Temporal and Spatial Patterns of Eukaryotic and Bacterial Communities Found in Vernal Pools

Abstract

In this study, we examined the effects of physicochemical variability on the microbial communities of vernal pools. Denaturing gradient gel electrophoresis revealed temporal changes to be more pronounced than spatial changes in eukaryotic and bacterial communities. Sequencing revealed high degrees of richness in decomposers, which supports the notion that vernal pools are heterotrophic habitats.

Vernal pools are seasonally flooded ecosystems that are subject to variability in abiotic conditions (8). Although previous studies have documented the importance of abiotic conditions for vernal-pool macroorganisms (3, 9, 10, 12, 18, 20), little is known about abiotic influences on microorganisms. Given the abundance of vernal pools in the landscape (e.g., reference 5) and their highly variable physicochemical natures (2, 4, 6, 17, 19), vernal pools can serve as a model system to explore abiotic influences on microbial communities. Further, while the diversity of vernal-pool macroorganisms has been well-documented, comparable studies of microbial biodiversity are rare (see reference 8 for a review). In one survey of a single snowmelt pool, 76% of the identified biota were protists or bacteria (15), which suggests that these organisms dominate temporary water bodies in terms of both species richness and abundance (8). The main objective of this study was to characterize the microbial (eukaryotic and bacterial) communities of vernal pools in northeastern Ohio and to illustrate any spatial and temporal variations in these communities in relation to abiotic factors.

This study was conducted at Case Western Reserve University's Squire Valleevue and Valley Ridge Farms (41°29′53″N, 81°25′27″W; elevation, 320 m), where five vernal pools (one man-made pool, called the salamander pond, and four natural pools), which ranged in size from 16 to 20 m², were studied. The pools were located within an 80-year-old secondary-growth forest, which is characterized by well-drained, silt loam soils and a canopy dominated by Acer saccharum and Fagus grandifolia. From each pool, overlying water, organic detritus (i.e., leaf matter), and soil were collected on 1 April, 4 May, and 2 June 2004. These three collection times were shortly after snow melting and prior to the drying of the pools. Three replicate samples (of each sample type) were collected from the shore of each pool and combined. Specifically, soil samples were cores collected with sterile 1-ml pipette tips, and water samples were collected with 0.2-μm-pore-size polycarbonate filters through which 50 ml of water was filtered. At the time of sampling, dissolved oxygen concentrations, pHs, average depths, and terrestrial light levels were measured in situ. Relative conductivities of water samples were determined in the lab.

Community profile analysis was carried out by PCR-denaturing gradient gel electrophoresis (DGGE). DGGE was performed first to compare a subset of pool samples (three pool samples from three different sampling dates) of each individual sample type (water, soil, and organic detritus) separately and then to compare spatial and temporal trends among the five pools (for which the DNA from the three sample types was combined to assess whole-pool changes). DNA from each sample type was extracted such that the total masses (averages for all extracted samples were as follows: water, 103.5 ± 55.9 mg; soil, 666.3 ± 127.9 mg; and detritus, 529.9 ± 96.6 mg) were kept consistent. The fast DNA spin kit for soil was used for DNA extractions according to the protocol of the manufacturer (MP Biomedicals LLC, Solon, OH), except that cell lysis was performed by grinding the samples under liquid nitrogen and shaking them on a vortex genie. For PCR, we targeted the 18S ribosomal DNA of eukaryotes and the 16S rRNA genes of bacteria with primer sets F37GC/R518 and F1427GC/R1616, respectively (21). PCR amplifications were performed on a MyCycler thermocycler (Bio-Rad Inc., Richmond, CA) using Platinum Taq high-fidelity DNA polymerase (Invitrogen Inc., Chicago, IL). PCRs and cycling were essentially as described by van Hannen et al. (21), except that the total reaction volume was 25 μl, the final extension time was 10 min, and the bacterial PCR included 30 cycles, in which the annealing step began at a temperature of 65°C, which decreased by 0.4°C every repeat for the first 25 cycles and remained at 55°C for the final 5 cycles. Because the PCR template concentration can influence the composition and distribution of mixed-species amplification products (7), four dilutions of the template were used: 1:5, 1:10, 1:20, and 1:40. Positive reaction mixtures were combined prior to DGGE, and approximately 800 ng of PCR products per lane, on average, was loaded. DGGE was performed with the DCode universal mutation detection system (Bio-Rad Inc., Richmond, CA) by following the basic protocol of van Hannen et al. (21). Nucleic acids were visualized by staining with a 1:10,000 dilution of Sybr gold (Molecular Probes, Eugene, OR). All visualized bands were excised with separate 1-ml pipette tips and stored in water at −20°C prior to preparations for sequencing.

Distinct bands were identified as operational taxonomic units (OTUs) with TotalLab gel analysis software (Nonlinear Dynamics Ltd., Durham, NC). OTU profiles were analyzed by nonmetric multidimensional scaling (NMDS) to assess changes in community composition (21). NMDS was performed with SAS (version 9.1, 2003; SAS Institute) using PROC MDS and PROC IML to calculate the Bray-Curtis distance metric (11). To assist in interpreting the changes in community profiles, Pearson correlations between dimension scores and the vernal-pool characteristics were calculated with SPSS (version 12.0.1, 2003; SPSS Inc.). To assess differences in degrees of richness (numbers of OTU) in the three sample types over time (n = 9; three pool samples from each of three different sampling dates), a two-way analysis of variance with repeated measures and a Tukey posthoc test were conducted with SigmaStat (version 3.5, 2006; Systat Software, Inc.). Prior to the analysis of variance, tests determined that the data had equality of variances.

NMDS revealed differences in the eukaryotic communities among the three sample types (Fig. 1). The eukaryotic communities on detritus formed a fairly distinct cluster, which separated especially from the soil communities. Degrees of eukaryotic richness varied significantly among sample types (P = 0.036) and were significantly greater in detritus samples than in water and soil (P = 0.050 for both Tukey posthoc tests), while no other significant differences were observed (comparison by date, P = 0.672; comparison by a combination of sample type and date, P = 0.150). In a study of the bacterial communities associated with the water, sediment, and rhizosphere of a wetland, Ibekwe et al. (14) found differences among these three substrates, similar to the results of our study. However, the difference between the water and the sediment in the wetland was most notable, while in our vernal-pool study, the eukaryotic community varied mostly between the detritus and soil. Our results are somewhat contrary to those of Verb et al. (22), who found little influence of the vernal-pool vegetative substrate on algal communities.

FIG. 1.

NMDS ordination diagram showing the changes in vernal-pool eukaryotic communities from three sample types over 3 months. The letters denote the three sample types (W, water; S, soil; and O, organic detritus). The stress value of the final configuration was 16.13%.

For whole-pool comparisons, strong correlations between NMDS dimension scores and physicochemical characteristics were observed (Table 1). This finding suggests that changes in the microbial community composition coincided with shifts in abiotic factors, which also changed over the study period (see Table S1 in the supplemental material). Changes in both eukaryotic and bacterial communities showed strong positive correlations with the pH, dissolved oxygen concentration, and depth and negative correlations with conductivity. The sampling time also affected microbial communities. NMDS revealed that the eukaryotic communities of all pools changed in June (Fig. 2A). Though the bacterial communities did not show as much clustering as the eukaryotic NMDS pattern, there were essentially three groups of bacterial communities roughly associated with the sampling month (Fig. 2B).

TABLE 1.

Pearson correlation coefficients showing the relationships between vernal-pool physicochemical characteristics and dimensions 1 (EUK dim 1) and 2 (EUK dim 2) from the NMDS analysis of the eukaryotic DGGE data and dimensions 1 (BAC dim 1) and 2 (BAC dim 2) from the NMDS analysis of the bacterial DGGE data^a

Characteristic	Coefficient for relationship with:
	EUK dim 1	EUK dim 2	BAC dim 1	BAC dim 2	Light	pH	DO concn	Conductivity	Depth
Light	0.261	0.470	0.183	−0.157	1
pH	0.563	−0.393	−0.234	0.621	−0.184	1
DO concn	0.535	−0.223	0.146	0.327	−0.107	0.709	1
Conductivity	−0.107	−0.645	−0.781	0.100	−0.192	0.180	0.013	1
Depth	0.393	0.265	0.556	−0.066	0.153	−0.212	−0.008	−0.456	1

^aLight, instantaneous canopy light levels; DO concn, dissolved oxygen concentration; depth, average depth for three arbitrary locations.

FIG. 2.

NMDS ordination diagrams showing the changes in profiles of eukaryotic communities (stress value, 13.78%) (A) and bacterial communities (stress value, 17.23%) (B) in vernal pools over 3 months. The letters denote the sample months (A, April; M, May; and J, June), and the subscripts denote the pools (S, salamander pond; 1, vernal pool 1; 2, vernal pool 2; 3, vernal pool 3; and 4, vernal pool 4).

From the whole-pool DGGE gels, one excised band representing each OTU was reamplified and used for DNA sequencing (see Fig. S1 in the supplemental material). These band sequences were screened against GenBank sequences last updated in June 2007 (http://www.ncbi.nlm.nih.gov) with BLAST. Biologically relevant sequences with the highest similarities were then aligned with the DGGE band sequences in BioEdit, version 7.0.5.2 (13), and used in a similarity index calculation, which omitted gaps and ambiguities. The results were confirmed with phylogenetic tree analyses (data not shown).

Of the 28 eukaryotic OTUs (see Fig. S1 in the supplemental material) and 32 bacterial OTUs, 23 eukaryotic OTUs and 18 bacterial OTUs were successfully sequenced (Table 2). The eukaryotic richness was well-represented by fungal groups, which are important to the decomposition of allochthonous carbon input into vernal pools (1). The pools were also rich in Alphaproteobacteria and Betaproteobacteria, which is consistent with the results of Olapade and Leff (16), who found that these groups dominated bacterial biofilm assemblages in a stream. Ibekwe et al. (14) also observed an abundance of Betaproteobacteria in the soil and rhizosphere of a wetland. These authors suggest that Alphaproteobacteria and Betaproteobacteria may be important in such systems for the degradation of organic matter (14, 16). The prevalence of fungal groups and Alphaproteobacteria and Betaproteobacteria in vernal pools suggests that microbial communities are rich in heterotrophic organisms.

TABLE 2.

Sequence similarities of excised OTUs^a

OTU no.	% Similarity to closest relative sequenceb	Species corresponding to closest relative sequence	Accession no. of closest relative sequence	Taxonomic classification corresponding to closest relative sequence
Eukaryotic sequences
1	99	Navicula ramosissima	AY485512	Bacillariophyceae
2	100	Achlya bisexualis	DQ403201	Oomycota
3	99	Bodomorpha sp.	DQ211595	Cercozoa
4	96	Gymnophrys cometa	AJ14866	Cercozoa
6	98	Uroglena sp.	EF165132	Synurophyceae
7	97	Chrysocapsa paludosa	EF165145	Chrysophyceae
8	98	Chaetonotus sp.	AJ001735	Gastrotricha
9	97	Hyphodiscus hymeniophilus	DQ227258	Ascomycota
11	100	Sporidiobolales sp.	EF060827	Basidomycota
12	98	Hyphodiscus hymeniophilus	DQ227258	Ascomycota
13	98	Rhizophydium elyensis	DQ536479	Chytridiomycota
14	100	Leptodontidium orchidicola	DQ521603	Ascomycota
15	99	Chalara cylindrosperma	AF222507	Ascomycota
16	97	Troposporella fumosa	AY856953	Ascomycota
17	100	Leptodontidium orchidicola	DQ521603	Ascomycota
18	100	Dothidea sambuci	AY544722	Ascomycota
19	98	Xylariales sp.	AB255193	Ascomycota
21	99	Cyprididae sp.	AY622196	Ostracoda
22	97	Physoderma dulichi	DQ536472	Chytridiomycota
23	100	Acidomyces richmondensis	AY374300	Ascomycota
25	99	Leotia lubrica	AY544687	Ascomycota
27	100	Bryocamptus pygmaeus	AY627015	Copepoda
28	100	Propappus volki	AY365457	Annelida
Bacterial sequences
6	99	Variovorax paradoxus	DQ257419	Betaproteobacteria
7	97	Flavobacteria bacterium	AB269814	Flavobacteria
10	94	Variovorax dokdonensis	DQ178978	Betaproteobacteria
11	97	Rhodoferax sp.	EF451708	Betaproteobacteria
12	86	Variovorax sp.	AM502921	Betaproteobacteria
13	99	Acidobacterium capsulatum	AM086241	Acidobacteria
14	98	Terriglobus roseus	DQ660895	Acidobacteria
16	97	Amphora delicatissima	U96445	Diatom chloroplast
17	100	Erythrobacter sp.	AY371411	Alphaproteobacteria
18	100	Erythrobacter sp.	EF581001	Alphaproteobacteria
19	96	Rhodoferax sp.	AB206450	Betaproteobacteria
20	99	Pseudomonas sp.	EF541138	Gammaproteobacteria
21	97	Methylobacterium extorquens	DQ870723	Alphaproteobacteria
22	100	Asticcacaulis sp.	AB093140	Alphaproteobacteria
23	96	Aquabacterium sp.	AF089859	Betaproteobacteria
28	92	Pseudonocardia sp.	EF588230	Actinobacteria
29	94	Flavobacterium columnare	DQ005510	Flavobacteria
31	96	Sphingomonas sp.	AF191021	Alphaproteobacteria

^aSequences were aligned to the closest relatives from the GenBank database.

^bGaps and ambiguities were not included in the similarity calculations.

Overall, our study revealed that vernal-pool microorganism communities are temporally variable and influenced by physicochemical conditions within the pools. The importance of the substrate type for the structure of eukaryotic communities and the prevalence of fungi and Alphaproteobacteria and Betaproteobacteria indicate that microbial communities in these pools are rich in heterotrophic organisms and decomposers.

Nucleotide sequence accession numbers.

The sequences determined in this study were deposited in GenBank under accession numbers EU000325 through EU000365.