Palaeo-water depth estimation for a 600-year record from Nam Co (Tibet) using an ostracod-based transfer function

Abstract

A modern training set of ostracods from the large slightly saline lake Nam Co on the southeastern Tibetan Plateau is presented. In total 44 samples covering the largest part of the lake and water depth from 2 to 93 m were taken. There are eight ostracod species in surface sediments of Nam Co: Leucocytherella sinensis, ? Leucocythere dorsotuberosa, Fabaeformiscandona gyirongensis, Candona xizangensis, Candona candida, Ilyocypris cf. mongolica, Eucypris gyirongensis and Limnocythere inopinata (in order of mean dominance). A morphological variant of ? L. dorsotuberosa, the forma postilirata, is an indicator form of deep water below the thermocline. The 36 samples comprise a sufficient number of ostracods for analysis covering water depths between 3.7 and 89.3 m. Two transfer function models based on weighted average binomial regression (WA-BR) and weighted average partial least square regression (WA-PLS) using the programs EXCEL and C2 were established to estimate past water depths of Nam Co. Transfer functions were applied to ostracod data from two short cores from Nam Co covering approximately the past 600 years. Model results show a lake level minimum occurring during the Little Ice Age (LIA) between about 1490 and 1760 AD. This minimum corresponds to results from previous stable isotope analysis and indicates a lake level which was about 5 m below the present-day level. The more exact transfer function model (WA-PLS) was set up using C2 and revealed a higher correlation index ( r) of 0.91 and a lower standard error of 6.7 m. However, the more selective (because of using indicator taxa only) WA-BR model using EXCEL ( r = 0.88, standard error 10.6 m) seems to represent palaeo-water depths better, at least for the studied short cores.