海洋浮游植物三维荧光光谱的逐层分类方法研究

Tiered Classification Approach for Three-Dimentional Fluorescent Fingerprints of Marine Phytoplankton

提出活体海洋浮游植物三维荧光光谱逐层分类法,将浮游植物区分为不同的"光谱类别"。选取中国近海常见的26种分属于甲藻、硅藻、黄藻、金藻、隐藻、绿藻、蓝藻的浮游植物,在室温20℃和3个光照条件下培养,在第3,6,9,12,15天测其三维荧光光谱。采用主成份分析结合非负最小二乘回归法(NNLSR),以浮游植物叶绿素a的发射峰680nm所对应的激发光谱作为第一特征谱以区分门类,以EX730谱作为硅藻的逐层分类的第二特征谱,SY350谱作为甲藻的第二特征谱,将实验所用9种硅藻和8种甲藻各划分为3个光谱类别;分析405个浮游植物混合物,门类识别正确率大于94%,光谱类别的识别正确率为75%,该方法可用于海洋浮游植物种群结构变化的快速宏观监测。

This paper presents a tiered classification approach for the three-dimensional fluorescence spectra of in vivo marine phytoplankton. Twenty-six phytoplankton species belonging to divisions of Dinophyta, Bacillariophyta, Xanthophyta, Chrysophyta, Cryptophyta, Chlorophyta and Cyanophyta were cultivated in lab under a temperature of 20℃ and three illumination intensities（15000 lux, 10000 lux, 6000 lux）, and three-dimensional fluorescence spectra were measured at 3, 6,9, 12, 15 days. Principle components analysis （PCA） combined with non-negative least-squares regression （NNLSR） method were used in data process. Firstly, the excitation spectra corresponding to the emission peaks of Chl a at wave- length of 680nm （EX680） were chosen as the primary feature spectra to discriminate the phytoplankton at the level of divisions. Subsequently, EX730 spectra were screened as the secondary feature spectra in the division of Bacillariophyta and the nine Bacillariophyta species were classified into three spectral catego- ries, while SY350 as the secondary feature spectra in the division of Dinophyta and the eight Dinophyta species into three spectral categories. Results showed that this tiered classification approach could identify the 405 mixed phytoplankton samples at the level of divisions and spectral categories with accuracies of 94% and 75%, respectively. Thus, this tiered classification approach might be applied in rapid monitoring of overall changes in marine phytoplankton populations.