Quaternary Research 53, 203–213 (2000)
doi:10.1006/qres.1999.2119, available online at http://www.idealibrary.com on
The Full-Glacial Forests of Central and Southeastern Europe
Katherine J. Willis
School of Geography, University of Oxford, Mansfield Road, Oxford, OX1 3TB, United Kingdom
and
Edina Rudner and Pal Sümegi
Department of Mineralogy and Geology, Kossuth Lajos University, P.O. Box 4, Debrecen, H-4010, Hungary
Received December 23, 1998
The presence of trees in central and southern Europe during the
last full-glaciation has long been a matter of debate. A low but
persistent presence of fossil tree pollen in central and southern
European full-glacial paleoecological sequences has been inter-
preted either as representing long-distance pollen transport from
southerly refuges or as representing in situ refugial populations.
Here we present macroscopic charcoal results from 31 sequences
located throughout Hungary that provide unequivocal evidence
for the presence of at least seven different tree types between
approximately 32,500 and 16,500 14C yr B.P. This evidence is
presented in conjunction with molluscan and pollen analyses to
indicate that during the last full-glaciation, trees grew as far north
as Hungary, probably in microenvironmentally favorable sites.
These areas provided an important cold-stage refugium for the
European flora and fauna. © 2000 University of Washington.
Key Words: full-glacial forests; central Europe; charcoal
analysis; molluscan analysis; pollen analysis; paleoecology.
INTRODUCTION
The question of whether forest communities existed in cen-
tral and southeastern Europe during the last full-glaciation has
long been a matter of debate. Tree pollen has been found in
many southern European sedimentary sequences that extend
back into the full-glaciation (Follieri et al., 1988). However, it
is uncertain whether it represents long-distance pollen transport
from more southerly refuges (e.g., the Near East, North Africa)
or production by in situ populations. The argument for long-
distance pollen transport is strengthened by examination of
pollen accumulation rates. These data suggest that during the
full-glaciation, the concentration of tree pollen declined dra-
matically, perhaps as much as three orders of magnitude,
(Magri, 1989). This trend is thought to be indicative of a
reduction in plant biomass and the presence of few, if any, trees
on the landscape.
One problem in accepting the long-distance pollen transport
hypothesis is the location of these so-called southerly refuges.
203
Little evidence exists from either the Near East or North Africa
for full-glacial tree refuges. Also, the presence in the Southern
European sequences of pollen types with extremely poor dis-
persal characteristics and/or low productivity, such as Ulmus,
Tilia, Acer, Larix, and Abies, is difficult to explain since these
taxa would likely not be transported great distances (Willis,
1994a). This debate highlights one of the main shortfalls in the
use of fossil pollen to reconstruct past vegetation. Pollen can
represent both local and long-distance flora, and other analyt-
ical methods must be sought, especially when the pollen ac-
cumulation rate is low, such as during the full-glaciation.
To date, alternative approaches for reconstructing past veg-
etation profiles have included analysis of plant macrofossil
assemblages from lake sediments (Birks, 1994) and packrat
middens (Betancourt et al., 1990). Both techniques have
proved extremely successful in providing detailed local vege-
tation profiles, although in such cases conditions for preserva-
tion are exceptional and many regions do not contain such
deposits. Another method has been the analysis of fossilized
wood preserved in the record as macroscopic charcoal (Vernet
and Thiebault, 1987; Marziani et al., 1991; Sander and Gee,
1990; Maspero, 1996). Although the microscopic and macro-
scopic charcoal record in lake sediments has been used exten-
sively by palaeoecologists to assess the histories, frequencies,
and intensities of fires (e.g., Whitlock and Millspaugh, 1996),
it has rarely been applied for identifying and establishing tree
species.
Macroscopic charcoal particles (⬎0.2 cm diameter) are as-
sumed to originate from the burnt lignaceous vegetation grow-
ing within 100 m of the site (Clark, 1988) and therefore can
provide a detailed record of local vegetation (similar to mac-
rofossil analysis). Charcoal is preserved in a number of differ-
ent sedimentary environments including fossil soils, archaeo-
logical sites, lake sediments, and loess sequences thereby
providing information beyond the lakeside environment.
This paper presents results from a macroscopic charcoal
study of 31 sites in Hungary that contain sedimentary se-
0033-5894/00 $35.00
Copyright © 2000 by the University of Washington.
All rights of reproduction in any form reserved.
204 WILLIS, RUDNER, AND SÜMEGI
FIG. 1. Location of macroscopic charcoal, molluscan, and palynological sites in Hungary (Willis et al., 1995, 1997; Willis, 1997). See Table 1 for site
information.
quences extending back into the full-glaciation (between ap-
proximately 32,500 and 16,500 14C yr B.P.). These records
provide unequivocal evidence for the presence of trees in
Hungary during this interval, and when considered in the light
of other paleoenvironmental reconstructions determined by
malacological and pollen analyses (Willis et al., 1995, 1997,
1998; Willis, 1997), provide a new interpretation of the full-
glacial and late-glacial landscapes of central and southeastern
Europe.
MATERIALS AND METHODS
Study Region
Hungary provides an important area of low relief within the
main mountain ranges of central and southeastern Europe. It is
effectively enclosed to the northeast by the Carpathians, to the
west by the Alps, and to the southwest by the Dinaric Alps.
During the last full-glaciation, Hungary remained unglaciated
(Denton et al., 1971). Conditions were cold and dry (Willis et
al., 1995, 1997; Hertelendi et al., 1992). Climate models and
paleoclimatic reconstructions suggest January temperatures
were as low as ⫺25°C (Kutzbach et al., 1993; Kordos, 1987).
Extensive permafrost extended across central Europe (Bell and
Walker, 1992), and much of Hungary was in a zone of discon-
tinuous permafrost (Maruszczak, 1987; Pécsi, 1961). Sporadic
deposition of loess also occurred throughout the full-glaciation
in this region (Pécsi, 1993; Pécsi and Schweitzer, 1991), ef-
fectively capping and preserving paleosols, charcoal layers,
and archaeological sites alike.
The 31 sequences described in this study (Fig. 1) include
data from new sites (21 samples) and from re-examined sam-
ples of previously published sites (Geyh et al., 1969; Marosi et
al., 1974; Vogel et al., 1964; Pécsi, 1977; Dobosi, 1967;
Vértes, 1964; Gábori-Csánk, 1960) (Table 1). The majority of
the macroscopic charcoal samples were extracted from pa-
leosols situated within loessic sequences. Five samples were
collected from buried layers found during archaeological ex-
cavations, and one sample was obtained from an organic lake
sedimentary sequence. A molluscan record was also obtained
for analysis from 16 of the loess sequences (Table 1).
Analyses
To sample the buried paleosols, a flat vertical section was
cut into the loess walls (between 100 and 200 m horizontally,
and 5 and 20 m vertically). Subsamples of ca 5 cm 3 were cut
from the cleaned profile for charcoal analyses and a larger
sample size of ca. 1 dm 3 (5 cm height ⫻ 20 cm width ⫻ 10 cm
depth) was collected for molluscan analysis. A similar strategy
was used in the collection of the samples from the archaeolog-
ical sites. The lake sediment sample was extracted from a core
that had been collected using a modified Livingstone piston
corer (Wright, 1967).
In the laboratory, the sediments were dried at 100°C for 24 h
and then sieved through a metal screen with distilled water
 EUROPEAN FULL-GLACIAL FORESTS 205

TABLE 1
Site Type, Radiocarbon Ages and Arboreal Vegetation in Macroscopic Charcoal Sites

Radiocarbon age Molluscan

Site no. Site type Tree types represented in macroscopic charcoal record ( 14C yr B.P.) record References

1 Ls Pinus sp., Picea sp., Juniperus sp. 32,500 ⫾ 2,170 x Geyh et al., 1969
2 Ls Pinus sylvestris 30,174 ⫾ 1,101 ⻫ new data
3 Ls Pinus sylvestris, Betula sp. 29,828 ⫾ 554 ⻫ new data
4 Ls Pinus sylvestris-P. cembra, 29,800 ⫾ 600 x Geyh et al., 1969
4 Ls Picea sp. 27,200 ⫾ 1,400 x new data
5 As Larix-Picea 28,700 ⫾ 3,000 x Krolopp, 1977
5 Ls Picea sp. 26,318 ⫾ 365 x new data
6 Ls Picea sp. 28,225 ⫾ 360 ⻫ new data
7 As Picea sp. 27,700 ⫾ 300 x new data
8 Ls Picea sp. 27,323 ⫾ 644 ⻫ new data
9 Ls Picea sp. 27,491 ⫾ 362 ⻫ new data
10 Ls Picea sp. 27,251 ⫾ 288 x new data
11 Ls Carpinus betulus 26,962 ⫾ 657 ⻫ new data
12 Ls Picea sp. 26,851 ⫾ 398 ⻫ new data
13 Ls Pinus sylvestris 26,736 ⫾ 629 ⻫ new data
14 Ls Picea sp. 26,618 ⫾ 532 ⻫ new data
15 Ls Picea-Larix, Pinus sylvestris-P. cembra 26,350 ⫾ 3,111 x Geyh et al., 1969
16 Ls Pinus sylvestris 25,200 ⫾ 300 ⻫ Krolopp et al., 1996
17 Ls Picea-Larix 21,165 ⫾ 865 x Pécsi, 1977
17 Ls Picea sp. 24,030 ⫾ 317 ⻫ new data
18 Ls Pinus sylvestris 23,749 ⫾ 494 ⻫ new data
19 Ls Picea sp. 23,571 ⫾ 486 ⻫ new data
20 Ls Picea-Larix 20,350 ⫾ 470 x Geyh et al., 1969
20 Ls Picea sp. 23,519 ⫾ 494 ⻫ new data
21 La Salix sp. 23,290 ⫾ 285 x new data
22 Ls Pinus sylvestris 22,110 ⫾ 300 ⻫ new data
23 Ls Pinus sylvestris 21,937 ⫾ 252 ⻫ new data
24 Ls Larix-Picea 21,740 ⫾ 320 x Krolopp, 1977
25 Ls Pinus sylvestris-P. cembra 21,725 ⫾ 560 x Marosi and Szilárd, 1974
26 Ls Pinus sylvestris-P. cembra 20,520 ⫾ 290 x Pécsi, 1975
27 As Pinus sylvestris-P. cembra, Picea-Larix 18,900 ⫾ 100 x Vogel-Waterbolk, 1964
28 As Pinus sylvestris-P. cembra 18,700 ⫾ 1,900 x Vértes, 1964
28 As Picea-Larix 17,050 ⫾ 350 x Vértes, 1964
29 As Pinus sylvestris-P. cembra, Picea-Larix 18,080 ⫾ 405 x Dobosi, 1967
30 As Pinus sylvestris-P. cembra, Picea-Larix 17,760 ⫾ 150 x Gábor-Csánk, 1960
31 Ls Pinus sylvestris-P. cembra, Picea-Larix, Betula sp. 16,750 ⫾ 400 x Geyh et al., 1969

Note. Ls, loess deposit; As, archaeological site; La, lake sediment. See Fig. 1 for site locations.

(mesh size of 0.8 mm). All charcoal particles ⬎0.2 cm 3 were
collected for analysis and prepared for identification using
scanning electron microscopy (Rudner, 1994). For each mac-
roscopic charcoal sample, transsection, longitudinal radial, and
tangential fracture faces were examined (Schweingruber,
1989). Most samples were identified to genus level although in
several cases it was possible to identify to species (e.g., Pinus
sylvestris). In some samples, it was impossible to distinguish
between Pinus sylvestris and Pinus cembra and in others
between Picea sp. and Latrix (Table 1).
In order to ensure a statistically significant molluscan sample
size (Lozek, 1987), a minimum of 100 individual/dm 3 were
counted. All molluscs were identified and categorized accord-
ing to their modern ecological preferences for air temperature,
humidity, and vegetation cover (Lozek, 1964; Meijer, 1985;
Sparks, 1964; Krolopp and Sümegi, 1995; Krolopp et al.,
1996).
Radiocarbon dating of the macroscopic charcoal was per-
formed by the Nuclear Research Centre of the Hungarian
Academy of Sciences, Debrecen. In addition, some dates
were taken from previous publications (Table 1). All dates
are presented as uncalibrated years before present ( 14 C yr
B.P.). Calibrated ages are used in comparisons with the
climate record provided by stable isotopes from the Nordic
Sea and the GISP ice-core (Voelker et al., 1998). Ages were
calibrated using the program CALIB 3.0 for dates up to
21,950 cal yr B.P. (Stuiver and Reimer, 1993) and the
conversion proposed by Laj et al. (1996) and van Andel
(1998) for dates between 35,500 and 21,950 cal yr B.P.
(Table 2).
206 WILLIS, RUDNER, AND SÜMEGI

TABLE 2 are no obvious gaps in samples during the stades, including the
Radiocarbon Ages proposed late-glacial maximum (at approximately 21,000 cal
yr B.P.) or the Heinrich events H2 (25,500 cal yr B.P.) and H3
Radiocarbon age Calibrated radiocarbon age
(30,000 cal yr B.P.) (Bond et al., 1993).
( 14C yr B.P.) Lab. code (cal yr B.P.)
Molluscan remains total 5137 individuals, comprising 33
16,750 ⫾ 400 Hv-1615 19,250 different species. The maximum number of individuals in one
17,050 ⫾ 350 GrN-4038 19,620 sample was 987 with a minimum of 102. All the molluscan
17,760 ⫾ 150 GrN-1957 20,630 remains identified were from land species, which are reliable
18,080 ⫾ 494 Hv-1619 21,070
indicators of local habitat conditions (Table 3) and usually
18,700 ⫾ 120 A-518 21,860
18,900 ⫾ 100 GrN-1783 21,900 represent populations contained within an area immediately
20,350 ⫾ 470 Hv-1775 23,300 around the site of deposition (⬍20 – 40 m) (Sparks, 1964;
20,520 ⫾ 290 Hv-2591 23,750 Lozek, 1987). Classification of the molluscan assemblages
21,165 ⫾ 865 Hv-5426 24,300 according to their modern climatic distribution (Table 4) indi-
21,725 ⫾ 560 Hv-2590 24,700
cates that in 15 out of the 16 sites, the highest percentage was
21,740 ⫾ 320 Hv-4189 24,750
21,937 ⫾ 252 deb-3104 25,000 from types usually found in mesophilous and/or thermophilous
22,110 ⫾ 300 deb-1562 25,250 conditions. Thus only one site (site 19) had a higher percentage
23,290 ⫾ 285 deb-4572 26,300 of cryophilous and cold-resistant types. When these data were
23,519 ⫾ 494 deb-4350 26,500 used to construct a paleotemperature curve (Sümegi, 1989), the
23,571 ⫾ 486 deb-1562 26,500
results predicted a mean July paleotemperature of between 16°
23,749 ⫾ 494 deb-3064 26,750
24,030 ⫾ 317 deb-3353 27,000 and 18°C. Classification according to present-day humidity
25,200 ⫾ 300 deb-2053 28,400 levels (Table 5) indicates a mixed signature. Types found today
26,318 ⫾ 365 deb-3051 29,500 in humid and semihumid conditions were present at all but one
26,350 ⫾ 3111 Hv-1777 29,550 site (site 2), but types that are aridity tolerant also occurred in
26,618 ⫾ 532 deb-3042 29,750
these sites. Finally, classification into groups according to
26,736 ⫾ 629 deb-4346 29,800
26,851 ⫾ 398 deb-3049 29,850 modern vegetational preferences (Table 6) indicates that those
26,962 ⫾ 657 deb-5052 30,100 types normally found in steppe were predominant at 11 sites.
27,200 ⫾ 1400 I-3430 30,300 However, woodland-loving and intermediate species were also
27,251 ⫾ 288 deb-4345 30,350 present at all sites and intermediate types (found associated
27,323 ⫾ 644 deb-2657 30,400
with both woodland and open vegetation sites) were predom-
27,491 ⫾ 362 deb-3034 30,500
27,700 ⫾ 300 deb-1901 30,750 inant at 5 sites.
28,225 ⫾ 360 deb-3035 31,500
28,700 ⫾ 3000 GXO-195 31,750 DISCUSSION
29,800 ⫾ 600 Mo-422 32,800
29,828 ⫾ 554 deb-3058 32,820
Previous palynological analyses suggested that trees were
30,174 ⫾ 1101 deb-4347 33,000
32,500 ⫾ 2170 Hv-1776 35,500 probably present in Hungary during the late-glaciation (ca.
16,000 –10,000 14C yr B.P.) (Willis et al., 1995). Evidence
Note. Dates were calibrated using CALIB for ages up to 21,950 cal yr B.P. from three radiocarbon-dated palynological sequences (Fig. 4),
(Stuiver and Reimer, 1993) and the conversion proposed by Laj et al. (1996) for example, indicated that over 50% of the total pollen during
and van Andel (1998) for ages between 35,500 and 21,950 cal yr B.P. the late-glaciation was from arboreal types. The highest input
was from Larix, Picea, and Pinus trees, but broad-leaved taxa,
such as Betula, Salix, Juniperus, Quercus, Corylus, Ulmus, and
RESULTS Carpinus orientalis, also occurred. In the interpretation of
these diagrams, it was proposed that the late-glacial vegetation
Thin-section analysis of the macroscopic charcoal data from of Hungary was similar to that of the southern margins of the
the 31 sites (Table 1) indicates that all material examined was present-day boreal forest (Pastor and Mladenoff, 1992), where
from arboreal vegetation. Types represented included Pinus there was an open coniferous forest containing small pockets of
sylvestris, Pinus cembra, Betula sp., Picea sp., Juniperus sp., deciduous trees (Willis et al., 1995, 1997; Willis, 1997). In
Larix sp., Carpinus betulus, and Salix sp. (Fig. 2). Samples addition, microscopic charcoal analyses from these sequences
ranged in date between 32,500 ⫾ 2170 and 16,750 ⫾ 400 14C indicated the vegetation was burned (Fig. 4). It was therefore
yr B.P. (approximately 35,500 and 19,200 cal yr B.P.). No suggested that similar to the modern boreal forest, wildfires
obvious clusters of samples (Fig. 3) occur around the were an important part of the ecological dynamics of the
“Dansgaard-Oeschger” interstades (Johnsen et al., 1992), late-glacial Hungarian landscape (Zackrisson et al., 1996).
which are calibrated to approximately 29,000, 27,500 and Evidence from the macroscopic charcoal record presented
23,500 cal yr B.P. (Voelker et al., 1998). Furthermore, there here confirms the presence of at least eight different tree types
 EUROPEAN FULL-GLACIAL FORESTS
FIG. 2. Tree types represented in the macroscopic charcoal record and their distributions within Hungary. See Table 1 for site information.
on the Hungarian landscape during the last full-glaciation
(between approximately 32,500 and 16,500 14C yr B.P., i.e.,
35,500 and 19,250 cal yr B.P.). Types present included Pinus
sylvestris, Pinus cembra, Picea sp., Betula sp., Juniperus sp.,
Salix sp., Larix, and Carpinus betulus (at one site). At many of
FIG. 3. Chronological position of macroscopic charcoal samples (crosses)
indicated against the ␦ 18O record of annual-layer-dated GISP2 ice core (re-
drawn from Voelker et al., 1998). The approximate positions of Heinrich
events (H1, H2, H3) (Bond et al., 1993), the Late-Glacial Maximum (LGM),
and the Dansgaard-Oeschger interstade (nos. 1– 8) (Johnsen et al., 1992) also
are indicated.
207
the sites, the in situ origin of the trees is further supported by
the presence of burnt sediment layers forming a silhouette of
the trunk and root positions. In some cases, it would appear that
the trees fell over either before or during burning, resulting in
a long thin charcoal layer that represents the fallen position of
the tree trunk. At two of the sites, burnt trunks, 50 –70 cm in
diameter, were found in a standing position and at right angles
to the surface of the paleosol. Dating of the macroscopic
samples indicates that trees were present during the coldest
intervals, such as the late-glacial maximum (i.e., 21,000 cal yr
B.P.) and Heinrich events (H2 and H3). No obvious clusters of
samples around the timing of the proposed late-glacial inter-
stades are present (Johnsen et al., 1992).
Paleoclimatic simulations (Kutzbach et al., 1993), although
only extending back to approximately 21,500 cal yr B.P., have
suggested full-glacial conditions in central and eastern Europe
that were extremely cold and arid, with predicted January
temperatures as low as ⫺20°C. Geomorphological evidence
indicates the evidence of extensive permafrost, but central and
eastern Europe remained unglaciated throughout the full-
glaciation (Denton and Hughes, 1971; Bell and Walker, 1992;
Székely, 1987; Hertelendi et al., 1992; Maruszczak, 1987). It is
these predicted cold but relatively ice-free conditions that led
to the suggestion that an open tundra or steppe landscape
prevailed in southeastern and central Europe at this time (Pécsi,
1993; Fink and Kukla, 1977; Stoilov, 1984).
Evidence from this study does not necessarily contradict
these climatic reconstructions but rather questions the classic
208 WILLIS, RUDNER, AND SÜMEGI

TABLE 3
Paleoecological Range of Molluscan Species

Species Paleoclimate group Paleohumidity group Paleovegetation group Biogeographical group

Vertigo modesta C H O Boreo-Alpine

Vertigo parcedentata C H O Boreo-Alpine
Vertigo alpestris Cr H W Boreo-Alpine
Vertigo angustior M H W/O European
Vertigo pygmaea M M O Holarctic
Succinea oblonga Cr H O Eurosiberian
Succinea putris Cr H O Eurosiberian
Columella edentula Cr H O Palearctic
Columella columella C H O Boreo-Alpine
Discus ruderatus Cr H W Boreo-Alpine
Semilimax kotulai C H W Alpi-Carpathian
Semilimax semilimax Cr H W W and Central European
Ariante arbustorum Cr H W W and Central European
Vestia turgida Cr H W Carpathian
Orcula dolium M H W Central European
Cochlicopa lubrica M M W/O Holarctic
Clausilia dubia M Sh W Central European
Vitrea cristallina M Sh W/O European
Nesovitra hammonis M Sh W/O Palearctic
Limacidae M Sh W/O Holarctic
Euconulus fulvus M Sh W/O Holarctic
Punctum pygmaeum M Sh W/O Holarctic
Vallonia pulchella M Sh W/O Holarctic
Vallonia costata M Sh W/O Holarctic
Vallonia enniensis T H O Central and SE European
Vallonia tenuilabris C D O N and Central Asian
Chondrula tridens T D O Central and SE European
Granaria frumentum T D O Central and SE European
Pupilla triplicata T D O S-SE European
Pupilla sterri C D O Eurasian
Pupilla cf. loessica C D O Extinct
Pupilla muscorum M M O Holarctic
Trichia hispida Cr Sh O European

Note. C, cryophilous; H, hygrophilous; W, woodland preference; Cr, cold resistant; Sh, subhygrophilous; W/O, intermediate (ecotone); M, mesophilous; O,
open habitat preference; T, thermophilous; D, aridity tolerant.

interpretation of a full-glacial vegetation dominated by steppe
or tundra for central and southeastern Europe. A number of
comparisons can be made between the composition of arboreal
species found in the modern European boreal forest (Nikolov
and Helmissaari, 1992) and those identified both from the
pollen and macroscopic charcoal evidence present in the Hun-
garian landscape during the late- and full-glaciation. A number
of these arboreal species are mesophilous and could withstand
the proposed cold and dry climate scenario. Picea and Larix
survive on semi-permanently frozen soils in many regions of
the northern European boreal forest (Korotaev, 1987; Nikolov
and Helmisaari, 1992), and species such as Pinus cembra and

TABLE 4
Classification of Molluscan Data (Percentages) from Loess Sequences According to Paleotemperature

Site number: 2 3 6 8 9 11 12 13 14 16 17 18 19 20 22 23

Cryophilous 0 3 22 14 29 2 0 0 7 4 4 20 30 24 0 0
Cold resistant 0 42 2 11 1 38 8 14 11 4 16 15 42 5 8 5
Mesophilous 56 43 74 72 10 38 67 54 80 38 61 52 27 57 85 86
Thermophilous 44 12 2 3 60 22 25 32 2 54 19 13 1 14 7 9

Note. See Fig. 2 for site locations.

 EUROPEAN FULL-GLACIAL FORESTS 209

TABLE 5
Classification of Molluscan Data (Percentages) from Loess Sequences According to Paleohumidity

Site number: 2 3 6 8 9 11 12 13 14 16 17 18 19 20 22 23

Hygrophilous 0 42 33 24 35 62 17 14 33 7 44 15 44 17 5 5
Subhygrophilous 8 18 17 15 1 13 25 52 43 12 19 2 24 9 3 4
Mesophlious 48 18 39 45 0 3 42 27 20 27 14 60 2 39 85 78
Aridity tolerant 44 22 11 16 64 22 16 7 4 54 23 33 30 35 7 13

Note. See Fig. 2 for site locations.

Pinus sylvestris grow today in regions where the air tempera-
ture and precipitation drop to values simulated for the full-
glacial environment.
The presence of thermophilous trees, such as Carpinus betu-
lus (from the macroscopic charcoal evidence) and Ulmus,
Quercus, Carpinus orientalis, and Corylus (from the pollen
evidence), is more difficult to reconcile with the predicted full-
and late-glacial climatic conditions. Although such trees are
found in small pockets at the southern edge of the modern
European boreal forest (Pastor and Mladenoff, 1992), climatic
conditions associated with this margin are much warmer and
moister than those predicted for the full-glaciation in Hungary.
The molluscan evidence from the macroscopic charcoal hori-
zons may provide the answer to this apparent anomaly.
Results from the paleoclimatic reconstructions based on the
16 molluscan records (dated between 33,000 and 19,250 cal yr
B.P.) presented in this study (Tables 3– 6) suggest that local
habitat conditions differed significantly from the regional pre-
dictions for the full-glacial climate (e.g., Kutzbach and Guetter,
1986; Kutzbach et al., 1993; Kordos, 1977, 1987). For exam-
ple, when classified according to paleoclimatological groups,
the highest percentage was from molluscs currently found in
mesophilous and/or thermophilous conditions (Table 4). The
predicted paleotemperature curve based on the molluscan ev-
idence (Sümegi, 1989) suggests a mean July paleotemperature
of between 16° and 18°C. Molluscan types found today in
humid and semihumid conditions were present at all but one
site. Thus these data would appear to contradict the paleocli-
matic models and other reconstructions. The molluscan paleo-
ecological record reflects local habitat conditions at the time of
deposition rather than broader regional patterns and therefore
indicates the microenvironmental conditions. The suggestion
can therefore be made that there were microenvironmental
“oases” that were of sufficient warmth and humidity to allow
the existence of small pockets of thermophilous fauna and flora
during the full-glaciation. Although the microenvironment cre-
ated by the coniferous trees is demonstrated in this record,
other areas that might have provided such oases include south-
facing slopes, high-altitude sites, and areas of well-drained
soils. Previous researchers suggested that small pockets of
microenvironmentally favorable conditions played an a impor-
tant refugial role in the survival of temperate tree taxa in
southern Europe during the last full-glacial (Bennett et al.,
1991; Willis, 1992; Tzedakis, 1993; Willis, 1994b). However,
until now there has been little evidence to support this sugges-
tion.
Evidence for tree populations surviving so far north dur-
ing the full-glacial maximum suggests that the in situ sur-
vival of trees at sites further south is also probable. A
number of southern European sites with paleoecological
records extending to the last full-glacial internal indicate a
low but continual presence of full-glacial tree pollen. At
Ioannina, Greece, for example, values of between 20 and
30% Quercus pollen (Tzedakis, 1993) have been recorded,
and at Grande Pile, France, there are values of between 15
and 25% Betula pollen (Woillard, 1979). Although these
values are often attributed to long-distance pollen transport,
similar to the Hungarian situation, there is increasing mac-
roscopic charcoal data (mainly from archaeological sites) to
support the suggestion of in situ tree populations. For ex-
ample, macroscopic charcoal samples of Pinus pinaster
have been recovered from Paleolithic rock-shelters in Por-
tugual that are dated to between 33,000 and 22,000 14 C yr
B.P. (approx. 36,000 and 25,000 cal yr B.P.) (Figueiral,

TABLE 6
Classification of Molluscan Data (Percentages) According to Paleovegetation

Site number: 2 3 6 8 9 11 12 13 14 16 17 18 19 20 22 23

Woodland preferring 0 5 24 13 14 27 25 0 17 6 30 5 28 14 0 5
Intermediate 56 25 17 15 1 17 25 53 46 25 31 7 24 9 85 55
Open habitat 44 70 59 72 85 56 50 47 37 69 39 88 48 77 15 40

Note. See Fig. 2 for site locations.

210 WILLIS, RUDNER, AND SÜMEGI

FIG. 4. Percentage pollen, spore, and microscope charcoal diagrams from: (top) Kis-Mohos Tó, (center) Bátorliget, and (bottom) Sárrétt plotted against age
( 14C yr B.P.) (redrawn from Willis et al., 1995, 1997; Willis, 1997).

1995). Macroscopic charcoal samples of Larix, Pinus syl-
vestris, Betula, Rhamnus, and Hippophae rhamnoides, dated
between 40,000 and 30,000 14 C yr B.P. (approx. 43,000 and
33,000 cal yr B.P.), have been recovered from a Paleolithic
rockshelter (Maspero, 1996), and at a Paleolithic site in
Slovenia, macroscopic charcoal of Pinus, Fagus, Ulmus,
Populus, Sorbus, and Rhamnus have been dated to between
approximately 25,500 and 21,800 14 C yr B.P. (approx.
29,000 and 24,800 cal yr B.P.) (Culiberg and Sercelj, 1995).
When these are viewed as individual cases, it is easy to
suggest that the samples are wrongly identified or the dating
is poor, but gradually a picture of a diverse assemblage of
 EUROPEAN FULL-GLACIAL FORESTS
FIG. 4—Continued
trees existing in southern and central Europe during the late-
and full-glaciation is starting to emerge.
If the pollen evidence can be taken as indicative of a
full-glacial presence of trees in southern and central Europe,
then the question as to why there is a dramatic reduction in
tree pollen accumulation rates must be addressed. One sug-
gestion has been that this decrease in the rate of tree pollen
in full-glacial sediments represents a reduction in total plant
biomass (Magri, 1994). To an extent, this interpretation is
probably correct because the pollen and molluscan results
do indicate a more open environment and a reduction in the
total number of trees. However, part of this decrease in tree
pollen accumulation rates may also represent a change in
plant physiology during periods of inclement weather con-
ditions. For example, work on pollen productivity in the
northern boreal forest has revealed that during years of
inclement weather conditions arboreal pollen productivity
can be reduced by up to one order of magnitude (Andersen,
1974; Hicks, 1985; Hicks, 1994). In full-glacial pollen dia-
grams this variability in pollen production could well ac-
count for some of the decline seen in pollen accumulation
rates.
CONCLUSIONS
Results from the macroscopic charcoal record of 31 sites
in Hungary dated to between 32,500 and 16,500 14 C yr B.P.
indicate that trees were growing on the Hungarian landscape
211
during the last full-glaciation. The types present were sim-
ilar to those found on the southern edge of the modern-day
northern boreal forest and include Pinus sylvestris, Pinus
cembra, Picea sp., Betula sp., Juniperus sp., Salix sp., and
Larix. An examination of the present-day ecological re-
quirements of these types (Korotaev, 1987) suggests that
most could have withstood the predicted extremes of the
full-glacial climate. However, thermophilous trees (e.g.,
Carpinus betulus, Quercus, Corylus, Ulmus, and Tilia),
which were also present, must have survived in microenvi-
ronmentally favorable pockets.
Paleoenvironmental reconstructions from the molluscan
record preserved alongside the macroscopic charcoal at 16
of the sites indicate that small microenvironmentally favor-
able pockets might have been created by the trees them-
selves. Paleoclimatic reconstructions suggest that local mi-
croenvironmental conditions in close proximity to the
arboreal vegetation were warmer and more humid than
regional simulations of the full-glacial climate (e.g.,
Kutzbach and Guetter, 1986). Therefore trees, such as those
types found in the boreal forest today, probably provided
microenvironmental oases. Within close proximity to the
trees, a thermophilous flora and fauna were also able to
survive the extremes of the full-glaciation. If, as the evi-
dence from this study suggests, trees were able to survive
this far north during the last full-glaciation, then it is highly
likely that paleoecological records from sites further south
212 WILLIS, RUDNER, AND SÜMEGI
containing tree pollen probably also represent in situ glacial
populations.
ACKNOWLEDGMENTS
The authors thank K.D. Bennett, H.J.B. Birks, B. Ammann, and A. Gardner
for helpful discussion and comments on this manuscript.
REFERENCES
Bell, M., and Walker, M. J. C. (1992). “Late Quaternary Environmental
Change.” Longman Group, UK.
Bennett, K. D., Tzedakis, P. C., and Willis, K. J. (1991). Quaternary refugia of
north European trees. Journal of Biogeography 18, 103–115.
Betancourt, J. L., van Devender, T. R., and Martin, P. S. (1990). Synthesis and
prospectus. In “Packrat Middens: The Last 40,000 Years of Biotic Change”
(J. L. Betancourt, T. R. van Devender, and P. S. Martin, Eds.), pp. 435– 447,
University of Arizona Press, Tucson.
Birks, H. H. (1994). Plant macrofossils and the nunatak theory of pre-glacial
survival. Dissertationes Botanicæ 234, 129 –143.
Bond, G., Broecker, W. S., Johnsen, S., McManus, J., Labeyrie, L., Jouzel, J.,
and Bonani, G. (1993). Correlations between climate records from the North
Atlantic sediments and Greenland ice. Nature 365, 143–147.
Clark, J. S. (1988). Particle motion and the theory of charcoal analysis: Source
area, transport, deposition and sampling. Quaternary Research 30, 67– 80.
Culiberg, M., and Sercelj, A. (1995). Anthracotomical and palynological
research in the Paleolithic site Sandalja II (Istria, Croatia). Razprave 4,
slovenska akademija Znanosti Utmentnosit 36, 49 – 47.
Denton, G. H., and Hughes, T. J. (1971). “The Last Great Ice Sheets.” Wiley,
New York.
Dobosi, V. (1967). Új felsõ-paleolit telep zs Alfóldön. Archeológiai Értesı́tõ
94, 184 –193.
Figueiral, I. (1995). Charcoal analysis and the history of Pinus pinaster (cluster
pine) in Portugal. Review of Paleobotany and Palynology 89, 441– 454.
Fink, J., and Kukla, G. (1977). Pleistocene climates in central Europe. Qua-
ternary Research 7, 363–371.
Follieri, M., Magri, D., and Sadori, L. (1988). 250,000-year pollen record from
Valle di Castiglione (Roma). Pollen et Spores 30, 329 –356.
Gábori-Csánk, V. (1960). A ságvári telep abszolút kormeghatározása. Archeo-
lógiai Értesı́tõ 87, 125–129.
Geyh, M. A., Schweitzer, F., Vértes, F., and Vogel, I. C. (1969). Neue
chronologische Angaben zur Würm-Vereisung in Ungarn. Földrajzi Értesı́tõ
18, 5–18.
Hertelendi, E., Sümegi, P., and Szöõr, G. (1992). Geochronologic and paleo-
climatic characterization of Quaternary sediments in the Great Hungarian
Plain. Radiocarbon 34, 833– 839.
Hicks, S. (1985). Modern pollen deposition records from Kuusamo, Finland.
Seasonal and annual variation. Grana 24, 167–184.
Hicks, S. (1994). Large and small scale distribution of pollen in the boreal
zone. PACT 33, 17–26s.
Johnsen, S. J., Clausen, H. B., Dansgaard, W., Fuher, K., Gundestrup, N.,
Hammer, C. U., Iversen, P., Jouzel, J., Stauffer, B., and Steffensen, J. P.
(1992). Irregular glacial interstadials recorded in a new Greenland ice core.
Nature 359, 311–313.
Kerney, M. P., Cameron, A. C. D., and Jungbluth, J. H. (1983). “Die Land-
schnecken Nord Und Mitteleuropas.” Parey, Hamburg.
Kordos, L. (1977). Changes in the Holocene climate of Hungary reflected by
the “vole-thermometer” method. Földrajzi Közlemények 25, 222–289.
Kordos, L. (1987). Climatostratigraphy of Upper Pleistocene vertebrates and
the conditions of loess formation in Hungary. Geojournal 15, 163–166.
Korotaev, A. A. (1987). Effect of soil temperature and moisture content on root
growth in coniferous cultures. Soviet Forest Science 1987, 52– 61.
Krolopp, E. (1977). Absolute chronological data of the Quaternary sediments
in Hungary. Földrajzi Közlemények 26, 228 –232.
Krolopp, E., and Sümegi, P. (1995). Paleoecological reconstruction of the Late
Pleistocene, based on loess malacofauna in Hungary. Geojournal 36, 213–
222.
Krolopp, E., Sümegi, P., Kuti, P., Hertelendi, E., and Kordos, L. (1996).
Paleoecological reconstruction of the Szeged-Öthalom loess-area. Földtani
Közlöny 125, 309 –361.
Kutzbach, J. E., Guetter, P. J., Behling, P. J., and Sehling, R. (1993). Simulated
climate changes: Results of the COHMAP climate-model experiments. In
“Global Climates since the last Glacial Maximum” (H. E. Wright, Jr., J. E.
Kutzbach, T. Webb III, W. F. Ruddimann, F. A. Street-Perrott, and P. J.
Bartlein, Eds.), pp. 24 –93. University of Minnesota Press, Minneapolis.
Kutzbach, J. F., and Guetter, P. J. (1986). The influence of changing orbital
parameters and surface boundary conditions on climate simulations for the
past 18,000 years. Journal of Atmospheric Science 43, 1726 –1759.
Laj, C., Mazaud, A., and Duplessy, J. C. (1996). Geomagnetic intensity and 14C
abundance in the atmosphere and ocean during the past 50 kyr. Geophysical
Research Letters 23, 2045–2048.
Lozek, V. (1964). Quartärmollusken der Tschechoslowakei. Rozpravy Ústred-
niko ústavu geologickeho 31, 1–375.
Lozek, V. (1987). Mollusca analysis. In “Handbook of Holocene Paleoecology
and Paleohydrology.” (B. E. Berglund, Ed.), pp. 729 –740. Wiley, London.
Magri, D. (1989). Interpreting long-term exponential growth of plant popula-
tions in a 250,000-year pollen record from Valle di Castiglione (Roma). New
Phytologist 112, 123–128.
Magri, D. (1994). Late-Quaternary changes in plant biomass as recorded by
pollen-stratigraphical data: A discussion of the problem at Valle di Castigli-
one, Italy. Review of Paleobotany and Palynology 81, 313–325.
Marosi, S., and Szilárd, J. (1974). Neuere Angabren über das Alter das
Balatons. Földrajzi Értesı́tõ 23, 333–346.
Maruszczak, H. (1987). Problems of paleographic interpretation of ice-wedge
casts in European loesses: SEM characterisation microfeatures on frost
shattered quartz grains. In “Loess and the Periglacial Phenomena” (M. Pési
and H. M. French, Eds.), pp. 285–302. Akademiai Kiado, Budapest.
Marziani, G. P., Iannone, A., Patrignani, G., and Schiattareggia, A. (1991).
Reconstruction of the tree vegetation near a Bronze Age site in Northern
Italy based on the analysis of charcoal fragments. Review of Paleobotany
and Palynology 70, 214 –246.
Maspero, A. (1996). Dati sulla vegetazione del periodo glaciale: Antracologia
dei siti paleolitici del nord Italia. II Quaternario 9, 591–598.
Meijer, T. (1985). The pre-Weichselian non-marine molluscan fauna from
Maastricht-Belvedere (southern Limburg of the Netherlands). Mededelingen
Rijks Gelogishe Dienst 39, 75–103.
Nikolov, N., and Helmisaari, H. (1992). Silvics of the circumpolar boreal
forest tree species. In “A Systems Analysis of the Global Boreal Forest”
(H. H. Shugart, R. Leemans, and G. B. Bonan, Eds.), pp. 13– 85. Cambridge
University Press, Cambridge, UK.
Pastor, J., and Mladenoff, D. J. (1992). The southern boreal-northern hardwood
forest border. In “A Systems Analysis of the Global Boreal Forest” (H. H.
Shugart, R. Leemans, and G. B. Bonan, Eds.), pp. 216 –240. Cambridge
University Press, Cambridge, UK.
Pécsi, M. (1961). The most important types of periglacial ground-frost phe-
nomena in Hungary. Földrajzi Közlemények 9, 1–24.
Pécsi, M. (1976). Lithostratigraphical subdivision of the loess sequences in
Hungary. Földrajzi Közlemények 23, 228 –239.
 EUROPEAN FULL-GLACIAL FORESTS
Pécsi, M. (1977). A hazai és az európai löszképzodmények paleogeográiai
kutatása és összehasonlı́tása. Geonómia és Bányászat 10, 183–221.
Pécsi, M. (1993). “Quaternary and Loess Research.” Akadémiai Kiadó, Budapest.
Pécsi, M., and Schweitzer, F. (1991). Short and long-term terrestrial records of
the Middle Danubian Basin. In “Quaternary Environment in Hungary” (M.
Pécsi and F. Schweitzer, Eds.), pp. 9 –26. Akadémiai Kiadó, Budapest.
Rudner, Z. E. (1994). “Upper Pleistocene Vegetation Reconstruction in Hun-
gary on the Basis of Anthracological Investigations.” Unpublished MSc
thesis, Kossuth Lajos University, Debrecen.
Sander, P. M., and Gee, C. T. (1990). Fossil charcoal: Techniques and
applications. Review of Paleobotany and Palynology 63, 269 –279.
Schweingruber, F. H. (1989). “Tree Rings—Basics and Applications of Den-
drochronology.” Kluwer Academic, Berlin.
Sparks, B. W. (1964). Non-marine mollusca and Quaternary ecology. Journal
of Animal Ecology 33, 87–98.
Stoilov, K. G. (1984). “The Loess Formation in Bulgaria.” Publishing House
of the Bulgarian Academy of Sciences, Sofia.
Stuiver, M., and Reimer, P. J. (1993). Extended 14C data base and revised
CALIB 3.0 14C age calibration program. Radiocarbon 35, 215–230.
Sümegi, P. (1989). “Upper Pleistocene Evolution History of the Hajdúsáag
(Hungary) Region, on the Basis of Stratigraphical, Paleontological, Sedi-
mentological and Geochemical Investigations.” Unpublished B.Sc. thesis,
Kossuth Lajos University, Debrecen.
Székely, A. (1987). Nature and extent of periglacial sculpturing of relief in the
Hungarian mountains. In “Loess and Periglacial Phenomena” (M. Pécsi and
H. M. French, Eds.), pp. 303–311. Akadémiai Kiadó, Budapest.
Tzedakis, P. C. (1993). Long-term tree populations in northwest Greece in
response to Quaternary climatic cycles. Nature 364, 437– 440.
van Andel, T. (1998). Middle and Upper Paleolithic environments and the
calibration of 14C dates beyond 10,000 B.P. Antiquity 72, 26 –33.
Vernet, J.-L., and Thiebault, S. (1987). An approach to the northwestern
Mediterranean recent prehistoric vegetation and ecological implications.
Journal of Biogeography 14, 117–127.
Vértes, L. (1964). Das Jung paläolithikum von Arka in Nord-Ungarn. Quartär
15/16, 132–139.
Voelker, A. H. L., Sarnthein, M., Grootes, P. M., Erlenkeuser, H., Laj, C.,
213
Mazaud, A., Nadeau, M.-J., and Schleicher, M. (1998). Correlation of
marine 14C ages from the Nordic seas with the GISP2 isotope record:
Implications for radiocarbon calibration beyond 25 ka B.P. Radiocarbon 40,
1–19.
Vogel, I. C., and Waterbolk, H. I. (1964). Groningen radiocarbon dates V.
Radiocarbon 6, 349 –369.
Whitlock, C., and Millspaugh, S. H. (1996). Testing the assumptions of
fire-history studies: An examination of modern charcoal accumulation in
Yellowstone National Park, USA. The Holocene 6, 7–15.
Willis, K. J. (1992). The late Quaternary vegetational history of northwest
Greece. III. A comparative study of two contrasting sites. New Phytologist
121, 139 –155.
Willis, K. J. (1994a). The vegetational history of the Balkans. Quaternary
Science Reviews 13, 769 –788.
Willis, K. J. (1994b). Altitudinal variation in the late Quaternary vegetational
history of northwest Greece. Historical Biology 9, 103–116.
Willis, K. J. (1997). The impact of early agriculture upon the Hungarian
landscape. In “Landscapes in Flux—Central and Eastern Europe in Antiq-
uity” (J. Chapman and P. Dolukhanov, Eds.), pp. 193–207. Oxbow Books,
Oxford.
Willis, K. J., Braun, M., Sümegi, P., and Tóth, A. (1997). Does soil change
cause vegetation change or vice versa? A temporal perspective from Hun-
gary. Ecology 78, 740 –750.
Willis, K. J., Sümegi, P., Braun, M., Bennett, K. D., and Tóth, A. (1998).
Prehistoric land degradation in Hungary: Who, how and why? Antiquity 72,
101–113.
Willis, K. J., Sümegi, P., Braun, M., and Tóth, A. (1995). The late Quaternary
environmental history of Bàtorliget, N.E. Hungary. Paleogeography, Paleo-
climatology, Paleoecology 118, 25– 47.
Woillard, G. (1979). Grand Pile peat bog: A continuous pollen record for the
last 140,000 years. Quaternary Research 9, 1–21.
Wright, H. E. (1967). A square rod piston sampler for lake sediments. Journal
of Sedimentary Petrology 37, 975–976.
Zackrisson, O., Nilsson, M., and Wardle, D. (1996). Key ecological function
of charcoal from wildfire in the boreal forest. Oikos 77, 10 –19.