KLISSOURA CAVE 1 AND THE UP PER PALEOLITHIC OF SOUTH ERN GREECE IN CUL TURAL AND ECOLOGICAL CONTEXT

Transcription

1 Eur asian Pre his tory, 7 (2): KLISSOURA CAVE 1 AND THE UP PER PALEOLITHIC OF SOUTH ERN GREECE IN CUL TURAL AND ECOLOGICAL CONTEXT Mary C. Stiner 1, Janusz K. Koz³owski 2, Ste ven L. Kuhn 1, Panagiotis Karkanas 3 and Margarita Koumouzelis 3 1 School of An thro pol ogy, P.O. Box , Uni ver sity of Ar i zona, Tuc son, Ar i zona , USA; mstiner@ .ar i zona.edu, skuhn@ .arizona.edu 2 Jagiellonian Uni ver sity, In sti tute of Ar chae ol ogy, Go³êbia 11, Kraków, Po land; janusz.kozlowski@uj.edu.pl 3 Ephoria for Palaeolanthropology and Spe le ol ogy, Ardittou 34b, 11636, Ath ens, Greece; pkarkanas@hua.gr Abstract Klissoura Cave 1 pre serves a long se ries of Mid dle Paleolithic, Up per Paleolithic and Mesolithic cul tural lay ers, in - ter rupted by at least three sig nif i cant ero sional hi a tuses. The sed i men tary fea tures, ar ti facts and an i mal re mains of the Up - per Paleolithic though Mesolithic lay ers tes tify a wide range of on-site ac tiv i ties, with com plex cy cles of fea ture con struc - tion and aban don ment. The in dus try of Layer V closely re sem bles Uluzzian as sem blages from south ern It aly. Its age re - mains un cer tain but al most cer tainly ex ceeds 39 kyrs BP. The most in tense use of the site oc curred dur ing the for ma tion of Aurig na cian lay ers IV and IIIe-g, dis tin guished by many su per im posed plain and clay-lined hearths, and in Layer IV, the rem nants of a small struc ture en vel op ing a dense con cen tra tion of per fo rated shell beads. Fire places were fed mainly with dicotyledonous wood and bark-pro duc ing plants, whereas grass re mains are con cen trated in other parts of the oc cu pied area. Post-dat ing the Aurig na cian are two non-aurig na cian lay ers, fol lowed by an ephem eral Epipaleolithic oc cu pa tion and sub stan tial Mesolithic oc cu pa tions. The bo tan i cal, fau nal and geo log i cal data iden tify a grad ual trend to ward cli ma tic cool ing through the Up per Paleolithic se quence. Warmer, wet ter con di tions re turned only well af ter MIS 2, dur ing the Mesolithic. Fau nal data in di cate op por tu nis tic hunt ing of a va ri ety of un gu late spe cies, but mainly fal low deer, one or a few an i mals at a time. The pat terns of small game ex ploi ta tion re veal a trend of in creas ing di etary breadth that be gan in the early Up per Paleolithic and in volved pro gres sively greater use of an i mals such as hares and/or birds with time. Land snail ex ploi ta tion be came im por tant in the later Up per Paleolithic phases and peaked in the Mesolithic. Per fo rated shell or na - ments are pres ent in the Uluzzian layer (V) and in all sub se quent lay ers. The or na ments con sist al most ex clu sively of fin - ished prod ucts, worn from use and lack ing ev i dence of pro duc tion de bris. Key words: Mesolithic, phytoliths, anthracology, zooarchaeology, lithic in dus tries, os se ous tech nol ogy, paleoecology. INTRODUCTION Klissoura Cave 1 pre serves a re gion ally unique se quence of Mid dle Paleolithic and Up per Paleolithic though Mesolithic cul tural lay ers dat - ing to the Late Pleis to cene (Koumouzelis et al., 1996, 2001; Koz³owski, 1999; Pawlikowski et al., 2000; Tomek and Bocheñski, 2002; Karkanas et al., 2004; Sitlivy et al., 2007). Klissoura 1 is one of sev eral ar chae o log i cal cave sites in the Klissoura Gorge (Koumouzelis et al., 1996, 2004; Runnels, 1996), and it con tains the deep est and ear li est Paleolithic se quence for the area. Here we re port only the ma te rial from the fi nal sed i men - tary cy cle in Klissoura 1, that rep re sent ing the Up per Paleolithic through Mesolithic oc cu pa - tions. The Up per Paleolithic is clearly dis tin - guished from the Mid dle Paleolithic on the ba sis of lithic and other artifactual con tents. The Up per Paleolithic layers contain ornament assemblages and os se ous tools, whereas the Mid dle Paleolithic lay ers do not, ex cept where mi nor post-deposi-

2 310 M. C. Stiner et al. tional mix ing at the MP-UP layer con tact is in di - cated. The for ma tion of the Up per Paleolithic through Mesolithic strati graphic se quence was dom i nated by anthropogenic pro cesses. Con structed hearths of di - verse forms, ash dumps, raked-out ash fea tures, and tram pled ash re mains are very com mon (Karkanas, this is sue). The ear li est Up per Paleolithic layer (V) trun cates the top of the Mid dle Paleolithic sed i men tary se ries. As of this writ ing, Layer V rep re sents the only Uluzzian oc cu pa tion doc u - mented in Greece (though un pub lished finds from Kephalari Cave may also be Uluzzian), and its in - dus try re sem bles es sen tially con tem po rary in dus - tries in It aly. The Aurig na cian lay ers have a gen - erally discrete appearance, and the earliest Aurignacian layer (IV) truncates the un der ly ing de - pos its. The Aurig na cian com po nents in Klissoura 1 are dis tin guished from all other cul tural lay ers by the re mark able com plex ity of in situ hearths, which range from sim ple small ba sin or stacked forms to clay-lined types (Karkanas et al., 2004). Rem nants of a small struc ture are de mar cated in the ear li est Aurig na cian layer (IV) by a roughly oval scat ter of large stones, a dis crete or ganic stained area within, and an ex cep tion ally dense con cen tra tion of per fo rated shell beads. Fol low ing the up per Aurig na cian lay ers are two enig matic cul tural ho ri zons. The def i ni tion of these ho ri zons suf fers in part from the lack of in - for ma tion for the re gion and pe riod, but also from am bi gu ities con cern ing the be hav ioral causes of variation in the lithic industries. Hypotheses for explaining this variation include ethnic differences and dis tinct tra di tions of tool man u fac ture and, al - ternatively, differences in the circumstances of oc - cupation by essentially similar groups by season. We offer some preferred explanations below but acknowledge that resolution of these questions ulti mately re quires more in for ma tion from as yet un - dis cov ered Up per Paleolithic sites in the Peloponnese. Mod er ately rich fau nal and lithic as sem - blages were ob tained from the Epigravettian and Mesolithic lay ers. These cul tural lay ers ex pe ri - enced fre quent trun ca tions and con sid er able dis - tur bance, mainly from hu man ac tiv ity, and dif fuse interfaces characterize the contacts between them. The sec ond ma jor hi a tus in the strati graphic se ries cor re sponds to the LGM, af ter which ap pears the typical middle (or early-middle) Epigravettian with con spic u ous links to south-cen tral It aly. The third hi a tus sep a rates the Epigravettian (lay ers IIa-d) from the Mesolithic (layer 5a). Here, how - ever, cul tural con ti nu ity can be seen de spite the pres ence of depositional hi a tus. The oc cur rence of the Ter mi nal Paleolithic marked by a Epigra - vet tian tra di tion in neigh bor ing caves of the Klissoura Gorge in di cates that Epigravettian groups did not aban don the re gion at this time, but rather oc cu pied dif fer ent caves and shel ters in the area. The rich and varied archaeological record of Klissoura Cave 1 pro vides an un prec e dented and for the mo ment unique body of in for ma tion about the Up per Paleolithic of south ern Greece. Co her - ent lithic, bone, shell and os se ous tool assemblages and many fea tures and spa tial data were re - corded and stud ied in this col lab o ra tive re search pro ject. This de tailed re cord of Up per Paleolithic ac tiv i ties yields several of sur prises and in sights on Up per Paleolithic be hav ior and cul tural di ver - sity in Eur asia, in clud ing the great age of the ear li - est Up per Paleolithic oc cu pa tion and the con texts of on-site ac tiv i ties through out the Up per Paleolithic. CULTURAL SEQUENCE BASED ON LITHIC AS SEM BLAGES The se quence of cul tural ho ri zons and lithic as sem blages does not doc u ment an un in ter rupted lo cal evo lu tion of Paleolithic cul tures (Kaczanowska et al., this is sue). In stead, it is char ac ter - ized by cul tural and oc cu pa tional dis con ti nu ities. The Up per Paleolithic be gins with the Uluzzian (layer V), re placed by the Aurig na cian as rep re - sented by layer IV and lay ers IIIg-a. Layer III con tains a non-aurig na cian Up per Paleolithic in - dus try. The Gravettoid com po nent in layer III con tains what we have de scribed as a Med i ter ra - nean backed bladelet/blade in dus try. The tech - nological character and stylistically distinct na - ture of the Gravettoid as sem blage rules out a lo cal evo lu tion, sug gest ing in stead or i gins from dif fer - ent re gions of the mid-north ern Med i ter ra nean re - gion. Lay ers 6, 6a, and 6/7 in fact rep re sent the mixed fill ing of what may be an anthropogenic ditch and will not be dis cussed fur ther.

3 Klissoura Cave The artifactual re cord in Klissoura Cave 1 af - fords im por tant in sights on the pro cess of cul tural evo lu tion and dif fer en ti a tion in the re gion. Above the Uluzzian layer V is a se ries of Aurig na cian lay ers (IV, IIIa-g) over lain by layer III that could rep re sent the fi nal phase of the Uluzzian. Higher in the strati graphic col umn, layer III con tains a Mediterranean backed bladelet/blade industry, which, after sedimentological/erosional hiatus, is over lain by the Epigravettian (layer II). These ob - ser va tions from the lithic as sem blages (Kaczanowska et al., this is sue) sug gest that the var i ous cul ture units iden ti fied in Klissoura 1 cor re spond not so much to the ad ap ta tion of the same for ag ers to specific raw materials and ecological conditions, but rather rep re sent dif fer ent cul tural tra di - tions cor re spond ing to dis tinct groups who pe ri - od i cally coex isted within the larger region. Culturally diagnostic artifact forms in Klissoura 1 in clude the fol low ing: arched backed blades and con vex trun ca tions for the Uluzzian; carenate cores/endscrapers and mi cro-re touched bladelets for the Aurig na cian; backed blades and bladelets for the Med i ter ra nean Early Gravettoid ; com po nent var i ous backed blades (also with ven tral re touch), para-geo met ric forms and shoul dered points for the Epigravettian; and geo - met ric mi croliths for the Mesolithic. Tech no log i cal di ver sity is quite nar row from the Up per Paleolithic through the Mesolithic as - sem blages and con trasts with the cul tural tax on - omy based on the in dic a tive ar ti fact classes. The limited technological variability in these assemblages is prob a bly due to the fact that a fairly ho - mog e nous group of pri mar ily lo cal raw ma te ri als (radiolarites and flints) were ex ploited through out the se quence (Koumouzelis et al., 1996). The pro - por tion of radiolarites re mains fairly sta ble (60 70%) through time, whereas the pro por tion of flint is more vari able; the low est fre quency of flint is found in the Aurig na cian lay ers (20 29%), and the high est in layer III (42%) and the Mesolithic (33 37%). Extra-local raw materials are rare through out the lay ers, with some of the red radiolarites in layer V be ing of the high est qual ity. Unquestionably exotic materials from the later cultural layers include Melian obsidian from Mesolithic layer 5a. The rel a tively uni form com - po si tion of the raw ma te rial across as sem blages, and the dom i nance of lo cal raw ma te ri als over all, sug gest a fairly lim ited range of mo bil ity of all of the groups while in hab it ing this cave. Lo cal raw materials were determined from field survey of raw ma te rial out crops over a ra dius of 3 4 km around the site. The near est known sources of higher qual ity stone are Me so zoic out crops lo - cated in the north east ern Peloponnese. Sev eral sites on the Argolid, particularly those of the Early Neolithic, contain these high quality raw materials, but Klissoura Cave 1 does not. The structure of the major technological catego ries also is sim i lar across most of the lithic as - sem blages. The large quan ti ties of lithic shat ter/ chunks in all lev els are clearly at trib ut able to the low qual ity of the lo cal raw ma te ri als. Tab u lar frag ments of both radiolarite and flint from the study area con tain many in ter nal frac tures and flaws as a rule. The high fre quency of chips and small flakes is also partly ex plained by the use of poor quality raw materials, but it simultaneously at tests to the in ten sive use of these ma te ri als, in - clud ing the frequent rejuvenation of retouched tools. Shat ter/chunks, chips and small flakes con sti - tute more than half of the lithic ar ti facts. This prop erty of the stone tool as sem blages re duces the in di ces of other tech no log i cal cat e go ries cores, blades, and re touched tools. In di ces for the lat ter categories nonetheless are consistent with values for other sites at which the full cy cle of blank and tool pro duc tion and di verse ac tiv i ties took place. The high in di ces of backed pieces used as in serts in the Uluzzian and Gravettoid in dus tries in di cate the use of hafts, ap par ently made from per ish able ma te ri als. In the Aurig na cian lev els, on the other hand, the bladelet in dex is low, and carenate cores/endscrapers from which the bladelets were de tached, oc cur in large quan ti ties. This pat tern sug gests that some bladelets were taken away from the site dur ing the Aurig na cian oc cu pa tions, hafted as ex change able parts of weap ons, tools or both of these. Mac ro scopic im pact frac tures and microscopic marginal scars indicate that both Aurig na cian mi cro-re touched (or unretouched) bladelets and Gravettoid backed bladelets were used for projectile hunting weapons and as inserts for cutting or scraping tools. The con trast be tween the Aurig na cian and other in dus tries is most ev i dent in the tools/cores ty po logi cal group. In the Aurig na cian lev els, end-

4 312 M. C. Stiner et al. scrap ers/cores (mainly carenate forms) are very com mon, with an in dex of about 60 for the in dus - tries of lay ers IV and IIIe-g. By con trast, the scraper/core in dex is be tween 16 and 36 in the Uluzzian, and about 30 in lev els III and II. Such high tool to core in di ces in the Aurig na cian are due to the dual roles of carenate pieces, which func tioned both as cores for bladelets and as endscrapers. In other lev els, bladelets were de - tached from other, spe cial ized core types, while endscrapers func tioned pri mar ily as tools for ac - tiv i ties such as hide processing. The in di ces of blades and bladelets do not fluc tu ate very much through the Up per Paleolithic se quence, os cil lat ing be tween the val ues of 4 and 6. It is only in the Late Mesolithic (layer 3) that the blade in dex is sig nif i cantly higher (9.5). More sig nif i cant dif fer ences can be seen among the cul - tural lay ers in the fre quency of flake tools (i.e. denticulated, notched and re touched flakes). An es pe cially high in dex of these tools (20) is found in layer III and in some of the Aurignacian layers. DATING The dates for the Gravettoid Aurig na cian se - quence (lay ers III, IIIe-g and IV) are rel a tively well con strained. They show a gen er ally monotonic trend of in creas ing age, from ca C kyrs BP to ca C kyrs BP. These ages are con sis tent with other clas sic or late Aurig na cian sites in south ern Eu rope (Kuhn et al., this issue). The Uluzzian in dus try in layer V clearly predates the Aurig na cian stratigraphically, but the ra - dio car bon re sults for layer V are am big u ous. Based on one date on a sam ple re ported to be from layer V (source lo ca tion is un clear, see Kuhn et al., this is sue), and two dates from layer VI (which is stratgraphically mixed), it is pos si ble to sug gest an age of >40,000 years for layer V. Other ra dio car bon data are sig nif i cantly youn ger and may or may not truly orig i nate from layer V, the lim its of which are lo cally dif fi cult to distinguish from layer IV. For tu nately, microtephra anal y sis con ducted as part of the RE SET pro ject has iden ti fied one ma jor peak con cen tra tion of tephra shards at the in ter face of lay ers IV and V, tail ing up wardly to layer III, along with a mi nor peak at the in ter face of lay ers VI and VII. At tempts cur rently are un - der way to cor re late the tephra(s) with spe cific erup tions of known age (Dustin White, per sonal com mu ni ca tion, 2010). In nearby Franchthi Cave, a wind-blown Y-5 tephra (Campanian Ignimbrite) was found in stra tum Q. Orig i nat ing from the Na - ples area of It aly, this tephra is dated to 39.28± 0.11 kyrs by 40 Ar/ 39 Ar (De Vivo et al., 2001). There is a strong pos si bil ity that Klissoura 1 may also contain this tephra. Im por tantly, the tephra seals layer V in Klissoura 1. The ra dio car bon re sults sug gest that the in dus try of layer V is about 6000 years older than most cur rently re ported ages for sim i lar lithic as - sem blages from Europe. Karkanas (this is sue) con cludes that the tem - po ral gap be tween lay ers VII and V is con sid er - ably greater than the gap be tween lay ers V and IV. Pub lished ages from Uluzzian sites in It aly tend to be sig nif i cantly youn ger than 39,000 years BP (re viewed by Kuhn et al., this is sue). How - ever, al most all pub lished dates for the Uluzzianin south ern Europe should be con sid ered min i mum es ti mates, and cur rent ef forts at re-dat ing the sites are likely to push age estimates backward in time. The ABOX pre-treat ment tech nique was used suc cess fully used to ob tain fi nite dates of >60,000 years on wood char coal from the Mid dle Paleolithic lay ers. Al though these should be also taken as min i mum ages, they il lus trate the tech nique s po ten tial for push ing back the max i mum lim its of ra dio car bon dat ing. ABOX proved less use ful when ap plied to more re cent Up per Paleolithic sam ples for rea sons that may in clude greater wa - ter cir cu la tion or other sources of contamination. The two im por tant depositional/oc cu pa tional hi a tuses in the later part of the Klissoura 1 se - quence cor re spond to ma jor paleoclimatic events. One of these oc curs within se quence B be tween the Med i ter ra nean Backed Bladelet layer III and the Epigravettian (layer II). Be cause layer III is dated at about Kyr and the Epigravettian can not be older than Kyrs BP, this hi a tus should include the Last Gla cial Max i mum (LGM) within Ma rine Ox y gen Iso tope Stage (MIS) 2. Sites dated to the LGM are ab sent in the gorge as well as in other parts of Argolid. The only cul tural en ti ties that are chro no log i cally close to LGM are lithic phase II in Franchthi Cave ( kyrs BP; see Perl s, 1987) and prob a bly layer D3 in

5 Klissoura Cave Kephalari Cave, which un for tu nately lacks ra dio - car bon dates. It is pos si ble that the in land ar eas were de pop u lated dur ing the LGM. The last ma jor hi a tus in Klissoura 1 oc curs be tween the geo log i cal se quences B and A, or be - tween the Epigravettian (layer II) and the lower Mesolithic (layer 5a). Given that the Epigravettian oc cu pa tion al most all cer tainly spans the time in ter val from 16 to 14 kyrs BP, and the Mesolithic is syn chro nous with the Early Ho lo - cene, the sec ond hi a tus must cover the end of the Late Gla cial, span ning at least 12 to 10 kyrs BP. This most re cent hi a tus in Klissoura Cave 1 may be unique to the site, since there are nu mer ous Late/Fi nal Paleolithic sites in the area, in clud ing the dated lay ers from Klissoura Caves 4 and 7 (Koumouzelis et al., 2004), sites in the Voidomatis Gorge, and in other parts of Argolid such as at Kephalari Cave (layer C) and Franchthi Cave (Perl s, 1987). PALEOENVIRONMENTS AND PALEOCOMMUNITIES OF THE ARGOLID The re sults from the bo tan i cal, fau nal and geo - logical analyses suggest a gradual trend toward clima tic cool ing through the Up per Paleolithic se - quence in Klissoura Cave 1. Warmer, wet ter con di tions re turned only well af ter MIS 2, or dur - ing the Mesolithic. These gen eral con clu sions can be qual i fied by in di ca tions about the broader en vi - ronment and the sedimentary environment inside the cave. Grass phytoliths are the most im por tant el e - ments in the phytolith as sem blages from Klissoura 1 (Al bert, this is sue). Most of the iden ti fied spec i mens cor re spond to the C3 festucoid subfam ily, which is very com mon in the Med i ter ra - nean ba sin. The grass phytolith as sem blages of lay ers IIIe-g and IV in di cate only a mod er ately hu mid en vi ron ment. Pytoliths rep re sent ing C4 grasses and prob a bly also reeds (Arundo donax) are pres ent in the Epipaleolithic (II) lay ers, and reed phytoliths oc cur in the up per most por tion of the III se quence (III-III ). Al though reeds re quire very wet con di tions, their pres ence may sim ply in di cate small pock ets of wet land some where in the area and pos si bly lo cal ized changes in wa ter ta bles caused by sea level changes, tec tonic events or other non-cli ma tic fac tors. The pres ence of C4 grasses, on the other hand, sug gests a sig - nif i cantly drier and more open environment during the Epipaleolithic. Wood char coal re mains from the Up per Paleolithic lay ers re flect a mo saic of pe ren nial veg e - ta tion types (Ntinou, this is sue). It is likely that dry park land veg e ta tion cov ered the rocky hills, giv ing way to open wood land with mesophilous and thermophilous trees in the foot hills and val ley floors. Burned wood rem nants of oak (Quercus sp., de cid u ous type) and elm (Ulmus), gen era that pre fer some what moister con di tions such as might oc cur in gul lies and small can yons, are most com - mon in lay ers V though IIIe-g. Elm all but dis ap - pears from the char coal as sem blages there af ter. The wood-char coal re cord of the early part of the Upper Paleolithic sequence indicates interstadial con di tions dur ing mid-mis 3 (40 30 kyrs BP) and grad ual cool ing and dry ing towards the end of the MIS 3 (after 30 kyrs BP). The scope of vari a tion in mois ture that would have been avail able to plants near Klissoura Cave 1 war rants some dis cus sion. Even the ear li est Aurignacian phases, in which some mois ture-de pend - ent tree spe cies are rep re sented, co in cided with generally dry con di tions. A mar ginal bal ance of moisture avail abil ity and wa ter up take by plants was enough dur ing this in ter val to sup port the de - vel op ment of some mesophilous and thermophilous veg e ta tion. While this situation helped to sup - press ero sion in the area dur ing the for ma tion of layer IV (and pos si bly layer V), con di tions were still suf fi ciently dry to pre vent ash in the site from be com ing ce mented by drip ping wa ter. Dur ing the later oc cu pa tions, the en vi ron ment be came very dry and, based on higher rates of clastic sed i - ment ac cu mu la tion, more prone to ero sion dur ing in fre quent storms. The main ex pla na tion for in - creased ero sion in the area is a de cline in pe ren - nial vegetation. Precipitation seems to have increased again only with the Mesolithic, but erosion was con sid er able un til for est de vel op ment caught up with wa ter avail abil ity. The com po si tion of the mam mal and avian fau nas in Klissoura 1 sug gests cor re spond ing changes in an i mal com mu nity struc ture dur ing the Up per Paleolithic through Mesolithic (Bocheñski and Tomek, this is sue; Starkovich and Stiner, this issue). The ungulate assemblages from layers

6 314 M. C. Stiner et al. IIIe-g, IV and V are rel a tively di verse for their sizes. Hares (and tor toises in V) dom i nate the small game frac tions. The as sem blages from the mid dle and up per Aurig na cian lay ers (IIId-a) are less rich in ungulates, and they are dom i nated by one un gu late spe cies in par tic u lar, Eu ro pean fal - low deer (Dama dama). Hunt ing of large and me - dium-sized ground birds (bus tard and par tridges, re spec tively) also be came im por tant in the in ter - val rep re sented by lay ers IIId-a. The dom i nance of fal low deer to gether with the high in ci dence of par tridges and bus tards in the later Aurig na cian lay ers sug gests an ex pan sion of open grassy ar - eas. Un gu late di ver sity ex pands again in the Mesolithic (and pos si bly in the Epipaleolithic, but this is a small sample), and hares once again dominate the small game fraction. It is un likely that the vari a tions in un gu late spe cies di ver sity stem mainly from hu man hunt - ing pref er ences. Changes in tem per a ture and sea level forced qual i ta tive shifts in the struc ture of terrestrial animal communities on the Argolid. A more het er o ge neous en vi ron ment would sup port a broader range and more even pro por tions of un - gu late spe cies, be cause greater macro-struc tural vari a tion in avail able hab i tats makes it more dif fi - cult for one spe cies to outcompete oth ers. Rel a - tively het er o ge neous hab i tats are in di cated by the fau nal and bo tan i cal re sults for the ear li est Up per Paleolithic. Sub se quently, drier or cooler con di - tions pre vailed and veg e ta tion be came more uni - form, al low ing Dama pop u la tions to dom i nate lo - cally. The small an i mal com po nent of the di etary spec trum poses the main con tra dic tion strictly cli - mate-driven pat terns in prey choice. The over all con tri bu tion of small an i mals to the meat diet in - creased dra mat i cally in layer III and above. This is clearly ap par ent within the ver te brate as sem - blages, but also from the ris ing eco nomic im por - tance of large land snails. Ex pan sions in di etary breadth are gen er ally thought to rep re sent ei ther tem po rary or long-term re sponses by con sum ers to the de cline in the most prof it able re sources (Stephens and Krebs, 1986) large game in the case of the Paleolithic. Such trends may rep re sent cul tural pref er ences only in the sense of their be - com ing per ma nent so lu tions with time, sup ported by sig nif i cant tech no log i cal in vest ments made in spite of other im por tant de mands on for ag ers time. It is likely, there fore, that the re lent less ex - pan sion in di etary breadth ev i denced in Klissoura Cave 1 re flects a grow ing hu man eco log i cal foot - print in the re gion and prob a bly also mild in - creases in hu man pop u la tion den si ties. That the trend is ev i denced prin ci pally within the small game frac tion of the fau nas is not sur pris ing, as these re sources were the main means for fill ing gaps in the avail abil ity of large game an i mals (Kelly, 1995; Kuhn and Stiner, 2006). The cul tural se quence of Franchthi Cave on the south ern Argolid partly over laps with the most re cent part of the Klissoura 1 chro nol ogy (Perl s, 1987; Far rand, 2000). The Aurig na cian fauna from Franchthi is small and as yet un - der-doc u mented, but the large Gravettoid as sem - blage dated to ca kyrs ap pears to fol low in time the III layer se ries of Klissoura 1. Al though the south ern Argolid ex pe ri enced sea changes much more di rectly than the Klissoura Gorge area, the Franchthi data are of com par a tive in ter - est. 1 Five of the un gu late spe cies found in the Klissoura 1 fau nas also oc cur in the up per Aurig - na cian, Gravettoid, Epipaleolithic, and Mesolithic lay ers of Franchthi Cave (Stiner and Munro, n.d.; Payne, 1975, 1982). Both red deer and Eu ro - pean wild ass were im por tant prey in the early part of the Franchthi se quence, but red deer was the only sig nif i cant large prey item in the later part. Other ungulates are rep re sented in low fre - quen cies, namely aurochs, wild pig and ibex. In - ter est ingly, no fal low deer re mains were found in Franchthi, 2 de spite the sin gu lar im por tance of this spe cies in Klissoura 1. The con trast in dom i nant deer spe cies in Klissoura 1 and Franchthi are one of sev eral lines of ev i dence that cli mate-driven en vi ron men tal con di tions were strongly in flu - enced by lo cal fac tors on the Peloponnese. An other con trast in un gu late rep re sen ta tion be tween the two sites con cerns the Eu ro pean wild ass. In Franchthi, wild ass re mains dom i nate during the Gravettoid oc cu pa tions (Stiner and Munro, n.d.). Deer dom i nate the en tire Up per Paleolithic Mesolithic se quence in Klissoura 1 where wild asses were al ways rare. The im por tance of wild ass at Franchthi Cave dur ing the Gravettoid oc cu pa tions must re late to more open con di tions on the south ern end of the pen in sula go ing into the LGM. Eu ro pean equids of the Late Pleis to - cene are thought to have pre ferred open and

7 Klissoura Cave steppic con di tions, par tic u larly E. hydruntinus. Mod ern fal low deer feed mainly in open grassy ar eas but must have some tree cover for pro tec tion from the el e ments and pred a tors. The op ti mal hab i tat for fal low deer there fore is de cid u ous and mixed wood lands on gently roll ing ter rain. Red deer, the dom i nant deer spe cies in Franchthi, are graz ers by pref er ence but can also feed on dwarf shrubs such as heather and other low qual ity browse, pro vided that con di tions are rel a tively moist. They are also more tol er ant of wet winds and cool, ex posed con di tions than are fal low deer. The find ings on the shell or na ments from Klissoura Cave 1 also speak to ques tions about the de gree of en vi ron men tal het er o ge ne ity in south - ern Greece dur ing the Late Pleis to cene (Stiner, this is sue). Al though Klissoura 1 was never sit u - ated on the Aegaen shore dur ing the Paleolithic or Mesolithic, the in hab it ants vis ited the sea and other aquatic hab i tats, and they brought many small or na men tal shells back to the site. These shells fall within a nar row range of sizes and shapes. How ever, the spe cies col lected dur ing the ear li est Up per Paleolithic phases are quite var ied, whereas few spe cies were uti lized for or na men tal pur poses in the later oc cu pa tions (above lay ers IIIg-e). The great va ri ety of or na men tal mol lusk spe cies in the as sem blages from lay ers IV and V re flect a mo saic of aquatic hab i tats, more com plex than ex ists in the Peloponnese to day. The tax o - nomic di ver sity of the early Up per Paleolithic or - na ment as sem blages from Klissoura 1 also greatly ex ceeds that of ev ery post-aurig na cian or na ment as sem blage from Franchthi Cave (Shackle ton, 1988; C. Perl s, per sonal com mu ni ca tion, 2010). The more re cent shell as sem blages from Klissoura 1 and Franchthi in vari ably are dom i nated by just a few brack ish wa ter and lagunal mol lusk species (Columbella rustica and Cyclope spp.). Re duc tion in tax o nomic di ver sity in the ma rine shell types was al most cer tainly linked to changes in sea level. Based on the dat ing and en vi ron men tal data, the early Up per Paleolithic part of the Klissoura 1 cul tural se quence cor re lates with an interstadial and the last mi nor high sea stand (ca. 35 kyrs cal i - brated BP), be fore the big drop in sea level that be gan about 30 kyrs (cal i brated BP) and cul mi - nated in the LGM (Chappel, 2002; Wright et al., 2009). The high sea stand in the early Up per Paleolithic helps to ex plain the great di ver sity of coastal and in land hab i tats, since el e vated wa ter ta bles would re sult in the for ma tion of marshes, la goons and sand bars. Ac cord ing to van Andel et al. (1990), for ex am ple, the north ern part of the Argolis Gulf is an ex tended flat shal low area less than 50 m deep. During the last high sea stand of MIS 3a this area would have been only partly sub - merged and thus would have sup ported a wider range of aquatic hab i tats. The de cline in sea lev els there af ter prob a bly also ex plains the abrupt de - cline in or na ment shell diversity in the layers above IIIe-g in Klissoura 1. HUNTING PATTERNS Klissoura Cave 1 lies at the in ter face of rug - ged hills and a large plain, thereby pro vid ing ac - cess to a var ied food sup ply. The site is lo cated stra te gi cally where the gorge opens onto the up - per Argos plain. There is no in di ca tion of mass hunt ing of ungulates at this site. Rather the data sug gest op por tu nis tic hunt ing of a va ri ety of spe - cies, one or a few at a time (Starkovich and Stiner, this is sue). As the lo cal an i mal com mu nity changed with cli mate and veg e ta tion, hunt ers responded op por tu nisti cally and pur sued what ever un - gu late spe cies were avail able. There is the ques - tion of where the Up per Paleolithic in hab it ants of Klissoura 1 ob tained the ibex and cham ois (layer IV only), since the area does not in clude true al - pine hab i tats. In fact ibex may in habit a much wider va ri ety of el e va tions, pro vided that the ter - rain is rug ged (Phoca-Cosmetatou, 2004). The low but per sis tent pres ence of ibex in the un gu late fau nas may sim ply re flect the ex tent to which hunt ers chose to search craggy up lands nearby. To day cham ois tend to oc cupy rocky or al pine ar - eas, but they along with wild goats may de scend to much lower, for ested pas tures in win ter (Forsyth, 2000; Baumann et al., 2005). Small game ex ploi ta tion at Klissoura 1 ex - hibit a pat tern of in creas ing di etary breadth. Gen - er ally sim i lar trends have been doc u mented in other re gions of the Med i ter ra nean Ba sin (Tchernov, 1992; Hockett and Bicho, 2000; Stiner et al., 2000; Stiner, 2001; Munro, 2004; Manne and Bicho, 2009). Spe cif i cally, there is a de crease through time in the pro por tion of small, slowmov ing game spe cies and an in creas ing re li ance

8 316 M. C. Stiner et al. of more pro duc tive quick small an i mals such as hares and/or birds. At Klissoura 1, this trend be - gins around the time of the Mid dle to Up per Paleolithic tran si tion (Starkovich, 2010), though oc ca sional hare ex ploi ta tion oc curred be fore this time, in the late Mid dle Paleolithic. The rel a tive con tri bu tion of small an i mals to the meat diet in - creased fur ther in the most re cent lay ers (3-5a, II), where the NISP counts for quick small game ac tu - ally sur pass those for me dium-sized ungulates (Bocheñski and Tomek, this issue; Starkovich and Stiner, this issue). The shells of the large ed ible land snail, Helix figulina, also are prev a lent in the late Up per Paleolithic through Mesolithic strata of Klissoura 1. Most of these were mod i fied by hu mans rather than small pred a tors, al though none is burned by fire (Koumouzelis et al., 2001; Starkovich and Stiner, this is sue). The rel a tive quan tity of land snails in the archaeofaunas in creases ex po nen - tially with time, and shell sizes be come larger and more uni form. Spe cif i cally, land snails are rare in lay ers IV or V and show no clear ev i dence of hu - man mod i fi ca tion, whereas hu man-mod i fied snail shells are mod er ately abun dant in layer IIIc and in crease greatly through lay ers III and III, and snail abun dance peaks in Mesolithic lay ers 3-5a. Epigravettian layer IIa-b rep re sents a strik ing ex - cep tion in that snails are un com mon and a wide range of tiny to large spe cies are rep re sented, sim - i lar to the nat u ral snail as sem blages that lit ter the ground in the site vi cin ity to day. Snails are not dif fi cult to find or col lect af ter heavy rains, but cook ing and ex trac tion is rel a tively la bor in ten - sive. Other find ings on Up per Paleolithic sub sis - tence at Klissoura 1 re late to large game hunt ing, spe cif i cally the pat terns of prey age se lec tion and food trans port. Mi nor bi ases were found for body part rep re sen ta tion in this site. These bi ases are not ex plained by in situ at tri tion and there fore must re flect hu man trans port de ci sions. The parts of un gu late skel e tons are fairly evenly rep re - sented, ex cept for the scar city of ax ial el e ments be low the neck. Nearly all meaty parts of car - casses were brought to the site for pro cess ing, and ax ial parts were of ten left at or near the kill sites. In fact the dressed weight for most of the prey an i - mals would not have ex ceeded what a few hunt ers could carry back to camp within a day. The mor - tal ity pat terns of the ungulates in di cate fairly even rep re sen ta tion of young, prime adult and old adult in di vid u als in all lay ers ex cept the early Aurig na - cian (IV), where old adults are less well rep re - sented. The lack of strong age bi ases in the un gu - late fau nas sug gests a con sis tently emphasis on encoun ter hunting, without focusing on sexsorted herds. Small quan ti ties of fe tal or ne o nate re mains are pres ent in the un gu late as sem blages through - out the se quence, and most of these rep re sent un - born an i mals. Hunt ing of preg nant fe male fal low deer must have taken place be fore or dur ing the spring birthing sea son (late May June). Other seasons of occupation are not necessarily ex - cluded by these find ings, how ever, since the phytolith ev i dence from lay ers IIIg, and to a lesser de - gree in IIIe, in di cates the pres ence of grass flow - ers. In mod ern Greece, grasses may be gin bloom - ing in March, but most inflorescenses de velop from April to June, and at much lower fre quen cies from July to Sep tem ber (Al bert, this is sue). Hearths and large vol umes of wood ash are also a ma jor com po nent of the Klissoura 1 re cord, and some of the oc cu pa tions must have included cool months of the year. Deer ant ler is pro por tion ately common in the Mesolithic (3-5a), though this is a small sam ple, and in lay ers III, IIIe-g, IV, and V (also a small sam ple). Worked ant ler and (rarely) bone ar ti facts oc cur in all of the lay ers, but most of these are from Aurig na cian lay ers IIIe-g and IV. The lay ers that con tain the most ant ler frag ments of any sort gen er ally also con tain the most worked ant ler ar ti - facts, with the ex cep tion of the Mesolithic (Starkovich and Stiner, this is sue). Mi cro scopic anal y - sis of the ant ler frag ments re vealed few un equiv o - cal ex am ples of trim ming de bris from ant ler work ing, but the lack of such ev i dence in this site may be ex plained by the ex ten sive breakage of the ant ler by the hu mans and post-depositional mi crosur face al ter ations (Christidou, personal communication, 2010). Male fal low deer pos sess ant lers from roughly July to April (Chap man and Chap man, 1975), and the ant lers harden in time for the au - tumn rut. Some of the male fal low deer in the Up - per Paleolithic Klissoura as sem blages there fore must have died in the colder months of the year (au tumn through early spring). It is pos si ble, how -

9 Klissoura Cave ever, that ant ler was col lected and curated over long pe ri ods in an tic i pa tion of tool-mak ing. Thus the pres ence of ant ler in di rect con nec tion with os se ous tool-mak ing may not pro vide a re li able in di ca tion of the season of occupation. PALEOLITHIC ORNAMENTS Shell or na ments oc cur through out the Up per Paleolithic, Epipaleolithic and Mesolithic lay ers (Stiner, this is sue). As is gen er ally true of the Up - per Paleolithic in Eu rope, the or na ment as sem - blages from Klissoura Cave 1 are well de vel oped in char ac ter. The ear li est Up per Paleolithic or na - ments oc cur in layer V in as so ci a tion with an Uluzzian in dus try. The larg est as sem blage of or - naments comes from the earliest Aurignacian or layer IV. There are a few or na ments in lay ers VI VII, but most of these were found im me di - ately be low the area of the man-made shel ter in layer IV. Layer V does not ex tend to this area of the ex ca va tion, and taphonomic ev i dence and di - rect dat ing of some of the shells (K. Douka, per - sonal com mu ni ca tion, Feb ru ary 2010) in di cate that the shells in VI VII represent very localized downward intrusions from layer IV. The Klissoura 1 or na ment as sem blages dif fer from those typ i cal of coastal sites in that the Klissoura 1 assemblages consist almost exclusively of finished products. There is considerable evidence of high-grad ing or hu man se lec tion of the as sem - blages for har mony in shell color, form and qual ity, and there are few, if any, ex am ples of man u fac tur - ing er rors. The prev a lence of cord-wear sug gests that many of the or na ments ar rived al ready at - tached to or ganic ma te ri als or hu man bod ies. What break age oc curred to the shells re sulted pri mar ily from long-term use. Faded or worn shells of spe - cies that would have orig i nally been red in color (Clanculus spp.) were re newed with red ochre. The ornament assemblages from the earliest Up per Paleolithic lay ers are par tic u larly rich in spe cies. The high fre quency of Dentalium (tusk) shells none the less sets the small Uluzzian or na - ment as sem blage apart from the Aurig na cian and later bead as sem blages in the site. Dentalium shells are also prom i nent in the Uluzzian ho ri zons of Grotta del Ca val lo in south ern It aly (Palma di Cesnola, 1966), though sam ple size vari a tion may be a complicating factor. The or na ment is con sid er able vari a tion in abun dance through the Up per Paleolithic and Mesolithic lay ers in Klissoura 1 is not ex plained by dif fer ences in the thick ness of the ex ca vated lay ers. As noted above, the as sem blage from the earliest Aurignacian layer IV is exceptionally large, and most of the or na ments from this layer oc cur within the in ferred per im e ter of the manmade shel ter, ringed by large stones. This fea ture is sur rounded by hearths but none oc curs within the inferred area of the shel ter where the or na - ments are concentrated. SITE FUNCTION AND OCCUPATION INTENSITY The va ri ety of fea tures, ar ti facts and fau nal re mains in the early Up per Paleolithic lay ers of Klissoura 1 in di cate that the site served as some kind of res i den tial base dur ing most or all of these oc cu pa tions. The in ten sity or du ra tion of the oc - cu pa tions prob a bly var ied greatly, how ever, with the most in tense use of the site oc cur ring dur ing the for ma tion of Aurig na cian lay ers IV and IIIe-g. In ad di tion to many clay-lined and un lined hearths, these lay ers con tain a di verse as sort ment of lenses, pits and other fea tures. Ant ler points and prob a ble man u fac tur ing de bris, mainly on ant ler, are par tic u larly abun dant in layer IV. Un - like the sit u a tion in the lay ers above, where os se - ous ar ti facts are widely scat tered among hor i zon - tal units, worked ant ler speciments are spa tially con cen trated within and around the im me di ate area of the shel ter fea ture in layer IV (Christidou, per sonal com mu ni ca tion, 2010). The in di ca tions of hu man ac tiv i ties are only some what less var ied in the later part of the Up per Paleolithic-Epipaleolithic sequence. Unfortunately, the up per lay ers suf fered from greater amounts of ero sion or dis tur bance, pos si bly re - ducing the diversity of visible activity areas. For ex am ple, the ma te rial in lay ers 6, 6a and 6/7, which in cludes a mix ture of Aurig na cian and other artifacts, represents material that was dumped ep i sod i cally into a large pit. Its sed i ment is ho mog e nous, gen er ally loose and po rous, and the ma trix is al most pure ash and con tains more than the usual pro por tion of frag mented lithic debitage. Nor mally such concentrated re fuse would as so ci ate with an in ten sive oc cu pa tion, but

10 318 M. C. Stiner et al. we find no hor i zon tal layer as so ci ated with the pit. A significant erosional hiatus is apparent in this part of the strati graphic column (Karkanas, this issue), however, and some sed i ment may have been lost during the LGM. Plant phytoliths are abun dant in most of the Up per Paleolithic sed i ment sam ples, but they gen - er ally are not found in the hearths. The con trast in phytolith dis tri bu tions be tween hearths and open areas testifies to the horizontal integrity of features in the Up per Paleolithic lay ers, con sis tent with the geological observations. The input of plant matter into fire places was se lec tive mainly dicotyledonous wood and bark-pro duc ing plants. It is the grasses that show the great est spa tial sep a ra tion from the hearths. The abun dant grass in flo res cen - ces in Layer IIIe-g may point to their use as food, or the har vest ing of ma ture stems for fi ber work - ing or bed ding. Small amounts of sedge phytoliths in layer IV, and reed phytoliths in lay ers II and III-III could also re late to fi ber work ing on site. There is considerable evidence of osseous tech - nol ogy use and pro duc tion on site (Christidou, per sonal com mu ni ca tion, 2010), par tic u larly in lay ers III IIIe-g and IV. Plant fi ber work ing and hunt ing may rank among the many po ten tial uses of these tools. Clearly, a wide range of ac tiv i ties took place in the cave, a sit u a tion typ i cal of base camps. The re cord of fire use in the Up per Paleolithic lay ers of Klissoura 1 is un usu ally com plex (Karkanas et al., 2004; Karkanas, this is sue). Hearths are of ten su per im posed in Klissoura 1, with re - peated build ing and main te nance of fires in cer - tain ar eas of the site. The func tion of the claylined hearths in the Aurig na cian lay ers is dif fi cult to in ter pret from the sparse wood char coal re - mains alone. How ever, the scar city of char coal in these lay ers might be the re sult of in ten tional pro - duc tion of hot em bers for use in the clay hearths. Em bers the in can des cent stage of a fire al low one to ex ploit the prop er ties of con duc tion and con vec tion to heat a small, semi-en closed space. Com plete com bus tion from con tin u ous stok ing of em bers would de stroy most of the wood char coal pro duced in these fires. Em bers may also be used to cook food in di rectly with ra di ant heat, for dry - ing and cur ing of food and other ma te ri als, and pos si bly heating sweat lodges (sau nas). Em ber trans fer and mul ti ple uses of hearths likely oc - curred dur ing the for ma tion of the Up per Paleolithic layers, particularly in layers where claylined hearths co-occur with unlined hearths. CONCLUSION The Paleolithic de pos its in Klissoura Cave 1 preserve a well-ordered stratigraphy that covers a large por tion of the Late Pleis to cene Mid dle Paleolithic and Up per Paleolithic, along with Epipaleolithic and Mesolithic com po nents. The Up - per Paleolithic re cord is es pe cially well pre served and com plex due to a dense as sort ment of in tact features and artifacts. These cultural components pro vide a rich and only partly over lap ping com - ple ment to the Paleolithic re cord of Franchthi Cave on the southern Argolid. Among the im por tant find ings of this re - search on the Up per Paleolithic through Mesolithic in Klissoura Cave 1 are the iden ti fi ca tion of an early Uluzzian oc cu pa tion in layer V that may be more than 39,000 years old, the first Up per Paleolithic oc cu pa tion in the strati graphic se ries, fol lowed by a long and rel a tively well-dated Aurignacian sequence. Ornament assemblages appear sud denly with the on set of the early Up per Paleolithic (Uluzzian) in this site, and all in tact (un - mixed) Mid dle Paleolithic lay ers that un der lie the Up per Paleolithic lack or na ments en tirely. Some inter-stratification of Aurignacian, later Uluzzianlike (III ) and Gravettoid (III ) com po nents is sug gested on the ba sis of for mal ar ti fact types. The lithic in dus tries in some lay ers of Klissoura 1 show de mon stra ble, pe ri odic links to Paleolithic populations in ltaly, particularly the Uluzzian of layer V. Zooarchaeological find ings in di cate early in - creases in di etary breath dur ing the Up per Paleolithic, con sis tent with trends ob served in other Mediterranean ar eas where this phe nom e non has been stud ied in ten sively. Di ets ex panded fur ther dur ing the later Paleolithic and Mesolithic oc cu - pa tions. The small game trends in Klissoura 1 and other Mediterranean faunal sequences are not ex - plained by cli mate driven en vi ron men tal changes, since they per sist through MIS 3 and gen er ally in - ten sify in MIS 2 (Stiner, 2001). 3 Paleoenvironmental data from the stud ies of in tact char coal frag ments, plant phytoliths, sed i - mentary characteristics, and animal species repre-

11 Klissoura Cave sen ta tion in di cate mod er ate changes in mois ture avail abil ity over the Up per Paleolithic through Mesolithic se quence. Over all, dry con di tions pre - vailed in the area through out the Up per Paleolithic, but mois ture avail abil ity was some what greater dur ing the for ma tion of lay ers V and IV. Lo cal en vi ron ments be came drier and cooler dur - ing the for ma tion of the mid dle and up per Aurig - na cian lay ers and more so through the Epipaleolithic. Mois ture availability increased again only in the Mesolithic. Im por tant hi a tuses marked by ero sion events sep a rate the Mid dle and Up per Paleolithic sed i - men tary groups. There is no seam less tran si tion be tween the two cul tural en ti ties in this site. Depositional hi a tuses also sep a rate the Uluzzian from the ear li est Aurigancian, and the Up per Paleolithic layer se ries from the Epipaleolithic. There is no re cord of hu man oc cu pa tion at Klissoura 1 dur ing the Last Gla cial Max i mum nor, ap - par ently, at the nearby site of Franchthi Cave (Farrand, 2000). Klissoura Cave 1 pro vides con sid er able de tail about the daily ex is tence of Up per Paleolithic for - ag ers at one site. The sed i men tary fea tures, ar ti - facts and an i mal re mains tes tify a wide range of on-site activities that is typical of residential bases. Anthropogenic pro cesses greatly shaped the char - ac ter of the Up per Paleolithic through Mesolithic sediments, par tic u larly the many cy cles of hearth building, cleaning, renovation and trampling. Humans were also the main sources of dis tur bance in the cave de pos its, as the re sult of cre at ing and cleaning hearth bas ins, clear ing ac tiv ity sur faces, and dig ging small pits. The re mark able clay hearths of Klissoura 1 Aurig na cian are unique as of this writ - ing, and may have been used as sat el lite fire in stal - la tions to which hot coals were moved for the pur - poses of cook ing or heat ing ac tiv ity ar eas. The in ten sity of the oc cu pa tions seems to have var ied over time, with those of layer IV be ing par tic u larly in tense. Greater post-depositional dis tur bances to the later lay ers may have un der mined the vis i bil ity of fea tures, and thus the rel a tive in ten sity of the later oc cu pa tions is more dif fi cult to judge, but site use does seem to have been lighter dur ing the Epipaleolithic and Mesolithic. The cul tural se quences iden ti fied at Klissoura Cave 1 and at other im por tant sites in south ern Greece (see Perl s, 1987, 1999 on Franchthi Cave; Panagopoulou et al., on Lakonis) testify to significant regional differences the clas sic chro nol o gies of west ern and cen tral Eu - rope be yond the Bal kans (e.g., Koz³owski, 1999). As might be ex pected for a part of the world de - fined by dis tinc tive eco sys tems and a uniquely bro ken and di verse to pog ra phy, there is much ev i - dence for re gional or en demic pat terns of cul - tural evo lu tion. While pro cesses that pro mote cul - tural di ver gence are to be ex pected for pen in su lar con di tions, as the Peloponnese cer tainly rep re - sents, there were also sig nif i cant in ter vals of in - creased con tact west wardly across the up per Adri atic sea bed, such as dur ing the Uluzzian. REFERENCES BAUMANN M., BABOTAL C., SCHIBLER J Na tive or nat u ral ized? Val i dat ing al pine cham ois hab i tat mod els with archaeozoological data. Eco - log i cal Ap pli ca tions 15(3), BERKE H The dis tri bu tions of bones from large mam mals at Petersfels. In: H. Berke, J. Hahn, C.-J. Kind (eds) Up per Palaeo lithic Set tle ment Pat terns in Eu rope. Verlag Archaeologica Venatoria, Institut für Urgeschichte der Universitat Tübingen, Tübingen, CHAP MAN D.I., CHAP MAN N.G Fal low Deer, their history, distribution and biology. Terence Dal ton Ltd., Lavenham, Suf folk. CHAPPELL J Sea level changes forced ice break outs in the Last Gla cial cy cle: new re sults from coral ter races. Qua ter nary Sci ence Re views 21, DE VIVO B., ROLANDI G., GANS P.B., CALVERT A., BOHRSON W.A., SPERA F. J., BELKIN H.E New con straints on the pyroclastic erup tive his tory of the Campanian vol ca nic Plain (It aly). Mineralogy and Petrology 73, FAR RAND W.R. (with JA COB SON T.W.) Depositional his tory of Franchthi Cave: stra tig ra - phy, sedimentology, and chro nol ogy. Excavations at Franchthi Cave, Greece, fasc. 7. Indiana University Press, Bloomington, IN. FORSYTH D.M Hab i tat se lec tion and co ex is - tence of the al pine cham ois (Rupicapra rupicapra) and Hi ma la yan tahr (Hemitragus jemlahicus) in the east ern South ern Alps, New Zea land. Jour nal of Zo - ology, London 252, HOCKETT B., BICHO N The rab bits of Picareiro Cave: Small mam mal hunt ing dur ing the late Up per Palaeo lithic in the Por tu guese Estremadura. Journal of Archaeological Science 27,