Veget Hist Archaeobot (2006) 16: 15–22 DOI 10.1007/s00334-006-0041-2

ORIGINAL ARTICLE

Spassimir Tonkov · G¨oran Possnert · Elissaveta Bozilova

The lateglacial vegetation and radiocarbon dating of Lake Trilistnika, Rila Mountains (Bulgaria)

Received: 3 February 2005 / Accepted: 19 December 2005 / Published online: 10 March 2006 C Springer-Verlag 2006


Abstract Pollen analysis, supplemented by 5 14 C dates, was carried out on the lateglacial section (160 cm) of a core retrieved from the glacial Lake Trilistnika (2216 m) in the northwestern Rila Mountains (Bulgaria). The reconstruction of the lateglacial vegetation history is linked for the first time to a chronological framework for the time window 13000–10000 b.p. The delimitation of an interstadial/stadial cycle, analogous with the Bølling/AllerødYounger Dryas from Western Europe, is based on important changes in the pollen stratigraphy. Mountain-steppe vegetation composed of Artemisia, Chenopodiaceae, Poaceae and other cold-resistant herbs, with isolated stands of Pinus and Juniperus-Ephedra shrubland, dominated the landscape after the ice-retreat. Interstadial conditions were established before 12815 ± 130 b.p. The spread of tree vegetation at lower elevation as a response to the climate improvement was confined to the time interval 12110 ± 95–11140 ± 75 b.p. The Younger Dryas stadial is characterised by a re-advance of the mountain-steppe vegetation. The results are compared with other sites of lateglacial age from the high mountains (Rila, northern Pirin, western Rhodopes) in southern Bulgaria. Keywords Pollen . Lateglacial vegetation . Radiocarbon dating . Lake Trilistnika . Rila Mountains . Bulgaria

Communicated by F. Bittmann S. Tonkov (
) · E. Bozilova Laboratory of Palynology, Department of Botany, Faculty of Biology, Sofia University “St. Kliment Ohridski”, 8 Dragan Tsankov blvd., 1164 Sofia, Bulgaria e-mail: [email protected] G. Possnert Angstrom Laboratory, Division of Ion Physics, 14 C–Lab, Uppsala University, Lagerhyddsv. 1, Uppsala, Sweden

Introduction The location of Bulgaria in the central part of the Balkans provides suitable high-altitude sites sensitive enough to record environmental changes since the last deglaciation. The intensive palynological research on continuous lateglacial and Holocene deposits from lakes and peat-bogs in the high mountains of southern Bulgaria (Rila, Pirin, western Rhodopes) contributes to the elucidation of issues related to dynamic patterns of vegetation and climate, location of possible tree refugia, migration processes, absolute chronology and human impact (Bozilova et al. 1989, 1990; Huttunen et al. 1992; Bozilova and Tonkov 2000; Tonkov et al. 2002a; Stefanova and Ammann 2003; Atanassova and Stefanova 2003; Filipovitch and Lazarova 2003) (Fig. 1). Such studies contribute to the understanding of the key role of the Balkan peninsula, with its complex topography, interaction of several climatic influences and rich flora and vegetation, in supporting one of the southernmost European survival areas for temperate deciduous and coniferous trees during the last glaciation (Lang 1994; Willis 1994; Taberlet and Cheddadi 2002; Willis and van Andel 2004). The Holocene vegetation development of the high mountains in southern Bulgaria is comparatively well known and complemented in most cases by consistent radiocarbon chronologies, unlike the existing blank spots in the characteristics of the lateglacial environmental changes and the driving mechanisms behind them. Recent attempts to fill in some of these gaps focused on tentative bio- and chronostratigraphical subdivion of sediments of lateglacial age (Bozilova and Tonkov 2000; Tonkov et al. 2002a; Atanassova and Stefanova 2003) but the insufficient radiocarbon control has hampered the presentation of more definite conclusions. The present paper reveals the results of pollen analysis and radiocarbon dating on the lateglacial section of a core from Lake Trilistnika in the Rila Mountains. This study was undertaken as a part of the continuing palaeoecological research programme on cores from three glacial lakes in the cirque of the Seven Rila Lakes. Of these, the

16

Fig. 1 A: Location of Bulgaria with Rila Mountains (filled circle) on the Balkan peninsula. B: Map of Bulgaria with location of lateglacial sites under discussion – 1 Lake Trilistnika and Lake Sedmo Rilsko (Bozilova and Tonkov 2000); 2 Lake Sucho Ezero (Bozilova et al. 1990); 3 Lake Ribno Banderishko (Tonkov et al. 2002a) and Lake Dalgoto (Stefanova and Ammann 2003); 4 Lake Kremensko5 (Atanassova and Stefanova 2003); 5 Kupena Mire (Bozilova et al. 1989; Huttunen et al. 1992); 6 Peat bog Shiroka Polyana–1 (Filipovitch and Lazarova 2003)

lowermost lake, Sedmo Rilsko, has already been investigated for pollen (Bozilova and Tonkov 2000) and diatom content (Lotter and Hofman 2003). A comparison with local and regional lateglacial pollen sequences from other mountainous sites is presented and discussed in the light of the new evidence from Lake Trilistnika. The study area Geology, geography, climate and modern vegetation The Rila Mountains (2925 m) are the highest massif on the Balkan peninsula and lie between 41◦ 52 30 – 42◦ 21 40 N and 23◦ 01 22 –24◦ 01 00 E in southwestern Bulgaria (Fig. 1). Geologically the massif is composed of Palaeozoic metamorphic rocks (crystalline schist and partly marble) and intrusive rocks (granite). During the last two glaciations typical glacial relief forms were shaped such as cirques, numerous lakes, moraines and trough riverbeds. The time of the last deglaciation is still under discussion (Velˇcev 1995). The climate below 1000 m is transitional between the continental and the submediterranean climates. Above this altitude it is montane, and the mean temperature decreases by 0.5 ◦ C for each 100 m. The mean annual precipitation above 1000 m is 800–1000 mm. The highest values, much of it snow, are recorded in the zone 1300–2400 m (Tishkov 1976). The present vegetation of the mountains has six vegetation altitudinal belts (Velˇcev and Tonkov 1986; Bondev 1991). A comprehensive description of the structure and composition of these vegetation belts in the study area has been presented in an earlier publication (Bozilova and Tonkov 2000) so only a brief summary follows. The nomenclature for vascular plants is after Flora of Republic of Bulgaria (Jordanov et al. 1963–95).

The lowermost xerothermic, xeromesophilous oak and hornbeam belts (up to 1000 m) shelter different oak species (Quercus cerris, Q. pubescens, Q. frainetto, Q. dalechampii) and hornbeams (Carpinus orientalis, C. betulus) with a mixture of Ulmus glabra, Acer platanoides and A. pseudoplatanus. Fragments of beech forests (1000 to 1600 m) occupy areas in the northern and western parts of the mountains. The dominant tree is Fagus sylvatica, which also forms mixed stands with Abies alba and Picea abies at higher altitudes. The coniferous belt (1600 to 2000–2200 m) is the bestdeveloped vegetation belt, composed mainly of forests of Pinus sylvestris and Picea abies. The Balkan endemic Pinus peuce forms a sub-belt and shapes the tree-line in the western and southern parts of the mountain. Communities of Abies alba, Betula pendula, Populus tremula and Juniperus sibirica also occur. The subalpine belt (2200 to 2500 m) on northern slopes is dominated by a thick formation of Pinus mugo (dwarf-pine) with Juniperus sibirica, Vaccinium myrtillus and herbaceous communities on open areas. Alnus viridis is restricted to the mountain brooks along the steep, rocky slopes. The arctic-alpine character of the vegetation is indicated by the presence of dwarf-willows such as Salix lapponum, S. reticulata and S. waldsteiniana. The alpine belt (2500 to 2925 m) is dominated by the herbaceous communities of Carex curvula, Festuca airoides, F. riloensis, Dryas octopetala, Salix herbacea etc. The vegetation of the subalpine and alpine belts is rich in endemics. Most of them are classified as glacial relicts that survived the harsh conditions of the glaciations. The present-day vegetation in almost all belts is highly influenced by anthropogenic activity, and the tree line has been artificially lowered. Study site The cirque of the Seven Rila Lakes is the largest one in the Rila Mountains, being elongated and extended in a northerly direction. The lakes are grouped in a tier on the slope between 2535 and 2095 m (Fig. 2). The site of the present study is the fifth lake, named Trilistnika (2216 m, 42◦ 12 01 N, 23◦ 18 56 E), located in the subalpine belt and surrounded by groups of Juniperus sibirica and Pinus mugo within herb vegetation. The lake collects water from the higher lakes. The inlet is at its southern edge while two outlets are found in the northern and eastern parts. The shore is flat, in some places marshy. Only the western slope is steep and rocky. The general characteristics of the lake are presented in Table 1. Material and methods Core and lithology In September 1994 a sediment core was recovered from a platform at the central deepest part of the lake with a

17 Table 1

General characteristics of Lake Trilistnika (Ivanov 1964)

Altitude (m) Legth (m) Maximum width (m) Mean width (m) Area of the lake (ha) Maximum depth (m) Mean depth (m) Volume (m3 )

2216 240 200 110 2.6 6.5 2 54000

pollen of aquatics and Cyperaceae. Their representation is expressed as percentages of the PS. The identification of spores and pollen was made using the reference collection of the Laboratory of Palynology, the keys in Beug (2004), Faegri and Iversen (1989) and Moore et al. (1991). The construction of the pollen diagrams (Figs. 3–5) was done with the program TGView version 2.0.2 (Grimm 2004). For the first time a determination of the zone boundaries by CONISS (Grimm 1987) was applied to the lateglacial sequences from the Rila Mountains in an attempt to synchronise the observed changes in pollen stratigraphy. Radiocarbon dating and chronostratigraphical considerations Conventional dating on 5 bulk sediment samples of silty clay was performed at the Angstrom Laboratory, Division of Ion Physics, 14 C-Lab, Uppsala University. The dates have been calibrated to calendar years ( ± 2σ range) with OxCal v3.9 (Bronk Ramsey 2001) using the atmospheric data from Stuiver et al. (1998). The results are shown in Table 3. A sedimentation plot with a linear fit is also presented (Fig. 6). Regarding the chronology the following points need noting:

I. Sediment accumulation rates for the interval 490– 470 cm average c. 90 yrs/cm while for the younger part (470–450 cm) a more variable sedimentation rate might be expected. II. The high NAP values in the interval 455–430 cm place it in a stadial phase (Younger Dryas), while the date Fig. 2 Map of the northwestern Rila Mountains showing the cirque 9620 ± 75 b.p. at level 449 cm, if accepted, indicates of the Seven Rila Lakes with their maximum depths (m), with Lake early Holocene (Preboreal) which is rather unlikely from Trilistnika and Lake Sedmo Rilsko a biostratigraphical point of view. Therefore it seems reasonable to disregard this date. square-rod piston sampler (Wright 1991). The core was 590 cm long and 5 cm in diameter. The water depth at the sampling site was 6.3 m. The lateglacial sediments were identified between 590 and 430 cm. The lithology is Results and discussion presented in Table 2. Lake Trilistnika – pollen stratigraphy and radiocarbon chronology (Fig. 3) Pollen analysis The pollen record (590–430 cm) reveals the vegetation hisSub-sampling for pollen analysis was carried out at 5– tory in the vicinity of the lake during the Late Pleniglacial 10 cm intervals and all 24 samples were processed with after the ice-retreat. At present it is difficult to suggest exHF and acetolysis (Faegri and Iversen 1989). The pollen actly when before 12800 b.p. the sedimentation in the lake sum (PS) used for percentage calculations was based on the started as the attempts to date samples of bulk sediment sum AP (arboreal pollen) + NAP (non-arboreal pollen). below level 500 cm or to find terrestrial plant macrofosIn most instances a PS of 370–540 was achieved. Excluded sils failed. Nevertheless, it appeared possible to link the from the PS are spores of mosses and pteridophytes and the vegetation and climate changes at high altitude in the Rila Mountains for the first time to a local chronology for the time window 13000–10000 b.p. The lateglacial climatic Table 2 Lake Trilistnika: Lithology of the core fluctuations are known at high resolution from Greenland 0–80 cm Grey olive gyttja ice cores such as GISP2 (Stuiver et al. 1995). The local ra80–410 cm Dark brown gyttja diocarbon chronology is compared, where possible, to the 410–430 cm Brown/Olive gyttja ice record and the traditional chronostratigraphical subdi430–500 cm Olive silty clay vision of the Lateglacial for Western Europe (Mangerud 500–590 cm Grey silt et al. 1974).

18 Table 3 Results of radiocarbon measurements from Lake Trilistnika

Lab. code (Ua)

Sample depth (cm)

14

20695 21599 20696 21600 21601

448–450 457–459 468–470 481–483 489–491

9620 ± 75 11140 ± 100 11365 ± 75 12110 ± 95 12815 ± 130

The first zone LT-1 (590–535 cm) is characterised by the deposition of grey silt. Low values of AP c. 15–20% are attributable mainly to Pinus Diploxylon-type and P. peuce and partly to Ephedra (distachya- and fragilis-type). In some pollen spectra Juniperus, Betula, Alnus, Corylus and Carpinis betulus are present. NAP taxa dominate, among them Artemisia with a characteristic peak (40–60%), Chenopodiaceae (20%), Poaceae (up to 17%) and a variety of herbs such as Achillea-type, Cichoriaceae, Galium-type, Rumex, Thalictrum, Apiaceae, Dianthus-type etc. The reconstruction for this zone reveals typical stadial conditions with the dominance of ArtemisiaChenopodiaceae-Poaceae open herb vegetation, composed of cold-resistant and heliophilous species distributed around the lake, isolated pine trees and sparse shrubland of Ephedra with Juniperus at lower altitude. The finding of Ephedra fragilis-type pollen complementary to E. distachya-type indicates that both taxa were important plant constituents on open sandy places. Today E. fragilis ssp. campylopoda occurs in only one place in SW Bulgaria and has not been reported for the Rila Mountains (Velˇcev 1984). In our opinion, the identification of pollen grains of trees other than Pinus (Quercus, Corylus, Carpinus betulus, Alnus), although in minimal quantities, suggests their probable survival in sheltered habitats at low altitudes with sufficient moisture and favorable temperature for growth.

Fig. 3 Lateglacial pollen diagram from Lake Trilistnika

C age (b.p.)

14

C age (cal b.p., 2σ)

11180–10730 13450–12850 13500–13000 15350–13650 15950–14350

Material dated Silty clay Silty clay Silty clay Silty clay Silty clay

The palynological records from other lateglacial sequences in the Rila Mountains (Bozilova et al. 1990; Bozilova and Tonkov 2000) and in the northern Pirin Mountains (Tonkov et al. 2002a) also contain comparable frequencies of deciduous tree pollen. Pollen analysis of recent surface moss samples, collected between 1900 and 2200 m in the subalpine lake area of the Rila and Pirin Mountains, shows an average of 2–4% pollen of deciduous trees (Quercus, Carpinus, Fagus, Corylus, etc.) which is efficiently transported high upslope through the dense coniferous forests (Tonkov et al. 2000, 2002b). Therefore, in the absence of a dense forest cover, the find of deciduous tree pollen in the lateglacial lake sediments is not unusual, and the possibility of reworking in the hard grey silt is most unlikely. However, other authors accept possible re-deposition of deciduous tree pollen in the lateglacial sediments from unknown deposits as in the case of Lake Kremensko-5 (2124 m) in the northern Pirin Mountains (Atanassova and Stefanova 2003). The next zone LT-2 (535–463 cm) features a characteristic increase in AP at level 500 cm coinciding with a change in the sediments from grey silt to olive silty clay. It is reasonable to divide this zone into two subzones (LT-2a and LT-2b). Typical of subzone LT-2a (535–505 cm) is the high proportion of Pinus Diploxylon-type (50–55%), accompanied by a slight increase of Pinus peuce, the appearance of continuous pollen curves for Betula and Alnus and the

19

Fig. 4 Lateglacial pollen diagram from Lake Sedmo Rilsko

Fig. 5 Lateglacial pollen diagram from Lake Sucho Ezero -II

decrease in Artemisia and Chenopodiaceae. Traces of Abies and Picea pollen are also recorded. The implied enlargement of the areas occupied by pines and the presence of Betula and Alnus, indicates a climate improvement that initiated the expansion of tree vegetation at lower altitude. The appearance of local plants such as ferns (Polypodiaceae) and sedges (Cyperaceae) in the pollen record is also notable. The rapid warming trend began 14500 years ago if compared with the palaeotemperature reconstruction for the Lateglacial derived from the GISP2 18 O and CH4

records (Stuiver et al. 1995; Blunier and Brook 2001). Also evidence from the Alps points to a large-scale recession and contraction to nearly half their size of the W¨urm glaciers that was underway by 15000 years b.p. (van Husen 1989). In subzone LT-2b (505–463 cm) the variety of arboreal taxa increases, suggesting that deciduous trees (Quercus, Corylus, Ulmus, Carpinus betulus, C. orientalis) spread slightly from their refugia as climate continued to improve. Following the GISP2 18 O climate record (Stuiver et al. 1995), the radiocarbon date of 12815 ± 130

20

Local and regional comparisons

Fig. 6 Age/depth sedimentation plot for the lateglacial sequence from Lake Trilistnika

b.p. for this subzone can be correlated with the transition Oldest Dryas-Bølling/Allerød while the radiocarbon date 12110 ± 95 b.p. confirms the interstadial age of the vegetation. Moreover, the deposition of olive silty clay probably points to a change in the trophic status of the lake or higher productivity of the surrounding vegetation. Zone LT-3 (463–437 cm) records a steep decline in AP of up to 20%, clearly manifested by the fall in Pinus Diploxylon-type (15–20%). Nearly all other arboreal taxa are absent from the pollen record, except for Betula and Juniperus. The proportion of Artemisia, Chenopodiaceae, Rumex, Galium-type and Dianthus-type again starts to increase. The radiocarbon dates indicate that the reversal from interstadial towards stadial conditions occurred between 11360 and 11140 b.p. conforming well with the termination of the Allerød phase in Western Europe according to the GISP218 O climate record (Stuiver et al. 1995). This stratigraphic interval is associated with the last abrupt and most intense climatic deterioration of the Lateglacial, i.e. the Younger Dryas stadial. The timing of this cold period of global importance lies between 10950 and 10150 b.p. (12900–11600 cal. b.p.) and its duration is estimated as c. 1100–1300 calender years (Wohlfarth 1996; Alley 2000). In the diagram this stadial is palynologically clearly defined by large increases in the herbs. The reduced quantity of pollen of Pinus Diploxylon-type and P. peuce, the decline of almost all the deciduous taxa and the high frequencies of Artemisia (45–55%) and Chenopodiaceae (15%) show this period to be the coldest phase before the final amelioration of the climate began. Typical stadial environmental conditions returned in the Rila Mountains. The rich mountain-steppe herb vegetation re-advanced and dominated at high altitudes while the tree stands moved down along the slopes.

The new lateglacial pollen record allows comparison with other insufficiently radiocarbon dated sequences from lakes and peat-bogs of similar age in the high mountains of southern Bulgaria. A 115 cm thick lateglacial record was studied from Lake Sedmo Rilsko (2095 m) located in the same cirque (Bozilova and Tonkov 2000) (Fig. 2). The lack of radiocarbon dates hinders the chronological subdivision but the biostratigraphy itself can be established from the percentages of the most important pollen taxa (Fig. 4). The basal part of the sequence (zone Ril-1) is poor in tree pollen and indicates stadial conditions of pre-Allerød age. The upper zone Ril-2 (503–433 cm) is considered to be an interstadial sequence, analogous with the Allerød, that can be divided into two subzones. The first, Ril-2a (503–443 cm), features the beginning of a climate improvement with AP gradually increasing from 10 to 30% (level 455 cm), contributed by Pinus (P. Diploxylon-type, P. peuce), Juniperus and a steady presence of Betula, Salix and Corylus. The second subzone Ril-2a (443–433 cm) is characterised by a peak in the AP curve (50%) and a decrease in the abundance of mountain-steppe species. The sparse tree vegetation of Pinus and Betula has extended in the approaches to the lake. In zone Ril-3 (433–413 cm) the AP curve sharply declines to 10–15% (level 425 cm) and high proportions of NAP dominated by Artemisia, Chenopodiaceae, Poaceae and other cold-resistant herbs are recorded. The sediments deposited are grey clay. The radiocarbon date 10170 ± 60 b.p. (level 410 cm) complements the pollen evidence that this zone can be associated with the stadial Younger Dryas. The transition Lateglacial/Holocene is manifested by a steep increase in AP and the deposition of brown gyttja. The LOI results from this core show that the organic matter steadily increases from ca. 25% to over 50% in the early Holocene, suggesting a change from predominantly allochthonous clastic sediment to autochthonous organic sediment in connection with increasing vegetation cover in the catchment area. The depth-age model sedimentation accumulation rates as estimated were too low for the Lateglacial and early Holocene (Lotter and Hofman 2003). The correlation of the two lateglacial pollen sections, which differ in altitude by 120 m, reveals that the lateglacial interstadial/stadial cycle recorded in the sediments of Lake Trilistnika is more clearly distinguished. The termination of the Lateglacial in the sediments of Lake Sedmo Rilsko is confirmed by the radiocarbon date of 10170 ± 60 b.p. and also by the palynostratigraphy (a steep increase in the AP curve). In the light of this conformity and according to the age/depth plot (Fig. 6), the radiocarbon date of 9620 ± 60 b.p. (level 449 cm) for the core from Lake Trilistnika appears to be too young and as already mentioned should not be considered. For a while the palaeoecological record from two cores from the present peat-bog Sucho Ezero (1900 m), a former lake in the southern Rila Mountains, served as the only source of information on the environmental changes throughout the Lateglacial and Holocene (Bozilova and

21

Smit 1979; Bozilova et al. 1990) (Fig. 1). Pollen analysis and plant macrofossil determination indicated a 3 m thick section of lateglacial sediments, poorly radiocarbondated (Bozilova et al. 1990) (Fig. 5). The palynostratigraphy of the lowermost 200 cm (zone P-1) suggests stadial environmental conditions after the Last Glacial Maximum. However, when sedimentation in the lake began remains obscure. The lateglacial interstadial (zone P-2) is clearly expressed by high values of AP (up to 45%), attributed mainly to Pinus Diploxylon-type (30%), Betula (10%), Pinus peuce (4%), Quercus, Corylus, Abies and Picea which indicate the most pronounced climatic improvement during the Lateglacial. The retreat of the mountain-steppe vegetation is documented by low proportions of Artemisia (10– 15%), Chenopodiaceae (below 10%) and many other herbs. The climate deterioration during the Younger Dryas (zone P-3), as discussed above, resulted in a wide spread of coldresistant herb communities at this altitude and in a decline of the sparse tree vegetation. The lateglacial plant macrofossil record is extremely poor, represented by a number of seeds and fruits from herbs and a fragment of Pinus peuce needle (level 850 cm). The radiocarbon date 10570 ± 220 b.p. is still the only one available for the Younger Dryas from the Rila Mountains. The termination of the Lateglacial above level 825 cm is marked by a steep rise in the AP curve, reaching 70%. In this profile the lateglacial palynostratigraphy appears quite similar to the results from Lake Trilistnika in that the lateglacial interstadial/stadial cycle is easily recognised but poorly radiocarbon dated. Earlier, sediments of lateglacial age were also investigated from three glacial lakes located in the subalpine belt of the northern Pirin Mountains (Fig. 1). The lowermost 50 cm sediments of grey silty clay/yellow silt from Lake Ribno Banderishko (2190 m) were deposited during the Lateglacial, as proved by the palynostratigraphy and the consistent available AMS radiocarbon chronology for the Holocene part of the sequence (Tonkov et al. 2002a). The lack of radiocarbon dates for the lateglacial section of the profile hampers the creation of a chronological framework; nevertheless the pollen stratigraphy provides a basis for a preliminary interstadial/stadial subdivision that can be correlated with the record from Lake Trilistnika. The Lateglacial in Lake Dalgoto (2310 m) spans only 20 cm of grey silt sediment. This is not radiocarbon dated but is assigned to the stadial Younger Dryas on the basis of the palynostratigraphy, characterised by the existence of open xerophytic herb vegetation (Stefanova and Ammann 2003). The core Lake Kremensko-5 (2124 m) yielded sediments from more than 13500 b.p.and a pollen stratigraphy that can be correlated with the interstadial/stadial cycle of the Lateglacial. The radiocarbon date 13350 ± 90 b.p. indicates a stadial phase that can be correlated with the Oldest Dryas (Atanassova and Stefanova 2003) and is close to the result of 12815 ± 130 b.p. from Lake Trilistnika. The climate amelioration during the interstadial is confirmed by the higher importance of tree taxa leading to enlargement of the coniferous woods around the lake and supported by the radiocarbon date 12360 ± 55 b.p. (Atanassova and

Stefanova 2003). A similar age of 12110 ± 95 b.p. for this stage was obtained from Lake Trilistnika. The upper part of the lateglacial record of Lake Kremensko-5 remains without an absolute age determination, but the Younger Dryas stadial, c. 25 cm in thickness, is delimited on the basis of the dominance of Artemisia, Chenopodiaceae and other various characteristic herb taxa and the occurrence of the Juniperus-Ephedra assemblages (Atanassova and Stefanova 2003). The palynological evidence for the Lateglacial from the western Rhodopes Mountains in southern Bulgaria, that in contrast to the Rila and northern Pirin Mountains were not glaciated during the Quaternary, originates from sites located in the coniferous forest belt (Fig. 1). Two pollen profiles were analysed from the Kupena Mire (1300 m) where the thickness of the lateglacial sediments, composed of clay gyttja, varies between 30 and 80 cm (Bozilova et al. 1989; Huttunen et al. 1992). The vegetation development dates back to an interstadial phase (11875 ± 55 b.p.), when the initial herb mountain-steppe vegetation was soon followed by steppe-forest composed of trees and shrubs such as Pinus, Betula, Acer, Corylus, Carpinus, Fagus, Fraxinus and many others, which had survived the harsh environmental conditions in local refugia. A re-advance of the herb vegetation, dominated by Poaceae, Artemisia and Chenopodiaceae, occurred during the Younger Dryas stadial after 10700 ± 64 b.p. and persisted till ca. 9320 ± 185 b.p. (Bozilova et al. 1989; Huttunen et al. 1992). The lateglacial sediments from the peat bog Shiroka Polyana-1 (1500 m), 100 cm in thickness, are dominated by NAP (Filipovitch and Lazarova 2003). The palynostratigraphy itself does not indicate an interstadial/stadial cycle. A variety of trees and shrubs, such as Pinus, Betula, Juniperus, Abies, Quercus, Fagus, Corylus etc., although in low frequences, is recorded continuously throughout the Lateglacial. The single radiocarbon date of 13080 ± 55 b.p. at the bottom part of the sequence marks the time when sedimentation began, but a radiocarbon chronology for this site is still lacking.

Conclusions The record from Lake Trilistnika in the Rila Mountains is a step towards the elucidation of the lateglacial vegetation and climate history at high altitudes and the establishment of a chronological control, based on pollen stratigraphy and radiocarbon dating, on environmental changes in the high Bulgarian mountains. It is possible to draw the following main conclusions concerning the lateglacial vegetation changes: 1. The steppe-mountain vegetation in the Rila Mountains dominated by Artemisia–Chenopodiaceae–Poaceae started to withdraw before 12800 b.p. as a response to the general warming trend after the Last Glacial Maximum. 2. During the lateglacial interstadial the steppe-mountain vegetation was partly replaced by stands of pines and

22

3. 4.

5.

6.

temperate deciduous trees began to spread at lower elevation. The lateglacial climatic reversal during the Younger Dryas stadial triggered the re-advance of the herb vegetation while the tree stands moved down the slopes. The comparison with other lateglacial Bulgarian mountainous sites reveals common tendencies in vegetation development bound to an interstadial/stadial cycle. The differences established depend on the local characteristics of the sites studied. Evidence is gathering in support of the survival through the glacial of various temperate trees in refugia where moisture and temperature were favorable for their growth. Further multi-proxy investigations are needed to define precisely the point in time when the mountain glaciers retreated and sedimentation in the glacial lakes began.

Acknowledgements The authors express their gratitude to H. Wright (Minnesota, USA) and B. Ammann, University of Bern (Switzerland) who brought the equipment and guided the coring of the lake. Invaluable help in the field was provided by A. Velˇcev, R. Penin and a group of students from Sofia University, D. Stojanov, D. Uzunov and E. Kozhuharova. The coring expedition to the Rila Mountains was funded by the Swiss National Science Foundation through Project N 7BUPJ041258. Both reviewers, H.-J. Beug and an anonymous colleague, and the editor-in-chief F. Bittmann, provided useful critical comments and suggestions to improve the manuscript. The copy editor J. Daniell refined the text. A. Tosheva helped with the drawing of Figs. 1 and 2. The present paper is a contribution to Project B-905/99 funded by the National Research Council in Sofia, Bulgaria.

References Alley RB (2000) The Younger Dryas cold interval as viewed from central Greenland. Quatern Sci Rev 19(1–5):213–226 Atanassova J, Stefanova I (2003) Late-glacial vegetational history of Lake Kremensko-5 in the northern Pirin Mountains, southwestern Bulgaria. Veget Hist Archaeobot 12:1–6 Beug H-J (2004) Leitfaden der Pollenbestimmung fur Mitteleuropa und angrenzende Gebiete. Verlag Dr. Friedrich Pfeil, M¨unchen Blunier T, Brook JE (2001) Timing of Millennial-Scale Climate Change in Antarctica and Greenland During the Last Glacial Period. Science 291:109–112 Bondev I (1991) Vegetation of Bulgaria. Map 1:600 000 with explanatory text (in Bulgarian with English summary). St. Kliment Ohridski University Press, Sofia Bozilova E, Smit A (1979) Palynology of Lake Suho Ezero in South Rila Mountain (Bulgaria). Fitologija 11:54–67 Bozilova E, Tonkov S (2000) Pollen from Lake Sedmo Rilsko reveals southeast European postglacial vegetation in the highest mountain area of the Balkans. Res New Phytol 148:315–325 Bozilova E, Panovska H, Tonkov S (1989) Pollenanalytical investigation in the Kupena National Reserve, West Rhodopes. Geographica Rhodopica 1:186–190 Bozilova E, Tonkov S, Pavlova D (1990) Pollen and plant macrofossil analyses of the Lake Suho Ezero in the South Rila mountains. Ann Univ Sofia Fac Biol 80:48–57 Bronk Ramsey C (2001) OxCal Program v3.9. University of Oxford, Radiocarbon Accelerator Unit. Faegri K, Iversen J (1989) Textbook of pollen analysis. John Wiley & Sons, Chichester Filipovitch L, Lazarova M (2003) Palynological research in the Shiroka Polyana locality (Western Rhodopes). Phytologia Balcanica 265–274

Grimm E (1987) CONISS: A Fortran 77 Program for stratigraphically constrained cluster analysis by the method of incremental squares. Comput Sci 13:13–35 Grimm E (2004) TGView (version 2.0.2). Springfield, Illinois State Museum Husen D van (1989) The last interglacial-glacial cycle in the eastern Alps. Quat Int 3/4:115–121 Huttunen A, Huttunen R-L, Vasari Y, Panovska H, Bozilova E (1992) Late-glacial and Holocene history of flora and vegetation in the Western Rhodopes Mountains, Bulgaria. Acta Bot Fennica 144:63–80 Ivanov K (1964) The lakes in Bulgaria (In Bulgarian). Nauka i Izkustvo, Sofia Jordanov D, Velˇcev V, Kozuharov S (eds) (1963–1995) Flora of Republic of Bulgaria I-X (in Bulgarian). Publ House of Bulg Acad Sci, Sofia Lang G (1994) Quartare Vegetationsgeschichte Europas. Gustav Fischer Verlag, Jena Lotter A, Hoffman G (2003) The development of the late-glacial and Holocene diatom flora in Lake Sedmo Rilsko (Rila mountains, Bulgaria) In: Tonkov S (ed) Aspects of Palynology and Palaeoecology. Pensoft Publ, Sofia-Moscow, pp. 171–184 Mangerud J, Andersen BE, Berglund BB, Donner JJ (1974) Quaternary stratigraphy of Norden, a proposal for terminology and classification. Boreas: 3:109–128 Moore P, Webb J, Collinson M (1991) Pollen analysis. 2nd edn. Blackwell Scientific Publications, Oxford Stefanova I, Ammann B (2003) Lateglacial and Holocene vegetation belts in the Pirin Mountains (southwestern Bulgaria). The Holocene 13:97–107 Stuiver M, Grootes PM, Braziunas TF (1995) The GISP2 18 O climate record of the past 16500 years and the role of the sun, ocean and volcanoes. Quat Res 44:341–354 Stuiver M, Reimer PJ, Bard E, Beck JW, Burr GS, Hughen KA, Kromer B, McCormac G, van der Plicht J, Spurk M (1998) INTCAL98 Radiocarbon Age Calibration, 24 000-0 cal b.p. Radiocarbon 40:1041–1083 Taberlet P, Cheddadi R (2002) Quaternary refugia and persistence of biodiversity. Science 297:2009–2010 Tishkov H (1976) Le climat des regions montagneuses de la Bulgarie structure et genese (in Bulgarian with French summary). Publ House of Bulg Acad Sci, Sofia Tonkov S, Bozilova E, Pavlova D, Kozuharova E (2000) Surface pollen samples from the valley of the Rilska Reka river, Central Rila Mountains (South-Western Bulgaria). Ann Univ Sofia Fac Biol 91(2):63–74 Tonkov S, Panovska H, Possnert G, Bozilova E (2002a) The Holocene vegetation history of Northern Pirin Mountain, southwestern Bulgaria: pollen analysis and radiocarbon dating of a core from Lake Ribno Banderishko. The Holocene 12:201–210 Tonkov S, Bozilova E, Dimitrov D, Atanassova J, Pavlova D (2002b) Surface pollen samples from the North-Western Pirin Mountain. Ann Univ Sofia Fac Biol 90:11–21 Velˇcev A (1995) The Pleistocene glaciations in the Bulgarian mountains (in Bulgarian with English summary). Ann Univ Sofia Fac Geol Geogr 87:53–65 Velˇcev V (ed) (1984) Red Data Book of the People’s Republic of Bulgaria. Vol. 1. Plants (in Bulgarian with English summary). Publ House of Bulg Acad Sci, Sofia Velˇcev V, Tonkov S (1986) Vegetation and flora of Southwestern Bulgaria. In: Botev B. (ed), Fauna of Southwestern Bulgaria (in Bulgarian with English summary). Publ House of Bulg Acad Sci, Sofia, pp 20–43 Willis KJ (1994) The vegetational history of the Balkans. Quatern Sci Rev 13:769–788 Willis KJ, van Andel TH (2004) Trees or not tress? The environments of central and eastern Europe during the Last Glaciation. Quatern Sci Rev 23:2369–2387 Wohlfarth B (1996) The Chronology of the Last Termination: A review of Radiocarbon-Dated, High-Resolution Terrestrial Stratigraphies. Quat Sci Rev 15:267–284 Wright HE (1991) Coring tips. Journal of Paleolimnology 6:7–49