Climatic Change in the Eastern Mediterranean Area During The Past 240,000 Years

The discretionary combination of both data sets provides a qualitative estimate on paleoclimatic changes relative to modern conditions, although the accuracy of any inferred detail can be only as good as various dating techniques applied to the field data permit. Information is more reliable for the past 10,000 years or so, but becomes less so for progressively older time periods, particularly as the accuracy limits of ¹⁴C dating techniques are surpassed. A summary of these inferred changes is presented here with an explanation of the assumptions and constraints applied to compiling the information, as well as a brief discussion of their implication on climatic conditions during the Late Bronze Age.

TECHNIQUES AND CONSTRAINTS ON DATA COLLATION AND INTERPRETATION

• Compilation of Data 

Paleoclimatological data for the eastern Mediterranean area (figure 1) have been summarized in terms of air temperature (figures 2, 3 and 4), atmospheric moisture (figures 5, 6 and 7), fluvial run-off or discharge (figure 8) and fluvial sedimentation cycles (figures 9, 10 and 11). Information has been compiled from the drainage area of rivers and streams draining into the eastern Mediterranean and southern Aegean Seas although comparisons are made with data from several adjoining areas, specifically northern Greece (Macedonia), the Jordan River - Dead Sea rift valley, and some lakes in the east African rift valley. Geographic areas are subdivided into three zones, the northern borderlands (Greece, Turkey), eastern borderlands (Syria, Israel, Jordan, Sinai Peninsula) and southern borderlands (Egypt, northeastern Libya, Sudan, Ethiopia, and portions of Uganda, Kenya and Tanzania). The most complete information comes from the southern borderlands, also the largest in area (figure 1), reflecting the excellent studies along the Nile River and its tributaries by Butzer and Hansen (1968), Butzer (1959) and many others. Water level variations in northeastern African rift-valley lakes are included because some of these lakes which were part of the Nile watershed during portions of late Quaternary time also provide inferences on rainfall variations.

Over 70 references giving paleoclimatic information have been critically reviewed for this compilation. Space limitations prohibit a complete listing of these references here and they will be given elsewhere with a more extensive analysis of the data and its interpretation. With information derived from a wide selection of disciplines - archaeology, marine geology, climatology, geography, paleontonoly, palynology, botany, to name only a few - it is difficult to be critical of reported data without being expert in all of these fields. Too often suggested paleoclimatic conditions are noted without adequate discussions or supporting arguments, but are simply summarized by statements such as "humid during the last interglacial", which then appear to be perpetuated by repeated citation and form a ukase for regional comparisons.

Omitting these wherever possible, there remains a consistency to the published paleoclimatic data, and their similarities do provide an indication of climatic changes that are reasonable within paleoclimatological and geological constraints. Correlation between these changes and glacial stages produces an interpretable sequence of patterns compatible with deep-sea data, modifications in sea-level oscillations and other oceanographic indicators of paleoclimatological variations. These correlations stand ready for correction and are presented here as a summary and as a basis for further modification and discussion.

• Terrestrial Data

A wealth of data exists on climatic changes in the borderlands surrounding the eastern Mediterranean Sea, much representing information derived from published archeological studies of flora and faunal components in non-marine sediments excavated at cultural sites. Additional information has been derived from geological mapping on land of loess deposits, littoral terrace sequences, fluvial terrace sequences, as well as stratigraphic relationships in peat bog and lake sediments. Such information, however, usually infers cyclic or rhythmic changes only - cold, temperate, warm; wet or dry; aggradation or degradation, etc. - but contributes much to our understanding of climatic perturbations over relatively brief time spans and forms an important part of this summation.

Interpretations of these paleoclimatic indicators must consider many other factors and processes, some quite transient and difficult to interpret from the geological record. Comparisons of fluvial terrace sequences between different river systems, for example, and interpretations of their significance, are influenced by numerous factors such as hydrodynamic processes of sediment movement and stream competency, bedrock lithology within the riverbed and the drainage basin, modifications in watershed vegetation cover, tectonic alterations to tributary networks, and stream gradient changes (base level) due to sea level changes. All influence the type and amount of sediment available for fluvial erosion, transportation and deposition, but can be difficult to distinguish. Perhaps the most geologically evanescent climatic category to interpret is atmoshperic moisture. Both humidity and rainfall, two quite different atmospheric attributes, are used synonymously in much of the literature and they have been accordingly combined as "moisture" in this study to describe atmospheric water concentrations, whether as a suspensate, precipitate, or both (figure 13). Humidity and rainfall variations appear somewhat concordant in Figures 5, 6 and 7, reflecting either this implied synonymity in published accounts or an expected natural relationship such as expressed by the presumption of lower evaporation rates during cooler periods leading to reduced rainfall on a regional basis (Flohn 1953; Fairbridge 1962).

• Oceanic Data

The terrestrial climatic indicators summarized here are compared and standardized to the extensive climatic studies conducted on deep-sea sediments, particularly by the CLIMAP (Climate: Long-Range Investigation, Mapping and Prediction) research effort. One of the valuable consequences of this research has been the ability to monitor global ice volume changes, thus indirectly determining growth and melting of northern hemisphere ice sheets (Shackleton & Opdyke 1973; Shackleton 1967; Dansgaard & Tauber 1969) based upon changes in the oxygen isotope (¹⁸O/¹⁶O) content of biogenic components in deep-sea sediments. An oxygen isotope stratigraphy developed from numerous cores obtained throughout the world represents a nearly continuous record of pelagic deposition over the past 5 million years or so with a discontinuous record over the past 100 million years. This stratigraphy has been correlated to the biostratigraphic and paleomagnetic stratigraphy in deep-sea sediments and land sections, both well-dated by geochemical dating techniques.

Accuracy of this ¹⁸O record is within the range of + 5000 years to -1000 years (Hays et al. 1976) due to oceanic water-mass mixing phenomena of about a 1000 years duration, to mixing of sediment by burrowing animals when older stratigraphic horizons formed the surface sediment, to dissolution of foraminifera texts, and to displacement of pelagic material by currents or slumping. Thus the ¹⁸O record can only provide a smoothed temperature or climatic curve, but forms a valuable criterion for comparison with terrestrial data, particularly where ¹⁴C dating techniques are inadequate. Age ranges for the ¹⁸O stages shown in Figure 12 are from Hays et al. (1976); the generalized ¹⁸O curve and stage designations have been taken from Shackleton and Opdyke (1973; 1976), Emiliani (1955), CLIMAP (1976), and Ninkovich and Shackleton (1975).

Sea-level fluctuations are an additional indicator of major climatic changes. Raised littoral terraces representing higher sea-level stands have been adequately dated for determining ocean level fluctuations through time, although complications due to tectonic activity and subsidence or progradation of continental margins by terrestrial sediment influx (Ryan 1976; Pitman 1978) must be considered. In the Mediterranean, for example, it is difficult to interpret uplifted marine terraces with respect to absolute elevations of sea level stands because of tectonic activity throughout the Pleistocene and Holocene in response to the continued underthrusting of the European lithospheric plate by the African plate, and the resulting movements by smaller lithospheric plates in the eastern Mediterranean area (Flemming 1978; and many others). With the exception of the late Miocene (Messinian) event some 5 to 6 million years ago, sea-level variations in the Mediterranean have been concordant with world ocean variations, thus the world-wide sea-level curve has been utilized for comparison to, and calibration of, the paleoclimatic trends summarized here. The sea-level curve depicted in Figure 12 is adapted from Bloom et al. (1974), Harmon et al. (1978), and Fairbanks and Matthews (1978), with the 0 - 30,000 year B.P. portion from Curry (1965) and modified according to McIntyre et al. (1978).

• Magnitude and Rapidity of Climatic Changes

As noted previously, much of the data on climatic fluctuations are qualitative and can only indicate cyclic or rhythmic modifications. Some quantization of the magnitude and rapidity of these modifications is possible, however. Temperatures can be estimated by reference to ¹⁸O values determined in marine sediments from oceanic areas or those on raised marine terraces, in calcareous cave deposits, as well as by the assumed ecological conditions satisfying biological residues from particular stratigraphic horizons such as pollens, spores, animal remains, etc. Moisture levels can be deduced by assuming oceanic atmospheric heat transfer, coupling and circulation relationships similar to those at present. Fluvial run-off can be assessed from the temperature-moisture information, and its magnitude inferred from geological mapping of fluvial terrace sequences suggesting aggradation-degradation cycles, presuming a minimal lag in response time between rainfall, run-off and depositional-erosional cycles.

Climatic changes may have been quite rapid, perhaps on the order of only a few hundred years. Sarnthein (1978) notes, for example, that modern west African dune fields became re-activated only 300 - 400 years ago; Portuguese explorers described a summer rainy season here that now occurs some 300 miles farther south. Sea-level fluctuations may occur at rates of 5 to 10 cm/1000 years (Harmon et al. 1978). Such short-period variations are difficult to interpret from the deep-sea record and thus the terrestrial data derived from non-marine sediments, whether in peat bogs or in archaeological excavations, are important.

Climatic variations in this paper are referenced relative to modern conditions. Suggested magnitute of change represents an interpretive response to inferences made in literature or to ¹⁸O values, but a scale of absolute paleotemperature values has not been attempted. Although rates of paleoclimatic change can be quite rapid, and those depicted in Figures 2 - 13 are not unreasonable with respect to contemporary indications, implied rates in some cases could represent an artifact of the data availability.

• Correlation with Pleistocene Time-Stratigraphic Subdivisions

The climatic deterioration leading to Plio-Pleistocene conditions commenced about 38 million years ago in the late Eocene (Kennett & Shackleton 1976), and the transition of glacial-interglacial fluctuations in the northern hemisphere began in the Pliocene about 3.2 million years ago (Shackleton & Opdyke 1977). Kukla (1977) has delineated seventeen glacial and seventeen interglacial stages during the past 1.7 million years, eight of which occurred during the past 700,000 years. The classical Alpine and northern European glacial terminology describes only four major glacial and four interglacial stages over the past 700,000 to 800,000 years, thus accounting for only some 15% of the major cooling sequences (Kukla 1977).

These problems, plus the difficulties in correlation between the Alpine, northern European and north American subdivisions (Woillard 1978; and others), despite the fact that large-scale climatic changes during the Pleistocene were by and large synchronous world-wide (U.S. National Academy of Sciences, 1975), are so severe that Kukla (1977) has "urgently recommended" the abandonment of the classical terminology "in all interregional correlations and to base the chronostratigraphic subdivision of the Pleistocene on the ¹⁸O record of deep-sea sediments". Accordingly, correlations are made here to the oxygen isotope record and no attempt has been made to identify specific stadial or interstadial sequences - those cited are mentioned for comparative reasons only and as a convenience for discussion purposes.

Dates on glacial-interglacial cycles in this paper are taken from Kukla (1977). The last glacial stage (Wisconsin; Würm) is given an upper boundary of 10,000 years B.P. Dates on the lower boundary of this stage remain in dispute and it is graphically shown here to be somewhere between 115,000 years B.P. and about 70,000 years B.P., the older date being preferable.

LATE QUATERNARY PALEOCLIMATOLOGY

• Temperature

Comparison of generalized temperature curves, derived from implied paleotemperature variations in published accounts for the northern borderlands (figure 2), eastern borderlands (figure 4) and southern borderlands (figure 5), correspond well with climatic variations deduced from central Europe and from deep-sea criteria (figure 12). In particular, correlation to the loess stratigraphy in central Europe (Kukla 1977) during the last glacial stage appears excellent, with warmer conditions in the Mediterranean borderlands corresponding to forests in central Europe and cooler conditions corresponding to episodes of steppe vegetation and loess deposition. These indicated warmer conditions for the eastern and southern borderlands graphically appear to encompass a significantly longer time-span in the figures, but only because the generalized eastern Mediterranean paleotemperature curves have not been drawn conformable to a lower time boundary of 115,000 years B.P. for the last glacial stage, whereas the central Europe and oceanic data are. Fine scale perturbations between different circum-Mediterranean borderland areas are difficult to delineate on a comparative basis, primarily because paleotemperature information for the northern borderland area is more detailed due to palynological studies on peat bog sediments in Macedonia (Van der Hammen et al. 1965, 1972; Wijmstra 1969).

Absolute air-temperature differences between those at present and those during the latest portion of the last glacial stage in this portion of the Mediterranean were about 4° to 6° C (Emiliani et al. 1963; Schroeder & Bada 1973), while Mediterranean Sea surface-water temperatures were about 1° to possibly as much as 8° C lower than modem conditions (CLIMAP 1976; Luz & Bernstein 1976; Cita et al. 1977; Thiede 1978).

• Moisture

There is some conformity to the somewhat disparate indications of humidity and rainfall variations. A period of pronounced aridity is evident between about 9,000 and 20,000 years B.P. during the later portion of the last glacial stage and the earliest Holocene. Numerical modeling of climates by Gates (1976) and Manabe and Hahn (1977) utilizing global temperature distributions 18,000 years B.P. (CLIMAP 1976) implies extensive aridity in tropical areas between about 30° N and 30° S, where sand dunes and deserts occupied nearly one-half of the world's land area (Sarnthein 1978), with narrow belts of tropical rain forests and savannah conditions compressed between this arid zone and the ice front (Newell et al. 1975; Williams 1975; and many others). Lake levels in east Africa (figure 13) also indicate this arid phase which was followed by a considerably wetter climate in the Holocene allowing these lakes to drain into the White Nile (Street & Grove 1976; Grove et al. 1975; Williams & Adamson 1974; Degens & Hecky 1973). Humid conditions here appear to have preceeded those in the circum-eastern Mediterranean area by a thousand years or so.

Overall aridity during glacial stages does not seem to be implied from this compilation, however. More moist conditions during earlier portions of the last glacial seem to be indicated for the eastern Mediterranean area, and possibly in eastern Africa as well (Gasse & Rognon 1973) - an interesting remnant or resuscitation of the Pluvial concept.

• Fluvial Run-off and Sedimentation

Inferences on variations in river and stream run-off during the late Quaternary, a transient phenomenon interpretable from moisture modifications and fluvial terrace deposits, is shown in figure 8, with the best data from the Nile River system and Dead Sea area. Cyclicity in the magnitude of fluvial run-off between the eastern and southern borderlands appears to be somewhat discordant during the latter portion of the last glacial stage, due perhaps to the more detailed studies in the latter borderland area but possibly reflecting brief climatic alterations in the Nile headwater regions. For most of the remaining late Quaternary - the earlier part of the last glacial, the Holocene-Pleistocene transition, and the Holocene - a closer correspondence exists between the run-off variations in various borderland areas with a distinct decrease in fluvial discharge evident about 20,000 to 9,000 years B.P. during the pronounced arid phase.

During the Holocene, numerous short-period alterations between wet and dry episodes seem to be reflected by persistently low fluvial run-off except for the Nile again, an indication of this river's source in a different climatic zone from most of the eastern Mediterranean area.

Modification of fluvial sedimentation cycles in response to fluvial run-off changes and wet-dry atmospheric changes does not appear to follow a persistent pattern (figures 9, 10, 11 and 13). In part this probably reflects inadequate dating of stratigraphic relationships in fluvial terraces, particularly in the northern borderland areas. A distinct depositional phase corresponds to diminished river discharge during the pronounced aridity between approximately 9,000 - 20,000 years B.P., and brief erosional cycles do correlate with increased atmospheric moisture and increased fluvial run-off during portions of the Holocene. Over longer time spans, on the order of a few thousand years, comparisons of moisture variations, fluvial run-off and fluvial sedimentation curves (figure 13) often also show a broad relationship between increased rainfall, river discharge and fluvial degradation, in agreement with assumptions by Fairbridge (1962) for the Nile. These are complex relationships, however, and cannot be further interpreted by the data summarized here, as has been noted previously. It is also interesting to note the differences between the upper (including the Nubian) and lower Nile, which reflect the two distinct climatic zones the river flows through now, as in the Pleistocene. Unfortunately data are adequate only for distinguishing very general atmospheric changes that accompanied the transition from glacial conditions to interglacial conditions between these two climatic zones, such as in the northern migration of higher moisture levels and their influence on fluvial processes.

INFERENCES ON LATE BRONZE AGE CLIMATES

Paleoclimatologic reconstructions inferred from these compilations do not imply a significantly different climate for the southern Aegean area during the Late Bronze Age as compared to those today. At the time of the eruption and caldera colapse on Thera, ca. 3430 years B.P. (1450 B.C.), air temperatures appear to have been only somewhat warmer than at present (figures 2, 3, 4 and 12). How much warmer cannot be determined from these data, but calibration of the inferred difference in temperature change to estimated maximum temperature declines of between 4^o and 6^o C in the Mediterranean area about 18,000 years ago from oceanic and terrestrial oxygen-isotope measurements (Emiliani et al. 1963; Schroder & Bada 1973; CLIMAP 1976: Cita et al. 1977; Thiede 1978), might suggest that air temperatures were possibly warmer by about 1^o C, at best, about 3500 years B.P. Average sea-surface temperatures decreased between 1^o and 4^o C, summer and winter estimates, in the seas surrounding Thera at the peak of the last glacial about 18,000 years ago (CLIMAP 1976; Thiede 1978), and by the Late Bronze Age sea-surface temperatures had probably risen to values close to those now, or possibly slightly higher in response to the warmer conditions suggested.

Information concerning atmospheric moisture changes is incomplete for the northern borderlands, which includes the southern Aegean area, but slightly more arid conditions about 3500 years B.P. than exist today are implied (figure 5). In the eastern and southern borderlands, increased aridity is apparent (figures 6 and 7), which is also suggested by the continued drop in east African lake levels (figure 13). Generally the late Bronze age in the Mediterranean area appears to have been characterized by decreased river discharge and a nondepositional cycle in fluvial sedimentation processes (figures 8 - 11, 13; see also Vita-Finzi 1969). Although data from Greece are conflicting (figures 9 and 13), new information from Attica of non-fluvial activity in rivers such as the Haradros between early Neolithic and Mycenean times would also imply somewhat increased aridity during Minoan times (Paepe & Meyer 1978).

It is therefore not surprising that indications of significant modifications in paleoclimatologic conditions during the late Bronze age on Thera are not interpretable from existing data there since the climate then was probably only somewhat warmer and more arid than that today. Accordingly, soil types and vegetation would not be much different from those at present, with due consideration for cultural modifications, which would be in agreement with conclusions reached by Davidson (1978; 1978a) in studying Minoan paleosols.

FINAL STATEMENT

It must be emphasized that all paleoclimatic indicators presented here are summarized from numerous published accounts, most of which cannot be identified due to space limitations in this volume. Numerous inconsistencies in these terrestrial data remain and are not discussed in detail, particularly in relationship to climatic oscillations interpretable from deep-sea data and to glacial-interglacial boundaries. Criteria suggestive of paleoclimatic conditions during older time periods is generalized and often vague; for example, persistently warmer conditions during the last interglacial stage is not particularly realistic nor does it imply that fine-scale variations did not occur as they have for the more recent past. Data from terrestrial areas simply does not allow this resolution yet. Time control is also often inadequate and what is depicted here represents geologic ages inferred by those who reported the information. Suggested paleoclimatology conditions during increasingly older time intervals assume landforms and tectonic activity similar to those today, whereas they could have been considerably different.

Nevertheless, this is a summary of data reported in much of the literature. Any interpretations must be done with constraint; clearly there is much work yet to be done.

------------------------------------------

----------------------------------------