Modeling the irradiance dependency of the quantum efficiency of photosynthesis
Limnol. Oceanogr. Methods 10:645-652 (2012) | DOI: 10.4319/lom.2012.10.645
ABSTRACT: Measures of the quantum efficiency of photosynthesis (fPSII) across an irradiance (E) gradient are an increasingly common physiological assay and alternative to traditional photosynthetic-irradiance (PE) assays. Routinely, the analysis and interpretation of these data are analogous to PE measurements. Relative electron transport rates (rETR = E × ΦPSII) are computed and fit to a PE curve to retrieve physiologically meaningful PE parameters. This widespread approach is statistically flawed as the response variable (rETR) is explicitly dependent on the predictor variable (E). Alternatively the E-dependency of ΦPSII can be modeled directly while retaining the desired PE parameters by normalizing a given PE model to E. This manuscript presents a robust analysis in support of this alternative procedure. First, we demonstrate that scaling ΦPSII to rETR unnecessarily amplifies the measurement error of ΦPSII and using a Monte-Carlo analysis on synthetic data induces significantly higher uncertainty in computed PE parameters relative to modeling the E-dependency of ΦPSII directly. Next a large dataset is simultaneously fitted to four PE models implemented in their original and E-normalized forms. Four statistical criteria used to evaluate the efficacy of nonlinear models demonstrate improved model fits and more precise PE parameters when data are modeled as E dependent changes in ΦPSII. The analysis presented in this manuscript clearly demonstrates that modeling the E-dependency of ΦPSII directly should be the norm for interpreting active fluorescence measures.