A new modern analogue reference dataset and its application to the 430-kyr pollen record from Lake Biwa

This study presents a newly compiled dataset of modern pollen and climate data from 798 sites across Japan and the Russian Far East. This comprehensive reference dataset combined with the modern analogue technique (MAT) provides a powerful tool for pollen-based reconstruction of the Quaternary Northwest Pacific climate. Pollen-derived reconstruction of the modern climate at the reference pollen-sampling sites matches well with the estimated modern climate values (R2 values vary between 0.79 and 0.95, and RMSEP values vary between 5.8 and 9.7% of the modern climatic range for all nine tested variables). The successful testing of the method encourages its application to the fossil pollen records. We used a coarse-resolution pollen record from Lake Biwa to reconstruct glacial-interglacial climate dynamics in central Japan since ~438 kyr and compared it to the earlier reconstruction based on a less representative reference dataset. The current and earlier results consistently demonstrate that the coldest glacial intervals experienced pronounced cooling in winter and moderate cooling in summer, supporting the growth of cool mixed forest (COMX) where warm mixed forest (WAMX) predominates today. During the last glacial, maximum (~24 kyr BP) mean temperatures of the coldest (MTCO) and warmest (MTWA) month were about -13 °C (RMSEP = 2.34 °C) and 21 °C (RMSEP = 1.66 °C) respectively, and annual precipitation (PANN) was about 800 mm (RMSEP = 158.06 mm). During the thermal optimums of the interglacial intervals, the temperatures of the coldest and warmest month were above 0 °C and 25 °C respectively, leading to the reconstruction of WAMX and temperate conifer forest (TECO). Although both these vegetation types grow in the southern part of Japan today, WAMX requires warmer space. The presence of WAMX during marine isotope stages (MIS) 11 and 1, and its absence during MIS 9 and MIS 5 contradict the marine isotope and Antarctic ice records, suggesting that the latter two interglacials were the warmest of the last 800 kyr. The apparent contradiction allows at least three different explanations including low temporal resolution of the pollen record; different trends in CO2 concentrations during 'short' and 'long' interglacials; and regional climate variability and non-linear response of different regions to the global forcing. More definitive conclusions will be possible on the basis of forthcoming high-resolution pollen records from central Japan