Diversity of planktonic photoautotrophic microorganisms along a salinity gradient as depicted by microscopy, flow cytometry, pigment analysis and DNA-based methods

Abstract

The diversity of prokaryotic and eukaryotic phytoplankton was studied along a gradient of salinity in the solar salterns of Bras del Port in Santa Pola (Alacant, Spain) using different community descriptors. Chlorophyll a, HPLC pigment composition, flow cytometrically-determined picoplankton concentration, taxonomic composition of phytoplankton (based on optical microscopy) and genetic fingerprint patterns of 16S (cyanobacteria- and chloroplast-specific primers) and 18S rRNA genes were determined for samples from ponds with salinities ranging from 4% to 37%. Both morphological and genetical descriptors of taxonomic composition showed a good agreement and indicated a major discontinuity at salinities between 15% and 22%. The number of classes and the Shannon diversity index corresponding to the different descriptors showed a consistent decreasing trend with increasing salinity. The results indicate a selective effect of extremely high salinities on phytoplanktonic assemblages.

Diversity, Solar salterns, Plankton, Pigments, Genetic fingerprinting

1 Introduction

Diversity, which can be defined as the richness of biological elements – such as genes, species or genera – in a community, has been linked to a number of ecosystem processes. However, in spite of much theoretical and empirical work, the relationships between different ecosystem properties such as diversity or productivity continue to be open to debate. For example, Lehman and Tilman [1] and Hector et al. [2] have suggested that increased diversity leads to increased productivity, while others [3,4] have challenged this conclusion. Another important open question is to what extent biotic factors, such as predation or competition, or abiotic factors such as habitat harshness, heterogeneity or size, control diversity in each particular system. Some theoretical work [5,6] has suggested that interactions within a community may lead to fluctuations in species abundances. King et al. [7] have shown that differences in species composition in vernal pools in California resulted from physico-chemical differences in the habitat. Moreover, Therriault and Kolasa [8] studied 49 coastal pools in Jamaica and concluded that much of the species richness was determined by the abiotic pool conditions either directly or indirectly (after modulation by biotic interactions), while biotic factors appeared to be more important in controlling species population densities. Based on general ecological theory, authors like Frontier [9] have suggested that an extreme environment could be expected to be less diverse.

As pointed out by Margalef [10], any attempt to quantify diversity needs to take into account both the spatial structure and temporal dynamics of ecosystems. For example, species richness in a sample of a particular ecosystem depends, among other factors, on the method used to examine the organisms and on sample size. In fact, the curvature of the relationship between species richness and sample size conveys important information on the ecosystem structure. This dependence of measurements on the chosen spatial and temporal scales hinders testing of hypotheses on the factors controlling biodiversity and its relationships with other ecosystem properties. In this context, the microbial communities inhabiting solar salterns offer attractive possibilities. Solar salterns comprise typically a gradient of environments with salinities ranging from that of seawater to sodium chloride saturation or even beyond. Thus, in a limited space, salterns may present conditions [11,12] ranging from those of one of the most common habitats in the planet (seawater) to one of the most extreme environments on earth (brines). The relatively small extension of salterns limits the number and scale of sources of variability to be considered. In addition, temperature and salinity conditions of each location are kept relatively constant at time scales of weeks, due to the mode of functioning of these systems, in which water evaporation in the high salinity ponds is compensated with less saline water from the sea or neighbouring ponds, fed by pumping or gravity.

The different ponds in the salterns provide habitats for many planktonic prokaryotic and eukaryotic protists. The autotrophic microbes (including prokaryotes such as cyanobacteria and eukaryotes like diatoms and other algae) found in hypersaline environments, have often been the subject of taxonomic and physiological research [11–16]. However, although the existence of a decreasing trend in the number of microbial species as salinity increases has been reported [11,17,18], there is a lack of quantitative attempts to test this relationship. One of the underlying problems is the need for robust measures of microbial diversity. Quantification of this parameter requires the grouping of individual elements into non-overlapping classes, according to a consistent classification criterion [19,20]. The use of species as the basic unit of grouping is in principle desirable, but in practice it is difficult to apply to microorganisms because special preparations or even culture techniques are required. As a consequence, studies of the diversity of microbial autotrophs have been generally based on morphological analyses leading to a combination of taxonomic identification and morphotypic categories. A new approach was introduced by Li [21], who applied diversity concepts to phytoplankton categorized by flow cytometry measurements related to size and pigment content. HPLC pigment analyses and more recently, molecular genetics techniques, have provided additional cultivation-independent methods for determining diversity, but there are very few examples of their parallel use in the same community [22].

The aims of this paper are: (1) to quantify the relationships between the salinity gradient and the diversity of phototrophic microorganisms, using an array of methods including morphology by microscopic observations, pigment composition by HPLC, flow cytometry-identified populations and DNA-based approaches (genetic fingerprinting based on the prokaryote and eukaryote small subunit (SSU) rDNA gene sequences), (2) to test whether measures of diversity based on different properties of the same group (like pigments or morphology for autotrophic microbes) offer consistent results, and (3) to determine whether the components of the microbial community described by the different diversity estimates (e.g. microalgal morphotypes, flow cytometric populations, prokaryotic and eukaryotic microorganisms represented by DGGE bands) show similar or different patterns of diversity variation. The work was part of a series of joint experiments, carried out between 17 and 28 May 1999, in the “Bras del Port” salterns in Santa Pola, Alacant, Spain (38°12′N, 0°36′W).

2 Materials and methods

2.1 Study site

The Bras del Port salterns, devoted to year-round commercial salt extraction, consist of over 100 shallow ponds (depth < 1 m). Research dealing with different aspects of the microbial food web and the composition of the bacterial and archaeal populations in the salterns has been published in [11,23–26]. The work described in this paper is based on two “salinity gradient” surveys carried out, respectively, on 18 and 26 May, 1999. Water samples for the surveys were taken with a plastic bucket, from eight to nine ponds with salinities ranging from 4% to 37% (the last one is called crystallizer). Salinities were determined with a hand-held refractometer [11]. Information dealing with the molecular biodiversity of microbes and trophic relationships in the salterns, during the study period, can be found in [27–32].

2.2 Pigment determinations

Pigment analyses were carried out by HPLC, according to the method of Wright et al. [33]. Water samples were kept in the dark, filtered through glass fibre filters (Advantec GF 75, Toyo Roshi Kaisha, Japan) and stored frozen in liquid nitrogen generally within three hours of sampling. Filters were subsequently transferred to 3 ml acetone, sonicated on ice for 15 min, and left to extract for 24 h at 4° C prior to filtering (0.2 μm) 1 ml extract into HPLC-vials and mixing with 300 μl water. HPLC analyses were performed on a Shimadzu LC 10A system with a Supelcosil C18 column (250 × 4.6 mm, 5 μm). Pigments were identified by retention times and absorption spectra identical to those of authentic standards, and quantified against standards purchased from DHI Water & Environment, Hørsholm, Denmark.

The contribution of different algal groups was estimated using the CHEMTAX program [34]. The pigment composition of Dunaliella cf. salina was calculated using pigment data from samples with different concentrations of Dunaliella. The resulting ratios were introduced into the initial pigment ratio matrix of the program, which was taken from Henriksen et al. [35]. Given that these initial ratios reflect mainly major trends in pigment composition, the results of CHEMTAX should be considered with caution when dealing with minor contributions. The CHEMTAX calculations were carried out to allow comparison with inverted microscopy counts. The number of algal classes present, as derived from CHEMTAX, was used in the diversity estimates. Other diversity indices (see below) were calculated directly from the different pigment concentrations.

2.3 Phytoplankton composition

The abundance of nano and microplankton was determined by the inverted microscope technique [36]. Samples of 100 ml of water were placed in Pyrex bottles and fixed immediately with 0.4% final concentration of formaldehyde neutralized with hexamethylenetetramine [37]. For microscopic observation, volumes of 10 ml were introduced in sedimentation chambers and allowed to settle for at least 24 h. The whole bottom of the chamber was examined to count the larger and less abundant organisms. Cells were assigned to the lowest possible taxonomic category. However, in many cases it was only feasible to classify the organisms into morphotypes. This happened, in particular, for flagellates and cyanobacteria (Table 1). All dinoflagellates were included, although some of them are heterotrophic or mixotrophic. Dunaliella spp. includes at least a large (Dunaliella cf. salina, approximately 16 μm × 12 μm) and a small form (9 μm × 5 μm). In the case of filamentous cyanobacteria, the total length of filaments of each morphotype was transformed into cell number by assuming an arbitrary cell length of 10 μm. It must be noted that the inverted microscope method is not adequate for picoplankton-sized organisms and that many forms degrade rapidly in fixed samples. For consistency, the term “phytoplankton” will be used to designate the set of organisms included in the inverted microscope counts, although some of them, like heterotrophic flagellates, may not be considered as “phytoplankton” sensu stricto.

1

Phytoplankton taxa and morphotypes identified by optical microscopy

1	Amphidinium sp.
2	Gymnodinium sanguineum(=Akashiwo sanguinea)
3	Oxyrrhis marina
4	Pentapharsodinium tyrrhenicum
5	Prorocentrum lima
6	Prorocentrum scutellum
7	Prorocentrum triestinum
8	Scrippsiella “trochoidea-like”, small
10	Scrippsiella sp.
11	Unidentified dinoflagellates, small
12	Unidentified dinoflagellates, large
13	Cysts? A
14	Cysts? B
15	Amphora sp.
16	Amphora coffaeformis
17	Nitzschia cf. closterium
18	Gyrosigma sp., large
19	Gyrosigma sp., small
20	Nitzschia“sigma-like”, large
21	Unidentified pennate diatoms A, small
22	Unidentified pennate diatoms B, small
23	Unidentified pennate diatoms, large
24	Dictyocha fibula
25	Aphanothece spp.
26	Spirulina spp.
27	Cyanobacteria, unicells < 2 μm diam.
28	Cyanobacteria, unicells >3 μm diam.
29	Cyanobacteria, unicells >6 μm diam.
30	Cyanobacteria, rectangular unicells
31	Cyanobacteria, filaments >20 μm diam. in 10 μm cell eq.
32	Cyanobacteria, filaments < 15 μm diam. in 10 μm cell eq.
33	Cyanobacteria, filaments, rectang. cells
34	Cyanobacteria, filaments, orange color
35	Chroococcales, rounded
36	Chroccocales, elongated
37	Unidentified flagellates, small
38	Green flagellates
39	Unidentified flagellates, large
40	Cryptophyceae
41	Leucocryptos?
42	Haptophyceae
43	Dunaliella cf. salina
44	Dunaliella“viridis”, small
45	Mesodinium sp.

2.4 Flow cytometry

Samples for flow cytometry were fixed with paraformaldehyde and glutaraldehyde (1%+ 0.05% final concentrations), deep frozen in liquid nitrogen and later stored at −80° C. In the laboratory, 1 μm yellow-green latex Polysciences beads were added to 0.4 ml subsamples as an internal standard and run in a Becton & Dickinson flow cytometer FACScalibur bench machine with a laser emitting at 488 nm. The subsamples were diluted 2× to 4× to reduce effective salinity and prevent optical problems related to the difference of salinities between the sheath fluid (MilliQ water) and the sample, although we did not use the values of forward scatter which are those most affected by density differences. Samples were run at the highest possible speed (around 60 μl min⁻¹) and 15,000 events were acquired in log mode. Abundances were calculated by the ratiometric method from the known amount of added beads, calibrated daily against TrueCount (Becton & Dickinson) beads. We differentiated several algal populations by their FL2 (orange fluorescence) vs. FL3 (red fluorescence) and by their SSC (side scatter) vs. FL3 signatures. For example, cells with a SSC similar to that of the beads, and with similar FL2 and FL3 signatures were considered to be Synechococcus following standard procedures [38,39].

2.5 Genetic fingerprinting

Genetic diversity determinations were carried out for the samples obtained from the different ponds on the survey of 18 May. The molecular methodology used in this study was based on denaturing gradient gel electrophoresis (DGGE) separation of 16S and 18S rRNA gene segments amplified by PCR. A detailed account of the PCR-fingerprinting procedures used can be found elsewhere [22,28,40]. Briefly, microbial biomass was collected from different water volumes (see Table 1 in [28]) using a peristaltic pump in 0.2 μm Sterivex filters (Millipore Corp., Bedford, MA) and DNA was further extracted with proteinase K, SDS and phenol–chloroform–isoamylalcohol, according to the ICM protocol detailed in [28]. Purified DNA was amplified by PCR using two primer combinations and protocols. Thus, fragments of the 16S rRNA gene suitable for subsequent DGGE-analysis were obtained with one primer combination specific for oxygenic phototrophs (CYA359 forward-GC clamp: 5′-CGC CCG CCG CGC CCC GCG CCC GTC CCG CCG CCC CCG CCC G- GGG GAA TYT TCC GCA ATG GG-3′, and CYA781 reverse: 5′-GAC TAC T/A GG GGT ATC TAA TCC C A/T T T-3′) targeting the 16S rRNA genes of cyanobacteria and algal chloroplasts [22]. In addition, some 16S rRNA gene bands were excised from the gel and sequenced as reported [40]. Sequences were submitted to BLAST (http://www.ncbi.nlm.nih.gov) for a first phylogenetic affiliation. Nucleotide sequence accession numbers at EMBL are AJ580966 to AJ580973.

The second primer combination targeted 18S rRNA eukaryotic genes and the genetic fingerprints were obtained from a former work (Fig. 3 in [28]). DGGE gels were stained with a solution of GelStar (1:5000 dilution; FMC BioProducts) and the band patterns were visualized under UV radiation with the Fluor-S MultiImager (Bio-Rad) and the Multi-Analyst software (Bio-Rad). High resolution digitized images were processed with the Diversity Database (Bio-Rad) software. The program carried out a density profile through each lane, detected the bands, and calculated the relative contribution of each band to the total band signal in the lane. A band was defined as a stain signal whose intensity was more than 0.2% of the total intensity for each lane.

3

Composition of the phytoplankton assemblage as determined by inverted microscopy, along the salinity gradient on 18 May 1999 (a) and 26 May 1999 (b).

2.6 Estimates of diversity

The number of classes (S_x) and their relative abundance was determined for each of the community descriptors (x) considered (x= M, light microscopy phytoplankton; x= F, flow cytometry picoplankton; x= P, HPLC pigments; x= 16S or 18S, DGGE bands for oxygenic phototrophs and eukaryotes, respectively).

In the case of light microscopy and flow cytometric counts, the abundance of each taxon or population was given as cell numbers per unit volume. For the HPLC analyses, the descriptors were the different pigments detected and diversity was calculated from their concentrations (μg l⁻¹). Numbers and proportional abundances of rRNA genes were estimated from the denaturing electrophoresis gels using the Multi-Analyst image analysis software facilities. We were aware of biases of PCR-based methods and we were cautious in the number of PCR cycles carried out and in using the same amount of template in each reaction. The samples that we compared were all run in the same PCR and analyzed in the same DGGE gel. Therefore, any biases should have been the same for all samples and the comparison would still be valid. The 16S rRNA fingerprints were taken as representative of the genetic diversity of cyanobacteria and plastids from algae. The 18S rRNA fingerprints are characteristic of eukaryotes, both autotrophic and heterotrophic.

The Shannon diversity index (D_x) was calculated for all descriptors [41] as:

where S is the number of classes, p_i is the relative abundance of class i (∑p_i= 1) and x stands for M, F, P 16S or 18S, depending on the descriptor considered.

In addition, the index K_M= logS_M/logN[42], where S_M is the number of taxa and N the total number of individuals, was also calculated for the inverted microscopy phytoplankton counts. This index is less influenced by the distribution of abundances of the different taxa (or evenness).

3 Results

3.1 Pigment analyses and CHEMTAX results

Chlorophyll a (Chl a) concentration (Fig. 1) presented maxima in the 8% pond and in the 37% crystallizers. Between the first and second surveys, Chl a concentrations changed significantly in some of the ponds, but the overall pattern of variation of chlorophyll with salinity remained similar.

1

Chlorophyll a concentration determined by HPLC analysis in the ponds sampled on 18 and 26 May 1999.

The HPLC analyses detected up to 13 different pigments, including chlorophylls a, b and c. β-Carotene was a major component of the highest salinity ponds, in which Dunaliella cf. salina was dominant (Table 2). The contribution of major phytoplankton groups to total Chl a, as derived from CHEMTAX, is shown in Fig. 2. There were marked differences between surveys in the contribution of some groups, but the general patterns were similar. Dunaliella spp. was dominant at the highest salinities, including the crystallizers, during both surveys, and also at the 31.6% pond on 26 May. Other chlorophytes were found at all salinities below 37% (first survey) or 31.6% (second survey). Dinoflagellates and cryptophytes were present between 4% and 8–11% salinity and diatoms were not present at salinities above 22.4% (disregarding the minor contribution at 31.6%, during the second survey). Cyanobacteria were only important in the 31.6% ponds and to a much lesser extent in the 8% pond.

2

Pigment composition, in μg l⁻¹, along the salinity gradient, as determined from HPLC analyses

Survey date	Salinity (%)	Chl c	Per	Fuc	Neo	Pra	Vio	Diadino	All	Lut	Zea	Chl b	Chl a	α-Carotene	β-Carotene
18/05/03	4	0.763	0.121	0.224	0.110	0.202	0.128	0.065	0.880	0.048	0	0. 855	4.064	0.137	0.289
	5.4	2.052	0.980	0.234	0	0	0.228	0.630	1.214	0.042	0	0.046	5.498	0. 064	0.181
	8	1.155	0.412	0.852	0.194	0	1	1.102	0.237	0.646	0.382	1.180	8.411	0	0. 588
	11	0.058	0	0.283	0.037	0	0	0.156	0	0.140	0.076	0.176	1.714	0	0.117
	15	0.200	0	0.515	0	0	0	0.422	0	0	0	0	2.008	0	0.102
	22.4	0	0	0	0	0	0	0	0	0	0	0	0.531	0.206	0.877
	31.6	0	0	0	0	0	0	0	0	0.219	0.561	0	2.374	1.193	17.38
	37	0	0	0	0	0	0	0	0	1.230	6.657	1.805	18.80	15.68	508.5
26/05/03	4	0.755	0.087	0.294	0.036	0	0.018	0.078	0.650	0.025	0	0.176	3. 220	0.118	0.152
	5.4	2.719	1.198	0.788	0	0	0.088	1.044	0.865	0.060	0	0.063	7.644	0. 119	0.316
	8	1.330	0.432	0.183	0.202	0	1.258	0.652	1.332	0.664	0.303	1.468	8. 595	0.168	0.464
	11	0.909	0.191	0.190	0.128	0	0.982	0.561	0.241	0.886	0.139	1.366	7. 982	0.022	0.381
	15	0.139	0	0.377	0	0	0	0.144	0	0	0.043	0	1.766	0	0.362
	22.4	0	0	0.181	0	0	0	0	0	0	0	0	3.492	0.150	0.870
	25	0	0	0	0	0	0	0	0	0	0	0	5.350	0.388	2.319
	31.6	0	0	0	0	0	0	0	0	0.540	2.847	0	9.410	4.684	182.0
	35.6	0	0	0	0	0	0	0	0	0.909	3.909	1.626	13.57	21.38	363.4

Chl c, chlorophyll c; Per, peridinin; Fuc, fucoxanthin; Neo, neoxanthin; Pra, prasinoxanthin; Vio, violaxanthin; Dia, diadinoxanthin; All, alloxanthin; Lut, lutein; Zea, zeaxanthin; Chl b, chlorophyll b; Chl a, chlorophyll. Diadinoxanthin and α-carotene were not included in the CHEMTAX calculations.

2

Contribution of different algal classes to total chlorophyll a along the salinity gradient, on 18 May 1999 (a) and 26 May 1999 (b), as determined by application of the CHEMTAX programme to the HPLC analyses.

3.2 Morphotypic composition of autotrophs

The distribution of total phytoplankton numbers (Fig. 3), derived from microscopic counts, was comparable to that of Chl a, with maxima around 8% salinity and in the crystallizers. A list of the identified taxa and morphotypes is given in Table 1. Dunaliella cf. salina was found at salinities of 25% and higher; on 18 May, its population reached 24,000 cells ml⁻¹ in one of the crystallizers, but decreased to 5300 cells ml⁻¹ in the second survey. Other important contributors to the phytoplankton community were filamentous and chroococcal cyanobacteria across all the salinity range, diatoms and dinoflagellates up to 15% and 11% salinities, respectively, and cryptophytes in the 5% pond on 18 May and from the 5% to the 11% ponds on 26 May. Mesodinium sp., an autotrophic ciliate with endosymbiotic cryptophyte chloroplasts, was present with 67 cells ml⁻¹ in the 4% pond, on 18 May, although most cells appeared to be in bad shape. The dinoflagellate assemblage of the 4, 5 and 8% ponds included Pentapharsodinium tyrrhenicum, Prorocentrum scutellum, Prorocentrum lima, and concentrations up to 88 cells ml⁻¹ of Gymnodinium sanguineum (=Akashiwo sanguinea, according to [43]), a typical red tide species. The most abundant diatoms were small pennates, Amphora coffaeformis, and several species of Nitzschia, including a large form (N. cf. sigma) which reached concentrations of 100 cells ml⁻¹ at 8% salinity on 18 May. The cyanobacteria included Aphanothece, Spirulina, and non-identified morphotypes of Chrooccocales (unicellular) and Oscillatoriales (filamentous).

3.3 Flow cytometry

The flow cytometric analyses of the survey samples allowed the detection of 16 distinct populations, characterized by their pigment and size signatures. Three of them contained phycobilins (orange fluorescence) and we considered that consisted of two cyanobacterial (example, population #A in Fig. 4) and one cryptomonad-like population (population #B in Fig. 4). One of the presumed cyanobacteria (#A) resembled Synechococcus and the other was composed of larger units, presumably corresponding to filamentous colonies. The other 13 populations were considered to be different picoeukaryotes although the possibility that some of them could be chlorophyll-positive (red fluorescent) and phycobilin-negative bacteria cannot be discarded. Fig. 4 shows three examples of different ponds, with the detected populations and the codes we assigned to them.

4

Red (chlorophyll) fluorescence vs. side scatter (left) and vs. orange (phycobilin) fluorescence (right) of three saltern samples selected as examples of the flow cytometric detection of different photosynthetic microbes. Saltern of 4% salinity (upper panel), 8% salinity (central) and 32% salinity (lower panels). The different populations are marked with letters. Population F corresponded to the most abundant picoalgae detected by flow cytometry in the samples (see also Fig. 5).

The total concentration of picoplankton presented a distribution pattern similar to that of all phytoplankton, with maxima at salinities of 8% and 32–37% (Fig. 5). The number of populations (Table 3) ranged from 6 for the 4% pond to 2–3 in the crystallizers. The most abundant population (population #F in Fig. 4), which peaked at 8% in both surveys (Fig. 5), reached concentrations of around 300,000 cells ml⁻¹.

5

Total picoplankton abundance as detected by flow cytometry in the experimental ponds sampled on 18 and 26 May 1999, as well as the contribution of some examples of the different picoplankton populations identified by flow cytometry on 18 May 1999 and 26 May 1999. Population #F was the most abundant picoplankton population appearing in salinities < 25% and contributing most of the picoplankton numbers in salterns from 5% to 15%. Population #C, a phycobilin-containing organism of a rather large size, in contrast, appeared in salterns of salinity >31% and contributed less than 10% to total picoplankton abundance.

3

Number of classes, detected for different descriptors (S_x, see Section 2) and CHEMTAX-derived taxa, along the salinity gradient

Survey	Salinity (%)	SM Phyto-plankton taxa	SF Pico-plankton populations	SP Pigments detected by HPLC	CHEMTAX-derived taxa	S16S 16S rRNA bands	S18S 18S rRNA bands
18/5/99	4	13	6	13	5	12	32
	5.4	17	5	11	4	9	28
	8	17	4	12	6	7	31
	11	9	4	9	2	4	21
	15	15	3	5	2	2	12
	22.4	6	3	3	1	2	10
	31.6	4	4	5	3	2	12
	37	2	2	6	1		11
26/5/99	4.0	16	6	12	4
	5.4	13	5	11	4
	8.0	17	4	13	6
	11	15	4	13	6
	15.0	7	2	6	2
	22.4	5	3	4	2
	25.0	9	2	3	1
	31.6	4	2	5	2
	36.0	7	3	6	1
	37.0	6	3

In the case of CHEMTAX-derived taxa, the potential maximum is 9. Only estimated contributions to total Chl a exceeding 0.2 μg l⁻¹ have been considered.

3.4 Genetic fingerprinting

Genetic fingerprinting of oxygenic phototrophs targeted cyanobacteria and algal chloroplasts. Heterotrophic bacteria were, therefore, mostly excluded from this analysis. The number of bands decreased from 12 to 2 along the gradient (Table 3). Sequences from excised bands indicated a shift from marine Cryptomonadaceae in the first two ponds to halophilic Cyanobacteria (99% similarity in 16S rRNA sequence to Euhalothece sp.) in the last two ponds, with a predominance of chlorophytes in the intermediate salinity ponds (Fig. 6). Given the dearth of sequences in the data base and the relatively low similarities of these bands to Chlorella, they could very well represent Dunaliella.

6

DGGE gel after PCR performed with 16S rRNA specific primers for oxygenic phototrophic microorganisms, in planktonic samples taken along the salinity gradient on 18 May 1999. Some minor bands, counted in Table 3, are not visible in the figure. Numbered bands (1–8) were excised from the gel and sequenced. Closest relative and percentage of similarity (in parenthesis) are: 1, Marine clone OM283* (97%); 2, Pyrenomonas salina* (94%); 3, marine clone OM283* (99%); 4, Cyanothece sp.** (92%); 5, marine clone OCS20* (98%); 6, Chlorarachnion sp. (98%); 7, Chlorella sp. (93%); 8, Euhalothece sp.** (99%). *, Cryptomonadaceae; **, Cyanobacteria. Nucleotide sequence accession numbers at EMBL are AJ580966 to AJ580973.

The 18S rRNA fingerprints, which targeted eukaryotic microorganisms (both auto and heterotrophs) yielded between 10 and 32 DGGE bands and each pond presented a particular fingerprint, indicating a quite different community composition for each salinity level (see [28] for details). The number of bands decreased from the 4% to 15% ponds and remained between 10 and 12 in the other ponds (Table 3). As discussed later, this high number could be influenced by the presence of heterotrophic microorganisms and by methodological biases.

3.5 Diversity indices

For all the descriptors (x) considered, the number of classes, S_x, presented a clear decreasing trend with salinity (Tables 3 and 4). The minimal number of classes identified ranged from 2 for phytoplankton, flow cytometry and 16S rRNA bands, to 10 for 18S rRNA; the maximal number ranged from 6 for flow cytometry to 32 for 18S rRNA. In spite of the differences in range of variation, all the S_x corresponding to the different descriptors were well correlated (Table 4).

4

Linear (Pearson) correlation coefficients among the salinity and the number of classes for the different descriptors

	Salinity	SM	SF	SP	S16S
SM	−0.85
SF	−0.66	0.58
SP	−0.77	0.78	0.74
S16S	−0.81	0.59	0.83	0.91
S18S	−0.83	0.73	0.76	0.98	0.92

The number of observations was 16–18 for pairs involving salinity, S_M, S_F and S_P and 7–8 for those with S_16S and S_18S. All values are significant (p < 0.05).

The K_M index for phytoplankton and the Shannon diversity indices for phytoplankton (D_M), pigments (D_P) and DGGE (D_16S and D_18S) presented in general the highest values at salinities below 10–15% (Figs. 7 and 8). Except for D_F (picoplankton) and for the correlation between the DGGE indices (D_16S and D_18S) and the phytoplankton ones (D_M and K_M), all these indices were negatively correlated with salinity and positively correlated among themselves (Table 5). In the case of DGGE bands, HPLC pigments and K_M index for phytoplankton, there was a clear trend of lower index values at higher salinities. The pattern was more complex for the picoplankton D_F index, with minima at 8–15% and 37% and for the phytoplankton D_M index, which is more sensitive than K_M to changes in relative abundance among groups and showed minima at salinities of 5%, 22% and 37% (on 26 May).

7

Distribution of the Shannon diversity indices on 18 (filled symbols) and 26 May 1999 (empty symbols) for (a) pigment concentrations determined by HPLC (D_P), (b) phytoplankton counts by the inverted microscope technique (D_M) and (d) fluorescent picoplankton counts (DF). (c) Distribution of the K_M diversity index for phytoplankton (see text).

8

Distribution of the Shannon diversity index based on the number and intensity of bands in DGGE gels from samples along the salinity gradient on 18 May, after a PCR with primers for 16S rRNA (D_16S, filled symbols) or for 18S rRNA (D_18S, empty symbols).

5

Linear (Pearson) correlation coefficients among salinity and several diversity indices

	Diversity indices
	Salinity	Phytoplankton		Picoplankton	Pigments	DGGE
		DM	KM	DF	DP	D16S
DM	−0.51
KM	−0.89	0.74
DF	n.s.	n.s.	n.s.
DP	−0.85	0.61	0.73	n.s.
D16S	−0.89	n.s.	0.69	n.s.	0.96
D18S	−0.79	n.s.	0.64	n.s.	0.89	0.95

The number of observations was 16–18 for pairs involving salinity, D_M, K_M and D_P, and 7–8 for those with D_16S and D_18S. All values, except those marked “n.s.” are significant (p < 0.05).

4 Discussion

4.1 Autotrophic biomass along the salinity gradient

All indicators of autotrophic biomass, including Chl a concentration, picoplankton and phytoplankton cell numbers (Figs. 1–3 and 5), presented a bimodal distribution, with maxima at salinities 5–11% and 37%. This distribution of Chl a concentration and phytoplankton abundance agrees with the findings of Pedrós-Alió et al. [11] in the same study area, although there were differences in the particular salinity at which the chlorophyll peaks were found and in the dominant groups along the salinity gradient. Given that the survey of [11] took place in July, 1993, part of these differences may be due to seasonal successional changes. Unfortunately, studies following the microbial community through a seasonal cycle have not been done in these salterns and there is no available background information to extract further conclusions. The Chl a peak at the lower salinities coincided with maxima of carbon fixation and Chl a-specific carbon fixation rates, as determined using the ¹⁴C method [29]. However, carbon fixation rates were very low in the crystallizers, in spite of their high Chl a concentration.

The comparison between the 18 and 26 May surveys showed significant variability, in spite of the relative stability of the temperature and salinity conditions. The changes affected both biomass and taxonomic composition, especially in the lower salinity ponds, and may reflect both spatial heterogeneity [29] and temporal variability. An important factor that must be considered in these shallow ponds is the resuspension of benthic organisms, due to wind events or to the activity of water birds. Such mechanisms could explain the high concentration of large Nitzschia cells found in the 8% pond on 18 May.

4.2 Comparison between optical microscopy observations and CHEMTAX results

Results of the CHEMTAX program for the contribution of dinoflagellates and diatoms were in general agreement with those of microscopy (Figs. 2 and 3). The correspondence was also good for Dunaliella spp., as could be expected because the corresponding initial pigment ratio had been adjusted with data from the present study. However, the relationship between cell numbers and pigment-derived contribution was not significant for cyanobacteria. The pigment ratios for cyanobacteria used as input to the CHEMTAX calculations were derived from a laboratory culture of the unicellular Synechococcus. This species was characterized by the presence of the carotenoid zeaxanthin and changes in irradiance were found to have a pronounced effect on the cellular ratio of zeaxanthin to chlorophyll a[35]. In the shallow salterns, irradiance by far exceed the experimental conditions and thus the zeaxanthin to chlorophyll a ratio of cyanobacteria used in the CHEMTAX calculations may not have been representative for the cyanobacterial community in the salterns. Furthermore, several filamentous cyanobacteria have different zeaxanthin to chlorophyll a ratios than the unicellular species [44] and therefore this group of organisms may have been consistently underestimated. Unfortunately no pigment ratios from species of Oscillatoriales were available for inclusion in the calculations. In the case of cryptomonads, the agreement was good for part of the samples but not for others, in which these organisms could not be detected by microscopy. In one of the samples (8% of 18 May) at least, this lack of correspondence could be due to the presence of the ciliate Mesodinium, which contains cryptophyte pigments. According to the CHEMTAX analyses, prasinophytes and chlorophytes were relatively important in some samples; organisms of these groups were also detected in the microscopic observations, but most of them could not be positively identified as such and were included in categories such as “others” (Fig. 3). As discussed in Section 2.2, the minor CHEMTAX-derived contribution of cryptophytes and diatoms in the 32% pond during the second survey should be considered as doubtful.

4.3 Phytoplankton composition along the salinity gradient

The results reported here confirm the dramatic ecological variability originating from the salinity gradient in the salterns. Basically, the phytoplankton communities in the studied salterns could be divided into a marine assemblage, occurring at salinities up to 15%, and a halophilic assemblage, found at higher salinities. There seemed to be a gap at intermediate salinities (22.4%), with minima of phytoplankton and picoplankton abundance and relative minima of phytoplankton diversity. Dinoflagellates were present up to salinities around 11% and diatoms practically disappeared at salinities >22%, in accordance with published data [16]. These observations were supported by the results of a perturbation experiment (data not shown) in which water from selected ponds (11%, 22.4% and 37%), either untreated or diluted to 85%, 75% and 60% of the original salinity with distilled water, was placed in 30 l tanks [30]. Due to evaporation, salinity in the tanks increased significantly between the beginning (20 May) and the end (27 May) of the experiments. Diatoms and dinoflagellates, originally present in the 11% community, disappeared in the untreated containers, which reached a salinity of 13.5%. Dinoflagellates disappeared also in the 85% dilution tank, which ended up with a salinity of 13%. The taxonomic changes around 22% salinity were also present in the picoplankton (e.g. population F in Fig. 5) and in the genetic composition of prokaryotes, which showed a consistent discontinuity between 10% and 22% salinity [28]. It is interesting to note that zooplankton presented a marked biomass maximum (due to Artemia salina) at salinities between 15% and 31.6%[30]. Probably, this Artemia peak contributed to low phytoplankton biomass at the 15%–22% salinity interval. However, Artemia abundance varied relatively little between 15% and 31.6% salinity, so that it seems unlikely that the qualitative changes in microbial composition around 22% could be attributed to grazing effects. Microzooplankton grazing was significant in the 4% and 8% salinity ponds while no significant microzooplankton grazing on the total phytoplankton community was found in the 11% pond and the crystallizer [29]. In addition to the potential effect on the total phytoplankton biomass, this grazing may have influenced the composition and diversity of the plankton community through selective grazing on some groups of phytoplankton. Thus, grazing in the 4% pond was primarily on prasinophytes and diatoms while in the 8% pond mainly cryptophytes, chlorophytes and possibly cyanobacteria were grazed by microzooplankton [29]. Unfortunately no data are available to quantify the abundance of microzooplankton along the salinity gradient.

4.4 Diversity patterns

As can be seen in Table 3, the number of classes of the different variables considered (S_x) tended to decrease with salinity and reached the minimal values in the crystallizers. All S_x indices were significantly correlated (Table 4). Due to the variety of methods used, different numbers of classes must be expected, even when dealing with the same organisms. For example, morphological differences among filamentous cyanobacteria, will not have been reflected by the HPLC analyses. Similarly, Mesodinium would not be separated from cryptophytes using information derived from only pigment analysis. The results of the molecular analyses deserve, however, some comments. In a parallel study carried out in the same salterns we detected that the number of groups of Bacteria and Archaea decreased as salinity increased, until only one group became dominant, but with a high degree of microdiversity [27]. Such microdiversity corresponds to clusters of closely related 16S rRNA sequences below the “species-level” (98–99.9% similarity in the sequence) and may represent the coexistence of several closely related clones of microorganisms that form ecologically distinct populations ([45,46] and references therein). On the other hand, it has been described that some DGGE bands (considered here as operational taxonomic units, OTUs) could correspond to artifacts, because simultaneous presence of several closely related 16S rRNA fragments may easily result in heteroduplex formation [47]. Therefore, in extremely low diversity assemblages, such as crystallizers, DGGE fingerprints require careful interpretation because the number of OTUs detected can overestimate the actual microbial richness [28]. In the case of eukaryotes, the relatively high number of 18S rRNA DGGE bands at the high salinity end of the gradient could be, in part, an artifact due to the formation of heteroduplexes derived from the presence of several closely related 18S rRNA sequences [28]. We cannot discard either the presence of eukaryal heterotrophs, such as yeasts, which have been recently reported in hypersaline waters [48]. Such organisms could have been overlooked in the microscopic counts or included in the “others” category as non-identified cells.

The values of the Shannon (D_x) and K_M diversity indices (Figs. 7 and 8), are affected both by the number of classes and by their individual abundances and, therefore, their distribution patterns should be more influenced than class numbers by the short-term ecological dynamics within the communities. In general, these indices showed decreasing values with increasing salinity, but D_M and D_F presented additional minima at intermediate salinities (Figs. 7 and 8). The phytoplankton D_M minimum at 22% can be related to the changes of community dominance and the minimum in Chl a concentration discussed above. The high diversity values of this index at 8% coincided with Chl a (Fig. 1) and primary production maxima as measured on 26 May [29] but the correspondence did not hold at the high salinity extreme, which presented very high Chl a concentration, very low primary production values and intermediate or low D_M diversity. The picoplankton D_F minimum at 8–15%, which is not reflected in the number of populations, can be related to the strong dominance of population F (Fig. 5).

Potential drawbacks of different diversity estimates have been considered by Margalef [42] and Nübel et al. [22] among others. As recognized by Nübel et al. [22], who attempted a quantification of microbial diversity based on morphotypes, 16S rRNA genes and carotenoid analyses, none of these approaches allows an exact determination of the number of existing classes and their abundance in the community. The problem does not lie only in the classification of elements, but starts with the selective effect of the sampling method and strategy adopted. In this context, it may be useful to adopt the proposal of Margalef [42], of distinguishing between diversity and biodiversity. Diversity is a measure of the richness of components of the biosphere which are active or abundant at a particular time and location, while biodiversity refers to set of non-redundant genetic information contained in this location. At any point in time, the differences between diversity and biodiversity are likely to be higher in a strongly dynamic environment. In the case of the salterns, the agreement of the results obtained with different approaches strongly suggests that there is a consistent trend of decreasing biodiversity with increasing salinity. The observation that this trend affects both prokaryotic and eukaryotic microautotrophs suggests that, as discussed by Brock [17] and Pedrós-Alió et al. [11], the underlying cause is likely the selective effect of extremely high salinities.

5 Conclusions

A varied set of diversity estimates based on microscopy, pigment analysis, flow cytometry, and DNA-based approaches confirmed a decrease in diversity with increasing salinity, indicating the selective effect of extreme environmental conditions on autotrophic microorganisms, both prokaryotic and eukaryotic. The numbers of elements of the different descriptors used (number of microalgal taxa counted by optical microscopy, flow cytometry-determined populations, pigment types, DGGE bands) were significantly correlated among themselves and negatively correlated with salinity. The Shannon diversity indices, which are influenced by the relative abundances of the different elements, also showed an overall decrease with salinity, but in the case of microalgal taxa and flow cytometric picoplankton presented marked minima at intermediate salinities. The phytoplankton diversity minimum around 22% salinity appeared to be related to a marked change in community composition, from an assemblage with mixed participation of dinoflagellates and diatoms, to another dominated by Dunaliella and cyanobacteria. In the case of picoplankton, low diversities (as measured by the Shannon index) at salinities 10–15% were due to the presence of a strongly dominant population that disappeared at salinities above 22%. A qualitative change around 22% was also clearly apparent from the fingerprinting analyses of 16S rRNA and 18S rRNA, indicating a similar response of the prokaryotic and eukaryotic communities. Our results indicate that general patterns of diversity variation along the salinity gradient in the salterns were comparable for different descriptors of the microautotrophic planktonic community.

Acknowledgements

This study was supported by the CSIC, by the Spanish project REN2001–2120/MAR (MicroDIFF) and by EU contract MAS3-CT97–0154 (MIDAS project), in the framework of MAST 3 Programme. We thank Mr. Miguel Cuervo-Arango for permission to work in the Santa Pola salterns. E.O.C. benefits from the Programa Ramón y Cajal of the Spanish Ministerio de Ciencia y Tecnología.