1.36 million years of Mediterranean forest refugium dynamics in response to glacial-interglacial cycle strength

Abstract

The sediment record from Lake Ohrid (Southwestern Balkans) represents the longest continuous lake archive in Europe, extending back to 1.36 Ma. We reconstruct the vegetation history based on pollen analysis of the DEEP core to reveal changes in vegetation cover and forest diversity during glacial-interglacial (G-IG) cycles and early basin development. The earliest lake phase saw a significantly different composition rich in relict tree taxa and few herbs. Subsequent establishment of a permanent steppic herb association around 1.2 Ma implies a threshold response to changes in moisture availability and temperature and gradual adjustment of the basin morphology. A change in the character of G-IG cycles during the Early-Middle Pleistocene Transition is reflected in the record by reorganization of the vegetation from obliquity- to eccentricity-paced cycles. Based on a quantitative analysis of tree taxa richness, the first large-scale decline in tree diversity occurred around 0.94 Ma. Subsequent variations in tree richness were largely driven by the amplitude and duration of G-IG cycles. Significant tree richness declines occurred in periods with abundant dry herb associations, pointing to aridity affecting tree population survival. Assessment of long-term legacy effects between global climate and regional vegetation change reveals a significant influence of cool interglacial conditions on subsequent glacial vegetation composition and diversity. This effect is contrary to observations at high latitudes, where glacial intensity is known to control subsequent interglacial vegetation, and the evidence demonstrates that the Lake Ohrid catchment functioned as a refugium for both thermophilous and temperate tree species.

Keywords: Pleistocene; glacial–interglacial cycles; pollen; tree diversity; vegetation dynamics.

Fig. 1.

(A) Location of LO and TP on the Balkan Peninsula. (B) Local setting around LO, bathymetry (81), and DEEP coring site (adapted from ref. 32).

Fig. 2.

(A) Pollen diagram of ecological groups (%) against chronology of Wagner et al. (33): montane trees (Abies, Betula, Cedrus, Fagus, Ilex, Picea, Taxus, Tsuga); mesophyllous trees (Acer, Buxus, Carpinus betulus, Carya, Castanea, Celtis, Corylus, Fraxinus excelsior/oxycarpa, Hedera, Ostrya/Carpinus orientalis, Pterocarya, Q. cerris–type, Q. robur–type, Tilia, Ulmus, Zelkova); Sclerophyllous trees (Arbutus, Cistus, Fraxinus ornus, Olea, Phillyrea, Pistacia, Quercus ilex–type, Rhamnus); wetland trees (Alnus, Liquidambar, Platanus, Populus, Salix, Tamarix, Taxodium-type); and semiarid shrubs (Ephedra, Ericaceae, Hippophaë, Cupressaceae [mostly Juniperus-type]). (B) TP (location see Fig. 1): AP percentages excluding Pinus, Betula, and Juniperus-type (–58, 60). (C) Model timeseries of annual precipitation and (D) mean SAT for the LO grid cell (33, 67). (E) MEDSTACK planktonic δ¹⁸O data (82). (F) Stacked benthic δ¹⁸O data for Ocean Drilling Program sites 967 and 968 from the eastern Mediterranean (83). (G) LR04 δ¹⁸O global benthic stack (24).

Fig. 3.

Relicts and richness. (A) Range and relative abundance of relict tree taxa in the DEEP site of LO together with (B) total % AP (filled blue) and % total relict taxa (filled gray, excl. Fagus). Bars represent % abundances for relict and rare tree taxa and are mirrored to enhance visibility. (C) Tree taxa richness of the DEEP site is based on taxa accumulation curves of tree pollen along a moving window. Single data points represent total richness from SAC (see Methods) with a sliding window of size 10 (red dotted line), 15 (black line), or 20 (green dashed line). All LO data are plotted against the chronology of Wagner et al. (33). (D) LR04 δ¹⁸O global benthic stack (24).