Coral Records Of Past Climate

The term "coral" is generally applied to members of the order Scleractinia, which have hard calcareous skeletons supporting softer tissues (Wood, 1983; Veron, 1993). For paleoclimatic studies, the important subgroup is the reef-building, massive corals in which the coral polyp lives symbiotically with unicellular algae (zoo-xanthellae); these are known as hermatypic corals (as opposed to ahermatypic, which contain no algal symbionts and are not reef-builders). The algae produce carbohydrates by photosynthesis and thus are affected by water depth (most growing between 0-20 m) as well as water turbidity and cloudiness. Much of the organic carbon fixed by the algae diffuses from the algal cells, providing food for the coral polyps, which in turn provide a protective environment for the algae. Reef-building corals are limited mainly by temperature and most are found within the 20 °C mean sea-surface temperature (SST) isotherm (generally between 30° N and 30° S). When temperatures fall to 18 °C, the rate of calcification (skeletal growth) is significantly reduced and lower temperatures may lead to death of the colony.

Coral growth rates vary over the course of a year; when sectioned and x-rayed, an alteration of high- and low-density bands can be seen (Fig. 6.39). High density layers are produced during times of highest SSTs (Fairbanks and Dodge, 1979; Lough and Barnes, 1990) providing a chronological framework for subsequent analyses. Dating in this way is fairly accurate as shown by the close correspondence between large excursions in 8isO and known El Nino-Southern Oscillation (ENSO) events (Cole el al., 1992) and by very close similarity between annual counts and precise 230Th dates on individual bands (Dunbar et al., 1994). In some regions, exceptional runoff events from adjacent continents are recorded by fluorescent bands in corals (visible under UV light) and provide a further cross-check on coral chronologies (Isdale, 1984; Isdale et al., 1998). These bands result from terrestrially derived fulvic acid being incorporated into the coral structure (Boto and Isdale, 1985). Such banding could also provide valuable estimates of the recurrence interval of extreme river discharge in certain regions.

Samples for analysis are generally drilled from the coral section at regularly spaced intervals along the coral growth axis. Assigning the precise seasonal time to each sample is problematic and is usually done by assuming a linear growth rate between the denser marker bands, the edge of which is assigned to the "onset" of high SSTs. However, if coral extension (growth) is nonlinear, very detailed sampling (6-10 samples per yr) is required or the samples may not cover the entire seasonal range, and provide only minimum estimates on total interannual variability. Furthermore, under extreme conditions (often associated with major El Nino-Southern Oscillation [ENSO] events) coral growth in some areas may even cease, so that the real extremes may go unrecorded in corals from those regions.

Coral studies have focused mainly on the environmental record in coral growth rates, isotopes, and trace elements. This has led to new information about paleo-SSTs, rainfall, river runoff, ocean circulation, and tropical wind systems. Many studies have been based on relatively short periods of time (the last few decades) to provide a better understanding or calibration of the parameter being analyzed, thereby increasing confidence in the paleoreconstructions. So far, only a few studies spanning more than

Coral Isotop Growth
FIGURE 6.39 Positive x-ray photographs of slabs of coral (Pontes lutea) from off the coast of Kenya showing annual banding.The slabs cover the period from 1994 (top left) back to < 1700 (bottom right). (Photograph kindly provided by Rob Dunbar).
HH TABLE 6.5 Long Coral-based Records of Past Climate

Site

Latitude

Longitude

Record length

Parameter

Indicator of:

Reference

Bermuda

32° N

65° W

-1180-1986

Growth rate

SST/upwelling

Patzold and Wefer, 1992

Cebu Island Philippines

10° N

124° E

-1860-1980

8180 813C

SST and rainfall/ cloudiness

Patzold, 1986

Gulf of Chiriqui Panama

8° N

82° W

1707-1984

8lsO

Rainfall/ ITCZ position

Linsley et al., 1994

Tarawa Atoll Kiribati

1° N

172° E

1893-1989

s18o

Rainfall

Cole et al., 1993

Isabela Island Galapagos Islands

0.4° S

91° W

1587-1953

5180

SST

Dunbar et al., 1994

Espiritu Santo Vanuatu

15° S

167° E

1806-1979

8180 813C

SST and rainfall/ cloudiness

Quinn et al., 1993

Great Barrier Reef Australia

22° S

153° E

1635-1957

A14C

Oceanic advection and/or upwelling

Druffel and Griffin, 1993

New Caledonia

22° S

166° E

1655-1990

8lsO

SST

Quinn et al., 1996

a century have been published (Table 6.5), but many more records are likely to be produced in the years ahead. Indeed, it is likely that the tropical oceans, having been almost totally unrepresented by high resolution paleoclimatic records in the past, may soon provide some of the best records, especially for the last few centuries (Dunbar and Cole, 1993). Furthermore, corals from raised marine terraces are found throughout the Tropics, some of which date back to the last interglacial, or even earlier. Providing diagenetic changes in the coral aragonite have not occurred (Bar-Matthews et al., 1993) it may be possible to reconstruct SSTs (and annual variations in SSTs) for selected intervals over the last 130,000 yr or more (Beck et al., 1992, 1997).

6.8.1 Paleoclimate from Coral Growth Rates

Coral growth rates are dependent on a variety of factors, including SSTs and nutrient availability (Lough et al., 1996). The longest record of coral growth rate variations is that of Patzold and Wefer (1992), who produced an 800-yr record from a massive coral head (Montastrea cavernosa) in Bermuda. In this region, growth rates are inversely related to SST, as cool upwelling water is nutrient-rich, which causes increased coral growth. The record shows that SSTs were generally above the long-term mean from -1250-1470. Coolest conditions were experienced from -14701710 and from -1760 to the end of the nineteenth century, followed by twentieth century warming. This is broadly similar to estimates of northern hemisphere summer temperature change over this period (Bradley and Jones, 1993). By contrast, in the Galapagos Islands coral growth rates generally increase with SSTs; growth in creased from -1600 to the 1860s, then declined, reaching the lowest rates from -1903-1940 (Dunbar et al., 1994). However, this record bears little relation to the 8lsO record of SSTs in the same corals, suggesting that in this area other factors are probably involved in growth rate besides water temperature.

6.8.2 8I80 in Corals

It has long been known that a temperature-dependent fractionation of oxygen isotopes occurs when biological carbonate is precipitated from solution (Epstein et al., 1953). The S180 decreases by ~0.22%o for each 1 °C increase in temperature. Seasonal variations in 5180 along the growth axis of a coral and their relationship to seasonal SST variations were first reported by Fairbanks and Dodge (1979). Subsequently, Dunbar and Wellington (1981) also showed that, if finely sampled, corals can provide an intra-annual record of SST changes (Fig. 6.40c). Offsets from predicted equilibrium values may be caused by vital effects (see Section 6.3.1) but these are constant for a given genus (Weber and Woodhead, 1972).

24 to to

1986 1987 1988 1989 1990 1991 1992 1993 Year

FIGURE 6.40 Comparison between recorded SSTs (3-week averages of daily temperatures) at a tidal station on the Ryukyu Islands, Japan (dashed lines) and geochemical variations in a nearby coral: A) Mg/Ca ratios (r = 0.92); B) Sr/Ca ratios (r = 0.85); C) 8lsO (r = 0.88).Analytical error bars are indicated; these amount to ±0.5, ±1.6, and ±0.4 °C, respectively (Mitsuguchi etal., 1996).

In those areas of the Tropics where seasonal changes in the isotopic composition of seawater occur, a simple SST-8180 relationship is not found. In areas with seasonally heavy rainfall, which is depleted in 8180 during convective activity, the ocean surface mixed layer becomes isotopically light during the wet season, producing a pronounced seasonal signal in coral 8180 (Cole and Fairbanks, 1990; Linsley et al., 1994). In some regions, this effect is brought about, or enhanced, by flooding of near-coastal waters by isotopically light river water discharged from the continents (Mc-Culloch et al., 1994). Conversely during prolonged hot, dry conditions, surface evaporation can increase sea-surface salinity (SSS) and lead to isotopic enrichment (more 8lsO) due to the preferential removal of 160. To avoid these complications, most studies either focus on areas with large annual changes in rainfall (Cole et al., 1993; Linsley et al., 1994) or on areas with little change in SSS, but large SST changes (Dunbar et al., 1994). For example, in parts of the western Pacific El Niños are associated with unusually heavy rainfall. At Tarawa Atoll (Io N, 172° E) negative 8lsO excursions of 0.6 ± 0.1 %o occur during ENSO events as a result of dilution of the mixed layer by isotopically depleted rainfall (Cole and Fairbanks, 1990). In this region, the anomalies provide a diagnostic signal of ENSO events over the past century. By identifying the appropriate ENSO signal in different parts of the Pacific Ocean, it should be possible to reconstruct the spatial and temporal characteristics of ENSO events (both "warm" and "cold") far back in time (Cole et al., 1992).

In those areas that experience extreme SST anomalies during ENSO events, 8180 in corals may provide a unique record of such occurrences (Carriquiry et al., 1994). Using a 8lsO record in the coral Pavona clavus from the Galapagos Islands, Dunbar et al. (1994) reconstructed SST variations over the past 380 yr (Fig. 6.41). This record shows that, of the the 100 largest negative 8lsO anomalies over the last 350 yr (indicating extremely high SSTs in the region) 88 corresponded

1600

1650

1700

1750 1800 Year A. D.

1850

1900

1950

2000

FIGURE 6.41 Annual 8lsO values from a specimen of Pavona clavus, a coral from the Galapagos Islands expressed as departures from the long-term mean.The standard deviation of the data is 0.07%o. Lower figure shows the data filtered by a 5-point running mean to emphasize lower frequency variations. Sea-surface paleotempera-ture estimates are given on the right-hand axis (Dunbar et al., 1996, Dunbar et al., 1994).

1600

1650

1700

1750 1800 Year A. D.

1850

1900

1950

2000

FIGURE 6.41 Annual 8lsO values from a specimen of Pavona clavus, a coral from the Galapagos Islands expressed as departures from the long-term mean.The standard deviation of the data is 0.07%o. Lower figure shows the data filtered by a 5-point running mean to emphasize lower frequency variations. Sea-surface paleotempera-ture estimates are given on the right-hand axis (Dunbar et al., 1996, Dunbar et al., 1994).

(±1 yr) to Quinn's (1992) chronology of El Niño events, derived from historical sources. They then examined the changing pattern of dominant perodicities in the record using evolutionary spectral analysis. This involves applying spectral analysis to the data sequentially, in overlapping intervals of time (in this case, 120 yr intervals, from 1610-1730 to 1862-1982) in order to map out the time/frequency response of SSTs in this area (Fig. 6.42). The analysis reveals several shifts in the dominant frequency modes; in the early 1700s, the quasi-periodic El Niño events shifted from the 4.6-7-yr band to 4.6 and 3 yr. A second shift occurred in the mid-1800s to a predominant period around 3.5 years. Similarly timed changes in the dominant lower frequency variance are also seen, especially in the mid-1800s, from -33 to -17 yr. It is interesting that a pronounced change (towards warmer and/or dryer conditions) also occurred in the mid 1800s (around 1866) in the South Pacific, as recorded in a coral record from Vanuatu (southwestern Pacific) (Quinn et al., 1993). Whether such changes are coincidental or represent major reorganizations of the tropical ocean-atmosphere climate system (as Dunbar et al., [1994] suggest) will become more apparent as new coral records are developed from throughout the tropical oceans.

FIGURE 6.42 Evolutionary spectral diagram of the Galapagos coral record shown in Fig. 6.41.The diagram shows spectral density as a function of time (x-axis) and frequency (y-axis) based on analysis carried out in 120-yr segments, each offset by 10 yr. Shaded areas correspond to frequencies at which significant variance occurs, the darker the shading the greater the statistical significance. Lowest frequencies (secular trends) are shown in the upper part of the diagram, and increasingly higher frequencies (shorter periods — see right-hand axis) are shown in the lower part of the diagram.The record indicates a shift to higher frequencies occurred around A.D. 1750 and again around A.D. 1850 (Dunbar et al., 1994).

Spectral density

n oto i

Centerpoint Year (A.D.)

FIGURE 6.42 Evolutionary spectral diagram of the Galapagos coral record shown in Fig. 6.41.The diagram shows spectral density as a function of time (x-axis) and frequency (y-axis) based on analysis carried out in 120-yr segments, each offset by 10 yr. Shaded areas correspond to frequencies at which significant variance occurs, the darker the shading the greater the statistical significance. Lowest frequencies (secular trends) are shown in the upper part of the diagram, and increasingly higher frequencies (shorter periods — see right-hand axis) are shown in the lower part of the diagram.The record indicates a shift to higher frequencies occurred around A.D. 1750 and again around A.D. 1850 (Dunbar et al., 1994).

6.8.3 8I3C in Corals

The 813C in coralline carbonates is affected by a variety of factors, including the 513C of seawater (related, in part, to the relative contributions of surface and upwelled waters) and fractionation of carbon isotopes during algal photosynthesis. Algae preferentially take up 12C from dissolved inorganic carbon (DIC) in ocean waters, so higher rates of photosynthesis lead to DIC becoming enriched in 13C (less negative 813C), which in turn affects the 813C of the skeletal carbonate being constructed (McConnaughey, 1989). Several studies have shown that 813C declines with water depth (Fairbanks and Dodge, 1979) and during cloudy months (i.e., as photosynthesis rates are reduced) (Shen et al., 1992a; Quinn et al., 1993), suggesting 813C in coral bands may provide a long-term index of cloudiness. However, complications related to coral geometry (varying growth rates and photosynthetic activity around a coral head) and possible nonlinear photosynthetic responses to changing light levels, have generally assigned 813C records a back seat to dlsO in stable isotope paleoclimatic reconstructions from corals.

6.8.4 AI4C in Corals

Changes in oceanic mixed layer A14C (that is 14C anomalies from long-term trends) are related to either changes in atmospheric 14C levels or to upwelling of 14C-depleted waters from the deep ocean. The A14C anomalies are also recorded in tree rings and, therefore, changes observed in corals, which are not seen in tree rings, are presumably related to changes in oceanic circulation, indicating either coral up-welling or advection of 14C depleted (or enriched) waters from other regions. Thus, Druffel and Griffin (1993) related unusually large excursions of A14C values in corals from the southwestern Great Barrier Reef between 1680 and 1730 to changes in the relative contributions of waters from the South Equatorial Current (A14C~ -60%o) and the East Australia current (A14C~ -38%o).

6.8.5 Trace Elements in Corals

Because certain elements (Sr, Ba, Mn, Cd, Mg) are chemically similar to Ca, trace amounts of these elements may be found in coral skeletal carbonate. Many studies have shown that the relative concentration of such elements (expressed as the ratio of the trace element to calcium) often provides a paleoclimatic, or paleoceano-graphic signal (Shen and Sanford, 1990). For example, because Cd levels are generally much higher below the mixed layer, Cd/Ca ratios in Galapagos corals increase in association with seasonal upwelling (Shen et al., 1987). The Ba/Ca ratios are inversely related to SSTs (Lea et al., 1989) so low Cd/Ca and Ba/Ca ratios provide a useful index of El Niño events (in the Galapagos area) because such events are associated with very high SSTs and minimal upwelling. The Mn/Ca ratios also provide valuable information in some regions; for example, in the west-central Pacific, Mn is remobilized from lagoonal sediments during strong episodes of equatorial westerly winds, (associated with El Niños) and thus large Mn/Ca ratios in corals are indicative of such conditions (Shen et al., 1992a, b). Elsewhere, Mn/Ca (and perhaps also Ba/Ca) ratios may provide information on runoff from continental regions because terrestrial material is rich in Mn and Ba compared to ambient levels in the oceanic mixed layer.

Other paleotemperature indicators are provided by Sr/Ca, U/Ca and Mg/Ca ratios in corals (see Fig. 6.40a, b) (McCulloch et al., 1994; Min et al., 1995; Mitsuguchi et al., 1996). This opens up the prospect of using multiple parameters to reconstruct paleotemperatures in both recent and fossil corals with high accuracy. However, recent studies by de Villiers et al. (1995) indicate that estimates based on Sr/Ca may be in error by several degrees. Sr/Ca (and perhaps Mg/Ca) ratios are very dependent on coral growth rate, leading to lower paleotemperature estimates in coral sections with low growth rates compared to those derived from faster-growing sections of the same coral. If such problems can be resolved, perhaps by a combination of growth rate measurements and the analysis of Sr/Ca, as well as Mg/Ca and/or U/Ca ratios, it would be of great value in helping to resolve controversial tropical SST estimates from the last glacial period. So far, a number of coral studies point to much lower tropical SSTs in glacial time than other oceanic paleotemperature indicators (Table 6.4). For example, Beck et al. (1992) estimated paleotemperatures from Sr/Ca in corals from a paleoreef on Espiritu Santo, Vanuatu that were dated -10,000 B.P. They calculated that SSTs were -5.5 °C cooler than today, similar to the results of Min et al. (1995), who used U/Ca to estimate a temperature difference of 4-5 °C (LGM today) in the same area. Studies of Sr/Ca in corals from Barbados also indicate SSTs were -5 °C lower at the LGM compared to today (Guilderson et al., 1994). These results stand in stark contrast to both foram- and alkenone-based SST reconstructions (Table 6.4); determining which approach provides the correct answer (or if all are correct with respect to the actual temperature being recorded) is a key issue in paleoclimatology today.

Was this article helpful?

+1 0

Responses

  • Leonie
    How are corals a part of paleoclimatology?
    3 years ago

Post a comment