MICROBIAL ECOLOGY OF SOUTH SLOUGH SEDIMENTS: COMMUNITY COMPOSITION OF BACTERIA AND PATTERNS OF OCCURRENCE by ERIC CHARLES MILBRANDT A DISSERTATION • • , • ~. ... ; : • ~ t ••••. , . I •• Presented to the Department ofBiology and the Graduate School of the University of Oregon in partial fulfillrm:nt cf tt.e re<;.uirements for the degree of Doctor ofPhilosophy June 2003 "Microbial Ecology of South Slough Sediments: Community Composition ofBacteria and Patterns of Occurrence," a dissertation prepared by Eric C. Milbrandt in partial fulfillment of the requirements for the Doctor of Philosophy degree in the Department of Biology. This dissertation has been approved and accepted by: Dr. Nora Terwilliger, Chair of the Examining Committee Date Committee in charge: Dr. Nora Terwilliger, Chair Dr. Lynda Shapiro Dr. Michelle Wood Dr. Richard Castenholz Dr. Mark Reed Dr. Stephen Rumrill Accepted by: Dean of the Gr:lduate School 11 © 2003 Eric Charles Milbrandt 111 Title: MICROBIAL ECOLOGY OF SOUTH SLOUGH SEDIMENTS: COMMUNITY COMPOSITION OF BACTERIA AND PATTERNS OF OCCURRENCE in the Department ofBiology Eric Charles Milbrandt An Abstract of the Dissertation of for the degree of to be taken IV Doctor ofPhilosophy June 2003 Approved: _ Dr. Lynda Shapiro The description of any community depends on the spatial, temporal, and organizational parameters chosen. It is essential to understand how patterns and dynamics vary at multiple scales in order to fully understand the community. In this dissertation, I exercised a spatially explicit approach to ask the following fundamental questions about microbial community structure in estuarine soft sediment habitats. (1) What are the patterns ofvariability within a site and among estuarine sites? (2) Are the same patterns found in all intertidal soft sediment habitats or only in the estuary? (3) Do restored and estuarine sediments have similar microbial assemblages? (4) Is there a relationship between the growth dynamics of algal mats and the organization of bacteria communities? The nuclear gene encoding the small subunit of the ribosome, 16S rDNA, was used to identify characteristics of the community and to identify unknown bacteria. The difference between estuarine and outer coast soft sediments was comparable to genetic diversity differences observed between restored and mature estuarine sites. vBacteria communities among mature estuarine sites differed, though to a lesser degree. Each mature site had a distinct community fingerprint, which suggested that the environment selects for a suite of species. Similarly, there was not an estuary wide community fingerprint associated with algal mats. However, when each site was analyzed separately, the bacteria associated with algal mats were distinct from bacteria in sediments with no mat. Observed differences in genetic diversity between a restored site and a mature estuarine suggested several testable hypotheses about succession of bacteria communities after habitat restoration. Restored estuarine sediments provide a greater supply of available energy to the community, and observed diversity differences may have resulted from reduced competition for substrate. A number ofnewly discovered sequences demonstrated that the majority ofbacteria from intertidal soft sediment habitats have not been cultured. The phylogenetic position of the newly contributed sequences may help optimize culture strategies, thus contribute to an understanding of the relationship between genetic diversity and ecosystem function. CURRICULUM VITA NAME OF AUTHOR: Eric Charles Milbrandt PLACE OF BIRTH: Buffalo, New York DATE OF BIRTH: March 8, 1974 GRADUATE AND UNDERGRADUATE SCHOOLS ATTENDED: University of Oregon Humboldt State University DEGREES AWARDED: Doctor ofPhilosophy in Biology, 2003, University of Oregon Bachelor of Science in Biology, 1997, Humboldt State University AREAS OF SPECIAL INTEREST: Ecology of Estuaries, Microbes, and Restoration PROFESSIONAL EXPERIENCE: Teaching Assistant, Oregon Institute ofMarine Biology, University of Oregon, Charleston, 1997-2002 Research Assistant, South Slough National Estuarine Research Reserve, Charleston, 1997-2002 AWARDS AND HONORS: Research Experience for Undergraduates (REU) Fellow, National Science Foundation, 1996 William Sistrom Memorial Scholarship, Department ofBiology, University of Oregon, 2000. VI vii Invited speaker, Estuarine Research Federation Biennial Meeting, St. Pete's Beach,2001 Student poster award, American Society for Limnology and Oceanography, Victoria, Canada, 2002 Doctoral Research Fellowship Nominee, Department ofBiology, University of Oregon, 2002 Oregon Sea Grant Travel Award, Multivariate statistics workshop, CSU- Fullerton, 2002 Student travel award, American Society for Limnology and Oceanography, Salt Lake City, 2003 GRANTS: Graduate Research Fellow, National Oceanic and Atmospheric Association, South Slough National Estuarine Research Reserve, 1998-2001 PUBLICATIONS: Milbrandt, B.C., 2002, "Communities ofbacteria as a metric for gauging restoration success," Earth Systems Monitor 12(2): 5-8. Nadeau, T-N., Milbrandt E.C., Castenholz RW., 2001, "Phylogeny ofAntarctic Oscillatorian Cyanobacteria," J. Phycology, 37:650-654. viii ACKNOWLEDGEMENTS The author is indebted to Dr. Lynda Shapiro for guidance and steadfast intellectual support. The data contained in this manuscript would not have been possible without Dr. Chuck Wimpee and Dr. Anna-Louise Reysenbach, who unselfishly shared their expertise in molecular biology. Special thanks to Barbara Butler, for sharing the currency of science. Victoria Poulton provided critical discussion and invaluable comments to improve this dissertation. Sincere thanks to the South Slough NERR for inspiration in the estuarine and restoration sciences. This research was supported in part by National Oceanic and Atmospheric Administration grant # NA870R0278 to Eric Charles Milbrandt and a grant from the University of Oregon to Dr. Lynda Shapiro. TABLE OF CONTENTS Chapter Page 1. GENERAL WTRODUCTION 1 II. LANDSCAPE-LEVEL PATTERNS OF OCCURRENCE OF BACTERIA FROM COASTAL SOFT-SEDIMENT HABITATS 4 Abstract 4 Introduction 5 Materials and Methods 8 Study Areas 8 Sampling Design 10 Extraction ofDNA from Estuarine Sediments 12 PCR Amplification of 16S rDNA gene 13 Separation of 16S rDNA by Denaturing Gradient Gel Electrophoresis (DGGE) 14 DGGE Gel Analysis and Statistics 14 Sequences ofDGGE Bands 15 Missing Data 15 Results 15 Variability in Bacterial Communities within Sites 15 Variability in Bacterial Communiities within South Slough Estuary 21 Variability in Bacterial Communities between Estuarine and Marine Sediments 21 Identity ofDGGE Bands 23 Discussion 23 III. BACTERIA DIVERSITY AND ANTHROPOGENIC HISTORY: A CASE STUDY OF RESTORED AND MATURE ESTUARWE WETLAND SEDIMENTS 31 Abstract 31 Introduction 32 Materials and Methods 33 Study Areas 33 Redox Microelectrode Measurements 34 DNA Extraction 35 PCR Amplification of the l6S rDNA Gene 35 Estimating Genetic Diversity using Denaturing Gradient Gel Electrophoresis 36 IX Chapter Page Clone Library Construction and Screening 36 Results 37 Discussion 41 N. SPATIAL AND TEMPORAL DISTRIBUTION OF MAT- FORMING ALGAE AND THE GENETIC DNERSITY OF THEIR ASSOCIATED MICROBIAL COMMUNITIES 47 Abstract 47 Introduction 48 Materials and Methods 49 Algal Mat Cover Sampling Design and Statistics 49 Microbial Community Analysis 50 Results 52 Seasonal Dynamics of Vaucheria longicaulis Mats 53 Bacterial Community Associated with Algal Mats 55 Discussion 58 V. PHYLOGENY OF UNCULTNATED BACTERIA FROM ESTUARINE AND MARINE INTERTIDAL MUDFLATS 64 Abstract 64 Introduction 65 Materials and Methods 67 Extraction of Genomic DNA 67 Amplification of the 16S rDNA Gene 68 Clone Library Construction and Screening 68 Phylogenetic Reconstruction 69 Identity ofDGGE Bands 70 Results 70 Discussion 76 VI. GENERAL CONCLUSIONS : 84 LITERATURE CITED 87 Chapter I 87 Chapter II 87 x Chapter Page Chapter III 91 Chapter N 94 Chapter V 97 Chapter VI 100 xi LIST OF FIGURES Figure Page 1. Map of the South Slough Arm ofCoos Bay, Oregon, USA 9 2. Sampling Design for Assessment of Within-site Variability 11 3. DGGE Gel Images from 2001 17-19 4. Cluster Analysis and MDS from 2001 DGGE Gels 20 5. Cluster Analysis, MDS plot, and DGGE Image of2002 Samples 22 6. DGGE Community Fingerprints and Position ofPicked Bands 24 7. Vertical Microelectrode Redox Potential Profiles from Restored and Mature Intertidal Mudflats 39 8. Cluster Analysis and DGGE Gel Images from Restored and Mature Intertidal Mudflats 40 9. Rarefaction Curve and Restriction Fragment Length Polymorphism Gel Images 42 10. The "Phytomegatron" 51 11. Probability Plot of Vaucheria longicaulis Mat Percent Cover 54 12. Seasonal Variability of Vaucheria longicaulis Mats in South Slough National Estuarine Research Reserve 56 13. Spatial and Temporal Variability of Vaucheria longicaulis Mats in South Slough National Estuarine Research Reserve 57 14. Phylogenetic Reconstruction ofUnknown Estuarine Bacteria with Selected Known Taxa 73 15. Phylogenetic Reconstruction of o-proteobacteria 78 16. Phylogenetic Reconstruction of 'Y-proteobacteria 79 17. Phylogenetic Reconstruction of a-proteobacteria 80 xii Table LIST OF TABLES Page xiii 1. Identity ofDGGE Bands from Estuarine and Marine Intertidal Mudflat Habitats 26 2. Two-way ANOVA of Yellow-green Algal Mat Vaucheria longicaulis Percent Cover 59 3. One-way ANOSIM of Mat-associated and Non-mat Bacteria Communities 60 4. Best matched 16S rDNA Sequences to the Unknown Sequences Found in South Slough National Estuarine Research Reserve 74-75 1CHAPTER I GENERAL INTRODUCTION Bacteria are extremely diverse and they occur in a range of habitat types. For example, bacteria have been discovered deep in the lithosphere (Ghiorse 1997), in hot springs (Pace 1997), and in close association with hydrothermal vents (Reysenbach and Cady 2001). Bacteria are a ubiquitous component of the biosphere, but their organization is not haphazard and random. These observations pose a series of fundamental questions: (1) how are communities ofbacteria organized, (2) what are the limits of distribution of a bacterium, (3) how do natural and anthropogenic gradients affect the spatial distributions of bacteria, and (4) how can variability in spatial distribution improve our understanding of estuarine ecosystems? Estuaries are ideal places to understand the organization and variability of microbial communities. Salinity, temperature, and redox potential are some of the gradients that affect distributions of organisms. Bacteria have an important role as prey for other estuarine dependant species and as the mediators of decomposition. Microbes, particularly bacteria, are central to the process of decomposition in marine and estuarine ecosystems. A description of the community composition across abiotic and anthropogenic gradients will improve our understanding of estuaries. The anoxic intertidal sediments of estuaries are also an ideal place to look for taxa that 2biodiversity and improve our understanding ofbacterial phylogeny. Estuaries are often developed for their capacities to deliver goods or provide rich soils for agriculture. Therefore, they have been impacted by human activities which have included dredged channels and drained salt marshes. The loss of coastal habitat in estuaries has prompted restoration activities, which provide an opportunity for improving degraded habitats. The pattern of recovery of bacteria in restored sediments may provide insights about restoration. This dissertation details the application of DNA-based markers to describe phylotype distribution patterns ofbacteria that occur in marine and estuarine habitats on the coast of southern Oregon, USA. The distribution ofbacteria phylotypes across natural and anthropogenic gradients was determined with denaturing gradient gel electrophoresis ofthe l6S rDNA molecule. l6S rDNA is a gene common to all prokaryotic life forms, plus mitochondria and chloroplasts. It is a well known marker that has an extensive database of published sequences that can be used for detecting the relatives of previously uncultured bacteria. It does not have information about the metabolic capacity or the physiological state, but can be amplified directly from the environment. The next chapter, Chapter II, uses a spatially explicit approach to understand the patterns of occurrence of microbial communities in a small estuarine system. The striking conclusion from this chapter is that the scale of an investigator's perspective reveals distinct patterns among estuarine sites. For example, there was no pattern across the redox gradient but bacterial communities were consistently different between raked plots (homogenized) and control plots. These within site patterns were insignificant when considering the variability among multiple estuarine, which showed a strong site specificity. On larger scales, bacteria communities in estuaries showed distinctive differences from communities in marine habitats. Chapter III examines genetic diversity following the restoration of tidal circulation to a degraded wetland I 3 examines genetic diversity following the restoration of tidal circulation to a degraded wetland habitat. The genetic diversity was greater in the restoration site than a mature site, which may reflect the pre-restoration impoundment. Chapter IV investigates the percent cover of an algal mat species, whose peak cover during the summer months follows the peak in biomass shown for other estuarine primary producers. Finally, Chapter V reconstructs a phylogenetic hypothesis of the phylotypes that were isolated from the intertidal mudflats. These results are the first known description of the uncultivated bacterial component from this habitat and represent a significant contribution to estuarine microbiology. Chapters II and III are co-authored with my major professor, Dr. Lynda Shapiro. Chapter III has been submitted for publication in Applied and Environmental Microbiology. 4CHAPTER II LANDSCAPE-LEVEL PATTERNS OF OCCURRENCE OF BACTERIA FROM COASTAL SOFT-SEDIMENT HABITATS Abstract The description of any community depends on the spatial, temporal, and organizational perspective chosen. It is essential to understand how patterns and dynamics vary at multiple scales in order to develop a predictive capability. In this study, I exercised a spatially explicit approach to ask fundamental questions about the structure of microbial communities in estuarine soft-sediments. The co-Author of Chapter II was my major professor, Lynda Shapiro, who has guided me through the thought process and assisted in editing drafts. The ideas and arguments that accompany these data were collected and analyzed by my hand. A bacterial community fingerprint was generated from amplified 16S rDNA genes that were separated with Denaturing Gradient Gel Electrophoresis (DGGE). Molecular sequence data for bacteria isolated from any given habitat circumvents the need for cultivation and provides a molecular-phylogenetic framework to describe microbial diversity based on 16S rRNA sequence. Results from this study demonstrate that patterns ofvariation at small scales were not universal and important only in select locales. The redox potential discontinuity layer (RPD) was expected to be a physiological boundary separating the aerobic community from the anaerobic community. Surprisingly, only Hidden Creek, showed evidence that the RPD layer was an important feature separating 5microbial communities. The amount of variability at larger spatial scales showed a striking pattern of site-specific identity. With few exceptions, samples collected within a site were more similar than samples from different sites. Bacterial communities collected from estuarine sites were distinct from the bacteria collected from intertidal marine mud. To clarify the differences in bacterial communities that inhabit marine and estuarine soft-sediments, DGGE bands were reamplified and sequenced. The higher salinity marine site yielded sequences that were related to the a-proteobacteria. We included an estuarine restoration site in the analysis and found that the restoration site is becoming more similar to other mature estuarine sites through time. This observation shows that communities of bacteria may be a promising metric for gauging success of future restoration projects. Transplant experiments and other field manipulation studies are needed to follow up on the growing list of hypotheses that have been developed to explain why specific lineages are restricted to precise environments. Introduction Communities of heterotrophic bacteria are associated with oxidation of deposited organic matter, regeneration of inorganic nutrients, and food web support (Fenchel et al. 1998). Cultivated isolates and biogeochemical field measurements have provided limited information about the role of microbes in marine and freshwater sediments (Capone and Kiene 1988). Estuarine and marine sediments were shown to have strong redox gradients in which steepness varies depending on location and sediment size (Fenchel and Riedl 1970). Theoretical calculations predict the occurrence ofthermodynamic zones of respiration (Chester 1990) Associated with these zones, is a progression ofbacterial functional groups. It is unknown whether the thermodynamic zones of respiration are also associated with a change in species composition. 6Historically, natural abundance and diversity of microbial populations were estimated by routine culture techniques. Anaerobic bacteria, common in estuarine and marine sediments, were difficult to identify due to the absence of information about enrichment conditions. Studies from several types of habitats estimate that more than 99% of microscopic organisms cannot be cultivated by routine techniques (Amann et ai. 1995, Pace 1997). Molecular sequence data isolated from any given habitat circumvents the requirement of cultivation and provides a molecular-phylogenetic framework (Pace 1997) to describe microbial diversity based on molecular sequence. The conservative properties of the 16S rDNA gene have allowed alignment of sequences spanning over 1 billion years of microbial evolution. It is an ideal molecule to use as a marker in the environment because it is shared by all bacteria. Recent research suggests that environmental conditions at distinct geographic locations select for a distinct assemblage ofprokaryotic organisms. Crump et al. (1999) used this postulate to explain why distinct bacteria assemblages were observed along a gradient in the Columbia River estuary, Oregon, USA. The hydrodynamic regime of the estuary mixed a distinct riverine and a marine bacterioplankton community. The free-living fraction never formed a distinct estuarine assemblage and was rapidly flushed out of the estuary. In contrast, the particle- associated fraction was retained long enough to develop a third, uniquely estuarine assemblage. Similarly, the spatial distribution of free-living heterotrophic bacterioplankton assemblage was observed in two sub-estuaries of the Chesapeake Bay (Bouvier and del Giorgio 2002). Members of the Class jJ-proteobacteria dominated the upper estuary, the class a-proteobacteria dominated the lower estuary and the Phylum Bacteroidetes prevailed in the middle estuary. This observation suggested that some lineages were restricted to specific environments because ofphysiological tolerances, growth optima, Dissolved Organic Carbon (DOC) concentrations, and associations with the estuarine turbidity maximum (ETS). Populations of small-sized organisms are not 7ubiquitous; they appear to be subject to selective forces at kilometer scales that influence diversity and abundance. Patterns at smaller scales, however, are less apparent due to technological restraints on a scientist's ability to sample at extremely fine scales. Spatial variability in distributions of soil organisms is thought to be accompanied by a predictable spatial structure (Ettema and Wardle 2002). Land use (Fromm et al. 1993), topography (Robertson et ai. 1997), tree species patch size (Saetre 1999) and corn plant spacing (Cavigelli et ai. 1995) have been suggested as major factors influencing microbial spatial patterns. In creek bank sediments (Franklin et ai. 2002), elevation above sea level and the degree of tidal flooding were postulated to be important determinants of community composition and structure. Field studies that described the distribution and genetic diversity of bacteria are particularly important in coastal and estuarine ecosystems where efforts are underway to restore drained and diked marshes (Zedler 2000). Sediments undergo visible and measurable changes following reintroduction of tidal flooding to previously diked marshes (Portnoy and Giblin 1997). The sediment and its biological components set an ecological foundation for later colonization by plants and invertebrates (Callaway 2001). Restoration success and the timing ofre-colonization rely on the return of proper soil conditions, including the presence of a bacterial community. Short and long term monitoring goals may be set more accurately with a description of the number of species, genetic diversity in the bacterial community, and the variability of bacterial communities in nature. The description of any community depends on the spatial, temporal, and organizational perspective chosen. It is essential to understand how patterns and dynamics vary at multiple scales in order to develop a predictive capability. In this study, the bacteria assemblage was observed and compared on several scales. A gene common to all bacteria, l6S rDNA, was 8chosen to track the distributions of taxa. I used a spatially explicit approach to ask the following fundamental questions about microbial community structure in estuarine soft sediment habitats. (1) Does the microbial community show variability above and below the redox potential discontinuity? (2) Do meter scale patches ofbenthic algae limit the distributions of microbial populations? (3) Are the same patterns found in all intertidal soft-sediment habitats or only in the estuary? I predicted that the redox potential discontinuity (RPD) layer would limit microbial populations so that the assemblage above the RPD would be different than below, and that similar taxa would be found above and below the RPD at a number of estuarine sites. I also postulated that tidal pools and algal mats at m-scales could explain the variation within each site. Finally, I predicted that a distinct estuarine community could be identified that was different from the intertidal sediment community outside the estuary. Materials and Methods Study Areas The South Slough National Estuarine Research Reserve (SSNERR) is a small drowned river mouth estuary that is heavily influenced by marine conditions (Figure 1). Summer months bring limited rainfall which produces a salinity range of27-32 PSU in the upper estuary and 30- 33 PSU in the mid estuary (NOAA, Central Data Management Office, 09/01). Nutrients (unpublished data), phytoplankton (Hughes 1999, Cowlishaw 2002), and chlorophyll a (Roegner and Shanks 2001) are advected into the estuary from the nearshore ocean during flood tides. The estuarine sites in SSNERR are characterized by broad mudflats, channelization, patches of algal mats and fringing salt marsh vegetation (Rumrill 2002). 9, . o Cape Arago lkm ! N Coos Bay Figure 1. Map of the South Slough Arm of Coos Bay, Oregon, USA. X indicates the location of 5 sample sites that span a gradient from fully marine sediments of Cape Arago to the estuarine sediments of Ferrie Ranch, Elliott Creek, Hidden Creek, and Kunz Marsh. 10 Unvegetated tideflats free of emergent salt marsh plants were chosen to measure communities of bacteria in the absence of any habitat complexity introduced by plant communities. These mudflats were devoid of vascular plants, but had a mosaic of mat forming algae (Oscillatoria spp., Vaucheria longicaulis, Rhizoclonium spp., Chaetomorpha spp.). The North Cove of Cape Arago (CA) is a marine site protected from the open ocean by an expansive rock reef that extends around the cape (Figure 1). Sediment has accumulated in the protected waters ofNorth Cove resulting in a soft-sediment macrofaunal community typical of an embayment or estuary. Sampling Design A. I focused on centimeter (cm) and meter (m) scale variability within South Slough National Estuarine Research Reserve (SSNERR) in August 2001. Core samples were collected in a cut off 140 cc syringe, with a diameter of3.2 cm (Walters and Moriarty 1993). I collected samples in four estuarine sites in two replicate groups separated by a distance of 3 m (Figure 2). In each group, two depth fractionated core samples (140 cc modified syringe) were collected at a separation distance of 3 cm. Genomic DNA was extracted from the sediments and used to amplify the 16S rDNA gene. Amplified sequences from PCR were separated with Denaturing Gradient Gel Electrophoresis (DGGE). I tested two a priori hypotheses to describe within site patterns of genetic composition: (1) All replicates within a depth fraction are more similar than between depth fractions. (2) Replicates collected within a patch are more similar than between patches within a site. B. A second component of the 2001 sampling was to determine if sampling within am-scale disturbance plot would yield less variability than sampling haphazardly in undisturbed plots. 11 3cm Disturbed plots -------r---...I 3cm Undisturbed plots 3cm 3m 3cm Figure 2. Sampling Design for Assessment ofWithin-site Variability. Core samples were collected in a spatially arranged pattern drawn in the above figure. Four replicates from undisturbed and four replicates from disturbed plots were collected from all four estuarine study sites in August 2001. 12 Mudflats in SSNERR are a patchwork of algae, water, and sediment. An artificial mosaic in all four estuarine sites was created by homogenizing the upper 2-3 cm of sediment in 2, I X I-m plots. Disturbance plots were separated by a distance of 3 m. A garden rake was used to homogenize the upper 2-3 cm of the sediment's surface and samples were collected 24-hours later. I hypothesized that similarity within the disturbance plot would be greater than between disturbance plots. C. The 2002 sampling regime focused on placement of the estuarine sites in context with a marine site. In order to do this, fewer replicates from Hidden Creek and Kunz Marsh and one from a marine 3 m. A garden rake was used to homogenize the upper 2-3 cm of the sediment's surface and samples were collected 24 hours later. It was hypothesized that replicates within the disturbance plots would show less variation than in surrounding undisturbed sediment. The success of 16S rDNA amplication in disturbed plots was greatly reduced, which resulted in missing data in Hidden Creek and Elliott Creek. Extraction of DNA from Estuarine Sediments DNA extraction was conducted by a direct lysis method modified from Zhou et al. (1996). Upon returning to the lab, DNA was extracted immediately from the sediments or stored at -80°C. We added 100 III of5MNaCl, 1 g of sediments, and then added 750 III ofCTAB Lysis buffer (pH 8.0, 1% CTAB, 0.05M Tris, 0.05M EDTA, 0.05M NaHzP04, 1.5M NaCl). The sediment slurry was mixed and incubated at 37°C for 30 min with Proteinase K (5 Ill, 20 mg/ml). We added 100 III of 20% SDS to the mixture followed by an incubation at 65°C for 1 hour, at- 80°C for 1 hour, and at 65°C for 1 hour. The boil/freezelboil sequence was followed by a 25:24:1 extraction in phenol/chloroform/isoamyl. The aqueous extract was removed, precipitated with 13 0.6 volumes isopropanol, and incubated overnight at -20°C. Precipitated nucleic acids were pelleted for 30 minutes at 15,000g, washed with 70% ethanol, and air dried. Pellets were resuspended in 100 Jll Sigma molecular-grade water and incubated overnight at 4°C. Crude extracts were not amendable to enzymatic reactions or restriction digestion, so they were agarose gel purified (Li and Ownby 1993). Genomic DNA extracts may not be uniform with regards to taxonomic affiliation. It is thought that some taxa are extracted and amplified more efficiently (Hugenholtz et al. 1998). It has been shown that heteroduplexes on DGGE gels can contain bands with multiple sequence identities (Sekiguchi et al. 2001). We attempted to minimize bias in the genomic DNA extraction by following the methods of Zhou et al. (1996). We also minimized the amount ofPCR inhibitors, common to sediment samples, by gel purifying all crude extractions prior to PCR analysis. PCR Amplification of 16S rDNA Gene Ten III (~40 ng) of gel purified template was used in a 50 III PCR reaction to amplify small subunit (SSU) 16S rDNA. Primers gc338F and 519R amplified a hypervariable region of the 16S rRNA gene (Muyzer et al. 1993, Ferrari and Hollibaugh 1999, Jackson et al. 2001). These PCR products were used in DGGE to observe community profiles ofbacteria. Each reaction contained 5 III lOX buffer, 5 Jll25 roM MgCh (Applied Biosystems) 1.2 Jll2.5 roM dNTP, 10 U Amplitaq LD, 1.5 III 10 pmol forward primer and 1.5 ul10 pmol reverse primer. The reaction was brought to 95° C for 3 minutes then cycled 35 times at 95°C for 30 seconds, annealed at 48°C for 30 seconds, and extended at noc for 30 seconds. After cycling, the reactions were held at noc for 10 minutes. Given the universal nature of the Domain level primers (gc338F, 5l9R) and the ubiquity ofbacteria in the environment, contamination of 14 samples was of great concern. We frequently checked for contamination by routinely running negative controls (PCR reactions with no template DNA added). Separation of 16S rDNA by Denaturing Gradient Gel Electrophoresis (DGGE) A Biorad D-Code gel rig was used to separate similarly sized, amplified rDNAs by melting domain (Biorad, Hercules, CA). All reagents were prepared as described in the D-Code Instruction manual and Applications Guide (Biorad, Hercules, CA). The denaturing gradient was 20% to 60% by weight urea and formamide (piceno et al. 2000). Gels were loaded with 15 III of the PCR reaction plus 5 Jll of loading dye (0.01% Bromphenol Blue, 40% glycerol), and run for 3 hours at 200 V. Gels were stained with SYBR Gold nucleic acid gel stain (Molecular Probes, Eugene, OR), and photographed on Polaroid film. Polaroid gel images were scanned into digital format and inverted for analysis. DGGE Gel Analysis and Statistics A series of DGGE gels were run in order to compare replicates from environmental samples. Gels were analyzed following van Hannen et al. (1998, 1999), and Schafer and Muyzer (2000). Individual bands that composed the bacterial community fingerprint were identified using Gel Pro 4.0, 1-D gel routine. All bands, including heteroduplexes, were included in the analyses since their appearance was characteristic of the sample and informative. A binary matrix was derived by identifying presence (1) or absence (0) of a band using 1.15% minimum peak height in GelPro Analyzer 4.0 (Media Cybernetics). A total of28 bands in 2001 and 27 bands in 2002 were used for analyses of similarity. Binary matrices were used to calculate Bray- Curtis similarity coefficients (PRIMER v.5.2, Clarke and Gorley 2002). A hierarchical cluster analysis and a non-metric MDS plot (Clarke and Warwick 2001) were used to visualize spatial 15 patterns of variability. To test the similarity within a priori-defined groups, a one-way ANOSIM (PRIMER-E) was applied. ANOSIM applies a non-parametric permutation procedure to test the null hypothesis that there are no differences in similarity between or among test groups (Clarke and Green 1998). ANOSIM was applied because the multivariate dataset did not meet the normality or homoscedasticity assumptions required by parametric statistics. Sequences of Prominent DGGE Bands DGGE bands of interest were selected for reamplification. Direct sequencing of reamplified bands was not reliable. Therefore, selected bands were reamplified with 338F and 5l9R and cloned with a Invitrogen TA cloning kit (Invitrogen CA). Clones were screened by DGGE and run alongside an environmental sample. Clones that contained the desired insert were prepared for sequencing with a mini-prep kit following manufacturer's instructions (CPG inc.). The insert was sequenced with M13 primer at the Center for Gene Research and Biotechnology, Oregon State University, Corvallis, OR. Missing Data There were cases where the PCR amplification did not produce a product that could be visualized on the DGGE gel. In 2001, the following lanes contained no data: KOb, FOe, FOd, FdOc, FdOd, Fdla, Fdlb, Eld. The disturbance treatment caused higher than usual failure of the PCR amplification and therefore no disturbance plot data was collected in either Hidden Creek or Elliott Creek. Results Variability in Bacterial Communities within Sites 16 Six DGGE gels were run in 2001 and used to examine within site (m, cm) spatial heterogeneity ofbacterial communities (Figure 3). After within site comparisons were made, the gels were compared with internal markers and a universal ladder to make between site comparisons (Figure 4). Three lanes containing a marker plus a set of internal standards were used to determine the position ofDGGE bands (Gel Pro 4.0). The marker lane contained four bands whose migration served as reference for the migration distance of unknowns. Bands that were used as internal standards were darkly staining among all estuarine sites. Bacterial communities collected from the Hidden Creek site showed high similarity among fingerprints from a single depth fraction (Figure 4). Hidden Creek 0-1 cm fraction and 1- 2 cm fraction were significantly different (p<0.03, Global R =0.81). The null hypothesis that the genetic composition of the bacterial community was identical in 0-1 cm and 1-2 cm depth fractions was rejected for the Hidden Creek site. The bacterial assemblage from depth fractions 0-1 cm and 1-2 cm in Kunz Marsh (p =0.54, Global R =0.03), Kunz Marsh disturbance plot (p<0.09, Global R = 0.354), Elliott Creek (p = 0.68, Global R 0.056), and Ferrie Ranch (p = 0.27, Global R 0.143), Ferrie Ranch disturbance plot (p = 0.33, Global R 0.0) were not significantly different. The samples collected 3 centimeters apart (a,b or c,d) were not more similar than those collected 3 meters apart (Figure 4). Replicates a,b,c,d were randomly grouped in control plots and therefore there was no visible pattern of organization at the meter scale. No differences were observed between patches and therefore statistical tests were not applied. Samples collected in disturbed plots showed variability between patches, but not within a patch. Samples collected 3 cm apart were more similar then those collected 3 m apart. Disturbed plots in Kunz Marsh (Kd) showed that replicates ~hhad high similarity (80) regardless of the depth fraction (Figure 4). There was variability between patches but not within patch. 17 A Figure 3. DGGE Gel Images from 2001. The first four lanes are replicates a, b, c, and d from the 0-1 cm fraction. The second four lanes are replicates from the 1-2 cm fraction. Lanes marked M were lanes with known sequences and served as a marker. (A) Ferrie Ranch, (B) Ferrie Ranch disturbance plots, (C) Kunz Marsh, (D) Kunz Marsh disturbance, (E) Hidden Creek, (F) Elliott Creek. co M 18 Figure 3 (cont.). DGGE Gel Images from 2001. EF Figure 3 (cont.). DGGE Gel Images from 2001. 19 A 40 z. .~ E60 US '":e :::l (( 80 ~ co $$$++$$+$+ ~ ••••••••++++++~+~~~~~ .... ............ DGGE Samples 20 B + + + Stress: 0.18 ++ •~ •• ~ • ••.... • • • • ~ ~ . 'f ~ ~ Figure 4. Cluster Analysis and MDS from 2001 DGGE Gels. A similarity table was calculated and used to plot dendrogram (A) and non-dimensional MDS plots (B). Site designations are: F, Ferrie Ranch, K, Kunz Marsh, H, Hidden Creek, E, Elliott Creek. Disturbed plot samples are marked with (d), depth fractions are marked: 0 (0-1 cm), or 1 (1-2 cm) and replicate is marked (a,b,c,d). Symbols in the MDS plot correspond to the designations in the cluster dendrogram. 21 Replicates M from the 0-1 cm fraction (80) and M from 1-2 cm fraction grouped together (100). The same pattern was found in Ferrie Ranch, site of a second set of disturbance plots. Disturbance (fd) replicates M grouped together (88) and~ grouped together (85). The number of bands that could be counted from DGGE gels was always higher in disturbance treated samples than in undisturbed samples. Variability in Bacterial Communities within South Slough Estuary Generally, sites exhibited a high degree of site-specificity supported by site groupings in cluster analyses (Figure 4). Similarity within site was highest within Elliott Creek (88), followed by Hidden Creek (78), Ferrie Ranch (68) and Kunz Marsh (52). Similarity among all estuarine sites in 2001 was only 47. Similarity values for comparing estuarine sites were higher when the samples from the disturbance treatment were excluded. A one-way ANOSIM was used to test the null hypothesis that communities of bacteria were identical at the four estuarine sites. The null hypothesis was rejected (pO.64). Variability in Bacterial Communities between Estuarine and Marine Sediments Replicates from two estuarine sites and a marine site were collected and run on a single DGGE gel (Figure 5). Branch tips are labeled according to site (CA, H, K), treatment (D), depth fraction (0,1), and replicate (a,b,c,d). Two marker lanes (M) were also included on the DGGE gel. DGGE bacterial community fingerprints collected in the South Slough estuary were more similar to each other than to marine sediment samples from Cape Arago. Similarity among estuarine sites was 50 while similarity between marine and estuarine sites was 30. 22 A 20 J!> 40'l: IV] (J) Vl 60 '€ ::J