Contrasting Effects of Local Environmental and Biogeographic Factors on the Composition and Structure of Bacterial Communities in Arid Monospecific Mangrove Soils

Abstract

Mangrove forests are important biotic sinks of atmospheric CO₂ and play an integral role in nutrient-cycling and decontamination of coastal waters, thereby mitigating climatic and anthropogenic stressors. These services are primarily regulated by the activity of the soil microbiome. To understand how environmental changes may affect this vital part of the ecosystem, it is key to understand the patterns that drive microbial community assembly in mangrove forest soils. High-throughput amplicon sequencing (16S rRNA) was applied on samples from arid Avicennia marina forests across different spatial scales from local to regional. Alongside conventional analyses of community ecology, microbial co-occurrence networks were assessed to investigate differences in composition and structure of the bacterial community. The bacterial community composition varied more strongly along an intertidal gradient within each mangrove forest, than between forests in different geographic regions (Australia/Saudi Arabia). In contrast, co-occurrence networks differed primarily between geographic regions, illustrating that the structure of the bacterial community is not necessarily linked to its composition. The local diversity in mangrove forest soils may have important implications for the quantification of biogeochemical processes and is important to consider when planning restoration activities. IMPORTANCE Mangrove ecosystems are increasingly being recognized for their potential to sequester atmospheric carbon, thereby mitigating the effects of anthropogenically driven greenhouse gas emissions. The bacterial community in the soils plays an important role in the breakdown and recycling of carbon and other nutrients. To assess and predict changes in carbon storage, it is important to understand how the bacterial community is shaped by its environment. Here, we compared the bacterial communities of mangrove forests on different spatial scales, from local within-forest to biogeographic comparisons. The bacterial community composition differed more between distinct intertidal zones of the same forest than between forests in distant geographic regions. The calculated network structure of theoretically interacting bacteria, however, differed most between the geographic regions. Our findings highlight the importance of local environmental factors in shaping the microbial soil community in mangroves and highlight a disconnect between community composition and structure in microbial soil assemblages.

Keywords: 16S rRNA; co-occurrence network analysis; community assembly; community structure; ecological processes; microbial biogeography; microbiome.

FIG 1

Within sample diversity of bacterial communities from arid mangrove soils. Alpha diversity measures of the bacterial communities detected in mangrove soils from Australia (green outline) and Saudi Arabia (orange outline) across zones (fringe, shrub) and depths (surface in light gray, subsurface in dark gray). (A) Rank-abundance relationship of ASVs by location. (B) Species richness described by the number of observed ASVs. (C) Shannon diversity index. (D) Phylogenetic diversity index (Faith’s PD). The boxplots indicate the median with the interquartile range (IQR) between the 25th and the 75th percentile and the whiskers extend 1.5*IQR. Boxes were plotted with the notch (+/- 1.58*IQR/sqrt[n]) to display likely statistical significance if notches do not overlap.

FIG 2

Comparison of different bacterial communities between the experimental factors geographic region, exposure, zone, and depth. (A) Relative abundance of phyla across the whole data set. The top 12 most prevalent phyla were displayed and the remaining grouped as “Other.” The most abundant classes within the Proteobacteria were additionally included (blue colors). (B) Schematic representation of the R2-values from the PERMANOVA. The large bubble represents the proportion of variation explained by the interaction of all experimental factors. The small bubbles show the proportion of variation explained by each individual factor. The arrow above indicates the environmental factors influencing the separation. (C) Nonmetric multidimensional scaling (NMDS) plots of the community matrix across experimental factors at ASV level, and (D) at genus level across geographic region, exposure, zone and depth.

FIG 3

Selected network metrics by phylum. (A) Betweenness centrality and (B) Topological coefficient. The 12 most prevalent phyla (within the networks) were selected and the remaining grouped as “Other”. Mean scores are displayed with standard errors calculated by the stat_summary function in ggplot2. Bars without standard errors display single occurrences.

FIG 4

Comparison of high degree scores in co-occurrence networks. Density plots of the relative proportion of nodes against the normalized node degree colored by phylum across geographic region, exposure, zone and depth. The proportion of nodes is calculated by the Kernel-Density function.

FIG 5

Theoretical functional assignments. Heat map showing the distributions of bacterial functions that were assigned by FAPROTAX across geographic region, exposure, zone and depth. Values were log-transformed with lighter values indicating higher abundances.

FIG 6

Map and sampling design of the nested factors. World map showing the global regions of the sampling sites on the Western Australian coast and in Saudi Arabia. The more detailed maps show the exposed and the sheltered sites in both geographic regions. An aerial photograph visualizes the separation between the zones of each forest, with actual sampling locations at the sheltered site in Saudi Arabia separated into three plots each at the tall fringe and the shrub as an example for sampling separation within each of the zones. The schematic representation of the sampling design below, shows the arrangement of the experimental factors and the numbers of replicates taken. The values in brackets represent the number of replicates for each sample. The numbers in the last row denote the sampling depth of the surface 0–2 cm (Surf) and the subsurface 2–10 cm (Sub). Asterisks (*) represents from which site a sample that has been removed for insufficient sequencing depth or cross-contamination, making the total number of samples n = 115.