46. CLIMATICALLY INDUCED CHANGES IN VERTICAL WATER MASS STRUCTURE OF THEVEMA CHANNEL DURING THE PLIOCENE: EVIDENCE FROM DEEP SEA

DRILLING PROJECT HOLES 516A, 517, AND 5181

David A. Hodell and James P. Kennett, Graduate School of Oceanography, University of Rhode Island,Narragansett, Rhode Island

andKathleen A. Leonard,2 Department of Geology, University of South Carolina, Columbia, South Carolina

ABSTRACT

Pliocene changes in the vertical water mass structure of the western South Atlantic are inferred from changes in ben-thic foraminiferal assemblages and stable isotopes from DSDP Holes 516A, 517, and 518. Factor analysis of 34 samplesfrom Site 518 reveals three distinct benthic foraminiferal assemblages that have been associated with specific subsurfacewater masses in the modern ocean. These include a Nuttalides umbonifera assemblage (Factor 1) associated with Ant-arctic Bottom Water (AABW), a Globocassidulina subglobosa-Uvigerina peregrina assemblage (Factor 2) associatedwith Circumpolar Deep Water (CPDW), and an Oridorsalis umbonatus-Epistominella exigua assemblage associatedwith North Atlantic Deep Water (NADW). Bathymetric gradients in Δ13C between Holes 516A (1313 m), 517 (2963 m),and 518 (3944 m) are calculated whenever possible to monitor the degree of similarity and/or difference in the apparentoxygen utilization (AOU) of water masses located at these depths during the Pliocene. Changes in bathymetric Δ13Cgradients coupled with benthic foraminiferal assemblages record fundamental changes in the vertical water mass struc-ture of the Vema Channel during the Pliocene from 4.1 to 2.7 Ma.

At Site 518, the interval from 4.1 to 3.6 Ma is dominated by the N. umbonifera (Factor 1) and O. umbonatus-E. ex-igua (Factor 3) assemblages. The Δ13C gradient between Holes 518 (3944 m) and 516A (1313 m) undergoes rapid oscilla-tions during this interval though no permanent increase in the gradient is observed. However, δ13C values at Site 518 areclearly lighter during this interval. These conditions may be related to increased bottom water activity associated withthe re-establishment of the West Antarctic Ice Sheet in the late Gilbert Chron (-4.2 to 3.6 Ma) (Osborn et al., 1982).

The interval from 3.6 to 3.2 Ma is marked by a dominance of the G. subglobosa-U. peregrina (Factor 2) assemblageand lack of a strong Δ13C gradient between Holes 518 (3944 m) and 516A (1313 m). We suggest that shallow circumpo-lar waters expanded to depths of a least 3944 m (Site 518) during this time.

The most profound faunal and isotopic change occurs at 3.2 Ma, and is marked by dominance of the N. umbonifera(Factor 1) and O. umbonatus-E. exigua (Factor 3) assemblages, a l.l‰ enrichment in δ 1 8 θ, and a large negative in-crease in the Δ13C gradient between Holes 518 and 516A. These changes at Site 518 record the vertical displacement ofcircumpolar waters by AABW and NADW. This change in vertical water mass structure at 3.2 Ma was probably relatedto a global cooling event and/or final closure of the Central American seaway. A comparison of the present-day δ13Cstructure of the Vema Channel with a reconstruction between 3.2 and 2.7 Ma indicates that circulation patterns duringthis late Pliocene interval were similar to those of the modern western South Atlantic.

INTRODUCTION

Leg 72 of the Deep Sea Drilling Project (DSDP) hascored three sites along a depth transect down the RioGrande Rise (Fig. 1). Each site intersects one of the ma-jor subsurface water masses of the western South Atlan-tic (Table 1). Leonard and others (this volume) report onthe Pliocene paleoceanography of Hole 516A which re-cords the history of changes in the position of AntarcticIntermediate Water (AAIW) and upper CircumpolarDeep Water (CPDW). Our study focuses upon the deep-er water masses associated with Sites 517 and 518. Site517 (2963 m) is presently positioned in the core of NorthAtlantic Deep Water (NADW), and Site 518 (3944 m) issituated near large physical and chemical gradients sep-arating Antarctic Bottom Water (AABW), lower Cir-

Barker, P. F., Carlson, R. L., Johnson, D. A., et al., Init. Repts. DSDP, 72: Wash-ington (U.S. Govt. Printing Office).

2 Present address: Marathon Oil Company, P.O. Box 3128, Houston, TX 77001.

cumpolar Deep Water (CPDW), and North AtlanticDeep Water (NADW) (Fig. 2). We have combined fau-nal and isotopic data from Holes 516A, 517, and 518 topotentially monitor changes in the vertical water massstructure of the Vema Channel during the Pliocene from4.1 Ma to 2.7 Ma.

We are especially interested in determining the re-sponse of subsurface water masses to times of rapid cli-matic change. The climatic history of the Pliocene ismarked by a high degree of stability punctuated by rap-id climatic deterioration associated with the late Plio-cene paleoceanographic events. The late Pliocene, atabout 3.2 and 2.5 Ma, represents a time of conspicuousglobal cooling (Kennett et al., 1979; Keigwin and Thun-ell, 1979; Thunell, 1979), and the latter event (2.5 Ma)has been inferred to represent the beginning of substan-tial ice buildup on northern hemisphere continents(Shackleton and Opdyke, 1977). The purpose of thisstudy is to investigate relationships between vertical mi-grations of water masses in the Vema Channel and timesof climatic change during the Pliocene.

907

D. A. HODELL, J. P. KENNETT, K. A. LEONARD

35° S

34

DSDP sitesMagnetic anomaliesBottom currents

Argentine Basin

45° W 40° 35° 30° W

Figure 1. The location of DSDP Holes 516A, 517, and 518, which were hydraulically piston cored along adepth transect down the Rio Grande Rise during Leg 72 in the southwestern Atlantic.

Table 1. Association of Leg 72 sites with subsurface water masses ofthe western South Atlantic.

Hole Latitude

Waterdepth

Longitude (m) Water mass

515 26°14.3'S 36°30.2'W 4250 AABW518 29°58.4'S 38°O8.1'W 3944 AABW, NADW, CPDW517 30°56.8'S 38°02.5'W 2963 NADW516A 30°16.6'S 35°37.1'W 1313 CPDW, AAIW

Note: AABW = Antarctic Bottom Water; NADW = North AtlanticDeep Water; CPDW = Circumpolar Deep Water; AAIW = Ant-arctic Intermediate Water.

HYDROGRAPHY

The physical oceanography of the western South At-lantic has been studied by Reid and others (1977). Thewestern South Atlantic contains five distinct subsurfacewater masses that have originated from the cooling andsinking of surface waters at high latitudes (Fig. 2). Thedeepest water mass is Weddell Sea Deep Water (WSDW),which is formed in the Weddell Sea or farther east alongthe Antarctic continent. We have retained the more fa-miliar term "AABW" in this paper to refer to WSDW.AABW is cold (0< - 0.4°C), highly saline (~34.7‰),nutrient depleted, and oxygen enriched (> 228 µM/kg).

AABW is either overlain by a thin lens of lower Circum-polar Deep Water or directly overlain by North AtlanticDeep Water (NADW). Where NADW meets CPDW inthe western South Atlantic, NADW splits CPDW into alower and an upper branch; the narrower density rangeof NADW fits into the broader range of CPDW. CPDWis characterized by an oxygen minimum (as low as ~ 200µM/kg) and nutrient maximum. The low oxygen valuesof CPDW result from lateral exchange with deeper wa-ters of the Indian and Pacific oceans while the relativelyhigh salinity is derived from the North Atlantic (Reid etal., 1977). NADW originates in the Norwegian-Green-land Sea and is recognized by high salinities (~34.9%0),

908

PLIOCENE ABYSSAL CIRCULATION

02(µM/kg)

220 260 Sites

1 -

2 -

3 -

4 -

'Salinity

- \

-

-

-

1 i

i i

-

y*

\

02 (µM/kg)^

%0^.>-C

^ ^ ^

Potentialtemperature 1

/

r—-

"̂

°C)/

1\\

MM —

N I

/

f

)b

>

-

-

-

\ "

\ -

) "

-

-

AAIW

UCPW

NADW

LCPW

WSDW

516

517

518

515

2T(°C)

I

34.2 34.6S (%„)

35.0

Figure 2. Position of Leg 72 sites relative to hydrographic data obtainedfrom GEOSECS Station #59 located in the axis of the Vema Chan-nel. AAIW = Antarctic Intermediate Water; UCPW = UpperCircumpolar Water; NADW = North Atlantic Deep Water; LCPW= Lower Circumpolar Water; WSDW = Weddell Sea Deep Wateror Antarctic Bottom Water. (After Barker et al., 1981.)

an oxygen maximum (250 µM/kg), and a nutrient mini-mum. NADW is overlain by the upper branch of CPDW.Antarctic Intermediate Water (AAIW) is the uppermostsubsurface water mass and is formed by surface coolingat subantarctic latitudes.

METHODS

Quantitative counts of benthic foraminifers in the > 150 µm sizefraction were made on 34 samples from Site 518 (3944 m). Sampleswere oven dried at 50°C, weighed, disaggregated in hot Calgon solu-tion, washed over a 63 µm Tyler screen, dried, and reweighed. Thirty-five taxa were counted according to the taxonomy outlined by Corliss,1979a (Table 2). No microsplitting of samples was necessary. Thenumber of specimens counted per sample ranged from 107 to 1045with an average of 319 counts per sample. Species percentages werecalculated within the total population. Q-mode factor analysis wasperformed on the data set using the program of Klovan and Imbrie(1971).

Isotopic analyses were performed on Planulina wuellerstorfl and/or Cibicidoides kullenbergi from Sites 517 and 518 (Tables 3, 4).Whenever possible, monospecific samples of P. wuellerstorfl wereanalyzed from the > 150 µm size fraction, but the paucity of this spe-cies in some samples occasionally required using a mixture off. wuel-lerstorfl and C. kullenbergi; however, these two species have beenshown to be isotopically similar (Duplessy et al., 1980; Woodruff et

al., 1981; Keigwin, 1982; and Leonard et al., this volume). Planktonicisotopic analyses were performed on Orbulina universa from Site 518(Table 5) and on Globigerinoides quadrilobatus from Holes 516A and517 (Leonard et al., this volume). Specimens for isotopic analysis wereultrasonically cleaned in deionized water, loaded into stainless steelcarrying boats, and roasted in vacuo at 400°C for one hour to vapor-ize organic matter. The samples were then loaded into a reaction vesseland reacted to completion in 100% orthophosphoric acid at 50°C.The resulting gas was distilled four times to remove water, and ana-lyzed using an on-line VG Micromass 602D mass spectrometer. Allisotopic results are expressed as per mil difference from PDB. Basedupon replicate analyses of a powdered standard (B-l), analytic preci-sion over a ten-month period was ±0.10% for δ 1 8θ and ±0.09‰ forδ13C.

STRATIGRAPHY

Precise intersite correlation is essential in this studybecause benthic carbon isotopic records are directly com-pared between sites. The best stratigraphic control forLeg 72 is contained within the shallowest Hole 516A(1313 m), which has detailed magnetostratigraphy andbiostratigraphy. Drilling at Hole 516A recovered a com-plete Pliocene section containing planktonic foraminif-eral Zones PL la through PL6 (Berggren et al., this vol-ume; Pujol, this volume). The oldest sediment recoveredat Site 517 is late Pliocene, and only Zones PL3 throughPL6 are present (Berggren et al., this volume; Pujol,this volume). The Pliocene section of Site 518 extendsfrom Zone PLlc through PL6 (Berggren et al., this vol-ume; Pujol, this volume). A hiatus of approximately1.5 Ma duration (5.5 to 4.0 Ma) spans the Miocene/Pliocene boundary at Site 518, and Zones PL la andPLlb are absent.

The late Pliocene sections of Site 517 and Hole 516Ahave been correlated using biostratigraphic datums andplanktonic stable isotopic records (Leonard et al., thisvolume). The Pliocene of Site 518 is directly correlatedwith the chronology of Hole 516A in a similar manner(Fig. 3). The planktonic stable isotopic records of Site518 and Hole 516A are in good agreement. Differencesbetween the two records can be attributed to differingsedimentation rates. For example, shoaling of the lyso-cline is likely to decrease sedimentation rates at thedeeper site (518) while producing no change at the shal-lower site (516A). The correlation of the planktonic iso-topic records also gives biostratigraphic datums that aregenerally consistent between Site 518 and Hole 516A.The isotopic correlations are more precise than thosederived using the biostratigraphy because the samplinginterval for isotopic analyses (1 sample/50 cm) is denserthan the interval used to determine placement of bio-stratigraphic events (1 sample/150 cm).

BENTHIC FORAMINIFERS

One of the most recent applications of deep-sea ben-thic foraminiferal ecology has been to study the rela-tionships between benthic foraminiferal distributionsand hydrographic properties of subsurface water mass-es. Numerous quantitative core-top studies of benthicforaminiferal distributions have now been conducted inall oceans (Streeter, 1973; Schnitker, 1974; Culp, 1977;Gofas, 1978; Lohmann, 1978; Ingle et al., 1980; Resig,1981). These studies form a basis for the extrapolation

909

D. A. HODELL, J. P. KENNETT, K. A. LEONARD

Table 2. Raw census data of 35 species of benthic foraminifers in the > 150 µm fraction from DSDP Site 518.

Sub-bottomdepth(m)

28.028.529.029.530.030.531.031.531.932.432.933.433.934.436.336.837.337.838.240.741.241.742.242.743.243.744.045.345.849.550.050.551.053.9

t1j1

230104128184524351

67240

5269343070795538214580174951373625

1182457312127

92939

.a

1ga.a

a951049419675384729

51413

141010008

150200

112111011

o

1aç"5

13827221918211621231687

1079

11106656

131012

125121614106

15104

O,

O

!

27172619102619141915284

1911

1845644632846214363

3

§|

1|cO138316052

1601293237543525222525251318128

432965341820

84540182519142219

111

I02005900

13000061213010030001021020

a,BOc

§I1

1000000000017

392914106700000001

312333160

2517

obos

a

|

|

oO

3620275010

985

5166

19264226272335342521592722

797

54485228212322

|"a•S

1|

α

0254611131319

1499759463655525

1817

5141410

dαaç

1s13

377350203112421142211132030010100

a

a

fs652000000100401

12300100103001010000

D.O.

a.β

i*l

7244

1411

311

315512

118037657

111093328436472

11

12012011011000000010210000001001520

1I||I0000000000000000000000000001000000

1δ

1•2

1100

244424905829371652102324273421

79

121013108

13568

1289

10121511

11&.c3&a

I1271410

82132

6161414

60

12161816

7111248

13654

1075785368

à

§Sj>

M5

102100000000010201004400622174

135424

1o

1δc

I16423504386

1267

1262249

108

1054367

10133

13157

•gα

1

!ε

0002140022812

119732005323200963278

10

S

s>-Si

a

110000000000000000200000000000100000

p.

o

1

222

153

1426

133810344403440101751

11552230

1.a.a

10000000009011323000000000003543000

1δ

j53

1171010

52723

65

127

104

104304324

1625

513

1336052354

1,S'a

§

11&17

016133116131611045045000102

13080090001004

a

1§

10000000000010

1283212000000003000012

lü

0010000020000110011000101001130002

ag

a.c5|

0000100000010030011101001002102101

jδ•5"a|

0100001000011031013000000001002100

rgo

!

i0000000000000000000000000000200001

D.

o.a

1824535234622174473907

11331120151010

i.a1

0011011000102000200100100100000000

oKa'S11ö

0013134152121340112263343204843517

ia0

1175146

10411029584316242332531344034

sKg

I11100002100207

154308033200011233112

O.

a

114121210012001020000023201000000200

Table 3. Oxygen and carbon isotopic results (in %0

PDB) of Planulina wuellerstorfi and/or Cibiá-doides kullenbergi from DSDP Site 517.

Core-section(interval in cm)

9-3, 20-2210-2, 20-2210-2, 120-12210-3, 70-7211-2, 20-2211-2, 70-7211-2, 120-12211-3,20-2211-3, 70-7212-1, 120-12212-2, 20-2212-2, 70-7212-2, 120-12212-3, 20-2212-3, 120-122

Depth(m)

36.537.938.941.443.844.344.845.345.847.748.248.749.249.750.7

Agea

(Ma)

2.812 .%2.993.013.073.093.103.113.133.183.193.203.213.233.25

δ 1 8 o

2.582.362.372.393.002.312.222.222.132.422.672.432.242.172.04

δ 1 3 c

1.111.061.080.870.261.061.030.970.970.910.820.860.871.170.62

a Age assignments have been rounded to the nearest10,000 years.

of modern assemblage-water mass relationships into thegeologic past. Although modern studies have shownthat benthic foraminiferal distributions tend to be con-trolled by specific water masses, it has been difficultto determine which of the many hydrographic proper-ties (temperature, alkalinity, silica, phosphate, nitrate,Σ C O 2 , etc.) is actually controlling their distribution.

This problem results from the fact that hydrographicproperties tend to be highly intercorrelated. Neverthe-less, quantitative analyses of benthic foraminifers stillprovide one of the best methods for studying the distri-bution of subsurface water masses in the past.

Factor analysis of 34 samples from Site 518 revealsthree distinct benthic foraminiferal assemblages, whichaccount for 92% of the total variance (Tables 6 and 7).Factor 1 is heavily dominated by Nuttalides umboni-fera; Factor 2 consists of a Globocassidulina subglobo-sa-Uvigerina peregrina assemblage; and Factor 3 islargely composed of an Oridorsalis umbonatus-Episto-minella exigua assemblage.

Changes in these benthic foraminiferal assemblagesduring the Pliocene of Site 518 are marked by three dis-tinct events (Fig. 4). From 4.1 to 3.6 Ma, the N. um-bonifera (Factor 1) and O. umbonatus-E. exigua (Fac-tor 3) assemblages dominate. At 3.6 Ma, these assem-blages are replaced by the G. subglobosa-U. peregrina(Factor 2) assemblage. This change at 3.6 Ma is best il-lustrated by the abundance of N. umbonifera, U. pere-grina, and E. exigua (Fig. 5). Between 3.6 and 3.2 Ma,U. peregrina increases to 10%, E. exigua decreases to< 1 %, and N. umbonifera averages 20%. The most dra-matic shift in benthic foraminiferal assemblages occursat 3.2 Ma when the G. subglobosa-U. peregrina assem-blage (Factor 2) is replaced by the N. umbonifera (Fac-tor 1) and O. umbonatus-E. exigua (Factor 3) assem-blages. U. peregrina disappears (0%), E. exigua perma-

910

PLIOCENE ABYSSAL CIRCULATION

Table 4. Oxygen and carbon isotopic results (in %0

PDB) of Planulina wuellerstorfi and/or Cibici-doides kullenbergi from Site 518.

Table 5. Oxygen and carbon isotopic results (in %0

PDB) of Orbulina universa from DSDP Site 518.

Core-section(interval in cm)

8-1, 70-728-1, 120-1228-2, 20-228-2, 70-728-2, 120-1228-3, 20-228-3, 70-728-3, 120-1229-1, 20-229-1, 70-729-1, 120-1229-2, 10-149-2,40-449-2, 70-729-2, 90-949-2, 120-1229-3, 10-1410-1, 70-7210-1, 120-12210-2, 20-2210-2, 51-5510-2, 59-6110-2, 90-9411-1, 10-1411-1,68-7211-1, 117-12211-2, 20-2211-2, 40-4411-2,70-7211-2, 120-12211-2, 130-13411-3, 19-2111-3, 31-3511-3, 50-52

Depth(m)

28.028.529.029.530.030.531.031.531.932.432.933.333.633.934.134.434.836.837.337.838.138.238.540.641.241.742.242.442.743.243.343.743.844.0

Agea

(Ma)

2.792.832.872.912.952.993.033.083.113.163.203.243.263.293.333.333.373.553.593.633.663.663.673.723.793.843.893.923.954.004.014.064.074.09

δ 1 8 θ

3.082.732.952.633.212.982.712.752.482.752.082.192.142.412.092.392.082.322.382.162.132.162.122.132.442.502.502.372.282.392.232.562.122.36

δ 1 3 C

0.500.43

-0.020.560.400.330.270.440.660.650.860.750.820.531.100.961.031.000.960.921.041.001.170.740.800.590.890.970.730.740.300.830.710.76

a Age assignments have been rounded to the nearest10,000 years.

nently increases to 10%, and N. umbonifera increasesby 30%.

In order to interpret these assemblage changes interms of water masses, the Pliocene assemblages mustbe compared with Recent core top studies. Lohmann(1978, 1981) studied benthic foraminiferal distributions(>250 µm) from the Rio Grande Rise, and recognizedfour taxonomically distinct assemblages which coincidewith the distribution of subsurface water masses: (1) aNuttalides umbonifera assemblage (>35OO m) coin-ciding with AABW; (2) a rmliohá-cibiciá-Oridorsalisassemblage (2100 to 3500 m) coinciding with NADW;(3) a Uvigerina assemblage (1500 to 2200 m) coincidingwith CPDW; (4) a Hoeglundina-Globocassidulina as-semblage (800 to 1500 m) coinciding with AAIW. Un-fortunately, these assemblage-water mass relationshipscannot be directly applied to the Site 518 data, owing toa difference in size fraction. Lohmann (1978, 1981) col-lected data at the > 250 µm size fraction, while countsfor the Site 518 samples were made at the > 150 µm sizefraction. Although the results of Lohmann (1978, 1981)are not directly applicable, some similarities do exist.

Core-section(interval in cm)

8-1, 70-728-1, 120-1228-2, 20-228-2, 70-728-3, 20-228-3, 70-728-3, 120-1229-1, 20-229-1, 70-729-1, 120-1229-2, 10-149-2, 20-229-2, 40-449-2, 70-729-2, 90-949-2, 120-1229-3, 10-1410-1, 20-2210-1, 40-4410-1, 70-7210-1, 120-12210-2, 20-2210-2, 51-5510-2, 59-6111-1, 10-1411-1, 20-2311-1, 68-7211-1, 117-12211-2,20-2211-2,40-4411-2, 70-7211-2, 90-9411-2, 120-12211-2, 130-13411-3, 19-21

Depth(m)

28.028.529.029.530.531.031.531.932.432.933.333.433.633.934.134.434.836.336.536.837.337.838.138.240.640.741.241.742.242.442.742.943.243.343.7

Agea

(Ma)

2.792.832.872.912.993.033.083.113.163.203.243.253.263.293.313.333.373.503.523.553.593.633.663.663.723.733.793.843.893.923.953.974.004.014.06

δ 1 8 θ

0.540.190.530.310.720.320.620.300.700.400.39

-0.090.430.37

-0.020.800.530.660.22

-0.090.240.630.300.320.81

-0.160.490.280.070.490.350.640.610.560.45

δ 1 3 c

2.212.362.021.962.212.492.532.402.231.981.732.141.892.192.302.601.851.832.012.072.052.262.122.482.031.881.821.381.671.591.881.711.951.772.15

a Age assignments have been rounded to the nearest10,000 years.

The N. umbonifera assemblage of our study (Factor1) is identical to that of Lohmann's (1978, 1981). Thepreference of this species for AABW has been demon-strated in all oceans (Streeter, 1973; Schnitker, 1974;Lohmann, 1978, 1981; Gofas, 1978; Corliss, 1978a,1978b, 1979b, 1983. The association of dominant N.umbonifera with AABW is partly the result of this spe-cies' tolerance for intense carbonate undersaturation ascompared with other species (Bremer and Lohmann,1982; Corliss, 1983).

Lohmann's (1981) miliotid-cibitid-Oràfo/ sαtó assem-blage may coincide with Factor 3 (O. umbonatus-E. ex-igua) from this study, thereby indicating NADW. Loh-mann (1981) didn't record E. exigua in his samples be-cause the average largest diameter of this species is about100 to 120 µm (Schnitker, 1980). However, a cursorycheck by Schnitker (p e r s o n a l communication) showedE. exigua to be present in Rio Grande Rise core tops.Evidence that Factor 3 (O. umbonatus-E. exigua) is as-sociated with NADW comes from Schnitker (1974) whofound an E. exigua assemblage associated with Norwe-gian Sea Outflow Water. The most important accessory

911

D. A. HODELL, J. P. KENNETT, K. A. LEONARD

618O(%oPDB)

0.4 0.0

δ13C (%o PDB)

2.4 2.0 1.8

Figure 3. Isotopic and biostratigraphic correlation between DSDP Hole 516A and Site 518. Isotopic data include oxygen andcarbon isotopes from Hole 516A (solid line) run on Globigerinoides sacculifera (Leonard et al., this volume) and from Site518 run on Orbulina universa (this study). Biostratigraphic data include first and last appearances of key planktonic fora-minifers from Hole 516A (solid line) and Site 518 (dashed line) (Berggren et al., this volume; Pujol, this volume). G. = Glo-borotalia; S. = Sphaeroidinella.

species in this assemblage was O. umbonatus. In theSouth Atlantic, this assemblage differs in that E. exiguadecreases in importance while O. umbonatus becomesdominant. This increased dominance of O. umbonatusover E. exigua at 30 °S may be the result of increasedmixing as NADW travels south.

Lastly, the G. subglobosa-U. peregrina assemblage(Factor 2) of our study may correspond to either theUvigerina (CPDW) or Globocassidulina (AAIW) as-semblage of Lohmann (1981). Corliss (1983) reports aG. subglobosa assemblage from the southeast IndianOcean which is associated with a mixture of AABW andCPDW with potential temperatures of 0.6 to 0.8°C.Schnitker (1974) also described a G. subglobosa-U. pe-regrina assemblage along the continental margin andmid-ocean ridge of the North Atlantic. This assemblageis associated with NADW with potential temperaturesbetween 2 and 4°C. We interpret the G. subglobosa-U.peregrina (Factor 2) assemblage in Site 518 as indicatinga shallower water mass of circumpolar origin. This wa-ter mass may have had slightly different physical andchemical properties than modern CPDW. For example,Corliss (1982) has presented benthic foraminiferal evi-dence for "glacial" CPDW which has distinctly differ-ent properties than its modern counterpart.

In summary, benthic foraminifers from the late Neo-gene of Site 518 can be grouped into three assemblageswhich are indicative of specific water masses in the geo-logic past (Fig. 4). These assemblages include: an N.umbonifera assemblage associated with AABW, a G.subglobosa-U. peregrina assemblage associated with

Table 6. Scaled varimax factor scores for each of the 35 taxa recordedin this study. The factor scores indicate the relative contributionof each of the benthic foraminiferal taxa to the five factor assem-blages.

Species

Nuttalides umboniferaEpistominella exiguaPlanutina wuellerstorflPyrgo spp.Oridorsalis umbonatusOridorsalis tenerUvigerina peregrinaGlobocassidulina subglobosaCibicidoides kullenbergiQuinquelocina spp.Quinqueloculina venustaFissurina spp.Paraflssurina spp.Nummoloculina irregularisPullenia bulloidesPullenia quinquelobaEhrenbergina spp.Gyroidinoides sodaniiGyroidinoides orbicularisHoeglundina elegansPullenia sp.Osangularia cullerMelonis pompiloidesSphaeroidina bulloidesBulimina rostrataCibicides spp.Laticarinina pauperataAnomalinoides sp.Opthalmidium acutimargoKarreriella spp.Eggerella spp.Cibicidoides cicatricosusLagena spp.Nonion spp.Lenticulina spp.

1

5.6610.7830.1210.280

-0.2880.061

-0.575-0.870-0.380

0.0870.0380.024

-0.103-0.002

0.6270.325

-0.237-0.480-0.170-0.019

0.126-0.048

0.1080.1650.011

-0.015-0.034

0.015-0.008

0.0720.004

-0.1050.138

-0.025-0.007

Factor assemblage

2

1.195-1.009

0.5330.1580.6090.0653.1044.1401.380

-0.0110.1710.3850.0340.0090.5320.7890.5010.7440.6760.0470.2490.261

-0.027-0.262

0.3090.1240.1080.0900.0460.2610.0070.3590.3340.455

-0.028

3

-0.0981.7210.5930.4735.171

-0.098-1.394

0.741-0.366

0.160-0.120

0.2920.0580.0010.284

-0.0480.0900.459

-0.179-0.001

0.100-0.155

0.9040.776

-0.250-0.007

0.006-0.089-0.022

0.1220.0570.100

-0.032-0.241

0.084

4

-0.7311.3020.1541.0220.3700.7122.445

-2.4600.224

-0.0460.217

-0.1580.035

-0.0054.0030.407

-0.1770.7970.713

-0.0180.2830.412

-0.6340.3200.525

-0.0570.0300.087

-0.010-0.171-0.041-0.169

0.3820.533

-0.039

5

-0.164-1.202

2.3710.159

-1.2340.302

-3.2860.6871.4370.2140.1940.7690.728

- 0.0222.4350.1850.1831.9240.606

-0.023-0.113-0.016

0.876-0.421-0.290-0.154

0.0190.139

-0.0290.1030.0090.606

-0.162-0.147

0.383

912

PLIOCENE ABYSSAL CIRCULATION

Table 7. Varimax factor loadings of the first five factors of 34 samplesfrom the late Neogene of Site 518. Only the first three factors,which explain ~ 92% of the total variance, were used in this study.

Sub-bottomdepth

(m)

28.028.529.029.530.030.531.031.531.932.432.933.433.934.436.336.837.337.838.240.741.241.742.242.743.243.744.045.345.849.550.050.551.053.9

Variance

Comm.

0.9840.9780.9770.9840.9860.9920.9370.9920.9360.9400.9480.9280.9760.9800.9700.9320.9440.9700.9830.9500.9610.9690.9720.9490.8870.9690.9560.9730.9700.9720.9810.9590.9540.959

Cumulative Variance

1

0.7600.8430.7870.8550.9140.8930.7590.9470.4570.7070.6710.5440.7460.6150.5550.5360.3790.6200.7890.2000.6540.3750.4930.5970.6270.9220.3990.4370.2660.0950.4090.1080.3600.561

39.36339.363

2

0.2850.3820.3230.3770.2260.2350.2360.2260.5250.3280.3290.4780.5490.7330.6990.7430.7800.7040.5740.5460.5010.5700.5950.5670.4550.3090.3050.7940.9090.9350.8240.6870.8080.739

33.35972.722

Factor

3

0.5310.3160.4780.3300.3110.3670.4870.2030.6680.3880.6020.5550.3070.1940.2460.1970.4290.2380.1100.7760.5070.6930.5840.4480.5320.0830.8270.3870.2490.2990.3630.4430.3280.297

19.58692.308

4

0.2010.0360.150

-0.0490.0470.0620.2570.046

-0.0210.3380.1310.2310.0500.1670.3270.1660.030

-0.141-0.093-0.011-0.007-0.140-0.087-0.034

0.012-0.085

0.139-0.015-0.044

0.0140.0450.0720.2390.0991.919

94.227

5

0.0330.1410.0540.014

-0.026-0.026

0.0530.0140.0770.2610.0960.2060.143

-0.0180.0760.1660.0880.1180.0940.1000.1600.0640.1630.2640.0640.100

-0.019-0.029

0.0930.0030.0020.5240.0810.0202.000

96.227

CPDW, and an O. umbonatus-E. exigua assemblage as-sociated with NADW.

BENTHIC OXYGEN ISOTOPIC RECORDS

Variations in the oxygen isotopic composition of fo-raminiferal tests are known to reflect changes in salin-ity, temperature, and ice volume. Although the controlson the δ 1 8 θ record are well understood, the degree towhich each affects the record is often debated.

The benthic δ 1 8 θ record of Site 518 is relatively stableuntil the late Pliocene. At 3.2 Ma, a large enrichment of~ l.l%o peak (0.45%0 average) occurs (Fig. 6). The ben-thic δ 1 8 θ record of Site 517 undergoes a 0.8%0 steplikeenrichment between 3.3 and 2.8 Ma (Fig. 7). In Hole516A, a O.5%o enrichment is also observed at this time(Leonard et al., this volume). Therefore, a benthic 18Oenrichment is observed in all Leg 72 sites at 3.2 Ma, andthe magnitude of the enrichment appears to increasefrom Hole 516A (1313 m water depth, 0.5%) to Site 517(2963 m water depth, 0.8‰) to Site 518 (3944 m waterdepth, 1.1 %0). Further isotopic work will be required tosubstantiate if the 18O enrichment at 3.2 Ma is indeeddepth dependent.

BENTHIC CARBON ISOTOPIC RECORDS

The δ13C distribution in the Atlantic provides a valu-able fingerprint for distinguishing water masses fromeach other. In most cases, the Atlantic δ13C data parallelthe distribution of dissolved O2. For example, AABWcan be identified by its low δ13C value north to the equa-tor; and the core of NADW, with its δ13C maximum,can be traced as far as 40° S (Kroopnick, 1980). Near thelatitude of the Vema Channel (~30°S), AABW has aδ13C value near 0.5%, NADW has a δ13C compositionnear 1.0%, and CPDW has values near 0.65% (Table 8).Because the differences in carbon isotopic values betweenwater masses are large, it should be possible to detectchanges in the vertical water mass structure of the VemaChannel by monitoring changes in bathymetric Δ13Cgradients.

Unfortunately, changes in the δ13C of the oceansthrough time are poorly understood. The synchroneityof carbon isotopic changes in different ocean basinssuggests that the δ13C record is at least partly controlledby a global mechanism. Recent theories have includedtransfer of terrestrial or shelf carbon to the oceans(Shackleton, 1977; Broecker, 1981), changes in the C:Pratio of seawater (Broecker, 1981), changes in organ-ic carbon burial rates (Scholl and Arthur, 1980), andchanges in oceanic circulation (Bender and Keigwin,1979). In this study, we are most interested in identify-ing those changes in the benthic δ13C record that are re-lated to changes in local deep circulation. In a similarstudy, Curry and Lohmann (1982) assumed that globalchanges in δ13C will equally affect all water masses with-in the mixing time of the oceans. Thus, changes in thebathymetric distribution of δ13C should be related tochanges in abyssal circulation.

Benthic carbon isotopic records have been generatedfor the Pliocene of Hole 516A (Leonard et al., this vol-ume), the late Pliocene of Site 517, and the Pliocene ofSite 518. Carbon isotopic gradients (Δ13C) have beencalculated between sites whenever possible. We assumethat the Δ13C gradients between Holes 516A (1313 m),517 (2963 m), and 518 (3944 m) reflect the degree of sim-ilarity and/or difference in the nutrient content of inter-mediate, deep, and bottom water masses, respectively.All isotopic analyses were made from specimens of P.wuellerstorfi and/or C. kullenbergi. P. wuellerstorfi ac-curately reflects the isotopic gradient (-1%) betweenAtlantic and Pacific oceans (Belanger et al., 1981; Gra-ham et al., 1981). In the Vema Channel, Curry and Loh-mann (1982) and Williams and Healy-Williams (1982)have shown that the δ13C of P. wuellerstorfi correctlyrecords the Δ13C gradient between AABW and NADW.

The benthic carbon isotopic record at Site 518 can bedivided into three segments (Fig. 8). The interval be-tween 4.1 and 3.6 Ma is marked by relatively light δ13Cvalues. At 3.6 Ma, a 0.5% peak (0.26% average) enrich-ment occurs, and heavy values persist till 3.3 Ma. At3.3 Ma, δ13C values begin a lightening trend culminatingin a 1.0% (0.49% average) depletion.

The Δ13C gradient between Holes 516A and 518 iscompletely dominated by δ13C changes in the deeper Site518 (Fig. 9). This suggests that fundamental circulation

913

D. A. HODELL, J. P. KENNETT, K. A. LEONARD

Factor scores squared0 +1

+1 0 +1T

Factor 2

r r

Factor 3

Figure 4. Relative importance of the three most significant benthic foraminiferal factorsversus age in Site 518. Rotated varimax factors have been squared and plotted versusage in millions of years. Factor 1 = Nuttalides umbonifera assemblage (Antarctic Bot-tom Water); Factor 2 Globocassidulina subglobosa-Uvigerina peregrina assem-blage (Circumpolar Deep Water); Factor 3 = Oridorsalis umbonatus-Epistominellaexigua assemblage (North Atlantic Deep Water). Major changes in the relative im-portance of benthic foraminiferal assemblages (indicated by dotted lines) occur at 3.6Ma, 3.3 Ma, and 3.1 Ma. The time scale was derived by isotopic and biostratigraphiccorrelation with Hole 516A.

changes are occurring within the deeper water masseswhile the shallower water masses are less affected. Themodern gradient between these two depths is - 0.15, re-flecting δ13C differences between AABW and CPDW.The Pliocene gradient undergoes rapid oscillations dur-ing the interval from 4.1 to 3.6 Ma, though no permanentchange is observed. From 3.6 to 3.2 Ma, the gradient issuch that δ13C values at Site 518 (3944 m) are slightlyheavier than those at Hole 516A(1313m).A large nega-tive increase in the Δ13C gradient occurs at 3.2 Ma whenSite 518 values lighten by 1.0%0 relative to values at Hole516A.

The modern Δ13C gradient between Sites 518 and 517is -0.5 reflecting the difference between AABW andNADW. The Δ13C gradient between these sites (Fig. 10)during the late Pliocene (3.3 to 2.8 Ma) also averagesabout -0.5, indicating that the AABW/NADW transi-tion during this interval was similar to its present-dayposition.

Lastly, the Δ13C gradient between Sites 517 (2963 m)and 516A (1313 m) averages +0.10 between 3.3 and2.8 Ma (Fig. 11) compared with the modern gradient of+ 0.35, reflecting the difference between NADW and

CPDW. The direction of the gradient during the late

914

10% 30 50 0 10

PLIOCENE ABYSSAL CIRCULATION

10 20

/V. umbonifera U. peregrina E. exigua

Figure 5. Percentages of Nuttαlides umbonifera, Uvigerina peregrina, and Epistominella exigua versus age in Site 518.Wavy line represents a 1.5 Ma hiatus (~5.5 to 4.0 Ma) separating the Miocene and Pliocene in Site 518. Dotted linesindicate times of major faunal turnovers which are coincident with major changes in the benthic stable isotopic record.

Pliocene is correct though the magnitude is slightly dif-ferent from the modern gradient.

DISCUSSION

Changes in the benthic foraminiferal assemblages atSite 518 are coincident with bathymetric carbon isotopicchanges, indicating a consistent response to circulationchanges. The Pliocene record can be divided into threesegments.

The interval between 4.1 and 3.6 Ma is marked by adominance of the N. umbonifera (AABW) and O. um-bonatus-E. exigua (NADW) assemblages with relativelylight δ13C values in Hole 518 (3944 m). The δ13C gradi-ent between Holes 518 (3944 m) and 516A (1313 m) un-dergoes rapid oscillations though no permanent changein the gradient is recorded. The vertical water mass struc-ture of the Vema Channel during this interval was prob-ably similar to the present, with both AABW andNADW present in the western South Atlantic. A majorcooling trend was initiated in the southern oceans at4.3 Ma (Ciesielski and Weaver, 1974). Evidence for thiscooling trend includes a northward displacement of sili-ceous sediments off East Antarctica at 3.86 Ma (Wea-ver, 1973), possible expansion of the west Antarctic icesheet to its present dimensions as early as 3.95 to 3.90 Ma(Ciesielski and Weaver, 1974), and major glaciation inArgentine Patagonia at 3.59 Ma (Mercer, 1976). Osbornand others (1982) have found a hiatus maximum in thesouthwest Indian Ocean which suggests increased bot-tom water activity in response to the re-establishment ofthe west Antarctic ice sheet in the late Gilbert Chron(-4.2 to 3.6 Ma).

From 3.6 to 3.2 Ma, the N. umbonifera (AABW) andO. umbonatus-E. exigua (NADW) assemblages are re-placed by a G. subglobosa-U. peregrina (CPDW) as-semblage, δ13C values lighten by 0.5‰ in Site 518, andno strong Δ13C gradient exists between Holes 518(3944 m) and 516A (1313 m). The presence of a faunalassemblage indicative of CPDW at 3944 m, coupledwith the absence of a strong Δ13C gradient between3944 m (Site 518) and 1313 m (Hole 516A) suggests thatshallow circumpolar waters expanded to depths of atleast 3944 m during this interval. This circulationchange may be related to diminished production ofAABW and NADW in response to the formation of theNorthern Hemisphere Barents Ice Sheet at -3.5 Ma(Osborn et al., 1982).

The most profound shift in all parameters at Site 518occurs between 3.3 and 3.2 Ma. The G. subglobosa-U.peregrina (CPDW) assemblage is rapidly replaced by theN. umbonifera (AABW) and O. umbonatus-E. exigua(NADW) assemblages, the Δ13C gradient between Holes518 and 516A records a large negative increase, and a1.1 %0 enrichment is recorded in benthic δ 1 8 θ. Thesechanges record the upward displacement by AABW andNADW of shallow circumpolar waters, which had ex-panded to depths of 3944 m prior to 3.2 Ma. This changein vertical water mass structure was probably related toa conspicuous global cooling event at 3.2 Ma (Kennett etal., 1979; Thunell, 1979) and/or the final closure of theCentral American Seaway by 3.0 Ma (Keigwin, 1982).

In order to reconstruct the vertical water mass struc-ture of the Vema Channel during this late Pliocene inter-val, we have averaged δ13C values for each site between

915

D. A. HODELL, J. P. KENNETT, K. A. LEONARD

+3.32.7

δ18O(%oPDB)

+2.9 +2.7 +2.0Table 8. Average modern δ^C composition (%0

PDB) of water masses at the latitude of theVema Channel in the southwestern Atlantic.

" c gradients are calculated by taking theδ^C difference between adjacent water masses.

δ 1 3 C (modern) Δ 1 3 C gradient(% PDB)a (modern) Water massb

1.800.900.651.000.50

-0.90-0.35+ 0.35-0.50

Surface waterAAIWCPDWNADWAABW

From Kroopnick, 1980.For abbreviations, see Table 1.

δ13C(%oPDB)

+0.8 +0.4 -0.1

Figure 6. Benthic δ 1 8 θ (‰ PDB) from analyses of Planulina wueller-storfi and/or Cibiddoides kullenbergi in DSDP Site 518 plottedagainst age in millions of years.

3.2δ18O(%oPDB)

2.8 2.4

Figure 7. Benthic δ 1 8 θ (‰ PDB) from analyses of Planulina wueller-storfl and/or Cibiddoides kullenbergi in DSDP Site 517 plottedagainst age in millions of years.

Figure 8. Benthic δ13C (%0 PDB) from analyses of Planulina wueller-storfi and/or Cibiddoides kullenbergi in DSDP Site 518 plottedagainst age in millions of years.

916

Δ1 3C

Figure 9. The Δ 1 3 C gradient between Site 518 (3944 m) and Hole 516A(1313 m) calculated by subtracting the benthic δ 1 3C values at Hole516A from values at Site 518. Δ 1 3 C values are plotted against agein millions of years.

3.2 and 2.7 Ma, calculated Δ13C differences betweensites, and compared the differences with modern Δ13Cgradients (Table 9). The agreement between the late Pli-ocene (3.2 to 2.7 Ma) and present-day gradients indicatesthat circulation patterns during this late Pliocene inter-val were similar to those of the modern western SouthAtlantic.

Ledbetter (1981) suggests that AABW production di-minished at 3.2 Ma in the southeastern Indian, theSouth Atlantic, and the South Pacific oceans. He pro-poses that decreased AABW activity at 3.2 Ma can beattributed to a decrease in NADW production, whichprovides part of the salinity component of AABW. Ourstudy does not record circulation intensity, but rathermonitors the positions of water masses in the VemaChannel through the Pliocene. Benthic foraminifers andΔ13C gradients between 3.2 and 2.7 Ma indicate thatboth AABW and NADW were present in the Vema

PLIOCENE ABYSSAL CIRCULATION

Δ 1 3 C

T-0.0

Figure 10. The Δ 1 3 C gradient between Sites 518 (3944 m) and 517(2963 m) calculated by subtracting benthic δ 1 3C values at Site 517from values at Site 518. Δ 1 3 C values are plotted against age in mil-lions of years.

1 3 c+0.4

Figure 11. The Δ 1 3 C gradient between Site 517 (2963 m) and Hole 516A(1313 m) calculated by subtracting benthic δ 1 3 C values at Hole516A from values at Site 517. Δ 1 3 C values are plotted against age inmillions of years.

Channel during this interval. The data suggest that thevertical water mass structure of the Vema Channel dur-ing the late Pliocene (3.2 to 2.7 Ma) was similar to thepresent-day; thus, we do not support decreased produc-tion of AABW and NADW during this interval. Keig-win (1982) has found an increased contrast between At-lantic and Pacific benthic foraminiferal δ13C at 3.0 Ma,which suggests that NADW production actually in-creased at this time.

CONCLUSIONS

1. Quantitative analyses of benthic foraminifers fromthe Pliocene of Site 518 reveal three assemblages whichare indicative of specific water masses in the geologicpast. These include: an AT. umbonifera assemblage asso-ciated with AABW, a G. subglobosa-U. peregrina as-

917

D. A. HODELL, J. P. KENNETT, K. A. LEONARD

Table 9. Averaged δ ^ C values (%0 PDB) from planktonic foraminifers and Holes 516A, 517, and518 benthic foraminifers for the interval between 3.2 and 2.7 Ma. Δ ^ C gradients are calculatedbetween Holes 516A, 517, and 518, and compared with modern Δ ^ C radients from Kroopnick(1980). The similarity of modern carbon isotopic gradients with those calculated between 3.2 and2.7 Ma indicates that the vertical water mass structure of the Vema Channel during the latePliocene was similar to that of the present day. Sites 516A, 517, and 518 were positioned inCPDW, NADW, and AABW respectively, during the interval from 3.2 to 2.7 Ma.

Hole depth(m)

Δ 1 3 C gradient

5 1 3 C (‰ PDB)a 3.2-2.7 Ma From Kroopnick, 1980

Modernwater masstransition*5

Planktonics516A517518

~50m131329633944

2.080.820.920.45

-1.26+ 0.10-0.47

-0.90-1.25-0.35-0.50

Surface/AAIWSurface/CPDWNADW/CPDWAABW/NADW

a Averaged data between 3.2 and 2.7 Ma.•3 For abbreviations, see Table 1.

semblage associated with CPDW, and an O. umbona-tus-E. exigua assemblage associated with NADW.Changes in benthic foraminiferal assemblages are coin-cident with changes in the carbon isotopic record, in-dicating a consistent response to abyssal circulationchanges.

2. Changes in the vertical water mass structure of theVema Channel during the Pliocene appear to be relatedto climatic events. Three fundamental changes in watermass structure have been identified.

3. From 4.1 to 3.6 Ma, the vertical water mass struc-ture was similar to today, with both AABW and NADWpresent in the western South Atlantic. This result is con-sistent with hiatus abundance information from thesouthwest Indian Ocean (Osborn et al., 1982), whichsuggests increased bottom water activity during this in-terval, perhaps associated with the re-establishment ofthe West Antarctic Ice Sheet in the late Gilbert Chron(-4.2 to 3.6 Ma).

4. From 3.6 to 3.2 Ma, shallow circumpolar watersexpanded to depths of at least 3944 m. Evidence for theexpansion of a shallower water mass to depths of— 4000 m include the dominance of a CPDW benthicforaminiferal assemblage at Site 518 (3944 m) and lackof a Δ13C gradient between Holes 516A (1313 m) and518 (3944 m).

5. At 3.2 Ma, circumpolar waters were vertically dis-placed by AABW and NADW. This change in verticalwater mass structure was probably related to climaticdeterioration and/or the final closure of the CentralAmerican Seaway at approximately 3.2 Ma. A compar-ison of the present-day vertical water mass structure ofthe Vema Channel with a reconstruction between 3.2and 2.7 Ma indicates that circulation patterns duringthis late Pliocene interval were similar to those of themodern western South Atlantic.

ACKNOWLEDGMENTS

The authors would like to thank W. B. Curry, D. A. Dunn, D. A.Johnson, G. P., Lohmann, G. A., Mead, and D. F. Williams, for dis-cussions on various aspects of this work; B. H. Corliss for reviewingthe manuscript; D. F. Williams for providing Site 516A data and Site517 samples; G. A. Mead for assistance with taxonomic identifica-tions; and D. R. Muerdter for computer assistance. This research wassupported by NSF Grant #OCE-7914594.

REFERENCES

Barker, P. F., Carlson, R. L., Johnson, D. A., and Shipboard Scien-tific Party, 1981. Deep Sea Drilling Project Leg 72: southwest At-lantic paleocirculation and Rio Grande Rise tectonics. Geol. Soc.Am. Bull., 92:294-309.

Belanger, P. E., Curry, W. B., and Matthews, R. K., 1981. Core-topevaluation of the paleo-oceanography utility of benthic foraminif-eral isotopic ratios. Palaeogeogr. Palaeoclimatol. Palaeoecol., 33:205-220.

Bender, M. L., and Keigwin, L. D., 1979. Speculation about the up-per Miocene change in abyssal Pacific dissolved bicarbonate δ1 3C.Earth Planet. Sci. Lett., 45:383-393.

Bremer, M. L., and Lohmann, G. P., 1982. Evidence for primary con-trol of the distribution of certain Atlantic Ocean benthonic Fora-minifera by degree of carbonate saturation. Deep Sea Res., 12:987-988.

Broecker, W. S., 1981. Glacial to interglacial changes in ocean and at-mosphere chemistry. In Berger, A. (Ed.), Climatic Variations andVariability: Facts and Theories: Boston (Reidel), pp. 111-121.

Ciesielski, P. F., and Weaver, F. M., 1974. Early Pliocene temperaturechanges in the Antarctic Seas. Geology, 2:511-515.

Corliss, B. H., 1978a. Studies of deep-sea benthonic Foraminifera inthe Southern Ocean (Ph.D. dissert.). University of Rhode Island.

, 1978b. Studies of deep-sea benthonic foraminifera in thesoutheast Indian Ocean. Antarctic J., 13:116-117.

, 1979a. Taxonomy of Recent deep-sea benthonic foraminif-era from the southeast Indian Ocean. Micropaleo., 25(1):1-19.

1979b. Recent deep-sea benthonic foraminiferal distribu-tions in the southeast Indian Ocean: Inferred bottom-water routesand ecological implications. Mar. Geol., 31:115-138.

1982. Linkage of North Atlantic and Southern Ocean deep-water circulation during glacial intervals. Nature, 298:458-460.

1983. Distribution of Holocene deep-sea benthonic forami-nifera in the southwest Indian Ocean. Deep-Sea Res., 30:95-117.

Culp, S. K., 1977. Recent benthic Foraminifera of the Ontong JavaPlateau [M.S. Thesis]. University of Hawaii, Honolulu.

Curry, W. B., and Lohmann, G. P., 1982. Carbon isotopic changes inbenthonic Foraminifera from the western South Atlantic: Recon-struction of glacial abyssal circulation patterns. Quat. Res., 18:218-235.

Duplessy, J. C , Moyes, J., and Pujol, C , 1980. Deep water formationin the North Atlantic Ocean during the last ice age. Nature, 286:479-482.

Gofas, S., 1978. Une approche du paléoenvironment océanique: Lesforaminifères benthiques calcaires, traceurs de la circulation abys-sale (Ph.D. dissert.). L'Université de Bretagne Occidentale, France.

Graham, D. W., Corliss, B. H., Bender, M. L., and Keigwin. L. D.,1981. Carbon and oxygen isotopic disequilibria of Recent deep-seabenthic foraminifera. Mar. Micropaleo., 6:483-497.

Ingle, J. C , Keller, G., and Kolpack, R. J., 1980. Benthic foraminif-eral biofacies, sediments, and water masses of the southern Peru-Chile Trench area, southeastern Pacific Ocean. Micropaleo., 26:113-150.

918

PLIOCENE ABYSSAL CIRCULATION

Keigwin, L. D., 1982. Isotopic paleoceanography of the Caribbeanand East Pacific: Role of Panama uplift in Late Neogene time. Sci-ence, 217:350-353.

Keigwin, L. D., and Thunell, R. C , 1979. Middle Pliocene climaticchange from the western Mediterranean from faunal and isotopictrends. Nature, 282:294-296.

Kennett, J. P., Shackleton, N. J., Margolis, S. V., Goodney, D. E,Dudley, W. C , and Kroopnick, P. M., 1979. Late Cenozoic oxy-gen and carbon isotopic history and volcanic ash stratigraphyDSDP Site 284, South Pacific. Am. J. Sci., 279:52-69.

Klovan, J. E., and Imbrie, J., 1971. An algorithm and Fortran IV forlarge scale Q-mode factor analysis. Int. Assoc. Math. Geol. /., 3:61-78.

Kroopnick, P. M., 1980. The distribution of δ13C in the AtlanticOcean. Earth Planet. Sci. Lett., 49:469-484.

Ledbetter, M. L., 1981. Palaeooceanographic significance of bottom-current fluctuations in the Southern Ocean. Nature, 294:554-556.

Lohmann, G. P., 1978. Abyssal benthic Foraminifera as hydrographicindicators in the western South Atlantic. J. Foram. Res., 8(1):6-34.

, 1981. Modern benthic foraminiferal biofacies: Rio GrandeRise. EOS, 62(45):903. (Abstract)

Mercer, J. H., 1976. Glacial history of southernmost South America.Quat. Res., 6:125-166.

Osborn, N. I., Ciesielski, P. F., and Ledbetter, M. T., 1982. Discon-formities and paleoceanography in the southwest Indian Oceanduring the last 5.4 million years. Geol. Soc. Am. Abst. Prog.,14:652.

Reid, J. L., Nowlin, W. D., and Patzert, W. C , 1977. On the charac-teristics and circulation of the Southwestern South Atlantic Ocean./. Phys. Oceanog., 7:62-91.

Resig, J. M., 1981. Biogeography of benthic Foraminifera of the north-ern Nazca Plate and adjacent continental margin. In Kulm, L. D.,Dymond, J., Dasch, E. J., and Hussong, D. M. (Eds.), Nazca

Plate: Crustal Formation and Andean Convergence, Geol. Soc.Am. Mem., 154:619-644.

Schnitker, D., 1974. West Atlantic abyssal circulation during the past120,000 years. Nature, 248:385-387.

, 1980. Quaternary deep-sea benthic foraminifers and bot-tom water masses. Ann. Rev. Earth Planet. Sci., 8:343-70.

Scholl, P. A, and Arthur, M. A., 1980. Carbon isotopic fluctuationsin Cretaceous pelagic limestones: Potential stratigraphic and pe-troleum exploration tool. AAPG Bull., 64:67-87.

Shackleton, N. J., 1977. Carbon-13 in Uvigerina: Tropical rainforesthistory and the equatorial Pacific carbonate dissolution cycles. InAndersen, N. R., and Malahoff, A. (Eds.), The Fate of Fossil FuelCO2 in the Oceans: New York (Plenum Press), pp. 401-427.

Shackleton, N. J., and Opdyke, N. D., 1977. Oxygen isotope and pa-leomagnetic evidence for early Northern Hemisphere glaciation.Nature, 270:216-219.

Streeter, S. S., 1973. Bottom water and benthonic Foraminifera in theNorth Atlantic: Glacial-interglacial contrasts. Quat. Res., 3(1):131-141.

Thunell, R. C , 1979. Pliocene-Pleistocene paleotemperature and pa-leosalinity history of the Mediterranean Sea: Results from DSDPSites 125 and 132. Mar. Micropaleo., 4:173.

Weaver, F. M., 1973. Pliocene paleoclimatic andpaleoglacial historyof East Antarctica recorded in deep-sea piston cores. Florida StateUniv., Sedimentological Res. Lab., Contrib. 36.

Williams, D. F., and Healy-Williams, N., 1982. Vertical isotope gra-dients in Recent benthic Foraminifera and the water column of theVema Channel. EOS, 63:359. (Abstract)

Woodruff, F., Savin, S. M., and Douglas, D. G., 1981. Miocene sta-ble isotope record: A detailed deep Pacific Ocean study and its pa-leoclimatic implications. Science, 212:665-668.

Date of Initial Receipt: November 2,1982

919