Hydrobiologia (2005) 534: 117130

Springer 2005

Depth distribution of photosynthetic pigments and diatoms in the sediments of a microtidal fjord
Angela Wul1,*, Sirje Vilbaste2 & Jaak Truu3
Department of Marine Ecology, Marine Botany, Goteborg University, P.O. Box 461, SE 405 30 Goteborg, Sweden Institute of Zoology and Botany, Estonian Agricultural University, Riia 181, 51014 Tartu, Estonia 3 Environmental Protection Institute, Estonian Agricultural University, Kreutzwaldi 5, 51014 Tartu, Estonia (*Author for correspondence: E-mail: angela.wul@marbot.gu.se)
2 1

Received 8 December 2003; in revised form 8 July 2004; accepted 15 July 2004

Key words: benthic, diatoms, photosynthetic pigments, sediment, HPLC

Abstract The depth distribution of photosynthetic pigments and benthic marine diatoms was investigated in late spring at three dierent sites on the Swedish west coast. At each site, sediment cores were taken at six depths (735 m) by scuba divers. It was hypothesized that (1) living benthic diatoms constitute a substantial part of the benthic microora even at depths where the light levels are <1% of the surface irradiance, and (2) the changing light environment along the depth gradient will be reected in (a) the composition of diatom assemblages, and (b) dierent pigment ratios. Sediment microalgal communities were analysed using epiuorescence microscopy (to study live cells), light microscopy and scanning electron microscopy (diatom preparations), and HPLC (photosynthetic pigments). Pigments were calculated as concentrations (mg m)2) and as ratios relative to chlorophyll a. Hypothesis (1) was accepted. At 20 m, the irradiance was 0.2% of surface irradiance and at 7 m, 1%. Living (epiuorescent) benthic diatoms were found down to 20 m at all sites. The cell counts corroborated the diatom pigment concentrations, decreasing with depth from 7 to 25 m, levelling out between 25 and 35 m. There were signicant positive correlations between chlorophyll a and living (epiuorescent) benthic diatoms and between the diatom pigment fucoxanthin and chlorophyll a. Hypothesis (2) was only partly accepted because it could not be shown that light was the main environmental factor. A principal component analysis on diatom species showed that pelagic forms characterized the deeper locations (2535 m), and epipelicepipsammic taxa the shallower sites (720 m). Redundancy analyses showed a signicant relationship between diatom taxa and environmental factors temperature, salinity, and light intensities explained 57% of diatom taxa variations.

Introduction Benthic microalgae play an important role in the carbon budget of coastal ecosystems, a fact that has been known for several decades (e.g. Teal, e & Hegeman, 1974); and in estuaries, 1962; Cade microphytobenthic communities can account for a substantial part (50%) of the total primary productivity (Underwood & Kromkamp, 1999). Furthermore, subtidal benthic microalgae on the

continental shelves can account for 42% of the total areal primary productivity (Nelson et al., 1999). Although microbenthic communities have gained increased attention (reected for example, in special sessions at international conferences), most studies have been conducted on tidal ats or in non-tidal shallow-water habitats (water depth <5 m). Very little, however, is known about the depth distribution of these communities, especially in the marine habitat (reviewed in Cahoon, 1999;

118 Totti, 2003). Under favourable conditions, benthic microalgae often form cohesive microbial mats on the surface of shallow-water sediments (Stal & Caumette, 1994; Sundba ck et al., 1996a), and diatom-dominated mats are the most common type of microbial mats in marine northern-temperate areas. The present study was inspired by observations of a brown diatom mat at 17 m depth while scuba diving on the Swedish west coast. The mat consisted mainly of the diatoms Gyrosigma sp., Pleurosigma sp., Haslea sp., and Nitzschia cf. sigmoidea (200300 lm). Gullmar Fjord has been extensively studied for more than 100 years, through studies of bottom stratigraphy, chemical components, macroalgae, foraminifera, and macrofauna. Little attention has so far been paid to the microphytobenthic part of the community. Because benthic diatoms are known to withstand low light conditions (Stevenson & Stoermer, 1981; Sundba ck & Jo nsson, 1988; Cahoon, 1999), and some species even tolerate anoxia (Admiraal et al., 1982), it was hypothesized that (1) living benthic diatoms constitute a substantial part of the benthic ora even at depths where the light levels <1% of the surface irradiance, and (2) the changing light environment along the depth gradient will be reected in (a) the composition of diatom assemblages, and (b) different pigment ratios. (i.d. 8.7 mm). Temperature and salinity were measured at each sampling depth. At each sampling site, a light (PAR) gradient (1 m resolution) was measured under clear sky conditions at noon using a quantum scalar irradiance meter. At each sampling depth, 50 ml of seawater was taken for analysis of inorganic nutrients (Table 1). At site A, only three depths were analysed. Inorganic nutrients (NO3 + NO2, PO4, and Si(OH)4) were analysed on a TRAACS 800 autoanalyser (Braun & Lubbe). Photosynthetic pigments Two samples (i.d. 8.7 mm) were taken from each core (2 4 subsamples from each depth, giving a total of 144 pigment samples), frozen in liquid nitrogen, transferred to a )80 C freezer, and stored up to 2 months before analysis. For extraction, 100% methanol was added to the samples. Extraction proceeded at )18 C for 48 h, and the samples were ultrasonicated and ltered (0.5 lm). The vials were kept on a cooled autosampler until analysis, which occurred within 12 h. Pigments were analysed by HPLC (Wright & Jeffrey, 1997) using a diode-array detector connected in series to a uorescence detector (both Spectraphysics). Absorbance detection was 436 nm. The uorescence detector was used to conrm the identity of chlorophyll degradation products, as some of them co-elute and interfere with carotenoids. Chl a, chl c1c2, fucoxanthin, and diadinoxanthin were quantied as mg l)1 according to Wright & Jerey (1997) and converted to mg m)2. Epiuorescing microalgae Two randomly chosen cores from each depth were subsampled, and two subsamples (i.d. 8.7 mm) from both cores (i.e. 3 sites 6 depths 2 cores 2 subsamples) were diluted with glutaraldehyde (nal concentration 2.5%) to keep the chloroplasts intact until analysed. Algal cells were detached from sediment particles through ultrasonication. The sample was shaken for 30 s; after ca. 5 s (to allow sand grains to settle) several individual subsamples (40 ll) of suspensions were pipetted onto a microscope slide, and using an epiuorescence microscope (500 magnication), epiuorescing cells were counted.

Materials and methods Study site and collection of material The study was carried out 813 May, 1997, in Gullmar Fjord, on the Swedish west coast (58 15 N, 11 27 E). The area has a maximum tidal amplitude of 2030 cm. The weather was calm and sunny on all sampling days. Sediment cores were sampled by scuba diving without disturbing the sediment surface. Three sites were sampled ca. 5 km apart. At each site, four sediment cores (i.d. 90 mm) were randomly taken at six dierent depths: 7, 15, 20, 25, 30, and 35 m, giving a total number of 72 cores. Because the aim was to study the depth distribution of benthic diatoms at low light conditions, no samples were taken at water depths <7 m. Subsamples were taken of the upper 5 mm of sediment using cut-o plastic syringes

119
Table 1. Dierent abiotic parameters at the dierent sites and depths Site Site A 1m 5m 7m 15 m 20 m 25 m 30 m 35 m Site B 1m 5m 7m 15 m 20 m 25 m 30 m 35 m Site C 1m 5m 7m 15 m 20 m 25 m 30 m 35 m 37 7.4 4.4 0.8 0.4 0.2 0.1 0 0.99 0.52 0.44 0.32 0.27 0.25 0.21 5 4.5 3 3 3 2.5 27 31 34 34 35 35 1.7 1.2 2.2 2.4 3.8 3.9 0.1 0.1 0.2 0.2 0.2 0.3 7.8 3.3 3.2 3.1 4.5 4.2 14.8 1.9 1.2 0.2 0.2 0.1 0.05 0 1.91 0.79 0.64 0.41 0.31 0.27 0.25 6 3 2 3 2 2 29 30 33 35 35 35 7.2 6.3 0.3 0.2 6.4 5.0 3.5 0.2 2.6 Light (%) Kd (m)1) Temp. (C) Salinity NO3+NO2 (lM) PO4 (lM) Si(OH)4 (lM)

16.6 2.4 1.2 0.2 0.2 0.1 0.05 0

1.80 0.75 0.63 0.41 0.31 0.27 0.25 5 4.5 2 2 2 2 27 27 29 33 34 33 0.5 0.8 2.0 4.1 4.9 7.2 0.2 0.1 0.1 0.2 0.3 0.4 3.1 3.1 4.1 4.1 4.4 6.6

Light measurements are presented as percentages of surface PAR irradiance.

Diatom preparation One sample (i.d. 8.7 mm) was taken from each core (four subsamples from each depth) and diluted with glutaraldehyde as above. Three of these four subsamples were randomly chosen, washed from the glutaraldehyde and treated with hot sulphuric acid (Vilbaste et al., 2000). The diatom suspension was pipetted onto ethanol-cleaned cover slips that were left to dry in the air. Naphrax (refractive index 1.7) was used as mounting medium. Diatom composition and diversity were examined at four depths (7, 15, 20, 25 m) at each site. On each slide, a minimum of 500 valves were identied and counted along a transect using interference microscopy (Zeiss Axiovert 135) using a 100 oil immersion

objective. The basic counting unit was a single valve, a complete frustule being counted as two units. The absolute numbers of the counted taxa were converted to relative abundance (RA, %). Scanning electron microscopy (SEM) was used to check the identity of dicult small taxa. Statistical analyses Multivariate analysis was used to identify the main gradients in the diatom community composition using the program CANOCO 3.1 (ter Braak, 1994; ter Braak & Verdonschot, 1995) and ADE-4 (Thioulouse et al., 1997). In order to study the relationship between diatom species and environmental factors, a redundancy analysis (RDA) was

120 carried out. RDA can be described as a multiple regression followed by a PCA on the tted values. The arc sine square root transformed values of the RA of taxa (the mean of three replicate cores) were used in the multivariate analysis. If a taxon was not present in at least four samples out of 36 with a numerical RA of at least 1% in a single sample, it was excluded. As a result, 106 taxa were included in the analyses. The following environmental factors were used in the RDA as explanatory variables: water temperature, salinity, and light (Table 1). The signicance of correlations between RDA axes and environmental variables as well as canonical coecients were assessed on the bases of approximate t-test values (t>|2.1|). The eect of sampling location as a grouping factor was estimated using the multivariate randomization test (Manly, 1997). The relationships between the diatom taxa and the pigment data table were dec & analysed with co-inertia analysis (Dole Chessel, 1994); the method allows one to relate two data tables and extract information common to both tables. The pigment data and the epiuorescing cells data were analysed by two-factor ANOVA with depth and place as the independent variables, n 4. Dierences were accepted as signicant at p < 0.05. Cochrans test was used to check homogeneity of variances and data with heterogeneous variances were transformed according to Underwood (1997). the sampling period, the weather conditions were stable and sunny with a clear sky. The surface PAR irradiance was around 1600 lmol photons m)2 s)1 at all sites and on all sampling occasions. The light penetration depth was similar at sites A and C, with ca. 15% of the incoming radiation left at 1 m depth, and the 1% level was reached at 7 m (Table 1). At site B, light penetrated deeper, and the 1% level was reached at 14 m. The attenuation coecients (Kd; m)1) (Table 1) for sites A and C were 0.25 m)1, and for site B, 0.21 m)1 (030 m). In general, the concentrations of all inorganic nutrients analysed increased with depth (Table 1). The purpose of the inorganic nutrient analysis was to nd a tendency of dierences between sites; since no dierences were found it will not be further discussed. Photosynthetic pigments The pigments were analysed as concentrations (mg m)2) and ratios relative to chl a. At all sites chl a showed a pattern of a signicant decrease with increasing depth (Fig. 1). These patterns agreed well with those found for microalgal counts, and the highest signicant correlation between chl a and epiuorescent cells was found for benthic diatoms (r 0.970.99, site A, B, p < 0.05) and for benthic plus pelagic diatoms (r 0.84, site C, p < 0.05). The chl a concentrations varied between 34 and 42 mg m)2 at 7 m, decreased to 1720 mg m)2 at 20 m; and at 25 35 m depth, the amounts varied between 3.6 and
50 Chlorophyll a, mg m-2 40 30 20 10 0 7 15 20 25 Depth, m 30 35 Site A Site B Site C

Results Study area Sampling sites A and C had similar bathymetry, whereas the slope was considerably steeper at site B. At all sites, the sediment appeared more occulent at depths >20 m, probably due to accumulation of organic material. A clearly visible brown mat, indicating the presence of diatoms, was found down to 7 m depth, and a mat-like structure was observed down to ca. 17 m. Mollusc shells were present at all depths. Salinity increased with depth down to 25 m and levelled out between 25 and 35 m (salinity 3435). The temperature decreased from 5 to 6 C at 7 m, levelling out at 20 m (23 C) (Table 1). During

Figure 1. Chlorophyll a concentrations in surface sediment samples (05 mm) from dierent sites and water depths. Symbols represent mean values of four replicate cores SE.

121 7.1 mg m)2 (Fig. 1). The concentrations of chlorophyll breakdown products showed a patchy distribution, and no signicant dierences between depths were found. For example, small peaks of pheophytin a and ca. six dierent peaks of pheophorbides or pheophorbide-like pigments were usually observed, especially at depths deeper than 15 m. The concentrations of the diatom pigments chl c1c2, fucoxanthin, and diadinoxanthin all decreased signicantly with increasing depth, but the magnitude of declines varied between sites (Fig. 2). The overall best correlation between pigments was found between fucoxanthin and chl a (r 0.95 0.99, p < 0.05). The carotenoid betacarotene showed a patchy distribution, and no signicant dierences between depths were found. It should be noted, however, that betacarotene was often mixed with breakdown products. Other pigments found were lutein, zeaxanthin, and chl b. Generally, lutein and chl b correlated very well, but

Figure 2. Concentrations (mg m)2) and pigment ratios (relative to chlorophyll a) (weight:weight) of dierent pigments in surface sediment samples (05 mm) from dierent sites and water depths. Symbols represent mean values of four replicate cores SE.

122 zeaxanthin diered, indicating a zeaxanthin source other than green algae. Both lutein and zeaxanthin were patchily distributed and decreased with depth. For pigment ratios relative to chl a, the outcome diered from the distribution pattern of the concentrations. Many pigment ratios varied between depths with no apparent decrease (Fig. 2). A signicant increase in chlorophyll degradation products with depth (ratios to chl a) was observed at all sites (not shown). When comparing ratios of the diatom pigments fucoxanthin and diadinoxanthin to chl a, only fucoxanthin showed a statistically signicant depth eect (Fig. 2). Microalgae The epiuoresence counting of cells was done mainly to corroborate the pigment analysis and to give an idea of what sediment taxa were present as live cells (i.e. with intact epiuorescing chloroplasts). Sites A and B closely resembled each other, with a maximum number of sediment diatoms at 7 m, 5293 106 cells m)2, decreasing to 27 31 106 cells m)2 at 20 m depth, levelling out from 25 m (Fig. 3). Site C diered, with similar cell numbers down to 20 m depth, decreasing to 25 m and levelling out (Fig. 3). Note, that cell numbers from deeper than 20 m are based on few counts (<50 cells), and these results should be considered with caution. Dominating taxa belonged to the genera Amphora, Nitzschia, Gyrosigma, Pleurosigma, Navicula, Cocconeis, Achnanthes, and Diploneis. In addition, planktonic cells were present in the sediment and unidentied centric taxa. Single coccoid cyanobacteria were found at all sites at depths shallower than 20 m. Unidentied cysts were rare but occasionally seen at all sites. The proportion of epiuorescing planktonic species was on average 17% for the depth range 720 m, and 53% for 2535 m. At 25 m at site C, however, the proportion of planktonic cells was 24%. Dominance and diversity of diatom species Altogether 198 diatom taxa were recorded from the sediment samples. Almost 45% of all counted cells were planktonic species. The dominating taxa (mean RA >3%) were Skeletonema costatum (pelagic and benthic; mean RA 34%), Achnanthes cf. delicatula (epipsammic; mean RA 4.0%), Opephora olsenii (epipsammic and epiphytic; mean RA 3.8%), Martyana atomus (epipsammic; mean RA 3.3%), and Aulacoseira spp. (pelagic; mean RA 3.1%). Skeletonema costatum is generally considered a pelagic species but it thrives in the benthic habitat for periods long enough not to be considered only a consequence of sedimentation following a pelagic bloom. Therefore we have chosen to treat S. costatum as a pelagic species in the statistical analyses but we discuss the possibility of it also being benthic. There were some other taxa that occurred in almost all samples and sporadically showed rather high RA values (>5%). Among them were representatives from dierent habitats: Achnanthes delicatula (epipsammic and epilithic), Fallacia cryptolyra (epipelic), Navicula gregaria (epipelic and epilithic), Navicula perminuta (epilithic), and Chaetoceros spp. (pelagic). The relative abundance of the nine commonest diatom taxa along the depth gradient is shown in Figure 4. The RA of Achnanthes delicatula, Martyana atomus, Opephora olsenii, Navicula gregaria, and N. perminuta was negatively related to depth; in contrast with the pelagic taxa Aulacoseira spp., Chaetoceros spp., and Skeletonema costatum, whose presence increased with increasing depth. An epipelic diatom, Fallacia cryptolyra, showed an interesting feature in this respect the RA of it was relatively low at 7 m, achieved the maximum at 15 m, and then started to decrease again (Fig. 4). The ShannonWeaver diversity index (H) was high (mean 4.38), varying between 2.57 and 5.39, and correlated (r 0.84, p < 0.01) with species

6 -2 Cell number *10 m

120 100 80 60 40 20 0 7 m 15 m 20 m 25 m 30 m 35 m Depth, m Site A Site B Site C

Figure 3. Cell numbers of uorescing benthic diatoms in surface sediment samples (05 mm) from dierent sites and water depths. Symbols represent mean values of two replicate cores SE.

123

Figure 4. Relative abundance (RA the mean values of 9 cores SE) of nine commonest taxa at four dierent depths shown, (a) epipsammic, (b) epipelic or epilithic, (c) pelagic taxa. Achnanthes delicatula and A. cf. delicatula are treated as one taxon. Note: RA scale is dierent in dierent graphs.

richness, which varied from 51 to 90, and evenness (J) (r 0.98, p < 0.01), which ranged from 0.45 to 0.85 (Fig. 5). The median diversity of the diatom community was lower in site A compared with B and C. The diatom community showed a signicant lower diversity with increasing depth (r )0.61, p < 0.01).

Multivariate analyses The PCA of pigment data was performed as a correlation matrix, because values for fucoxanthin and chl a were dominating and together made up ca. 90% of total pigment weight. These two pigments, together with diadinoxanthin and zeaxan-

124 to dierences between sites. The multivariate randomization test indicated signicant dierences between three sampling locations ( p < 0.01) as well as between sampling depth ( p < 0.001). The dierence between sampling locations was bigger in comparisons involving shallower depths. The ordination of the RA of 106 diatom species by PCA arranged the sites along gradients of depth and sampling area (Fig. 7). The rst axis were related to the depth gradient and explains 43% of the variation; the second axis reects the dierences between the three sampling sites. The later was not statistically signicant. Again, as for the PCA of pigments, the dierences between sampling areas were larger at shallower depths. The most important species for dening the rst PCA axis was S. costatum it characterized the deeper location. For shallow sites, the typical taxa were Achnanthes cf. delicatula, Martyana atomus, and Achnanthes sp. (all epipsammic taxa). In the ordination based on the RDA (Fig. 8), the length and direction of the variable arrows indicate their importance and their approximate correlation to the ordination axis, respectively. There was a signicant relationship between diatom data and environmental variables, wherein temperature, salinity, and light explained 49.9% of

Figure 5. Dierences in ShannonWeaver diversity index (H) values shown as boxplots by (a) location and (b) depth.

thin, explained most of the data variation in the PCA (Fig. 6). The rst axis of the PCA reects the depth gradient, and the second axis can be related
C15m

0.4 -0.5 -0.4 0.5

F2
1

1.6 -2.1 2.9 -0.8

C20m

A7
C25m F1 (42.6%) C7m
4

B7m

B15m

B20m B25m

A25
3

A15

A20 B25 8 C15 B20

F1
C7
F2 (15.5%) A15m

A20m

A25m

B15

B7

A7m

C20

Figure 6. Position of sampling location according to PCA analysis of pigment data. Each location is represented by a centroid of four replicate measurements. Lines join samples taken along the same transect (depth gradient).

Figure 7. Ordination of diatom samples based on PCA of the relative abundance of 106 diatom taxa. Lines join samples taken along the same transect (depth gradient).

125 important variables contributing signicantly to the RDA axes were salinity (r )0.85, t )3.1) and light (r 0.68, t 3.2). The rst RDA axis was statistically signicant (p < 0.01). The eigenvalues of the rst three axes were 0.352, 0.104, and 0.043, respectively. Four distribution patterns of taxa can be discerned from the RDA. (1) The species with scores situated in the left-hand part near the salinity arrow of the ordination plot (Fig. 8) were more abundant in deeper water. Odontella aurita, Cocconeis costata, Thalassionema nitzschiodes, and Rhabdonema arcuatum are all marine species. (2) The species with scores in the upper right quadrant of the diagram prefer lower salinity and are brackish species (e.g. Navicula phyllepta, Nitzschia dissipata, Pinnularia quadratarea). (3) The diatoms recorded near the arrow for light occurred mainly in shallow water where the light conditions were better. These are typical epipsammic taxa, such as Achnanthes sp., Opephora olsenii, and Navicula germanopolonica, and epipelic ones like Staurophora salina and Amphora staurophora. (4) The taxa situated on the upper left-hand corner of the graph (negatively correlated with light) are mostly planktonic forms (Cyclostephanus dubius, Chaetoceros spp., Thalassiosira spp., Detonula sp.). According to co-inertia analysis (explained as a correlation between two PCA analyses), the relationship between diatom taxa and pigment data was signicant ( p < 0.01) (not shown). The dominant trend was related to depth. Epipelic and epipsammic diatom species Achnanthes sp., Amphora abludens, A. luciae, A. staurophora, and Staurophora salina were positively correlated with e.g. fucoxanthin and chl a. Pelagic taxa such as Chaetoceros spp., a small centric diatom, Detonula sp., and Thalassiosira angulata had a negative relationship with these pigments. In fact, most of the later named taxa were dead or inactive (cysts) on the bottom.

Figure 8. Ordination biplot based on RDA of diatom species data with respect to light measurements, water temperature (Temp), and salinity. The displayed species are selected on the basis that more than 30% of their variance is accounted for by the diagram. Achnanthes cf. delicatula (ACH CFDE); Achnanthes lemmermanni (ACH LEMM) Achnanthes sp. (ACH SP); Amphora beaufortiana (AMP BEAU); Amphora copulata (AMP COPU); Amphora staurophora (AMP STAU); Aulacoseira spp. (AUL SP); Berkeleya spp. (BER SP); Biremis lucens (BIR LUCE); Chaetoceros spp. + cysts (CHA SPP); Centric sp. (CEN SP); Cocconeis costata (COC COST); Cyclostephanus dubius (CYCS DUB); Cylindrotheca closterium (CYL CLOS); Detonula sp. (DET SP); Diatoma moniliformis (DIA MONI); Diploneis smithii (DIP SMIT); Entomoneis sp.( ENT SP); Fallacia cryptolyra (FAL CRYP); Fallacia forcipata (FAL FORC); Fallacia litoricola (FAL LITO); Fallacia tenera (FAL TENE); Gomphonema parvulum (GOM PARV); Gyrosigma peisonis (GYR PEIS); Gyrosigma tenuissima (GYR TENU); Navicula bipustulata (NAV BIPU); Navicula capitata (NAV CAPI); Navicula clamans (NAV CLAM); Navicula directa (NAV DIRE); Navicula cf. fauta (NAV FAUT); Navicula anatica (NAV FLAN); Navicula germanopolonica (NAV GERM); Navicula gregaria (NAV GREG); Navicula phyllepta (NAV PHYL); Navicula portnova (NAV PORT); Navicula ramosissima (NAV RAMO); Nitzschia dissipata (NIT DISS); Nitzschia cf. distans (NIT DIST); Nitzschia hybrida (NIT HYBR); Odontella aurita (ODO AURI); Opephora olsenii (OPE OLSE); Pinnularia quadratarea (PIN QUAD); Pleurosigma aestuarii (PLE AEST); Psammodictyon panduriforme (PSA PAND); Rhabdonema arcuatum (RHA ARCU); Staurophora salina (STA SALI); Tabularia fasciculata (TABU FAS); Tabularia investiens (TABU INV); Thalassionema nitzschoides (THAL NIT); Thalassiosira angulata (THA ANGU); Thalassiosira eccentrica (THA ECCE).

Discussion Our experimental design was well replicated and successful on a spatial scale but can only be considered a snapshot in time. However, the abiotic values fall within the range found by others in the area, and therefore we believe our results can be

diatom taxa variations. The location of species scores relative the arrows indicate the environmental preferences of each species. The two most

126 considered representative for the season in the area studied. We like to mention some methodological considerations. Small epipsammic species were found in the diatom analysis, but very few were found when counting epiuorescent cells, and it is likely they were overlooked or did not detach from the sand grains. The chl a concentration per cell further support this conclusion, because chl a per cell was up to 10 times too high, even for a shadeadapted community. In dead cells, chl a degrades over hours (Cahoon et al., 1994), and it is not likely the chl a concentrations were overestimated. Because we found these small cells at all sites and depths we do not expect them to have made an impact on the statistical analyses. Furthermore, it can be argued that the sediment sampling depth (5 mm) was too deep, because the euphotic zone in a sandy sediment usually is <3 mm (MacIntyre et al., 1996). However, as pointed out by MacIntyre & Cullen (1995), photosynthetically competent diatoms are present well below the sediment light penetration depth. In contrast to the turbid water column in tidal areas during high tide, in non-tidal or microtidal areas the water column can be clear all day, and there is a good potential for a photosynthetically active microphytobenthic community. However, occasionally the water column also in non-tidal waters can be turbid due to river discharge during long periods of rain. Although the weather conditions during this study, and 2 weeks preceding, were stable and without rainfall, the higher attenuation coecients found in the upper water layer might be explained by river discharge. This is further indicated by the dierent salinities among the sites. In a study in the same area (Site C) by Engelsen (unpublished), the salinity was 17, 23, 27, and 33 at depths 1, 5, 10, and 15 m. At a nearby site in the fjord, salinity has been reported to be between 15 and 28 in the upper meter during May, 19951998. Hypothesis 1 Intact epiuorescing cells were found at depths well below the euphotic zone, and therefore Hypothesis 1 can be accepted. Further support for accepting the hypothesis was obtained from the pigment analysis. Our assumption that fucoxanthin and chl a mostly originated from living benthic diatoms was supported by the co-inertia analyses, where fucoxanthin and chl a were positively related to typical epipelicepipsammic diatoms, and the opposite was found for pelagic taxa. The chl a concentrations at 7 m depth (3442 mg m)2) were well within the range previously found for shallowwater sandy sediments (<1 m) on the Swedish west coast (Wul et al., 1997; Odmark et al., 1998; Wul et al., 1999), when using the same HPLC technique as in the present study. In this study, it is likely that chl a concentrations (as well as the other photosynthetic pigments) from 25 m depths and deeper represent background values pigments in dierent stages of degradation, originating from sedimented planktonic species or possibly from microalgal resting stages (cf. Cahoon et al., 1994). The low numbers of epiuorescent cells as well as the relative dominance of cysts and planktonic species support this assumption. One can argue that such a background value also exists at shallower depths, but if the chl a concentrations from 2535 m depth are subtracted from pigment concentrations at 720 m depth the chl a concentrations are still within the range 837 mg m)2; furthermore, the cell counts conrmed a high proportion of epiuorescing benthic diatoms. Although pelagic forms constituted ca. 1045% of total counts (diatom species), the dominating taxon was Skeletonema costatum, a pelagic species that thrives in the benthic habitat for periods long enough not to be considered only a consequence of sedimentation following a pelagic bloom (Sundba ck & Jo nsson, 1988; Yap, 1991; Sundba ck et al., 1996b; Sundba ck & Miles, 2002). For example, Sundba ck & Jo nsson (1988) observed, a bloom of the spring bloom diatom S. costatum was induced on the sediment surface. Furthermore, 49% of the total cell numbers at 1416 m depth were S. costatum. At shallower depths (25 m), however, only 5% was attributed to this species or other centric diatoms. These ndings agree well with our results showing that epiuorescing cells of S. costatum were observed down to 20 m depth. In another study, a 20% contribution to total microalgal biomass was attributed to S. costatum at 4 m depth in April, and it was suggested that this species contributes to the benthic primary productivity (Sundba ck et al., 1996b).

127 Despite the absence of data on planktonic species from the Gullmar Fjord, data from adjacent areas and fjords show that in May 1997, the planktonic ora was dominated by dinoagellates; and diatoms were abundant at only one site of eight. In early April, however, S. costatum was the most abundant species, but planktonic algal biomass was low at this time. In 1997, an unusual planktonic bloom occurred during winter along the Swedish west coast, and again S. costatum was the dominating species. Based on these data, we conclude that the S. costatum that was found in the upper 5 mm of sediment originated from a planktonic bloom 13 months prior to this study, a time period long enough to assume that the epiuorescing cells of S. costatum must have continued to photosynthesize there. The observation that diatoms can grow (survive) under very low light intensities (single lmol photons m)2 s)1) is an ability already described by others, both from laboratory work (Peters, 1996; Jochem, 1999) and in situ (reviewed in Cahoon, 1999). A preference for mid-depths was found along a depth gradient (627 m) in Lake Michigan (Stevenson & Stoermer, 1981), where diatom abundance was found to be greater at 9 and 15 m, compared with shallower depths. The diatom preference for mid-depths was probably due to less physical disturbance such as wave action (Stevenson & Stoermer, 1981), which in the present study might be applicable to site C, since a small marina is located nearby. A similar trend was found on the Swedish west coast, with higher biomass (measured as chl a) found in the depth interval 1416 m, compared with shallower depths (Sundba ck & Jo nsson, 1988). In our study, a negative correlation (r )0.61) was observed between the diversity and depth (725 m). So far, we have concluded that benthic diatoms have the capability to survive and contribute to the benthic ora. But are they active primary producers? On the basis of two recent studies at Site C, the answer is yes. In the rst study, Engelsen (unpublished) measured in situ primary productivity (14C-incubations) and concluded that, at 10 and 15 m depth, the primary productivity was between 0.23 and 0.09 mg C m)2 h)1 (May, 1.40.3 lmol photons m)2 s)1) and from 1.32 to 0.7 mg C m)2 h)1 (September, 4018 lmol photons m)2 s)1). The cores were incubated in situ for 2.53.5 h. In the second study, Sundba ck et al. (2004) found microbenthic oxygen production down to 15 m depth (maximum depth studied). Hypothesis 2 Hypothesis 2a (the changing light environment along the depth gradient will be reected in the composition of diatom assemblages) is accepted. A changed diatom assemblage with depth was found, however, it could not be related to light only. Of the nine most common diatom taxa, the epipelic taxa were negatively correlated with depth, and the opposite was found for the pelagic taxa. Thus, we have a diatom assemblage dominated by epipelic taxa at depths 720 m, and deeper down the pelagic inuence increases (also supported by the proportion of epiuorescent planktonic cells). We assume that the pelagic taxa, apart from S. costatum, make only a minor contribution to the benthic primary productivity, because they epiuoresce for a short period when sedimented down from the water column. In addition, Barranguet et al. (1996) concluded that, in deeper bottom waters, phytoplankton production could be absent due to light limitation, but microphytobenthos could still be productive. We cannot conclude that light was the most important environmental factor, since salinity was found to be equally important in the RDA. Furthermore, among the 15 most abundant epipelic species in a muddy estuary on the east coast of the UK, Underwood et al. (1998) identied Navicula phyllepta, N. gregaria, and Nitzschia dissipata, three species observed also in this study. The two Navicula species were found to be positively correlated to nutrients and negatively correlated to salinity (Underwood et al., 1998). In our study, no sediment nutrients were measured, but a tendency of decreasing abundance of these species with increased salinity was found. According to the RDA, some species seemed more aected by light than salinity, and those were typically epipsammic epipelic species. Pelagic species were negatively correlated to light, and several were found dead or inactive (cysts) on the bottom. The primary productivity data from Engelsen (unpublished) supports our original theory that light was the main environmental variable; more specically, they found that primary productivity was positively correlated to light intensity.

128 Hypothesis 2b (the changing light environment along the depth gradient will be reected in different pigment ratios) holds for the fucoxanthin:chl a ratio. The fucoxanthin:chl a ratio increased, with a simultaneous decrease in fucoxanthin concentrations to 20 m depth, indicating a photoacclimation for benthic diatoms. A ratio of 0.7 was found in a shade-adapted diatom mat (Sundba ck et al., 1996a), and this value was also typical of those reported from a diatom-dominated tidal at (in spring) in SW Netherlands (Barranguet et al., 1997); in our study, we found similar values also. Engelsen (unpublished) found that the microphytobenthos from 1 to 15 m depth was shade adapted, because the microalgae required a lower light intensity to reach a light-saturated photosynthetic rate. Thus, one can argue that the community was shade-adapted already at 7 m, and a change in the ratio should have reected not a photoacclimation but a larger proportion of diatoms compared with other algal groups. The use of pigment ratios in natural communities is complicated by the fact that changes in pigment ratios can reect not only the acclimation to environmental conditions but can also reect a change in community composition (cf. Goericke & Montoya, 1998). Theoretically, other fucoxanthin-containing microalgae such as Dinophyceae, Chrysophyceae, and Haptophyceae could have contributed to the amount of fucoxanthin present in the sediment. However, of these groups, only members of Dinophyceae were likely to be present, and furthermore, the benthic dinoagellate taxa usually found in the experimental areas contain not fucoxanthin but peridinin as the major carotenoid (Wul unpublished). Therefore, the fucoxanthin:chl a ratio should have reected a physiological adaptation of the benthic diatoms. In addition, fucoxanthin is among the most labile carotenoids, with a half-life of a few days if oxygen and light are present (Leavitt & Carpenter, 1990; Leavitt, 1993), and is therefore not usually preserved in the sediment. This implies that in the present study, old fucoxanthin should not have interfered with pigments from living cells. The occurrence of breakdown products or unidentied pigments complicated quantication of carotenoids, and we could not exclude the possibility that, in some cases, an overestimation occurred. A terrestrial origin of typical green algal pigments (chl b, lutein, and zeaxanthin) is not likely, because these pigments usually degrade before deposition in the sediment (Leavitt, 1993). Furthermore, the sampling took place in late spring where a terrestrial input was unlikely. The occurrence of several dierent peaks of pheophorbides or pheophorbide-like pigments is typical for marine sediments and reported from several studies where they suggested to originate from grazing activities (Cariou-Le Gall & Blanchard, 1995; Brotas & Plante-Cuny, 1998). Similar chl-like pigments have been observed also in lakes (Leavitt & Carpenter, 1990). In our study, the scuba divers observed several small burrows; and small macrofauna, e.g. snails and polychaetes, were occasionally observed in the sediment cores. The occurrence of several chloropigments stresses the fact that, for pigment measurements in sediments, it is advantageous to use a uorescence detector in connection with the absorbance detector in order to better identify chloropigments and avoid interference from the non-uorescent carotenoids. Although dierences between sites were found, they were not statistically signicant in any variable other than pigments, where the multivariate randomization test showed a statistically signicant dierence between the three sampling locations. In the ANOVA, however, no statistically signicant dierences were found. We therefore conclude that our sampling design was successful in allowing us to draw general conclusions about the inner part of the fjord. In conclusion, we have shown that benthic diatoms constitute a substantial part of the microbenthic ora and are active (epiuorescing) under very low light conditions (single lmol photons m)2 s)1) in situ. Our ndings stress the fact that, when using pigment analysis in sediments, it is not reliable to assume that benthic microalgal production can be excluded below the euphotic zone (1% of surface irradiance).

Acknowledgements We thank K. Wul for diving assistance and the landowner W. Nilsson for letting us use his property. This work was nancially supported by the hlstro funds: Birgit och Birger Wa m, the Swedish

129 Institute, Kapten Carl Stenholm, UddenbergNordingska, Per Adolf Larsson, Kungliga och Hvitfeldtska, and the SYS-RESOURCE Program at The Natural History Museum, London. They are all greatly acknowledged for their support.
low-water benthos. Marine Ecology Progress Series 122: 227237. MacIntyre, H. L., R. J. Geider & D. C. Miller, 1996. Microphytobenthos: The ecological role of the secret garden of unvegetated, shallow-water marine habitats. I. Distribution, abundance and primary production. Estuaries 19: 186201. Manly, B. F. J., 1997. Distance matrices and spatial data. In Randomization, Bootstrap and Monte Carlo Methods in Biology, Chapman and Hall, London. Nelson, J., J. Eckman, C. Robertson, R. Marinelli & R. Jahnke, 1999. Benthic microalgal biomass and irradiance at the sea oor on the continental shelf of the South Atlantic Bight: spatial and temporal variability and storm eects. Continental Shelf Research 19: 477505. . Wa Odmark, S., A. Wul, S.-A ngberg, C. Nilsson & K. Sundba ck, 1998. Eects of UV-B radiation in a microbenthic community of a marine shallow-water sandy sediment. Marine Biology 132: 335345. Peters, E., 1996. Prolonged darkness and diatom mortality: II. Marine temperate species. Journal of Experimental Marine Biology and Ecology 207: 4358. Stal, L. J. & P. Caumette, 1994. Microbial Mats. Structure, Development and Environmental Signicance. NATO ASI Series G: Ecological Sciences, Vol. 35. Springer-Verlag, Berlin. Stevenson, R. J. & E. F. Stoermer, 1981. Quantitative dierences between benthic algal communities along a depth gradient in Lake Michigan. Journal of Phycology 17: 2936. Sundba ck, K. & B. Jo nsson, 1988. Microphytobenthic productivity and biomass in sublittoral sediments of a stratied bay, southeastern Kattegat. Journal of Experimental Marine Biology and Ecology 122: 6381. Sundba ck, K. & A. Miles, 2002. Role of microphytobenthos and denitrication for nutrient turnover in embayments with oating macroalgal mats: a spring situation. Aquatic Microbial Ecology 30: 91101. Sundba ck, K., L. Carlson, C. Nilsson, B. Jo nsson, A. Wul & S. Odmark, 1996a. Response of benthic microbial mats to drifting green algal mats. Aquatic Microbial Ecology 10: 195208. Sundba ck, K., P. Nilsson, C. Nilsson & B. Jo nsson, 1996b. Balance between autotrophic and heterotrophic components and processes in microbenthic communities of sandy sediments: a eld study. Estuarine and Coastal Shelf Science 43: 689706. Sundba ck, K., F. Linares, F. Larson, A. Wul & A. Engelsen, 2004. Benthic nitrogen uxes along a depth gradient in a microtidal fjord: the role of denitrication and microphytobenthos. Limnology and Oceanography 49: 10951107. Teal, J. M., 1962. Energy ow in the salt marsh ecosystem of Georgia. Ecology 43: 614624. ter Braak, C. J. F., 1994. Canonical community ordination. Part 1. Basic theory and linear methods. Ecoscience 1: 127140. ter Braak, C. J. F. & P. F. M. Verdonschot, 1995. Canonical correspondence analysis and related multivariate methods in aquatic ecology. Aquatic Sciences 57: 255289. dec & J. M. Olivier, 1997. Thioulouse, J., D. Chessel, S. Dole ADE-4: a multivariate analysis and graphical display software. Statistics and Computing 7: 7583.

References
Admiraal, W., H. Peletier & H. Zomer, 1982. Observations and experiments on the population dynamics of epipelic diatoms from an estuarine mudat. Estuarine Coastal and Shelf Sciences 14: 471487. Barranguet, C., M. R. Plante-Cuny & E. Alivon, 1996. Microphytobenthos production in the Gulf of Fos, French Mediterranean coast. Hydrobiologia 333: 181193. Barranguet, C., P. M. J. Herman & J. J. Sinke, 1997. Microphytobenthos biomass and community composition studied by pigment biomarkers: importance and fate in the carbon cycle of the tidal at. Journal of Sea Research 38: 5970. Brotas, V. & M. Plante-Cuny, 1998. Spatial and temporal patterns of microphytobenthic taxa of estuarine tidal ats in the Tagus Estuary (Portugal) using pigment analyses by HPLC. Marine Ecology Progress Series 171: 4357. e, G. C. & J. Hegeman, 1974. Primary production of the Cade benthic microora living on tidal ats in the Dutch Wadden Sea. Netherlands Journal of Sea Research 8: 260291. Cahoon, L., 1999. The role of benthic microalgae in neritic ecosystems. Oceanography and Marine Biology: an Annual Review 37: 4786. Cahoon, L., R. Laws & C. Thomas, 1994. Viable diatoms and chlorophyll a in continental-slope sediments o Cape Hatteras, North Carolina. Deep Sea Research II 41: 767782. Cariou-Le Gall, V. & G. F. Blanchard, 1995. Monthly HPLC measurements of pigment concentration from an intertidal ron Bay, France. Marine muddy sediment of Marennes-Ole Ecology Progress Series 121: 171179. dec, S. & D. Chessel, 1994. Co-inertia analysis: an alterDole native method for studying speciesenvironment relationships. Freshwater Biology 31: 277294. Goericke, R. & J. P. Montoya, 1998. Estimating the contribution of microalgal taxa to chlorophyll a in the eld-variations of pigment ratios under nutrient- and light-limited growth. Marine Ecology Progress Series 169: 97112. Jochem, F. J., 1999. Dark survival strategies in marine phytoplankton assessed by cytometric measurement of metabolic activity with uorescein diacetate. Marine Biology 135: 721 728. Leavitt, P. R., 1993. A review of factors that regulate carotenoid and chlorophyll deposition and fossil pigment abundance. Journal of Paleolimnology 9: 109127. Leavitt, P. R. & S. R. Carpenter, 1990. Aphotic pigment degradation in the hypolimnion: implications for sedimentation studies and paleolimnology. Limnology and Oceanography 35: 520534. MacIntyre, H. L. & J. Cullen, 1995. Fine-scale vertical resolution of chlorophyll and photosynthetic parameters in shal-

130
Totti, C., 2003. Inuence of the plume of the River Po on the distribution of subtidal microphytobenthos in the northern Adriatic Sea. Botanica Marina 46: 161178. Underwood, A. J., 1997. Experiments in Ecology. Cambridge University Press, Cambridge. Underwood, G. J. C. & J. Kromkamp, 1999. Primary production by phytoplankton and microphytobenthos in estuaries. Advanced Ecological Research 29: 93153. Underwood, G., J. Phillips & K. Saunders, 1998. Distribution of estaurine benthic diatom species along salinity and nutrient gradients. European Journal of Phycology 33: 173183. Vilbaste, S., K. Sundba ck, C. Nilsson & J. Truu, 2000. Distribution of benthic diatoms in the littoral zone of the Gulf of Riga, the Baltic Sea. European Journal of Phycology 35: 373385. Wright, S.W. & S. W. Jerey, 1997. High-resolution HPLC system for chlorophylls and carotenoids of marine phytoplankton. In Jerey, S. W., R. F. C. Mantoura & S. W. Wright (eds), Phytoplankton Pigments in Oceanography. UNESCO Publishing, Paris. Wul, A., K. Sundba ck, C. Nilsson, B. Jo nsson & L. Carlson, 1997. Eect of sediment load on the microbenthic community of a shallow-water sandy sediment. Estuaries 20: 547 558. . Wa Wul, A., C. Nilsson, K. Sundba ck, S.-A ngberg & S. Odmark, 1999. UV eects on microbenthos a four-month eld experiment. Aquatic Microbial Ecology 19: 269278. Yap, H. T., 1991. Benthic energy dynamics in a southern Baltic ecosystem. Marine Biology 108: 477484.