Aquat.sci.61 (1999) 307–322 1015-1621/99/040307-16 $ 1.50+0.20/0 © Birkhäuser Verlag, Basel, 1999

Aquatic Sciences

Comparative analysis of Lake Periphyton communities using high performance liquid chromatography (HPLC) and light microscope counts Karl E. Havens 1, *, Alan D. Steinman 1, H.J. Carrick 1, J.William Louda 2, Nancy M. Winfree 2 and Earl W. Baker 2 1

2

South Florida Water Management District, West Palm Beach, Florida 33416-4680, USA Department of Chemistry and Biochemistry, Florida Atlantic University, 777 Glades Road, Boca Raton, Florida 33406, USA

Key words: Periphyton, pigment analysis, HPLC, light microscopy. ABSTRACT High performance liquid chromatography (HPLC) and light microscope counts were used to characterize the taxonomic composition of epiphyton and epipelon at seven locations in a subtropical lake. Both methods indicated that algae were dominated by diatoms and cyanobacteria. However, the methods often gave dramatically different estimates of relative biomass among algal divisions, and there was no consistent pattern of co-variation. Large differences in underwater irradiance may have caused variation in accessory pigment to chlorophyll a ratios, invalidating application of the generic HPLC-based model. In large heterogeneous lakes, it may be necessary to use a suite of models, tailored to site-specific environmental conditions, if HPLC is to be used for evaluation of algal community structure.

Introduction Although periphyton plays an important role in the nutrient cycling and trophic dynamics of lakes (Wetzel, 1996), these communities have received less attention than phytoplankton in the limnological literature (Lowe, 1996). In shallow regions of lakes, sufficient quantities of light may reach the benthos to permit extensive growth of periphyton on plants and the sediment surface (Steinman et al., 1997). These communities can sequester nutrients from the water column, and potentially affect growth rates, competitive interactions, and seasonal succession in the phytoplankton (Sand-Jensen and Borum, 1991). Periphyton and its associated invertebrate fauna also may form the principal base of a lake’s food web (Hecky and Hesslein, 1995). * Corresponding author, e-mail: [email protected].

308

Havens et al.

To fully understand the role of periphyton in a lake ecosystem, it is important to quantify seasonal and spatial variations in taxonomic structure, which typically is done by microscopic counting (Lowe and LaLiberte, 1996). At present, there is no substitute for microscopy when one requires information on species composition and bio-diversity. However, other procedures might suffice if only coarse-scale information (e.g., relative biomass of algal divisions) is needed. An approach that has been widely used in studies of phytoplankton is high performance liquid chromatography (HPLC), whereby pigments are extracted from a natural algal assemblage and quantified using absorbance spectroscopy (Jeffries et al., 1997). To employ this method, pigment signatures of selected unialgal cultures must also be evaluated, to determine ratios of accessory pigments to chlorophyll a for the various algal divisions. A multiple regression equation derived from the pigment data then is used to estimate the contribution of each algal division to the total community biomass. This HPLC method has been used to quantify phytoplankton seasonal succession in a variety of aquatic ecosystems, including lakes (Wilhelm et al., 1991), estuaries (Roy et al., 1996), and rivers (Pinckney et al., 1997). It now is routinely used to quantify phytoplankton taxonomic structure in the open ocean (e.g., Wright et al., 1996). In recent studies of freshwater systems (Pickney et al., 1997, Suzuki et al., 1997), phytoplankton community structure has been evaluated using HPLC alone, without independent confirmation by microscopic counts. This may not always be prudent, as we explain below. The HPLC method has rarely been used in studies of lake periphyton. In a recent survey of periphyton in Lake Okeechobee, Florida, Steinman et al. (1998) used HPLC to elucidate seasonal and spatial patterns in pigment composition. In the present study, we compared the HPLC results with those obtained from microscopic counts, to determine whether HPLC can serve as a tool for routine evaluation of periphyton taxonomic composition. We also considered whether disparities between the two methods could be explained by environmental conditions.

Methods Study site Sampling was conducted both in the pelagic and littoral regions of Lake Okeechobee (surface area 1,800 km2, mean depth < 3 m), located in south Florida, USA, at 27°00¢ N Latitude and 80°50¢ W Longitude. The pelagic region (1,400 km2) of this lake is comprised of: (a) a deeper (up to 5 m), turbid central zone with sediments of unconsolidated mud, phosphorus-rich water, and a low phytoplankton biomass (Olila and Reddy, 1993; Phlips et al., 1997); (b) a shallower (2–3 m) mud-bottomed north zone, which receives large inputs of phosphorus from nearby tributaries and often has phytoplankton blooms (Havens et al., 1994); and (c) sand and peatbottomed south and west zones, located between the pelagic and littoral regions. At the interface of these pelagic and littoral zones, the underwater irradiance is favorable both for phytoplankton and periphyton growth (Phlips et al., 1993). The relatively shallow (<1 m) interior littoral zone (400 km2) is oligotrophic, and supports a

Periphyton analysis by HPLC vs. microscopic counts

309

Figure 1. Map of Lake Okeechobee, showing the locations where periphyton samples were collected during 1994–95. Solid circles indicate sites where both epiphyton and benthic algae were sampled, and open circles indicate sites where only benthic algae were sampled. Shaded area of map indicates a large zone of emergent macrophytes

diverse assemblage of macrophytes with high epiphyton biomass (Steinman et al., 1997). We collected benthic samples from seven locations, including north and central pelagic sites, four pelagic-littoral interface sites, and one interior littoral site (Fig. 1). At four of the sites (POLES, STAKE, FB-OUT and MH-IN), where macrophytes occurred, epiphyton samples also were taken. Collectively, the sampling sites characterized the wide range of limnological conditions that occur in Lake Okeechobee (Table 1). Sample collection Sites were sampled during four seven-day periods in December 1994 and March, June and September 1995. Prior to collecting periphyton, measurements at the top and bottom of the water column were made for temperature, pH, and dissolved oxy-

310

Havens et al.

Table 1. Average limnological conditions at the sampling sites in Lake Okeechobee. Station codes and locations are given in Fig. 1. Data are means from four seasonal observations, with pH values reflecting mean hydrogen ion concentrations. Iz = irradiance at the sediment surface, TP = total phosphorus concentration, TN = total nitrogen concentration Station

Sediment type

Macrophyte type

Depth (m)

Iz (µmol m–2 s–1)

pH

TP (µg L–1)

TN (µg L–1)

MH-IN FB-OUT STAKE POLES SBAY L001 LZ40

Sand/Marl Sand Sand Sand Peat Mud Mud

Eleocharis Scirpus Scirpus Scirpus Vallisneria – none – – none–

1.0 1.7 2.0 2.0 2.3 4.3 5.3

240 26 30 18 33 <1 <1

7.2 8.0 7.7 7.8 8.3 7.6 8.0

8 52 53 72 22 86 113

1,310 1,400 1,420 1,300 1,380 1,270 1,620

gen, using a Hydrolab Surveyor III water analyzer. Instantaneous irradiance measurements were taken from just above the water surface and at the bottom of the water column using a Li-Cor spherical quantum sensor attached to a LiCor LI-1000 data logger. Samples for chemical analyses (total phosphorus and nitrogen) were collected from approximately mid-depth with a Van Dorn bottle. Benthic algae were sampled using a hand-held, 3.8-cm diameter, gravity corer with a clear acrylic collection tube (Davis and Steinman, 1998). This device allowed the sediment surface to be visually inspected while still in the sampler. If the sediments were compact (as in littoral and near-littoral sites), the upper 1 cm was sectioned from the core and placed into a plastic bag, and kept on ice until return to the laboratory. If sediments were soft mud (as in pelagic sites L001 and LZ40), making collection of a discrete section impossible, the entire core sample (approximately 5 cm length) was collected. Analysis of the longer cores introduces potential problems because they may include pigments that have been buried for a longer period of time, and thus may have more degraded pigments. In the present study we believe that this is not a serious problem because the 5 cm cores consisted of fluid mud sediments, which are frequently resuspended by wind (Reddy and Fisher, 1991), and therefore are not prone to long-term burial. Epiphyton was sampled from the dominant macrophytes (Table 1) at the POLES, STAKE, FB-OUT, and MH-IN sites. Three shoots were collected from each site by cutting at the sediment interface, allowing the shoots to float to the water surface, and then collecting them in plastic bags. These samples also were stored on ice until processing. Chemical analyses and periphyton sample preparation Water samples were analyzed using standard methods. Total phosphorus concentrations were determined according to USEPA (1979) and total nitrogen was assayed according to USEPA (1987), in both cases using a flow-injected autoanalyzer. Benthic algae samples were placed into a graduated cylinder, the volume determined, and homogenized for several minutes using a vortex mixer. Sub-samples (10

Periphyton analysis by HPLC vs. microscopic counts

311

to 50 mL) of the material then were removed for independent analyses of pigments by HPLC, and algal species biovolumes by light microscopy. The pigment sub-samples were frozen (–20°C) and the samples for microscopic analysis were placed into opaque plastic bottles and preserved with acid Lugol’s solution. Epiphyton was removed from the host macrophytes by hand, or by gentle scraping with a brush. It was not determined whether 100% of the epiphyton community was collected in this manner. However, because the data were used to compare results of two analytical methods applied to the same sample material, 100% removal was not a critical assumption. The collected epiphyton material was placed into a graduated cylinder, the volume determined, and homogenized as described above. Sub-samples for pigment analysis were filtered onto GF/F filters, placed into plastic bags, and frozen. Sub-samples for microscopic analysis were collected and preserved as described above. HPLC pigment analysis Procedures used for HPLC pigment analysis were described in detail by Steinman et al. (1998), and therefore, only an overview is provided here. Just prior to analysis, frozen benthic samples were thawed to 0–2°C and filtered using GF/F filters to remove excess water. Samples were ground with their filters in 95–100% acetone, under yellow light and at ice-bath temperature, to achieve a 90% acetone extract solution. The solution was centrifuged at 1400 g for 2–3 min, and the supernatant was removed and passed through a 0.45 µm polycarbonate filter. The final extract material was combined with an ion-pairing solution (Mantoura and Llewellyn, 1983) and an internal standard (IS) comprised of copper mesoporphyrin-IX dimethyl ester. The IS allowed for correction of measured pigment quantities based on the ratio of IS added to IS found (calculated from the HPLC absorption integration at 394 nm). The system used for HPLC analysis was comprised of a Thermo-Separations Products 4100 quaternary gradient pump, a Rheodyne 7125 injector with a 100 µl injection loop, a 150 ¥ 3.9 mm, 4 µm column (Waters NovaPak™), and a Waters 990 photodiode array detector, for screening of absorbances from 190 to 800 nm. The reverse-phase HPLC protocol was based on Mantoura and Llewellyn (1983) and Kraay et al. (1992), with modifications described by Steinman et al. (1998). External standards (53 different pigments) used for calibrations were obtained commercially, extracted from uni-algal cultures, or obtained as gifts. Molar ratios of accessory pigments to chlorophyll a were determined from commercially available unialgal cultures. This information was supplemented by published ratios from studies spanning a wide range of marine and freshwater habitats. From the newly measured and published data, the following regression equation was used to estimate relative abundances of the various algal divisions: [CHLA] = 6.7 [MYXO] + 2.0 [CHLB] + 1.2 [FUCO] + 2.8 [ALLO] + 1.7 [PERID]

(1)

where the [MYXO], [CHLB], [FUCO], [ALLO] and [PERID] indicate the molar concentrations of myxoxanthophyll, chlorophyll b, fucoxanthin, alloxanthin, and

312

Havens et al.

peridinin, respectively, in a natural community sample, and the numeric coefficients are molar ratios of pigments to chlorophyll a. Calculated values for each pigment type were then divided by the HPLC-measured molar chlorophyll a concentration [CHLA] to estimate the relative biomass of cyanobacteria, chlorophytes, diatoms, cryptophytes and dinoflagellates, respectively, in the community.

Microscopic counts Periphyton cell abundances and taxonomic composition were determined by light microscopy. After measuring the volume of each preserved epiphyton or epipelon sample, the contents were mixed and 0.1 ml aliquots were transferred into a PalmerMaloney counting chamber. Samples were enumerated to the species level at 450X magnification, by counting the number of cells in random fields. Only cells containing a visible protoplast were counted. Between 300 and 5,000 individual cells were counted in each sample, and the counting error (based on Poisson statistics) was below 4%. Replicate counts gave coefficients of variation below 20%. Cell biovolumes (µm3) were determined by measuring ten randomly chosen individuals of each species, and using formulae for regular geometric solids having shapes similar to the observed cells. Results were converted to population biovolumes using the following formula: Population Biovolume = C ¥ SB ¥ (TA/FA) ¥ (SV/CV)

(2)

where C = number of cells counted, SB = the species cell biovolume (µm3), TA = total area of counting chamber base (mm3), FA = combined area of random fields counted (mm3), SV = sample volume (ml), and CV = aliquot volume (ml). Total biovolumes of the dominant algal divisions were determined by summation of respective species’ biovolumes.

Data analysis and statistics Results from the HPLC analyses were compared with those obtained from microscopic counts using three approaches. First, side-by-side stacked bar graphs were developed, to allow direct comparisons between the relative biomass (or biovolume in the case of counts) of cyanobacteria, chlorophytes, diatoms, cryptophytes, and dinoflagellates, as estimated using the two different approaches. A Wilcoxon ranksum test (SAS, 1993) then was used to evaluate, for each algal division, whether the two methods indicated significantly different relative biomasses. This test was performed separately for the epiphyton and benthic algae. The data also were subjected to nonparametric (Kendall’s) rank correlation analysis, to determine whether estimates of relative biomass of algal divisions from the two methods were correlated. Where the HPLC and microscope-based estimates did not match, we inspected the data to determine whether these instances corresponded to a low total pigment yield (i.e., near or below the lower limit of HPLC detectibility). No such corre-

Periphyton analysis by HPLC vs. microscopic counts

313

spondence was observed. We then considered whether the HPLC regression model over-estimated marker pigments in some communities. This was done by plotting HPLC measured concentrations of chlorophyll a vs. those calculated using equation (2). If the data points fall tightly around a 1:1 line on such a plot, the regression model consistently accounts for the variation in total chlorophyll a. Data having a slope substantially greater or lower than unity indicate that accessory pigment to chlorophyll a ratio(s) are consistently being under- or over-estimated. Widely scattered data indicate that pigment ratios vary from site to site, and that a generic model is not applicable. These considerations still do not address the underlying environmental conditions that may determine whether the two analytical approaches are in agreement. To address this issue, we performed a simple comparison of inter-method error vs. selected environmental parameters (water temperature, irradiance, dissolved oxygen, total phosphorus, total nitrogen, and pH) known to affect the pigment content of algal cells. All parameters were measured near the sediment surface. For each algal sample, a sum of squares error (SSE) estimate was derived, as follows: n

SSE = ∑ (pigmenti – countsi)2

(3)

i=1

Where i refers to the ith algal division, pigmenti is the relative biomass of the ith division determined from the HPLC pigment method, and countsi is the relative biomass of the ith division determined from the microscopic counts. The SSE estimates then were used as the dependent variables in a forward-selection stepwise regression analysis (SAS, 1993). Because our aim was to elucidate possible causes for situations with particularly high or low SSE, we used a relatively high error rate (r = 0.10) for variable entry and removal from the regression model. Results Physical and chemical conditions Mean water depth ranged from 1.0 to 5.3 m, and mean irradiance at the sediment surface ranged from 1 to 240 µmol photons m–2 s–1 (Table 1). All sites had circumneutral pH, and high dissolved oxygen concentration (data not shown). Mean total phosphorus concentration was below 10 µg L–1 at the interior marsh site, ranged from 22 to 72 µg L–1 at the littoral-pelagic interface sites, and exceeded 80 µg L–1 at the deeper pelagic sites. Mean total nitrogen concentration varied little, except that it was higher near mid-lake. Steinman et al. (1997) give more information regarding seasonal patterns of lake water chemistry.

Algal taxonomic structure and dominant pigments Light microscope counts indicated that both the epiphyton and benthic algal communities were dominated by diatoms and cyanobacteria, had lesser amounts of chlorophytes, and had a virtual absence of cryptophytes and dinoflagellates. Acces-

314

Havens et al.

sory pigments were dominated by fucoxanthin, myxoxanthophyll, and periodically, high concentrations of chlorophyll b. A more detailed description of accessory pigments in Lake Okeechobee is provided in Steinman et al. (1998).

Community structure estimates from pigments vs. microscopic counts Among the epiphyte samples (Fig. 2), HPLC and microscope-based estimates of division relative biomass (indexed as biovolume in the case of microscope counts) did not consistently agree. We observed close agreement in only 6 of the 15 samples. At the MH-IN site (Fig. 2A), there was close agreement of results in March, June, and September, but not in December. At the FB-OUT site (Fig. 2B), both methods indicated a diatom dominated community in December, but in March and June, HPLC gave a much higher estimate of chlorophyte relative biomass than the counts. At the STAKE site (Fig. 2C) both methods indicated similar proportions of cyanobacteria, chlorophytes, and diatoms in December and March, but HPLC gave a lower estimate of cyanobacteria relative biomass in June and September. At the POLES site (Fig. 2D), there was close agreement among methods in December and September, but not in March and June.

Figure 2. Relative biomass of algal divisions in epiphyton samples, based on HPLC pigment analysis (P), and relative biovolumes based on microscopic counts (M) at the MH-IN (A), FB-OUT (B), STAKE (C), and POLES (D) sampling sites. Crypto = cryptophytes, Diatom = diatoms, Chloro = chlorophytes, and Cyano = cyanobacteria. Dinoflagellates are not included because their relative biomass was always estimated as zero

Periphyton analysis by HPLC vs. microscopic counts

315

Table 2. Results of Wilcoxon rank-sum test comparing estimates of periphyton division relative biovolumes derived from microscopic counts vs. estimates of relative biomass derived from HPLC pigment analysis. Dinoflagellates and cryptophytes were too rare (most data = 0) to be included in the analysis Sample type

Algal division

S (linear rank)

r > |Z|

Epiphyton (n = 15)

Cyanobacteria Chlorophytes Diatoms

222 259 209

0.67 0.28 0.34

Benthic (n = 26)

Cyanobacteria Chlorophytes Diatoms

769 595 600

0.14 0.06 0.10

Results of the Wilcoxon rank-sum test indicated that overall, there were no significant differences in the estimates of relative epiphyton biomass for cyanobacteria, chlorophytes, or diatoms between the two methods (Table 2). However, these results likely reflect the inconsistent nature of contrasts between the two approaches (e.g., on some occasions HPLC gave higher estimates for cyanobacteria, while on other occasions it gave lower estimates, compared to microscopic counts). These errors tend to “cancel” one another, so that correlation analysis is a more suitable test. Kendall’s nonparametric correlation analysis (Table 3) indicated that there were not significant correlations between the relative biomass of algal divisions estimated from the two methods (all r values > 0.19); i.e., the two analytical methods did not agree in space and time. In the benthic samples (Fig. 3) there generally was a poor agreement between HPLC and microscopic estimates of division-level taxonomic structure. Only 7 of the 26 samples displayed a close agreement between the two methods (MH-IN March; FB-OUT December; POLES June and September; L001 December and March; LZ40 September). There were no apparent seasonal or spatial patterns in the extent of agreement. At the MH-IN site (Fig. 3A), both HPLC and microscopic methods indicated strong diatom dominance in March and September, but they indicated disparate Table 3. Results of Kendall’s nonparametric correlation analysis, considering correlations between microscopic and HPLC estimates of periphyton division relative biovolumes and biomasses. Dinoflagellates and cryptophytes were too rare (most data = 0) to be included in the analysis Sample type

Algal division

Kendall’s t

r

Epiphyton (n=15)

Cyanobacteria Chlorophytes Diatoms

0.23 0.26 0.07

0.27 0.19 0.72

Benthic (n=26)

Cyanobacteria Chlorophytes Diatoms

0.01 0.19 0.08

0.92 0.27 0.59

316

Havens et al.

Figure 3. Relative biomass of algal divisions in benthic samples, based on HPLC pigment analysis (P), and relative biovolumes based on microscopic counts (M) at the MH-IN (A), FBOUT (B), STAKE (C), POLES (D), SBAY (E), L001 (F), and LZ40 (G) sampling sites. Crypto = cryptophytes, Diatom = diatoms, Chloro = chlorophytes, and Cyano = cyanobacteria. Dinoflagellates are not included because their relative biomass was always estimated as zero

Periphyton analysis by HPLC vs. microscopic counts

317

community structures in December and June. At the FB-OUT site (Fig. 3B), both methods indicated an approximately equal proportion of diatoms and cyanobacteria in December, but results for March and June were in disagreement. At the STAKE site (Fig. 3C), the two methods gave different estimates of community structure in December, March and June, but were in general agreement in September. At the POLES site (Fig. 3D), agreement between the methods was observed only in June and September. At the SBAY site (Fig. 3E), there was relatively close agreement between methods in December and March, but not in June and September. The best match between HPLC and microscope results occurred at the L001 site (Fig. 3F). At the mid-lake LZ40 site (Fig. 3G), the methods were in relatively close agreement in December and September, but drastically different in June. In the case of benthic algae, Wilcoxon test results (Table 2) indicated differences between microscopic count and HPLC-based estimates of biomass that were marginally significant for all three taxonomic groups considered (cyanobacteria, chlorophytes, diatoms). Furthermore, Kendall’s correlation coefficients were extremely low (Table 3), with r values always in excess of 0.27. Error analysis Use of HPLC or microscopic counts to quantify algal biomass and community structure depends on a set of assumptions that include the following: (1) microscopic counts include all living cells in the community, but no dead cell remnants, and they provide accurate estimates of biomass; (2) selected marker pigments collectively characterize all of the dominant taxa in the assemblage; (3) ratios of marker pigments to chlorophyll a, derived from studies of laboratory-cultured algae, reflect ratios that occur in the natural community; (4) ratios of marker pigments to chlorophyll a do not vary with season or location; and (5) pigments measured with HPLC are associated with living algal cells. Although we cannot quantitatively assess the first assumption, great care was taken to count only cells with visible protoplasts. Therefore, the inclusion of substantial amounts of dead material in biomass estimates is unlikely. Some cells did appear to be in resting stages, and may have had low cellular pigment contents relative to active cells. This could result in a discrepancy between HPLC and microscope-based estimates of biomass for algal divisions dominated by resting forms. The regression of calculated vs. HPLC measured chlorophyll a was highly significant (r2 = 0.92, r < 0.001), but had a slope of only 0.66 (Fig. 4A). This suggested that the marker pigment to chlorophyll a ratio did not vary dramatically among samples, but that laboratory estimates of ratios for one or more of the pigments generally underestimated those occurring in the natural community. Five epiphyton samples displaying a close match (samples in the lower 30th percentile of the calculated error sums of squares for the 15 samples, as defined previously) between HPLC and microscope techniques fell closer to the 1:1 line than did the remaining samples. The slope of the regression line through those data points was approximately 0.90, indicating a better agreement between marker pigment to chlorophyll a ratios with those derived in the laboratory.

318

Havens et al.

Figure 4. Comparisons of measured HPLC chlorophyll a concentrations vs. chlorophyll a concentrations calculated as the sum of products in equation (1), for the epiphyton (A) and benthic algae (B). Least-squares linear regression models are shown as a solid line, and the dashed line indicates the 1 : 1 relationship. A subset of samples, representing the lower 30th percentile of error sums-ofsquares between microscope and HPLC estimates of periphyton division relative biovolumes and biomasses, are indicated in black

Periphyton analysis by HPLC vs. microscopic counts

319

A significant regression equation could not be derived for the benthic algal data (Fig. 4B). On considering only the eight samples in the lower 30th percentile of error sums of squares calculated for the 26 epiphyton samples, the data points appeared to fall closer to the 1:1 line, but there still was not a significant regression relationship. These results indicate that in samples with large differences between pigment and microscope-based results, variations in accessory pigment to chlorophyll a ratios may have been partially responsible. A stepwise multiple regression analysis was performed to determine whether these differences between methods could be attributed to variations in environmental conditions. The analysis considered water temperature, pH, total phosphorus, total nitrogen, dissolved oxygen, and underwater irradiance measures. No significant regression model could be derived for either the epiphyton or benthic algae.

Discussion In this study, we evaluated whether periphyton community structure in a shallow subtropical lake could be accurately estimated using HPLC pigment signatures (Mantoura and Llewellyn, 1983) and a simple multiple regression approach (Gieskes and Kraay, 1983). Several common attributes were indicated by HPLC and standard light microscope methods. Diatoms and cyanobacteria were found to be dominant in both the epiphyton and benthic algal communities, as observed in studies of other wetland periphyton communities in south Florida (McCormick et al., 1996). There also was agreement among methods that dinoflagellates and cryptophytes were rare or absent, and that chlorophytes sometimes were important contributors to periphyton biomass. However, beyond these general similarities, there were striking disagreements between HPLC and microscope-based estimates of community structure. The discrepancies did not lend themselves to a simple explanation, such as a consistent overestimation of marker pigment to chlorophyll a ratios for a particular algal division. Better agreement between methods generally was found for epiphyton than for benthic algae. In a sense, this corresponds to better results for a simpler community, because the benthic samples contained more detritus, and had contributions to the community from both resident algal taxa and settling phytoplankton (Steinman et al., 1998). Benthic samples, from 5 cm deep cores, also appeared to include a greater percentage of resting cells, based on microscopic counts, than was observed in the epiphyton. These factors, along with the possible presence of pigments not associated with intact algal cells, make interpretation of HPLC results more problematic for benthic communities. Some error may have arisen due to difficulties in estimating relative biomass from counts of preserved algal material comprised of diverse morphologies. However, this sort of error would not explain the many cases where HPLC and microscopic counts gave nearly opposite estimates of community structure. Likewise, the absence of auto-fluorescent counts of pico-plankton may have underestimated the biomass of that group (Carrick and Schelske, 1997), but not in a manner that could explain the mis-match between methods.

320

Havens et al.

Perhaps the major reason that HPLC and microscope-based results did not agree is that accessory pigment to chlorophyll a ratios varied among species within the major divisions, depending on environmental conditions. The HPLC method makes use of a diverse array of accessory pigments, each of which may have a unique physiological role. Thus the ratios of pigments can be expected to change with alterations in the environment. For example, Schofield et al. (1998) documented that the relative proportion of the photosynthetic carotenoid, violaxanthin, to the photoprotective carotenoid, zeaxanthin, decreased during daylight. Deventer and Heckman (1996) found pronounced differences in pigment ratios between two genera of chlorophytes (Scenedesmus and Hematococcus), and between two genera of diatoms (Fragilaria and Thalassiosira). Rosen and Lowe (1984) found that the chlorophyll a content of Cyclotella meneghiniana varied with both light and nutrient deficiency. Nitrogen or phosphorus deficiency suppressed the level of cellular chlorophyll a, while low light conditions enhanced it. These results indicate why broad application of a simple pigment-based regression equation across multiple sites (varying in nutrient content, irradiance, and species composition) may not be a suitable method for estimating periphyton community structure in Lake Okeechobee, or for other heterogeneous benthic communities. The success of HPLC in open-ocean systems (Wright et al., 1996, Jeffrey et al., 1997) likely is a function of a less variable environment, in particular with respect to irradiance. In order to establish a similarly successful HPLC method for lake periphyton, a more complex approach may be called for. One solution might be to determine a suite of ratios between marker pigments and chlorophyll a, using cultures of dominant algae from the natural system, grown under different light, nutrient, and perhaps temperature, regimes. A set of regression models could then be derived, and for any given community sample, the “appropriate” model chosen on the basis of measured environmental conditions. Initially, this would be a rather laborious approach, but once the necessary data were obtained and the new models validated, application would be simple. The use of matrix factorization programs, such as CHEMTAX (Wright et al., 1996), also might be considered, although these also would need to be adjusted to consider spatially variable relationships between pigment ratios and taxonomic structure under different irradiances. A possible complication is in specifying the environmental conditions under which the community developed its pigment complement. In the present study, we found no statistically significant relationship between nutrients, light, or any other environmental variable with the error term describing how well HPLC and countbased community structures matched. It may be that environmental conditions measured at collection time did not reflect conditions that existed during growth and pigment formation in the algal communities. Owing to its shallow depth, the water column of Lake Okeechobee frequently experiences complete mixing, due to winds associated with frontal events or afternoon thunderstorms. These mixing events can transport benthic material into the water column, and dramatically affect the underwater light climate, as well as concentrations of nutrients such as phosphorus (Maceina and Soballe, 1991). The events often are short-lived, with conditions returning to their previous state within hours (Hanlon et al., 1998). It previously has been recommended (Roy et al., 1996) that the best approach for quantifying algal dynamics is to use HPLC and microscopic counts in tandem,

Periphyton analysis by HPLC vs. microscopic counts

321

taking advantages of the unique information that can be provided by each approach. In the present study, HPLC data indicated the presence of photosynthetic purple sulfur bacteria at the littoral site (Steinman et al., 1998). We previously were not aware of their presence, and they would not likely have been detected by standard bacterial counts. On the other hand, subtle changes in species composition within certain algal divisions, the diatoms in particular, are good indicators of early eutrophication trends (McCormick et al., 1996). The HPLC method cannot provide information at this level of detail, no matter how detailed the interpretation of chromatographs. ACKNOWLEDGEMENTS The authors are grateful to Andrew Rodusky, Richard Meeker, Therese East, Patrick Davis, and Bruce Sharfstein for providing support in the field and laboratory. Comments from Susan Gray, Sue Newman, Paul McCormick, David Millie, Barry Rosen, David Rosenberg, and two anonymous reviewers led to improvements in a previous version of the manuscript. REFERENCES Carrick, H.J. and C.L. Schelske, 1997. Have we overlooked the importance of small phytoplankton in productive waters? Limnol. Oceanogr. 42: 1613–1621. Davis, W.P. and A.D. Steinman. 1998, A light-weight, inexpensive benthic core sampler for use in shallow water. J. Freshwat. Ecol. 13: 475–479. Deventer, B. and C.W. Heckman, 1996. Effects of prolonged darkness on the relative pigment content of cultured diatoms and green algae. Aquat. Sci. 58: 241–252. Gieskes, W.W.C. and G.W. Kraay, 1983. Unknown chlorophyll a derivatives in the North Sea and the tropical Atlantic Ocean revealed by HPLC analysis. Limnol. Oceanogr. 28: 757–766. Hanlon, C.G., R.L. Miller and B.F. McPherson, 1998. Relationships between wind velocity and submersed PAR in a shallow lake (Lake Okeechobee, Florida, USA). Water Res. Bull. 34: 951–961. Havens, K.E., C. Hanlon and R.T. James, 1994. Seasonal and spatial variation in algal bloom frequencies in Lake Okeechobee, Florida, USA. Lake Reserv. Manage. 10: 139–148. Hecky, R.E. and R.H. Hesslein, 1995. Contributions of benthic algae to lake food webs as revealed by stable isotope analysis. J. N. Amer. Benthol. Soc. 14: 631–653. Jeffrey, S.W., R.F.C. Mantoura and S.W. Wright, 1997. Phytoplankton pigments in oceanography: guidelines to modern methods. UNESCO, Paris. Kraay, G.W., M. Zapata and M.J.W. Veldhuis, 1992. Separation of chlorophylls c1, c2, and c3 of marine phytoplankton by reversed-phase-C18-high-performance liquid chromatography. J. Phycol. 28: 708–712. Lowe, R., 1996. Periphyton patterns in lakes. Pages 57–76 in R.J. Stevenson, M.L. Bothwell, and R.L. Lowe (editors). Algal ecology: freshwater benthic ecosystems. Academic Press, San Diego, CA. Lowe, R. and LaLiberte, 1996. Benthic stream algae: distribution and structure. Pages 269–294 in F.R. Hauer, and G.A. Lamberti (editors). Methods in stream ecology. Academic Press, San Diego, CA. Maceina, M.J. and D.M. Soballe, 1991. Wind-related limnological variation in Lake Okeechobee, Florida. Lake Reserv. Manage. 6: 93–100. Mantoura, R.F.C. and C.A. Llewellyn, 1983. The rapid determination of algal chlorophyll and carotenoid pigments and their breakdown products in natural waters by reverse-phase highperformance liquid chromatography. Anal. Chem. Acta 151: 297–394. McCormick, P.V., P.S. Rawlik, K. Lurding, E.P. Smith and F.H. Sklar, 1996. Periphyton-water quality relationships along a nutrient gradient in the northern Florida Everglades. J. N. Am. Benthol. Soc. 15: 433–449.

322

Havens et al.

Olila, O.G. and K.R. Reddy, 1993. Phosphorus sorption characteristics of sediments in shallow eutrophic lakes of Florida. Arch. Hydrobiol. 129: 45–65. Phlips, E.J., P.V. Zimba, M.S. Hopson and T.L. Crisman, 1993. Dynamics of the plankton community in submerged plant dominated regions of Lake Okeechobee, Florida. Verh. Int. Ver. Limnol. 25: 423–426. Phlips, E.J., M. Cicha, K.E. Havens, C. Hanlon, S. Badylak, B. Rueter, M. Randall and P. Hansen, 1997. Relationships between phytoplankton dynamics and the availability of light and nutrients in a shallow subtropical lake. J. Plankton Res. 19: 319–342. Pinckney, J.L., D.F. Millie, B.T. Vinyard and H.W. Paerl, 1997. Environmental controls of phytoplankton bloom dynamics in the Neuse River Estuary, North Carolina, USA. Can. J. Fish. Aquat. Sci. 54: 2491–2501. Reddy, K.R. and M.M. Fisher, 1991. Sediment resuspension effects on phosphorus flux across the sediment-water interface. Technical report, South Florida Water Management District, West Palm Beach, Florida, USA. Rosen, B.H. and R.L. Lowe, 1984. Physiological and ultrastructural responses of Cyclotella meneghiniana (Bacillariophyta) to light intensity and nutrient limitation. J. Phycol. 20: 173–183. Roy, S., J.P. Chanut, M. Gosselin and T. Sime-Ngando, 1996. Characterization of phytoplankton communities in the lower St. Lawrence Estuary using HPLC-detected pigments and cell microscopy. Mar. Ecol. Prog. Ser. 142: 55–73. Sand-Jensen, K. and J. Borum, 1991. Interactions among phytoplankton, periphyton, and macrophytes in temperate freshwaters and estuaries. Aq. Botany 41: 137–175. SAS, 1993. SAS/STAT Users Guide (4th edition). SAS Institute Inc., Cary , NC. Schofield, O., T.J. Evans and D.F. Millie, 1998. Photosystem II quantum yields and xanthophyllcycle pigments of the macroalga Sargassum natans (Phaeophyceae): responses under natural sunlight. J. Phycol. 34: 104–112. Steinman, A.D., R.H. Meeker, A.J. Rodusky, W.P. Davis and C.D. McIntire, 1997. Spatial and temporal distribution of algal biomass in a large, subtropical lake. Arch. Hydrobiol. 139: 29–50. Steinman, A.D., K.E. Havens, J.W. Louda, N.M. Winifree and E.W. Baker, 1998. Characterization of the photoautotrophic algal and bacterial communities in a large shallow subtropical lake using HPLC-PDA based pigment analysis. Can. J. Fish. Aquat. Sci. 55: 206–219. Suzuki, K., N. Handa, T. Nishida and C.S. Wong, 1997. Estimation of phytoplankton succession in a fertilized mesocosm during summer using high-performance liquid chromatographic analysis of pigments. J. Exp. Mar. Biol. Ecol. 214: 1–17. USEPA, 1979. Methods for Chemical Analysis of Water and Waste Water. United States Environmental Protection Agency, Washington, DC. USEPA, 1987. Handbook of Methods for Acid Deposition Studies. United States Environmental Protection Agency, Washington, DC. Wetzel, R.G., 1996. Benthic algae and nutrient cycling in lentic freshwater ecosystems. Pages 641– 663 in R.J. Stevenson, M.L. Bothwell and R.L. Lowe (editors). Algal ecology: freshwater benthic ecosystems. Academic Press, San Diego, CA. Wilhelm, C., I. Rudolph and W. Renner, 1991. A quantitative method based on HPLC-aided pigment analysis to monitor structure and dynamics of the phytoplankton assemblage – a study from Lake Meerfelder Maar (Eifel, Germany). Arch. Hydrobiol. 123: 21–35. Wright, S.W., D.P. Thomas, H.J. Marchant, H.W. Higgins, M.D. Mackey and D.J. Mackey, 1996. Analysis of phytoplankton of the Australian sector of the Southern Ocean: comparisons of microscopy and size frequency data with interpretations of pigment HPLC data using the CHEMTAX matrix factorization program. Mar. Ecol. Prog. Ser. 144: 285–298. Received 26 October 1998; revised manuscript accepted 22 June 1999.