Niche Differentiation of Active Methane-Oxidizing Bacteria in Estuarine Mangrove Forest Soils in Taiwan

Mangrove forests are one of the important ecosystems in tropical coasts because of their high primary production, which they sustain by sequestering a substantial amount of CO2 into plant biomass. These forests often experience various levels of inundation and play an important role in CH4 emissions, but the taxonomy of methanotrophs in these systems remains poorly understood. In this study, DNA-based stable isotope probing showed significant niche differentiation in active aerobic methanotrophs in response to niche differentiation in upstream and downstream mangrove soils of the Tamsui estuary in northwestern Taiwan, in which salinity levels differ between winter and summer. Methylobacter and Methylomicrobium-like Type I methanotrophs dominated methane-oxidizing communities in the field conditions and were significantly 13C-labeled in both upstream and downstream sites, while Methylobacter were well adapted to high salinity and low temperature. The Type II methanotroph Methylocystis comprised only 10–15% of all the methane oxidizers in the upstream site but less than 5% at the downstream site under field conditions. 13C-DNA levels in Methylocystis were significantly lower than those in Type I methanotrophs, while phylogenetic analysis further revealed the presence of novel methane oxidizers that are phylogenetically distantly related to Type Ia in fresh and incubated soils at a downstream site. These results suggest that Type I methanotrophs display niche differentiation associated with environmental differences between upstream and downstream mangrove soils.


Introduction
Global warming is caused by anthropogenic greenhouse gas emissions, and the major greenhouse gas, methane (CH 4 ), traps 25 times more heat than carbon dioxide (CO 2 ) over a 100-year time horizon [1]. Globally, about 40% of CH 4 is released from natural ecosystems, mainly anaerobic environments such as rice paddies and wetlands [2,3]. However, the global CH 4 budget varies greatly, which is in part because biogeochemical models do not include microbially-mediated processes. For instance, it has been widely accepted that 30-90% of CH 4 produced in anaerobic ecosystems is consumed by aerobic methane-oxidizing bacteria (methanotrophs) when it diffuses through shallow soils [4]. Linking the taxonomic identity of aerobic methanotrophs to the ecologically important process of CH 4 Figure 1. Map of studied mangrove sites (black stars) in the Tamsui estuary, Taiwan. The grey star indicates the site from a previous study [19].
Soil samples were collected in January and August of 2018 to estimate the dynamics of methanotroph communities across seasons. During each field sampling, 25 soil core samples were collected from each site with polyvinyl chloride (PVC) tubes 1.5 cm in diameter and 30 cm long. Core samples were collected about every 5-10 m on a random walk, covering about 0.7 and 0.5 ha of the upstream and downstream mangrove forests, respectively. The soil samples were stored in ice and transported to the laboratory for analysis. After the collected soils were retrieved from the tubes, the top 2 cm of the bulk soil samples were sliced off using a knife pretreated with 70% ethanol. Any visible litter/roots were manually removed from the samples. Then, the 2-cm soil samples from the 25 cores from each site were mixed carefully to form a composite sample and stored in a plastic sealed bag at 4 °C for further analyses.
The soil texture was determined by a pipette method [27]. A 1:1 soil to water slurry was created and measured by a pH meter with a glass electrode (Jenco 6009, Jenco Instruments, San Diego, CA, USA). Soil total organic C (TOC) and total N (TN) were evaluated by the combustion method with an elemental analyzer (Fisons NA1500, ThermoQuest Italia, Milan, Italy). Soil labile C and N (soluble organic C: SbOC; soluble organic N: SbON; ammonium: NH4 + ; nitrate: NO3 − ; total dissolved N: TDN) were extracted using a 2 M KCl extraction method [28,29]. Briefly, extracts were filtered from 5 g of soil and soaked with 50 mL KCl for an hour at 150 rpm. Then, the extracts were analyzed with the cadmium reduction method for NO3 − [30], indophenol method for NH4 + [31], and persulfate method for TDN [32]. SbOC was analyzed with a TOC analyzer (1010, Analytical, TX, USA). Soil potential mineralizable N (PMN) was measured using a waterlogging incubation method [33].

Aerobic CH4 Oxidation Experiment and DNA-SIP Gradient Fractionation
Six replicates of soil samples from each site in January and August were incubated under oxic conditions. For each replicate, 10 g of soil was placed into a 125-mL serum bottle, then the bottle was sealed with a rubber stopper and covered tightly with an aluminum cap. Three replicates were injected with 3 mL of pure (99 atom% 13 C) 13 CH4 (Sigma-Aldrich, St. Louis, MO, USA) as the isotopelabeled set, and the other three replicates were injected with the same amount (99 atom% 12 C) of 12 CH4 to create a headspace of about 1.8% CH4. Then, the samples were placed in a dark incubator at 20 °C for 7 days to mimic the average soil temperature in the field (LM-570RD, Yihder Technology Co., Taiwan).
To measure the CH4 concentrations, a syringe was used to extract a 0.5-mL gas sample from the headspace of each bottle. The CH4 concentration of the sample was immediately measured by gas  [19].
Soil samples were collected in January and August of 2018 to estimate the dynamics of methanotroph communities across seasons. During each field sampling, 25 soil core samples were collected from each site with polyvinyl chloride (PVC) tubes 1.5 cm in diameter and 30 cm long. Core samples were collected about every 5-10 m on a random walk, covering about 0.7 and 0.5 ha of the upstream and downstream mangrove forests, respectively. The soil samples were stored in ice and transported to the laboratory for analysis. After the collected soils were retrieved from the tubes, the top 2 cm of the bulk soil samples were sliced off using a knife pretreated with 70% ethanol. Any visible litter/roots were manually removed from the samples. Then, the 2-cm soil samples from the 25 cores from each site were mixed carefully to form a composite sample and stored in a plastic sealed bag at 4 • C for further analyses.
The soil texture was determined by a pipette method [27]. A 1:1 soil to water slurry was created and measured by a pH meter with a glass electrode (Jenco 6009, Jenco Instruments, San Diego, CA, USA). Soil total organic C (TOC) and total N (TN) were evaluated by the combustion method with an elemental analyzer (Fisons NA1500, ThermoQuest Italia, Milan, Italy). Soil labile C and N (soluble organic C: S b OC; soluble organic N: S b ON; ammonium: NH 4 + ; nitrate: NO 3 − ; total dissolved N: TDN) were extracted using a 2 M KCl extraction method [28,29]. Briefly, extracts were filtered from 5 g of soil and soaked with 50 mL KCl for an hour at 150 rpm. Then, the extracts were analyzed with the cadmium reduction method for NO 3 − [30], indophenol method for NH 4 + [31], and persulfate method for TDN [32]. S b OC was analyzed with a TOC analyzer (1010, Analytical, TX, USA). Soil potential mineralizable N (PMN) was measured using a waterlogging incubation method [33].

Aerobic CH 4 Oxidation Experiment and DNA-SIP Gradient Fractionation
Six replicates of soil samples from each site in January and August were incubated under oxic conditions. For each replicate, 10 g of soil was placed into a 125-mL serum bottle, then the bottle was sealed with a rubber stopper and covered tightly with an aluminum cap. Three replicates were injected with 3 mL of pure (99 atom% 13 C) 13 CH 4 (Sigma-Aldrich, St. Louis, MO, USA) as the isotope-labeled set, and the other three replicates were injected with the same amount (99 atom% 12 C) of 12 CH 4 to create a headspace of about 1.8% CH 4 . Then, the samples were placed in a dark incubator at 20 • C for 7 days to mimic the average soil temperature in the field (LM-570RD, Yihder Technology Co., Taiwan).
To measure the CH 4 concentrations, a syringe was used to extract a 0.5-mL gas sample from the headspace of each bottle. The CH 4 concentration of the sample was immediately measured by gas chromatography with a flame ionization detector (GC-FID, GC9720, Fuli Instruments, Zhejiang, China). All the CH 4 had nearly depleted by day 7, so the CH 4 uptake rate was determined based on two data points between Day 0 (initial) and Day 7.
After the incubation experiment was completed, about 0.8 g of incubated soil from each replicate was extracted and its total genomic DNA was isolated with the PowerSoil ® DNA isolation kit (MO BIO Laboratories, Inc., Carlsbad, CA, USA) based on the manufacturer's instructions. DNA was also extracted from the fresh soils under field conditions. All the extracted DNA was stored at −20 • C until sequence analysis. Subsamples of the total genomic DNA were used to evaluate the absolute abundance of pmoA genes by real-time quantitative polymerase chain reaction (qPCR) analysis with a primer pair (A189F/mb661r) [34] and qPCR reagent kit (RR420A, SYBR ® Premix Ex Taq™, Takara Bio Inc., Japan). The samples were amplified using the following cycling steps: 5 min at 95 • C, followed by 40 cycles at 92 • C for 10 s, 55 • C for 30 s, 72 • C for 30 s, and 80 • C for 12 s with acquisitions. Analysis of the pmoA copies yielded 98% efficiency and R 2 = 0.9997.
The total DNA extracted from 13 CH 4 and 12 CH 4 incubated samples was separated into 15 density gradient fractions with an ultracentrifuge (Optima XPN-80, Beckman Coulter, Brea, CA, USA) with isopycnic density gradient centrifugation using the fractionation technique summarized in Lu and Jia [35]. Then, a digital refractometer (AR200, Reichert Tech., Ametek Inc., NY, USA) was used to measure the buoyant density for each DNA fraction. The same qPCR procedure was applied to evaluate the abundance of pmoA genes in fractions 2-14. The active methanotroph communities under high CH 4 concentrations were identified by comparing the abundance of pmoA genes in fractions 2-14 between the 13 CH 4 and 12 CH 4 samples.

Analyses of Active Methanotrophic Communities
The fraction of 13 CH 4 samples with the highest abundance of pmoA genes was analyzed by high-throughput sequencing to evaluate active methanotroph compositions. The methanotrophic species from fresh soils were also investigated using the total extracted DNA. The polymerase chain reaction (PCR) technique was used to amplify pmoA gene fragments with the primers A189F/mb661r and reagent kit RR902A, Premix Ex Taq™ (Takara Bio Inc., Kusatsu, Shiga, Japan) [16]. The PCR amplicons were sequenced on an Illumina MiSeq instrument with a MiSeq Reagent Kit v3 (2 × 300 bp paired-end reads).
Mothur was applied to process and classify the sequences based on the instructions and reference database in Dumont et al. (2014). The FunGene Pipeline [36] was used with USEARCH 6.0 Chimera Check [37] and Expand Mappings Programs to remove chimeric reads and obtain representative operational taxonomic units (OTUs), respectively. The microbial community diversity was calculated with the Shannon diversity index through the FunGene Pipeline.
To evaluate the total bacterial community in soil under field and incubated conditions, a universal primer pair (515F/907R) and reagent kit (KAPA HiFi HotStart ReadyMix, KAPA Biosystems, MA, USA) were used to amplify the V4 region of the 16S rRNA gene; the total bacterial composition was also analyzed by the Illumina Miseq platform. The sequences were analyzed by the Ribosomal Database Project (RDP) pipeline (http://pyro.cme.msu.edu/) and the microbial communities were classified by the naïve Bayesian rRNA classifier [38] to determine the soil microbial compositions. The microbial communities associated with methanotrophs were selected for analysis.

Statistical Analyses
The differences in soil physiochemical properties, CH 4 oxidation potential, and pmoA gene abundance between the two studied mangrove sites were tested with a one-way analysis of variance (one-way ANOVA) and Tukey's honestly significant difference (HSD) with JMP 11.0 (SAS Inc., Cary, NC, USA). p-value < 0.05 was considered significant. Based on the suggestions from Shiau et al. [20], the qPCR copies of pmoA were log transformed before performing any statistical analyses to normalize the data.
To evaluate the relationships between soil physiochemical properties and in situ methanotrophic communities, a canonical correspondence analysis (CCA) was applied to analyze the effects of environmental factors on methanotroph composition with XLSTAT 2019 (Addinsoft Inc., New York, NY, USA). Phylogenetic trees of methanotrophic communities in the studied mangrove forests were analyzed using MEGA X (Pennsylvania State University, State College, PA, USA). The representative sequences can be accessed on the National Taiwan University Scholars website (https://scholars.lib.ntu. edu.tw/handle/123456789/510011) and in the supplementary file.

Results
The total soil TOC and TN were not significantly different between the two seasons (Table 1). For the KCl extracts, the S b OC appeared to be higher in winter than summer (p < 0.05) and the upstream site than the downstream site. In addition, soil soluble N-i.e., NH 4 and PMN (p < 0.05)-were higher downstream in summer, while labile N was similar at both sites in winter (p > 0.05). The CH 4 oxidation potentials were similar between mangrove soils, except in samples from the downstream site in winter ( Figure 2). This difference in CH 4 oxidation potentials at the downstream site in winter was also reflected in their methanotrophic bacterial communities. qPCR results showed that the pmoA copy numbers in the fresh mangrove soils were highest at the downstream site in winter. In addition, pmoA copy number in the incubated soils increased by approximately one order of magnitude in both mangrove soils for both seasons ( Figure 3).
Gradient fractionation of the 13 CH 4 -incubated soil DNA showed that the 13 C-labeled pmoA genes were mostly concentrated in the 7-8 fractions (density: 1.733-1.737 g mL −1 ), except in the soil collected upstream in summer, in which the heavy fraction of the pmoA genes was located in the 10th fraction (density: 1.724 g mL −1 ) ( Figure 4). In addition, regional peak fractions (i.e., 10th fractions in Figure 4c,d) of the pmoA gene were also found in the Bali mangrove soils.
New York, NY, USA). Phylogenetic trees of methanotrophic communities in the studied mangrove forests were analyzed using MEGA X (Pennsylvania State University, State College, PA, USA). The representative sequences can be accessed on the National Taiwan University Scholars website (https://scholars.lib.ntu.edu.tw/handle/123456789/510011) and in the supplementary file.

Results
The total soil TOC and TN were not significantly different between the two seasons (Table 1). For the KCl extracts, the SbOC appeared to be higher in winter than summer (p < 0.05) and the upstream site than the downstream site. In addition, soil soluble N-i.e., NH4 + (p < 0.001), TDN (p < 0.02), and PMN (p < 0.05)-were higher downstream in summer, while labile N was similar at both sites in winter (p > 0.05). The CH4 oxidation potentials were similar between mangrove soils, except in samples from the downstream site in winter ( Figure 2). This difference in CH4 oxidation potentials at the downstream site in winter was also reflected in their methanotrophic bacterial communities. qPCR results showed that the pmoA copy numbers in the fresh mangrove soils were highest at the downstream site in winter. In addition, pmoA copy number in the incubated soils increased by approximately one order of magnitude in both mangrove soils for both seasons (Figure 3).     Gradient fractionation of the 13 CH4-incubated soil DNA showed that the 13 C-labeled pmoA genes were mostly concentrated in the 7-8 fractions (density: 1.733-1.737 g mL −1 ), except in the soil collected upstream in summer, in which the heavy fraction of the pmoA genes was located in the 10th fraction (density: 1.724 g mL −1 ) (Figure 4). In addition, regional peak fractions (i.e., 10th fractions in Figure  4c,d) of the pmoA gene were also found in the Bali mangrove soils.   Gradient fractionation of the 13 CH4-incubated soil DNA showed that the 13 C-labeled pmoA genes were mostly concentrated in the 7-8 fractions (density: 1.733-1.737 g mL −1 ), except in the soil collected upstream in summer, in which the heavy fraction of the pmoA genes was located in the 10th fraction (density: 1.724 g mL −1 ) (Figure 4). In addition, regional peak fractions (i.e., 10th fractions in Figure  4c,d) of the pmoA gene were also found in the Bali mangrove soils.   Results of high-throughput sequencing of the pmoA genes revealed that Type Ia Methylobacter was the most dominant methanotrophic population, making up 55-58% and 54-56% of bacterial methanotroph abundance at the upstream and downstream sites under field conditions, respectively ( Figure 5). Through the amended 13 CH 4 concentrations, Methylobacter increased in relative abundance by 14-29% in winter and 0-12% in summer. In addition, the pmoA sequencing of two peak fractions in the Bali mangrove soils yielded similar methanotrophic compositions.
Results of high-throughput sequencing of the pmoA genes revealed that Type Ia Methylobacter was the most dominant methanotrophic population, making up 55-58% and 54-56% of bacterial methanotroph abundance at the upstream and downstream sites under field conditions, respectively ( Figure 5). Through the amended 13 CH4 concentrations, Methylobacter increased in relative abundance by 14-29% in winter and 0-12% in summer. In addition, the pmoA sequencing of two peak fractions in the Bali mangrove soils yielded similar methanotrophic compositions.  Figure 5. The relative abundance of methanotrophic communities identified with the pmoA genes in the average of triplicate fresh soils under field conditions and in 13 C-DNA from 13 CH 4 -enriched microcosms of (a) upstream (Guandu) and (b) downstream (Bali) mangrove forest soils, Taipei, Taiwan. Two regional peak fractions of pmoA genes were found in the 13 CH 4 -amended Bali mangrove soils in one season (Figure 4), so the pmoA genes in both fractions were sequenced to identify the potential active methanotrophs.
In addition, the Type II methanotroph Methylocystis accounted for 10-15% of the relative abundance in the fresh soils from the upstream site but less than 5% in the fresh soils from the downstream site in both winter and summer under field conditions. Moreover, the relative abundance of Methylocystis in 13 C-DNA decreased after the incubation experiment with 13 CH 4 in both mangrove soils in winter. However, the abundance remained almost constant in both mangrove soils in summer.
The results from high-throughput sequencing of the 16S rRNA genes revealed very different community structures. Overall, 2-20% of the methanotrophic populations were identified in the 13 CH 4 -amended heavy fractions based on the 16S rRNA sequencing. Although Type Ia methanotrophs were still mostly dominant in both mangrove soils, 16S rRNA gene sequencing revealed that Methylobacter and Methylomonas were the major Type Ia methanotrophs in the collected fresh soils (p < 0.05), while pmoA gene sequencing suggested that it was just Methylobacter (Figure 6). Moreover, the two methanotrophs, Methylobacter and Methylomonas, were both active after being incubated in 13 CH 4 -amended environments (p < 0.05). In addition, about 10-40% of the active methanotrophs belonged to the unclassified Methylococcaceae and were dominant in the fresh and incubated mangrove soils in both seasons (p < 0.05). Moreover, the 16S sequencing of the two peak fractions in the Bali mangrove soils showed slightly more Methylomonas and Methyloprofundus in fraction 7 than fraction 10, while both Type Ib and Type II methanotrophs had negligible relative abundances in both fractions.
Microorganisms 2020, 8, x FOR PEER REVIEW 9 of 16 Figure 6. The relative abundance of methanotrophic communities identified with the 16S rRNA genes in fresh and 13 CH4-enriched (a) upstream (Guandu) and (b) downstream (Bali) mangrove forest soils, Taipei, Taiwan. Two regional peak fractions of pmoA genes were found in the 13 CH4-amended Bali mangrove soils in one season (Figure 4), so the 16S rRNA in both fractions were sequenced to identify the potential active methanotrophs.
The results from CCA showed the associations among environmental factors and different methanotrophic communities (Figure 7). Type I methanotrophs such as Methylobacter and Methylomonas were not influenced by the increasing salinity, while the Type II Methylocystis decreased with the increasing salinity.  6. The relative abundance of methanotrophic communities identified with the 16S rRNA genes in fresh and 13 CH 4 -enriched (a) upstream (Guandu) and (b) downstream (Bali) mangrove forest soils, Taipei, Taiwan. Two regional peak fractions of pmoA genes were found in the 13 CH 4 -amended Bali mangrove soils in one season (Figure 4), so the 16S rRNA in both fractions were sequenced to identify the potential active methanotrophs.
The results from CCA showed the associations among environmental factors and different methanotrophic communities (Figure 7). Type I methanotrophs such as Methylobacter and Methylomonas were not influenced by the increasing salinity, while the Type II Methylocystis decreased with the increasing salinity.

Composition and Adaptation of Methanotrophs in Field Mangrove Soils
This study demonstrates that different sections of an estuary can provide distinct ecological niches for methanotrophs in mangrove soils and influence methanotroph compositions. This may be because there is a wider variation in soil inundation statuses at the downstream than the upstream site, making the methanotrophic bacteria more active and resulting in the downstream site having the most pmoA copies in winter. Moreover, because methanotrophs can use CH4 as their sole C source [39], the high TOC and SbOC in the upstream site soils appears to be independent of the CH4 oxidation potentials in the studied mangrove ecosystem.
In addition, the qPCR experiments showed that pmoA gene copy numbers were highest at the downstream site in winter, which is more evidence that there is ecological niche differentiation between the downstream and upstream sites (e.g., the downstream site has a more dynamic soil aeration status). Furthermore, the inconsistencies in methanotrophic activities and pmoA gene increments may indicate that methanotrophic community composition is more important than abundance when determining the CH4 oxidation rates in mangrove soils. In addition, the heavy fraction enriched with 13 C-pmoA genes in the upstream in summer was found at the 10th layer, which may indicate that little 13 C was assimilated into the microbial genes and instead may have just respired as 13 CO2, and thus many of the pmoA genes in fraction 10 may not been 13 C labeled. This assumption may be confirmed by the two sequenced fractions (i.e., 7th and 10th fractions) in the Bali mangrove soils, because both fractions showed similar relative methanotrophic abundances. Moreover, the low percentages (i.e., 2%) of methanotrophic populations in the overall microbial community based on the 16S rRNA sequencing in the Guandu soils in summer may also confirm that few methanotrophs were 13 C labeled.
The methanotrophic community compositions, measured using high-throughput sequencing,

Composition and Adaptation of Methanotrophs in Field Mangrove Soils
This study demonstrates that different sections of an estuary can provide distinct ecological niches for methanotrophs in mangrove soils and influence methanotroph compositions. This may be because there is a wider variation in soil inundation statuses at the downstream than the upstream site, making the methanotrophic bacteria more active and resulting in the downstream site having the most pmoA copies in winter. Moreover, because methanotrophs can use CH 4 as their sole C source [39], the high TOC and S b OC in the upstream site soils appears to be independent of the CH 4 oxidation potentials in the studied mangrove ecosystem.
In addition, the qPCR experiments showed that pmoA gene copy numbers were highest at the downstream site in winter, which is more evidence that there is ecological niche differentiation between the downstream and upstream sites (e.g., the downstream site has a more dynamic soil aeration status). Furthermore, the inconsistencies in methanotrophic activities and pmoA gene increments may indicate that methanotrophic community composition is more important than abundance when determining the CH 4 oxidation rates in mangrove soils. In addition, the heavy fraction enriched with 13 C-pmoA genes in the upstream in summer was found at the 10th layer, which may indicate that little 13 C was assimilated into the microbial genes and instead may have just respired as 13 CO 2 , and thus many of the pmoA genes in fraction 10 may not been 13 C labeled. This assumption may be confirmed by the two sequenced fractions (i.e., 7th and 10th fractions) in the Bali mangrove soils, because both fractions showed similar relative methanotrophic abundances. Moreover, the low percentages (i.e., 2%) of methanotrophic populations in the overall microbial community based on the 16S rRNA sequencing in the Guandu soils in summer may also confirm that few methanotrophs were 13 C labeled.
The methanotrophic community compositions, measured using high-throughput sequencing, further revealed a potential reason for the high CH 4 oxidation potentials in the downstream mangrove soils. Although the results showed that Type I methanotrophs dominate both mangrove soils, the Type II methanotrophs decreased in relative abundance from 10-20% upstream to 3-6% downstream (closer to the estuary of the Tamsui River). Type II methanotrophs are reported to have lower CH 4 oxidation potentials than Type I methanotrophs in environments where the CH 4 concentration frequently fluctuates [20,21], so the higher Type I methanotrophs in the downstream soils may be a result of heterogeneous CH 4 oxidation at microsites under inundation and salinity stresses due to daily tides.
There were low counts of Type Ib methanotrophs belonging to the OSC-rel group and the Type II methanotroph Methylocystis, and high counts of Type Ia Methylobacter in the studied mangrove forests, suggesting that soil salinity regulates the distribution and growth of methanotrophic bacteria. Previous studies have indicated that Type Ib and Type II methanotrophs have low salinity resistance [23,40], while Type Ia methanotrophs are commonly found in saline and hypersaline environments [18,19,22,23]. Thus, our current study demonstrates that the changes in ecological niches can influence the compositions of methanotrophic communities at a spatial scale from upstream to downstream. Furthermore, the Type I methanotrophs levels were positively correlated with soil labile nutrient concentrations (i.e., TDN, NH 4 , and S b ON) from CCA, supporting the theory that they are r-selected organisms [9]. However, the observed methanotrophic compositions from our present study showed that the relative abundance of Methylobacter did not change much across the river sites. Instead, other methane-oxidizing bacteria that are yet to be classified (i.e., methane-oxidizing bacteria like (MOB-like)) appeared to be more abundant at the downstream site. This implies that more undiscovered salt-tolerant methanotrophs exist in coastal and marine ecosystems.
In addition, Type Ia Methylobacter was the dominant methanotrophic bacterium in both upstream and downstream sites. However, results from this study and our previous study located between the two mangrove sites in the present study [19] show that there is no transitional change in compositions. In our previous study sites, Type Ia Methylosarcina and uncultured Type Ib methanotrophs in the deep-sea-5 cluster were the dominant genera and shared >50% of the relative abundance. Moreover, one of the sites contained nearly 20% Type II methanotrophs, including Mehtylocystis, Methylosinus, and Methylosinus/cystis. Although the exact mechanism for this inconsistency in methanotrophic compositions is still unclear, we noticed that mangroves in the previous study grew on something resembling a deposition bar, which was only inundated during monthly high tides, and thus had low salinity. This difference in soil salinity along the Tamsui estuary may cause niche differentiation in the mangrove soils and alter methanotrophic community compositions. This assumption was supported by the differences in soil redox potentials: 180-370 mV in our previous study sites and about −200 mV in the present study sites.
Climate change and warming global temperatures appear to increase litter decomposition in mangrove forests [41], which likely accelerates the C cycle in and increase greenhouse gas emissions from mangrove soils [42]. On the other hand, the global warming-induced salt water intrusions in coastal ecosystems may also increase the salinity by 1-3 psu [43], which likely increases the Type Ia methanotrophic population and methane uptake capacity. Moreover, the competitions between sulfate reducing bacteria and methanogens for C sources may also shift under increased SO 4 2due to global warming [42,44]. All of these factors likely reduce the overall CH 4 flux and increase CO 2 flux from mangrove ecosystems. However, more studies are needed to confirm these above-mentioned hypotheses.

Active Methanotrophs Identified with 13 CH 4 DNA-SIP
High-throughput sequencing of 13 CH 4 -labeled pmoA heavy fraction DNA revealed that Type Ia Methylobacter is the most active methanotroph in the mangrove forest soils in both winter and summer. In addition, the CH 4 oxidation potential at the downstream site in winter was found to be significantly higher than that in summer. However, the abundance of methanotrophic bacteria ( Figure 3) and soil chemical properties (Table 1) remained largely unchanged between the two sampling seasons. Moreover, both methanotrophic communities were dominated by Methylobacter (>50%). Although we did not measure the soil temperature during the field soil sampling, the air and water temperature typically vary 10-15 • C in the Tamsui estuary between winter (i.e., 18.3-18.6 • C) and summer (i.e., 30.5-31.5 • C). Thus, it is plausible that Methylobacter is more active at low temperatures. A previous study showed that Methylobacter psychrophilus grows optimally at 6-10 • C in a pure culture laboratory [45]. Furthermore, previous studies observed an uncultured Methylobacter sp. from the cold Zoigê Tibetan Plateau wetland in China that was adapted to environmental temperatures of 4-21 • C [18,46]. The Methylobacter identified from this research was distantly related to the Methylobacter sp. found in the Zoigê wetland ( Figure S1), and this may strengthen our assumption that the observed high CH 4 oxidation potential is associated with high Methylobacter abundance in the Bali mangrove forests in winter.
However, the Methylobacter genes identified in our previously studied mangrove forest between the two sites in the present study was more closely related to Methylomicrobium-jap-rel genes than the Methylobacter ones in this study. This result implies that species in Methylobacter genera may be adapted to different soil salinities and temperature ranges. Moreover, Methylomicrobium was categorized as a sub-branch of Methylobacter and was divided into an adjacent subbranch until 1995 [47], so the Methylobacter and Methylomicrobium-like methanotrophs found in our previous study may be less active in polyhaline environments.
In addition, the~10% of unclassified Type Ia methanotrophs at the downstream site increased to a relative abundance of 20% after amended CH 4 incubation. Because a similar observation was reported in the tidal-influenced mangrove soil in our previous study [19], the unclassified methanotrophs in this study may also be new methanotrophic bacteria adapted to highly saline environments.
One notable point is that, because we only sequenced DNA in the fractions with the most abundant pmoA gene copies, we might have overlooked some Type Ia methanotrophs with more T and A nucleotides and some Type Ib and Type II methanotrophs with more C and G nucleotides.

pmoA and 16S rRNA Sequences
The high-throughput sequencing of 16S rRNA genes resulted in a considerable number of unclassified Methylococcaceae after amended 13 CH 4 incubation. This observation was similar to our previous finding in another mangrove forest [19] in which unclassified Methylococcaceae made up more than 50% of the sequences at all sites. The mangrove forests in this present study were spatially connected to the previously-analyzed mangrove forests, so the unclassified methanotrophs in the upstream and downstream sites may also be novel methanotrophs that are important to CH 4 oxidation in the mangrove forest ecosystems [19].
In addition, the Type Ib Methyloparacoccus was identified only from 16S rRNA gene sequencing but not from the pmoA gene because the database we used for classification did not include the pmoA gene fraction of Methyloparacoccus [48]. However, the phylogenetic tree of methanotrophic communities using pmoA sequencing showed some unclassified Type Ib methanotrophs in the study sites that were distantly related to Methyloparacoccus ( Figure S1), implying that the unclassified Type Ib methanotrophs are likely Methyloparacoccus in the present studied sites.
The 13 C labeled 16S rRNA gene analysis revealed a considerable number of Type Ia Methylomonas that were active at the downstream site. The reason for the high abundance of this phylotype remains unclear, but it highlights the importance of cultivating novel methanotrophs using state-of-the-art techniques such as the single-cell method. Furthermore, phylogenetic congruency between pmoA and 16S rRNA genes warrant further study to better represent and track the population dynamics of methanotrophs in natural settings [49]. Indeed, the phylogenetic tree of methanotrophic communities using 16S rRNA sequencing found that Methylobacter, Methylomonas, and Methylosarcina identified in this study are distantly related ( Figure S2). This may explain the inconsistencies between the methanotrophic communities identified with pmoA and 16s rRNA genes. Further studies may consider sequencing the entire 16S rRNA (V1-V9) in the mangrove soils to identify members of the methanotrophic community at more depth. Furthermore, the phylogenetic congruency between pmoA and 16S rRNA genes warrants further study to better represent and track the population dynamics of methanotrophs in natural settings [49].

Conclusions
Our study indicates that methanotrophic species and abundance can vary across space in mangrove forests along a river due to different niches. The high salinity in the downstream estuary may facilitate the growth and activity of Type Ia methanotrophs such as Methylobacter and Methylomicrobium and certain Type Ib methanotrophs yet to be classified. On the other hand, Type II methanotrophs did not grow well in polyhaline environments. Moreover, the high salinity-tolerant Methylobacter identified in the mangrove forests may also be more active in low temperature soil environments. Overall, this study demonstrates that predominant methanotrophs vary widely depending on not only the type of freshwater ecosystem they live in but also fine-scale soil physiochemical properties.