Phytoplankton spectral absorption as influenced by community size structure and pigment composition

CORRESPONDING AUTHOR: 0 NAVAL RESEARCH LABORATORY 1 DEPARTMENT OF MARINE SCIENCE 03 lohrenz (147J)(ds) Assessments were made of the relative importance of package effects and pigment composition in contributing to variations in spectral absorption in shelf waters off North Carolina during May 1997 and off west Florida during October 1998. Measurements of spectral absorption of sizefractionated particulate material on glass fibre filters were made using two methods, the transmittance-reflectance (T-R) method and the quantitative filter technique (QFT). Spectral absorption of phytoplankton pigments was decomposed into a series of 13 Gaussian absorption bands, and absorption band peak heights were related to concentrations of major pigment classes. Maximum weightspecific pigment absorption coefficients for individual absorption bands (p*m) derived from the fit of a hyperbolic tangent function to the data were found to be similar for North Carolina and west Florida shelf waters. The values were used to reconstruct spectral absorption in the absence of pigment packaging, which was then compared to measured absorption to provide an assessment of pigment packaging. Package effects were found to be responsible for up to a 62% reduction in the amplitude of major absorption bands, particularly for samples from low-salinity waters and for populations dominated by larger (>3 m) phytoplankton. Variations in pigment composition were also found to have an impact, although it was smaller (10-28%), on variations in total absorption. Potential biooptical applications of the Gaussian decomposition approach include the estimation of pigment concentrations from in situ or remotely sensed ocean colour observations. Alternatively, where pigment concentrations are known, it may be possible to estimate absorption. Successful application of such techniques may necessitate characterizations of coefficients specific to a given region and time. - NUMBER PAGES Phytoplankton light absorption is a major factor contributing to the variation in optical properties of oceanic and coastal waters. The properties of phytoplankton spectral absorption form an integral part of a variety of bio-optical algorithms to estimate phytoplankton biomass and other constituents (Roesler and Perry, 1995; Garver and Siegel, 1997; Bricaud et al., 1998; He et al., 2000; Sathyendranath et al., 2001) and model primary production (Platt and Sathyendranath, 1988; Morel, 1991; Sosik, 1996; Bouman et al., 2000a,b; Lohrenz et al., 2002). Continued improvements in the performance and spectral resolution of in situ optical instruments and airborne and satellite sensors have generated increasingly sophisticated information about the optical properties of natural waters. Derivations of meaningful bio-optical information from these data hinge on improved understanding of the variability in absorption characteristics of phytoplankton and its relationship to phytoplankton community composition, size structure, and pigment concentrations. Phytoplankton spectral absorption can vary as a consequence of composition and concentration of pigments as well as of pigment packaging (Duysens, 1956; Sathyendranath et al., 1987; Hoepffner and Sathyendranath, 1991, 1993; Nelson et al., 1993) and macromolecular interactions (Geider and Osborne, 1987; Kirk, 1994). Such factors are known to differ among species and in relation to cell size. Variability in these factors may be more pronounced in coastal waters (Nelson et al., 1993; Bricaud et al., 1995; Cleveland, 1995; Arbones et al., 1996) VOLUME NUMBER PAGES where gradients in species and physiology may occur over smaller spatial and temporal scales. Several investigators have described an increase in chlorophyll-specific absorption, with decreasing chlorophyll concentrations (Bricaud et al., 1995, 1998; Cleveland, 1995). Such relationships reflect a general trend of decreasing cell size with decreasing chlorophyll concentrations, but are also representative of a relative increase in concentrations of accessory pigments in waters with low chlorophyll concentrations (Bricaud et al., 1995, 1997; Sakshaug et al., 1997; Ciotti et al., 1999). The performance of semi-analytical models in describing relationships of ocean colour to chlorophyll (Carder et al., 1999) and diffuse attenuation (Ciotti et al., 1999) has been shown to be sensitive to variations in the chlorophyllspecific absorption coefficients that accompany changes in phytoplankton cell size and pigment composition. In addition, examination of large data sets from laboratory and field studies has shown that a metric related to phytoplankton cell size can be used as a parameter to explain a large proportion of the variation in the shape of the phytoplankton absorption spectrum (Ciotti et al., 2002). These prior studies did not explicitly separate the relative contributions of pigment packaging and accessory pigment effects, but instead relied on the fact that variations in cell size and accessory pigmentation co-vary in a predictable manner (Ciotti et al., 1999, 2002; Trees et al., 2000). Knowledge of the extent to which variations in pigmentation related to taxonomic and physiological variations can account for residual variation in size-based empirical relationships may help in refining semi-analytical models and may improve our understanding of the basis for variations in phytoplankton spectral absorption. Variations in accessory pigment concentrations and composition, exclusive of packaging effects, can have a substantial impact on chlorophyll-specific absorption, particularly in the blue region of the spectrum. Prior studies have estimated that 1442% of the variation in chlorophyll-specific absorption at 440 nm can be attributed to pigment composition effects (Babin et al., 1996; Lazzara et al., 1996; Allali et al., 1997; Stuart et al., 1998). Information about the absorption properties of individual pigments can be derived by decomposition of phytoplankton absorption into bands attributable to individual pigments. The observed features can then be normalized to pigment concentrations determined by high-performance liquid chromatography (HPLC) to obtain weight-specific absorption coefficients for pigments in their particulate form, a*i() (m 2 mg pigment1). Methods that have been applied to the decomposition of in vivo phytoplankton absorption spectra into contributions by pigment groups include derivative analysis (Bidigare et al., 1988, 1989), spectral reconstruction methods (Bidigare et al., 1990; Babin et al., 1996; Allali et al., 1997) and the Gaussian decomposition approach (Hoepffner and Sathyendranath, 1993). We chose to use the Gaussian model because it is firmly rooted in the physical processes of electronic absorptions and can be used to reduce hyperspectral data to a series of Gaussian curves with almost no loss of information (Cloutis, 1996). The Gaussian model is relatively insensitive to noise, and Gaussian bands (...truncated)