response of Pleistocene coral reefs to environ-mental change over long temporal scales. American Zoologist39:113–130.

Pandolfi, J. M., and B. J. Greenstein. 1997a. Taphonomic alter-ation of reef corals: effects of reef environment and coralgrowth form. I: The Great Barrier Reef. Palaios 12:27–42.

———. 1997b. Preservation of community structure in death as-semblages of deep-water Caribbean reef corals. Limnologyand Oceanography 42:1505–1516.

Pandolfi, J. M., and J. B. C. Jackson. 2001. Community structureof Pleistocene coral reefs of Curacao, Netherlands Antilles.Ecological Monographs 71:49–67.

Pandolfi, J. M., and P. R. Minchin. 1995. A comparison of taxo-nomic composition and diversity between reef coral life anddeath assemblages in Madang Lagoon, Papua New Guinea.Palaeogeography, Palaeoclimatology, Palaeoecology 119:321–341.

Paulay, G. 1996. Dynamic clams: changes in the bivalve fauna ofPacific islands as a result of sea-level fluctuations. AmericanMalacological Bulletin 12:45–57.

Perrin, C., D. W. J. Bosence, and B. R. Rosen. 1995. Quantitativeapproaches to palaeozonation and palaeobathymetry of cor-als and coralline algae in Cenozoic reefs. Pp. 181–229 in D. W.J. Bosence and P. Allison, eds. Marine palaeoenvironmental

analysis from fossils. Geological Society of London SpecialPublication 83.

Peterson, C. H. 1976. Relative abundances of living and deadmolluscs in two California lagoons. Lethaia 9:958–965.

Rahel, F. J. 1990. The hierarchical nature of community persis-tence: a problem of scale. American Naturalist 136:328–344.

Scoffin, T. P. 1981. Aspects of the preservation of deep and shal-low water reefs. Proceedings of the Fourth International CoralReef Symposium 1:499–501.

———. 1992. Taphonomy of coral reefs: a review. Coral Reefs11:57–77.

Scoffin, T. P., A. W. Tudhope, B. E. Brown, H. Changsang, andR. F. Cheeney. 1992. Patterns and possible environmental con-trols of skeletogenesis of Porites lutea, South Thailand. CoralReefs 11:1–11.

Stearn, C. W., T. P. Scoffin, and W. Martindale. 1977. Calciumcarbonate budget of a fringing reef on the west coast of Bar-bados, I. Zonation and productivity. Bulletin of Marine Sci-ence 27:479–510.

Taylor, J. D. 1978. Faunal response to the instability of reef hab-itats: Pleistocene molluscan assemblages of Aldabra atoll. Pa-laeontology 21:1–30.

Tudhope, R. W., D. W. Lea, G. B. Shimmield, C. P. Chilcott, T. P.Scoffin, A. E. Fallick, and M. Jebb. 1997. Climatic records frommassive Porites corals in Papua New Guinea: a comparison ofskeletal Ba/Ca, Skeletal d18), and coastal rainfall. Proceed-ings of the Eighth International Coral Reef Symposium 2:1719–1724.

Tudhope, A. W., R. W. Buddemeier, C. P. Chilcott, K. R. Berry-man, D. G. Fautin, M. Jebb, J. H. Lipps, R. G. Pearce, T. P. Scof-fin, and G. B. Shimmield. 2000. Alternating seismic uplift andsubsidence in the late Holocene at Madang, Papua New Guin-ea: evidence from raised reefs. Journal of Geophysical Re-search (Solid Earth) 105(B6):13797–13807.

Veron, J. E. N. 1993. A biogeographic database of hermatypiccorals. Species of the central Indo-Pacific and genera of theworld. Australian Institute of Marine Science Monograph 10.Townsville, Australia.

Veron, J. E. N., and M. Pichon. 1976. Scleractinia of eastern Aus-tralia, Vol. I. Families Thamnasteriidae, Astrocoeniidae, Po-cilloporidae. Australian Institute of Marine Science Mono-graph 1. Townsville, Australia.

———. 1980. Scleractinia of eastern Australia, Vol. III. FamiliesAgariciidae, Siderastreidae, Fungiidae, Oculinidae, Merulini-dae, Mussidae, Pectinidae, Caryophylliidae, Dendrophylli-idae. Australian Institute of Marine Science Monograph 4.Townsville, Australia.

———. 1982. Scleractinia of eastern Australia, Vol. IV. FamilyPoritidae. Australian Institute of Marine Science Monograph5. Townsville, Australia.

Veron, J. E. N., and C. C. Wallace. 1984. Scleractinia of easternAustralia, Vol. V. Family Acroporidae. Australian Institute ofMarine Science Monograph 6. Townsville, Australia.

Veron, J. E. N., M. Pichon, and M. Wijsman-Best. 1977. Sclerac-tinia of eastern Australia, Vol. II. Families Faviidae, Trachy-phylliidae. Australian Institute of Marine Science Monograph3. Townsville, Australia.

Walker, K. R., and L. P. Alberstadt. 1975. Ecological successionas an aspect of structure in fossil communities. Paleobiology1:238–257.

Zuchsin, M., J. Hohenegger, and F. F. Steininger. 2000. A com-parison of living and dead molluscs on coral reef associatedhard substrata in the northern Red Sea—implications for thefossil record. Palaeogeography, Palaeoclimatology, Palaeoe-cology 159:167–190.


Appendix 1Taxa most responsible for dissimilarity between fossil and Recent assemblages. Discriminating taxa are those for

which the ratio of average abundance to standard deviation of abundance is .1. Abundances are double square-root transformed counts.

A. Taxa most responsible for dissimilarity between fossil vs. shallow life assemblage, taxon relative abundances,JAR. ANOSIM p 5 0.008.

TaxonGrowth

formLive

abundanceFossil

abundance%

DissimilarityCumulative

% dissimilarity

Acropora spp.Montipora digitataMillepora sp.Porites sp.Faviidae

BBBMM

2.27.206.04.0

16.44.64.88.60.6

11.088.057.957.055.46

11.0819.1327.0834.1339.59

Heliopora sp.Seriatopora sp.Montipora spp.Acropora paliferaGoniastrea retiformisFungiidae

COLBFRCOLMFL

01.81.60.41.80

1.62.20.22.42.01.0

5.064.873.653.413.311.81

44.6549.5253.1756.5859.8961.70

B. Taxa most responsible for dissimilarity between fossil vs. shallow death assemblage, JAR, taxon relative abun-dances. ANOSIM p 5 0.016.

TaxonGrowth

form

Deathassemblageabundance

Fossilabundance

%Dissimilarity

Comulative% dissimilarity

Montipora digitataAcropora spp.Porites sp.Millepora sp.Seriatopora sp.Heliopora sp.Fungiidae

BBMBBCOLFL

105.427.22.80.4

17.400.6

4.616.4

8.64.82.21.61.0

15.8313.15

9.297.486.535.342.55

15.8328.9838.2745.7552.2857.6260.17

C. Taxa most responsible for dissimilarity between fossil vs. deep life assemblage, JAR, taxon relative abundances.ANOSIM p 5 0.008.

TaxonGrowth

formLive

abundanceFossil

abundance%


% dissimilarity

Acropora spp.FungiidaePorites sp.Montipora digitataMillepora sp.Heliopora sp.

BFLMBBCOL

2.23.62.80.22.800

16.41.08.64.64.81.6

11.085.775.755.494.424.14

11.0816.6522.4027.8932.3136.45

Seriatopora sp.Acropora paliferaGalaxea sp.Porites cylindricaGoniastrea retiformisPocillopora damicornisFaviidae

BCOLMBMBM

5.01.80.63.80.20.60.8

2.22.41.602.01.00.6

3.893.743.703.662.572.392.29

40.3444.0847.7851.4454.0156.4058.69


Appendix 1Continued.

D. Taxa most responsible for dissimilarity between fossil vs. deep death assemblage, JAR, taxon relative abundances.ANOSIM p 5 0.008.

TaxonGrowth

form

Deathassemblageabundance

Fossilabundance

%Dissimilarity

Cumulative% dissimilarity

Seriatopora sp.Acropora spp.Porites sp.Millepora sp.FungiidaeHeliopora sp.Acropora palifera

BBMBFLCOLCOL

111.644.41.8

60.24.800.4

2.216.4

8.64.81.01.62.4

10.828.077.047.004.553.932.83

10.8218.8925.9332.9337.4841.4144.24

Goniastrea retiformisPocillopora damicornis

MB

01.2

2.01.0

2.612.59

46.8549.44

E. Taxa most responsible for dissimilarity between fossil vs. shallow life assemblage, WIW, taxon relative abun-dances. ANOSIM p 5 0.024.

TaxonGrowth

formLive

abundanceFossil

abundance%


% dissimilarity

Acropora spp.Stylophora sp.Porites sp.Montipora sp.Pocillopora sp.Millepora sp.FaviidaeGalaxea sp.

BBMFRBBMM

5.40.66.81.400.41.00.2

18.753.254.50.51.00.7500.5

10.867.626.756.493.773.492.892.18

10.8618.4825.2331.7235.4938.9841.8744.05

F. Taxa most responsible for dissimilarity between fossil vs. deep life assemblage, WIW, taxon relative abundances.ANOSIM p 5 0.024.

TaxonGrowth

formLive

abundanceFossil

abundance%


% dissimilarity

Acropora spp.Stylophora sp.Porites cylindricaPorites sp.Pocillopora sp.Millepora sp.Montipora sp.Pocillopora damicornisPorites lichen

BBSBMBBFRBEN

2.203.82.801.00.80.81.2

16.43.2504.51.00.750.51.00

15.577.896.776.353.393.242.532.402.18

15.5723.4630.2336.5839.9743.2145.7448.1450.32

G. Taxa most responsible for dissimilarity between fossil vs. death assemblage, depths pooled, WIW, taxon relativeabundances. ANOSIM p 5 0.024. Depths pooled because ANOSIMs for fossil assemblage vs. death assemblage withdepths separated were not significant.

TaxonGrowth

formLive

abundanceFossil

abundance%


% dissimilarity

Seriatopora sp.Stylophora sp.Millepora sp.Porites sp.Pocillopora sp.Montipora spp.Galaxea sp.

BBBMBFRM

34.750.446.22400.560.22

03.250.754.51.00.50.5

18.426.976.956.163.282.362.14

18.4225.3932.3435.5038.7841.1443.28


Appendix 2Rare taxa. Rare taxa are those that contributed less than 1% of the total coral abundance in the life assemblages.

Number of transects on which a given species occurred and its abundance in each assemblage are shown for JARand WIW combined.

TaxonNo. of live

occurrences% Live

abundanceNo. of deadoccurrences

% Deadabundance

No. of fossiloccurrences

% Fossilabundance

Symphyllia sp.Acropora echinataFavites abditaStylophora sp.Porites lichen

43143

0.960.960.140.830.83

21021

0.120.0200.100.02

10141

0.2400.943.070.24

Porites rusGalaxea sp.Hydnophora pilosaTurbinaria sp.Goniastrea sp.

44110

0.830.690.280.550

13000

0.020.10000

04011

02.3600.240.47

Pachyseris sp.Favia sp.Acropora validaAcropora cerealisAcropora tenuis

24212

0.550.550.280.280.69

20000

0.220000

00000

00000

Acropora elseyiAcropora cf. solitaryensisAcropora pulchraAcropora valencienesiAcropora puertogalerae

21111

0.280.140.280.550.14

00010

0000.170

00000

00000

Goniastrea asperaMontastraea sp.Cyphastrea sp.Hydnophora sp.Mussismilia sp.

11221

0.280.140.410.280.14

10001

0.020000.02

20201

0.4700.7100.24

Pectinia paeoniaPectinia lactucaPavona sp.Anacropora sp.Culicia sp.

11100

0.410.140.1400

02231

00.070.05

11.770.02

00000

00000

Oxypora sp.Acropora hyacinthusPocillopora sp.Oulophyllia sp.Coeloseris mayeri

00000

00000

10200

0.0700.0500

02432

00.941.421.420.71

Platygyra sinensisFavites complanataSymphyllia radiansAstreopora sp.Lobophyllia sp.

00000

00000

00000

00000

21111

0.470.242.120.240.24

Documents

Community structure of Quaternary coral reefs compared ...marinepalaeoecology.org/.../2011/...Paleobiology.pdf · Paleobiology,27(4), 2001, pp. 669–694 Community structure of Quaternary