Sea-levelchange and shore-line evolution in Aegean Greecesince UPPer Palaeolithic time
Kunr LAvtgncr*
- not been 'As the glaciation ended, the ice melted and the sea-Ievelrose'' Yes but it has surface-Ioads as simp"leas that, ss the'Earth has adjusted in several ways to the changing in a it siffers under ice and under weight of water. The important issuesare set out Greeceand simple'mathematical treatment, and theii varied consequencesare shown fot islands, where the impact on Lspecialtyfor the Greek coastalplains and the Greek human settlement has been latge'
about 20,000 to Nature and consequences ofPostglacial sea- the Last Glacial Maximum, lowerthan level change 18,000 years ago, were sufficiently plains that have Sea-Ievelshave changed significantly since Late today tó leave exposed coastal appearc to have Palaeolithic time, primarily in response to the since flooded. But less attention and rates of change melting of the gréat ice-sheets that covered been focused on the timing the great ice-sheets. northern Europe and North America. These ice- after the onset of melting of - with the exception sheets*et" of a sufficiently large volume that, What discussion there is by van Andel & upon melting, sea-level rose globally by about of the important paper (1982) (see van Andel tg89J tàoaso m. Releaseof this ice into the oceans Shackletor also - that this was initiated at about 18,000 years before present often leavesthe distinct impression quickly with (b.p.), although the majority of the-melting oc- change in level occurred early and cuired betwebn about 16,000 and 8000 b.p.1 rather minimal human imPact. using the Aegean Ratesof global sea-levelrise reached 15-20 mm This paper sets out to describe, model of p*r y"ut-dn.ing this interval. As the sea-levels Sea region as an example, a realistic migration for the ior", to did the shorelines migrate with time, sea-IevLl change and shoreline a frame- at rates that for some low-lying regions reached past 20,000 ye¿Irs,one that canprovide of such chalge on about a kilometre per year. Examples of where work for discussing impacts such rapid ettcro"chtoent of the sea occurred huma¡r movements ald settlement. ice sheets, the include the Persian GuIf and the Gulf of If, during the decay of the uniformly over Carpentaria of northern Australia, between about meltwater vòIume is distributed change at time f 12,ô00 and t0,000 years b.p. In some areasof the oceans, then the sea-level the world the sea-Ievelspeaked at about 6000 would be years b.p., inundating now low-Iying areasbe- A("(t)= changein oceanvolume/ocean surface area (1) iore falling slowly to their present position, The consequencesof these changes on human.set- 'eustatic This sea-levelchange'is a function of tlement ánd movement have been recognized time. It represents only a zero-order approxi- in the archaeological and pre-historic records' mation because sea-Ievel does not respond Thus it is widely accepted that levels during uniformly to the melting of the ice caps: the rates of rise are spatially variable, and in some are in conventional radiocarbon years unless 1 All ages may actually be falling rela- otherwise specified. Iocalities sea-levél
ACT 0200, Australia' * Research school ofEarthsciences, Australian Nalional university, canberra
Received 23 January 1996, accepted 26 April 1996' ANrrQUrrY 70 (1996): SEA-LEVEL CHANGE & SHORE-LINE EVOLUTION IN,A,EGEAN GREECE SINCE UPPER PALAEOLITHIC TIME 58S tive to the land. This is a consequence of the Because of the viscous nature of the planet's adjustment of the Earth to the changing sur- interior, the crustal defo¡mation continues lons face-Ioads of ice and meltwater. fhé eãrth,s after the melting of the Late pleistocene icel response can be ilescribed as that of an elastic sheetshas ceased;and shorelines have contin- (the layer lithosphere which includes the crust) ued to change up to the present. overlying a viscoelastic mantle. When changes This combined behaviour of the eustatic in the mass distribution occur on the Eart-h's change a¡ld the crustal rebound describes a first- surface,the lithosphere and mantle respond to order solution to the question of what happens the new stress state in different *uys: elastic to sea-levelwhen ice sheetsmelt. A more ôurrr- deformation primarily occurs in the lithosphere plete description requires consideration of two and viscous flow occurs prirnarily in thè un- additional effects: derlying mantle. The characteristic time-scale . the fact that newly added meltwater loads of this flow is of the order of a few thousand the sea floor and stressesthe mantle. re- years. The deformation of the Earth's surface sulting in further mantle flow and surface under a changing load therefore exhibits both displacement; an instantaneous elastic responseand a viscous r the gravity field of the earth is changing; first, response. Such behaviour iì well documented because of the changing gravitalioñal at- by other geophysical observations: gravitational traction between the ice, water and land attraction of the Sun and Moon raises tides in as the ice sheetsmelt, and second,because the solid Earth; ocean tides load the sea floor of theredishibution of masswithinthe solid and contribute to the deformation of the Earth's Earth in response to the mantle flow and zurface; atmospheric pressure fluctuations over surface deflection. the continents induce deformations in the solid The total effect ofthese various processesis Earth. These displacements, when measured called glacio-hydro-isostasy'after the (regional) with precision scientific instruments, show both isostatic response of the Earth's surfacJto the an.elasticand a viscous component, with the ice and water loads. It leads to a complex spa- latter becoming increasingly important as the tial temporal pattern of sea-level change duration of the load or force .and increãses.Changes as the ice sheets wax and wane, making tñe in the ice and water loads occurring a! times concept of a uniform global eustatic change of of major deglaciation are much largãr and oc- only very limited value. This complexity is ñtore cur on longer time-scales than these small at- than a scientific curiosity, as iti understand- mospheric and tidal perturbations, butthe tleory ing has broad scientific impacts. In geophys- underpinning the formulation of the surface re- ics, for example, the spatial changesinieallevel bound problem is similar and has been tested provide a valuable insight into physical prop- against a range of different observations (e.g, erties of the Earth that cannot otherwiie 6e Lambeck 198s). measured. In environmental studies they pro- When the ice-sheets melt, the surface load vide a framework for separating naturai 6om is changedin two ways. Ice is removed, thereby man-caused changesin sea-level.In prehistory unloading the formerly glaciated crust, and thä and archaeology, they provide a baiis for de- meltwater is added into the oceans,loading the tailed geographic reconstructions of coastal oceanic _lithosphere. The resulting changã in regions and for assessingthe impact of chang- sea-level relative to the land is a cãmbinátion ing sea-levelson past societies. of two reactions: the increase in ocean volume, and the deflection of the land surface.Depending Models for sea-level change on which contribution is greater, sea-level ii The evaluation of the evolution of shorelines seen to rise or fall relative to the land. In re- during-times of glaciation and deglaciation re- gions near or within the former centres of gla- quires knowledge of several quantities: ciation, the crustal adjustment is one of uplift; ¡ the change in sea-level relãtive to the land it is usually the more important, and sea-level for the region in question, a change that here is seen to fall. Further away from the ice is usually regionally variable; sheets,the crustal rebound is small and gener- . the present topography of the low-lying land ally one of subsidence;here the chanqeinãcean areas and shallow sea-floorbathymetry for volume dominates, and sea-levelis séento rise. the region; KURT LAMBECK
= a("(t)+^Çía,Ð+AÇ@'Ð (2) . any information on tectonics, sedimentation ^ç@,t) or erosion that may have changed the ge- where A("(t) is the previously defined-eustatic ometry of the coastal zone. "(t) sea-Ievel , AÇ,@,t)is the combined glacio- Because of the spatial variability of the sea-level hydro-isostatic contributi on, and A(r(E,t) is any change, the beJt way to establish a record of additionai tectonic contribution. This last con- the pãst positions at a particular location is to tribution is discussed further below and the infei it frbm the geoiogical, geomorphological emphasis here is on the isostatic function, which and archaeological record. But observations that b" written as the sum of three parts: permit a quantitative estimate of the changing "ati exist for the þosition olfthe sea surface seldom Açía,Ð = ^Ç@,Ð+ ^(,@,Ð + a(*(tP,t) t3l ärea of interest and - at best - it probably consists of a partial time series. Also, such The total sea-levelchange is expressedas: records become increasingly rare and impre- cise the further one goesback in time. What is AKa,t)= A("(t) + A(,,@,t)+ Aqq,t) + Aç,@,t)+ AÇ{EÐ t+l required is a physical model for inte-rpolation in is a function of beiween the available fragments of information The eustatic term A("(t) [a) volume a¡rd its evalu- so as to be able to predict sea-Ievelat any place the rate of change in oðean of the rate of melt- and time. For tectonically stable regions the ation requires ã knowledge ice caps. The second glacio-hydro-isostatic theory provides- a rela- ing of thi totality of the the departures of sea- [ively smoothly varying and predictable func- rclrn A(,(E,t) describes on a rigid planet by tion ôf such change in time and spacethat can level chänge lrom eustasy in gravitational attrac- be used to interpolate between isolated obser- allowing for the change water as the ice sheets vations. But when active vertical tectonic proc- tion between the ice and water and land as the essesare important, resulting in surface faulting melt, and between the During times of ice- or warping, the interpolation model beco-mes ocean volumes increase. gravitational at- -rrch *o.ã difficult becausethe crustal displace- sheet growth, the increasing water towards the ments can change abruptly over compalatively traction of the ice pulls the inthe absenceof the other short distancesa¡d need notbe uniform in time expanding ice dome; rises in the neighbour- when viewed over periods of the order of mil- coiritibutiotts, sea-Ievel away it falls. /Ç lennia. hood of the ice while further the ice sheet geom- The approach used here is to develop mod- is therefore a function of of the geometry of the els of uútì"ty and glacio-hydro isostasy for etry through time and the water is extracted' tectonically stable areas so that physical pa- ocõan basin lrom which in areasclose to the rametets describing the Earth response and the This term, most significant out to con- ice sheetscan be evaluated' The result is a com- ice sheets, remains non-negligible the ice margins' The prehensive portrayal results of gìobal sea-level siderable distancesbeyond ice is still sufficient ähung" in response to the growth and decay of attraction by the northern to 'pull up'the Medi- the lãrge ice sheets. When these models are in Greece,fôr example, this correction is positive apptied to tectonically active areasany depar- terranean water, and (see Ftcunn 2a be- tuies of observed sea-levels from the model- at times of major glaciation early epochs would predicted values can be attributed to tectonic low); sea-le-teis Jt these the present if the other iactors. The successof this approach does de- have been higher than at The major pend on whether the isostatic predictions are contributions had been ignored. Fennoscandian ice, but iepresentative of the region ulder considera- contribution is from the the North American and üón. If this can be d.emonstrated,then the model contributions from are not negligible if sea- can be used to predict the positions of former Antarctic ice sheets to a metre or less are shorelines and palaeo-water depths. level models accurate sought. (+) describes the co-nse- The sea-level equation The term l(,(E,Ð in the deformation of the In the presence of vertical tectonic motions of quence on sea-ievel of ice load; it includes the Earth's surface, the relative sea-Ievel change Ëarth under the changing in the gravity field Aç@,t) at a site g and time ú can be written a contribution from changes redistribution of mass within schematically as: associatedwith the SEA-LEVEL CH,TNGE & SHORE-LINE EVOLUTION IN ,TEGEAN GREECE SINCE UPPER PALAEOLITHIC TIME 591
'1,. FrcuRn Examples of predicted (solid lines) and observed (with error bars, right-hand side) relative sea-level change in late- and post-glacial times from different localities. The curves on the fight- hand side illustrate the total predicted change and the obsented change at selected sites. The geographical variation in these changes is inter- preted as the combination of the eustatic change (indicated by e.s.l. on the Left-hand side) and the gl a ci o -hy dr o -i so st ati c adjustment of the crust in response to the glacial o unloading (indicated by ice the glacio-term n for o and water for the hydro- q term). d a The Angerman River, Sweden: the glacio- isostatic rebound domi- nates over the eustatic change. b The upper Forth Valley, Scotland: the isostatic and eustatic -80 tems are of comparable magnitude but opposite -ró signs. -t2Ã c Southern England: the dominant contribution is -140 the eustatic component but the glacio- and hydro- i so stati c c ontributi ons are also significant. t0 d Karumbø in the Gulf of 0 Carpentaria, Australia, a continental margin site :10 -20 far from the former ice Ioads: the dominant -30 isostatic perturbation is 40 the hydro-isostatic 0 108 component. Time(x1000 years b.p.) KURT LAMBECK
'glacio-isostatic times. At continental the deforming planet. This ice sheets at postglacial hydro-isostatic correction' is a function of the elastic and vis- margin sites, the characteristic as the seafloor slowly cous properties of the Earth, as well as of the signãl is a falling sea-Ievel gibrid"r water load (Frcuns td), temporal and spatial distribution of the ice under the new the past 6000 years does it sheeis.This term is also most important for the but only for about contributions; relative former areas of glaciation where the crust has dominate over the other higher than today from been depressedby the ice load. As the ice melts, sea-levelswill have been to the present. The'am- the crust rebounds at a rate determined by the about 6000 years b.p. are small, up to-about mantle viscosity and by the ice thickness' For plitudes of the highstands coastal geometry,but in large ice loads, crustal rebound exceeds the ã m depending on the such as lower Meso- eulatic change so that sea-level appearsto fell flat and low-Iying areas shore of the GuIf of relative to the land. This is seen in the GuIf of potamia or thê sõuthern have led to substantial Bothnia, for example, whete there are raised ðarpentaria, this can time. shoreline features whose ages increase with inundation in mid-Holocene aII three contributions height (Ftcun¡ 1a). Rates of relative sea-level For the Aegean region (3) are important, as FIc- falún excess of 20 mm per year have been in- to the isostatic term model discussed in de- ferred for some localities. Because of the vis- un¡ 2 illustrates, for a (Lambeck tggsa). Flcunn 2a cous nature of the mantle flow, the rebound tail elsewhere in the rigid term- /Ç and relative sea-Ievel change continues long illustrates the variation and 10,000years b'p' after deglaciation has ceased.For small ice sheets over the region at 18,000 is positive, indicating that such as that formerly over Britain, or regions This corrective term on the water by the ice near the margins of the ice sheets, as in south- the gravitational puII important out to distances ern Norway ór Denmark, the crustal rebound on"t Fenttoscandia is from the centre of the and the chánge in relative sea-Ievelwas much of about 2000-2500 km the ice load term smaller, the lãtter remaining within a few tens ice }oad. FIGURE2b illustrates This term is nega- of meires of the present level since Late Gla- A(.for the same two epochs' subsidence upon cial time (Flcun¡ 1b). Hete, the crustal rebound tiíe in this region ofirustal (c/. 1c), and partly can- is less than for the sites closer to the former ice unloading Frcun¡ The water-load term is centres of ice loading; it is of comparable mag- cels out the rigid term. This correction is mainly nitude but of opposite sign to the eustatic con- illustated inFicunn 2c. and negative at sea,and tribution. Further away again, beyond the areas positive over the land follow the coastline, their of former glaciation, the response of the crust ihe contours tend to illustrating the fil- to the rem-oval of the ice is one of subsidence relatively smooth character lithosphere on the in reaction to the flow of the underlying man- tering efiect of the elastic tle material towards the formerly loaded areas' crustal deformation. terms in the Here sea-level appears to be rising to the present, Together, the three corrective in sea-level at the time of even though all melting ceased much earlier' Aegeãn region result varying between -115 This is the case;for example, in southern Eng- the Last GIaciaI Maximum -13s to the eustatic value of land, or along the Atlantic margin of France and m, compaled -125m 10,000 years (Ftcuna 1c). In the Aegean, sea-Ievel also ap- about for this epoch' At over the region-v¿lry pears to be rising throughout the-Holocene be- ago the predicted sea-Ievels -43 -55 to a eustatic level äause of the ctnJt"I response to the melting 9f, from to m, compared -43 amplitudes and primarily, the Scandinavian ice sheet. of about m. The actual patterns depend on the The hydro-isostatic term lÇ.(P,Ð describes details of the spatial parameters describe the ice load and the contiibution to the sea-level change from used to the overall trends and the meltwater loading of the crust and the as- ihe narth's rheology but of the departures sociated change in gravitational potential' The magnitudes are representative from the simple eustatic term is a function of the Earth's elastic and vis- of tÃe sea-levelchange cous parameters (its rheology) and-of the rela- approximation.'bther temporal and spatial tive Jea-Ievel change itself, as well as of the than predicting the glacio-hydro-isostatic shifting coastlines' SmaII compared with the changesin sea.Ievel, location of past shore- maximum glacio-isostatic and eustatic terms, model also predicts the depths if the present it becomes dominant in areas away from the' lines and palaeo water SEA-LEVEL CHANGE & SHORE-LINE EVOLUTION IN AEGEAN GREECE SINCE UPPER PALAEOLITHIC TIME 5S3 18,000b.p. 10,000b.p.
b
d FIGURE2. Spatial variation of the ri.gid (a eb), glacio- isostatic (c ù d) and hydro-isostatic (e ù f) contributions to 40' relative sea-IeveÌ change across Greece and the Aegean at 1-8,000(a, c, e) and L0,000 {b, d, fj ¡p. The first two compo- nents are for the northern hemisphere 36" ice-sheet only; the third component is for the meltwater from both hemi- spheres. KURT LAMBECK
such as the water depth H"(E) at location E is krrown. That tions a¡e sought for a region is, the water depth at a time f and position E eastern Mediterranean. This is usually are related to AÇ@,t)by achieved by iterating between equations 4 and 5, H(E,t) = H.(tP)- t((E,Ð (5) The eustatic sea-level curve To evaluate the spatial and temporal distribu- Published estimates of the eustatic sea-Ievel tion of sea-Ievel and shoreline change during curve Érrevaried, the reasonsfor which are sev- a glacial cycle several requirements have to be eral. Some are associated with interpreting of met. These include: the observational evidence. Corals, as for ex- . A model of the eustatic sea-level function ample used to estimate the sea-Ievel curve at A( (Ð -uguallv inferred from the sea-level Baibados (Fairbanks 1989), provide estimates oú"servations ihemselves because inde- of a lower limit to the sea-level curve, while pendent estimates of the volumes of the submerged in situ terrestrial vegetation such former ice sheets do not exist. These esti- as fresh-water peats or tree stumps provide an mates are derived from localities where upper limit. Estimates based on the depth-age the isostatic corections are believed to relãtions of molluscs are often unreliable: their be small such that they can be estimated original depth-range relative to mean sea-level in a¡ iterative manner (Nakada & Lambeck is large, or they may have been transported from 1988). Many published estimates of the their growth position to new levels. The dif- eustatic sea-Ievel curve have ignored the ferent published results may also reflect the isostatic contributions: the resulting er- neglect, or inadequate elimination, of the glacio- rors may be particularly important when hydro-isostatic factors. For example, a number estimatesare based on sea-Ieveldata from oi the still-quoted estimates used in sea-Ievel the North Atlantic margin of the United studies in the Mediterranean are based on old Statesof America where the glacio-isostatic observationb from the Atlantic coast of North effects are significant, America (e.g. Curray 1965; Milliman & Emery . A detailed description of the growth and 1968) where the relative sea-Ievelis influenced decay ofthe ice sheets,required to calcu- by the glacio-isostatic rebound of the Laurentide Iate the glacio-isostatic term. For locali- region such that the observed levels in Late ties near or within former ice-sheetlimits, Glacial time generally lie above the eustatic a detailed description ofthe evolution of curve (c/. FIcuna tc). the ice load is required whereas for re- Across continental margins, corrections for gions away from the ice margins, such as the hydro-isostatic factors in Late Glacial times ihe Mediterranean, approximate spatial are also significant, reaching 20-30 m at the (Lambeck descriptions of the changing ice-sheets time of the Last Glacial Maximum & suffice. Nakada 1990). Observations from islands far A model of the Earth's rheology, describing from both continental margins and ice sheets the responseof the planet to surface loading orovide better estimates of the eustatic sea-Ievel, on time-scales of thousands of years. The 6ut even these are not immune to the isostatic elastic behaviour of the Earth is well knor,'¡n factors (e.g.Nakada & Lambeck 1987; Mitrovica from seismological studies but the visco¡us & Peltier 1991). As the establishment of a glo- properties, IessweII known, are estimated bal eustatic sea-Ievelcurve requires models for from comparisons of isostatic-model pre- the isostatic corrections,the procedure adopted dictions with observations from regions here is to apply the model (4) to areas of the of otherwise tectonic stabilitY. world where other tectonic processesproduc- A description of the coastLinegeometry and ing vertical movements a¡e believed insignifi- the shallow-water batþmetry, required to cant, or where corrections for these movements estimate the palaeo-shorelines through the can be independently made (e.g.Lambeck rgge; relation (s). The time-dependence of this 1995a). Ftcunr 3 illustrates the best estimate coastline must also be considered in the of the eustatic sea-level curve since the time of calculation of the hydro-isostatic term, the Last Glacial Maximum; it is based on analy- particularly if high accuracy local solu- ses of sea-level change from locations far foom SEA-LEVEL CHANGE & SHORE-LINE EVOLUTION IN AEGEAN GREECE SINCE UPPER PALAEOLITHIC TIME 595
L.Ig K_ W ______>lH
õ40 c) g -60 o
(! -Rll o S -100
-lû 150 50 time (x1000years b.p.)
FIGURE3. Estimate of eustatic sea-Ievel change (dashed curve) for the past 150,000 years based on obsen¡ed shore-Iine age-depth relations and models of ice-sheet melting for the lasi 18,000 yeaß and on oxygen isotope data from deep-sea cores (after Shackleton 1gB7) for the earlier period. The solid cunte teptesents the predicted sea-level variation for Póros in the Cycladean island group (this prediction is discussed below). (L.G.M. = Last GIaciaI Maximum, L.Ig = Lss¡ Trterglacial, S = SaàI¡an Giacial Maximum, W = Weichselian or Mousterian, H = Holocene).
the ice sheets, along the continental margin of for such oscillations once the major ice sheets Australia and elsewhere. This result is consistent have melted or approached their present vol- with the upward coral reef grov\,'th-rate curve umes (Chappell 1983; Hyvärinen 1980). Like- from Barbados (Fairbanks 1989) when that is wise, estimates of the eustatic sea-Ìevel function 'l,2,OOO corrected for the isostatic perturbations, as well for the period between and 6000 years as with the results from the rapidly uplifting b.p. suggestthat any oscillations in this inter- Huon Peninsula of Papua New Guinea (Chappell val are likely to be less than about 2 m (Lambeck & Pollach 1991.Jonce tectonic and isostatic 1993;in press). corrections are applied, Major contributions to Becauseof the viscous natu-reof the response this globally integrated rate of melting comes of the Earth to changing surface loads, the from the Laurentian, Fennoscandian, Barents isostatic contributions to the sea-level change Sea and Antarctic ice sheets. are effective long after the changes in loading One characteristic ofthis sea-level curve is have occurred, as is best seen in the on-going that the bulk of melting ceased at about 6000 uplift of the Fennoscandian area (Frcuru La). years ago when the last ice from Laurentia van- The prediction ofsea-Ievel at any period, there- ished. (The Fennoscandian ice sheet vanished fore, requires information about the earlier ice soon after 9000 years b.p.) But a small amåunt sheet histories. For studies of sea-levelnear the of additional meltwater continued to be added time of the Last Glacial Maximum, this requires into the oceansafter this time, probably from a a knowledge of the ice sheetsduring their growth small reduction in the Antarctic ice volume, phase. As this information is generally scanty, so as to raise the eustatic level bv about 2-3 m reliable models for the changing volumes of during the past 6000 years (Nakada & Lambeck the individual major ice sheetsbefore the Last 1988), Another characteristic of this eustatic Glacial Maximum do not exist. Estimates of the sea-level curve is that the function is devoid early eustatic sea-level curve also cannot be of rapid oscillations. Detailed studies from the established from sea-level information alone same or close-by sites of sea-level change dur- because of the great paucity of older observa- ing the past 6000 years indicate little evidence tions preserved in the geological record and KURT LAMBECK which can be reliably dated. Instead, an ap- the glacial maximum (Zwartz et al.inprepara- oroximate estimate of this function is obtained tionJ but the evidence is inadequate to estab- ho- oxygen isotope records of the sea-floor lish a quantitative model. Instead, inferences sediments scaledby the more directly observed of past Antarctic ice volumes arebased on more changesfor the past 18,000 years (e,g.Shackleton indirect indicators, such as the need for the total 1987; Chappell a Shackleton 1986)' FIGURE3 ice volume to correspond to the observed illustrates this result back to 150,000 years b.p' eustatic sea-Ievel curve and from the pattern The last time sea-levelswere near their present ofLate Glacial and Postglacial sea-levelchange value occurred about 120,000 years ago when in southern latitudes. climate and ice volumes were last similar to those of the latter part of the Holocene. In be- Earth model parameters tween these two interglacial periods, the oxy- Essentialto predictions ofthe isostatic sea-level gen isotope information shows the peak glacial corrections is a model for the elastic and vis- õonditions persisting for only a relatively short cous properties of the Ea¡th. The elastic param- interval, although for much of the time between eters and the radial density distribution are about 70,000years b.p. and the Glacial Maxi- obtained from the analysis of seismic wave mum, ice volumes significantly exceededthose velocities through the planet, quantities suffi- oftoday such that the global sea-levels did not ciently well known for the elastic deformation rise above 40-50 m below present level during to be calculated and tested against knovrm short- this interval period tidal deformations of the Earth. The vis- ãosity structure, Iess well known, has to be Ice sheet models inferred from the isostatic analysis itself' It is The main sources of meltwater were the ice- this geophysical problem that has provided the sheets over North America, northern Europe majoì impetus for the study of past sea-Ievel including the Barents Sea, and a larger-than- (CathlesL975;Peltier & Andrews 1976;Nakada presentice-sheet over Antarctica. Theseice caps & Lambeck 1987; Mitrovica & Peltier L993).The õontributed about 70, 25 and 25 m respectively assumed viscosity models are relatively sim- to the eustatic sea-Ievel change since the time ple, comprising a mantle with a few layers of of the Last Glacial Maximum. The retreat of different viscosity values. For most predictions the northern ice sheetsis reasonably weII con- of palaeosea-Ievelsand shorelines, a three-lay- strained over the continents by geological and eréd model gives adequatepredictive capabili- geomorphological observations (e'g. Delton & ties; studies from different localities produce Èughes 1981), but the extent over the shallow Iargely consistent results (Lambeck eúc1. 1990; seassuch as the Barents Seais less well known, Lambeck & Nakada 1990; Lambeck 1993). Such although observationsof raised shorelines from models comprise an essentially elastic islands in such areas can be used to constrain lithosphere of 60-80 km thickness, an upper the ice volumes (Lambeck 1995b). Even over mantle (down to the 670 km depth seismic dis- (z- the continents the ice thickness is usually not continuity) with an effective viscosity of well constrained from independent observations 5)10'z0Pa ð, and a lower mantle with a viscosity and is usually inferred from ice-model consid- of about 1O2?Pa s. erations and from the interpretation of raised Whether the lithospheric thickness and shorelines within the limits of the areasof former mantle viscosity parameters are laterally uni- glaciation (e.g. Tushingham & Peltier rgdt; form is an important geophysical question and Lambeck 1993). Observational evidence ofsea- some evidence suggeststhey may vary from Ievel change within and immediately outside region to region, depending on the tectonic the former areas of glaciation has generallybeen history ofthe crust. For tectonically active ar- sufficient to anive at consistent northern ice eas such as the Aegean, the lithosphere can be sheet models since the time of the Last Glacial expected to be thinner than for the stable tec- Maximum. More problematic has been the es- toñic teuains of northern Europe; Iikewise, the timation of the volume changes in the Antarc- upper mantle viscosity for the tectonic regions tic ice sheet since this time. Some observations aie tikely to be less in the Aegean as well' Ide- of raised shorelines from the margin of this ice ally the mantle parameters should be estimated sheet suggesta reduction in ice volumes since from sea-level data from the region of interest. SEA-LEVEL CHANGE & SHORE-LINE EVOLUTION IN AEGEAN GREECE SINCE UPPER PALAEOLiTHIC TIME 5g7
But for the Aegean the observational data-base dence from the lagoonal and terrestrial sedi, is limited and, furthermore, the evidence is ments comes mainly from the shallow upper contaminated by tectonic contributions. But - regions of the Gulfs of Messenia (Kraft & Rapp by a largely fortuitous trade-off between the 1975), Lakonia and Argolis [Kraft ef al.1gT7), earth-model parameters- sea-levelpredictions and from the Bay of Navarone (Iftaft ef a1,1980) for thin-Iithosphere, lower-mantle-viscosity near Pylos, all in the Peloponnese.At these sites models are very similar to those for thick- sea-levels appear to havè been rising relative Iithosphere, higher-mantle-viscosity models to the land for at least the last 10,000 years, at flambeck 199ba; Lambeck ef o/. 1s96) and it rates that may have been temporally ánd spa- appears that parameters estimated from well- tially variable, although the quality of the data constrained rebound problems for northern is generally insufficient to quantify more than Europe constrain reasonably well the sea-level the general trends. An important inference from change models for the eastern Mediterranean. seismic stratigraphy is that sea-level at about Until abetter observational record is established, 18,000 years ago offshore from the southern '1,20 the earth models for the northern European Argolis Peninsula was about mlower than region are used for the Aegean predictions be- today (van Andel & Lianos 1983).The solution- low. notches produced by marine borers are pre- served only when rapid uplift events occur zuch The Aegean region that the notch, formed at about mean sea-level, Observations of sea-level change is lifted beyond the tidal range.Well-developed The evidence for sea-level change along the sequencesoccur, for example, in western Crete, Greek coastline is inferred from a variety of Rhodes, Karpathos and the Kythera islands geological and archaeological indicators, the (Flemming 1978;Thomm erctet al. 198L) as well latter being more plentiful for about the last as in the Gulf of Corinth (Pirazzoli et al.1.gg4), 4000 years. The archaeological evidence has at elevations of up to 0 m and with estimated been examined in detail by Flemming (1928), agesof up to 4000-5000 years. particularly for the coast of the Peloponnese and southwestern Turkey. The evidence con- Tectonic contributions to sea-level chanse sists of the positions of structures that, at their Tectonic movements and deformation ðf the time of construction, are believed to bear a crust occur over a wide spectrum of length and known relation to the position of the sea. In time-scales, from the slow and nearly uniform other instances, the provides only global plate tectonic motions to localized and "-ridence constraints on sea-level.For example, offshore sudden earthquake displacements. These move- from Franchthi Cave, near Koilás, in the south- ments are surface expressions of geophysical ern Argolis Peninsula, a Neolithic site points mechanisms operating internally to the Earth, to the seahaving been at least 11 m lower than and their records contain much information on today between 7610+L50and 6220+130years the processesshaping the planet. The geologi- b.p. fiacobsen& Farrard 1987; van Andel r-Sez). cal record provides many examples of these Likewise, at Saliagos on the Cycladean island surface displacements, particularly in the ver- of Andipáros, now-submerged Early Bronze Age tical direction which contribute to the relative remains point to sea-levels before b000 years change in the positions of the land and sea b,p. having been at least 5 m lower than þday surfaces.The easternMediterranean is a region (Morrison 1968). of active tectonics and Greece in particulãr is The geological evidence for past.sea-levels one of the most rapidly deforming continental in the Aegean and adjacent areastakes a number areas on Earth (Jackson1994), Other exampies of forms, including the depths of submerged of vertical movements occur in response to terrestrial or lagoonal vegetation and sediments, changes in the surface loads when crustal ma- inferences drawn from seismic reflectors in terial is eroded and deposited elsewhere. This shallow offshore sediments, and the age-height process, similar to the glacial unloading prob- relation of marine solution notches. The evi- lem, is usually more localized. In Greece,the sediment transport into the shallow bays and 2 Spelling of pliace-names generally follows that of the gulfs is comparatively small and this loading Times Atlas of the World. contribution is unlikely to have been impor- KURT LAMBECK
identified, it becomes possible to deter- tant on time-scales of 10,000 or so years. The can be whether vertical crustal tectonics has been deposition of sediments may,-however, have mine - if it has - to correct for it on locally to the build-up of the land important and not always substantiated assumption that surface"oitttibnt"d and migration of shorelines in some bays' the- tectonic iates have been uniform through Usually the processesproducing these ver- the for at least the last 120,000 years. tical movômettis ate only qualitatively under- time stood; quantitative models with sufficient seq-IeveLsin the Aegean region accutacli to correct observed sea-level oscilla- Predicted the sea-levelobservations for Greece tions foi the tectonic components rarely exi'st' Generally, the A"g"utt are inadequate to estimate b-oth Analyses of relative sea-Ievel observations tor and paramãters describing the glacio-hydro- isostätic parameters or for the eustatic func- the theory and the tectonic movements of tion are therefore less accurateif tectonic move- isosàtic Instéad, the former part is estimated ments of the crust are also important. But some the crust. the above outlined theory and model separation is made possible by the different using then it is tested against data f'rom chãracteristics of the two contributions to sea- paraireters; inferred to be tectonically stable foom Ievel change.While the tectonic displacements äreas of the position of the Last Inter- on the sho.-rtterm may be episodic and local- observations shoreline. If these comparisons are sat- ized, they persist for long periods of time, and glacial then predictions are also made for on time-scalesof 100,000-1,000'000yea-rs' they lsfactory, active sites to estimate the rates appear as slow but persistent movements' In the tectónically uplift or subsidence (Lambeck rggsa)' cãtttrast, changes asiociated with the growth of vertical FIcun¡ + illustrates the predicted past sea- and decay of the ice sheets are relatively uni- acrossthe region in the form of contours form and global on short time-scales (-1000 Ievels relative sea-level position for a few years), buithey are of a cyclic-nature on the of equal epochs. The contours at 18,000 b.p', iotrg"t time-scale' Thus, in the absence of seleited example, indicate the amount by which sea- chalges in ocean volume, a tectonic-uplift or for Ievels at that time were lower than the present subsiãence of 1 mm/year means that shorelines mean sea-level. For all epochs there is a formed 10,000 years ago would now be L00 m day of the offshore areasof the,tegean' above or below present sea-Ievel An indicator subsidence and Mediterranean Seasrelative to the of long-term tectonic stability is therefore pro- Ionian the contour values increasing with videdby the present position oJthe Last Inter- mainland, from the mainland, This reflects the glacial shoreline which formed about 120'000 distance effect with subsidence of the Ocean and ice volumes at that time hvdro-isostatic !e"tt and uplift of the land areas (c/' Ftc- *"t" titttilur"go. to those of today, and sea-Ievels sea floor 2). Superimposed upon this is a north-south were rìear their present level. In the Mediter- uRE - imposed by the glacio-isostatic effect, ranean Sea, the Last Interglacial shoreline trend frõm the Fennoscandian ice load in Late known regionally as the Tyrrhenian Shoreline mainly time (c/. Ftcun¡ 2). Throughout the re- - o""rrtJitt many localities to within a few GlaciãI gion sea-Ievel continues to rise, initially rap- metres abovepresént sea-Ievel.This is reported ur the ice sheets are disintegrating and then to be the case, for example, in the upper re- IdIy relätively slowly for the last 6000 years when sions of the Gulfs of Messenia, Lakonia and to the change come from Árgolis of the southern Peloponn-ese(KelleJat the mailcontributions isostatic component. FIGURE5 illustrates etáL. tgzo; van Andel 1.987)'Although a new the the changes in sea-level at four sites in the re- evaluation of the evidence is desirable, Iittle PrJdictions for the areas of Thrace and vertical tectonic movement appears to have gion. (FIcun¡ 5a) differ because Crete is fur- occurred in these localities during the past örete the ice sheet and the hydro-isostatic 120,000 or so years, Along the southern shore ther from is that of an island,whereas the predic- of the GuIf of Corinth, where the Last Intergla- effect tion for Thrace is characteristic of a continen- cial shoreline is found at elevations of 30-70 site. At the time of the Last GIaciaI m (Keraudren& Sorel 1987;CoIIier et al' L992), tal margin the predictions for the two sites differ tectonic uplift at average rates of 0'25 to 0'6 Maximùm 20 m, with the Th¡ace curve lying above mm per year appears to have occurred during by about prediction at aII times' The east-west the same intervãi. Thus, where this shoreline the Crete SEA-LEVEL CHANGE & SHORE-LINE EVOLUTION IN AEGEAN GREECE SINCE UPPER PAL,{EOLITHIC TIME 599 L8,000b.p. 10,000b.p.
6,000b.p. 2,000b.p.
28" 20' 24. FIGURE4. Ptedicted relative se,aJevelchange for Greeceand adjacent areasat (a) 18,000Bp, (b) L0,000 (c) (d) BP' 6000 BP, 2000 ae. Tlte c-ontoursrepresent the position of mean sea-làve|at these;i;rh, relative to the present mean level. KURT LAMBECK
risen above their present leveÌs. As a whole, the region wóuld appear to experience a gradual -20 subsidence at averagerates ranging from about -0.6 to -1.2 mm per year. This is largely con- € -40 sistent with the observations of an apparent o encroachment of the seain the Argolis, Lakonia' * -øo Kavalla, Th¡ace and Navarone Bay areas. Tectonic ¿J Messenia, subsidence does not appeal to have been sig- 'È -ao nificant, in agreement with the observation of E /,T=- at only ,':' e.s.l, the location of the Tyrrhenian shoreline -100 present sea-level at several -/ .i.i\ a few metres above Iraklion' Crete of these localities. The model also predicts a -120 -..1.1'" . sea-level throughout the Cycladean Is- 1l-"' rising IandJ, consistent with the observation of the -140L 20 submerged Early Bronze Age site near Saliagos (Morrisón 1968); there is no need to invoke tectonic subsidence ofthis area of sea floor. Where the observations point to relative uplift the estimated rates of tectonic displacements are increased over and abovethe rates that would - -')(\ be inferred if the glacio-hydro-isostatic correc- Likewise, areas of small ap- o tions are ignored. oarent subsidence, at rates that are less than Ø _?o o -- ihe isostatic rates, will also be undergoing tec- tonic uplift albeit at slow rates' The main area o * -40 of systématic and substantial tectonic uplift forms an arc extending from Rhodes in the east to Karpathos, Crete, Kythera and the southern- most peninsulas of the Peloponnese (Lambeck This pattern follows closely the con- -60' 1995á). ""r086420vergent boundary between the major tect^onic b time(xlooo years b.p.) unils in the area and the maximum rate of up- 4 mm per year, occurs in south- FIGURE5. Predicted sea-level at four sites since the lift, exceeding The time of the Last Glacial Maximum compated with western Crete close to the plate margin. eustatic sea-IeveL. other area of major tectonic uplift occurs along a Sites near Kavalla (Thruce) and ltaklion (Ctete) the southern shore and eastern end of the Gulf illustrate the north-south vatiability resulting of Corinthos, with uplift-rate estimates ap- mainly from the glacio-isostatic effect. proaching 1'5 mm/year. For central Ewoia, the b Sifãs in the Argolis Gulf and the Cycladean iates of uplift are also estimated to be about L- the east-west vaúability island of Póros illustrate 1.5 mm pet yeat. Elsewhere in the Pelopon- resulting mainly the hydro-isostatic effect' from nese attd Aegean no systematic patterns of (Note the different time-scales fot the two tectonic uplift or subsidence appear;those move- illustrations.) ments aré either smaller in amplitude or of than the glacio- variability is illustrated by the comparison for a shorter horizontal length scales In the south oJ the site intheArgolis Gulf with a site inthe Cycladean hydro-isostatic changes. near Porto Kheli, subsidence isla¡d gronp lfrcunn 5b). The east-west spatial Aigolis Peninsula rate of about 1 variation, largely the result of the hydro-isostatic apþears to have occurred at a few thousand years conection, remains significalt, about 5 m at 10,000 mm per year for the last just the north, at the years b.p. At both sites the predictions lie well (Flemming 1978) but to movement appears below the eustatic sea-Ievel function. Franchthi site, Iittle tectonic Lianos (1-983J If no other tectonic processes were active, to have occurred if the van Andel & sea-Ievel of at no epoch, particularly during the past 6000 estimate of Last GIaciaI Maximum -120 yeats, áte the sea-Ievelshere predicted to have about m is correct. SEA-LEVEL CHANGE & SHORE-LINE EVOLUTION IN AEGEAN GREECE SINCE UPPER PALÂEOLITHIC TIME 601
Palaeo shorelines in the Aegean region extending from Andros in the north to Ios in Once a predictive model for sea-levelexists that the south over a distance of about 160 km. To adequately reflects the observed changes, it the north, beyond the area illustrated in Frc- becomes possible to predict the location of URE6, Iarge tracts of the Thermaïkós Gulf were shorelines through time using the relation (S) exposed, a¡rd the plain of Thrace extended more for areaswhere tectonic movements are inferred than 60 km to the south of the present shore- to have been small - as appears to have been Iine. The broad and shallow ãrea between the casefor the central Aegean islands and much Limnos and Turkey would also have been ex- of the Peloporìnese.As the topography and water posed, cutting off the Sea of Marmara from depth vary much more rapidly than the sea- marine influence: (SeeFrcunn + for the predicted level change function, a much higher spatial sea-level change in these areas.) description of the former is required for pre- At this time of maximum glaciation the 'Cycladean cise palaeogeographicreconsbuctions. The high- Island' effectively divides the Aegean resolution data compiled here are based on: Sea into two: the Aegean Seaproper in the north . digitized bathymetric charts for the central from the Mirtoan Sea and the Sea of Crete in Aegeansea-floor at scalesof i.:100,000and the south and southwest. The two basins are 1:150,000(HNHS L993), connected via a narrow channel (-6 km wide) a 500-m resolution digital terrain model for between Evvoia and Andros in the north and central Greece and the Aegean islands by further channels (-12 km wide)between the provided by the National Technical Uni- islands of Amorgós and Kalimnos. On the versity in Athens, and Cycladean Island itself, a broad central and for an area around Franchthi on the Argolid relatively flat plain was punctuated by hills and Peninsula, digitized data (including bathy- mountains that now form the residual islands. metry) from the 1:50,000topographic map The averageheight of the plain was about 1b- for the area (HMGS IS92). Beyond the 20 m above sea-levetr;topographic lows within marine areas covered by the digitized it, between Náxos and Mikonos, for example, 1:100,000and 1:t50,000 maps,the bathy- could have held shallow freshwater environ- metry included in the Greek digital ter- ment$. To the west, the islands of Milos, Sifnos, rain model is used. Sérifos and Kithnos remained senarateboth from These three data-setshave been combined and each other and from the mainland, although gridded on to a 0.01' grid using the Delaunay the minimum widths of the sea passageswere triangulation and natural neighbour element considerably less than today. The single entity methods (Sambridgeeú o1. 1995). Bathymetric made by the islands of Milos, Kimolos and features larger than about 0.02' (or about 2 km) Poliaigos, for example, were separatedfrom the are generally well resolved in this gridded data- Cycladeanlaadmass by relatively shallow (
KURT LAMBECK that 4000 years earlier (FIcun¡ 6b) although some able, the shoreline evolution for this period is ofthe narrow channels separating the various best estimated by superpositioning upon them island groups have eíxpandedmarginalty. Af- the sea-level predictions given in FIGURE4. ter about 14,000 b.p. the sea-Ievel rise is more rapid. The Cycladean Island begins to break Earþ Palaeolithic to Early Upper Palaeolithic up into northern and southern parts that are shorelines separatedby shallow sea,break-up being com- FIGURE3 illustrates the predicted sea-Ievel os- ptéted soon after 1'2,5oOb.p. At this time the cillations for Páros near the centre of the present Milos group is separated from the southern Cycladean island group from the time of the Cycladean island, via Sifnos, by a minimum penultimate Glacial Maximum (saalian) at about water crossingof about 14 km, By 12,000b.p' 140,000 yeats ago to the present. Spatial vari- the two main parts of the earlier Cycladean Is- ability of sea-level occuring across the region Iand are separated by a minimum distance of wiII generally be small when compared with about 17 km. At 10,000 b.p. the original Cycla- the larger uncertainties introduced by incom- dean Island has broken up further, and the geo- plete knowledge of the ice sheetsbefore about graphy of the region is beginning to resemble 20,000 years b.p. As the Last GIaciaI Maximum the present configuration. The southern part appears to have been of relatively short dura- of the Evvöikós Gulf is first subjected to ma- tiãn, the coastalplains were fuIly exposed from rine influence at about 9000 b.p. and the north- only a little before 20,000 years b.p, to about ern sector, through the Dhiavlos Strait, before 16,000years b.p. The maximum development 8000 b.p. The islands of Milos, Kimolos and of the Cycladean Island would therefore have Poliaigos remain together with decreasing sur- been restricted to between about 21,000 and face area until about 8500 b.p' The final sepa- 14,000years b.p. The only other time in the last ration of Náxos, Páros and Andipráros also occurs 150.000vears that these low levels would have at about this time. Likewise, the area of the been reached is during the penultimate Glacial coastal plains around the periphery of the Maximum before about 135,000years b.p. From present Argolis Peninsula is gradually reduced about 70,000years b.p. to the lead-up to the Last throughout this period (FIGURE8J. Glacial Maximum, sea-Ievelsoscillated between about -50 and -80 m and shoreline locations similar to those that occurred Shorelines from Late Neolithic time to the would have been present laterbetween about t2,ooo and 10,000years b.p. During Late Neolithic and Bronze Age times (Frcunn 6). Between about 110,000 and 70,000 (after about 6000 years b.p'), the predicted vears b.p. the water depths a¡rd shorelines are change is a sea-level rising at a rate of about äharacteiistic of conditions last experienced be- O.7-1,.Omm per year, small compared with that tween about 9000 and 7000 Yearsb.P, for the earlier period when the rates wete as high as 12mmlyear (FIcun¡ a). The Late Holo- Limitqtions of the shoreline rcconsttuctions cene rates are comparable to the vertical tec- The reconstructions of past topography and tonic rates inferred for some localities in the shoreline locations are model dependent, no- region, an effective separation of the two con- tably on the assumed eustatic sea-Ievel func- tributions to relative sea-level change is now tion, on the appropriate earth-model parameters more problematical, For example, the local that define the isostatic contributions, and on modification of shorelines by sedimentation may the assumed absenceof other tectonic and sedi- now become important in areas such as the mentary processes,Constraints on the model upper reaches of the Argolis Gulf. Neverthe- parametersare providedby observationsof past lesi, the characteristicregional pattern is a slow iea-levels, and the practice is first to calibrate encroachment of the sea throughout this pe- the eustatic a¡d isostatic models with data from riod, consistentwith observations of Bronze Age stable regions and then to apply them to the and more recent sites that are now a few me- tectonically active areas. Solutions ofthe sea- tes below sea-levelin, for example, Milos (Bintliff level equation (4) have therefore tended to be 1976),Andipáros (Morrison 1968) and near the iterative: observations of sea-Ievelchange from Franchthi Palaeolithic cave site (Jacobsen & both the a¡ea of interest and from areas out- Fa¡rand 1987).If localbathl'rnehic maps are avail- side of it are compared with the model predic- SEA-LEVEL CHANGE & SHORE-LINE EVOLUTION IN AEGE,{N GREECE SINCE UPPER PALAEOLITHIC TIME 607
tions to improve the estimates of the param- inhabitants of Greece. With some notable ex- eters that define both the isostatic model and ceptions, the fuII recognition of this input has the tectonic components, Predictions for the been neglected. That sea-Levelsduring the Last Aegean point to potential tests to check the Glacial Maximum were some 100-150 m lower correct timing of the shoreline movements. For than today is not a matter for dispute in the example, the model predicts that the Saronikós archaeological and prehistory liteiature. Nor Gulf was separated from the Aegean Sea and is the fact that at that time extensive coastal from marine influence between about 23,000 plains would have been exposed, plains that and 13,000 years ago and sedimentary cores would generally have seen much human ac- from the central areas of the depression may tivity (e.g.Gamble 1986).Less attentionhasbeen contain information on the timing of the first paid to the timing of the sea-level rise from then - marine transgression and through the sea- up to about 6000 years ago, and to the fact that - Ievel equation on the eustatic sea-level func- this rise occurred over a substantial time in- tion. Other potential target areasfor testing the terval. Van Andel & Shackleton (1982) and van model are the Peialión Gulf, or possibly the now- Andel (1989),in neglectedpapers, attempted shallow sea floor depression between Náxos t-oquantify sea-Ievel and shoreline changes in and Mikonos (Frcune O). the eastern Mediterranean during this Late GIa- As noted above, resolution of the digitized cial period arld recognized that the rise occr¡rred and gridded nautical charts gives a spatial reso- over an extended period, with important hu- Iution of about 2 km, so some verv fine detail man implications as the progressive reduction plcunn is lost in the reconstructions, Bc. for ex- ofthe relatively hospitable coastal plain envi- ample, shows Idrha connected to the mainland ronment causeda concomitant loss of resources. at about 10,000b.p., whereas there would ac- The consequences would have been particu- tually have been a narrow channel - less larly severe for the Cycladean Island, where - than about 500 m wide between the two. an extensive, relatively flat and low-lying plain Similar resolution is lost between those islands was progressively reduced to a few rocky is- of the Cycladean group that are separated by lands over a period of about 6000 years [Frc- very narrow and deep channels, trRE6). Likewùe, the coastal plains of the palaeo The model predictions, most reliable for the Argolis Peninsula (Frcune B) could have offered time since the Last Glacial Maximum, are also a hospitable environment for Palaeolithic and indicative of conditions throughout Early and Neolithic dwellers. Certainly throughout the Mid Palaeolithic time. Thus the reconstruction Aegean region, these now-flooded plains are in Frcun¡ 6 for the interval 12,soo-L4,O00 vea¡s Iikely to have been more conducive to human b,p. also represents conditions between ãbout activities than much of the present coastal zone 25,OOO-22,O00years b.p, Some question does and human traffic patterns and migration routes remain about the time-scale appropriate to ob- may well have been quite different in the past. servation and prediction. All ageshere refer to Can this, for example, explain the general pau- the conventional radiocarbon time-scale since city ofevidence for Palaeolithic occupation of most observations of sea-level change and of the region. Did the encroaching seacover much the retreat ofice sheets use radiocarbon deter- of the evidence for the activities of earlv peo- minations. Insofar as archaeological deter- ple until about 10,000-9000 years b.p.í Éven minations for Palaeoiithic and Neolithic contexts in Early Bronze Age times, sea-levelswere lower also refer to this time-scale, the modelË are than today by as much as 5 m and, as the re- consistent with those data. Care is required when gion has been a zone oflow tidal range tluough- using calendar ages for the more recent times out, much of the archaeological evidence for because discrepancies between this and the this interval may also be below today's sea-level. radiocarbon time-scale can be substantial (e.g. Locally, the coastal changes may have been qI.'t982; Klein ef Bard eúa1, rgsO). significant. In Milos the coastal plain was con- siderably more extensive at the time when ob- Shorelines and the human environment sidian from this island first appears on the The examples set out here indicate that recon- mainland in latest Palaeolithic time (FrcuRx7). structions of the contemporaneods geography Can this explain the apparent absence ofLate should be integral to discussion ofearly coastal Palaeolithic and Early Neolithic settlement on
610 KURT LÂMBECK the island (Renfoew L972) despite the quarry- of 11th millennium b.c.). With the lower sea- ing that must have occurred to obtain the ob- levels of Late Palaeolithic time, the island would sidian found on the mainland? Likewise - as have been much more accessiblefrom the main- rose. At already emphasizedby Jacobsen(1969) and van land than it became later as sea-Ievels Andel (1987)-the coastalgeometry atthe Late the time of the maximum glaciation and until Palaeolithic site at Franchthi on the Argolis about 16,000 b.p, the minimum water crossing Peninsula, would have been distinctively dif- reouired to travel foom Milos to the mainland ferent in the past. At the peak of the last gla- would have been about B km (Ftcunrs 6 & 7), ciation, the coast would have been at least 6 with the largest crossing being between Sifnos km to the west (FIcunr 9); if much of the evi- and the Cycladean Island; this configuration dence for the activities of Palaeolithic and wouÌd have been largely preserved until about Neolithic people is now submerged, Franchthi L3,000b.p.By 11,000b.p., the time of the old- Cave represents only a partial record of human est obsidian finds recorded so far, the Cycladean activity in the area. Certainly any interpreta- Island had broken up into northern and south- tion ofthe cave record needs to consider these ern parts centred on Páros-Andipáros and shifting shorelines. Does the appearance of Mikonos-Tinos-Andros although at this time the marine molluscs and small fishbones by about minimum water separation acrossshallow seas 1-2,000-10,000b.p,, and the first much larger was only about 20 km, with clear inter-island fish vertebrae by about 9200 b'p. reflect more visibility for a daylight crossing. on the coastal evolution than on the fishing methods of the early inhabitants, as lacobsen (L976) suggested? Acknowledgemenfs. I thank Dr C.L. Smither and Mrs C. effort in digitizing the nautical from Milos appearc in the Frarchthi Krayshek for their major Obsidian preparing the figures. I thank Professo¡ G. (Perlès charts and for strata after about 1L,000 b.p. 1987; Ren- Veis and Dr Lysandros Tsoulos of ihe National Technical frew & Aspinall 1987) [Perlès(L979) and Cherry University of Athens for making available the non-marine & Torrence (1982), quoting Perlès, give an age topographic data.
References glacio-eustaticsea-level record: BARD,E., B. HAMELIN,R.G, FAIRBANKS & A. ZINDLER.1990' CAIi- FAIRBANKS,R.G. 1989. A 17,000-year glacial melting dateson the Younger Dryas event bration of the 14Ctimescale over the Past 30,000years influence of 342i 63742' using mass spectrometric U-Th agesf¡om Barbados cor- and deep ocean circulation, Nature and coastal als, Nofure 345: 405-10. FLEMMING,NC. rszg. Holocene eustaticchanges implications BINTLIFF, 1976.The history of archaeo-geographicstudies of tectonics in the no¡theast Mediterranean: J. Tîans- prehistoric Greece,and recent fieldwork, in Mycenaean for models of c¡ustal consumption, Plu'losophical Cambridge:Cambridge University Press actions of the Royal Society of London A289:405-58. eeography:3-16. of Europe. Cam- g Siitiih Associationfor MycenaeanStudies, GAMBLE,C. L986. The Palaeolithic settlement Press. CATHLES,L.M.r1975. The viscosity of the Eaùh's mantle. bridge: Cambridge University SERVICE.1992' Princeton (NJ):Princeton University Press. HMGS = HELLENICMILITARY GEOGRAPHTCAL 1:50,000map sheetSPeúsoÍ. Athens. CHAPPELL,J. 1983. Evidence for a smoothly falling sea-level = HYDROGR.{PHICSERVICE. 1993. 1:1OO,OOO relative to ncirth Queensland, Australia, during the past HNHS HELLENICNAVY Nauticaì Charts 415, 6000 years,Ndfu¡e 302: 406-8. Nautical Cha¡ts 412-414 and 150,000 CHAppELL,i. a H. Pol,l,¿'cH.1991. Post-glacialsea-level rise 42L,423, At}]ens. ' near Helsinki, from a coral ¡ecord at Huon Peninsula,Papua New Guinea, HYVÁRINEN,H. 1980.Relative sea-leveì changes time, Bulletin of Nature 349:1-47-9' southern Finland, during early Litorina Society of Finland 52" 207-1'9. CHAPPELL,]. & N.I. SHACKLEToN.1986. Oxygen isotopes and the Geological Aegeanregion, án- sea-level,Naturc 324: 73740. JAcKsoN,J. 1994. Active tectonics of the Science22:239-71" CHERRY, & R. TORRENCE.1982. The earliest prehistory oT nual ReviewofEarth and Planetary J.F. and vicinity, Melos, in C. Renf¡ew & M. Wagstaff (ed,.), An island pol- JAcoBsEN,T.W. 1.969.Excavations at Porto Cheli Cave,1967-1968' ity: the archaeology and exploitation in Melos: 24-34' preliminæy Report,II: The Franchthi Cambridge: Cambridge Univelsity Press. Hesperia 38: 343-81. 'J.g76. prehistory, Scientific Ameri' CoLLTER,R.E.L., M,R. LEEDER,P.J. RowE & T.C.ArKINsoN. 1992. L7,Ooo years of Greek Rates of tectonic uplift in the Corinth and Megara Ba- can 234(6):76-87 'J.'L: i Cavea nd Parali a. sins. Central Gteece, Tectonics 7!59-67 ' JAcoBsEN,T.w. & WR. FAPGAN.198 7 Franchth (IN): Indiana University Press,Excavations CURRAY, 1965. Late Quaternary history, continental shelves Bloomington J.R. 1. of the Uníted States,in H.E. Wright & D.G' Frey (ed.), at Franchthi Cave,Greece Fascicle & K.P WINTER.1976' The ofthe United States:723-35.Princeton KELLETAT,D., G. KOWALCZYK,B. SCHRÖDER Quaternary of the (NJ):Princeton UniversitY Press' A synoptic view on the neotectonicdevelopment Zeitschrift der Deutsche DENroN,G.H. & T.i. HucHEs(ed.). 7981..Thelast geat ice sheets. Pelópoirnesian coastal regions, New York (NY): Wiley. Geologische Gesellschaft 27: 447-65' SEA.LEVEL CHANGE & SHORE-LINE EVOLUTION IN.A,EGEAN GREECE SINCE UPPER P,{L,ITEOLITHIC TIME 611
KERÂUDRIN,B. & D. SoREL.1982. The tenacesofCorinth (Greece) NAKADA,M. & K. LAMBECK.1982. Glacial ¡ebound and relative - a detailed record of eustatic sea-levelva¡iations dur- sea-levelvariations: a new appraisal,Geophysical lour- ing the last 500,000years. Marine GeologyZZ I gg-1,O2. naÌ of the Royal Astronomical Society 90: 171,-224. KLEIN,J., J.C. LERMAN, PE. DAMoN& E.K. RALPH.1982. Cali- NAKADA,M. & K. LAMBECK.LgBB. The melting history of the bration of radiocarbon dates,Radiocarbon 24: 1,O}-SO. Late PleistoceneA.ntatctic ice sheet,Nafure 333: 36-40. paleo- KRÂFT,J.C., S.E. ASCHENBRINNER& c.J. RApp.1922. PELTTER,W,R. & t.T. ANDREWS.1926. Glacial-isostaticadiust- geographicteconstructions - of coastal Aegean archaeo- ment 1. The forward problem, Geophysical lourn'al of logical sites,Science L95i 94L-47. the Royal Astronomical Society 46: 605-46. KRAFT, J.C.& G.J.RApp. 192b. Late Holocenepaleogeography PERLÈS,C. 1979.Des navigateurs méditerranéens il V a 10,000 of the coastalplain of the Gulf of Messenia,Greece, and ans,La Recherche1O:82-3. its relationships to archaeologicalsettings and coastal L987à. Les industies lithiques taillés de Franchthi (Argolide, '1,1,9'),- change,Geological Society of America Bulletin 86i Grèce),in Perlès(ed.): 142-3. 1204. (Ed.). 1987b. Présentationgénérale et industries paléo- KRAFT,J.C., G.R. RApp & S.E.ASCHENBRENNER. 1980. Late HoIo- lithiques. Bloomington (IN): Indima University Press.Ex- cenepalaeogeomorphic reconstructions in the a¡eaofthe cavationsat F¡anchthi Cave,Greece Fascicle B(1). Bay of Navarino: Sandy Pylos, /ournal of Archaeological PIRÂZZoLI,PA,, S.C.STIRos, M. ARNoLD,J. LABoRIL,F. LABoPGL. Science 7: 187-21,0. DEGUEN& S. PAPAGEoRGIoU.1994. Episodic uplift deduced LAMBEcK,K. 1988. Geophysical geodesy: the slow deforma- from Holocene shorelines in the Þerachorã peninsuJ.a, tions of the ea¡fñ. Oxford: Oxford Universitv Press. Corinth area,Greece, Tectonophvsics ZZg: 2O'L-9. 1993.Glacial rebound ofthe British IslesII: A hieh resolu- RENFREW,C. 1,972.The emergence;f civilisation: the Cyclades tion, high-precision model, Geophysical loulnat Inter- and the Aegean in the 3rd millennium Bc. London: national 115: 960-90. Methuen. 1995a.Late Pleistoceneand Holocene sea-levelchanee in RxMREw,C. & A. ASPI.JALL.1982. Aegean obsidim and Fræchthi Greeceand southwesternTurkey: a separationofeust=atic, Cave,in Perlès(ed):257-zO. isostatic and tectonic contributions, Geophysical Jour- SAMBRTDGE,M., J. BRAUN& H. MceuEEN.1995. Geophysical nal International 1,22:Io22-44. parametrizationand interpolation of irregula data us- 1995b. Constraintson the Late Weichselianice sheetover ing natural neighbours, Geophysicallournal Intemational the BarentsSea from observationsof raised shorelines, . 122:837-57. Quaternary Science Review L4:'J.-1-6. SHACKLEToN,N.J. 1987.Oxygen isotopes, ice volume and sea- In press. Sea-level change along the French Atlantic coast Ievel, Quaternary Science Review 6: 183-90. since the time of the Lastglacial Maximum, Palaeo- THOMMERET,Y., J. L,{BOREL,L.F. MONTAGGIONI& PA. PIRAZZOLI. geo grap hy, Pal aeo c I i mato I ogy, Pal ae o e c o I ogy. 1981. Late Holoceneshoreline chanqesand seismo-tec- LAMBECK,K., P JoHNSToN& M. NAKADA.1990. Holocene gla- tonic displacementsin western Crcþl,Gteece),Zeitschrift cial rebound and sea-leveìchange in NW Europe, Geo- für GeomorphoÌogre (neue Folge) 40: 127-49. physical /ournal International 103: 4S1-68. TusHrNcHAM,A.M. & W.R.PELTTER. 1991. ice 3G: a new slobal LA.MBECK,K., P. IOHNSToN,C. SMITHER& M. NAKADA.1996. model of Late Pleistocenedeglaciation based upoñ geo- Glacial rebound of the British Isles III: Constraintson physical predictions of postglacialsea-level chmge,/ouma1 mantle viscosity, Geophysical Journal Intemational'J,ZE: of Geophysical Reseo¡cå 96: 4492-529. 340-54. vANANDEL, T.H. 1987.The adjacentsea, in T.H. van Andel & pleistocene LAMBECK,K. & M. NAKADA.1990. Late and HoIo- S.B. Sutton (ed.), Landscapesand people ofthe Franchthi cene sea-leveì.change aìong the Australian coast,pa,laeo- region'. 3-64. Bloomington (IN): Indiana Univetsity press. geo gr ap hy, Pal aeo c I im atology, Pal ae o e c ol ogy (Glob al an d Excavationsat Franchthi Cave,Greece Fascicle 2. PlanetaryChange Section) 89: 143-26. vAN ANDIL, T.H. 1989. Late Quaternaty sea-level changesand MILLTMAN,J.D. & K.O. EMERY.1968. Sea-levelsduring the past archaeology, Antiquity 63: 733-46. 35,000years, Science 1,62'. 1121-3. vAN ANDEL,T.H. & N. LIANos.1983. p¡ehistoric and histonc MITRovrcA,I.X.& W.R.PELTTER. 1991. On postglacialgeoid sub- shorelines ofthe southern Argolid Peninsula: a subbotton sidence over the equatorial oceans,/ournal of Geophysi profiler study, NauticalArcÍtaeology 1.2(a)t gOB-24. caI Research 96: 20,059-7 L. vAN,{NDEL,T.H. & J.C.SHACKLETON. 1982. Late paleolithic and 1993,The inferenceof mantle viscosity from an inversion Mesolithic coastlinesof Greeceand the Aegean,/ournaJ of the Fennoscandian ¡elaxation spe¿trum, Geophysical of Field ArchaeoloW g: 445-54. lournal International L14i 4s-62. ZwAR'rz,D., K. LAMBECK,M. BIRD& J. SToNE.IN pREpARATroN. MoRRtsoN,I.A. 1968.Appendix L Relativesea-level change in Constraints on the forme¡ Antarctic ice sheet f¡om sea- the Saliagosarea since Neolithic times, in J.D.Evans & level observations and geodynamic modelling, in pro- C. Renfrew (ed.), Excavations at Saliagosnear Antiparos: ceedingsofthe VII International Symposium on Antarctic 92-8. London: Thames& Hudson. Earth Sciences, Siena, ltaly.