Barking up the right tree: ecological insights into the microbiome of bald cypress tree bark

Abstract

Background

Trees and their associated microbes provide numerous ecosystem services including carbon sequestration, nutrient cycling and phytoremediation. Tree bark represents a large and seasonably stable habitat for microbial communities. However, the tree bark microbiome remains largely understudied, particularly for wetland tree species. In the Lower Mississippi River Basin, bald cypress (Taxodium distichum) are the predominant tree species in many wetlands, including lakes and streams connected to large agroecosystems dominated by row-crop agriculture. These water bodies are often managed for irrigation and drainage needs and are subject to agrochemical runoff from adjacent fields. Thus, we sought to understand how hydrology affects the bald cypress bark microbiome.

Results

We collected 278 bark samples over six months from 18 trees located in three different lakes. Using 16S rRNA gene sequencing, we found that the bald cypress tree bark microbiome was largely consistent between trees within a lake as well as between different lakes, with a core microbiome that includes bacterial taxa that were present in over 95% of samples collected. Hydrology had a significant influence on microbiome structure, with different sections of bark having distinct bacterial communities depending on if the bark was submerged, just above the water, or dry. Water quality was significantly correlated with alpha diversity of wet bark, which was more diverse than dry bark and had higher relative abundances of bacteria that may be providing relevant ecosystem services such as denitrification, methane oxidation, and pollutant degradation.

Conclusions

Wetlands are important for nutrient cycling and water quality regulation. Our study provides insights into microbial dynamics of these ecosystems and how hydrology can impact the microbial communities present, which in turn may be impacting water quality. This work is the first to the describe the bark microbiome of a wetland tree species and lays the groundwork for future studies assessing the functional role of the microbiome in wetland ecosystem services.

Supplementary Information

The online version contains supplementary material available at 10.1186/s40793-026-00862-2.

Background

Bald cypress (Taxodium distichum) is a deciduous conifer endemic to the southeastern United States, where it is the dominant tree species in many coastal wetlands and shallow lakes. Bald cypress grow in both continuously or seasonally flooded habitats and can experience extreme changes in hydrology from natural and anthropogenic causes [1, 2]. In Mississippi, bald cypress often grow in wetlands and shallow agricultural lakes that are highly eutrophic and receive agrochemical runoff from adjacent agricultural fields. These field-adjacent water bodies may be heavily managed for water storage and irrigation needs [3]. Excessive water level fluctuations and low water levels can impact ecosystem function and water quality, resulting in increased turbidity and eutrophication [4, 5]. This has the potential to impact the microbial communities present on bald cypress trees, and in turn those same communities may have an impact on water quality through nutrient processing and bioremediation of agrochemicals.

Trees are a dominant and crucial component of many ecosystems, with an estimated three trillion trees globally [6]. The various ecological, economical, and social benefits they provide are manifold and generally well-studied. It is estimated that a single tree can harbor a trillion microbes in its above ground tissues [7]. However, their role as a microbial habitat, particularly in above ground tissues other than leaves, remains understudied. Tree bark provides a large, and seasonally stable, surface area for microbial communities to form [8]. For bald cypress growing in wetlands, a large portion of their tree trunk will be submerged in water either continuously or seasonally. This provides ample opportunity for microbial community formation on bark and for microbes to interact with chemicals in the water column.

Numerous ecosystem services, including nutrient cycling and pollutant degradation, are driven by microbial communities present in the soil, water, and on plants [9–12]. Plant-associated microbes have long been recognized for their pivotal role in plant health, but there is increasing evidence for their role in phytoremediation of atmospheric pollution [13], pesticides [14, 15], and other pollutants [9, 10, 16]. Plant associated methanotrophic bacteria can play a role in carbon cycling and reduction of methane emissions from peatlands [17] and tree stems [18, 19]. Despite the importance of plant-based microbes in ecosystem function, there have been limited investigations into tree bark microbial communities using high-throughput sequencing. Moreover, none of the studies to date have examined a wetland tree species, and have focused on fruit trees [20–25], diseased forest trees [26–29], and other non-wetland species [30–34].

In this study, we sought to understand how the bark microbiome of bald cypress trees responds to hydrochemical dynamics. We collected bark samples (n = 278) over a six‑month period from 18 trees located in three oxbow lakes in the Lower Mississippi River Basin (LMRB; Fig. 1). Beasley Lake and Roundaway Lake served as the primary study sites because they are representative of the shallow (depth < 3 m), turbid, and eutrophic lakes common in the region [3, 35]. These lakes have also been the focus of long‑term monitoring by our research group, with water samples collected bi-weekly to evaluate how hydrology and agricultural practices influence water quality [36–38]. Although not part of the original study design, Sky Lake was later included when an unusual drought provided an opportunity to sample some of the oldest bald cypress trees in Mississippi. Additionally, a subset of soil, water, and Bald Cypress leaf samples (n = 16) were collected to compare their community composition to that of adjacent bark samples.

Fig. 1.

Bald cypress trees were sampled from three lakes (Beasley Lake (B), Roundaway Lake (C), Sky Lake (D)) in the Mississippi Delta subregion within the Lower Mississippi River Basin (A). Six trees were samples from each lake (locations numbered 1–6). Bark from each tree was collected from dry, splash, and submerged sections

From each tree, bark was collected from three positions: below the water surface, above the waterline, and at the water–air interface (the “splash zone”). We used 16S rRNA gene sequencing to assess how inundation conditions shape epiphytic bacterial communities. We hypothesized that the three bark sections (dry, splash, and submerged) would harbor bacterial communities that were distinct from one another. For Beasley Lake and Roundaway Lake, we also sought to identify abiotic drivers of community composition by correlating water‑quality parameters with alpha diversity. In addition, we considered the potential ecosystem services provided by these communities within the context of environmental and management factors. Finally, we analyzed bark samples from all three lakes collectively to identify a core bald cypress microbiome.

Methods

Bark collection overview

Bark samples were collected from three oxbow lakes in the Mississippi Delta subregion of the LMRB (Fig. 1A, and dark gray outline of Fig. 1A inset): Beasley Lake (33°23’53.08”N, 90°40’36.56 W, Fig. 1B), Roundaway Lake (34°01’06.2”N, 90°35’41.4”W, Fig. 1C), and Sky Lake (33°16’58.55”N, 90°30’19.19”W, Fig. 1D). Six trees were repeatedly sampled at each lake, unless stated otherwise, with three bark sections sampled at each tree: a dry bark Section (80 cm above the water line), a splash Sect. (20 cm above the water line), and a submerged Sects. (2–20 cm below the water; Fig. 1). Bark was collected from the south side of the tree by either peeling off the outer bark with sterile gloves or with an ethanol-sterilized chisel. An approximately 8 × 5 cm section of bark was removed for each sample and placed into a sterile Whirl-Pak bag. Bagged samples were immediately placed into a cooler with dry ice and transferred to a -80° C freezer upon return to the laboratory. GPS coordinates were recorded for each tree and labeled flagging tape was tied to a branch to ensure the same trees were sampled each time.

Study sites and sampling

Beasley Lake is a 25-ha oxbow lake formed by the adjacent Sunflower River (Fig. 1B). It is located within a 625-ha watershed in Sunflower County, MS that is predominantly row crop acreage in addition to a 125-ha forested riparian wetland. Numerous studies on water quality have been conducted at Beasley Lake [39–42], and it is subject to long-term monitoring as part of the Conservation Effects Assessment Project (CEAP) and recently the USDA-ARS Long-Term Agroecosystem Research (LTAR) Network [43]. Because of this, biweekly monitoring of water quality has occurred for over 25 years [44]. Tree bark was sampled monthly from 11/27/2023 to 7/1/2024, except for January and February because of inaccessible winter conditions. Bark collection was conducted via boat and occurred on days in which water samples were collected for routine water quality monitoring. Two trees were sampled on east and west ends of the lake, along with two trees towards the middle of the lake (Fig. 1B). Additionally, bald cypress leaves were collected on 5/6/2024 in sterile Whirl-Pak bags that were immediately placed on dry ice. In late May 2024, a beaver dam causing the water level to be abnormally high was scheduled for removal. This provided the opportunity to assess how a rapid drop in water levels would affect bark microbial communities. On 5/29/2024, 1 h before the dam was destroyed, bark samples were collected from all six trees along with 250 mL water samples, which were collected in bleach-sterilized Nalgene bottles. Samples were collected again on 6/3/2024 when the water level had dropped > 50 cm. At this time, submerged bark could not be collected, so an extra sample was taken approximately 40 cm above the water line. This bark would have been submerged for several months prior to the rapid drop in water level.

Roundaway Lake is an approximately 160-ha riverine oxbow lake located in Coahoma County, MS (Fig. 1C). It has a generally U-shaped channel that is deepest in the middle with steeply sloping banks lined with bald cypress trees. Water level is controlled by an adjustable weir [38]. Several studies assessing water quality have been conducted at this site [3, 37, 38, 45], and water quality monitoring occurs bi-weekly throughout the year. Six trees located on the south end of the lake were sampled monthly from 12/11/2023 through 5/20/2024, although submerged bark sections could not be collected on 12/11/2023 due to low water levels. Samples were collected on days that coincided with routine water quality sampling. Bald cypress leaves were collected from three trees on 4/22/2024 and 5/20/2024.

Sky Lake is an oxbow lake-wetland system in Humphreys County, MS (Fig. 1D). The lake is bordered by a forested riparian fringe containing some of the oldest, and largest, bald cypress trees in Mississippi. The lake receives runoff from approximately 1,900 ha of primarily agricultural land. The riparian fringe is flooded most of the year and is accessed via a boardwalk or boat, although it can go dry during periods of drought. The lake and bald cypress trees have been the subject of several studies [46–49]. Bark samples were first collected 12/6/2023, when the riparian fringe was dry with no standing water, and accessible by foot. Bark was collected from four trees and three samples were collected per tree: one at the base of the tree (within 20 cm of the soil line), one at 1.6 m, and one at ~ 3.8 m (an area typically above the water line). Additionally, soil samples were collected near the base of each tree by pushing a sterile 50 mL centrifuge tube through the top 5 cm of soil. Bark was collected from the same four trees again on 2/13/2024 when the trees were in approximately 2.2 to 2.8 m of standing water. On 2/20/2024, we collected bark from an additional two trees at the Delta Wind Birds Nature Reserve at Sky Lake (https://www.deltawindbirds.org/initiatives/sky-lake). All six trees were sampled once more on 4/23/2024, along with a 250 mL water sample that was collected from around each tree.

Water quality data

In-situ water quality parameters [pH, dissolved oxygen (DO), and conductivity] were measured with a calibrated Yellow Springs Instrument (YSI) Professional Plus multimeter. Water samples (1 L) were collected in the field, transported to the laboratory on ice, and analyzed for turbidity, total dissolved organic carbon (TDOC), and total nitrogen (TN) using a Hach 2100 P Turbidimeter, Shimadzu TOC-L, and flow injection analysis (FIALab), respectively. Each procedure was conducted according to Standard Methods [50].

DNA extraction and sequencing

Each thawed bark sample was transferred to a 50 mL centrifuge tube and 15 mL of cold sterile phosphate buffered saline (PBS; Gibco) added. The tube was vortexed for 30s and the PBS washing solution transferred to a 15 mL centrifuge tube. The washing solution was then centrifuged at 2147 RCF for 10 min at 4 °C. The resulting supernatant was decanted, and the pellet resuspended in 800 µL of solution CD1 from the Dneasy PowerSoil Pro Kit (Qiagen). The resuspended pellet was transferred to a PowerSoil bead tube and processed according to the manufacturer’s protocol. Bald cypress leaves were initially processed in an identical manner but yielded insufficient DNA. This was remedied by omitting steps 5–6 (inhibitor removal) of the PowerSoil Kit. Soil samples were mixed and passed through a 2-mm sieve to remove rocks and plant litter and 0.5 g of soil per sample was used for DNA extraction with the PowerSoil Kit. Water samples were filtered through 0.22 μm membrane filters (Millipore) and DNA extracted with the DNeasy PowerWater Kit (Qiagen) according to the manufacture’s protocol.

Controls consisted of three field blanks of sterile cotton swabs waved in the air for 10 s while outdoors collecting bark samples; three laboratory blanks of empty PowerSoil kit reactions; and three replicates each of the ZymoBIOMICS Microbial community Standard and ZymoBIOMICS Microbial Community DNA Standard. Field blanks and mock communities were processed with the PowerSoil kit according to the manufacture’s protocol with the exception that steps 5–6 (inhibitor removal) were skipped.

Purity of extracted DNA was assessed with a Nanodrop 2000 spectrophotometer (Thermo Scientific) and yield quantified with a Qubit 4 using the 1x dsDNA broad-range kit (Invitrogen). DNA samples were normalized to 10 ng/uL and sent to the University of Mississippi Medical Center, Molecular & Genomics Core Facility (Jackson, MS) for amplification and sequencing of the V3-V4 region of 16S rRNA gene. The 300 bp paired-end sequencing was performed on an Illumina NextSeq1000 using XLEAP P1 reagents.

Bioinformatics

Demultiplexed reads were processed in R (v4.4.1) using the DADA2 (v1.34.0) pipeline (https://bioconductor.org/packages/devel/bioc/vignettes/dada2/inst/doc/dada2-intro.html) [51, 52]. The quality scores of the forward and reverse reads were visualized using plotQualityProfile() and low-quality sequences removed with filterAndTrim(truncLen = c(290, 200), maxN = 0, maxEE = 1). Taxonomy was assigned to amplicon sequence variants (ASVs) using the Silva nr99 v138.1 reference database [53]. Only ASVs assigned to the Kingdom “Bacteria” were retained and chloroplast or mitochondria were removed along with ASVs with only 1 or 2 total counts. The resulting ASV count table and taxonomy were imported into phyloseq (v1.50.0) [54]. Negative control samples were used to identify and remove contaminant DNA sequences using decontam (v1.26.0) [55]. To account for uneven sequencing depth between samples, the phyloseq::rarefy_even_depth() function was used to normalize samples to 15,332 sequences each. Alpha diversity measures (Observed number of species, Shannon and InvSimpson) were estimated using these rarified ASV counts in phyloseq and visualized with ggplot2 (v3.5.1) [56]. Both Shannon and Inverse Simpson (InvSimpson) take species richness and evenness into account, with the InvSimpson index placing more emphasis on evenness [57]. For beta diversity, rarefied ASV counts were transformed to relative abundance and Bray-Curtis dissimilarity was calculated using phyloseq::distance() and visualized with phyloseq::plot_ordination(). To calculate weighted-unifrac, ASV sequences were aligned with mafft (v7.505), a phylogenetic tree constructed with fasttreeemp (v2.1.11) and loaded into R with phytools (v2.4.4) [58–60].

Taxonomy was explored by aggregating the ASVs at either the phylum or genus level using the tax_glom() function from speedyseq (v0.5.3.9021) [61]. Bar plots were constructed using phyloseq::plot_bar(). Core taxa were identified using the core() function in the microbiome (v1.28.0) package [62]. The idea of a “core microbiome” is not well defined [63, 64], so we defined it as genera that were present in > 95% of samples at a minimum relative abundance of 0.033%. The abundance level was chosen as only 40 Mbp of sequencing would be required to get 30x coverage of the 440 bp V3-V4 region (440 bp x 30 × 100 / 40 Mbp = 0.033). This approach was used to define the core microbiome for all bald cypress bark samples collected, as well as for each lake and bark sample type. Upset plots were constructed with ComplexUpset (v1.3.3) to visual overlaps in core genera between sample types and locations [65]. The relative abundance of methane oxidizing bacteria (MOB) in each sample type was determined from rarified counts of known MOB. This included the families Methylococcaceae, Methylomirabilaceae, and Methylomonadaceae, as well as the genera Methylocystis, Methylocella, Methyloferula, Methylocapsa, Methylosinus, and Methyloceanibacter. Additional data wrangling, processing, and graphing assistance was provided by the packages tidyverse (v2.0.0), biostrings (v2.74.1), genefilter (v1.86.0), cowplot (v1.1.3), and ggpubr (v0.6.0) [66–70].

Statistical analysis

Significant differences in alpha diversity metrics between sample types and locations were determined using the nonparametric Kruskal-Wallis test followed by FDR-adjusted Wilcoxon rank sum tests. Permutational multivariate analysis of variance (PERMANOVA) tests were employed to assess differences in community structure using Bray-Curtis dissimilarities and weighted UniFrac distances via the adonis2 function from vegan (v2.6.8) with 9999 permutations [71]. Bray-Curtis dissimilarity matrices were used with the vegan::betadisper() function to calculate the homogeneity of group dispersions using spatial medians and significant differences were identified with vegan::permutest(). Bacteria that are differentially abundant between sample types or lakes were identified with DESeq2 (v1.44.0) using a negative binomial generalized linear model [72]. Unrarefied counts that were aggregated at the genus level were used in the analysis. Genera were considered significantly differentially abundant if they had a Benjamini-Hochberg FDR adjusted p-value < 0.05 and an absolute log2-fold change > 1. The influence of water quality metrics on tree bark alpha diversity was assessed with linear mixed effect models using lmerTest (v3.1-3) which provides approximate p-values [73]. Only splash and submerged bark samples from Beasley and Roundaway lakes that had time matched water quality data were included (n = 123). The height at which the sample was collected (cm from the base of the tree) was included as a predictor variable. Model quality and collinearity between predictor variables was evaluated with performance (v0.13.0) and correlation heatmaps constructed with corrplot (v0.95) [74, 75]. Predictor variables were set as fixed effects with tree ID and site as random effects (e.g. y ~ w + x + z + (1|TreeNumber) + (1|Site)). Predictor variables were transformed to z-scores using R’s built-in score() function to improve model performance and interpretability [76].

Results

Sequencing summary

A total of 294 samples were sequenced (278 bark, 8 water, 5 leaf, 3 soil) along with 12 controls. Excluding controls, an average of 215,502 ± 32,685 sequences were generated per sample. The six mock community controls had an average of 182,561 + 29,171 sequences. The three field blanks and three lab extraction blanks had an average of 23,975 ± 26,359 sequences. After the initial screening, 163,078 ASVs were identified, which was reduced to 124,508 ASVs after the samples were normalized to 15,332 sequences to account for differences in sequencing depth.

Bark community composition at Beasley lake and Roundaway lake

Bacterial communities on bald cypress tree bark varied significantly between bark sections with regard to all three alpha diversity metrics at Beasley Lake and Roundaway Lake (Fig. 2). Submerged bark exhibited the highest average alpha diversity (Beasley: Observed = 3022 ± 501; Shannon = 7.1 ± 0.4; InvSimpson = 404.2 ± 252.4, Roundaway: Observed = 2711 ± 460; Shannon = 6.8 ± 0.5; InvSimpson = 247.8 ± 180.9) while dry bark consistently had the lowest alpha diversity (Beasley: Observed = 1345 ± 338; Shannon = 6.1 ± 0.4; InvSimpson = 174.4 ± 73.3, Roundaway: Observed = 1264 ± 407; Shannon = 5.9 ± 0.4; InvSimpson = 151.3 ± 90.2). Splash bark had alpha diversity metrics that were significantly different than dry or submerged sections (Beasley: Observed = 2284 ± 670; Shannon = 6.7 ± 0.6; InvSimpson = 302.1 ± 175.8, Roundaway: Observed = 2465 ± 764; Shannon = 6.3 ± 0.4; InvSimpson = 213.2 ± 239.9).

Fig. 2.

Alpha diversity (Observed richness, Shannon, and Inverse Simpson) of bacterial communities on dry, splash, and submerged bald cypress bark collected from Beasley Lake (A) and Roundaway Lake (B). Diversity metrics were calculated from normalized 16 S rRNA gene sequences. Asterisks represent significant differences between bark sections (Wilcoxon; * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.001)

The diversity of dry bark sections remained consistent across the six months of sample collection, while the diversity of the splash and submerged sections increased with time from Winter to Spring/Summer (Supp Fig. S1 & S2).

Non-metric multidimensional scaling (NMDS) ordination showed that samples clustered according to bark section (Fig. 3) for both Beasley Lake (Bray-Curtis: R² = 0.22, p = 0.0001; UniFrac: R² = 0.50, p = 0.0001) and Roundaway Lake (Bray-Curtis: R² = 0.25, p = 0.0001; UniFrac: R² = 0.54, p = 0.0001). Significant differences between bacterial communities in the splash and submerged sections were found using pairwise PERMANOVA (p = 0.0048). Differences in multivariate dispersion between bark sections were significant for Beasley Lake (betadisper permutest, F = 15.41, p = 0.001) but not for Roundaway Lake (betadisper permutest, F = 0.95, p = 0.372). Comparison of individual trees within a lake found no significant differences except for dry bark sections, which were significantly different between trees in both Beasley Lake (Bray-Curtis: R² = 0.20, p = 0.0001) and Roundaway Lake (Bray-Curtis: R² = 0.34, p = 0.0001).

Fig. 3.

Non-metric multidimensional scaling (NMDS) ordination of bacterial genera present on dry, splash, and submerged bald cypress tree back sections. PERMANOVA analysis identified significant differences in bacterial community structure between bark sections at both Beasley Lake: A Bray-Curtis dissimilarity (R² = 0.22, p = 0.0001), B weighted UniFrac (R² = 0.50, p = 0.0001) and Roundaway Lake: C Bray-Curtis dissimilarity (Bray-Curtis: R² = 0.25, p = 0.0001), (D) weighted UniFrac (R² = 0.54, p = 0.0001)

At the phylum level, bacterial communities in the different bark sections showed consistent patterns both between trees within a lake, and between lakes. Proteobacteria was the phylum with the most sequences in all three bark sections for both Beasley Lake and Roundaway Lake (Fig. 4), accounting for 25–32% of the sequences recovered. Plantomycetota, Actinobacteriota, and Verrucomicrobiota were also major components of all bark sections in both lakes. Acidobacteriota was a greater component of the bacterial community in dry bark Sects. (27–28% of sequences) but less so in the splash zone (4–11%) and submerged Sects. (4–5%), which had higher percentages of Bacteroidota, Chloroflexi, Cyanobacteria, and Firmicutes (Fig. 4).

Fig. 4.

The relative abundance (%) of the ten most abundant bacterial phyla on dry, splash, and submerged bald cypress tree bark from Beasley Lake and Roundaway Lake

More variation in bacterial community composition was seen at the genus level, particularly for the splash and submerged sections (Fig. 5). Dry sections of bark had high relative abundances of sequences identified as Bryobacter (Acidobacteriota; 13.8–14.8%), Bryocella (Acidobacteriota; 6.3–6.7%), Acidiphilium (Proteobacteria; 5.3–6.7%), Tundrisphaera (Planctomycetota; 3.7–4.3%), and 1174-901-12 (Proteobacteria; 3.4–3.6%), an undescribed genus in the order Rhizobiales. This contrasts with submerged bark sections, which had significantly lower relative abundance of Bryobacter (DESeq2; log2-FC = -3.85, p < 0.0001), Bryocella (DESeq2; log2-FC = -7.65, p < 0.0001), Acidiphilium (DESeq2; log2-FC = -4.62, p < 0.0001), and 1174-901-12 (DESeq2; log2-FC = -4.83, p < 0.0001).

Fig. 5.

Average relative abundance (%) of the 15 most abundant bacterial genera present on dry, splash, and submerged bald cypress tree bark collected from Beasley Lake (top row) and Roundaway Lake (bottom row)

Compared to the dry bark sections, the splash and submerged bark sections had significantly higher relative abundances of cyanobacterial genera (Wilmottia, Leptolyngbya, Cyanobium, Tychonema, Calothrix, Pseudanabaena, and Nostoc), pollutant degraders (Exiguobacterium, Sphingorhabdus, Phenylobacterium, Nocardioides, Sphingomonas, Novosphingobium, Rhodopseudomonas, Steroidobacter, Georgfuchsia, Agromyces, Pseudarthrobacter), ammonia oxidizing and denitrifying bacteria (Bacillus, Massilia, Arenimonas, Hyphomicrobium, Rhodoferax, Candidatus Anammoximicrobium, Nitrospira, and numerous members of the Nitrosomonadaceae), and polyphosphate accumulating organisms (Dechloromonas, Microlunatus) (Fig. 6A–B, S3 and S4). Bacterial genera that oxidize methane were also at higher relative abundances in splash and submerged bark sections (Fig. 6C).

Fig. 6.

Average relative abundance (%) of genera identified as putative pollutant degraders (A), denitrifiers (B), or methane oxidizing bacteria (C) from dry, splash, and submerged bald cypress tree bark

Samples (n = 254) from Beasley Lake, Roundaway Lake, and Sky Lake were analyzed together. Asterisks represent significant differences between bark sections (Wilcoxon; * p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.001).

The effects of water quality on bacterial alpha diversity at Beasley lake and Roundaway lake

Total nitrogen was the only parameter that had a significant positive influence on bacterial alpha diversity, being positively correlated with both Shannon diversity and InvSimpson (Table 1). Specific conductivity (serving as a proxy for salinity), pH and TDOC had a significant negative influence on all three alpha diversity metrics, while turbidity was negatively related to Shannon diversity and InvSimpson but not observed richness. The height on the tree in which the bark sample was taken was negatively related to observed richness and Shannon diversity, but its effect on InvSimpson was not significant.

Table 1.

Linear mixed model of water quality parameters affecting wet bald Cypress tree bark alpha diversity (Observed Richness, Shannon, and inverse Simpson Diversity)

Effect	Estimate	Std. error	t value	P
Richness
(Intercept)	2511	291	8.62	0.10
Height	− 388	84	− 4.63	6.09e−05
pH	− 155	67	− 2.33	0.02
Turbidity	− 106	54	− 1.95	0.05
TN	13	114	0.12	0.91
DO	− 8	58	− 0.14	0.89
TDOC	− 237	99	− 2.41	0.02
Conductivity	− 499	74	− 6.73	9.48e−10

Estimate	Std. error	t value	P
Shannon
6.71	0.13	51.45	0.06
− 0.26	0.07	− 3.87	4.30e−04
− 0.20	0.06	− 3.63	4.43e−04
− 0.21	0.05	− 4.61	1.12e−05
0.27	0.09	2.88	5.49e−03
0.03	0.05	0.71	0.48
− 0.35	0.08	− 4.18	6.11e−05
− 0.49	0.06	− 8.03	7.55e−11

Effect	Estimate	Std. Error	t value	P
Inverse Simpson
(Intercept)	265.35	36.65	7.24	0.31
Height	− 44.37	26.22	− 1.69	0.10
pH	− 72.96	21.89	− 3.33	1.18e−03
Turbidity	− 105.81	18.33	− 5.77	7.73e−08
TN	167.43	36.28	4.61	5.12e−05
DO	5.20	19.58	0.27	0.79
TDOC	− 182.70	33.22	− 5.50	2.69e−07
Conductivity	− 57.78	23.20	− 2.49	0.02

TN, total nitrogen; DO, dissolved oxygen; TDOC, total dissolved organic carbon

Bold p-values indicate statistical significance

Bark community composition at Sky lake

Similar to what was observed at Beasley and Roundaway lakes, dry bark sections from Sky Lake had the lowest alpha diversity and high relative abundances of Bryobacter, Bryocella, Acidiphilium, Tundrisphaera, and 1174-901-12 (Fig. 7). Bacterial communities on bark collected at the base of the tree and at 1.6 m, when the forested riparian fringe was unusually dry, was more similar to communities on submerged bark samples collected from Beasley and Roundaway in terms of alpha diversity and taxonomy. When the forest was flooded with > 2.2 m of water, the splash and submerged bark sections still had high relative abundances of Bryobacter, Bryocella, Acidiphilium, Tundrisphaera, and 1174-901-12, and the three bark sections weren’t as dissimilar in alpha and beta diversity, as observed at Beasley and Roundaway lakes. In February and April, the alpha diversity of dry, splash, and submerged bark sections was not significantly different (Kruskal-Wallis, p > 0.05).

Fig. 7.

A Observed richness of bacterial communities present on bald cypress tree bark collected at Sky Lake. Note that samples from December (left box) were collected when no standing water was present, and the labels refer to different heights from the base of the tree (High = 3.8 m, Medium = 1.6 m, and Low = 20 cm). B Relative abundance (%) of the 15 most abundant bacterial genera present on Sky Lake bald cypress tree bark

The core microbiome of bald cypress tree bark

Across all samples, 10 bacterial genera (Bryobacter, Methylobacterium-Methylorubrum, Sphingomonas, Candidatus Solibacter, Conexibacter, Tundrisphaera, Mycobacterium, Aquisphaera, Gemmata, and Chthoniobacter) were found in > 95% of bark sections. At a finer scale, 10–28 bacterial genera were identified in the core microbiome when samples were partitioned by bark section and lake (Fig. 8, Supp Table S2). Of the 10 genera identified in the overall core microbiome, five were also identified as members of the core microbiome when the different bark sections and lakes were considered individually (Bryobacter, Conexibacter, Tundrisphaera, Gemmata, and Chthoniobacter). Submerged bark, collected from all three lakes, had the highest number of core genera at 28, and the highest number of core members at 15 that were only present in that section.

Fig. 8.

Upset plot of the core microbiome of bald cypress tree bark. Horizontal bars (left) indicate the total number of core bacterial genera in each bark section or lake. Vertical bars (top) represent the number of core bacterial genera that are unique or shared among bark sections or lakes

Comparison of leaf, water, and soil samples

Bacterial communities in leaf samples, and adjacent water and soil, had significant differences in alpha diversity (Wilcoxon, padj < 0.05) compared to time-matched bark samples, and were clearly discriminated via ordination (Supp Figs S3–S5). Bacterial communities in water were dominated by Cyanobacteria, and Chthoniobacter and Mycobacterium were amongst the 10 most common genera. Soil samples from Sky Lake had the highest average alpha diversity (Observed = 3691 ± 306, Shannon = 7.6 ± 0.02, InvSimpson = 779.6 ± 305.5). However, the genus accounting for the most sequences was Candidatus Solibacter (2.9%), and several other genera identified as core members of the bald cypress microbiome were prevalent, including Mycobacterium (1.4%), Bryobacter (1.2%), Aquisphaera (1.2%), and Sphingomonas (1.1%). Leaf samples exhibited the lowest alpha diversity (Observed = 1035 ± 63, Shannon = 5.7 ± 0.3, InvSimpson = 116.1 ± 63.2) and were taxonomically most similar to the dry bark sections. Genus 1174-901-12 had the highest relative abundance (14–15%), with Acidiphilium (2.5–2.7%), Bryocella (2.5–2.7%), Methylobacterium-Methylorubrum (3.4–4.7%) and Sphingomonas (3.2–3.5%) also being in the ten most relatively abundant genera in leaf communities.

Discussion

Our initial prediction for this study was that different sections of bald cypress tree bark would have distinct bacterial communities depending on whether the bark was submerged, in the splash zone just above the water surface, or dry. Our results clearly support this, with differences in community composition and diversity between sections. As lake water level rises, bark communities likely encounter free living aquatic microbes and sediment-bound communities, such that splash and submerged bark likely has the opportunity to be colonized from a greater diversity of source communities. The three lakes in this study are highly turbid and noticeable amounts of sediment were recovered from nearly every bark sample processed, including those from the dry sections. This may explain the prevalence of soil-associated bacterial taxa in our bark samples, as well as water and plant-associated taxa. Water may also homogenize the bacterial communities by acting as a more efficient conduit for the transfer of microbes between trees than air. This would explain why dry bark sections had different community structure between trees within a lake, but splash and submerged sections did not.

Rising water levels can also provide bark microbial communities with access to nutrients in the water column and may be buffer these communities from changes in temperature, light, moisture, and nutrients that dry bark communities experience. Alternatively, wet bark communities are also exposed to a more dynamic environment in terms of pH, salinity, and oxygen availability. We observed that TN, TDOC, pH, specific conductivity, and turbidity could all be related to bark alpha diversity. Of these, only TN had a positive effect. This largely aligns with what has been observed in other studies examining effects of water quality on microbial diversity [77–80].

The microbial communities on bald cypress tree bark must be able to survive prolonged cycles of flooding and drought. Despite this, we observed that bark bacterial communities were quite stable over short-term, or even seasonal, changes in water level. This was most evident in Sky Lake, which went from no standing water in December to the trees being submerged in > 2 m of water by February. Although the bark was dry, December samples collected at the base of the tree and at 1.6 m had similar alpha diversity to wet bark from Beasley Lake and Roundaway Lake, and had lower relative abundance of 1174-901-12, Bryocella, Bryobacter, Acidiphilium, and Candidatus Solibacter, genera that are characteristically dominant in dry bark samples. Those same genera were observed at high relative abundances in all three bark sections collected at Sky Lake when it was flooded in February and April. Additionally, no significant differences in alpha diversity were found between bark sections collected in those months. This suggests some persistent vertical stratification of cypress bark microbial communities that can withstand seasonal flooding dynamics. Hydrology may also explain the differences in alpha diversity and beta dispersion found between Beasley Lake and Roundaway Lake. During field collection, we noted that Roundaway Lake experiences more wind-driven wave action and larger quantities of sediment were encrusted on the splash section than what was observed at Beasley Lake or Sky Lake. More frequent splashing may lessen the difference between splash and submerged bark sections and explain why alpha diversity and beta dispersion, as measured by average distance to the median, was not significantly different between the two.

A novel aspect of our study is the wetland habitat of bald cypress. The majority of this habitat has been converted to farmland over last century [81], and what remains is now fragmented and “embedded within an agricultural matrix” [82]. The three lakes in this study, like many in the LMRB, are eutrophic and subject to agricultural runoff from adjacent fields [38, 83, 84]. The increased loading of suspended sediment, along with nutrients and pesticides, influences lake water quality which can impact, and in turn be impacted, by aquatic microbes. This is typified by excess N and P loadings leading to algal blooms and hypoxic zones [38, 85]. A goal of this project was to assess the potential for bacterial communities on bald cypress bark to provide relevant ecosystem services. We observed greater proportions of known denitrifying bacterial genera in wet bark samples. Organic C is the primary limiting factor for denitrification in aquatic systems [86–88], and wood mulch has been shown to significantly increase denitrification rates in agricultural runoff [89, 90]. Bald cypress are likely facilitating aquatic denitrification by serving as substrate and source of organic C for aquatic denitrifiers. Conversely, nitrogen fixing bacteria present on bald cypress bark may be increasing N, particularly in summer when demand is high and inorganic N is low [91, 92]. Cyanobacteria was the phyla with the second highest relative abundance in submerged bark samples, and some species of 1174-901-12, which was prevalent in our data, have the capacity to fix N [93].

Tree stems, particularly from wetland species, have recently been shown to be an important source of methane emissions [94, 95]. Bark-dwelling MOB can significantly reduce methane emissions from trees [18] and even artificial submerged surfaces can be hot spots of methanotrophy [96]. We observed a significantly higher relative abundance of MOB in submerged bark relative to dry, including a 14-fold increase in the genus Methylomonas which was identified as the dominant MOB in Melaleuca quinquenervia tree bark [18].

We also observed greater proportions of bacterial genera known to be involved in pollutant degradation in our submerged and splash zone bark samples. While more work is needed, submerged bark microbiota are more likely than dry bark microbial communities to be metabolically active and therefore may have greater remediation potential. Biofilm growth on submerged back also suggests another avenue for the valorization of tree bark residue, a largely unutilized waste product of the forestry industry [97, 98]. Biologically active carbon is used extensively in wastewater treatment with bacterial biofilms providing enhanced biodegradation [99, 100]. Pine (Pinus Silvestris) bark has been proposed as a cheap alternative to activated carbon in water treatment and used successfully to treat 2,4,6-trinitrotoluene (TNT) contaminated waters [101]. Bark from other tree species has been used as a “passive sampler” for atmospheric pollution [102–104] and the large surface area and high lipid content of bark can trap pollutants such as heavy metals, polycyclic aromatic hydrocarbons, pesticides, and chlorinated organics. Aquatic vegetation has been shown to mitigate pesticides runoff [105, 106] and submerged bark could serve a similar role.

The dry bark microbiome was consistent in both alpha diversity and taxonomy, with minimal variation between lakes or over the six-month sample collection period. Bark is the outermost protective layer of the tree and serves as a barrier against biotic and abiotic stressors [107]. Bark tends to be acidic with pH values of 2.2–6.8 [108–110]. Outer bark is compositionally distinct from wood, and is predominantly composed of cellulose, hemicellulose, and lignin [111]. Bark contains high levels (~ 15–30%) of extractive compounds, which include resin acids, tannins, sterol esters, and terpenoids [111, 112]. Many of these compounds have antimicrobial properties [113–115] and bark microbial communities may also be exposed to high levels of UV radiation, such that bark-dwelling Cyanobacteria have been shown to possess photoprotective compounds [116, 117]. Stem bark may be wetted during rain events, but may experience low water availability as most of the precipitation that a tree receives (> 70%) is intercepted by the canopy and falls to the ground as throughfall with only a small fraction draining down the tree stem as stemflow [118]. These environmental factors may place a severe restraint on the bark microbiome and could explain the lower observed alpha diversity and the dominance of specific bacterial taxa on dry bark.

Proteobacteria was the predominant phyla in all bark sections and lakes, with an average relative abundance of ~ 30%. This is consistent with what is reported in other tree bark studies, with Proteobacteria often comprising 40–50% of the tree bark bacterial community [20, 23–26, 28, 30, 31, 33, 119]. Acidobacteriota accounted for the second highest percentage of sequences (27–28%) in the dry bark sections, the portion that would most closely match the bark habitat from other systems. This is higher than that reported for other tree species, such as 13% for communities on Ginko biloba tree bark [30]. At the genus level we identified several core microbiome members that have been noted as major components of tree bark communities in other studies: Sphingomonas, Methylobacterium-Methylorubrum, Conexibacter, and 1174-901-12. However, we also found a high prevalence of genera such as Bryobacter, Bryocella (both in phylum Acidobacteriota), and Acidiphilium (Proteobacteria), which have not been reported as major components of the bark microbiome for other tree species, as well as relatively low abundances of Pantoea and Hymenobacter compared to other studies. Whether these differences represent a phylogenetic signal or are simply the product of the wetland habitat is unclear. Bald cypress growth is to not restricted to wetlands [120] and analyzing the bark microbiome of bald cypress growing on dry land would help clarify this.

Conclusions

Our study provides a window into the microbial dynamics of a rapidly vanishing ecosystem. The floodplain and bottomland hardwood forests that once dominated the Mississippi River Alluval Valley provide important ecological services that help mediate water quality, and their decline has resulted in increased eutrophication of aquatic ecosystems. Wetland trees such as bald cypress represent a huge amount of surface area for microbial colonization, growth, and interaction with aquatic contaminates. We found that hydrology was significantly correlated with alpha diversity and community composition of wet bark, which hosted a microbiome that was distinct from dry bark. A core bald cypress bark microbiome was identified that includes members present across six months of sample collection, and on trees in three different lakes. Wet bark hosted bacterial taxa that may be functionally involved in nitrogen cycling, methane oxidation, and pollutant degradation. Future studies should assess ability of wetland trees to facilitate remediation of aquatic contaminants and contribute to nutrient cycling in eutrophic systems.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

D.B. and L.H. designed the experiment and collected samples. L.H. prepared Fig. 1, and D.B. prepared Figs. 2, 3, 4, 5, 6, 7 and 8. D.B. processed the samples, analyzed the data, and wrote the main manuscript text. G.D. assisted with sample collection. L.H., G.D., C.J., and M.M. reviewed and edited the manuscript.

Funding

This research was supported by USDA-ARS Project Number 6060-13660-009-00D.

Data availability

Raw sequence reads are available from the NCBI Sequence Read Archive (SRA) under the BioProject ID PRJNA1285833.

Declarations

Competing interests

The authors declare no competing interests.