Speleothems

Speleothems are mineral formations occurring in limestone caves, most commonly as stalagmites and stalactites, or slab-like deposits known as flowstones. They are composed primarily of calcium carbonate, precipitated from ground water that has percolated through the adjacent carbonate host rock. Certain trace elements may also be present (often giving the deposit a characteristic color); one of these, uranium, can be used to determine the age of a speleothem, as discussed in what follows. Deposition of a speleothem may result from evaporation of water or by degassing of carbon dioxide from water droplets. Evaporation is normally only an important process near cave entrances; most speleothems therefore result from the degassing process. Water that has percolated through soil and been in contact with decaying organic matter usually accrues a partial pressure of carbon dioxide exceeding that of the cave atmosphere. Thus, when water enters the cave, degassing of carbon dioxide occurs, causing the water to become supersaturated with calcite, which is thus precipitated (Atkinson et al., 1978).

The deposition of speleothems is dependent on a number of factors — geological, hydrological, chemical, and climatic. A change in any one of these factors could cause water percolation to cease, terminating speleothem growth at a particular drip site. However, cessation of speleothem growth over a large geographical area is more likely to be due to a climatic factor than anything else, so dating periods of speleothem growth can provide useful paleoclimatic information (Harmon et al., 1977). Uninterrupted speleothem growth is recognizable in a polished section as a series of very fine growth layers; major hiatuses in deposition are usually marked by erosional surfaces, desiccation and chalkification, dirt bands, and sometimes color changes. In speleothems deposited close to sea level, a rise in relative sea level may result in an overgrowth of marine aragonite on the deposit.

Speleothems grow in sheltered environments that often have escaped the radical surface alterations resulting from glaciation. For example, speleothems dated at up to 350,000 yr B.P. are found beneath the present-day Columbia Icefield of Alberta, whereas the geomorphology of the surface has been repeatedly altered by glacial events. Speleothems may thus provide a long, often continuous record of past environmental conditions. Furthermore, the extensive distribution of karst landscapes (Fig. 7.28) means that studies can be undertaken on a worldwide basis.

Paleoclimatic studies have focused on the timing of speleothem growth periods, their isotopic composition (of both the minerals and fluid inclusions) and their relationship to sea-level fluctuations (Gascoyne, 1992). These are discussed separately in the following sections. Pollen extracted from speleothems may also provide a record of regional vegetation change related to past climatic conditions (Burney et al., 1994).

7.8.2 Dating of Speleothems and the Significance of Depositional Intervals

Speleothems are most commonly dated by uranium-series disequilibrium methods (generally 230Th/234U) described in Chapter 3. Isotopes of uranium leached from the carbonate bedrock are co-precipitated as uranyl carbonate with the calcite of the speleothems. Normally, the precipitating solution contains no 230Th because thorium ions are either adsorbed onto clay minerals or remain in place as insoluble hydrolysates (Harmon et al., 1975). Thus, providing the speleothem contains no clay or other insoluble detritus, which are carriers of detrital thorium, the activity

FIGURE 7.28 Distribution of karst in the world, showing the potential sources of paleoclimatic information from speleothems.

7.8.1 Paleoclimatic Information from Periods of Speleothem Growth

World Karst Distribution

FIGURE 7.28 Distribution of karst in the world, showing the potential sources of paleoclimatic information from speleothems.

ratio of 234U to its decay product 230Th will give the sample age (Harmon et al., 1975). The method is useful over the time range 350,000-10,000 yr B.P. A number of precautions are taken to ensure a reliable age estimate, most notably that any samples containing more than 1% of acid-insoluble detritus are rejected. Also, any indication that recrystallization has occurred (suggesting that the sample may not have remained a closed system) would cause it to be rejected. Recent developments in U-series dating by thermal ionization mass spectrometry (TIMS) have meant that smaller samples (3-5 g of calcite) can be used, yielding more precise dates (a 1 o precision of <1%) and extending the potentially useful dating horizon to >400,000 yr (Edwards et al., 1987a; Li et al., 1989). The TIMS dating also opens up the prospect of high-resolution studies of samples spanning episodes of rapid environmental change (Baker et al., 1995; Goede et al., 1996). However, the temporal limit of such studies is limited by the residence time of water in the aquifers, which link surface climatic conditions to subsurface speleothem growth (Schwarcz, 1996). This effectively acts as a lowpass filter on the environmental record in speleothems, but if flow rates are high enough annual or even subannual variations may be resolved (Baker et al., 1993; Shopov et al., 1994).

Dating the onset and termination of speleothem growth, using samples from a wide area enables regional chronologies to be built up and may indicate large-scale (climatically related) controls on periods of speleothem growth (Hennig et al., 1983). This is most successful when dating samples from subalpine or subarctic sites that are currently marginal for speleothem growth. During glacial periods, colder conditions, less snowmelt, and more extensive permafrost would result in a marked reduction, even a cessation, of groundwater percolation. Furthermore, decreased biotic activity would lead to a decrease in the partial pressure of carbon dioxide in the soil atmosphere, and therefore less carbonate in solution; consequently speleothem growth might cease (Harmon et al., 1977; Atkinson et al., 1978). Uranium-series dates on alpine speleothems from western Canada, collected at sites that currently have a surface mean annual temperature close to 0 °C and are hence marginal for speleothem growth today reveal four periods of speleothem deposition. These are assumed to represent interglacials, when conditions for speleothem growth were most favorable. Periods of relatively warm climate occurred from -320,000 to 285,000 yr B.P., from -235,000 to 185,000 yr B.P., from 150,000 to 90,000 yr B.P., and from -15,000 yr B.P. to the present (Harmon et al., 1977). In addition, a brief interval, possibly an interstadial, was identified at -60,000 yr B.P. The earliest period may, in fact, have begun >350,000 yr B.P., representing a long warm interval of massive spleothem deposition (Harmon, 1976) possibly correlative with the prolonged interglacial episode from 460-560,000 yr B.P. represented in China by a very thick paleosol Ss (An et al., 1987).

These dates compare favorably with periods of speleothem growth observed in caves in the North of England from which a large number of U-series dates have been obtained (Fig. 7.29) (Gascoyne et al., 1983). Speleothems formed extensively during the periods 130-90 ka B.P. and from 15 ka B.P. to the present (Atkinson et al., 1978, 1986a; Gordon et al., 1989). Deposition appears to have ceased entirely from 165-140 ka B.P. and from 30-15 ka B.P., when the area was too cold for water movement in the caves (Gascoyne, 1992). Limited speleothem deposition occurred during an interstadial episode, from -80-30 ka B.P. It is interesting that these periods of speleothem growth correspond reasonably well with warm intervals noted in the S180 record of foraminifera in ocean sediments and also with dates on corals that grew during interglacial periods of high relative sea-level stands (at 10,000 yr B.P. to the present and between 145,000 and 85,000 yr B.P.). The period from -300-165 ka B.P. seems to have been warm enough for cave waters to circulate, although the larger error bars on many older dates make it difficult to resolve any short cold episodes. More dates on speleothems from selected localities are needed to clarify the record further.

7.8.3 Isotopic Variations in Speleothems

In addition to using periods of speleothem growth as a rather crude index of paleo-climatic conditions, attempts have also been made to use oxygen isotope variations along the speleothem growth axis as an indicator of paleotemperatures. When air and water movement in a cave is relatively slow, a thermal equilibrium is established between the bedrock temperature and that of the air in the cave, approximating the mean annual surface temperature. During deposition of calcite from seepage (drip) water, as C02 is lost, fractionation of oxygen isotopes occurs that is dependent on the temperature of deposition. Thus, in theory oxygen isotopic variations in the speleothem calcite (S18Oc) should provide a proxy of surface temperature through time. Unfortunately, the situation is not quite so simple. First, isotopic paleotemperatures are recorded only if the calcite (or aragonite) is deposited in isotopic equilibrium with the drip-water solution. This can be assessed by determining if 818Oc is constant along a growth layer; if values vary for the same depositional interval, it indicates that deposition was affected by evaporation, not just the slow degassing of C02; this would alter the simple temperature-dependent fractionation relationship.

Interstadials Since 140ka

FIGURE 7.29 Histogram of -180 uranium series dates on speleothems from the North of England (Yorkshire Dales). Speleothem formation is associated with warm interglacial or interstadial conditions when groundwater movement was not impeded by temperatures below freezing. No samples dated between -140-165 kyr B.P. or from 15-30 kyr B.P., indicating cold, glacial conditions at those times (Gascoyne, 1992).

FIGURE 7.29 Histogram of -180 uranium series dates on speleothems from the North of England (Yorkshire Dales). Speleothem formation is associated with warm interglacial or interstadial conditions when groundwater movement was not impeded by temperatures below freezing. No samples dated between -140-165 kyr B.P. or from 15-30 kyr B.P., indicating cold, glacial conditions at those times (Gascoyne, 1992).

Another test of isotopic equilibrium involves comparing variations in carbon and oxygen isotopes along individual growth layers (Hendy and Wilson, 1968; Hendy, 1970). If a nonequilibrium situation existed, the isotopic composition would be controlled by kinetic factors and the same fluctuations would be found for both carbon and oxygen isotopes. If no correlation between these two isotopes is found, it can be assumed that the carbonate speleothem was deposited in equilibrium. Some indication of the likelihood of equilibrium conditions being present in the past can be obtained by analyzing the 180/160 ratios in present-day ground water and in calcite deposited from it, which should indicate deposition in isotopic equilibrium.

Interpretation of 180/160 variations in speleothems is not easy because a number of climatic factors other than cave temperature can influence observed 818Oc values. First, with a decrease in cave temperature, the fractionation factor between calcite and water increases, causing an increase in the 818Oc values. However, as air temperature at the surface decreases, so the 180/160 of precipitation, and thus the 8180 value of drip water tends to decrease. Finally, during glacial periods, the growth of lsO-depleted continental ice sheets results in an increase of 8180 values of oceanic water and hence also of precipitation (see Section 5.2.3). Thus, for a given climatic shift several opposing factors come into play, and it is difficult to assess a priori in which direction the 818Oc will change (Thompson et al., 1976; Harmon et al.; 1978a); indeed the calcite 8180-temperature relationship may not even be constant through time. This has led to diametrically opposite interpretations of 818Oc variations in speleothems. Duplessy et al. (1970b, 1971) for example, assumed that measured 8lsO variations were the result of variations in the lsO content of precipitation; hence lower 8lsO values were interpreted as indicating colder conditions. This was disputed by Emiliani (1972), who observed that the speleothem 818Oc record, as interpreted by Duplessy et al. was the inverse of paleotemperatures derived from oceanic fora-minifera. He therefore concluded that the speleothem 818Oc variations were not controlled by variations in the 8180 of precipitation, but due to the dominant effect of temperature-dependent fractionation. This has subsequently been confirmed in other (but not all) localities by the analysis of the isotopic composition of drip water trapped as tiny liquid inclusions as the speleothem grew (Schwarcz et al., 1976). These inclusions vary in abundance and, when present in large amounts (>1% by weight), give speleothems a milky appearance; by isolating and analyzing the liquid at successive levels along the growth axis of a speleothem it is possible to assess, directly, whether isotopic variations of precipitation have occurred (Thompson et al., 1976; Harmon et al, 1979).

Because it is possible that the inclusion water may have continued to exchange oxygen isotopes with the surrounding calcite following its entrapment, it is of no value to measure oxygen isotopes directly. A measure of the oxygen isotope fractionation between calcite and inclusion water would probably give a temperature close to present-day ambient temperature levels. Instead, the deuterium-hydrogen (D/H) ratio is measured because there is no hydrogen in the calcite with which hydrogen in the water might have exchanged. It is assumed that the relationship between 8D and 8lgO in drip water approximates that noted in meteoric water by Dansgaard (1964):

In this somewhat circuitous manner it is possible to estimate the former 8lsO values of meteoric water over very long periods of time. By thus controlling for changes in the isotopic composition of precipitation, it is then possible to use the S18Oc values of the surrounding calcite to estimate paleotemperatures. Thompson et al. (1976) have carried out such studies on speleothem calcite and inclusion waters from caves in West Virginia. Their results show that the oxygen isotopic composition of inclusion water at this site has changed very little over time, supporting the view of Emiliani (1972) that changes in 8lsO were largely controlled by variations in calcite-water fractionation factors (i.e., temperature changes at the site). Thus, at least in this case, the 818Oc values of speleothem calcite increase with falling temperatures. Using this interpretation, the 818Oc data indicate that West Virginia experienced three major warm episodes in the last 200,000 yr —at <10,000 yr B.P., at 110,000-100,000 yr B.P., and at 175,000 ± 10,000 yr B.P. Cold intervals appear to have occurred prior to 200,000 yr B.P. and at -180,000, 165,000-110,000, and 95,000-15,000 yr B.P., the last-mentioned period perhaps interrupted by a warmer interval at -50,000 yr B.P. These records, although incomplete, do show some similarities with other isotopic records from sites in Alberta, Canada; Iowa, Kentucky, and Bermuda (Harmon et al., 1978b) and are in reasonable agreement with marine isotopic records.

A very detailed 8lsO analysis of a stalagmite from northern Norway (at a sampling interval of 20-30 yr for the last 5000 yr) has been interpreted in terms of pa-leotemperature by Lauritzen (1996) (Fig. 7.30). He "calibrated" the 8lsO variations by reference to both modern temperatures (compared to 8180 in calcite forming today) and to conditions in the mid-18th century when other data indicate temperatures were 1.5 °C lower. At that time (one of the coldest in the entire Holocene), speleothem growth ceased. These points provided a crude yardstick on which to hang the overall Holocene temperature changes, assuming the 8180 record primarily reflects temperature-related shifts in the isotopic composition of meteoric waters (precipitation). Some reassurance that this may well be the case is provided by a very similar (though less detailed) Holocene paleotemperature record for west-central Norway, based on entirely separate and independent data (Nesje and Kvamme, 1991). However, further studies are required before this speleothem paleotemperature reconstruction can be accepted with confidence.

Elsewhere, 818Oc data indicate that temperature-related effects on the isotopic composition of precipitation are more important than the temperature-dependent fractionation effects in calcite deposition (Dorale et al., 1992). In such cases, 818Oc and temperature are positively correlated (that is, colder temperatures are indicated by isotopically lighter calcite). Thus, Goede (1994) suggested that declining values of 818Oc in a Tasmanian speleothem were indicative of a cooling trend from 98-60,000 yr B.P., followed by an increase in temperatures to -55,000 yr B.P. (Fig. 7.31). Other proxy data suggest that the overall change in temperature over this interval was -6 °C, which would mean that 818Oc in this cave varied by + 0.26%o /°C.

The longest isotopic speleothem record currently available is from Devil's Hole, Nevada, where a calcite vein precipitated from groundwater supersaturated in calcite spans almost 0.5 Ma (60 ka to -560 ka B.P.)(Ludwig et al., 1992). In this system, 8lsO in calcite is considered to reflect changes in the isotopic composition of precipitation feeding the groundwater, so that higher values reflect warmer conditions

0123456789 10 11 103 calendar years B.P.

FIGURE 7.30 Paleotemperature reconstruction from oxygen isotopes in calcite sampled along the growth axis of a stalagmite from a cave at Mo i Rana, in northern Norway. Samples were taken every I mm at the top, corresponding to a resolution of ~25-30 yr. Linear interpolation between 12 TIMS U-series dates along the stalagmite provide a timescale in calendar years, with an average error of 10-50 yr.The record is very similar to that reconstructed independently from pollen and glaciological data (Lauritzen, 1996; Lauritzen and Lundberg, 1998).

0123456789 10 11 103 calendar years B.P.

FIGURE 7.30 Paleotemperature reconstruction from oxygen isotopes in calcite sampled along the growth axis of a stalagmite from a cave at Mo i Rana, in northern Norway. Samples were taken every I mm at the top, corresponding to a resolution of ~25-30 yr. Linear interpolation between 12 TIMS U-series dates along the stalagmite provide a timescale in calendar years, with an average error of 10-50 yr.The record is very similar to that reconstructed independently from pollen and glaciological data (Lauritzen, 1996; Lauritzen and Lundberg, 1998).

(Winograd et al., 1988,1992; Johnson and Wright, 1989). The record shows very similar variations to the SPECMAP marine isotope record (Fig. 7.32) as well as to 8D in the Vostok ice core suggesting that the signal recorded is of more than regional significance. However, the Devil's Hole record has generated considerable controversy because the timing of the transition from the penultimate glaciation to the last interglacial occurred earlier at Devil's Hole than in the marine isotope record; further the interglacial maximum was -140 ka B.P. at Devil's Hole, compared to the SPECMAP isotopic minima at -128 ka B.P. It is worth noting that the Devil's Hole calcite is extremely well dated (by U-series, including many high-precision dates derived by TIMS) whereas the SPECMAP record was tuned to orbital frequencies, assuming orbital forcing was the dominant control on continental ice volume changes.27 Furthermore, the Devil's Hole calcite was only deposited after precipitation had been transferred through the groundwater system, which must have taken at least several thousand years (and by some estimates >10 ka), making the discrepancy between the two records even larger. Winograd et al. (1992) argue that the warming that led to

27 However, the last interglacial isotopic minimum in marine sediments corresponds to the high sea-level stand, recorded by corals (uplifted by tectonic activity) in several locations around the world and consistently dated at -125,000 ± 2500 yr B.P. by U-series TIMS (Gallup et al, 1994; Stirling et al, 1995). Direct U-series dating of marine sediments also confirms this time (123,500 ± 4500 yr B.P.) as the last interglacial peak, when continental ice volume was at a minimum (Slowey et al., 1996).

FIGURE 7.31 Variations of 8,8Oc in aTasmanian speleothem spanning the interval from 98,000-55,000 yr B.P., based on uranium-series dating methods. In this record temperature and 8l8Oc are positively correlated so temperatures generally declined from 98,000 to -60,000 years B.P.; independent evidence suggests that this temperature change was -6 °C (Goede, 1994).

FIGURE 7.31 Variations of 8,8Oc in aTasmanian speleothem spanning the interval from 98,000-55,000 yr B.P., based on uranium-series dating methods. In this record temperature and 8l8Oc are positively correlated so temperatures generally declined from 98,000 to -60,000 years B.P.; independent evidence suggests that this temperature change was -6 °C (Goede, 1994).

FIGURE 7.32 The 8l8Oc record in a calcite vein from Devil's Hole, Nevada compared to the SPECMAP marine isotope record. Selected marine isotope substages are numbered.Values are expressed as departures from the overall mean of each series, in standard deviation units (computed over the full record).The value of zero thus represents the mean for each record. Note that the sign of the SPECMAP time series has been reversed so that interglaciations appear as peaks (Winograd et al„ 1997).

FIGURE 7.32 The 8l8Oc record in a calcite vein from Devil's Hole, Nevada compared to the SPECMAP marine isotope record. Selected marine isotope substages are numbered.Values are expressed as departures from the overall mean of each series, in standard deviation units (computed over the full record).The value of zero thus represents the mean for each record. Note that the sign of the SPECMAP time series has been reversed so that interglaciations appear as peaks (Winograd et al„ 1997).

the last interglacial maximum (stage 5e in the marine isotope record) actually began around 150 ka B.P. and eustatic sea level had reached modern levels by -135 ka B.P. The critical implication of this argument is that sea level started to rise when or-bitally driven northern hemisphere insolation anomalies were low (and were actually declining) so that the conventional view of solar insolation forcing deglaciations is thus not tenable. This assault on the Milankovitch hypothesis launched numerous counterarguments, none of which have entirely settled the matter (Johnson and Wright, 1989 and reply by Winograd and Coplen, 1993; Shackleton, 1993 and reply by Ludwig et al., 1992; Edwards and Gallup, 1993 and reply by Ludwig et al., 1992; Imbrie et al., 1993c and reply by Winograd and Landwehr, 1993). Additional precisely dated records are needed to duplicate and establish the global significance of the Devil's Hole record, and to resolve the important questions this controversy has raised (Hamelin et al., 1991).

7.8.4 Speleothems as Indicators of Sea-Level Variations

Speleothem growth in carbonate island locations close to sea level can provide an extremely valuable indicator of former sea-level position, and hence ice volume changes on land. As speleothems must form above sea level in air-filled caves, the occurrence of speleothems in locations presently below sea level provides an upper limit to sea level at the time of formation (Fig. 7.33). Similarly, those speleothems

Mediterranean Sea, offToscana (Central Italy).The stalagmite formed during glacial time (22,670 ± 460 yrs cal BP) when sea-level was up to 120m lower, but was subsequently submerged due to eustatic sea-level rise. Dating the inner parts of the overgrowths on several submarine stalagmites, sampled at different depths, has enabled a well-constrained sea-level history to be established (Alessio et al., 1998). (photograph kindly provided by Fabrizio Antonioli, ENEA-Environmental Dept., Roma Italy).

Mediterranean Sea, offToscana (Central Italy).The stalagmite formed during glacial time (22,670 ± 460 yrs cal BP) when sea-level was up to 120m lower, but was subsequently submerged due to eustatic sea-level rise. Dating the inner parts of the overgrowths on several submarine stalagmites, sampled at different depths, has enabled a well-constrained sea-level history to be established (Alessio et al., 1998). (photograph kindly provided by Fabrizio Antonioli, ENEA-Environmental Dept., Roma Italy).

exposed today that have overgrowths of marine aragonite indicate unequivocally that sea level was formerly higher. Uranium-series dating of the speleothems and coral deposits enables a picture of relative sea-level variations through time to be built up. Thus, Harmon et al. (1978b) were able to conclude from studies of Bermuda speleothems that interglacial conditions (high sea-level stands) occurred at around 120,000 and 97,000 yr B.P.; between these events, a lower sea-level stand (-8 m) occurred at -114,000 yr B.P. Li et al. (1989) and Lundberg and Ford (1994) extended this approach by applying high-precision (TIMS) dating to a flow-stone from the Bahamas (at -15 m water depth), which revealed growth hiatuses attributable to pre-Holocene high sea-level episodes at >280 ka, -230 ka, -215 ka, -125 ka, and -100 ka B.P. Not all hiatuses are attributable to sea-level rise; a reduction in precipitation could limit groundwater flow and lead to cessation of cal-cite deposition (Richards et al., 1994). Nevertheless, evidence of continuous speleothem growth below current sea level in areas of platform stability is a definitive indication that sea level could not have been above the threshold level during that time. Thus, Richards et al. (1994) were able to eliminate the possibility that sea level rose above -18 m at any time between 93 ka and 15 ka B.P. Furthermore, their data constrains sea level at marine isotope stage 5a as having been between -15 and -18 m, based on the onset of speleothem growth at these levels at 93 ka and 80 ka B.P., respectively. Further (submarine) sampling holds the potential of revealing in considerable detail the precise magnitude of former eustatic sea-level changes, which in turn has important implications for interpretation of ice sheet growth and decay rates.

This Page Intentionally Left Blank

+1 0

Responses

  • SAVANNA
    What causes periods of increased speleothem growth?
    6 years ago
  • luce
    What does degassing of carbon dioxide mean for caves?
    6 years ago
  • Pirkko
    Why are speleothems useful?
    5 years ago

Post a comment