Odekon M. Encyclopedia of paleoclimatology and ancient environments

Подождите немного. Документ загружается.

Figure P85 Pollen percentage diagram from borehole Ioannina 284. The spectra are presented as histograms connected by a sawtooth curve.

Depth is on the ordinate axis. Ages are given for specific points: the upper seven ages are calibrated radiocarbon ages, the lower ages are based on

astronomical calibration of pollen-vegetation abundance patterns (Tzedakis et al., 2002). Reprinted with permission from Tzedakis et al., 2002.

Figure P84 Percentage pollen diagram from Amoco Production Company’s Bridge well, Great Salt Lake. The volcanic ashes utilized for age control

are at the far left, along with a chronostratigraphic ordinate axis. The modern analog index has been plotted, along with ten common pollen

taxa utilized in the pollen sum. The shaded sawtooth curve for “Ephedra and Artemisia” represents a 10 exaggeration of the percentages.

“Cyperaceae + Myriophyllum + Typha” are wetland plants. “Botryococcus + Pediastrum” are algae. Geological time periods appear to the far right

(from Davis and Moutoux, 1998). Reproduced with permission from the author.

POLLEN ANALYSIS 821

often been used as biostratigraphic tools. The correlation of

zones amongst sites is a frequent point of discussion and con-

troversy in post-glacial pollen analysis, especially in Western

Europe. The recent increased utilization of radiometric dating,

most commonly radiocarbon applied to late Quaternary sedi-

ments, has somewhat replaced the reliance on zones for site-

to-site correlation.

Interpretation

The researcher is striving to interpret pollen assemblages and

their changes over time, or from one locale to another, as

changes in vegetation. “The vegetation is the central object in

pollen-analytic investigations” ... and yet there is “no one-to-

one correlation between the representation of a species in the

vegetation cover and in the pollen spectrum (Fægri et al.,

1989, p. 129).” Therefore, the soundest interpretations will

depend on the researcher’s knowledge of all the ecological and

geological factors that have affected the final pollen record.

The knowledge and skills required making a reasonable inter-

pretation of past vegetation and climate based upon a pollen

record will vary as greatly as the vegetation formations and

geological regimes encompassed by the pollen record.

Two common methodologies utilized to interpret the pollen

assemblages are the “modern analog” and the “indicator spe-

cies” methods. The modern analog method seeks to elucidate

the relationship between the vegetation that covers an area

and the pollen assemblage found at a point within that area.

Much field work has involved the collecting of small surface

sediment samples from within different types of vegetation for-

mations: the pollen are separated from the sediment, identified

with a microscope, and a spectrum of the identified pollen

taxa is produced; the surrounding vegetation is also identified

and classified. With sufficient samples, one can at least qualita-

tively correlate a pollen spectrum with a vegetation forma-

tion. Several complications arise: the existing array of pollen

surface samples is extensive in some vegetation types and not

others, the pollen surface samples are useful only insofar as

the vegetation (which produced the pollen) is in a state of

equilibrium with the environment (in all aspects), some impor-

tant geographical regions can yield no information on the

modern pollen-vegetation relationships since anthropogenic

activity has eliminated all the natural vegetation, and not

all pollen assemblages in the geological record have a modern

analog. However, in areas with adequate pollen surface sam-

ples, quantitative relationships between climate and pollen

may be attempted and the specialist literature contains exam-

ples of these (for a more detailed introduction see Bradley,

1999).

The indicator species method relies on the ecological or cli-

matological preferences of one (or a few) key taxa. If a taxon’s

habitat is highly specific, ecologically or climatologically, then

the presence of that single taxon may indicate the habitat, even

in the past. Alternatively, several taxa might be used to narrow

the climatic or ecological range of the spectrum. The indicator

species method is valid only if the fossil pollen truly indicates

the presence of the plant in the immediate vicinity of the

sampled site. Thus, application of this method should also utilize

modern pollen distribution studies showing correlations

between actual presence of a taxon, and a numerical pollen per-

centage threshold. The identification of plant macrofossils

(seeds, leaves, etc.) can substantiate a specific taxon’s presence,

which may be doubtful if based on pollen percentages alone.

Conclusions: examples of pollen records

The following examples show pollen analysis applied to late

Cenozoic sediments. The pollen taxa identified belong to extant

genera (or families) and hence interpretation may be based on

modern analogs and indicator species. Chronological resolution

varies with the basin cored; chronological control also varies,

from volcanic ashes to radiocarbon dates and to correlation

with marine records.

The longest continuous terrestrial sediment cores in North

America on which pollen analysis have been conducted come

from the Great Salt Lake, northern Bonneville Basin region

(northwest Utah; Davis and Moutoux, 1998; Davis, 2002).

Pollen analysis was conducted from sediment cuttings (which

integrate sediment from a range of depths) retrieved during

drilling conducted for geophysical exploration. The wells are

exceptionally deep, exceeding 3 km. The sampling interval is

from 3 to 30 m, and thus chronological resolution ranges from

8 to 23 kyr. The Bridge well (Figure P84) contains the longest

accumulation of pollen and sediments. The interpretation of the

relationship between vegetation and pollen spectra is based on

modern analogs; the fossil pollen samples were compared to

1,367 modern pollen samples from the western United States.

The squared chord distance metric statistical method, also

called square chord distance, was used to determine the ana-

logy between modern and the fossil pollen samples quantita-

tively. The environmental development of the Great Basin can

be interpreted from these pollen diagrams. Since the pollen

spectra were dominated by xerophytic vegetation, Sarcobatus

and Ephedra, from >5 Ma ago to the present, the region’s

present-day desert vegetation is interpreted to have existed

since the late Miocene. In addition, the poor pollen preserva-

tion and scarce pollen of wetland plants indicates that lakes

and marshes in the Great Salt Lake basin were shallow and

intermittent before 2.6 Ma. Between 5 and 0.75 Ma, the Sarco-

batus and Ephedra pollen are replaced by pollen from the

wetland plants, “Chenopodiaceae + Amaranthus” (these pollen

types are indicative of greasewood steppe vegetation), and

Artemisia (sagebrush), which indicates a regional cooling and

increase in moisture. After 0.75 Ma, Artemisia and Pinus pol-

len predominate, which is interpreted as downward displace-

ment of vegetation zones by approximately 500 m in

elevation, and the replacement of the greasewood steppe by

sagebrush steppe in the intermountain basins (Davis, 2002).

The Ioannina basin, Pindus Mountain Range, Greece, con-

tains a 430,000 year record of pollen in which fluctuating per-

centages can be used to understand the Quaternary climate

cycles and the reactions of tree populations (Tzedakis, 1993).

It provides a rare combination of stratigraphic continuity, and

high sedimentation accumulation rates for the late Quaternary.

Several lacustrine cores have been taken from the basin: the

increasing detail of the pollen analysis is evident, as the results

have been published. The first analysis shows the pollen

spectra alternating between forest and open vegetation com-

munities, reflecting the interglacial and glacial climate, res-

pectively. Four cycles are evident, and they are correlated

with glacial-interglacial changes registered in other southern

European pollen records. However, a unique feature of the Ioan-

nina cores is the continuous presence of Quercus and Abies,

followed by Pinus, Ulmus/Zelkova and, to a certain extent,

Corylus: these are temperate trees (Tzedakis, 1993). These

temperate tree populations persist even during glacial periods.

This record of the continuous presence of temperate tree

822 POLLEN ANALYSIS

pollen is used to support the ecological hypothesis that these

mountains provided long-term refugia for the temperate taxa

(Tzedakis, 1993).

Increased detail is apparent in the pollen analysis conducted

on a borehole called Ioannina 284 (Figure P85) (Tzedakis et al.,

2002, 2003). The mean sampling interval has been increased to

approximately 225 years. Because of the increased temporal

resolution of the record, greater detail within a single glacial

or interglacial becomes evident. The onset of the last intergla-

cial has been defined by the authors as occurring when tree pol-

len consistently rises above 50% frequency, and when tree

pollen concentration exceeds 200,000 grains per cm

; the onset

of the ensuing colder stadial occurs when tree pollen con-

sistently drops below 50%. Thus, the Last Interglacial is rep-

resented in the interval from 96.7 to 83.4 cm depth, or

approximately 127.3–111.7 kyr ago (Tzedakis et al., 2003).

After the chronology has been determined, the pollen changes

in Ioannina 284 are compared with other climatic data

from the North Atlantic region. The data indicate that the

millennial-scale expansions and contractions of tree popula-

tions may be in phase with climate events in the North Atlantic

region and Greenland regions. However, sufficient moisture to

maintain refugial temperate tree populations is proposed to be

linked to the topographic variability of the Pindus Mountains

(Tzedakis et al., 2002).

The pollen analyses from these two sites show how both

paleoclimatological and paleoecological interpretations may

be extracted from the studies. Furthermore, the means of chron-

ological control will vary with the age of the sediments. With

basins that encompass the Holocene epoch, laminated sedi-

ments may allow high temporal resolution. For older basins

(including marine sediments), age may be based on biostrati-

graphic correlations. Regionally significant climate changes

usually cannot be derived from a single pollen record: several

fossil pollen records together from a user-defined region are

necessary in order to confirm that patterns in pollen spectrum

have significance beyond the single site examined.

Margaret Kneller

Bibliography

Bradley, R.S., 1999. Paleoclimatology, Reconstructing Climates of the

Quaternary. Burlington, MA: Harcourt/Academic Press, 613pp.

Davis, M., 1969. Climatic changes in southern Connecticut recorded by

pollen deposition at Rogers Lake. Ecology, 50, 409–422.

Davis, O.K., 2002. Late Neogene environmental history of the Northern

Bonneville Basin: A review of palynological studies. In Hershler, R.,

Madsen, D.B., and Currey, D.R. (eds.), Great Basin Aquatic Systems

History. Washington, D.C.: Smithsonian Institution, pp. 295–307.

Davis, O.K., and Moutoux, T.E., 1998. Tertiary and Quaternary vegetation

history of the Great Salt Lake, Utah, USA. J. Paleolimnol., 19,

417–427.

Deevey, E.S., 1969. Coaxing history to conduct experiments. Bioscience,

19,40–43.

Erdtman, G., 1943. An Introduction to Pollen Analysis. Waltham, MA:

Chronica Botanica Company, 239pp.

Fægri, K., and Iversen, J., 1975. Textbook of Pollen Analysis, third revised

edition. New York: Hafner Press, 295pp.

Fægri, K., Kaland, P.E., and Krzywinski, K., 1989. Textbook of Pollen Ana-

lysis, by Knut Fægri and Johs. Iversen. Caldwell, NJ: The Blackburn

Press, 328pp.

McAndrews, J.H., Berti, B., and Norris, G., 1973. Key to the Quaternary

Pollen and Spores of the Great Lakes Region. Toronto, ON: The Royal

Ontario Museum, 64pp.

Pennisi, E., 2003. Modernizing the Tree of Life. Science, 300, 1692–1697.

Roberts, N., 1998. The Holocene: An Environmental History, 2nd edn.

Oxford, UK: Blackwell Publishers Ltd, 316pp.

Traverse, A., 1988. Paleopalynology. Winchester, MA: Allen & Unwin

Inc., 600pp.

Twell, D., 2000. Pollen: Structure, Development and Function. In Nature

Encyclopedia of Life Sciences. London, UK: Nature Publishing Group.

http://80-www.els.net.osiyou.cc.columbia.edu:2048 /[doi:10.1038 /npg.

els.0002039].

Tzedakis, P.C., 1993. Long-term tree populations in northwest Greece

through multiple Quaternary climatic cycles. Nature, 364, 437–440.

Tzedakis, P.C., Lawson, I.T., Frogley, M.R., Hewitt, G.M., and Preece, R.C.,

2002. Buffered tree population changes in a Quaternary refugium:

Evolutionary implications. Science, 297, 2044–2047.

Tzedakis, P.C., Frogley, M.R., and Heaton, T.H.E., 2003. Late Interglacial

conditions in southern Europe: Evidence from Ioannina, northwest

Greece. Glob Planet. Change, 367, 157–170.

von Post, L., 1967 [1916]. Forest tree pollen in south Swedish peat bog

deposits, a translation by M.B. Davis and K. Fægri of Om skogstrad-

spollen i sydsvenska torfmosselagerfoljder (foredragsreferat) (Geolo-

giska Foereningen i Skockholm. Foerhandlingar 38:384), with an

introduction by Knut Fægri and Johs. Iversen. Pollen et Spores,

9, 378–401.

Cross-references

Dating, Biostratigraphic Methods

Late Quaternary-Holocene Vegetation Modeling

Paleobotany

Palynology

Quaternary Vegetation Distribution

POTASSIUM-ARGON/ARGON-ARGON DATING

General principl es

The K-Ar dating method utilizes the decay of the naturally-

occurring radioactive isotope of potassium,

K, into an isotope

of the noble gas, argon (i.e.,

Ar).

K decays into two different

daughter products: the most common path (89.5%) is via beta

decay into

Ca; the remainder of the decay paths arrive at

Ar, mostly via electron capture, but also rarely via positive

beta (positron emission) decay. The total decay half life for

K is about 1.25 billion years, which means that there is still

some of the isotope remaining from the formation of the solar

system. Because of the branching decay scheme, the standard

radiometric age equation needs to be modified as follows:

t ¼

ln 1 þ



ð1Þ

where l

is the decay constant leading to

Ar (0.581  10

10

1

) and l is the total decay constant for both branches

(5.543  10

10

1

Much of the natural radioactivity surrounding us is due to

K, but fortunately, it is a quite rare isotope, only accounting

for about one atom in 8,570 atoms of potassium. Despite the

rarity of

K and the fact that only about one atom in ten decays

into

Ar, the K-Ar system is arguably the most versatile of the

long-lived decay scheme dating methods. It is nearly an ideal

chronometer because K is a major element present in almost

all rocks and minerals, but the act of mineral formation is

usually exceptionally efficient at excluding atmospheric argon

from the mineral’s crystal structure. Radiogenic

Ar is well

POTASSIUM-ARGON/ ARGON-ARGON DATING 823

accommodated in many minerals, which can often quantita-

tively retain their Ar to temperatures of 300–400



C or higher.

Therefore, K-Ar dating usually has a very clear-cut resetting of

the radiometric clock and the clock can faithfully record geolo-

gical time for durations ranging from thousands to billions of

years. Indeed, K-Ar and its cousin Ar-Ar dating have been used

successfully to calibrate all of the geomagnetic reversal time

scale, much of the history of early humanity, and most of the

Phanerozoic timescale, plus ancient samples from Earth, the

Moon and Mars.

K-Ar method

In the conventional K-Ar dating method, a part of the sample is

analyzed via wet chemistry techniques for K concentration; the

remainder of the sample is placed into an ultra-high vacuum

system and fused to release all of its trapped gases. The

released water vapor is usually trapped in a “cold trap” at dry

ice or liquid N

temperatures, and the remaining active gases

are removed using a “getter” pump, which consists of a highly

reactive metal alloy kept at an elevated temperature. The getter

removes O

,CO

,CH

, etc., but leaves the Ar untouched.

The remaining “cleaned” Ar sample is usually mixed with a

known quantity of the rare argon isotope

Ar so that the con-

centration of

Ar can be estimated from the

Ar/

Ar ratio

(i.e., isotope dilution). The combined Ar sample is then

admitted to a noble gas mass spectrometer. The presence of

any atmospheric Ar is corrected by monitoring the

Ar signal

and subtracting an amount 295.5 times larger from the

peak (i.e., the

Ar/

Ar ratio of the atmosphere).

The precision of the K-Ar method is limited by the errors

associated with the isotope dilution technique, weighing errors,

and the precision of the chemical measurement of K concentra-

tion. Typical error estimates are about 1–2% for this method.

Accuracy and the reliability of K-Ar ages are further hampered

by so-called “nugget effects,” where there may be unusual

phases in either the K or Ar sample split. This possibility and

the requirement for two separate measurements limit the use

of the technique to relatively large and homogeneous samples.

Ar-Ar or the

Ar/

Ar method

In 1966, the team of Merrihue and Turner, working in John

Reynold’s laboratory, devised a remarkable enhancement of

the K-Ar method that has greatly increased its scope and power

(Merrihue and Turner, 1966). In the

Ar/

Ar (or Ar-Ar) dat-

ing method, a single split of the sample is sent to a nuclear

reactor and irradiated with high-energy neutrons. Some of the

most common isotope of potassium, the stable isotope

undergoes a reaction where it briefly absorbs a neutron and

expels a proton, thereby transforming into

Ar. This isotope

of argon is radioactive and is not present in the atmosphere in

appreciable quantities since it has a half-life of only ca. 270

years. The sample now has both its original radiogenic

and another argon isotope, which can effectively be used as a

proxy for potassium. It is possible to measure the

Ar/

ratio, which is proportional to the

Ar/K ratio, which is in turn

proportional to the

K ratio. The proportionality constant

is estimated by analyzing a standard mineral of known K-Ar

apparent age during the same neutron irradiation as the sample.

The

Ar/

Ar age equation can be written as:

t ¼

ln 1 þ J





ð2Þ

The superscript * denotes radiogenic

Ar. J is calculated as:

J ¼

standard

 1



ðÞ

standard

ð3Þ

The suffix “K” is put on

Ar to indicate that only the com-

ponent from potassium is used. There are interfering nuclear

reactions, usually from Ca, which produce small amounts of

Ar and

Ar, but these can be corrected by calibration ana-

lyses of pure Ca salts. Ca also produces significant quantities

of the radioactive isotope

Ar and it is possible to infer the

relative concentrations of Ca, Cl, and K by the enhanced abun-

dances of

Ar,

Ar, and

Ar, respectively.

The

Ar/

Ar method eliminates the problem of sample het-

erogeneity and it has significantly better precision relative to

K-Ar dating because all of the measurements are based on iso-

tope ratio measurements, not absolute concentration estimates,

and mass spectrometers are much more accurate with the for-

mer style of measurement. Age estimates that are precise to

within a few tenths of a percent are commonplace, but even

more important than laboratory precision are the extra interpre-

tation capabilities of the technique, which lead to more accurate

age estimates.

Since age information comes from Ar isotope ratios, it is

possible to step-heat a sample in gradually increasing tempera-

ture steps and calculate an apparent age for each gas fraction.

The sequence of apparent ages is plotted as an “age spectrum,”

which can be extremely useful in testing whether or not the

sample has remained a closed system during its lifetime. Ideally,

the age spectrum will yield a flat sequence of ages or a perfect

“plateau age” which can be used with much greater confidence

than a comparable single K-Ar age. Alternatively, there are

other typical age spectral patterns such as that expected from

diffusive loss of Ar or the so-called “saddle shaped” spectrum,

which is diagnostic for the presence of excess

Ar. Sometimes

it is possible to derive useful age information event from these

disturbed age spectra (e.g., Snyder et al., 1996), but at the very

least a disturbed age spectrum is a red flag indicating that the

data need to be interpreted with care.

Ar/

Ar step-heating also makes it possible to do an iso-

chron style analysis on a single sample, which was impossible

with the K-Ar method. Usually in an argon isochron diagram,

the

Ar/

Ar ratio is plotted against the

Ar/

Ar ratio, and the

intercepts of the fitted line indicate the sample’sinitial

Ar/

ratio and apparent age. This kind of analysis is particularly useful

for young samples (a few Ma) where the usual assumption of

atmospheric composition initial Ar can be rigorously tested.

Finally, the combination of

Ar/

Ar dating with a new

generation of low-blank laser fusion systems (York et al.,

1981) and high sensitivity mass spectrometers has led to a

whole generation of highly automated laboratories capable of

routinely analyzing extremely small and extremely young sam-

ples. A crucial application has been to revolutionize the dating

of volcanic tephra (Lo Bello et al., 1987) by enabling the

researcher to date single crystals of volcanic feldspar, thereby

eliminating contamination problems, which had long plagued

bulk sample analyses. This in turn has allowed for extremely

accurate ages for studies ranging from calibrating the geomag-

netic reversal timescale to dating early human specimens. For a

more complete description of the

Ar/

Ar method, see

McDougall and Harrison (1999).

Chris M. Hall

824 POTASSIUM-ARGON/ ARGON-ARGON DATING

Bibliography

Lo Bello, P., Fraud, G., Hall, C.M., York, D., Lavina, P., and Bernat, M.,

1987.

Ar/

Ar step-heating and laser fusion dating of a Quaternary

volcanic from Neschers, Massif Central, France: The defeat of xeno-

crystic contamination. Chem. Geol. (Isotope Geosci.), 66,61–71.

McDougall, I., and Harrison, T.M., 1999. Geochronology and thermochronol-

ogybythe

Ar/

Ar method. New York: Oxford University Press, 269pp.

Merrihue, C., and Turner, G., 1966. Potassium-argon dating by activation

with fast neutrons. J. Geophys. Res., 71, 2852–2859.

Snyder, G.A., Hall, C.M., Lee, D.-C., Taylor, L.A., and Halliday, A.N.,

1996. Earliest high-Ti volcanism on the moon:

Ar-

Ar, Sm-Nd and

Rb-Sr isoptopic studies of Group D basalts from the Apollo 11 landing

site. Meteoritics, 31, 328–334.

York, D., Hall, C.M., Yanase, Y., Hanes, J.A., and Kenyon, W.J., 1981.

Ar/

Ar dating of terrestrial minerals with a continuous laser. Geo-

phys. Res. Lett., 8(11), 1136–1138.

Cross-reference

Dating, Radiometric Methods

PRECESSION, CLIMATIC

The celestial equator intersects the ecliptic in two points. The ver-

nal equinox is the one at which the Sun in its apparent annual

motion crosses from South to North of the equator. The vernal equi-

nox can be used as a reference point to define the celestial longitude

locating the Earth along its orbit during the year. The longitude of

the perihelion (ϖ) represents the angular position of the perihelion

of the orbit from the vernal equinox. This angle gives the position

of the Earth on its orbit at the spring equinox. Knowing the longi-

tude of the perihelion, it is possible to determine the Earth-Sun dis-

tance at any time in the year. Climatic precession (e sin ϖ)is

usually used instead of the longitude of the perihelion. It is the main

factor driving the 24 h-mean irradiance.

Astronomers of old civilizations kept records of the direc-

tion of the Earth’s axis of rotation in the sky. Five thousand

years ago, stars circled around a-Draco’s in the night sky, while

it is now around a Ursa Minor (polaris). Indeed the Earth’s axis

of rotation does not point towards a fixed direction through

time but describes a clockwise circle in space. This happens

because the torque of the Moon and the Sun on the Earth’s

equatorial bulge causes the axis of rotation to wobble (the

Earth’s equatorial radius is close to 6,375 km while its polar

radius is only 6,355 km). This motion of the Earth’s axis

of rotation, called astronomical precession, is similar to the

precession of a tilted top and it induces the motion of the vernal

equinox, with a period of 25,700 years. However, from a paleo-

climatic point of view, the situation is not so simple. Indeed,

the ecliptic is not fixed in time either, leading to an anticlock-

wise motion of the perihelion. Consequently the climatic pre-

cession, which describes the position of the perihelion with

respect to the vernal equinox, varies with a quasi-period of

21,000 years, resulting from the combination of several peri-

ods, the most important being around 23,000 and 19,000 years

(Berger, 1978; Berger and Loutre, 1991) (Figure P86).

The present-day value of the longitude of the perihelion is

ϖ = 102.04



, which gives a climatic precession of e sin

ϖ = 0.01636. In this situation the Earth is at the closest dis-

tance to the Sun in early January (Berger and Loutre, 1994).

Therefore the mean solar irradiance received during the south-

ern summer in the Southern Hemisphere is larger than the mean

solar irradiance received during the boreal summer in the

Northern Hemisphere. Some 11,000 years ago, the Earth

was at the closest distance to the Sun around the June solstice.

This increased the solar irradiance received during boreal sum-

mer and reduced it during Southern Hemisphere summer. The

climatic precession therefore plays opposite roles in both hemi-

spheres. Any insolation change in one hemisphere due solely

to a variation in climatic precession corresponds to an insola-

tion change of opposite sign and same amplitude in the other

hemisphere six months later (Figure P87).

Moreover, the length of the seasons, which vary in time

according to Kepler’s law, is also strongly dependent on the

climatic precession (Berger and Loutre, 1994). Presently the

length of the Northern Hemisphere seasons is respectively

92.8 days for spring, 93.6 days for summer, 89.8 days for

autumn, and 89 days for winter. In about 4,500 years from

now, the Northern Hemisphere spring and winter will have

about the same short duration (89.6 days), and summer and fall

will also be equally long (93.0 days).

Marie-France Loutre

Bibliography

Berger, A.L., 1978. Long-term variations of daily insolation and Quatern-

ary climatic changes. J. Atmos. Sci., 35(12), 2362–2367.

Berger, A., and Loutre, M.F., 1991. Insolation values for the climate of the

last 10 million years. Quaternary Sci. Rev., 10, 297–317.

Berger, A., and Loutre, M.F., 1994. Precession, eccentricity, obliquity, inso-

lation and paleoclimates. In Duplessy, J.C., and Spyridakis, M.-T.

(eds.), Long Term Climatic Variations, Data and Modelling, vol. 22.

Nato ASI series, Serie I: Global Environmental Change, Berlin:

Springer, pp. 107–151.

Figure P86 Long-term variations of the climatic precession over the last 800,000 years.

PRECESSION, CLIMATIC 825

Cross-references

Astronomical Theory of Climate Change

Climate Forcing

Eccentricity

Obliquity

Pre-Quaternary Milankovitch Cycles and Climate Variability

Quaternary Climate Transitions and Cycles

SPECMAP

PRE-QUATERNARY MILANKOVITCH CYCLES AND

CLIMATE VARIABILITY

Introduction

Cyclic variations in insolation, caused by the precession and

obliquity of Earth’s spin axis and variations in the eccentricity

of Earth’s orbit, have been stressed as a driver of Quaternary cli-

mate since Hays et al. (1976). The Quaternary, however, com-

prises only 0.4% of Earth history, and is characterized by

unusual “ice house” conditions. In addition, neither the mechan-

ism through which ice house conditions were initiated, nor the

mechanism whereby the 100 kyr cycle has come to dominate

glacial oscillations are understood (see Muller and MacDonald,

2000). Analysis of pre-Quaternary Milankovitch cycles provides

a more synoptic view of the role of astronomically controlled cli-

mate change, a context for understanding the present state of the

Earth system and anthropogenic modifications, a mechanism for

producing high-resolution timescales, and data for calibrating the

dynamic evolution of the solar system (see reviews by Berger

et al., 1989; Fischer et al., 1991; Schwarzacher, 1993; Weedon,

1993, 2003;Hinnov,2000, 2005).

Methods

Two fundamentally different, although mutually illuminating,

approaches to the analysis of pre-Quaternary Milankovitch

cyclicity have been developed. The first, cycle counting, pre-

dates acceptance of the Milankovitch theory for Quaternary

climate change by nearly a century (e.g., Gilbert, 1895)

(Figure P88). Cycle counting is an inherently typological

approach requiring both identification of each sedimentary or

proxy cycle and its boundaries within a section and a timescale

to determine cycle duration. Bundles of cycles of variable thick-

ness allow recognition of the characteristic Milankovitch cycle

hierarchy based on the thickness ratio of short to long cycles,

assuming the highest frequency cycles have been correctly iden-

tified. A variant of the cycle counting approach is named for

Fischer’s(1964) analysis of the Triassic asymmetric carbonate

Lofer cycles (Figure P89). These so-called Fischer plots are

graphs of cumulative departure from average cycle thickness

plotted against cycle number, assuming a uniform period for

each cycle and modified for assumed constant subsidence during

each cycle (see Sadler et al., 1993; Boss and Rasmussen, 1995

for criticism). While the typological cycle counting method

allows easy visualization of sedimentological patterns and cycle

identification in the field, the numerous ad hoc assumptions that

must be made about the variability of the cyclicity makes its

practical use treacherous unless there is a known target pattern

for correlation, such as an age-appropriate and accurate insola-

tion curve. Additionally, often used as a measure of relative

sea-level change, Fischer plots appear to be a poor predictor

for Quaternary examples where sea-level change is well under-

stood (e.g., Boss and Rasmussen, 1995).

The second approach, time series analysis, is commonly

used in Quaternary analyses and was inspired by the pivotal

paper by Hays et al. (1976) that established Milankovitch cycli-

city as the “Pacemaker of the Ice Ages, ” although there were

earlier applications. Here, a variable (lithological or climate

proxy) is plotted against thickness or time, and Fourier or other

quantitative methods are used to determine the frequency prop-

erties of the data that can then be compared to a hypothesis of

orbital forcing (Figure P90). This method dispenses with the

need to identify the cycle type, has the advantage of not requir-

ing variability in the data to be discarded, and thus does not

presume the mode of cyclicity (i.e., precession vs. obliquity).

The disadvantage is that the variable must have a unimodal

relationship to climate or a precise timescale must be available;

these become progressively much less common in more ancient

deposits. Sequences have typically been analyzed by Fourier

methods with the data transformed from the depth or time

domain to the frequency domain. The results are usually

expressed as a periodogram (a graph of frequency against a

measure of the importance of that frequency, such as power).

This method, however, presumes little variation in accumula-

tion rate; that is, the frequencies are stationary with respect to

the depth scale. This assumption can be considerably relaxed

with use of evolutive or depth-frequency analysis. Here, the

frequency properties of the section are analyzed by use of

a moving window and are plotted with respect to depth

(Figure P91). A plethora of techniques is available in which

accumulation rate changes can be determined directly from

the internal frequency properties of the data.

Timescales

Both cycle counting and time series methods require a time-

scale to help assess accumulation rates. Accumulation rates

are quantified by relying largely on direct dating by annual

rhythms (varves) and radiometric methods, indirect dating by

correlation to other well established timescales, such as d

curves or magnetic polarity transitions, and tuning to insolation

curves based on celestial mechanics. In older stratigraphic

Figure P87 Variation of the mean irradiance (W m

2

) following a

change in the climatic precession from perihelion at winter solstice to

perihelion at summer solstice (at present-day eccentricity, e = 0.0167

and obliquity, e 23.446



826 PRE-QUATERNARY MILANKOVITCH CYCLES AND CLIMATE VARIABILITY

records, it becomes progressively harder to apply these proxies

and dating methods successfully and other, more inferential,

methods must be used.

Varves are annual rhythms that can be marine or non-marine

and generally consist of couplets or, more rarely, of triplets

of heterogeneous laminae (see review by Anderson, 1964;

Anderson and Dean, 1988). It is usually assumed that a single

couplet or lamina series represents one year, corresponding to

summer and winter or wet and dry seasons, although some

lacustrine environments could theoretically possess two cou-

plets of laminae series per year, and there can be confusion

with tidal banding. Although varve calibration provides direct

and very high-resolution records of accumulation rates, varved

sequences are rare and when present often constitute only a

small part of the sedimentary column.

Radiometric dates provide another direct way to calibrate

accumulation rates and a basis for a timescale. To be useful,

there must be several dates in a section separated in time

sufficiently to exceed error limits, and they must be geologically

accurate. Very few sections meet these criteria. It is possible

to obtain paleontological or other time-correlative means to

tie radiometric dates to sections under analysis. Possible difficul-

ties are exemplified by debates on the Triassic Latemar and

the Eocene Green River Formation (Hinnov and Goldhammer,

1991; Brack et al., 1996; Pietras et al., 2003; Machlus

et al., 2004, 2008).

As is true for Quaternary records, the relatively well-estab-

lished marine d

O record provides a powerful tie to published

timescales for older Neogene marine sections (e.g., Miller et al.,

1987). The temporal accuracy, however, of correlations to the

Figure P88 Cycle counting and matching to an astronomical record; left: marine Pliocene-Miocene example with astronomical calibration from

Sicily modified from Hilgen (1991b); right: lacustrine Triassic example, modified from Van Houten (1964) with 400 kyr cycle added.

PRE-QUATERNARY MILANKOVITCH CYCLES AND CLIMATE VARIABILITY 827

O record can be no better than the accuracy of the d

record itself, which is undergoing significant modifications.

Using this technique for accumulation rate calibration, only

open marine sections can be meaningfully correlated, and the

section being correlated needs sufficient temporal scope to

encompass enough of the character of the d

O curve.

Correlation and tuning to insolation curves has been exten-

sively used in strata of Neogene age, but less commonly in

the Paleogene (e.g., Hilgen, 1991a,b; Shackleton et al., 2000),

and not at all in the Mesozoic and older strata (Figure P88).

The production of insolation curves to which ancient proxy

curves can be matched is limited by uncertainties in observa-

tions, chaotic behavior of the planets, lack of knowledge of the

evolution of the Earth-Moon system, and dynamic aspects of

mass distribution within the Earth. Insolation curves for the

higher frequency climatic precession and obliquity cycles are

accurate to 40–50 Ma, and the 405 kyr cycle to 250 Ma. The

variations of the 405 kyr cycle beyond 250 Ma are relatively

small, and could be used, recognizing the larger uncertainties,

back to the age of the Earth (see Laskar, 1990, 1999; Berger

et al., 1992;Laskaretal.,1993, 2004; Palike et al., 2004). If

the 405 kyr cycle can be identified, it can be used to calibrate

average accumulation rate and hence provide a mechanism to

establish the period of higher frequency cycles independent of

their drift through deep time (e.g., Olsen and Kent, 1999).

For the late Jurassic through Neogene, the well-established

marine magnetic anomaly timescale (Cande and Kent, 1995)

establishes broad constraints for sections with internal polarity

data. Although the average duration of a chron is ~0.25 million

years (Lowrie and Kent, 2004), correlation to the marine mag-

netic anomaly timescale is useful only for sections involving

millions of years. As there is no marine magnetic anomaly

timescale for strata older than late Jurassic, Milankovitch

cycles have proved more useful for calibrating the magnetic

polarity record than vice versa (e.g., Kent and Olsen, 1999).

One of the most powerful methods of calibrating accumula-

tion rate is via the internal frequency structure of the data series

itself, and this takes two basic forms. First, the celestial

mechanical theory predicts a very specific relationship between

the higher and lower frequencies present in data. The lower fre-

quencies are beat cycles of the higher frequencies within the

precession-related bands or the obliquity-related bands. The

frequencies of the ~100 kyr eccentricity cycles are equal to

the differences between the frequencies of the climatic preces-

sion cycles because these cycles result from the interaction of

the Earth’s precession and the gravitational attraction of the

planets and the eccentricity cycles result from the gravitational

interaction of the very same planets. At present, the main peri-

ods that can be recognized in climatic precession are about 19,

23, and 24 kyr. The periods of the linear combinations of dif-

ferences of the frequencies of these periods are about 93,

125, and 405 kyr (calculated using appropriate precision). This

is a mathematical relationship that holds regardless of changes

in the value of the precession constant in the deep past (due

to the Earth-Moon system evolution or mass distribution within

the Earth), or chaotic changes in planetary orbits. The same

relationship is maintained for the obliquity-related cycles.

However, only the most high-fidelity records (e.g., Triassic

Lockatong Formation; Olsen and Kent, 1996; Hinnov, 2005)

have this level of detail (Figure P89). The second, more widely

used concept is that the ratio of the shorter to the longer periods

within the Milankovitch bands, such as precession to eccentri-

city, changes very slowly through Earth history, presently there

is a 1:5:20 ratio of precession (~20 kyr) to short (~100 kyr) and

long (~400 kyr) eccentricity cycles. This change is theoretically

a steady decrease in the precession period due to tidal friction

that should increase the ratio of climatic precession to eccentri-

city cycles (Figure P92). The same effect is seen in obliquity.

The actual change in precession, however, may be more com-

plicated due to other factors, especially the changing dynamic

Figure P89 Fischer plot of Fischer’s example of Triassic Lofer cycles in marine Triassic carbonates of Austria (modified from Fischer, 1964) with a

modern Milankovitch cycle interpretation overlain.

828 PRE-QUATERNARY MILANKOVITCH CYCLES AND CLIMATE VARIABILITY

ellipticity of the Earth caused by movement of continents,

mantle, or ice sheets. There should be little measurable change

caused by the chaotic diffusion of the planetary orbits for the

typically cited 405 kyr and 100 kyr eccentricity periods, and

thus the main observable secular change in orbital frequencies

should be in precession.

Many authors have used the canonical 1:5:20 ratio of the

shorter to longer cycles as strong evidence of the Milankovitch

origin of pre-Quaternary cycles, despite uncertainty in the actual

ratio, especially in strata hundreds of millions of years old. In

the absence of any other time constraints, these ratios may not

be unique (e.g., Latemar controversy, see below). Presuming that

the average ratios are broadly consistent with available con-

straints, and the section covers millions of years, evolutive

depth-frequency analysis (as opposed to time-frequency) can

reveal shifts in the absolute values of frequencies through a

section, while maintaining the ratios of the higher to lower

frequencies. Only Milankovitch theory predicts constancy of

those frequencies independent of actual accumulation rate. The

advantage to this analysis is that the changes in accumulation

rate can be derived directly from the spectrogram. At present,

only a few sections have been examined this way (e.g., see

Olsen and Kent, 1999; Preto et al., 2001;Weedon,2003).

The interaction of climatic and sedimentological processes

should produce a distortion of accumulation rates, resulting in

periodograms and time- or depth-frequency spectrograms that

are noisy or distorted beyond interpretation. Two techniques,

gamma analysis (Kominz and Bond, 1990) and frequency mod-

ulation (FM) analysis (Hinnov and Park, 1998) were developed

to deal with a situation in which accumulation rate may be a

function of Milankovitch cycles themselves. Both involve a

version of cycle counting. In gamma analysis, an approximate

solution to the relationship between time and facies is found

for a succession of sedimentary cycles. Individual facies speci-

fic to certain environments are identified within a section and

measured in thickness. Assuming that each cycle is of constant

duration and each facies is characterized by a specific effective

accumulation rate (g), an approximation of the accumulation

rate of each facies can be found by solving a series of simulta-

neous equations representing the cumulative time of the differ-

ent facies within each cycle. The fit of these approximations is

assessed by an inverse method using the accumulation rates

derived from the measured section and Fourier analysis of the

resulting time series. If the accumulation rates are identified

correctly and the sequence is of Milankovitch origin, the Four-

ier spectrum should reflect a better fit to the expectations of

Milankovitch theory than the original data. Although this

method incorrectly assumes a constant duration for the preces-

sional cycles, it might reveal the presence of another cycle to

which the section could be tuned, such as the 40 kyr obliquity

cycle of relatively constant frequency. This method has been

applied to Cambrian marine, and Triassic and Jurassic lacus-

trine sequences, significantly improving the spectral properties

of the data.

FM analysis begins with the identification of the presumed

geological expression of precessional (or obliquity) cycles

and estimation of the length of each cycle, creating a new time

series of depth against cycle thickness. Fourier analysis of this

new time series reveals the FM modulating cycles. Since the

precession cycle is frequency modulated by the “eccentricity”

cycles, FM analysis of a real insolation curve reveals the

eccentricity cycles (see Hinnov, 2000). FM analysis of the data

curve also has this characteristic shape if it is the result of

Milankovitch processes. A tacit assumption of this method is

that there is a strong and amplified link between the frequency

modulators of accumulation rate and the eccentricity cycles

themselves. A major disadvantage to this method is that the

higher frequency cycles must be identified, which is nearly

impossible in noisy, low amplitude, or clipped records. In addi-

tion, the Fourier FM spectral pattern may not be unique to

the frequency band characterized by the “usual” Milankovitch

frequencies; it may be shared with sub-Milankovitch, yet still

quasiperiodic, climatic processes.

Figure P90 Power spectra of pre-Quaternary paleoclimate proxy time

series. Top: Thompson multitaper spectral estimate of 3 million year of

Triassic lake level data (depth ranks) from New Jersey, USA, modified

from Olsen and Kent (1996), shaded area is 90% confidence limits;

middle: discrete Fourier transform power spectrum of 1.7 million year of

wt% CaCO

data from the marine Belemnite Marls, England (modified

from Weedon et al., 1999); bottom: linear Blackman Tukey spectral

analysis of grey scale reflectance of a 720 kyr section of Cretaceous

marine strata from the Crimea, Ukraine (from Gale et al., 1999).

PRE-QUATERNARY MILANKOVITCH CYCLES AND CLIMATE VARIABILITY 829

Proxies

Since climate parameters cannot be measured in sedimentary

strata, a reasonably direct recorder of a response to climate

change, or climate proxy, is necessary. For Quaternary sequen-

ces, a vast array of possible climate proxies can be reasonably

calibrated because neither the tectonic plate configuration nor

species composition has changed much over the last 2 million

years. These proxies include d

O, a proxy mostly for ice

volume (e.g., Hays et al., 1976), eolian dust, a proxy mainly

for aridity and wind intensity, and pollen and spores, a reflection

of climate-related changes in plant species distributions. Deeper

in time these proxies can be affected by diagenesis, lithification,

and lack of extant taxa that can be calibrated. Proxies less

prone to time-related processes range from sedimentary facies

classifications sensitive to water depth (e.g., Olsen, 1986)to

geochemical proxies such as d

C of organic matter, carbonate

sensitive to climate-related oceanographic processes, and oil-

shale yield, related to water depth (e.g., Bradley, 1929).

Examples of pre-Quaternary Milankovitch cycles

Neogene

Largely circum-Mediterranean and deep sea core sections are the

basis of a high resolution Miocene to Quaternary Milankovitch-

cycle-calibrated timescale (Shackleton et al., 1990; Hilgen,

1991b; Hilgen et al., 1995, 1997). The Mediterranean marine

sections exhibit the marl-sapropel cyclicity originally described

from Mediterranean cores. The largest-scale bundles of cycles

visible are matched to the 400 kyr of a precession-dominated

isolation curve, in a portion of the section that is already well

dated. Progressively smaller bundles are matched to the insola-

tion curve, followed by precession-related individual marl-

sapropel cycles (Figure P88). Older sections are then spliced.

The hierarchical nature of both the bundling and the insolation

curve limits potential miscorrelations. Radiometric dates have

been used to test the timescale. The astronomical calibration is

so robust that it has allowed for a recalibration of the

Kdecay

constant, itself (Hilgen et al., 1999) now widely adopted.

Hilgen’s method has also been extended to non-marine strata,

particularly in the Miocene (Aziz et al., 2003).

Shackleton et al. (1999, 2000), Palike et al. (2004), and

Billups et al. (2004) have used marine cores showing strong

Figure P91 Evolutive spectral analysis of 22 million year of lake level (depth rank) data from the continuously cored Late Triassic age lacustrine

Lockatong and Passaic formations of the Newark Rift basin, New Jersey, USA. Above: continuous wavelet transform of lake level data (wavelet

provided by C. Torrence and G. Compo; http://paos.colorado.edu / research/ wavelets/ run by M. Machlus); bottom: depth rank data in depth domain

(from Olsen and Kent, 1996).

Figure P92 Change in the period of the Earth’s axial precession due to

tidal friction over the last 900 million years, and the expected changes

in climatic precessional cycles. The sum of the frequency of axial

precession, k, and the various fundamental frequencies of the planets,

g1–g5 (only the most important 5 shown here) is equal to the

frequency of climatic precession. For the present day, there is a cluster

of 3 periods around 23 kyr and another 2 periods around 19 kyr.

Derived from Berger et al. (1992).

830 PRE-QUATERNARY MILANKOVITCH CYCLES AND CLIMATE VARIABILITY