Determination of phytoplankton absorption coefficient in natural seawater samples: evidence of a unique equation to correct the pathlength amplification on glass-fiber filters

Abstract

Estimations of the spectral absorption coefficients of phytoplankton in the sea, based on the spectrum of absorption of particles retained on a filter, require a correction for pathlength amplification in the glass fibre filter (beta factor). The relationship between the optical density in suspension and 'on filter' is misleading due to the hysteresis effect. This effect is reduced on eliminating the absorption due to non-pigmentary material, minimising the dependency of the beta factor on wavelength. A general quadratic equation, ODsus(lambda) = 0.38ODfilter(lambda) + 0.42OD2filter(lambda) (r2 = 0.97, n = 15600), was obtained on comparing 52 pairs of spectra of 9 marine phytoplankton species. The algorithm was validated against natural samples, obtaining a 1% average error. Comparisons between Cleveland & Weidemann's equation (1993, Limnol Oceanogr 38:1321-1327) and the equation obtained in this study provided a 2% average difference, suggesting that a unique equation can be useful in determining the phytoplankton absorption coefficient in seawater samples. We compared the absorption coefficients obtained by spectra reconstructed from the pigment content in seawater samples with those obtained with spectra on GF/F filters and beta-corrected. Our results indicate that the spectra-reconstructed approach is only suitable when the sample consists of small quasi-spherical cells. For coastal-upwelling samples which are mainly composed of large cells the approach overestimates the package effect.