A total of 15 genera and 50 species of planktonic foraminifers were identified at this site (Table 1). Our uppermost sample (Sample 164-997A-1H-1, 39-41 cm, 0.39 mbsf) contains pink specimens of Globigerinoides ruber, Globorotalia fimbriata, Globorotalia ungulata, and Bolliella calida calida, and is late Pleistocene to Holocene in age. Although the extinction of pink G. ruber is dated at 120 ka in the Pacific Ocean (Thompson et al., 1979), this datum is not useful in the Atlantic Ocean. Bé and Hamlin (1967) and Bé and Tolderlund (1971) reported the common occurrence of living pink G. ruber in modern plankton tows from the North Atlantic Ocean. No reliable markers of late Pleistocene such as Globorotalia menardii flexuosa and Pulleniatina finalis are present in the core top sample. However, the dominant occurrence of Emiliania huxleyi in Samples 164-997A-1H-1, 0-1 cm, and 1H-CC indicates latest Pleistocene or Holocene age (<85 ka; Shipboard Scientific Party, 1996). The Leg 164 Shipboard Scientific Party also reported the existence of a significant coring gap or a hiatus between Cores 164-997A-1H and 2H because E. huxleyi decreases in abundance in Sample 164-997A-2H-1, 0-1 cm (2.90 mbsf). The first occurrence of E. huxleyi is placed at Sample 164-997A-2H-3, 39 cm (6.29 mbsf), and the lower part of Core 164-997A-2H is assigned to upper Pleistocene Subzone CN14b.
In sediments older than late Pleistocene, the following five datums of planktonic foraminifers are recognized at this site (Fig. 2); the first occurrence (FO) of Globorotalia hirsuta (0.45 Ma), the last occurrence (LO) of Globorotalia tosaensis (0.65 Ma), the LO of Globigerinoides obliquus (1.77 Ma), the FO of Globorotalia truncatulinoides (2.0 Ma) and the LO of Globorotalia exilis (2.15 Ma) in descending order. The age of these datums are based on Berggren et al. (1995a, 1995b). The lowest sample in this study contains no specimens of Globorotalia miocenica and is assigned to the PL6 zone of Berggren et al. (1995b). Therefore, the section studied here ranges from latest Pliocene to Holocene in age (0-2.15 Ma).
The lower boundary of the Jaramillo normal chron is placed at 52 mbsf, and the Pleistocene/Pliocene boundary (the upper boundary of the Olduvai normal chron) is at 72 mbsf. The Brunhes/Matuyama boundary and top of Jaramillo normal chron are tentatively defined at 36 and 41 mbsf, respectively (Shipboard Scientific Party, 1996). Hence, three nannofossil datums ([1] extinction of large Gephyrocapsa, [2] reappearance of medium-size Gephyrocapsa, and [3] extinction of Discoaster brouweri) do not match the magnetostratigraphy (Fig. 2). Globigerinoides obliquus was also found to be unreliable as a marker of the Pleistocene/Pliocene boundary at this site because this species disappeared at 70.94 mbsf, about 1.06 m from the top of the Olduvai Subchron. The discrepancies between magnetostratigraphy and biostratigraphy of calcareous fossils may indicate redeposition of marker species or an unreliable magnetostratigraphy. The Shipboard Science Party identified a short hiatus in the CN13b subzone, around Core 164-997A-7H, just below the lower boundary of Jaramillo Subchron. Although the magnetostratigraphic study demonstrated that the base of Jaramillo Subchron is well defined, the biostratigraphic ages of Samples 164-997A-7H-1, 39-41 cm, to 8H-2, 54.89 cm (52-55.97 mbsf; 1.07- to 1.46-Ma interval) are not clearly defined in this study.
The presence/absence behavior of the G. menardii complex in Pleistocene sediments from the tropical Atlantic has been well documented in many previous studies (Ericson and Wollin, 1956, 1968; Ruddiman, 1971). The abundance peaks and barren zones of this group correspond, to some extent, with warm isotopic stages (odd numbered) and cooling isotopic stages (even numbered). Hays et al. (1969) correlated the G. menardii "Y," "W," and "U" zones proposed by Ericson and Wollin (1956, 1968) with isotope Stages 2-4, 6, and 15 of Emiliani (1966).
We used the abundance of the Globorotalia menardii complex (menardii, tumida, ungulata, and exilis) as time stratigraphic markers and climatic indices during the Brunhes instead of isotope stages, because detailed isotopic records are not yet available for Site 997. We regard the occurrence of G. menardii as an indicator of warm interglacial conditions, whereas barren zones represent cold, glacial stages. Furthermore, we recognized a temperate assemblage that is characterized by the occurrence of G. hirsuta, associated with a few to common G. truncatulinoides, G. crassaformis, and Globigerinita glutinata. This group is considered to reflect transitional conditions, because G. truncatulinoides and G. hirsuta are particularly abundant in the central gyre of the temperate North Atlantic (Kipp, 1976; Bé, 1977). The climatic curve in Hole 997A and its correlation with a standard oxygen isotope curve are shown in Figure 2. In the upper part of the Brunhes (0-0.26 Ma), the presence/absence cycle of G. menardii is repeated in a pattern similar to that of the G. menardii zones of Ericson and Wollin (1956, 1968; Fig. 2). However, the fluctuations in G. menardii abundance in Hole 997A do not correlate with the standard oxygen isotope curve and Ericson and Wollin's G. menardii zones below the lower part of Brunhes (Fig. 2). Prior to the Brunhes, fluctuations in G. menardii abundance may be controlled by other aspects of the ocean surface conditions, rather than just surface temperature in Hole 997A.
Coiling changes in Pulleniatina are excellent biostratigraphic markers in the Indo-Pacific Ocean during the Pleistocene to late Pliocene. There are eight coiling-shift events numbered L1 to L8 in the Pacific Ocean, whereas the coiling patterns in the Atlantic Ocean are relatively constant, and only two events (AL1 and AL2) are recognized (Saito, 1976). In Hole 997A, the left-coiling form of Pulleniatina was observed in the interval between 82.39 and 83.89 mbsf (Samples 164-997A-11H-2, 39-41 cm, and 164-997A-11H-3, 39-41 cm), close to the bottom of the Olduvai Subchron (Fig. 2). This interval is identical with the AL2 event of Saito (1976). Thus, AL1 is probably missing.
On the other hand, most Pleistocene globorotaliid species display preferred coiling patterns and sometimes change their coiling mode quickly. For example, G. crassaformis is usually sinistral, but there are two intervals of random coiling at 18.79 mbsf (Sample 164-997A-3H-5, 39-41 cm) and dextral coiling from 56.39 to 59.44 mbsf (Samples 164-997A-8H-3, 49-51 cm, to 8H-5, 49-51 cm; Fig. 2). Globorotalia viola switched its coiling pattern several times during the Pleistocene. However, a systematic coiling trend in this species cannot be clearly identified because of its discontinuous occurrence throughout Hole 997A.
Two coiling populations of the temperate species Globorotalia truncatulinoides exist in the surface sediments of the North and South Atlantic. In the North Atlantic, right-coiling provinces appear to be separated from each other by a left-coiling province in the middle of the subtropical gyre (Bé and Tolderlund, 1971). The left-coiling specimens are dominant in the modern northwestern Sargasso Sea, comprising 5%-98% of the total in the surfaces waters, whereas right-coiling forms are abundant in the tropical Atlantic and northeast Atlantic (Ericson et al., 1954; Bé and Hamlin, 1967; Bé and Tolderlund, 1971; Kipp, 1976). The two coiling variants of G. truncatulinoides have been utilized as stratigraphic markers in equatorial Pleistocene sediments of the Atlantic Ocean (Ericson and Wollin, 1956, 1968; Ruddiman, 1971). Two abundance peaks of left-coiling G. truncatulinoides are known during the Brunhes Epoch, within the "X" and "U" zones of Ericson and Wollin's zones. During the Jaramillo event, an additional left-coiling spike is present in the lowermost "T" zone.
In Hole 997A, dextral specimens of G. truncatulinoides are dominant in many samples, but sinistral specimens are found in Sample 164-997A-1H-1, 39-41 cm (0.39 mbsf, S1), 2H-3, 39-41 cm (6.29s mbsf, S2), and 4H-6, 114-116 cm (30.54 mbsf, S3), and random-coiled populations are present in Sample 6H-4, 114-116 cm (46.54 mbsf, R; Fig. 2). The stratigraphic levels of sinistral G. truncatulinoides probably correlate with the "Z," "V," and "T" zones of Ericson and Wollin (1956, 1968). We suggest that the coiling shift from dextral to sinistral in G. truncatulinoides is not useful as a stratigraphic marker in the temperate region in Hole 997A. The coiling shift may be related with water mass change between tropical Atlantic water (or western boundary current water) and the North Atlantic subtropical gyre over the Blake Ridge.
The total abundance (per 10-cm3 sample) of planktonic foraminifers shows extremely large fluctuations, ranging from 62 (Sample 164-997A-6H-1, 39-41 cm, 41.29 mbsf) to about 14,370 specimens (Sample 164-997A-3H-6, 39-41 cm, 20.29 mbsf). There are 11 abundance peaks of over 2000 specimens/sample within the Brunhes Chron and five in the Matuyama (Fig. 3). Species of Globigerinoides are dominant throughout the Pliocene-Pleistocene and fluctuate in abundance between 35% and 55% (Fig. 4). The genus Globorotalia comprises 10%-20% of the planktonic foraminifer assemblage, but decreases to less than 10% in the interval between the base of the Jaramillo and the top of the Olduvai Subchron (Sections 164-997A-6H-5 through 9H-5). Species of Neogloboquadrina also occur abundantly (10%-20%), and increase rapidly to over 40% of the assemblage within the Jaramillo Subchron (Core 164-997A-6H). The globular forms of Globigerina, Globigerinita, and Globigerinella are common (5%-10%), sometimes reaching 13%-15% within the Brunhes Chron. Pulleniatina (obliquiloculata and primalis) and Orbulina (universa and bilobata) are generally few (<5%), except during the Jaramillo when both genera exhibited high abundance (9%-10%; Fig. 4). Orbulina and Pulleniatina tend to display similar patterns of abundance to each other. Sphaeroidinella dehiscens is present in trace amounts or absent in sediments younger than the Jaramillo, but is a common part of foraminiferal faunas below this chron, reaching a maximum abundance of 7% within the Olduvai Subchron.
Globigerinoides ruber is the most abundant species and fluctuates between 20% and 40% of the total foraminifer assemblage (Fig. 5). The other dominant species are Globigerinoides sacculifer and Neogloboquadrina dutertrei whose abundance are generally 5%-15%, but frequently exceed 20% (Fig. 5, Fig. 6). Other common members of the foraminiferal faunas are Globorotalia inflata and Globigerinita glutinata, which constitute 5%-10% of the total assemblage. Globorotalia inflata displays two high-abundance intervals (15%-20%) within Cores 164-997A-1H through 2H (CN15) and Core 164-997A-5H (between the lower Brunhes and top of the Jaramillo; Fig. 7), whereas larger specimens (>177 µm) of Globigerinita glutinata decrease to 5% in this interval (Fig. 6). Globorotalia truncatulinoides and G. crassaformis occur consistently (<5%-10%) above the Jaramillo (Cores 164-997A-1H through 7H), but are absent or rare in sediments below this chron (Fig. 7). Despite their generally low abundance, species in the Globorotalia menardii complex (menardii, tumida, ungulata, and exilis) are characterized by several significant abundance peaks, which suggest warm or interglacial periods (Fig. 8). Forms of G. hirsuta (G. hirsuta and G. cf. hirsuta) occurred abundantly in the restricted intervals of Cores 164-997A-2H (G. hirsuta) and 11H (G. cf. hirsuta; Fig. 7). The species Globigerina bulloides, G. falconensis, and Globigerinella aequilateralis exhibit a constant, but low abundance with small amplitude and short-term fluctuations (Fig. 5, Fig. 6).
Planktonic foraminifers are moderate to well preserved in many samples, whereas broken tests are common to abundant in several samples of two intervals of Cores 164-997A-2H through 4H, and below Core 164-997A-8H (Fig. 3). Decreases in the abundance of foraminifer shells was not always synchronous with that of fragments of tests or with decreasing abundance of dissolution-susceptible species such as the members of the genus Globigerinoides (Fig. 3, Fig. 4).
For each sample, species
richness, Shannon/Weaver diversity, and equitability were calculated (Fig.
9; Buzas and Gibson, 1969). Species richness is the number of species in
each sample, whereas the Shannon/Weaver information index describes diversity,
taking into account the relative population of each species within a sample (H =
-
Pi lnPi; where Pi is
the proportion of each species). Equitability is a measure of the evenness of
the species distribution within a sample (E = eH/S in which S =
number of species in a sample). Equitability equals one if all species are
present in the same population and approaches zero when one species dominates
the fauna.
The richness of planktonic foraminifers in Hole 997A is nearly constant (21-31 species) throughout the studied cores. Shannon/Weaver diversity exceeds 2.0 in many samples, except for lower values between 0.65 and 0.75 Ma (the lower Brunhes, the middle part of Core 164-997A-4H) and between 1.0 and 1.6 Ma (the Jaramillo to top of the Olduvai Subchrons, Cores 164-997A-6H through 8H). Equitability is nearly synchronous with Shannon/Weaver diversity such that intervals of low equitability (0.55-0.65) coincide with the low intervals of Shannon/Weaver diversity (Fig. 9). The general trends of these diversity indices suggest that Shannon/Weaver diversity is controlled more by equitability fluctuations than by species richness.
Ottens (1991) examined species diversity of planktonic foraminifers in four different surface-water masses in the eastern North Atlantic Basin. The boundary between these water masses corresponds approximately with the boundary of faunal assemblages in planktonic foraminifers. The highest Shannon/Weaver diversity (over 2.0) and highest equitability (close to 0.6-0.7) are recorded around frontal zones between major surface water masses, in agreement with work in the Indian Ocean (Ottens, 1991).
The high-diversity and high-equitability assemblages in Hole 997A, therefore, suggest proximity to the frontal zones between two different water masses, probably the subtropical gyre and Gulf Stream. Low values of both indices during the intervals of 0.65-0.75 Ma and 1.0-1.6 Ma suggest that Site 997 was intermittently under the edge of the subtropical gyre (Fig. 9).
Hole 997A is located at 31º51'N and 75º3'W and is within the subtropical bioprovince of Holocene planktonic foraminifers (Bé, 1977). Hence, it is not surprising that the most abundant to common taxa are members of the genus Globigerinoides (G. ruber and G. sacculifer) during the Pliocene-Pleistocene. We used Q-mode factor analysis with varimax rotation to identify foraminifer assemblages in our downhole abundance data (Table 2, Table 3). Principal component analysis of Hole 997A data identifies at least three factor assemblages in this subtropical fauna (Table 2). The calculated eigenvalues for the first three components explain 38.5%, 22.8%, and 16.3% of the total variance, respectively.
The first assemblage is dominated by mixed-layer species (G. ruber and G. glutinata), with rare G. inflata and N. dutertrei (Table 2), and represents a warm-water mass with a stable, deep mixed layer. Globigerinoides ruber dominates assemblages in the central Sargasso Sea and G. sacculifer (which negatively loads on the first principal component) is abundant in the southern Caribbean Sea and equatorial Atlantic in the surface sediments (Bé and Tolderlung, 1971; Kipp, 1976; Bé, 1977). Therefore, the first component describes assemblages from the central subtropical gyre (Sargasso Sea). Principal component one describes our "subtropical gyre assemblage."
The second principal component is dominated by N. dutertrei, and contains common G. hirsuta and O. universa, and few G. glutinata with rare specimens of G. sacculifer, G. crassaformis, G. truncatulinoides, G. inflata, and G. ruber (Table 2). Neogloboquadrina pachyderma (right-coiling) is also present in this group. According to Bé (1977), N. dutertrei is a tropical to subtropical species that occurs abundantly in active current systems along the continental margin and upwelling regions. Kipp (1976) also showed that N. dutertrei is very abundant along the western boundary of the subtropical gyre.
Globorotalia hirsuta
is a subtropical species, occurring abundantly in subsurface depths of the
central water masses during the winter or spring seasons. Although the generally
low abundance of this species makes its difficult to define its distribution
limits in the modern oceans, it is rare or absent poleward of the Northern and
Southern Transition Zones. The maximum abundance of G. hirsuta in the
North Atlantic is in the northwestern Sargasso Sea (Bé and Tolderlund, 1971;
Kipp, 1976). High concentrations in the western North Atlantic are observed in
February/March when the seasonal thermocline breaks down (Bé, 1960; Bé and
Tolderlund, 1971). The water here is well mixed, with salinity of almost 36.5
and temperature of 18.5º-20.1ºC. Below 400 m, a permanent thermocline exists
and water temperature decreases to 10ºC at about 800 m (Morris et al., 1977). Globorotalia
hirsuta is also found to be normally more abundant in deep tows (0-300 m)
than in surface hauls (0-10 m; Bé and Tolderlund, 1971). Globorotalia
hirsuta thrives in deep mixing conditions and is a deep-living planktonic
foraminifer in the gyres (Lohmann, 1992). We consider G. hirsuta to be
an index of seasonal breakdown of surface-water stratification. Hence, the
second principal component suggests thermocline waters with upwelling or deep
mixing as in the northern subtropical gyre. We call the fauna characterized by
the second principal component a "gyre margin assemblage."
The third principal component is dominated by G. inflata and G. truncatulinoides with lesser abundance of N. dutertrei, G. aequilateralis, G. ruber, and very rare G. sacculifer (Table 3). Globorotalia inflata is only abundant in the transition zone between the subpolar and subtropical provinces (Kipp, 1976). The highest frequencies are recorded between 35ºN and 45ºN in the North Atlantic (Bé, 1977). Fairbanks et al. (1980) found that living G. inflata are abundant within slope waters, and rare or absent within Gulf Stream cold core rings and the Sargasso Sea in November.
Globorotalia truncatulinoides is a mid-latitude species, and living populations show a distinct preference for winter conditions, predominantly between December and April (Bé, 1977). In the Sargasso Sea off Bermuda, the abundance peak of this species is in January before G. hirsuta and G. inflata become locally abundant in March and April (Bé, 1960; Bé and Tolderlund, 1971). Today, G. truncatulinoides is found in the western Mediterranean where there is seasonal deep mixing down to 600 m, but not in the eastern Mediterranean where deep mixing conditions do not develop (Lohmann, 1992). Moreover, the distribution of left-coiling and right-coiling varieties of this species are related to deep mixing of water masses because the abundance of right-coiling specimens increases towards the tropics, paralleling the decrease in vertical mixing (Bé and Tolderlund, 1971; Bé, 1977; Lohmann, 1992). Therefore we interpret the third principal component as a "slope-water assemblage," indicating cooler conditions or weaker seasonal deep-mixing conditions than the second factor assemblage.
Our statistically identified assemblages resemble some of the faunal groups recognized by Imbrie and Kipp (1971) and Kipp (1976). These authors analyzed the core top assemblage of planktonic foraminifers in the Atlantic Ocean using factor analysis, and classified them into five assemblages:
Based on comparisons of our data with the distributional and ecological studies of planktonic foraminiferal faunas in the modern oceans, the following five assemblages representing different hydrographic conditions are recognized at Site 997.
The tropical fauna is characterized by the occurrence of the G. menardii complex (menardii, tumida, ungulata, and exilis), suggesting the influx of warm-water species from the Gulf Stream that advanced out of the Gulf of Mexico and Caribbean regions. Other warm or tropical indices consist of Sphaeroidinella dehiscens, G. tenellus, and Globigerina rubescens, which are also distributed abundantly in the tropical Gulf of Mexico and Caribbean regions. The abundance of G. sacculifer is difficult to understand, but may also represent a warm-water index, because G. sacculifer is the most prolific tropical species, whose peak abundance occurs throughout the topics and warm subtropics between 20ºN and 20ºS in the modern ocean (Kipp, 1976; Bé, 1977).
The gyre margin assemblage lives along the frontal zone between the subtropical gyre and Gulf Stream and has the highest diversity and equitability of any assemblage recognized in this study. One group of this assemblage contains abundant temperate or cooler globorotaliids composed of G. hirsuta, G. crassaformis, G. truncatulinoides, and G. inflata. They are considered to be dwellers within very deep mixing conditions, suggesting no thermocline or a seasonal breakdown of surface-water stratification. In particular, the abundant occurrences of G. hirsuta or G. inflata suggest times when the surface-water stratification decreased and the gyre edge or slope water moved over the Blake-Bahama Outer Ridge, perhaps because of decreases in Gulf Stream flow. The other group associated with this assemblage consists of typical thermocline/upwelling species, including N. dutertrei, Pulleniatina (primalis and obliquiloculata), Orbulina (universa and bilobata), and Globigerinoides conglobatus.
The subtropical gyre assemblage consists of such abundant mixed-layer species as G. ruber, G. glutinata, and G. falconensis, suggesting oligotrophic, seasonal conditions with stable deep-mixed layers. The cool-or high-latitude indices in the subpolar assemblage are characterized by the occurrence of Neogloboquadrina pachyderma (dextral coiling) and Globigerina bulloides.
During the Brunhes Chron, the warm-water (Gulf Stream) assemblage was extremely rare (<2%), except during two interglacial times (presumably Stages 1 and 5). The assemblage reached higher average abundance (generally 3%) in the lower Brunhes, where distinct abundance maxima occurred during the interval between 0.6 and 0.78 Ma (Fig. 10). This group was very rare or absent in the Jaramillo Subchron, and fluctuated between 0% and 7% before 1.55 Ma.
The gyre margin/slope-water assemblage is one of the dominant groups of planktonic foraminifer faunas in Hole 997A. During the middle to late Pleistocene (0.2-0.3 Ma), the relative abundance of this assemblage increased during glacial/cooler times and decreased during interglacial/warmer times (Fig. 10). This pattern, however, is difficult to recognize in the lower Pleistocene and Pliocene. The relative abundance of this assemblage dropped to 10%-15% in the intervals between 0.5-0.7 Ma and 1.3-1.5 Ma, but increased to as much as 40%-45% in other intervals (Fig. 10). Within the gyre margin assemblages, the abundance of the slope-water globorotaliid group (G. inflata, G. hirsuta, and G. truncatulinoides) increased in the two intervals of 1.6-1.9 Ma and <1.0 Ma, whereas the other gyre margin assemblage dominated by N. dutertrei occurred consistently throughout the Pliocene-Pleistocene (Fig. 3, Fig. 11). The increase in the globorotaliid group between these intervals may suggest cooling in the frontal zone between the gyre and Gulf Stream because G. inflata characterizes a transitional assemblage between subpolar and subtropical waters in the modern ocean.
The subtropical gyre assemblage is the dominant group with high-amplitude fluctuations of 25%-60% (Fig. 10). During the late Pleistocene, the abundance peaks of this assemblage coincided with warm/interglacial times, representing a trend opposite to gyre margin species. The correspondence between the abundance peaks and warm-cool cycles, however, is not clear in the Pliocene and early Pleistocene. The subtropical gyre assemblage became relatively rare (<20%) in the interval between 0.78 and 1.0 Ma (from the Brunhes/Matuyama boundary to the top of the Jaramillo Subchron). This assemblage also exhibited abundance fluctuations before 1.5 Ma similar to the abundance histories of the gyre margin assemblage in this interval (Fig. 10).
The subpolar assemblage is a minor element (generally <5%) and displays high-frequency fluctuations with an amplitude of 1%-5% throughout the studied section (Fig. 10).
Temporal trends in these assemblages are summarized as follows (Fig. 12):
