Estimation of the photosynthesis/irradiance (P/I) curve parameters from light and dark bottle experiments

Light and dark bottle experiments, carried out in three systems in the Netherlands, were used to estimate the parameters of models relating the oxygen production rate to incident light intensity. The maximum production rate (P-M) and the light saturation constant (I-S) were estimated using both gross and nett oxygen production data. In the latter case, the community oxygen consumption rate (R-ox) was also estimated. Eight models were compared with respect to goodness of fit, as has been accomplished previously by Jassby and Platt (Limnol. Oceanogr., 21, 540-547, 1976) and many others. This study, however, emphasizes the problem of parameter correlation and the consequential usefulness of these type of experiments to identify the P/I curve parameters. The results show that at a 90% level of confidence, the models cannot be distinguished with respect to goodness of fit. However, the models do show distinct differences in parameter correlation. Parameter correlation was shown to be related to the shape of the curve. P/I curves with high convexity were shown to be less sensitive for parameter correlation. In particular, models showing low convexity suffer from over-parameterization, which means that on the basis of the observed production rates it is difficult to discriminate between the parameters. Also, alternative model formulations, using P-M and the initial slope (alpha), were investigated. These formulations produced less parameter correlation. However, for models with low convexity, showing high parameter correlation anyway, the reduction is limited. The use of nett oxygen production data does not show a significant difference in fit at a 90% confidence level. However, measured R-ox from dark bottle experiments tends to be higher than the values found by estimating R-ox from nett oxygen production.