Marine Micropaleontology 60 (2006) 269 – 294 www.elsevier.com/locate/marmicro
Comparison of radiolarian and sedimentologic paleoproductivity proxies in the latest Miocene–Recent Benguela Upwelling System D. Lazarus a,⁎, B. Bittniok a , L. Diester-Haass b , P. Meyers c , K. Billups d a
c
Museum für Naturkunde, Humboldt University Berlin, Invalidenstrasse 43, 10115 Berlin, Germany b Zentrum für Umweltforschung, Universität des Saarlandes, 66041 Saarbruecken, Germany Department of Geological Sciences, University of Michigan, 425 East University Ave., Ann Arbor, Michigan 48109, USA d College of Marine Studies, University of Delaware, 700 Pilottown Road, Lewes, DE 19958, USA Received 23 January 2006; received in revised form 2 June 2006; accepted 6 June 2006
Abstract Estimating past ocean productivity from ocean sediments often gives different results depending on the measurement used. We have examined a suite of paleoproductivity proxies in latest Miocene–Recent sediments from DSDP Site 532 and ODP Site 1084, two deep-sea sections underlying the Benguela Upwelling System off the Atlantic coast of southern Africa. The productivity history of this system has been previously established via organic carbon concentration, diatom floras and alkenone based estimates of surface water temperature, and shows a change from low productivity in the early Pliocene to sustain high productivity in the late Pliocene–Recent. Each of our samples was split and simultaneously analysed for several proxies of ocean productivity, including organic carbon (TOC%), carbonate, abundance of opaline radiolarians, accumulation rate of benthic foraminifera (BFAR); the radiolarian faunal composition indices Upwelling Radiolarian Index (URI) and the Water Depth Ecology index (WADE); other proxies for opal and carbonate dissolution, plus stable isotopes of benthic foraminifera. Comparisons between proxies in the same measured samples, between sites in downcore plots and to the published productivity record for this region suggest that TOC and radiolarian faunal composition, particularly the WADE index, are good indicators of past productivity, albeit with different sensitivities (log–linear correlation WADE–TOC% r = 0.78, n = 65, p < 0.01). In contrast, carbonate, and carbonatebased proxies such as BFAR primarily reflect changes in dissolution. Radiolarian faunal composition indices do not appear to be affected by bulk opal accumulation or changes in opal preservation. WADE analysis of radiolarian faunas and TOC% measurements appear to be useful proxies for productivity in late Neogene sediments, particularly for sections where opal or carbonate dissolution is significant. © 2006 Elsevier B.V. All rights reserved. Keywords: Micropaleontology; Nutrient cycling; Paleoproductivity; South Atlantic; Neogene; Organic carbon
1. Introduction Oceanic productivity plays an important role in the global carbon cycle, particularly by export of biologically fixed carbon to deep waters and sequestration of ⁎ Corresponding author. E-mail address:
[email protected] (D. Lazarus). 0377-8398/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.marmicro.2006.06.003
carbon in ocean sediments. Understanding this complex process is best achieved using a variety of approaches, including study of past changes in ocean productivity as recorded by paleoproductivity proxies in marine sediments. Many different paleoproductivity proxies exist, but, for various reasons, no single proxy gives consistently reliable results. This issue has led to the use of multiple proxies, in the hope that a reasonably consistent
270
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
consensus picture of past productivity change will emerge (Wefer et al., 1999). Recent research on the paleoproductivity history of the Benguela Upwelling System exemplifies this multi-proxy approach. The Benguela Upwelling System is a region of nearshore upwelling cells and offshore transport of highproductivity water filaments in the Benguela eastern boun-
dary current region of the South Atlantic off the southwestern coast of Africa (Hay and Brock, 1992; Fig. 1). Dominantly offshore wind stress near the coast draws nutrient-rich waters from ca. 100–200 water depth to the surface in coastal upwelling cells. High algal production in this water is exported offshore in long, narrow filaments, is carried north by the Benguela Current and related current
Fig. 1. Map showing major currents, extent of upwelling near coast and extent of upwelling influenced water further offshore from transport of filaments, and DSDP–ODP Site locations. Sites in bold/yellow have substantial downcore radiolarian data (prior work plus this study). Redrawn from Wefer et al. (1998). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
systems, and is eventually mixed into the more oligotrophic waters of the oceanic subtropical South Atlantic. This complex system of cool, upwelling water and high productivity has a significant effect on regional climate and fisheries, and, via the export of biologically fixed carbon to deep waters and sediments, plays a significant role in the global carbon budget. Sediments underlying this region vary considerably in composition, but typically are enriched in organic carbon, and have significant amounts of biogenic opal, generally considered an indicator of elevated marine productivity. This sedimentary record of upwelling productivity has been extensively studied, most notably by three deep-sea drilling legs: DSDP Legs 40 and 75, and more recently ODP Leg 175. This work (Berger et al., 2002; Fig. 2) has shown that the Benguela Upwelling System most probably first developed in the late Miocene, when organic carbon and opal enriched sediments are first recorded in the region (Siesser, 1980). The detailed history of productivity in the system has however been difficult to reconstruct, due to geographic differences and in particular to inconsistent behavior of different proxies for marine productivity in
271
these sediments over time, including TOC%, opal, benthic foram accumulation rates (BFAR) and other bulk sediment parameters. In this paper we wish to explore the use of an additional paleoproductivity proxy, radiolarian faunal composition, that may help clarify past productivity change in the Benguela Upwelling System, and other similar late Neogene upwelling environments. Radiolarians are protist zooplankton that live in both surface and deeper water layers in most of the world's oceans. They occur commonly in the sediment record, and are more resistant to dissolution than are diatoms. In recent years radiolarian faunal composition has been exploited in several studies of Pleistocene sediments to estimate productivity. Two methods have been proposed — URI and WADE (see also review in Lazarus, 2005). The URI (Upwelling Radiolarian Index) method is based on the observation that in lowlatitude upwelling regions, some radiolarian species are found that are either endemic to these regions, or are more common there than in similar low-latitude non-upwelling environments (Nigrini and Caulet, 1992). WADE (Water Depth Ecology) methods are based on the original insights
Fig. 2. Summary of Benguela Upwelling System history over the last 5 million years from selected sediment proxies (Carbonate, TOC%, bulk opal, diatom floral composition, surface water temperature from Talkenone). Talkenone and upwelling diatom relative abundance for Site 1084 are from Marlow et al. (2000). Other data is from Site 1084 (blue, Meyers in Wefer et al., 1998) and Site 532 (orange, Meyers, 2002, black, Hay and Sibuet, 1984). Opal estimates from Dean and Parduhn (1984) shown in green, estimates from reprocessed smear-slide data in red. For details of opal data sources and synthesis methods see Bulk opal measurements in text. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
272
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
of Casey (Casey et al., 1982) who proposed that an increase in the ratio of deep-dwelling to surface species should reflect the relative increase of organic carbon export from surface waters due to elevated productivity. Specific questions addressed in this study are: • Do WADE and/or URI methods produce coherent, consistent signals in this long time-series sample set? • Do these indices correlate well with other paleoproductivity proxies, and in particular, with the TOC record from late Neogene Benguela sediments? • Are radiolarian faunal methods significantly affected by changes in bulk opal accumulation and/or changes in the quality of opal preservation? 1.1. Previous studies of late Neogene tropical–subtropical Atlantic radiolarians 1.1.1. Living and recent There are only a few studies of living radiolarians from the tropical Atlantic water column. The most detailed information was collected by Boltovskoy et al. (1996) in a single set of 0–300 m depth-stratified tows from ca. 0–17°S off the coast of Africa in February– March, 1988. The full data set was kindly provided to the authors by Dr. Boltovskoy. Being a single ‘time-point’ sample of the water column the generality of patterns in the data could not be determined from this study alone, but these data helped to confirm depth ecology interpretations from other sources, primarily Pacific studies (see Materials and methods below). Biogeographic distributions of radiolarians in surface sediments are also available, particularly from Lombari and Boden (1985) and Boltovskoy (1998). These data were used together with other sources to determine ecologic preferences for some taxa (see Materials and methods). 1.1.2. Fossil Stratigraphic studies of late Neogene radiolarians from tropical–subtropical Atlantic sediments are rare, as, outside of upwelling regions, preservation of biogenic opal in late Neogene sediments from this region is generally poor. Other than the study of Weinheimer (2001), a detailed stratigraphic study of latest Miocene–Recent radiolarians from ODP Leg 175 Site 1082 has been published by Motoyama (2001). This report provided useful information on taxonomy and morphologic variation in taxa over time, although, due to its stratigraphic focus, only a limited number of ecologic indicator taxa are included in his report. Motoyama also carried out a very preliminary analysis of downcore variation in upwelling/productivity
Table 1 List of parameters measured Measurement Lab1 Group2 name
Unit
1
BFAR
S
Foram
2
BF/allF
S
Foram
3
PFtotalCF
S
Foram
4
Uvig/BF
S
Foram
5
fragF
S
Foram
6
Age
B
General
7
Depth mbsf
P
General
8 9
DBD LSR
P P
General General
10 B18O
D
Geochem
11 CaCO3
M
Geochem
12 TOC
M
Geochem
13 CaCO3 AR
M
Geochem
14 TOC AR
M
Geochem
15 Rads/gSed
B
Opal
16 R preservation
B
Opal
17 R AR
B
Opal
18 R in sand
S
Opal
19 Warm
B
Rads
20 Temperate
B
Rads
N/cm2 kyr Benthic foraminifera accumulation rate % Benthic foraminifera in all foraminifera % Planktonic foraminifera in sediment coarse fraction % Uvigerina in benthic foraminifera % Fragmented foraminifera in all foram shells my Sample ages (million years) m Depth, meters below sea floor g/cm3 Dry bulk density cm/kyr Linear sedimentation rate ‰ VPDB Oxygen isotope composition % Calcium carbonate in bulk sediment % Total organic carbon mg/cm2 yr Calcium carbonate accumulation rate mg/cm2 yr Total organic carbon accumulation rate Ng Radiolarian shell abundance in bulk sediment 1 (very Radiolarian poor)–7 preservation (semi(excellent) quantitative visual estimates of assemblages) N/cm2 yr Radiolarian shells accumulation rate % Radiolarian shell abundance in coarse fraction of sediment % Warm surface water radiolarians ecologic group abundance in total rad assemblage % Temperate radiolarians ecologic group abundance
Comments
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294 Table 1 (continued ) Measurement Lab1 Group2 name 21 Intermediate
B
Rads
22 Upwelling
B
Rads
23 WADE
B
Rads
Unit
Comments
%
Intermediate water depth radiolarians ecologic group abundance % Upwelling radiolarians ecologic group abundance Ratio W/I Water Depth Ecology ratio (intermediate radiolarians/warm radiolarians)
1
Lab (author) responsible: B—Berlin, D—Delaware, M—Michigan, S—Saarbrücken, P—published. 2 Group — type of measurement/category for data presentation discussion.
indicator radiolarian taxa, using core-catcher samples from Site 1075 off the Congo River (Wefer et al., 1998). 1.2. Previous studies of other parameters (carbonate, organic carbon, opal, benthic foraminifera and stable isotopes) There are numerous studies of these sediment components in Benguela piston and drill cores, but it is not our purpose here to review or synthesize this literature. Detailed reviews are given by Dean et al. (1984) and Berger et al. (2002). Here, we primarily wish to explore the use of radiolarian faunas as a proxy for productivity, and thus we limit our reference to previously published non-radiolarian data to that needed to compare our radiolarian results to them. Data for these parameters, to the extent available from the same sites used in our study, have however been plotted against our own data for comparison purposes. 2. Materials and methods 2.1. Strategy To investigate our study's questions we have used the following strategy. We assume that the current consensus view of the Benguela Upwelling System is essentially correct — that it has displayed continuous (on longer time scales) high productivity from the mid Pliocene to the Recent, with lower productivity in the early Pliocene and latest Miocene. We also assume that TOC% is, at least for this system, a reasonably reliable marker for this productivity. TOC% can be influenced by many other
273
factors than productivity, including diagenesis within the sediment column, lateral transport of organic carbon in bottom waters — particularly near continental margins, and other factors (Wefer et al., 1999). Diester-Haass et al. (1992) have suggested that TOC% in Benguela sediments could reflect increased down-slope transport, at least during glacials. Berger et al. (2002) however argue, on the basis of increased BFAR values during glacials, that TOC % in these sections primarily reflects surface water export productivity. This latter conclusion is strengthened by the close correspondence between long-term trends in TOC% and alkenone derived surface water temperatures reported by Marlow et al. (2000). Our goal has been to obtain a similar time-series of radiolarian faunal indices to compare with this long-term temporal record of productivity change. The similarity of temporal patterns between these parameters is taken as an indication that radiolarian faunal changes can be used as a productivity proxy in these sediments, as is TOC% (at least over longer time intervals). It is also possible that any observed correlation in temporal trends is coincidental and not due to similar response to a common cause (productivity). Only additional studies from other sections, in other regions, can fully examine this potential problem. We have measured TOC% and several other productivity proxies – BFAR, opal content etc – in the same set of samples used for radiolarian faunal analysis. Paired measurements from the same set of samples permit us to analyse the different productivity proxies for patterns of covariation, at least in part independently of longer-term temporal trends, although with the low resolution of our sample set we cannot explicitly identify and separately compare true short-term (glacial–interglacial) variation from longerterm trends in the system. 2.2. Sampling and measurements Samples have been taken in two time-series from two different sites in the Benguela system, at a resolution of ca. 150 kyr. Samples (ca. 20 cc) were taken from ca 3–4 cm of core section, and thus represent < 1000 years of sedimentation (plus bioturbation effects). Each sample was homogenised and split into two subsamples that were analysed separately for several different proxies. Radiolarian slides of the >45 μm acidized fraction were made from one subsample using standard techniques (Moore, 1973; Lazarus, 1994a). Counts of all radiolarians and selected taxa (see below) were carried out on at least 300 specimens/sample, except for a few samples where radiolarian abundances were very low. Reproducibility (standard deviation) of counts for individual taxa between observers (analysis not shown) is approximately 5%
274
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
absolute. Absolute radiolarian abundance as radiolarians/ g dry sediment was also computed. Radiolarian preservation (7 point scale) was estimated semi-quantitatively in these samples. These two parameters serve in our study as estimates for bulk opal accumulation and dissolution. The other subsample was further split. One fraction was used for TOC% and carbonate measurements using previously published methods (Meyers, 2002). The other part was analysed for coarse fraction abundances using methods published by Diester-Haass et al. (2002). Several components were determined in the coarse fraction, including percent radiolarians (also used as an opal indicator), and several components of the foraminiferal fraction: benthic foraminifera accumulation rate (BFAR), benthic–planktic ratios; fragmentation index and relative abundance of Uvigerina. Benthic foraminiferal oxygen isotope measurements were made on 1–5 individuals of Cibicidoides or Uvigerina selected from the >250 mg size fraction. All samples were sonicated in deionized water to remove adhering sediments and oven-dried for at least 24 h prior to measurement. Measurements were performed at the University of Delaware using a GVI IsoPrime equipped with a multiprep peripheral for the automated reaction of individual samples with phosphoric acid. Using NBS-19 and an in-house standard (Carrara Marble) our precision for δ18O values, reported vs. VPDB, is better than 0.08‰. In a few samples isotope measurements could not be made due to rarity of suitable specimens. A full list of the measurements collected is given in Table 1. 2.3. WADE and URI indices Although the basic concept for both URI and WADE analysis of radiolarian faunas is now well established, there is no standardised list of taxa that are used in such studies. The URI taxa listed by Nigrini and Caulet (1992) are a reference point for URI analysis, but rarity of individual species, biogeographic differences between regions and a Pleistocene origin for some forms all affect the choice of species used. To date, no previous studies of tropical Atlantic sediments using URI methods have been published, nor are such taxa given in the radiolarian faunal lists of tropical Atlantic plankton so far published. In our study only a few taxa could be identified that are included in Nigrini and Caulet's list, i.e. Plectacantha cremastoplegma, Acrosphaera murrayana, Pterocanium auritum, Dictyophimus infabricatus, Lamprocyrtis nigriniae and Pterocorys minythorax; most of these species were very rare, with only few specimens being seen in the entire sample set. We refer to the percent abundance of these species as the URI or ‘upwelling’ component in our study. The absence of other forms, despite the obvious upwelling
environment, could be due to any of the above factors (particularly age and rarity), as well as preservation problems and difficulties in distinguishing some species from morphologically very similar forms. Taxa lists for use in WADE analysis are less well established. The original suggestions of Casey et al. (1982) did not specify in detail which species would be most appropriate, and Jacot des Combes' studies (e.g. Jacot Des Combes et al., 1999) used only a very small number of species to form her Thermocline to Surface Radiolarian Index (TSRI). Some of these taxa were not present, or extremely rare, in our materials (including two of her three warm surface water indicators : Siphonosphaera polysiphonia and Lithopera bacca). Weinheimer's (2001) study of Site 1082 sediments is based on a much larger number of species whose depth preferences were in most cases previously established from earlier studies of plankton and/or sediment distributions, largely from the North Pacific (Kling, 1977, 1979), and southwest Atlantic (Boltovskoy and Riedel, 1980; Abelmann and Gowing, 1996). For our study we have resurveyed the published literature on radiolarian water depth ecology, and have extended Weinheimer's (2001) taxa list with additional surface and intermediate water taxa. Biogeographic distributional information was also obtained from searches of the Neptune database (Lazarus, 1994b) to supplement published biogeographic syntheses. We maintain Weinheimer's category of (cool) temperate water taxa for those species with a clear extra-tropical modern biogeographic distribution, but whose water depth preferences in lower latitudes have not been clearly established. It should be noted that our ecologic assignments are still preliminary as different sources of information are sometimes in conflict. This problem may in part be due to the current incompleteness of knowledge for modern radiolarian biology; different depth distribution of water masses at different locations; taxonomic uncertainties; and often highly incomplete recording of species occurrences in the literature (this latter clearly affects the biogeographic patterns reported by the Neptune database, where high latitude faunas are more completely reported than low-latitude ones, and in general non-stratigraphic indicator taxa are under-represented). All taxa used are listed, together with the ecologic and taxonomic source information, in Table 2. The WADE index itself is merely the ratio between the sum of warm water taxa (percent of assemblage) divided by the percent sum of intermediate (i.e. deeper water) taxa. 2.4. Analysis Data were first plotted downcore to determine primary patterns of variation. Accumulation rates were also
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
275
Table 2 List of taxa and sources of ecologic and taxonomic information for each Taxon
This study
Weinheimer L175
Nigrini and Caulet
Addnl ecologic info and sources
Acrosphaera murrayana (Haeckel) Benson (1983) Actinomma antarctica (Haeckel) Nigrini (1967)
U T
Antarctic–subantarctic distribution in 2. Common and reported only in Antarctic–subantarctic in 19
1
Actinomma mediana Nigrini (1967)
T
1
Actinomma leptoderma (Jørgensen) Bjørklund (1976)/haysi Bjørklund (1977) Actinomma popofskii (Petrushevskaya) Morley and Nigrini (1995) /Haliometta miocenica Campbell and Clark (1944) Actinomma boreale Cleve (1899)
T
Cool temperate-transitional distribution in 2 and 19; surface species in 3 N. Atlantic in 19. Temperate-transitional in 2
T
19-only in Antarctic sites; A.pop. grp has the same stratigraphic abundance distribution as L. minor (a Wein. T) in Motoyama L175 data
5;6
T
4;7
Heliodiscus asteriscus Haeckel 1887
W
Didymocyrtis tetrathalamus (Haeckel) Sanfilippo and Riedel (1980)
W
W
Amphirhopalum ypsilon Haeckel (1887)
W
W
Dictyocorne spp.
W
W
Spongocore puella Haeckel (1887)
W
W
Only reported by Norwegian– Greenland Sea workers in Quaternary, in Nrn N. Atlantic DSDP in 19. syn? Crom. antarctica in 8 as AA/SAA Considered a warm surface water species in TSRI in 9. Tropical– subtropical in 2. Temperate-polar in 19. Common in warm (0–50 m) surface waters in 3. Tropical– subtropical in 2. Widely reported in tropical–temperate in 19. Found in warm (0–50 m) surface waters in 3. Widely reported, primarily tropical–subtropical in 19. Rare, possibly tropical– subtropical in 2 Common group in surface-100 m in 3. Tropical–subtropical in 2. Cosmopolitan in 19. Subtropical–cool temperate in 19; subtropical in 2. 0–100 m+ in 3.
Spongopyle osculosa Dreyer (1889) grp
I
I
Subantarctic/gyre margin; >200 m in 3; thermocline species in TSRI in 9
8
Spongotrochus? venustum (Bailey) Nigrini and Moore (1979) Octopyle stenozona Haeckel (1887) /Tetrapyle octacantha Müller (1858)
T
T
Temperate in 19. 0–300 m in 3.
1;7
W
W
Tropical distribution in Pacific in 2. Mostly 0–50 m in 3 (T. octacantha only).
1;7
U
Taxonomic Comments reference 1;7 Correct species ending for this and other Actinomma taxa is -a, not -um; as indicated in 18
1
Broad category similar to usage in 4
7
1;7
1;7
1;7
Also includes Euchitonia spp.
8;1;7
Synonymous with S. cylindrica of some authors Probably includes other spongodiscid species as preservation often incomplete
Broad category for several pylonid taxa. See also comments in 12 (continued on next page)
276
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
Table 2 (continued ) Taxon
This study
Weinheimer L175
Larcopyle buetschlii Dreyer (1889)
T
T
Lithelius minor Jørgensen (1889) Larcopyle hayesi irregularis (Chen) Lazarus et al. (2005) Arachnocorys circumtexta Haeckel (1860) Plectacantha cremastoplegma Nigrini (1968) Helotholus histricosa Jørgensen (1905) Lithomelissa setosa Jørgensen (1900)
T
T
Lithomelissa hystrix Jørgensen (1900) Antarctissa spp.
Nigrini and Caulet
T
W
W
U
Addnl ecologic info and sources
Taxonomic Comments reference
Given as thermocline taxa (in tropical Indian Ocean) in 9. > 100 m? in 3. Frequent in N. Pacific and Antarctic, rare in tropics in 10 Cool temperate distribution in 2. 0–100 m but variable in 3. 10: common cool water form but not endemic to Antarctic Miocene as is L. h. hayesi >100 m but very rare in 3
1;7;10
U
T
T
I
I
T T
1;7 10
8 11
Polar distribution in 19. 0–200 m in 3. Common in Norwegian–Greenland Sea sediments. Only report in 19 is from N. Atlantic Reported from Nordic Seas in 7 Primarily Antarctic form in 19; but also seen in cool gyre margin settings Polar distribution in 19. 50–200 m but variable in 3
8
As Ceratocyrtis histricosa in 7
7
7 7
Pseudodictyophimus gracilipes (Bailey) Petrushevskaya (1971)
I
I
Botryocyrtis scutum (Harting) Nigrini (1967) Carpocaniid spp.
W
W
Tropical in 2
1;7
W
W
Tropical–temperate in 19. Subtropical–temperate in 2. 0–100 m in 3 (as Carpocanium spp.) Maximum abundance in subantarctic according to 8. Reported only in Antarctic sediments in 19. Cosmopolitan biogeography. >200 m in 3. Cosmopolitan biogeography
1
As Carpocanistrum spp. in 1
8
Also includes closely related forms
Cool temperate–subpolar in 2. Polar distribution in 19, widely reported. >200 m in 3 (as Theocalyptra. b.) Polar distribution in 19, widely reported.>200 m in 3 (as Theocalyptra davisiana cornutoides) Transitional fronts/gyre margin modern distribution in 2. 100–200 m in 3 (as Theocalyptra d. davisiana)
13
Saccospyris antarctica Haecker (1907)
T
Cornutella profunda Ehrenberg (1854) Peripyramis circumtexta Haeckel (1887) Cycladophora bicornis (Popofsky) Lombari and Lazarus (1988)
I
I
I
I
I
I
Cycladophora conica/ cosma both Lombari and Lazarus (1988)
I
T
Cycladophora davisiana Ehrenberg (1862)
I
I
Dictyophimus crisiae Ehrenberg (1854)
T
T
Primarily polar distribution in 19 but also reported off coast of N. Africa
7
Equals Dictyophimus gracilipes of Weinheimer etc
7 1;7
13
13
1;7
Does not include former subspecies C. d. cornutoides. First appears in Atlantic ca. 2.6 Ma
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
277
Table 2 (continued ) Taxon
This study
Weinheimer L175
Nigrini and Caulet
Addnl ecologic info and sources
Taxonomic Comments reference
Dictyophimus infabricatus Nigrini (1968) Pterocanium trilobum (Haeckel) Popofsky (1913)/Pterocanium praetextum eucolpum Haeckel (1887)
U
T
U
100–300 m in 3
11
W
T
20: cosmopolitan but most common in subtropical regions. Mostly polar distribution in 19 but tropical/gyre margin in 2. 0–100 m in 3 but variable
1
Pterocanium praetextum praetextum (Ehrenberg) Haeckel (1887) Pterocanium auritum Nigrini and Caulet (1992) Dictyophimus hirundo (Haeckel) Nigrini and Moore (1979) grp Lipmanella dictyoceras (Haeckel) Kling (1973) Eucyrtidium acuminatum (Ehrenberg) Nigrini (1967)
W
W
Tropical distribution in 20. 0–50 m in 3 (as P. praetextum)
1
Eucyrtidium calvertense Martin (1904)
I
Eucyrtidium teuscherei Caulet (1986) gr.
T
Lithostrobus hexagonalis Haeckel (1887) Cyrtopera laguncula Haeckel (1887)
W
Stichocorys peregrina (Riedel) Riedel and Sanfilippo (1970) Theocorys veneris Haeckel (1887) Anthocyrtidium ophirense (Ehrenberg) Nigrini (1967) Anthocyrtidium zanguebaricum (Ehrenberg) Haeckel (1887) Lamprocyclas maritalis Haeckel (1887)
W
Lamprocyrtis heteroporos (Hays) Kling (1973) /neoheteroporos Kling (1973) Lamprocyrtis nigriniae (Caulet) Kling (1977)
I
U
U
I
W
W
W
W
I
1;7
7 1
15
7
7
Given as a thermocline species in 9. 7 Temperate–polar distribution in 19. >100 m? in 3 (as Cyrtolagena l.) Fairly cosmopolitan in 19, 7 could also be classified as T.
W W
W
W
W
I
I
I
11 Given as deep water by 14 in Sea of Okhotsk. Polar distribution in 19. >100 m? in 3 (as D. sp. 2) Only records in 19 are in N. Pacific and Benguela. 0–50 m in 3 Tropical–subtropical in 2. Widespread in mid-latitudes in 19. 0–50 m in 3 TSRI taxon in 9, depth given in 9 as >200 m in tropics; 19-temperate/cold distribution May be more cosmopolitan than true transitional form. Occurs both Antarctic and tropics in 19 50–200 but variable in 3
W
I
U
Grouped due to similarity of local P. trilobum morphs to P. p. eucolpum. P. trilobum is polymorphic and may be more than one species
Surface waters (0–100 m) in 3 and 16 Tropical–subtropical in 2. 0–50 m in 3 Tropical–subtropical in 2. 50–100 m in 3 At or above thermocline depth (50–100 m) in 3. Subtropical–temperate in 2. Cosmopolitan distribution in 19 Both L. heteroporos and L. neoheteroporos assumed to be ecologically equivalent to L. nigriniae Given as upwelling species in 11 but note widespread occurrence also outside upwelling regions. Upwelling/gyre margin in 2. High productivity, including eq. Pacific in 19. 100–500 m in 3
Extinct in late Pliocene
1;7 1;7 1;7
17
1;7
(continued on next page)
278
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
Table 2 (continued ) Taxon
This study
Pterocorys minithorax (Nigrini) U Nigrini and Moore (1979) Pterocorys zancleus (Müller) W Nigrini and Moore (1979) Theocorythium trachelium (Ehrenberg) Nigrini (1967) Botryrostrobus aquilonaris (Bailey) Nigrini (1977) Botyrostrobus auritus/australis (Ehrenberg) Nigrini (1977) Phormostichoartus corbula (Harting) Nigrini (1977) Siphocampe arachnea (Ehrenberg) Nigrini (1977)/S. lineata (Ehrenberg) Haeckel (1887) Spirocyrtis spp.
Lophospyris pentagona (Ehrenberg) Goll (1977) Phormospyris stabilis (Goll) Goll (1976)
Zygocircus spp.
Weinheimer L175
Nigrini and Caulet
Addnl ecologic info and sources
Taxonomic Comments reference
W
U
0–100 m in 3
1;7
Warm and temperate waters in 19; temperate/transitional in 2. 0–50 m in 3 (as P. sp. 2) Warm and temperate waters in 19. Temperate in 2. 0–50 m in 3 Subpolar in 2. Warm to subpolar waters in 19. >200 m in 3 Tropical in 2. 50–100 m in 3 but broad depth range Widely reported in low to mid latitudes in 19. Rare, possibly tropical in 2. Deep-dwelling species (max ca. 1000 m) in 3. Cosmopolitan but mostly polar regions in 19
1
W
W
W
I
I
W
W
W
W
I
W
W
1;7 1;7
1;7 1;7
7
But not S. lineata sensu 7
Uniformly R throughout tropics and subtropics, a bsent in polar regions in 19 W
W
W
References: 1-Nigrini and Moore (1979) 2-Lombari and Boden (1985) 3-Kling and Boltovskoy (1995) 4-Motoyama (2001) 5-Chen (1975) 6-Morley and Nigrini (1995) 7-www.radiolaria.org 8-Boltovskoy (1998) 9-Jacot Des Combes et al. (1999) 10-Lazarus et al. (2005) 11-Nigrini and Caulet (1992) 12-Weinheimer (2001)
1;7 Cosmopolitan in 19. 50–100 m in 3 but primarily for different subspecies than seen in this study Maximum abundance 0–50 or 0–100 m in 3 and 8. Only in high latitudes in 19, but other literature reports in tropics
1;7;8
Includes several subspecies
7;8
13-Lombari and Lazarus (1988) 14-Nimmergut and Abelmann (2002) 15-Hays (1965) 16-Kling (1979) 17-Nigrini and Sanfilippo (2001) 18-Cortese and Bjorklund (1998) 19-Neptune database 20-Lazarus et al. (1985)
W—warm, T—temperate, I—intermediate (deeper water), U—upwelling.
computed for appropriate proxies using the age models established for each site (see below). Plots show raw data values and smoothed curves for each site. Smoothing was done using a 3-point moving average. All results, having been taken from identical (‘paired’) samples, were then compared to each other by simple visual cross-correlation
(scatterplot matrix) and pairwise correlation coefficient analysis (R for Mac OS X), with significant correlations being investigated further as appropriate. Multivariate analyses (e.g. factor analysis) that explore more general relationships between large sets of variables were not done, given the specific focus of our research question. In
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
addition we compared our own results for the long-term behavior of the Benguela system to similar long-term time-series published by other authors (e.g., Marlow et al., 2000; Berger et al., 2002). 2.5. Sites 2.5.1. 532 DSDP Site 532 (19° 45′ S, 10° 31′ E) was drilled near the older, incompletely recovered section of DSDP Site 362 at a water depth of 1331 m on Walvis Ridge (Hay and Sibuet, 1984), ca 270 km off the coast of Africa (Fig. 1). This location is today well offshore from the coastal upwelling cells but is significantly affected by high productivity water entrained in advected filaments from coastal locations (Berger et al., 2002). Sediments at Site 532 are bioturbated clayey calcareous oozes and marls with varying amounts of biogenic silica. Organic carbon and pyrite are minor but significant components of the sediment. Site 532 was drilled using an early version of hydraulic piston core
279
technology (4.5 m lengths) with three holes (532, 532A, 532B). The cores from 532A and cores 1–56 from 532B from this site were reserved for special DSDP studies and not generally sampled (indeed their current location is uncertain-ODP curator, pers. comm. 2001); our samples were taken from holes 532 and 532B (below 250 mbsf). The age model for Site 532 has long been problematic. Planktonic foraminifera are generally either too poorly preserved or contain too few low-latitude stratigraphic indicators for detailed biostratigraphy (Boersma, 1984). No detailed diatom or radiolarian biostratigraphic studies have been published for this site. Calcareous nannofossils are by contrast relatively common and better preserved and are the primary basis for biostratigraphy (Steinmetz et al., 1984). Sediments from this site, having been recovered with a hydraulic piston corer, are suitable for paleomagnetic studies, but the published paleomagnetic interpretation for the site (individual sample measurements using a cyrogenic magnetometer by Keating, in Hay and Sibuet, 1984) has been at variance with the biostratigraphic data. Specifically,
Fig. 3. Site 532 age model. Age scale on bottom after Berggren et al. (1995). Interpretation (this paper) of paleomagnetic data from Keating in Hay and Sibuet (1984) on left, core recovery on right. Details of calcareous nannofossil events (N) and magnetostratigraphic age interpretations (M) given in Table 3. Graphic made with ADP v 2.08 (Lazarus, 1992).
280
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
the published paleomagnetic interpretation of Keating identifies only a small number of polarity intervals and suggests a significantly younger age than that by biostratigraphy for much of the deeper part of the recovered section. Subsequent paleoceanographic studies of this site have mostly used the alternate age model of Gardner et al. (1984), although this model makes use only of a few control points, primarily chosen to make the resultant age model fit the expected gradually varying sedimentation rate. In this paper we present a new age model for Site 532 (Fig. 3; Table 3), that is based on recalibrated local ages from Leg 175 for nannofossil events that were earlier determined for 532 (Berger et al., 2002), and a reinterpretation of the paleomagnetic data published by Keating (Hay and Sibuet, 1984). Our age model differs from those published previously primarily in the interpretation of the paleomagnetic data. Unlike Keating, who extrapolated uniform polarity between rather widely spaced paleomagnetic measurements, we have treated intervals longer than ca. 5 m without measurements as ‘no data’. This change in interpretation allows us to much better match observed polarity reversal levels to biostratigraphic ages, in that some polarity Table 3 List of bio-and magnetostratigraphic datums used in 532 age model (Fig. 3) Group Name/Comments
Plotcode Event age
N1 N1
bEHux TCmac
0.27 1.63
2.1 21.6
3.65 26
TDbrw
2.07
74.4
75.65
TSspp
3.7
149.1
TDqui
5.5
246.11 246.8
THsel TDsur
1.23 2.7
39.2 95.8
43.6 96.4
TRpsu
3.84
92
94.31
TAtri BBru TGau BGau TgilA BgilC2
4.09 0.78 2.58 3.58 4.29 5.23
N1 N1 N1 N2 N2 N2 N2 M M M M M
FO Emiliana huxleyi LO Calcidiscus macintyrei LO Discoaster brouweri LO Sphenolithus spp. (=albes in 532) LO Discoaster quinqueramis LO Helicosphaera selli LO D. surculus (L consist. Occ.) LO R. pseudoumbilicus (L Consist. Occ.) LO Am. tricorniculatus B Brunhes T 2AN1 (Gauss) B 2AN3N (Gauss) B 3N1 (A, Gilbert) B 3N4 (C2, Gilbert)
Top depth
Bot depth
reversals are inferred to be present in intervals of no data, thus there are more polarity zones, and more time inferred in the section than in the original interpretation. Our age model thus reconciles the published biostratigraphic and paleomagnetic data for this site. The sedimentation rate history inferred for this site (see below) is indeed quite similar to that proposed by Gardner et al. (1984). 2.5.2. 1084 ODP Site 1084 (25° 31′ S, 13° 2′ E) was drilled in the northern Cape Basin near the Lüderitz coastal upwelling cell at a water depth of 1992 m on the continental slope off Africa, approximately 690 km S–SE of Site 532 (Fig. 1). While the bulk composition of the sediments at Site 1084 is broadly similar to Site 532, being a mixture of nannofossil ooze, clay, and biogenic opal, the sediments at this site have one of the highest organic carbon contents of all sites drilled in the Benguela region, as well as high contents of gas, extensive development of diatom mats and other indicators of very high productivity (Berger et al., 2002). Site 1084 was triple APC cored to a depth of ca 200 mbsf with essentially 100% section recovery, and XCB cored (Hole A only) to a total depth of 605 mbsf with ca 80% recovery (Wefer et al., 1998). Our samples were taken from Hole 1084A core catchers. Site 1084 has an excellent age model (Berger et al., 2002), constrained by mutually coherent biostratigraphy from several microfossil groups as well as detailed paleomagnetic data. The bottom of the hole is within the early Pliocene (ca. 4.5–4.7 Ma). 3. Results 3.1. Sedimentation rates
150.6
199.28 199.6 23 83 143 185 244
Group codes: N1 — better constrained events; N2 — less well defined events: M — magnetostratigraphic polarity interpretations (this paper). Source: Biostratigraphy: DSDP Leg 75, part 2 nannofossil report, ODP Leg 175 Scientific Results Summary Chapter Table 5. Magnetostratigraphy polarity data: DSDP Leg 75, part 1 Site 532 chapter.
Sedimentation rates and dry bulk density values for the two studied sites are shown in Fig. 4. Sedimentation rates are similar and moderately high (ca. 5 cm/kyr) prior to 3.5 Ma. They remain stable or even decline slightly in Site 532 in the remainder of the section, but in Site 1084 they increase significantly in the later Pliocene, and again in the late Pleistocene to rates > 15 cm/kyr. Dry bulk density (DBD) was computed from the relevant Initial Reports site chapters in Hay and Sibuet (1984) and Wefer et al. (1999) using the formula DBD = W(et)BD − 1026 ⁎ porosity / 100. Dry bulk density values for both sites increase downcore as expected due to compaction but not in a uniform way. In both sites, there is a step-like increase in density downcore between 2.5 and 3.5 Ma, due presumably to the reduced abundance of low density siliceous microfossil tests in the older sediments. Both
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
281
Fig. 4. Sedimentation rates and dry bulk density, DSDP Site 532 (red triangles) and ODP Site 1084 (blue circles). Lines are 3-point running averages with end point duplication for top and bottom of series. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)
sites, but particularly Site 1084 also show significant fluctuations in density that vary inversely with sedimentation rate in the upper part of the sections, again presumably reflecting changing concentration of opaline microfossil tests in the sediment. 3.2. Lithology and geochemistry measurements Carbonate and TOC concentration trends in both sites are similar, but for both measurements values are somewhat more variable in Site 1084 (Fig. 5).
Carbonate in the early Pliocene in both sections is fairly high – between 50–70% – but declines gradually to values of only 30% or less (and occasionally < 10% in Site 1084) in the latest Pliocene and earliest Pleistocene. Carbonate values increase in the later Pleistocene to values similar to those of the early Pliocene but are much more variable on short time scales than in the early Pliocene, though this may also reflect the higher general variability of Site 1084 and the different stratigraphic coverage of data between the two Sites. TOC values in both sites are low in the early Pliocene (generally <1%)
282
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
Fig. 5. Lithologic and geochemical measurements. Symbol and line conventions as in Fig. 4.
but increase gradually throughout the section, reaching values > 10% in the late Pleistocene. Site 1084 TOC values in the Pleistocene tend to be higher than those in coeval Site 532 samples. The same components show much the same shortterm variations when expressed as accumulation rates, but the long-term trends are slightly different. Carbonate accumulation at Site 1084 shows no net trend over time (values in both the early Pliocene and late Pleistocene are similar at ca. 5 mg/cm2 yr) but are lower (ca 2 mg/cm2 yr) in the 3–1 Ma time interval. Carbonate accumulation in Site 532 shows a net decline with time, from ca 4 mg/cm2 yr in the basal Pliocene to < 2 mg/cm2 yr in mid Pliocene and younger sediments. TOC% accumulation rates in the two sites also show this contrast. While TOC% accumulation rates in Site 1084 increase significantly (from ca 0.2 mg/cm2 yr to > 0.5 mg/cm2 yr), and rather abruptly at ca. 2 Ma, TOC% accumulation rates in Site 532 do not, remaining at low levels typical of the early Pliocene for both sites (0.1–0.2 mg/cm2 yr) throughout the Pleistocene. Our results for carbonate and TOC% compare well with previous measurements of these parameters for Site 532 (summarised in Meyers, 2002).
Stable oxygen isotope values for benthic foraminifera from both Sites are also shown in Fig. 5. At Site 532, Cibicidoides were abundant in the majority of the intervals sampled (there are only 4 gaps), but at Site 1084, a continuous oxygen isotope record to parallel the faunal counts and geochemical data could only be constructed by combining Cibicidoides with Uvigerina measurements (Fig. 5). We were able to analyse tests of both genera in eight intervals allowing us to determine an offset of 0.61‰, which is in excellent agreement with published Cibicidoides–Uvigerina δ18O differences of 0.64‰ (Shackleton, 1974). The standard deviation of these eight measurements is 0.11‰ suggesting that the scatter about the average offset is relatively low, and we add 0.61‰ to Cibicidoides δ18O values to bring them in line with the Uvigerina measurements and equilibrium calcite δ18O values (Shackleton, 1974). At both sites, δ18O values follow the global trend of increasing values after about 3 Ma (Fig. 5). The trend parallels an increase in ice volume corresponding to the onset of Northern Hemisphere glaciation beginning at about 3 Ma. The agreement between the two sites is particularly good between 3–2 Ma, when both records
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
display a parallel, step-wise δ18O increase. The Site 1084 age model is based on orbital tuning. The good agreement of the Site 532 with the Site 1084 δ18O record supports the revised biostratigraphic age model at Site 532 and our reinterpretation of the Site 532 paleomagnetic record. 3.3. Bulk opal measurements Given the importance of opal accumulation records in understanding Benguela system history, there are surprisingly few syntheses of the published data available. For this reason we present or at least discuss in somewhat more detail the available Neogene (i.e., pre late Quaternary) data for our two study sites. Previous Neogene bulk opal abundance estimates for Site 532 include the original shipboard smear-slide analyses of biogenic opal components, which have been reprocessed into a total biogenic silica value by NGDC (‘total siliceous components’ at http://www.ngdc.noaa.gov/ mgg/geology/dsdp/data/75/532/smear.txt), and chemical estimates of opal abundance by Dean and Parduhn (1984), made indirectly by subtracting 3.3 times the % Al2O3 value from the total percent SiO2, both as measured by XRF and ICP spectrometry. The value of 3.3 was empirically determined by Dean and Parduhn (1984) for this site by calculating the SiO2:Al2O3 ratio for each sample and noting that values were nearly constant at ca. 3.3 for samples where smear-slide estimates noted no biogenic opal. Such indirect methods are less accurate than direct estimates (indeed, the calculated values occasionally drop slightly below zero), but are useful nonetheless to suggest general trends, even if absolute values are less reliable (in this instance, they are probably significant underestimates of the true value). Opal data for Site 1084 includes semiquantitative estimates of diatom abundance (ca 10 point scale) and biogenic opal measured by chemical leaching, both data sets presented in Lange et al. (1999). In addition to these, we present our own measurements, albeit on many fewer samples, for radiolarian abundance (rads/g sed and as accumulation rate n/cm2 yr) and radiolarian preservation, plus a coarse fraction estimate of radiolarian abundance. The opal parameters show a very similar pattern at both sites (Fig. 6). Generally low opal content and flux is recorded prior to ca. 3.6 Ma by most parameters, including percent abundance of radiolarians in the coarse fraction, as well as radiolarians/g bulk sediment. Opal preservation is poor, as indicated by radiolarian preservation data in Site 532. All these parameters show a shift in both sites to higher rates of opal flux and preservation in sediments younger than ca. 3.8 Ma. An interval between
283
2 Ma and 1 Ma is marked by relatively high variability in opal parameters, particularly in Site 1084 radiolarians/g sediment, and a decline in radiolarians/g sediment and radiolarian preservation in Site 532. Radiolarians show a decline in abundance and preservation beginning at ca. 0.7 Ma that persists to the top of the section. These patterns are in very good agreement with those for opal accumulation published by previous authors, including bulk opal percent and diatom accumulation index (DAI) values from Lange et al. (1999) for Site 1084; Dean and Parduhn's (1984) biogenic SiO2 estimates for Site 532, and Site 532 smear-slide opal estimates (Hay and Sibuet, 1984). The main features of our data are also seen in these data sets, including the drop in accumulation between 2 and 1 Ma, and the further decline in younger sediments (at 0.66 Ma in the data of Lange et al., using the shipboard age model). The initial increase in opal accumulation at ca. 3.6 Ma in our data also matches well with the increase in the values of the Diatom Abundance Index of Lange et al. (1999) shortly above 4 Ma. Intriguingly, even though they are probably within the measurement error, some of the minor fluctuations in downcore patterns appear to be reproduced by the different data sets, such as the small peak in opal abundance at ca 4.1 Ma, which is most notable in the non-smoothed individual data points, in Site 532 smear-slide opal abundances and in Site 1084 DAI values. 3.4. Foraminiferal measurements Foraminiferal parameters are shown in Fig. 7. Benthic foraminiferal accumulation rates (BFAR) are considered to be a proxy for export productivity in sediments that are not markedly affected by carbonate dissolution (Herguera, 2000). As we will show, all our foraminiferal parameters, benthic as well as planktic ones, reveal strong carbonate dissolution and thus our BFAR values cannot be used as a productivity proxy. BFAR — with the exception of a single anomalously high value of 19.1 k n/cm2 ky in Sample 532-2-2, 36 cm (not shown), are all <10 k n/cm2 ky and show a consistent pattern in both sites. In the oldest part of the section (> 4 Ma) BFAR values are moderate (2–4 k n/ cm2 ky) but decline from 4–3 Ma and remain low (mostly < 2 k n/cm2 ky) until ca 1.3 Ma, after which they again increase, reaching values >6 k n/cm2 ky in late Pleistocene samples. Variability is also noticeably higher in this part of the section. We use two proxies for carbonate dissolution, based on the well-known fact that planktonic foraminifera are more easily dissolved than benthic ones (Diester-Haass et al.,
284 D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
Fig. 6. Bulk opal measurements. Symbol and line conventions as in Figs. 4 and 2.
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
285
Fig. 7. Foraminiferal measurements. Symbol and line conventions as in Fig. 4.
2004): benthic/planktic foraminiferal ratios (B/P) (calculated as [%BF / %BF + %PlF] ⁎ 100), and fragmentation of planktic foraminifera (calculated as [%fragments / % fragments + %wholetests] ⁎ 100). These proxies agree in showing most intense dissolution in the 3.5 to 2 Ma period. The fact that BFAR drops during this interval despite higher productivity is a further indication that also benthics are affected by carbonate dissolution. The ratio of Uvigerina to total benthic foraminifera in Site 532 shows a rough pattern of higher values between ca 3.2 and 1.5 Ma, and below 5 Ma, but also a fairly high degree of scatter. Values of this parameter from Site 1084 are too scattered to interpret any temporal pattern. 3.5. Radiolarian fauna Relative (percent) abundances of different ecologic components of the counted radiolarian assemblages, as well as the computed WADE ratio, are shown in Fig. 8.
The pattern seen at the two sites is very similar. The relative abundance of both warm surface water and temperate water taxa decreases in the basal to mid Pliocene, while the relative abundance of intermediate water taxa increases. Faunal composition undergoes relatively little further change between ca 3 Ma and the top of the section. The WADE ratio shows the corresponding shift towards lower values, indicating higher export productivity, throughout the lower to mid Pliocene, reaching stable low index values (high productivity) by 3 Ma, and remaining so until the top of the section. Although small sample sizes are common at both the base and top of the section, due to low radiolarian abundances and poor preservation, these do not appear to have significantly different values than coeval samples with larger sample counts. Unusually high abundances of upwelling taxa are seen in the late Pleistocene in both Sites in one short interval (0.33–0.35 Ma), and, at least in Site 1084, the sample size of 249 specimens is adequate. Thus, this appears to reflect a real event.
286
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
Fig. 8. Rad ecology assemblages. Symbol and line conventions as in Fig. 4.
3.6. Analysis of covariation of data A correlation coefficient matrix for all 23 variables is shown in Table 4. With 65 measurements in total, any value of r > 0.32 is statistically different than the null hypothesis of no correlation. However, we restrict our discussion to an arbitrary cutoff of correlations ≥0.60 to focus on the paleoceanographically most significant signals. Most variables have, in this usage of the term, significant correlation coefficients to several other variables, and these correlations were explored further by means of a scatterplot matrix (due to the large number of plots not shown). Selected plots are illustrated (Fig. 9) to document the more important correlation patterns. A few variables did not show any significant correlations to other parameters, and are thus not discussed separately in this section. These include Upwelling and Temperate rad fauna factors, BFAR, and Uvigerina/BF ratio. The sedimentation rate is correlated with several measured components, although the coefficients are not very high. In part this reflects the near constant sedimentation rate in Site 532 and/or the relatively low-
resolution age model for this site. Data for Site 1084 shows a closer correlation between sedimentation rate and TOC %, and between oxygen isotopes and sedimentation rate. Radiolarian faunal components ‘warm’ and ‘intermediate’ show similar, if opposite, correlations to the oxygen isotope curve, TOC%, carbonate percent and radiolarian abundance (as measured either by rads/g sediment or percent rads in the coarse fraction). A strong computationally driven, and thus artifactual correlation is of course also seen between these two radiolarian components and their ratio (WADE index). The WADE ratio itself shows a set of correlation patterns similar to those of the individual components, although the correlation tends to be somewhat less linear in several cases, particularly WADE vs. TOC%. A plot of log-transformed variables (Fig. 10) however shows a strong log–linear correlation (r = 0.78, p < 0.01). This pattern of non-linear data relationships and linear patterns in log-transformed data is due to the high skew in some of the variables, including in particular the WADE ratio and TOC%. Carbonate values, in addition to the above noted correlation with radiolarian faunal characteristics, also
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
show significant correlation to several other parameters. Interestingly, the correlation of carbonate is better to percent radiolarians in sand than to radiolarians/g sediment. Carbonate percent also has a significant inverse relationship to the relative abundance of benthic foraminifera in the total foraminifera assemblage (benthic plus planktonic). TOC% values also show a relationship with both carbonate and the relative abundance of benthic foraminifera in the foraminiferal assemblage. The benthic–planktonic ratio (BF/BF + PF) shows a pattern of consistent offset in the correlation pattern between the two sites, to a degree greater than for most other variables. This suggests a consistent difference in the benthic/planktonic ratio between sites. Given that there is a strong temporal trend in many parameters, including δ18O, it is not surprising that there is a noticeable correlation between the latter and the former, including sedimentation rate, radiolarian faunal composition, etc, and to these parameters and age. By contrast, the fragmentation index does not show a strong correlation – positive or negative – with any of the other variables. However this is due in part to compiling data from two sites. There is a noticeable inverse correlation visible in the scatterplot (not shown) between the fragmentation ratio and percent carbonate, but only in Site 532 data. Lastly, we re-examined the pairwise scatterplots to see if older samples and younger samples showed distinctly different, significant patterns of covariation, by the use of appropriate data point markers for samples younger and older than 2.5 Ma. Such shifts in system behavior have been noted for example in the glacial–interglacial response of opal, and could in principle so negate each other that no overall pattern of covariation is seen in the full, unseparated data set. No such patterns were seen, suggesting that changes in short-term dynamic behavior, to the extent that they exist, are masked by larger, longterm trends in our measured parameters. 4. Discussion 4.1. Generality of patterns Despite some differences between the two sites in a few data sets, one of the most important results of our study is the generality of patterns seen in the data in both sites, including the close correspondence in changes through time for individual measured characters, and also in patterns of covariation between measurement types. This suggests that the patterns do reflect general processes and not just coincidences due to the particular history of sedimentation at each site.
287
4.2. Indicators of productivity Several variables in our data show similar patterns, both in downcore plots and in the covariation analysis. This group includes one measurement thought to reflect productivity – the TOC% in the sediment – and the WADE ratio. Low TOC% values and high values of the WADE ratio in the basal Pliocene give way in the mid Pliocene (between 4 and 3 Ma) to higher TOC% values and lower WADE ratio values that are sustained throughout the remainder of the section. TOC% and the WADE ratio also show a log–linear pattern of covariation, and a strong positive skew in the distribution of values. Whether this represents some sort of power–law relationship is not known, nor, if so, what the underlying mechanisms might be. In any case, the close correspondence between TOC% and WADE in our study data, from two geographically separated sites, strongly suggests that WADE is a useful indicator of paleoproductivity, albeit with a different range of sensitivity, as suggested by the non-linear relationship of it to TOC%. Further work, and in particular, calibration studies of WADE, TOC% and productivity in recent environments will be needed to clarify the details of how these proxies respond to productivity. The upwelling factor (=URI), although responding in a broad way to the inferred increase in upwelling and productivity between 4 and 3 Ma, shows no detailed correspondence to either TOC% or WADE. This may be due to the very small percent abundance of URI taxa in our samples, or it may reflect a more basic lack of control of this parameter by productivity (see also discussion below). Bulk opal shows variable behavior as a component of Benguela system sediments, being enriched in sediments within the late Miocene and mid Pliocene, but less common in the latest Miocene/basal Pliocene, and in the late Pliocene–Recent. Further, although TOC% and bulk opal positively covary in older sediments on glacial– interglacial scales, a negative relationship exists between these proxies in glacial–interglacial comparisons in more recent sediments from the region (ca. <2 Ma) (Berger et al., 2002). The inconsistent behavior of bulk opal content of sediments and other paleoproductivity indicators such as TOC% in Benguela system sediments was coined the Opal Paradox by Leg 175 scientists (Berger et al., 2002). Examination of additional proxies, in particular the changing species composition of diatom floras, has largely resolved this discrepancy in favor of the TOC% record, and indeed this model serves as the basis for our own study. The reasons for the anomalous behavior of the bulk opal record are not known, but this phenomenon might possibly be related to increasing limitation of opaline phytoplankton productivity by reductions in dissolved
288
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
Table 4 Correlation coefficients between parameters Age Age Depth mbsf LSR cm.ky DBD g.cm3 Warm Temperate Intermediate Upwelling WADE Rads gSed RadAR.cm2.yr R.pres Rad.in.Sand. CaCO3. CaCO3.ARmg.cm2 TOC. TOC.ARmg.cm2 BFAR.n.cm2.kyr fragF UvigBF PFtotalCF d180Uvi
0.45 − 0.55 0.82 0.71 0.37 −0.76 − 0.27 0.75 − 0.52 − 0.47 − 0.47 − 0.34 0.51 0.27 −0.72 − 0.58 − 0.34 − 0.23 − 0.00 0.35 0.26 −0.83
Depth. mbsf
LSR. cm.ky
DBD. g.cm3
Warm
Temperate
Intermediate
Upwelling
WADE
Rads. g.Sed
RadAR. cm2.yr
0.45
−0.55 0.10
0.10 0.28 0.06 0.23 − 0.13 − 0.35 0.08 0.06 0.11 − 0.39 0.10 0.08 0.32 − 0.31 − 0.13 − 0.23 − 0.18 − 0.00 − 0.02 0.02 − 0.36
−0.52 −0.42 −0.14 0.46 −0.16 −0.40 0.54 0.70 0.24 0.27 −0.44 0.27 0.54 0.74 0.16 −0.03 −0.16 −0.39 −0.07 0.63
0.82 0.28 − 0.52
0.71 0.06 − 0.42 0.67
0.37 0.23 −0.14 0.25 0.04
−0.76 − 0.13 0.46 −0.66 −0.86 − 0.52
− 0.27 − 0.35 − 0.16 − 0.14 − 0.09 − 0.10 0.04
0.75 0.08 − 0.40 0.68 0.91 0.30 −0.90 − 0.13
− 0.52 0.06 0.54 − 0.56 − 0.53 − 0.27 0.59 − 0.15 − 0.52
− 0.47 0.11 0.70 − 0.42 − 0.45 − 0.21 0.50 − 0.17 − 0.43 0.92
0.67 0.25 −0.66 − 0.14 0.68 − 0.56 − 0.42 − 0.56 − 0.49 0.55 0.37 − 0.52 − 0.37 − 0.20 − 0.22 − 0.18 0.19 0.32 − 0.65
0.04 − 0.86 − 0.09 0.91 − 0.53 − 0.45 − 0.29 − 0.48 0.62 0.40 −0.63 − 0.53 − 0.16 − 0.38 − 0.12 0.26 0.44 − 0.63
−0.52 −0.10 0.30 −0.27 −0.21 −0.33 −0.32 0.42 0.25 −0.33 −0.25 0.11 0.33 −0.20 −0.11 0.39 −0.21
0.04 −0.90 0.59 0.50 0.37 0.54 −0.73 − 0.44 0.71 0.61 0.05 0.48 0.16 − 0.20 − 0.55 0.64
− 0.13 − 0.15 − 0.17 0.01 − 0.19 0.16 − 0.03 0.01 − 0.11 .56 − 0.11 − 0.03 − 0.07 0.19 0.25
− 0.52 − 0.43 − 0.34 − 0.50 0.63 0.39 −0.60 − 0.47 − 0.11 − 0.41 − 0.18 0.22 0.49 − 0.58
0.92 0.53 0.59 − 0.53 − 0.15 0.40 0.40 − 0.05 0.26 0.12 − 0.15 − 0.38 0.45
0.38 0.44 − 0.45 0.07 0.42 0.55 0.05 0.10 − 0.06 − 0.24 − 0.20 0.48
Absolute values >0.6 shown in bold.
silica concentration in the intermediate-depth upwelled water that is the source of nutrients for the system. The various measures of bulk opal in the sediment that we obtained, though in broad agreement with each other (r values generally > 0.5), also show differences. The radiolarian percentage in the coarse fraction appears to be not only partially controlled by radiolarian abundance in the sediment, but also in part (as indicated by the stronger correlation of coarse fraction radiolarians to carbonate) by the degree of dilution of radiolarians by other carbonate components (primarily planktonic foraminifera) in the coarse fraction residue. A particularly strong correlation is seen between radiolarian abundance estimators and radiolarian preservation. This correlation reflects the well known link between opal preservation in sediments and opal flux (Ragueneau et al., 2000). The most reliable bulk opal measurements are direct geochemical estimates on bulk sediment, such as those made by Lange et al. (1999) for Site 1084 and shown in Fig. 6. Since these measurements were not made on the same set of samples as ours, we cannot however quantitatively analyse the extent of correlation to our measurements but only refer to general trends. These bulk opal measurements, although strongly correlated to TOC% and WADE throughout much of the section, become decoupled in the
late Pleistocene. While TOC% and WADE remain high, bulk opal values, as indicated by direct geochemical measurements (Lange et al., 1999; Fig. 6), radiolarian/g sed and radiolarian preservation decline. This pattern was christened the Walvis Paradox by Leg 175 scientists and is thought to reflect decreased availability of dissolved silica for opal primary productivity (mostly diatoms). It is also worth noting that the decline in the abundance of radiolarians/g sed in the late Pliocene and Pleistocene in Site 1084 is less dramatic than the decline in bulk sediment opal in the same section. This difference may be due to the relatively greater availability of opal to radiolarians living at or below the thermocline, in comparison to diatoms which are largely restricted to the uppermost surface layer; or to differences in the basic growth strategy of the two organism groups — diatom ecology depends on rapid growth (cell division times measured in days, or even hours) in competition with other phytoplankton for nutrient sources (r-selection). Radiolarians by contrast have a much longer individual life span (typically around one month) and thus may not be as constrained by somewhat lower concentrations of nutrients, or, as zooplankton, may be better able to harvest them in solid form. Carbonate accumulation rates, as well as the accumulation rates of benthic foraminifera, show complex patterns of behavior in Benguela system sediments, but in general
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
289
R.pres
Rad.in. Sand
CaCO3
CaCO3. ARmg.cm2yr
TOC%
TOC.ARmg. cm2yr
BFAR.n. cm2.kyr
BF. allF
fragF
Uvig. BF
PFtotalCF
d180. Uvi
− 0.47 − 0.39 0.24 − 0.56 − 0.29 − 0.33 0.37 0.01 − 0.34 0.53 0.38
− 0.34 0.10 0.27 − 0.49 − 0.48 − 0.32 0.54 − 0.19 − 0.50 0.59 0.44 0.41
0.51 0.08 −0.44 0.55 0.62 0.42 −0.73 0.16 0.63 −0.53 −0.45 −0.30 −0.75
0.27 0.32 0.27 0.37 0.40 0.25 − 0.44 − 0.03 0.39 − 0.15 0.07 − 0.18 − 0.53 0.66
−0.72 −0.31 0.54 −0.52 −0.63 −0.33 0.71 0.01 −0.60 0.40 0.42 0.19 0.41 − 0.71 −0.38
− 0.58 − 0.13 0.74 − 0.37 − 0.53 − 0.25 0.61 − 0.11 − 0.47 0.40 0.55 0.11 0.35 −0.63 − 0.12 0.90
− 0.34 − 0.23 0.16 − 0.20 − 0.16 0.11 0.05 0.56 − 0.11 − 0.05 0.05 − 0.15 − 0.39 0.19 0.17 0.19 0.14
− 0.23 − 0.18 − 0.03 − 0.22 − 0.38 − 0.33 0.48 − 0.11 − 0.41 0.26 0.10 0.18 0.57 −0.66 −0.63 0.54 0.32 − 0.15
− 0.00 − 0.00 − 0.16 − 0.18 − 0.12 − 0.20 0.16 − 0.03 − 0.18 0.12 − 0.06 0.22 0.45 − 0.27 − 0.29 0.08 − 0.05 − 0.23 0.50
0.35 − 0.02 − 0.39 0.19 0.26 − 0.11 − 0.20 − 0.07 0.22 − 0.15 − 0.24 0.06 0.03 0.07 − 0.10 − 0.25 − 0.28 − 0.24 0.13 0.16
0.26 0.02 −0.07 0.32 0.44 0.39 −0.55 0.19 0.49 −0.38 −0.20 −0.27 −0.78 0.75 0.63 −0.50 −0.33 0.30 −0.93 −0.53 −0.11
−0.83 − 0.36 0.63 −0.65 −0.63 − 0.21 0.64 0.25 − 0.58 0.45 0.48 0.45 0.28 − 0.41 − 0.05 0.66 0.63 0.34 0.13 − 0.09 − 0.30 − 0.15
0.41 − 0.30 − 0.18 0.19 0.11 − 0.15 0.18 0.22 0.06 − 0.27 0.45
−0.75 − 0.53 0.41 0.35 − 0.39 0.57 0.45 0.03 −0.78 0.28
0.66 −0.71 −0.63 0.19 −0.66 −0.27 0.07 0.75 −0.41
− 0.38 − 0.12 0.17 − 0.63 − 0.29 − 0.10 0.63 − 0.05
0.90 0.19 0.54 0.08 −0.25 −0.50 0.66
0.14 0.32 − 0.05 − 0.28 − 0.33 0.63
show both long-term (my) and short-term (glacial– interglacial) inverse correlations to TOC%. TOC% contents of sediments show by contrast a more consistent increase from the late Miocene to Recent, suggesting gradually increasing productivity of the system over time. Changes in the abundance and composition of carbonate phases in the sediment in our study appear to reflect not only productivity change but also to a great degree change in carbonate dissolution. This effect is seen in the substantial inverse relationship between TOC% and WADE, on the one hand, and carbonate percent and BFAR, on the other. This interpretation is also supported by the inverse correlation of carbonate content and dissolution indicators such as the benthic/planktonic foraminiferal ratio, and (albeit only for Site 532) the fragmentation index. The use of the ratio of Uvigerina to other benthics as a productivity indicator was previously proposed by Diester-Haass et al. (2004). Berger and Wefer (2002) suggest that benthic foraminiferal faunal composition in general may not be of much use in high productivity (upwelling) regions, based on the results by Anderson et al. (2002) from Quaternary sediments from Site 1085. However, in Anderson et al. (2002) only Uvigerina/g sediment values are presented, not differences in faunal composition: the U/g values closely follow the patterns seen for BFAR as a whole. Our results show
− 0.15 − 0.23 − 0.24 0.30 0.34
0.50 0.13 −0.93 0.13
0.16 − 0.53 − 0.09
− 0.11 − 0.30
−0.15
clearly different patterns of response for BFAR and the U/ BF ratio, although neither parameter shows substantial correlations to any of the other measured parameters in our study. Licari and Mackensen (2005) showed that the species composition of benthic foraminifera in Recent sediments does reflect changes in productivity in the Benguela region, suggesting that, similar to radiolarians, a taxonomic/ecologic approach may also be beneficial for study of Neogene age benthic foraminifera from Benguela region sediments. 4.3. Comparison to previous paleoproductivity studies using radiolarians The URI method has been validated by several studies, mostly in Indian Ocean sediments (Caulet et al., 1992; Venec-Peyre et al., 1995). However, several of the species appear to be restricted to the Pleistocene–Recent, limiting use in older sediments. In our study we observed only a very broad correlation between URI and other indices for paleoproductivity, and no strong correlations to any other parameter in our study, including WADE and other radiolarian faunal components. Published URI curves have also often showed complex patterns of change, with correlation to other parameters such as glacial–interglacial change in carbon and oxygen isotopes being only
290
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
Fig. 9. Scatterplots of selected parameters. Symbol conventions as in Fig. 4.
moderate, and sometimes phase shifted as well. We cannot comment on previous comparisons between WADE and URI methods because, to our knowledge, no published
data using both methods exist. The low resolution of our sampling also precludes a more direct comparison between URI and other parameters. All these observations suggest
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
291
particular, shows a strong correlation to TOC% in paired sediment samples. The WADE ratio thus appears to be a reliable indicator of past productivity in Benguela sediments, at least back to the latest Miocene. The relation between TOC% and WADE however is not linear. Whether this non-linearity reflects differences in the relative sensitivity of the two methods, artifacts due to the computation of WADE as a ratio, or other factors are not known. 4.4. Radiolarian faunal methods and opal accumulation
Fig. 10. Log–log plot of WADE and TOC%. Symbol conventions as in Fig. 4.
that use of URI as a proxy for paleoproductivity needs further work, and may be of use only in Pleistocene studies. URI taxa may for example simply reflect short-term, possibly seasonal aspects of upwelling conditions in the water column, even if the upwelling is not associated with strongly enhanced productivity. Indeed, as noted above, the actual oceanographic parameters that control the occurrence of URI taxa have not been explored either theoretically or by measurement in modern environments. WADE (Water Depth Ecology) methods are based on ratios of deep-dwelling to surface species, which should reflect the relative increase of organic carbon export from surface waters due to elevated productivity. This idea was first quantitatively employed by Jacot Des Combes et al. (1999) in the form of a Thermocline Surface Radiolarian Index (TSRI) in Pleistocene Indian Ocean sediments. Weinheimer (2001) broadened the method to include a much larger number of surface and intermediate water depth species in a short pilot study of late Pleistocene Benguela sediments from Leg 175 Site 1082. Her study reported faunal changes only by depth in section, however Berger et al. (2002, p. 25) estimate ages for her data, and note that (as expected in the model proposed by Casey et al. (1982)) the increase in intermediate taxa of radiolarians seen by Weinheimer (2001) corresponds to the last glacial interval, when other indicators at this site also suggest increased productivity. In our study we have extended the WADE method used by Weinheimer (2001) to include additional taxa, and have applied it to latest Miocene–Recent sediments in two sediment cores from the Benguela region. Our results show that the WADE ratio reflects the long-term trend in productivity change in these sediments, and in
Our results suggest that the WADE ratio does not appear to be tied to bulk opal productivity as such. Although our own measurements do not fully reflect bulk opal accumulation, comparison of our radiolarian faunal results, particularly WADE, to published bulk opal accumulation in our study sites suggests that opal accumulation rates do not significantly affect WADE values. This is reinforced by direct estimates of radiolarian preservation in our samples. Here, despite poor preservation in late Pleistocene samples, WADE ratios remain unchanged from values in underlying, better preserved samples where, from TOC% and other estimates (Marlow et al., 2000) productivity was relatively high in both intervals. Admittedly this observation is based on only a few samples, and needs to be tested with additional observations in the future. 4.5. Proxies of export productivity vs proxies for carbon burial It is also worth noting one consequence of our observation above, that the patterns seen in the parameters discussed above are so similar between the two sites studied, despite the major difference between the sites in the sedimentation rate in the younger part of the section (<2 Ma). This similarity is particularly true for the radiolarian faunal composition, but it also holds for TOC% and other concentration/percent measurements, which are very similar between the two sites, despite, at least in the younger part of the sections, major (>3×) differences in accumulation rate. To the extent that TOC%, radiolarian faunal characteristics etc are used as proxies not just for export productivity in the ocean to intermediate/deep waters, but also as proxies for export of carbon into the sedimentary record, this is a cause for concern, as the burial rate of carbon in younger, much higher sedimentation rate Site 1084 sediments clearly is much greater than in coeval sediments from Site 532, as is shown by the accumulation rate data for both carbonate and TOC%. The differences in sedimentation rate in the upper part of the sections do not substantially affect the relative strength of correlation
292
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
between parameters in our statistical analysis, but this may be due in part to the small number of samples in the <2 Ma part of Site 532, so that the correlation values are dominated by the signal from the higher sedimentation rate Site 1084 data. A larger data set from these younger sediments will be needed to examine this issue. More generally, it needs to be remembered that productivity proxies measure, however indirectly, a surface water biologic phenomenon, and are in principle independent of the complex processes of sedimentation and diagenesis at, and below the ocean bottom that determine the degree to which this oceanic signal is transformed into a geologic one. 5. Summary and conclusions We present our conclusions by first returning to the questions posed at the beginning of this paper. Do WADE and/or URI methods produce coherent, consistent signals in this long time-series data set? Yes, although more for WADE than URI indices. WADE values show a remarkable degree of consistency between the two studied sites, and they show a gradual shift towards lower values from the latest Miocene to the latest Pleistocene. URI values are much more variable within the section and primarily provide a simple ‘binary’ type of signal — URI taxa are either present, or generally absent. The switch between the two types of behavior appears to occur in both sites at the same time, although the absence of older sediment in Site 1084 makes this interpretation tentative. The absolute abundance of URI taxa appears to be somewhat higher on average in Site 532 than in Site 1084. This is surprising, since 1084 in general is more strongly affected by upwelling than Site 532. Do these indices correlate well with other paleoproductivity proxies, and in particular, with the TOC record from late Neogene Benguela sediments? Yes, though again, primarily for WADE. A clear correlation exists between other productivity proxies and WADE values. Although complicated by a non-linear relationship and skewed value distributions, WADE is closely correlated to TOC%, and thus appears to reflect changing productivity. Are radiolarian faunal methods significantly affected by changes in bulk opal accumulation and/or changes in the quality of opal preservation? No. Although in our own study we have only relatively few samples to test this, radiolarian faunal signals for both WADE and URI better track the general long-term trend of productivity than either bulk opal accumulation or opal preservation, as measured by the radiolarian preservation index. Weinheimer's (2001) results for late Pleistocene sediments in the Benguela system also support this conclusion.
In general, the WADE ratio may be considered to be analogous to BFAR, in that both reflect enhanced protist consumption of export organic flux from surface waters. WADE however is not affected by errors in age models as it does not incorporate a rate computation; nor, apparently, is it influenced to the same degree as BFAR by dissolution, and certainly not by carbonate dissolution. Lastly, WADE measures export productivity in near thermocline waters rather than at the sediment–water interface, and thus is much less likely to be affected by down-slope transport of phases as can be the case with TOC%. The most obvious limitation of WADE (or any other radiolarian faunal method) is the limited distribution of radiolarians in sediments, particularly in the carbonate rich, opal poor Atlantic basin. However it provides an important complementary method for estimating paleoproductivity, and is of particular value in regions where carbonate preservation is poor, including upwelling regions in the Atlantic. The results of this first application of radiolarian faunal methods to paleoproductivity estimation in prePleistocene sediments are encouraging. Many questions however remain unanswered and will need to be addressed in future work. First, the generality of our results is so far based only on two sites from the same subtropical region of one upwelling system. Additional sections, from other upwelling regions will be needed to confirm the broader utility of WADE or other indices. Second, although a good correlation exists between WADE and TOC%, we do not as yet have any independent calibration of WADE to productivity in modern ocean environments. Lastly, as noted above, the robustness of WADE to changing opal abundance and preservation is currently based on a relatively small number of poorly preserved late Pleistocene samples, and will need further testing. Despite these caveats, our results suggest that radiolarian faunal analysis can contribute useful complementary data to the study of pre-Pleistocene paleoproductivity in low-latitude upwelling regions. Acknowledgements The authors wish to thank Dr. Demetrio Boltovskoy for sharing his unpublished plankton data and Dr. Isao Motoyama for sending us the Site 1084 samples used in our study as well as unpublished analyses of his Site 1084 radiolarian data. Jörn Slotta, Saarbücken; Silvia Salzmann, Berlin and John Yuen, Ann Arbor are thanked for their careful sample preparation. Guiseppe Cortese and Kozo Takahashi provided thoughtful reviews of the draft version of the ms.
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j. marmicro.2006.06.003. References Abelmann, A., Gowing, M.M., 1996. Horizontal and vertical distribution pattern of living radiolarians along a transect from the Southern Ocean to the South Atlantic subtropical region. Deep-Sea Research. Part 1. Oceanographic Research Papers 43 (3), 361–382. Anderson, P.A., Charles, C.D., Berger, W.H., 2002. In: Wefer, G., Berger, W.H., Richter, C. (Eds.), Walvis Paradox Confirmed for the Early Quaternary at the Southern End of the Namibia Upwelling System, ODP Site 1085. Proceedings of the Ocean Drilling Program, Scientific Results, vol. 175. Ocean Drilling Program, College Station, TX, pp. 1–31. Berger, W.H., Wefer, G., 2002. On the reconstruction of upwelling history: namibia upwelling in context. Marine Geology 180, 3–28. Berger, W.H., Lange, C.B., Wefer, G., 2002. Upwelling history of the Benguela–Namibia system: a synthesis of leg 175 results. Proceedings of the ODP, Scientific Results 175, 1–103. Berggren, W.A., Kent, D.V., Swisher, I.C.C., Aubry, M.-P., 1995. A revised Cenozoic geochronology and chronostratigraphy. In: Berggren, W.A., Kent, D.V., Aubry, M.P., Hardenbol, J. (Eds.), Geochronology, Time Scales and Stratigraphic Correlation: Framework for an Historical Geology. Special Publication. Society of Economic Paleontologists and Mineralogists, Tulsa, pp. 129–212. Boersma, A., 1984. In: Hay, W.W., Sibuet, J.-C. (Eds.), Pliocene planktonic and benthic foraminifers from the southeastern Atlantic Angola margin: Leg 75, Site 532, Deep Sea Drilling Project. Initial Reports of the Deep Sea Drilling Project, vol. 75. U. S. Government Printing Office, Washington, D.C., pp. 657–669. Boltovskoy, D., 1998. Classification and distribution of South Atlantic Recent polycystine Radiolaria. Paleontologica Electronica 1 (2), 116 (web). Boltovskoy, D., Riedel, W.R., 1980. Polycystine Radiolaria from the southwestern Atlantic ocean plankton. Revista Espanola de Micropaleontologia 12 (1), 99–146. Boltovskoy, D., Oberhänsli, H., Wefer, G., 1996. Radiolarian assemblages in the eastern tropical Atlantic: patterns in the plankton and in sediment trap samples. Journal of Marine Systems 8, 31–51. Casey, R.E., Spaw, J.M., Kunze, F.R., 1982. Polycystine radiolarian distributions and enhancements related to oceanographic conditions in a hypothetical ocean. Transactions of the Gulf Coast Association of Geological Societies 32, 319–332. Caulet, J.P., Venec-Peyre, M.-T., Vergnaud-Grazzini, C., Nigrini, C., 1992. Variation of South Somalian upwelling during the last 160 ka: radiolarian and foraminifera records in core MD 85674. In: Summerhayes, C.P., Prell, W.L., Emeis, K.C. (Eds.), Upwelling Systems: Evolution Since the Early Miocene. Geological Society of London, Special Publication, pp. 379–389. London, UK. Chen, P.H., 1975. Antarctic Radiolaria. In: Hayes, D.E., Frakes, L.A. (Eds.), Initial Reports of the Deep Sea Drilling Project. U. S. Government Printing Office, Washington, D.C., pp. 437–513. Cortese, G., Bjorklund, K.R., 1998. The taxonomy of boreal Atlantic Ocean Actinommida (Radiolaria). Micropaleontology 44 (2), 149–160. Dean, W.E., Parduhn, N.L., 1984. In: Hay, W.W., Sibuet, J.-C. (Eds.), Inorganic geochemistry of sediments and rocks recovered from the
293
southern Angola basin and adjacent Walvis Ridge, sites 530 and 532, Deep Sea Drilling Project Leg 75. Initial Reports of the Deep Sea Drilling Project, vol. 75. U. S. Government Printing Office, Washington, D.C., pp. 923–958. Dean, W.E., Hay, W.W., Sibuet, J.-C., 1984. In: Hay, W.W., Sibuet, J.-C. (Eds.), Geologic evolution, sedimentation, and paleoenvironments of the Angola basin and adjacent walvis ridge: Synthesis of results of Deep Sea Drilling Project Leg 75. Initial Reports of the Deep Sea Drilling Project, vol. 75. U. S. Government Printing Office, Washington, D.C., pp. 509–544. Diester-Haass, L., Meyers, P.A., Roth, P., 1992. The Benguela Current and associated upwelling on the southwest African Margin: a synthesis of the Neogene–Quaternary sedimentary record at DSDP sites 362 and 532. In: Summerhayes, C.P., Prell, W.L., Emeis, K.C. (Eds.), Upwelling Systems: Evolution Since the Early Miocene. Special Publications. The Geological Society, London, pp. 331–342. Diester-Haass, L., Meyers, P.A., Vidal, L., 2002. The Late Miocene onset of high productivity in the Benguela Current upwelling system as part of a global pattern. Marine Geology 180, 87–103. Diester-Haass, L., Meyers, P.A., Bickert, T., 2004. Carbonate crash and biogenic bloom in the late Miocene: evidence from ODP Sites 1085, 1086, and 1087 in the Cape Basin, southeast Atlantic Ocean. Paleoceanography 19, 1–19 (web). Gardner, J.V., Dean, W., Wilson, C.R., 1984. In: Hay, W.W., Sibuet, J.-C. (Eds.), Carbonate and organic-carbon cycles and the history of upwelling at Deep Sea Drilling Project Site 532, Walvis Ridge, South Atlantic Ocean. Initial Reports of the Deep Sea Drilling Project, vol. 75. U. S. Government Printing Office, Washington, D.C., pp. 905–922. Hay, W.W., Brock, J.C., 1992. Temporal variation in intensity of upwelling off southwest Africa. In: Summerhayes, C.P., Prell, W.L., Emeis, K.C. (Eds.), Upwelling Systems: Evolution Since the Early Miocene. Special publications. Geological Society, London, pp. 463–497. Hay, W.W., Sibuet, J.-C. (Eds.), 1984. Initial Reports of the Deep Sea Drilling Project, vol. 75. U. S. Government Printing Office, Washington, D.C., p. 1303. Hays, J.D., 1965. Radiolaria and late Tertiary and Quaternary history of Antarctic Seas. In: Llano, G.A. (Ed.), Biology of Antarctic Seas II American Geophysical Union, Antarctic Research Series, pp. 125–184. Herguera, J.C., 2000. Last glacial paleoproductivity patterns in the eastern equatorial Pacific: benthic foraminifera records. Marine Micropaleontology 40, 259–275. Jacot Des Combes, H., Caulet, J.P., Tribovillard, N.P., 1999. Pelagic productivity changes in the equatorial area of the northwest Indian Ocean during the last 400,000 years. Marine Geology 158, 27–55. Kling, S.A., 1977. Local and regional imprints on radiolarian assemblages from California coastal basin sediments. Marine Micropaleontology 2 (3), 207–221. Kling, S.A., 1979. Vertical distribution of polycystine radiolarians in the central North Pacific. Marine Micropaleontology 4 (4), 295–318. Kling, S.A., Boltovskoy, D., 1995. Radiolarian vertical distribution patterns across the southern California Current. Deep-Sea Research 42 (2), 191–231. Lange, C.B., Berger, W.H., Lin, H.-L., Wefer, G., 1999. The early Matuyama Diatom Maximum off SW Africa, Benguela Current System (ODP Leg 175). Marine Geology 161 (2–4), 93–114. Lazarus, D.B., 1992. Age depth plot and age maker: age depth modelling on the Macintosh series of computers. Geobyte 7 (1), 7–13. Lazarus, D., 1994a. An improved cover slip holder for preparing microslides of randomly distributed particles. Journal of Sedimentary Petrology A64 (3), 686.
294
D. Lazarus et al. / Marine Micropaleontology 60 (2006) 269–294
Lazarus, D.B., 1994b. The Neptune Project — a marine micropaleontology database. Mathematical Geology 26 (7), 817–832. Lazarus, D.B., 2005. A brief review of radiolarian research. Paläontologische Zeitschrift 79 (1), 183–200. Lazarus, D.B., Scherer, R.P., Prothero, D.R., 1985. Evolution of the radiolarian species-complex Pterocanium: a preliminary survey. Journal of Paleontology 59 (1), 183–220. Lazarus, D., Faust, K., Popova-Goll, I., 2005. New species of prunoid radiolarians from the Antarctic Pliocene and Miocene. Journal of Micropaleontology 00, 00-00. Licari, L., Mackensen, A., 2005. Benthic foraminifera off West Africa (1° N to 32° S): do live assemblages from the topmost sediment reliably record environmental variability? Marine Micropaleontology 55, 205–234. Lombari, G., Boden, G., 1985. Modern Radiolarian Global Distributions, Special Publication, vol. 16A. Cushman Foundation for Foraminiferal Research. Lombari, G., Lazarus, D.B., 1988. Neogene cycladophorid radiolarians from the North Atlantic, Antarctic, and North Pacific deep-sea sediments. Micropaleontology 34 (2), 97–135. Marlow, J.R., Lange, C.B., Wefer, G., Rosell-Melé, A., 2000. Upwelling intensification as part of the Pliocene–Pleistocene climate transition. Science 290, 2288–2291. Meyers, P.A., 2002. In: Wefer, G., Berger, W.H., Richter, C. (Eds.), Miocene–Pleistocene sedimentary record of carbon burial under the Benguela Current upwelling system, southwestern margin of Africa. Proceedings of the Ocean Drilling Program, Scientific Results, vol. 175. Ocean Drilling Program, College Station, TX, pp. 1–19. Moore, T.C.J., 1973. Method of randomly distributing grains for microscopic examination. Journal of Sedimentary Petrology 43, 904–906. Morley, J.J., Nigrini, C., 1995. Miocene to Pleistocene radiolarian biostratigraphy of North Pacific sites 881, 884, 885, 886 and 887. In: Rea, D.K., Basov, I.A., Scholl, D.W., Allan, J.F. (Eds.), Proceedings of the Ocean Drilling Program, Scientific Results. Ocean Drilling Program, College Station, TX, pp. 55–91. Motoyama, I., 2001. Late Cenozoic radiolarians from South Atlantic Hole 1082A, Leg 175. In: Wefer, G., Berger, W.H., Richter, C. (Eds.), Proceedings of the Ocean Drilling Program, Science Results, vol. 175. Ocean Drilling Program, College Station, TX, pp. 1–26. www-odp.tamu.edu, web. Neptune database. www.chronos.org. Nigrini, C., Caulet, J.P., 1992. Late Neogene radiolarian assemblages characteristic of Indo-Pacific areas of upwelling. Micropaleontology 38 (2), 139–164.
Nigrini, C., Moore, T.C., 1979. A guide to modern Radiolaria, Special Publication. Cushman Foundation for Foraminiferal Research. pp. i–xii, S1-S142, N1–N106, bibliography 10 pages, Index 8 pages. Nigrini, C., Sanfilippo, A., 2001. Cenozoic radiolarian stratigraphy for low and middle latitudes. ODP Technical Notes, vol. 27. Ocean Drilling Program, College Station, TX. web. Nimmergut, A., Abelmann, A., 2002. Spatial and seasonal changes of radiolarian standing stocks in the Sea of Okhotsk. Deep-Sea Research. Part 1. Oceanographic Research Papers 49 (3), 463–493. Ragueneau, O., Tréguer, P., Leynaert, A., Anderson, R.F., Brzezinski, M.A., DeMaster, D.J., Dugdale, R.C., Dymond, J., Fischer, G., François, R., Heinze, C., Maier-Reimer, E., Martin-Jézéquel, V., Nelson, D.M., Quéguiner, B., 2000. A review of the Si cycle in the modern ocean: recent progress and missing gaps in the application of biogenic opal as a paleoproductivity proxy. Global and Planetary Change 26, 317–365. Shackleton, N.J., 1974. Attainment of isotopic equilibrium between ocean water and the benthonic foraminifera genus Uvigerina: isotopic changes in the ocean during the last glacial. Colloques Internationaux du Centre National du Recherche Scientifique 219, 203–210. Siesser, W.G., 1980. Late Miocene origin of the Benguela upwelling system off northern Namibia. Science 208, 283–285. Steinmetz, J.C., et al., 1984. In: Hay, W.W., Sibuet, J.-C. (Eds.), Summary of biostratigraphy and magnetostratigraphy of Deep Sea Drilling Project Leg 75. Initial Reports of the Deep Sea Drilling Project, vol. 75. U. S. Government Printing Office, Washington, D.C., pp. 449–458. Venec-Peyre, M.-T., Caulet, J.-P., Vergnaud Grazzini, C., 1995. Paleohydrographic changes in the Somali Basin (5 degree N upwelling and equatorial areas) during the last 160 kyr, based on correspondence analysis of foraminiferal and radiolarian assemblages. Paleoceanography 10 (3), 473–491. Wefer, G., et al., 1998. Proceedings of the Ocean Drilling Program, Initial Reports, vol. 175. Ocean Drilling Program, College Station, TX, p. 1477. Wefer, G., Berger, W.H., Bijma, J., Fischer, G., 1999. Clues to ocean history: a brief overview of proxies. In: Fischer, G., Wefer, G. (Eds.), Use of Proxies in Paleoceanography: Examples from the South Atlantic. Springer, Berlin, pp. 1–68. Weinheimer, A., 2001. Radiolarians from the Northern Cape Basin, Site 1082. In: Wefer, G., Berger, W.H., Richter, C. (Eds.), Proceedings of the Ocean Drilling Program, Science Results, 175. Ocean Drilling Program, College Station, TX, pp. 1–16. www-odp.tamu.edu, web.