and data sets are compared. It is stressed that for a given (chl), the ap(it) coefficients show large residual variability around the regression lines (for instance, by a factor of 3 at 440 nm). The consequences of such a variability, when predicting or interpreting the diffuse reflectance of the ocean, are examined, according to whether or not these variations in ap are associated with concomitant variations in particle scattering. In most situations the deviations in ap actually are not compensated by those in particle scattering, so that the amplitude of reflectance is affected by these variations.

Abstract. Chlorophyll-specific absorption coefficients of particles, a • (X), and of phytoplankton, a ;h(•-), were determined using the glass-fiber filter technique along 150øW in the equatorial Pacific (13øS-løN). A site-specific algorithm for correcting the path length amplification effect was derived from field measurements. Then a decomposition technique using the high-performance liquid chromatography pigment information and taking into account the package effect was used to partition a;h into the contributions of photosynthetic pigments (a The spectrophotometer automatically corrected for the baseline which was stored prior to analyses. All spectra were set to 0 at 750 nm to minimize differences between sample and reference filters. Such an approximation, which would be questionable for detrital samples, is justified here because total absorption was largely dominated by living phytoplankton (see Figures 4a, 4b, 4e, and 4f). The optical densities measured on the filter, ODf(X), were corrected for the path length amplification effect using a site-specific algorithm, as described in the Appendix (equation (8) Partitioning Absorption into Photosynthetic and Nonphotosynthetic ComponentsThe chl a-specific absorption spectrum of phytoplankton, a,oh(X), was further partitioned into its photosynthetic and nonphotosynthetic components, a ps(X) and a np•(X) (see (1) Pigment Analyses and Other MeasurementsSamples for pigment measurements were collected on Whatman GF/F filters and either analyzed immediately or stored in liquid nitrogen for later analysis. Pigments were quantified using HPLC according to the procedure described by Vidussi et al. [1996]. The algal pigments identified and quantified include the following photosynthetic pigments: chlorophylis a, b, c, DV chl a, divinyl chlorophyll b (DV chl b), peridinin, 19'-hexanoyloxyfucoxanthin (19'-HF), 19'-butanoyloxyfucoxanthin (19'-BF), prasinoxanthin, fucoxanthin, a carotene, and three nonphotosynthetic pigments: zeaxanthin, diatoxanthin, and diadinoxanthin. The /3 carotene, which is a nonphotosynthetic pigment, was not discriminated from a carotene, which results in underestimating a* and overestimat-* Analyses previously performed on oligotrophic waters ing a p•.in the tropical North Atlantic, however, showed that the /3 carotene-to-zeaxanthin ratio was <10% throughout the water column, so that the error on these coefficients is thought to be very limited.Cell number densities for Prochlorococcus, Synechococcus, and picoeukaryots were determined using a FACSort flow cy-

The photosynthetic prokaryote Prochlorococcus appears to have a high capacity to modify its physiological and optical properties in response to changes in available irradiance. In order to study the time scale of variations in light absorption induced by photoacclimation, 2 strains (MED, clone CCMP 1378, and SARG, unialgal strain) were grown in batch cultures at high (56 7 pm01 quanta m-* S-') or low (8.4 1.1rnol quanta m-' S-') irradiance. Then the temporal changes over 3 to 4 d in spectral absorption coefficients, pigment composition, cell number density and size distribution were followed for cultures transferred from high to low irradiance, and vice versa. Both strains experienced sign~ficant changes in their divinyl-chlorophyll a-specific absorption coefficients (a') in response to the transfers. For the MED strain, photoacclimation appeared to be achieved within about 40 h (covering 2 cell generations) for the low to high irradiance transfer, while 3 to 4 d (correspondmg to l doubling of the population) were necessary for the high to low irradiance transfer For the SARG strain transferred from h~g h to low irradiance, in spite of a rapid change in absorption during the first 25 h (i.e. within the same cell generation), full photoacclimation was not achieved after 3 d. The efficiency factors for absorption, Qa(h), and the a'(h) coefficients, were reconstructed from theory at the different photoacclimation stages from the cell characteristics, i.e. intracellular concentrations of the various pigments and cell size distribution. This permitted the determination of the parameters which are mainly responsible for the observed changes. The Q&) values (and therefore the package effect) are enhanced at low irradiance by 2 effects resulting from photoacclimation: the increase of the intracellular concentration of divinylchlorophyll a, and (for the SARG straln only) the increase of the intracellular concentration of chlorophyll b. In addition, thegresence of zeaxanthin, in stable amounts within the cells whatever the irradlance, enhances the Q,$) values for all light conditions, and thus 'moderates' its variations with irradiance. Contrary to what is commonly admitted because of its tiny size, the absorption efficiency of Proch1orococcus (per pigment unit) is not always maximal, but can be reduced by 20 to 25 % at low irrad i a n c e~, such as those prevailing in the lower part of the euphotlc zone. This reduction directly affects the amount of absorbed energy usable for photosynthesis.

A new method has been developed to measure the light absorption coefficients of marine particles. The procedure, adapted from one proposed earlier for microscopic observation of nanoplankton, consists of concentrating particles onto a Nuclepore filter, transferring the filtered material to a glass microscope slide using liquid nitrogen freezing, and finally measuring the particle absorption spectrum on the slide. Measurements on various phytoplanktonic species show that the transfer efficiency is 94–98% and that no alteration of absorption spectra occurs. Our new method has been tested successfully on both phytoplankton‐dominated and detritus‐dominated field samples. In comparison with the widely used glass‐fiber filter technique, the major advantage of our procedure is to provide more accurate absorption coefficients, since the pathlength amplification effect (β factor) and its attached uncertainties are completely eliminated.