Coral growth bands - CiteSeerX

Click Here

PALEOCEANOGRAPHY, VOL. 21, PA4217, doi:10.1029/2006PA001288, 2006

for

Full Article

Coral growth bands: A new and easy to use paleothermometer in paleoenvironment analysis and paleoceanography (late Miocene, Greece) Thomas C. Brachert,1 Markus Reuter,1,2 Karsten F. Kroeger,1,3 and Janice M. Lough4 Received 9 March 2006; revised 11 July 2006; accepted 20 July 2006; published 19 December 2006.

[1] Modern scleractinian corals are classical components of marine shallow warm water ecosystems. Their occurrence and diversity patterns in the geological record have been widely used to infer past climates and environmental conditions. Coral skeletal composition data reflecting the nature of the coral environment are often affected by diagenetic alteration. Ghost structures of annual growth rhythms are, however, often well preserved in the transformed skeleton. We show that these relicts represent a valuable source of information on growth conditions of fossil corals. Annual growth bands were measured in massive hemispherical Porites of late Miocene age from the island of Crete (Greece) that were found in patch reefs and level bottom associations of attached mixed clastic environments as well as isolated carbonate environments. The Miocene corals grew slowly, about 2–4 mm yr1, compatible with present-day Porites from high-latitude reefs. Slow annual growth of the Miocene corals is in good agreement with the position of Crete at the margin of the Miocene reef belt. Within a given time slice, extension rates were lowest in level bottom environments and highest in attached inshore reef systems. Because sea surface temperatures (SSTs) can be expected to be uniform within a time slice, spatial variations in extension rates must reflect local variations in light levels (low in the level bottom communities) and nutrients (high in the attached reef systems). During the late Miocene (Tortonian–early Messinian), maximum linear extension rates remained remarkably constant within seven chronostratigraphic units, and if the relationship of SSTs and annual growth rates observed for modern massive Indo-Pacific Porites spp. applies to the Neogene, minimum (winter) SSTs were 20°–21°C. Although our paleoclimatic record has a low resolution, it fits the trends revealed by global data sets. In the near future we expect this new and easy to use Porites thermometer to add important new information to our understanding of Neogene climate. Citation: Brachert, T. C., M. Reuter, K. F. Kroeger, and J. M. Lough (2006), Coral growth bands: A new and easy to use paleothermometer in paleoenvironment analysis and paleoceanography (late Miocene, Greece), Paleoceanography, 21, PA4217, doi:10.1029/2006PA001288.

1. Introduction [2] Massive zooxanthellate scleractinian corals are highresolution archives of past climates and coral reef environments. The geochemical signatures of the coral skeleton are most commonly used as proxy records of atmospheric and oceanic variations and carbon budgets of the ocean [Druffel, 1997; Felis and Pa¨tzold, 2004; Grottoli, 2001; Knutson et al., 1972]. In addition to annual calcification and linear extension rates being directly linked to ambient water temperature and water depth, other factors, such as nutrient concentration or wave exposure play a role [Barnes and Lough, 1993; Heiss, 1994; Logan and Tomascik, 1991; 1 Institut fu¨r Geowissenschaften, Johannes Gutenberg-Universita¨t Mainz, Mainz, Germany. 2 Now at Institut fu¨r Erdwissenschaften, Bereich Geologie und Palaöntologie, Karl-Franzens-Universita¨t Graz, Graz, Austria. 3 Now at Geoforschungszentrum Potsdam, Potsdam, Germany. 4 Australian Institute of Marine Science, Townsville, Queensland, Australia.

Copyright 2006 by the American Geophysical Union. 0883-8305/06/2006PA001288$12.00

Logan et al., 1994; Lough and Barnes, 2000; Rosenfeld et al., 2003]. However, the scleractinian aragonite skeleton is unstable in most diagenetic environments and undergoes rapid textural change and chemical alteration [Constantz, 1986; Dullo, 1984; Dullo and Mehl, 1989; McGregor and Gagan, 2003; Reuter et al., 2005]. Consequently, geochemical proxy records from corals are limited to the more recent geological past and usually not older than Pleistocene [Brachert et al., 2006; Roulier and Quinn, 1995; Tudhope et al., 2001]. Paleoenvironmentally, the annual growth increments are of particular significance, because in diagenetically altered scleractinian skeletons they are most commonly preserved as ghost structures and are easy to measure in the field or in the laboratory [Reuter et al., 2005]. A few studies have explored the potential of using ghost structures of growth bands in fossil corals in comparative paleoenvironment analyses, or for tracing the history of algal symbiosis [Ali, 1984; Geister, 1984, 1989; Helmle and Stanley, 2003; Insalaco, 1996; Shinn, 1966]. Here we describe massive hemispherical Porites of late Miocene age from the island of Crete (eastern Mediterranean), that appear particularly useful for testing the potential of the

PA4217

1 of 12

PA4217

BRACHERT ET AL.: GROWTH BAND ANALYSIS

PA4217

Figure 1. Late Miocene shallow marine biofacies of the Mediterranean region. Numbers show modern February water temperatures at 10 m depth (available from http://ingrid.ldeo.columbia.edu/). EMWP is the Eastern Mediterranean Warm Pool. C, I, and T give the location of Crete, southern Italy, and southern Turkey, respectively. For references see auxiliary material. growth band approach in paleoenvironment analysis and paleoclimatology. 1.1. Mediterranean Reef Environments of the Late Miocene [3] Cenozoic plate movements transformed the central Tethys Ocean into the Mediterranean Sea. By the end of the middle Miocene, the sea straits connecting the Indo-Pacific and Mediterranean had closed, and the Mediterranean formed a deep marginal sea of the Atlantic. Marine passages in the Betic and Rifian domain were increasingly restricted in the course of the late Miocene, eventually leading to complete isolation of the Mediterranean during the Messinian Salinity Crisis [Meulenkamp and Sissingh, 2003; Ro¨gl and Steininger, 1984]. Late Miocene eastern Mediterranean oceanography was affected by high-frequency changes on astronomical timescales [Hilgen et al., 2003; Kouwenhoven et al., 1999; Schenau et al., 1999]. Finally, the new and efficient Gibraltar passage into the Atlantic was established by the Zanclean (early Pliocene) [Meulenkamp and Sissingh, 2003; Ro¨gl and Steininger, 1984]. At the present day, incoming surface waters from the Atlantic become more saline and warmer toward the east, where they form the oligotrophic Eastern Mediterranean Warm Pool (EMWP, new term) with winter sea surface temperatures (SST) remaining warm temperate (17°C at 10 m depth, available from http://ingrid.ldeo.columbia.edu/) but still below the critical threshold for reef development (18°C [Abram et al., 2001], Figure 1). A shallow sill at Gibraltar prevents inflow of cool deep water into the basin and allows formation of a remarkably homiothermal water column in the Mediterranean.

[4] Late Miocene climate was transitional between the middle Miocene optimum and the icehouse state of the late Neogene [Zachos et al., 2001]. In addition to late Miocene warm climate, the Mediterranean was positioned slightly further to the south and has subsequently moved into its present location [Perrin, 2002]. Late Miocene warm SSTs in the Mediterranean area are documented by widespread coral reef growth, but no reefs existed in the northern Mediterranean Sea (Gulf of Lyons, northern Aegean Sea) nor to the west of the Betic-Rifian chain in the eastern Atlantic [Esteban, 1996]. Reef growth was intermittent in the western Mediterranean, where constructional coral communities changed with level bottom carbonate communities [Brachert et al., 1996; Martıń and Braga, 1994; Montenat, 1990]. Toward the east, carbonates formed by nonreefal level bottom communities become less abundant (Figure 1). The Mediterranean reef biogeography, the remarkably low diversity of reef corals and the discontinuous growth patterns of the reefs is generally believed to reflect a position of the Mediterranean at the northern margin of the global coral reef belt [Brachert et al., 1996; Esteban, 1996]. Final extinction of the reef ecosystem in the Mediterranean area by the Pliocene is then regarded a consequence of climatic cooling [Esteban, 1996; Martıń and Braga, 1994]. 1.2. Case Study: Central Crete, Eastern Mediterranean [5] Neogene basins of central Crete (Figure 2) were formed by extensional tectonics during the middle Miocene [Fassoulas, 2001; Meulenkamp and Sissingh, 2003; Rahl, 2004; Reuter et al., 2006]. Subsequent to an episode of

2 of 12

PA4217


PA4217

Figure 2. Location of sampling sites in central Crete. Longitude and latitude for map window on top left are N35° to N36°, E23° to E27°.

nonmarine sedimentation, during the late middle Miocene, the basins of Iraklion and Messara in central Crete were inundated by the sea [Meulenkamp et al., 1979]. After an episode of clastic sedimentation during the early Tortonian, climatic aridification led to increasing sediment starvation of the basins. Remaining alluvial sediment load was trapped in the marginal zone of the basin, when carbonates and finally evaporites became dominant during the Messinian [Reuter et al., 2006; Zidianakis et al., 2004]. [6] The upper Miocene section has been assigned to three lithostratigraphic units, the Ambelouzos Formation of Tor-

tonian age, and the Varvara Formation and Pirgos Member of late Tortonian-Messinian age [Krijgsman et al., 1999; Kuiper et al., 2004; Meulenkamp et al., 1979; Santarelli et al., 1998; ten Veen and Postma, 1996, 1999] (Figure 3). The chronostratigraphic classification and age model for late Miocene shallow water deposits of the western margin of the Iraklion Basin relies on stratal architectures and stacking patterns of seven unconformity bound depositional units (‘‘coral levels’’) grouped into three larger-scale depositional sequences [Reuter et al., 2006] (Figure 3). According to the given time frame constraints [Kroeger, 2004; ten Veen and

Figure 3. Lithostratigraphy, chronostratigraphy, and age model of coral levels, late Miocene, western Iraklion Basin, Crete. Eustatic sea level and timescale are from Hardenbol et al. [1998]. 3 of 12


PA4217

Table 1. Cross Table Showing Results of One-Way Analysis of Variancea Coral Site 1.1 1.2 1.3 2.1 3.1 3.2 4.1 4.2 4.3 5.1 5.2 5.3 6.1 7.1 7.2 7.3 1.1 1.2 1.3 2.1 3.1 3.2 4.1 4.2 4.3 5.1 5.2 5.3 6.1 7.1 7.2 7.3

NS S 0 0 0 0 0 0 0 0 0 0 0 0 0

S 0 0 0 0 0 0 0 0 0 0 0 0 0

S 0 0 0 0 0 0 0 0 0 0 0 0

NS 0 0 0 0 0 0 0 0 0 0 0

NS 0 0 0 0 0 0 0 0 0 0

NS 0 0 0 0 0 0 0 0 0

S NS NS S S 0 0 0 0

NS S NS NS 0 0 0 0

NS S NS 0 0 0 0

S S NS 0 0 0

S S 0 0 0

S 0 0 0

S 0 0

NS 0 NS -

a NS indicates no significant difference; S indicates significant difference (p < 0.05); and 0 indicates not analyzed.

Postma, 1999], these larger-scale sequences are a reflection of third-order eustatic changes [Reuter et al., 2006] that represent a suitable age model for coral growth rate analyses. Coral levels 1 – 3 were assigned to the early Tortonian sea level cycle, coral levels 4– 6 to the late Tortonian, and coral level 7 to the early Messinian eustatic cycle [Haq et al., 1988; Hardenbol et al., 1998] (Figure 3). [7] Systematic assessments of average skeletal growth rates in modern Indo-Pacific Porites have demonstrated a significant relationship between coral calcification and linear extension rate with average sea surface temperatures (SSTs) [Highsmith, 1979; Jokiel and Coles, 1990]. These relationships were quantified for shallow water ( 0.05) against the alternate hypothesis H1 that there are significant differences (p < 0.05) (Table 1). SSTs were calculated from these average extension rates using the linear regression of modern Porites form the Indo-Pacific [Lough and Barnes, 2000]. Variation in SSTs not accounted for by the best fit line is given by the 95% confidence intervals, both for the intercept and slope of linear regression.

2. Results 2.1. Coral Environments [9] Coral outcrops were selected in the Asites-Kroussonas area of the western Iraklion basin (Figure 2 and Table S1 in auxiliary material). Excellent outcrop conditions in this area allow an exact calibration of the coral levels in the chronostratigraphic scheme. A detailed account of depositional geometries and lithologies of the geological transects studied has been given by Reuter et al. [2006]. Schematic representations of stratal geometries for the coral sites are given in auxiliary material Figure S1. Except for small

1 Auxiliary material data sets are available at ftp://ftp.agu.org/apend/pa/ 2006pa001288. Other auxiliary material files are in the HTML.

4 of 12

PA4217


Figure 4. Photographs of annual bands in massive Porites. Arrows delineate growth increments. (a) X-ray radiograph (positive) from specimen in a pristine type of preservation showing annual density bands: Site 1.3, early Tortonian. (b) Field photograph of recrystallized massive Porites showing rhythmic bands interpreted to represent ghost structures of annual growth increments: Site 4.2, late Tortonian. The mode of transformation of the skeleton has been described in detail by Reuter et al. [2005].

fragments, no in situ coral fauna or reef is present in the coastal conglomerates of coral level 1. To cover the earliest Tortonian period, additional outcrops were chosen from the Ambelouzos-Apomarma area (northern Messara Basin; Figure 2). Outcrop conditions in this area were not good enough to document stratal geometries. All coral sites from this area used here that interfinger with coastal conglomerates were combined into one single unit equivalent with coral level 1 (Figure 3). Such a classification is consistent with the Sr chronostratigraphy of central Crete [Kroeger, 2004]. 2.1.1. Corals of Coral Level 1 [10] In the Ambelouzos-Apomarma area (Figure 2), the lower Ambelouzos Formation exhibits spectacular outcrops of coral biostromes and reefs. Dense stands of columnar or massive Porites (0.5 m high) with minor contents of Tarbellastrea, Acanthastrea and Siderastrea form biostromes of meter thickness that are laterally continuous over hundreds of meters. In Psalidha, the biostromes are stacked vertically to form a low-mounded coral buildup (sites 1.1, 1.2, 1.3) [Reuter et al., 2005, 2006]. 2.1.2. Corals From Coral Levels 2 – 7 [11] In the Asites-Kroussonas area of central Crete (Figure 2) patch reefs predominate. Nearshore reefs formed on ramps in a mixed carbonate– clastic environment. Reef frameworks result from mutual coral-coral incrustations of massive hemispherical Porites and Tarbellastrea (decimeter to meter size) with some patchy intergrowths of Porites sticks (1 m). Binding by coralline algae or boring by sponges and bivalves is insignificant. Colonies with an inverted or oblique orientation, interpreted as reworked, are rare (sites 2.1, 3.1, 3.2, 4.3, 5.2). Nonetheless, some isolated heads of corals (Porites, Tarbellastrea) occur basinward from nearshore reefs in a poorly classified, muddy fossiliferous sandstone. Here the massive corals exhibit ragged outlines and complex intergrowth with sediment,

PA4217

nongeniculate coralline algae and encrusting bryozoa. They are considered to represent a parautochthonous level bottom coral association in the muddy offshore zone (site 4.1). [12] On top of uplifted fault blocks located within Iraklion Basin, the Ambelouzos Formation is covered by limestone of the Pirgos Member or is absent. Coralline red algae with additions of bryozoans, bivalves and larger foraminifers are the principal constituents of the limestone. Locally, in a dip slope ramp setting, nonconstructional coral associations of Porites and Tarbellastrea may be present (site 5.1). In a few locations, positioned along fault scarps of tilted blocks, the red algal facies changes into reefal talus (site 4.2) and coral bioherms (site 5.3) [Reuter et al., 2006]. [13] On top of the Ambelouzos Formation and Pirgos Member is the Varvara Formation. It is formed of rhythmically bedded light grey marl and limestone, grading into a unit of laminated gypsum. At the base of the Varvara Formation some nonframework coral associations occur (site 6.1). Up section, in proximal settings (Moni Gorgolaini) the Varvara Formation contains massive beds with angular lithic and skeletal fragments supported by a fine-grained matrix (debrites), and graded calcarenites (calciturbidites). The debrites contain decimeter-sized angular fragments of massive Porites and Tarbellastrea, thus documenting reef growth in an up-slope direction in shallow water (sites 7.0, 7.1, 7.2) [Reuter et al., 2006]. 2.2. Growth Banding and Vertical Extension Rates in Late Miocene Porites [14] Preservation of the original aragonite mineralogy and microstructures is rather unusual in the geological record of scleractinian corals [Constantz, 1986; Dullo, 1984; Dullo and Mehl, 1989; Reuter et al., 2005]. In central Crete, it is restricted to the uppermost reef level in Psalidha outcrop and to debrites in the Moni Gorgolaini outcrop (Figure 2 and Table S1 of auxiliary material). The vast majority of corals from the Ambelouzos and Varvara Formations are either leached or transformed and recrystallized into coarse blocky calcite spar. This is the typical state of preservation of zooxanthellate corals from Neogene carbonate sediments [Martıń et al., 1989]. However, preservation of corals transitional between original aragonite and secondary calcite reveals that conspicuous banding in recrystallized corals represents ghost structures of the original annual density bands [Reuter et al., 2005]. This type of preservation is common in massive Porites from the late Miocene of Crete and allows reliable measurements of extension rates of corals in the field (Figure 4). [15] Coral extension rates were measured at 16 sites (n = 93) assigned to seven chronostratigraphic coral levels covering the Tortonian and early Messinian time periods (Figure 3 and Table 1). In coral level 1, measured corals derive from biostromes. Mean extension rates were 2.1 to 2.6 mm yr1 with a maximum of 3.8 mm yr1. With a mean of 4.3 mm yr1 and a maximum of 5.2 mm yr1, extension rates encountered were substantially higher in the stratigraphically topmost biostrome of the AmbelouzosApomarma area (coral level 1).

5 of 12

PA4217


PA4217

paleogeography of the basin suggests a reefal source [Reuter et al., 2006].

3. Discussion

Figure 5. Extension rates of massive hemispherical Porites in coral levels 1 through 7 (Tortonian, early Messinian). Solid symbols indicate framework-forming Porites; open symbols indicate nonframework-forming Porites. Arithmetic mean with plus/minus standard deviation is given. N = 93.

[16] In coral levels 2 and 3, we analyzed Porites from small patch reefs within clastic nearshore environments that exhibit low extension rates of 1.8 –2.3 mm yr1 (2.3 – 3.0 mm yr1 maximum). Low extension rates of massive Porites (mean 2.2 mm yr1, maximum 2.3 mm yr1) were also encountered in a level bottom coral association of a nearshore environment of level 4. In coral reefs from the same stratigraphic level, in positions proximal to the paleocoast and on an uplifted block in an offshore position, extension rates were 2.9 and 3.5 mm yr1 respectively (maxima 4.9 and 4.4 mm yr1). In coral level 5, there were again significant differences between massive Porites from reefs and level bottom communities: Extension rates were comparatively low in isolated Porites from a level bottom community (mean = 2.3 mm yr1, maximum = 2.8 mm yr1), and high in coral reefs from both, a nearshore clastic environment and an offshore carbonate environment. In the former, mean extension rates were 4.5 mm yr1 (maximum = 5.0 mm yr1), in the latter 3.1 mm yr1 (maximum = 3.3 mm yr1). In coral level 6, we were unable to document the nearshore zone and a coral reef environment, probably because of more recent erosion of the Neogene sediments. Nonetheless, massive Porites from a level bottom community in a distal paleogeographic setting show low mean annual extension rates of 1.8 mm yr1 (maximum is 2.3 mm yr1), as in the older level bottom communities (Figure 5). Extension rates of corals recovered from three successive debris flow deposits of coral level 7 of early Messinian age averaged 4.0 to 4.3 mm yr1 (maximum 4.3 to 4.4 mm yr1) respectively. The source area for the debrites has been destroyed by subrecent erosion, however, the overall

3.1. Paleoenvironmental Significance of the Annual Coral Growth Increments [17] The rock record of late Miocene coral environments in Crete displays mostly corals preserved in situ, both in the attached mixed clastic-carbonate systems and isolated carbonate environments of the central basin, whereas reworked coral material is subordinate and was encountered associated with steep submarine scarps (Figure 6). In growth framework pores and peripheral to the reefs, reef debris is subordinate. The geological record of the coral communities within a single reef therefore does not appear to be strongly biased by incomplete preservation [Reuter et al., 2006], and, given the preserved mutual coral incrustations and small thickness, the reef outcrops must cover a short time interval lasting some thousands of years [cf. Geister, 1989; Shinn, 2001]. Correspondingly, measured extension rates were assumed to be representative of a given outcrop and paleoenvironment. [18] During the late Tortonian (coral level 4 to 5) coral reefs and level bottom coral communities occurred both in mixed clastic carbonate and pure carbonate environments (Figure 6). Extension rates in massive Porites were not significantly different between reef sites and between sites analyzed in level bottom communities (Table 1). However, extension rates in level bottom communities were significantly lower by 1 mm yr1 compared to reefal communities (Figure 6 and Table 1). Low extension rates in the level bottom corals (2.2 and 2.3 mm yr1 per site) can be attributed therefore to low-light conditions (water depth, turbidity) and not to low SSTs (Figure 6). A dominant water depth effect on extension rate is plausible for site 5.1 situated on a distal carbonate ramp (Figure 6) and is analogous to the relationship between mean extension rates of Porites and depth in modern high-latitude reefs (Bermuda [Logan and Tomascik, 1991] and Red Sea [Heiss, 1994]). Surprisingly, in coral level 5 extension rates of massive Porites were significantly higher in coeval reefs from the clastic nearshore environments than those in pure carbonate environments located offshore (Figure 6 and Table 1). Consequently, clastic turbidity in the nearshore zone did not have a negative effect on extension rates of the corals, and/or high nutrient availability in the nearshore zone appears to have stimulated coral growth [cf. Barnes and Lough, 1993; Lough and Barnes, 2000] (Figure 6). 3.2. Growth Increment Variability in Porites During the Late Miocene [19] Annual growth patterns of massive hemispherical Porites documented for seven coral levels (CLs) representing the Tortonian and early Messinian (late Miocene) exhibit surprisingly little variation (Figure 5). Nonetheless, a significant positive excursion (4 mm yr1) exists at the topmost site from CL 1 (site 1.3; Table 1). Porites from lower levels of CL1 and from levels CL2 to CL3 remain remarkably constant and similar at 2 mm yr1. Annual extension rates were also significantly higher compared to

6 of 12

PA4217


PA4217

Figure 6. Annual extension rates of massive hemispherical Porites spp. during late Tortonian coral levels 4 – 5, central Crete. (top) Solid symbol indicates reef community; open symbol indicates level bottom community. (bottom) Numbers refer to coral sites. Arithmetic mean with plus/minus standard deviation is given. N = 32. the lower levels at site 4.2 (3.5 mm yr1; CL 4), site 5.2 (4.5 mm yr1; CL 5), and all sites from CL 7 ( 4 mm yr1) (Table 1). Annual extension rates encountered in CL 6 were lower than in earlier and later levels (2 mm yr1). However, these data derive from a level bottom community, and therefore potentially do not record maximum extension rates for this chronostratigraphic unit (Figure 5 and Table 1).

extension rate and SST, rather than annual maximum SST [Lough and Barnes, 2000]. [21] The linear regressions between linear annual extension rate and (1) for minimum SSTs and (2) for maximum SSTs are:

3.3. Porites Thermometry [20] Modern Porites are cosmopolitan in the tropics [Veron, 1993]. Atlantic massive Porites astreoides and related taxa are typically small with encrusting or columnar to branching growth habits and are not widely used in coral growth studies [Harriot, 1999; Logan, 1988; Logan and Tomascik, 1991]. Although a subgeneric classification of Indo-Pacific Porites is difficult [Veron, 1995], the more massive and hemispherical growth forms of Indo-Pacific Porites spp. have stimulated extensive work on their growth characteristics [i.e., Grigg, 1982; Veron and Minchin, 1992]. In a synoptic study covering a variety of shallow water locations of the Indo-Pacific, calcification rates and linear extension rates of massive hemispherical Porites spp. were found to be principally a function of SST [Lough and Barnes, 2000] (Figure 7). For average annual SST, average annual extension rates change by 3.1 mm yr1 per 1°C change (over a range of 22°– 29°C) [Lough and Barnes, 2000]. This relationship can therefore be used as a paleothermometer to estimate SST from a given extension rate of massive Porites spp. Annual average minimum SST was the main contributor to the relationship between average annual

ð1Þ

SST½ Cminimum ¼ ð0:33 0:06Þ G þ ð19:35 0:68Þ

SST½ Cmaximum ¼ ð0:18 0:05Þ G þ ð26:28 0:60Þ ð2Þ where G is the vertical extension rate measured in mm yr1. These relationships explain 77% (r = 0.88, P < 0.001) in (1) and 57% (r = 0.76, P < 0.001) in (2) of the variation in SST, respectively. [22] To test predicted SSTs using of the Porites thermometer for the late Miocene, we used a stable oxygen isotope time series (d 18O) with a quarterly resolution from a massive Porites from site 1.3 [Brachert et al., 2006]. From an average extension rate of 4.7 mm yr1 we infer average SSTsminimum = 20.9 ± 1°C (equation (1)) and average SSTsmaximum = 27.2 ± 0.9°C (equation (2)) with a seasonality of 6.3°C (Figure 8). Mean d 18O seasonality in the coral record is 1.14%. Assuming no seasonal changes in d18Oseawater, this variability is equivalent to 6.3°– 7.5°C as

7 of 12

PA4217


PA4217

Figure 7. Porites thermometer: scatterplots and linear regression of relationship between average sea surface temperature (