Use of principal component scores in multiple linear regression models for simulation of chlorophyll-a and phytoplankton abundance at a karst deep reservoir, southwest of China

The relationships between chlorophyll-a, phytoplankton abundance and 20 chemical, physical and biological water quality variables were studied by using principal component scores (PCs) in stepwise linear regression analysis (SLR) to simulate chlorophyll-a and phytoplankton abundance at a karst deep reservoir, southwest of China. Score values obtained by PC scores were used as independent variables in multiple linear regression models. The following models were used to simulate chlorophyll-a and abundance of Cyanobacteria, Chlorophyta, Bacillariophyta, and Pyrrophyta respectively: chlorophyll-a1 = 10.501 + 1.390 (score 1) (P < 0.01), chlorophyll-a 2 = 10.501 + 1.102 (score 1)−0.877 (score 2) (P < 0.05), log10 (Cyanobacteria) = 1.277−0.726 (score 2) (P < 0.05), log10 (Chlorophyta) = 3.927−0.150 (score 2) (P < 0.01), log10 (Bacillariophyta) = 4.872−0.131 (score 4) (P < 0.01) and log10 (Pyrrophyta) = 2.463 + 0.578 (score 1) (P < 0.05). The models could be used to simulate chlorophyll-a and phytoplankton abundance levels successfully, and revealed that DO, WD, Tem, TD, pH, NH4–N and TSS were the most important factors regulating the composition of chlorophyll-a and Pyrrophyta abundance. ORP, Cl−, , TN were the main factors affecting Chlorophyta and Cyanobacteria abundance. F− and Ca2+ were the main factors influencing the Bacillariophyta abundance.