Flood signals in tree-ring δO and wood anatomical parameters of Lagerstroemia speciosa: Implications for developing flood management strategies in Bangladesh.

Bangladesh consists of 80% of the flood plain of the Ganges-Brahmaputra-Meghna river system (GBM), making the country one of the highest flood prone countries of the world. Due to the high rate of discharge of the GBM caused by the summer monsoon and the snowmelt of the Eastern Himalaya and Southern Tibetan Plateau due to climate change, Bangladesh witnessed 16 flood events over 1954-2017. We performed a multiproxy tree-ring analysis to investigate the impact of extreme flood events on tree growth, xylem anatomical parameters and oxygen isotope composition of tree-ring cellulose (δO) in a Bangladeshi moist tropical forest and to establish relationships between water level of the regional rivers and tree-ring parameters. By using pointer year analysis and comparing the pointer years with historical flood records (a cut-off threshold of the country's flooded land area of 33.3%), we identified the three extreme flood events (hereafter called flood years) 1974, 1988, and 1998 in Bangladesh. Superposed epoch analysis revealed significant changes in Tree-ring width (TRW), total vessel area (TVA), vessel density (VD), and δO during flood years. Flood associated hypoxic soil conditions reduced TRW up to 53% and TVA up to 28%, varying with flood events. In contrast, VD increased by 23% as a safety mechanism against flood induced hydraulic failure. Tree-ring δO significantly decreased during the flood years due to the amount effect in regional precipitation. Bootstrapped Pearson correlation analysis showed that wood anatomical variables encoded stronger river level signals than TRW and δO. Among the wood anatomical parameters, VD showed a strong relationship (r = -0.58, p < 0.01) with the water level of the Manu River, a regional river of the north-eastern part of Bangladesh, indicating that VD can be used as a reliable proxy for river level reconstruction. Our analyses suggest that multiproxy tree-ring analysis is a potential tool to study tropical moist forest responses to extreme flood events and to identify suitable proxies for reconstructing hydrological characteristics of South Asian rivers.