Adapting the CHEMTAX Method for Assessing Phytoplankton Taxonomic Composition in Southeastern U.S. Estuaries

CHEMTAX is a matrix factorization program used to derive taxonomic structure of phytoplankton from photosynthetic pigment ratios. The program was originally developed from and applied to the analysis of oceanic phytoplankton assemblages. We found that application of the original CHEMTAX reference matrix to southeastern United States estuarine systems produced inaccurate results, as verified by microscopy. Modification of the matrix, based primarily on the pigment ratios of 33 estuarine isolates, improved the predictive capabilities of CHEMTAX for our samples. Limitations of the method included an overestimation of diatom biomass (due to the inability to differentiate diatoms from taxa with chloroplasts derived from diatom endosymbionts, notably some dinoflagellates) and a tendency to exclude some raphidophyte species. In complement with microscopic verification, the method was shown to improve assessment of phytoplankton taxonomic composition. Introduction Quantitative taxonomy of phytoplankton assemblages traditionally have been determined by microscopic examination of water samples. This process is time consuming, requires great skill and expertise, and is often subjective. Nanoplankton and picoplankton dominated assemblages are particularly difficult to identify and quantify, and require a combination of light, epifluorescent, and electron microscopy. In recent years, high performance liquid chromatography (HPLC) has been used to estimate phytoplankton composition by identifying photosynthetic pigments of chemotaxonomic relevance. New analytical techniques allow an increasing number of algal pigments to be resolved, and HPLC is used extensively (Wright et al. 1991; Van Heukelem et al. 1994; Jeffrey et al. 1997; Van Heukelem and Thomas 2001). Advantages to * Corresponding author; tele: 843/762-8868; fax: 843/7628737; e-mail: [email protected] † Current address: NOAA, Hollings Marine Laboratory, 331 Ft. Johnson Road, Charleston, South Carolina 29412. ‡ Current address: Department of Family Medicine, Medical University of South Carolina, Charleston, South Carolina 29425. HPLC pigment analyses include rapid turnover and reproducible results (Millie et al. 1993; Wright et al. 1996). HPLC does not provide the same taxonomic resolution as microscopy. Early HPLC analyses were qualitative and relied on the detection of diagnostic (marker) pigments present in only one or two groups; e.g., peridinin for Dinophyceae, alloxanthin for Cryptophyceae, prasinoxanthin for some Prasinophyceae (Gieskes and Kraay 1986; Klein and Sournia 1987). Quantitative methods were developed later to estimate the abundance of various phytoplankton classes from ratios of marker pigments to chlorophyll a (chl a; Gieskes et al. 1988; Everitt et al. 1990; Letelier et al. 1993). These methods gave unreliable results for taxa containing ambiguous pigment markers (Mackey et al. 1996). Mackey et al. (1996) introduced CHEMTAX, a MATLAB (The MathWorks, Inc., Natick, Massachusetts) program for describing the relative abundances of taxonomic groups. Rather than using simple ratios of marker pigments, CHEMTAX uses a steepest-descent algorithm to fit a matrix of expected pigment ratios for several taxa, to one consisting of the actual pigment ratios from unknown Application of CHEMTAX to Estuaries 161 samples. It has been used to describe oceanic phytoplankton assemblages, using pigment data obtained from literature (Wright et al. 1996; Mackey et al. 1998; Higgins and Mackey 2000; Riegman and Kraay 2001; Furuya et al. 2003). Although it does not rely on marker pigments to form groups, greater resolution and accuracy is achieved when taxa with distinctive markers are present. CHEMTAX requires that the number of pigments be slightly greater than the number of taxonomic groupings (i.e., phytoplankton classes). The reference matrix must also include each major class likely to be present in the samples (Mackey et al. 1996). Because the predictive capabilities of CHEMTAX depend on the pigment ratios used for the reference matrix, it is critical to calibrate CHEMTAX to the assemblages from which samples will be taken (Mackey et al. 1996; DiTullio et al. 2003). These pigment ratios should be obtained from isolates representing taxa local to the area of investigation. In our study, we applied CHEMTAX analyses to generate estimates of southeastern United States estuarine phytoplankton assemblages. We performed dual analyses, comparing results from Mackey et al.’s (1996) matrix (based on oceanic assemblages), to our matrix of pigment ratios primarily from estuarine phytoplankton isolates. Materials and Methods CULTURES A pigment matrix was developed that included 12 taxonomic groups (Table 1). Nine of these groups were based on pigment compositions of 33 estuarine isolates. Estuarine representatives of prasinoxanthin-containing prasinophytes, and flagellates with 199-hexanoyloxyfucoxanthin or 199-butanoyloxyfucoxanthin, were not available to the project. Groups 5 (Prasino-B), 9 (Hapto-B), and 10 (Chryso-B) were based on Mackey et al.’s (1996) Prasinophyceae Type 2, Haptophyceae Type 4, and Chrysophyceae Type 2, respectively. All cultures were routinely maintained at 248C, on a 14:10 h light:dark cycle (60–80 mE m22 s21). Culture media was f/2-enriched (nutrient, trace metal, and vitamin additions following Guillard 1975, but without silicate) filtered North Inlet, South Carolina, water with salinity adjusted to 16‰ or 30‰, depending on the species. The effects of physiological status on pigment composition were examined in 10 isolates (designated by * in Table 1) by growing cultures under light-limiting or light-saturating conditions and harvesting cultures during exponential (i.e., nutrient-replete) and stationary (nutrient-deplete) growth phases. Isolates were grown in 250 ml batch cultures in f/2 media at 248C (n 5 3 per light treatment). A 14:10 h light:dark cycle was used under two light treatments. Low light (LL) was set at 25 mE m22 s21, and high light (HL) was set at 250 mE m22 s21. Heterosigma akashiwo did not grow under LL, so an intermediate irradiance of 65 mE m22 s21 was used. Cells were kept in suspension with gentle shaking (100 rpm). Cultures were first acclimated to ambient conditions by making two successive transfers into fresh media, during the exponential growth stage. Each transfer resulted in an abundance of 104 cell ml21. Following a third transfer, cell abundance in each culture was monitored daily, using a hemacytometer (Hauser Scientific, Horsham, Pennsylvania). Subsamples of each culture were withdrawn during mid exponential and stationary growth phases and filtered under gentle vacuum (220 mm Hg) onto 25 mm glass fiber filters (GF/F; 0.7 mm nominal pore size). For H. akashiwo cultures grown at 65 mE m22 s21, samples were taken during stationary growth phase only. FIELD SAMPLES Water samples were collected from the North Inlet (Georgetown, South Carolina) and Murrells Inlet (Murrells Inlet, South Carolina) saltmarsh estuaries (Table 2). Sample bottles were placed in coolers at ambient water temperature and transported immediately to the Baruch Marine Laboratory (Georgetown, South Carolina). Aliquots were either preserved with Lugols iodine (3 ml in 100 ml sample), preserved with 50% glutaraldehyde solution (200 ml in 10 ml sample), or filtered onto GF/Fs for HPLC analyses. Diatoms and dinoflagellates were enumerated from Lugols-preserved samples, using Utermöhl chambers, and examined with an inverted microscope. All other cells were counted by filtering a known volume of glutaraldehyde-fixed water onto 0.4 mm black polycarbonate membrane filters (Osmonics, Inc.) and enumerated by epifluorescent microscopy. Biovolumes were estimated using the geometric shapes and mathematical formulas presented in Hillebrand et al. (1999). A biovolume estimate of 1.3 mm3 was used for Synechococcus spp., based on Kana and Glibert (1987). The effects of preservation on biovolume can be significant and vary with fixative, species, and analytical method (Booth 1987; Verity et al. 1992; Montagnes et al. 1994; Menden-Deuer et al. 2001). We did not attempt to apply correction factors to biovolume because of the great variability of effects on individual species, which can swell or shrink depending on fixative or physiological state. The overall correction factor of Montagnes et al. (1994) for analysis of Lugols-fixed samples by light microscopy (25% shrinkage) and 162 A. J. Lewitus et al. T A B L E 1. Sp ec ie s u se d to d er iv e p ig m en t ra ti o fo r C H E M T A X m at ri x ba se d on 12 ta xo n om ic gr ou p s. C u lt u re co lle ct io n s: C C M P 5 P ro va so liG u ill ar d N at io n al C en te r fo r C u lt u re of M ar in e P h yt op la n kt on , H P 5 H or n P oi n t, an d SC A E L 5 So u th C ar ol in a A lg al E co lo gy L ab or at or y. M A 5 M as sa ch u se tt s, M D 5 M ar yl an d , an d SC 5 So u th C ar ol in a. al lo 5 al lo xa n th in , bu t 5 19 9-b u ta n oy lo xy fu co xa n th in , ch l b 5 ch lo ro p h yl l b, ch l c1 5 ch lo ro p h yl l c 1 , ch l c2 5 ch lo ro p h yl l c 2 , d ia d in o 5 d ia d in ox an th in ,d ia to 5 d ia to xa n th in ,f u co 5 fu co xa n th in ,h ex 5 19 9-h ex an oy lo xy fu co xa n th in ,l u t 5 lu te in ,n eo 5 n eo xa n th in ,p er 5 p er id in in ,p ra si n o 5 p ra si n ox an th in ,v io la 5 vi ol ax an th in ,a n d ze a 5 ze ax an th in .u n d es 5 u n d es cr ib ed . * 5 st ra in s u se d in lig h t-g ro w th p h as e ex p er im en ts (s ee Fi gs . 1 an d 2) . G ro u p C la ss Sp ec ie s So u rc e P ig m en ts 1 5 D ia to m / D in oA B ac ill ar io p h yc ea e B ac ill ar io p h yc ea e B ac ill ar io p h yc ea e B ac ill ar io p h yc ea e D in op h yc ea eA *T ha la ss io si ra cf . m in is cu la H P 91 01 C yl in dr ot he ca cl os te ri um (u n d es st ra in ) N itz sc hi a sp . H P 91 01 C yl in dr ot he ca cl os te ri um (u n d es st ra in ) K ry pt op er id in iu m fo lia ce um C h op ta n k R iv er , M D