Testing the REVEALS-based vegetation reconstruction around Lake Biwa, western Japan, using absolute pollen productivity estimates derived from the flower counting method

Lake Biwa pollen records and the REVEALS model provide new insight into the past dynamic of Japanese vegetation. Instead of relying solely on pollen percentages, we employed the REVEALS model with absolute pollen productivity estimates (aPPEs) derived from flower counting approaches. The locally-obtained aPPEs, combined with a Lagrangian stochastic dispersal model (LSM), yielded results that were dramatically closer to the modern vegetation survey data within 100 km of the lake. For instance, Japanese cedar pollen greatly over-represents the actual landscape composition greatly (63.8% pollen vs 17.5% tree cover); however, the REVEALS-LSM estimate was much closer to a more accurate 24.7 ± 3.4%, mirroring the vegetation data far more faithfully. Notably, the Gaussian Plume model within REVEALS with aPPEs proved ineffective for the Lake Biwa pollen record, particularly for plants that produce heavy pollen, such as fir. Compared to the relative pollen productivity estimates obtained and applied in Europe, China, and elsewhere, flower-counting-derived aPPEs combined with the LSM appear optimal for REVEALS-based reconstruction in Japan. This opens exciting opportunities for exploring forest dynamics and human impacts throughout the 20th century, the Holocene in Japan, and possibly beyond.