Sediment bacterial communities reflect the history of a sea basin

Abstract

How entire microbial communities are structured across stratified sediments from the historical standpoint is unknown. The Baltic Sea is an ideal research object for historical reconstruction, since it has experienced many fresh- and brackish water periods and is depleted of dissolved oxygen, which increases the sediment's preservation potential. We investigated the bacterial communities, chemical elements (e.g. Cr, Pb Na, P, Sr and U) and sediment composition in a stratified sediment core dated by radiocarbon and spanning 8000 years of Baltic Sea history, using up-to-date multivariate statistics. The communities were analysed by 16S rRNA gene terminal restriction fragment length polymorphism. The communities of the deep Early Litorina and surface Late Litorina Sea laminae were separated from the communities of the middle Litorina Sea laminae, which were associated with elevated concentrations of U and Sr trace elements, palaeo-oxygen and palaeosalinity proxies. Thus, the Litorina Sea laminae were characterized by past oxygen deficiency and salinity increase. The communities of the laminae, bioturbated and homogeneous sediments were differentiated, based on the same historical sea phases, with correct classifications of 90%. Palaeosalinity was one of the major parameters that separated the bacterial communities of the stratified sediments. A discontinuous spatial structure with a surprising increase in community heterogeneity was detected in Litorina Sea sediments from 388 to 422 cm deep, which suggests that a salinity maximum occurred in the central Gulf of Finland app. 6200-6600 years ago. The community heterogeneity decreased from the surface down to 306 cm, which reflected downcore mineralization. The plateau of the decrease was in the app. 2000-year-old sediment layers. Bacterial community data may be used as an additional tool in ocean-drilling projects, in which it is important to detect mineralization plateaus both to determine historically comparable portions of sediment samples and historical events, such as sea-level rise culminations.

Figure 1. Association between bacterial sediment communities and chemical variables in Baltic Sea laminae.

The black-filled circles indicate the sediment samples from the Early Litorina, red from the Litorina and blue from the Late Litorina sea phases. The numbers are rounded to the nearest integer. Canonical analysis of principal coordinates was based on the Bray Curtis dissimilarity matrix (n = 32×158) of bacterial terminal restriction fragments of the 16S rRNA gene. Samples and chemical parameters (purple arrows) were plotted against canonical axis scores 1 and 2. The chemical variables chosen explained 45% of the variation. Test statistics with 9999 permutations resulted in a highly significant p value (p = 1e-04), which allows rejection of the null hypothesis that there are no relationships between the bacterial communities and chemical variables. The length of the arrow indicates the strength of the correlation between the sediment samples and chemical parameter. An arrow direction indicates the increasing concentration of the chemical parameter.

Figure 2. Differentiation of Baltic Sea bacterial communities based on sea phases, depth and sediment composition.

(A) Shea phases: black filled circles indicate sediment samples from the Early Litorina, red from the Litorina and blue from the Late Litorina sea phases. (B) Depth classes: pink-filled circles indicate sediment samples from 533.5 to 500.75 cm, yellow from 494.85 to 403.75 cm, black from 393.7 to 301.35 cm, turquoise from 297.5 to 203.50 cm, blue from 199 to 101 cm and red from 96 to 18.25 cm below the seafloor. (C) Sediment composition: black-filled circles indicate laminae, yellow bioturbated and red homogeneous sediment layers. Generalized discriminant analysis was based on Bray and Curtis distances of bacterial terminal restriction fragments of the 16S rRNA gene (n = 148×199). Correct classification was a) 90.5%, b) 78% and c) 52.7%. Test statistics with 9999 permutations resulted in a highly significant p value for a) and b) (p = 1e-04) which allows rejection of the null hypothesis of no differences among the a priori groups. The default option of canonical analysis of principal coordinates, which combines unconstrained and constrained multivariate methods, resulted in a) 24, b) 23 and c) 13 principal component axes that were further analyzed by generalized discriminant analysis. The first and second canonical axes were plotted here.

Figure 3. Spatial structure of bacterial communities along a sediment sample covering 8000 years of Baltic Sea history.

The piecewise Mantel correlogram was based on the Bray and Curtis dissimilarity matrix (n = 148×219) of bacterial terminal restriction fragments of the 16S rRNA genes. Autocorrelations (Mantel r) were plotted against distance classes (Table S6), which were based on the Euclidean distance between sampling depths. Red, green and black circles represent the distinct trends detected. Filled circle: P<0.05. Open circle: nonsignificant value.

Figure 4. Neighbor-joining phylogenetic tree of the partial 16S rRNA genes from stratified Baltic Sea sediments.

In the inverted circular tree with a stacked bar chart (sea phases), red bars indicate the abundance of a leaf (sequence) found in the Litorina Sea laminae (depths 330 and 422 cm), blue bars the abundance of a leaf found in the Late Litorina Sea laminae (depths 91 and 101 cm) and the black bars the abundance of a leaf found in the Early Litorina lamina (depth 534 cm). The leaves without bars are reference sequences obtained using the NAST tool. Reliable branches are shown in the bootstrap tree (Fig. S5). The clone sequences were assigned to phylum or class level (Proteobacteria) or to genus level (Pseudomonas) by the RDP classifier with an 80% threshold. The putative JS1 clones were assigned, based on the closest sequence matches, using the RDP seqmatch tool. Scale bar represents the genetic distance, i. e. the nucleotide substitutions per site.