In-depth characterization of denitrifier communities across different soil ecosystems in the tundra
在苔原不同土壤生态系统中对反硝化细菌群落进行深入表征

Keywords: Arctic, Denitrification, Genome-resolved metagenomics, Nitrous oxide
关键词：北极，脱氮，基因组解析的宏基因组学，一氧化二氮

Nitrous oxide is a greenhouse gas (GHG) that has approximately 300 times the global warming potential of carbon dioxide on a 100-year scale [1]. Atmospheric concentrations have increased by nearly since preindustrial times, with soils-both natural and anthropogenic-accounting for up to of the global emissions
一氧化二氮 是一种温室气体（GHG），在 100 年的时间尺度上，其全球变暖潜力约为二氧化碳的 300 倍[1]。大气 浓度自工业化前几乎增加了 ，其中包括天然和人为的土壤占全球排放量的高达 。

[2]. Despite being nitrogen limited and enduring low temperatures throughout most of the year, tundra soils are increasingly recognized as important sources of [3-7]. The relative contribution of tundra soils to global GHG emissions is predicted to increase in the future , 9], as the warming rate at high latitude environments is more than twice as high than in other regions [10].
尽管苔原土壤受氮素限制且大部分时间内温度较低，但越来越多地认识到苔原土壤是重要的温室气体排放源[3-7]。预测未来苔原土壤对全球温室气体排放的相对贡献将增加[9]，因为高纬度环境的升温速度是其他地区的两倍以上[10]。

Microbial denitrification is an important source of [11]. Denitrification is a series of enzymatic steps in which nitrate is sequentially reduced to nitrite , nitric oxide , and dinitrogen via the activity of the Nar, Nir, Nor, and Nos enzymes, respectively. The denitrification trait is common across a wide range of archaea, bacteria, and some fungi, most of which are facultative anaerobes that switch to oxides as electron acceptor when oxygen becomes limiting [12]. Denitrification is a modular community process performed in synergy by different microbial taxa that execute only a subset of the complete denitrification pathway . With the growing number of microbial genomes sequenced in recent years, it has become evident that only a fraction of the microorganisms involved in the denitrification pathway encode the enzymatic machinery needed for complete denitrification .
微生物反硝化是一种重要的 来源[11]。反硝化是一系列酶催化的步骤，其中硝酸盐 依次还原为亚硝酸盐 、一氧化氮 和二氮 ，通过 Nar、Nir、Nor 和 Nos 酶的活性分别完成。反硝化特征在广泛的古菌、细菌和一些真菌中普遍存在，其中大多数是需氧性厌氧菌，在氧气限制时切换为 氧化物作为电子受体[12]。反硝化是一个模块化的群落过程，由不同微生物类群协同执行，只执行完整反硝化途径 的子集。随着近年来微生物基因组测序数量的增加，已经明显地看到，参与反硝化途径的微生物只有一小部分编码了完成反硝化所需的酶机制 。

Compared to high -emitting systems such as agricultural and tropical soils, our knowledge of denitrifier communities in tundra soils is limited. As denitrification leads to the loss of to the atmosphere, it enhances the -limited status of tundra systems thus impacting both microbial and plant communities [16, 17]. Investigations of denitrifier diversity in the tundra have been largely limited to gene-centric surveys using microarrays, amplicon sequencing, qPCR, and read-based metagenomics, which provide limited information on the taxonomic identity and genomic composition of community members. These studies have shown that denitrifier communities in the tundra are dominated by members of the phyla Proteobacteria, Actinobacteria, and Bacteroidetes, and that the potential for complete denitrification is usually present at the community level [18-22]. However, it is not known whether the complete denitrification potential occurs within discrete microbial populations or is widespread throughout populations of truncated denitrifiers lacking one or more denitrification genes. In addition, tundra soils encompass many different ecosystems, some of which are notorious sources (e.g. bare peat surfaces [3]). consumption is usually favoured in wetlands, where low availability due to anoxia promotes the reduction of to [23]. In upland soils, fluxes vary in both time and space. Strong sinks have been observed specially in sparsely vegetated upland soils [7], but the microbial processes underlying consumption in these systems are largely unknown [24]. Altogether, these large differences in fluxes across tundra ecosystems indicate differences in the structure of microbial communities, but a comprehensive understanding of the microorganisms driving fluxes in tundra soils is lacking.
与高排放系统（如农业和热带土壤）相比，我们对苔原土壤中反硝化细菌群落的了解有限。由于反硝化导致 流失到大气中，它增强了苔原系统的 限制状态，从而影响了微生物和植物群落[16, 17]。对苔原中反硝化细菌多样性的研究主要限于基因为中心的调查，包括微阵列、引物测序、qPCR 和基于读取的宏基因组学，这些方法提供了有限的关于群落成员的分类身份和基因组组成的信息。这些研究表明，苔原中的反硝化细菌群落主要由变形菌门、放线菌门和拟杆菌门的成员主导，而完全反硝化的潜力通常存在于群落水平[18-22]。然而，目前尚不清楚完全反硝化潜力是在离散微生物种群内发生，还是在缺少一个或多个反硝化基因的截短反硝化细菌种群中广泛存在。 此外，苔原土壤包括许多不同的生态系统，其中一些是臭名昭著的来源（例如裸露的泥炭表面）。在湿地中通常偏好消耗，因为缺氧导致低氧化还原电位促进氧化还原反应。在高地土壤中，气体通量在时间和空间上都有所变化。尤其在稀疏植被的高地土壤中观察到强烈的甲烷汇，但这些系统中甲烷消耗的微生物过程大部分是未知的。总的来说，苔原生态系统中甲烷通量的巨大差异表明微生物群落结构存在差异，但对驱动苔原土壤甲烷通量的微生物的全面理解仍然缺乏。

Modelling emissions based on microbial community structure is challenging. fluxes are characterized by a high temporal and spatial heterogeneity driven by several environmental constraints related to soil , , moisture, and oxygen content [11]. In addition, our knowledge of the regulation of the denitrification process is largely based on the activity of model organisms such as the complete denitrifier Paracoccus denitrificans [25].
基于微生物群落结构建模 排放是具有挑战性的。 通量的特点是高时空异质性，受到与土壤 、 、湿度和氧含量相关的几个环境约束的驱动[11]。此外，我们对反硝化过程调控的认识主要基于模式生物的活性，如完全反硝化菌 Paracoccus denitrificans[25]。
It has been suggested that incomplete denitrifiers that contain Nir and Nor but lack Nos contribute substantially to soil emissions [26], while non-denitrifying reducers, i.e., microorganisms that contain Nos but lack Nir, can represent an important sink [27-29]. Furthermore, the partitioning of metabolic pathways across different populations with truncated pathwaysalso known as metabolic handoffs [30]-has been linked to higher efficiencies in substrate consumption compared to complete pathways . However, it remains largely unclear how populations of truncated denitrifiers with different sets of denitrification genes interact with each other and the environment impacting in situ emissions.
已有研究表明，含有 Nir 和 Nor 但缺乏 Nos 的不完全反硝化细菌在土壤 排放中起着重要作用[26]，而非反硝化 还原细菌，即含有 Nos 但缺乏 Nir 的微生物，可能代表一个重要的 汇[27-29]。此外，代谢途径在不同群体之间的分配，即被称为代谢交接[30]，已被证明与废物利用效率比完整途径 更高有关。然而，截断反硝化细菌群体如何相互作用以及如何影响环境从而影响原位 排放仍然不太清楚。

The paucity of in-depth knowledge on denitrifying communities in the tundra impairs our ability to model current and future fluxes from this biome. A better understanding of the ecological, metabolic, and functional traits of denitrifiers is thus critical for improving current models and mitigating emissions [32]. This invariably relies on the characterization of the so-called uncultured majority, i.e., microorganisms that have not been cultured to date but which comprise a high proportion of the microbial diversity in complex ecosystems [33, 34]. Genome-resolved metagenomics is a powerful tool to access the genomes of uncultured microorganisms and has provided important insights into carbon cycling processes in tundra soils [35-37]. However, this approach has not yet been applied to investigate the mechanisms driving fluxes in the tundra. Here, we used genomeresolved metagenomics to investigate the diversity and metabolic capabilities of denitrifiers across different tundra soil ecosystems characterised by a high variability in net fluxes in an area of mountain tundra in Kilpisjärvi, northern Finland.
缺乏对苔原反硝化群落的深入了解，影响了我们模拟当前和未来从这一生物群落中排放的氮气通量的能力。因此，更好地理解反硝化细菌的生态、代谢和功能特征对于改进当前模型和减少氮气排放至关重要。这不可避免地依赖于所谓的未培养多数的表征，即迄今尚未培养的微生物，但它们在复杂生态系统中占据着很高的微生物多样性比例。基因组解析宏基因组学是一种强大的工具，可用于获取未培养微生物的基因组，并已为苔原土壤中的碳循环过程提供了重要见解。然而，这种方法尚未应用于研究驱动苔原氮气通量的机制。在这里，我们使用基因组解析宏基因组学研究了不同苔原土壤生态系统中反硝化细菌的多样性和代谢能力，这些生态系统的净氮气通量在芬兰北部 Kilpisjärvi 的山地苔原地区具有很高的变异性。

The Saana Nature Reserve ( is located in Kilpisjärvi, northern Finland (Fig. 1a). The area is part of the mountain tundra biome and is characterized by a mean annual temperature of and annual precipitation of [38]. Sampling was performed across 43 sites during the peak of the growing season in the northern hemisphere. Our study sites are distributed across Mount Saana and Mount Korkea-Jehkas and the valley in between (Fig. 1b), and include barren soils ), heathlands (dominated by evergreen and deciduous shrubs) ), meadows (dominated by graminoids and forbs) , and fens (Fig. 1c). Elevation across the sampling sites varies from 586.6 to 904.5 m.a.s.l. (Additional file 1: Table S1). Fen sites were sampled in July 2018 and all other sites in July 2017. Samples were obtained
萨阿纳自然保护区（ 位于芬兰北部基尔皮斯亚尔维（图 1a）。该地区属于山地苔原生物群落，其特点是平均年温度为 ，年降水量为 [38]。在北半球生长季节的高峰期间，我们在 43 个地点进行了采样。我们的研究地点分布在萨阿纳山和科尔凯亚-耶赫卡斯山之间的山谷（图 1b），包括贫瘠土壤 ），荒地（以常绿和落叶灌木为主） ），草地（以禾本科植物和草本植物为主） ，以及沼泽地 （图 1c）。采样地点的海拔从 586.6 米至 904.5 米不等（附加文件 1：表 S1）。沼泽地点于 2018 年 7 月进行采样，其他所有地点于 2017 年 7 月进行采样。样本已获得。
with a soil corer sterilized with ethanol and, when possible, cores were split into organic and mineral samples using a sterilized spatula. In total, 69 samples ( 41 organic and 28 mineral) were obtained from the 43 sites (Fig. 1c, Additional file 1: Table S1). Samples were transferred to a whirl-pack bag and immediately frozen in dry ice. Samples were transported frozen to the laboratory at the University of Helsinki and kept at until analyses.
使用用 乙醇消毒的土壤取样器，尽可能地，使用消毒过的铲子将样品分为有机和矿物样品。总共从 43 个站点（图 1c，附加文件 1：表 S1）获得了 69 个样品（41 个有机样品和 28 个矿物样品）。样品被转移到旋转包装袋中，并立即在干冰中冷冻。样品被冷冻运送到赫尔辛基大学实验室，并保存在 进行分析。

Soil , moisture, and soil organic matter (SOM) content were measured from the 69 samples according to Finnish (SFS) and international (ISO) standards (SFS 300, ISO 10390, and SFS 3008). Carbon (C) and N content were measured using a Vario Micro Cube machine (Elementar, Langenselbold, Germany). In situ ecosystem-level and methane fluxes were measured from the 43 sites using a static, non-steady state, non-flow-through system composed of a darkened acrylic chamber diameter, height) [4,39]. Measurements were conducted between 2nd July and 2nd August 2018, between 10 am and . Simultaneous measurement of GHG fluxes and sampling for metagenomic sequencing was not possible due to limited resources and logistic constraints. At each site, five gas samples were taken during a 50 -min chamber closure and transferred to evacuated Exetainer vials (Labco, Lampeter, UK). Gas samples were analysed using an Agilent 7890B gas chromatograph (Agilent Technologies, Santa Clara, CA, USA) equipped with an autosampler (Gilson, Middleton, WI, USA) and a flame ionization detector for and an electron capture detector for . Gas concentrations were calculated from the gas chromatograph peak areas based on standard curves with a concentration of and a concentration of .
土壤 ，湿度和土壤有机质（SOM）含量根据芬兰（SFS）和国际（ISO）标准（SFS 300，ISO 10390 和 SFS 3008）从 69 个样品中测量。碳（C）和氮含量使用 Vario Micro Cube 机器（Elementar，Langenselbold，德国）测量。在 43 个站点使用静态、非稳态、非流通系统（由一个暗色丙烯酸室组成，直径 ，高 ）测量原位生态系统水平 和甲烷 通量[4,39]。测量时间为 2018 年 7 月 2 日至 8 月 2 日，时间为上午 10 点至 。由于资源有限和后勤限制，无法同时测量温室气体通量并进行元基因组测序取样。在每个站点，通过 50 分钟的室内关闭取五个 气体样品，并转移到抽空的 Exetainer 瓶（Labco，Lampeter，英国）。气体样品使用装有自动进样器（Gilson，Middleton，WI，美国）和用于 的火焰离子化检测器以及用于 的电子俘获检测器的 Agilent 7890B 气相色谱仪（Agilent Technologies，Santa Clara，CA，美国）进行分析。 气体浓度是根据气相色谱峰面积和标准曲线计算得出的，其中 浓度为 ， 浓度为 。

Differences in physicochemical composition and rates of GHG fluxes across soil ecosystems and depths were assessed using one-way analysis of variance (ANOVA) followed by Tukey's HSD test with the and TukeyHSD functions in R v3.6.3 [40]. The relationship between soil ecosystem, depth, and physicochemical properties was also verified using a multivariate approach consisting of principal component analysis (PCA) and permutational ANOVA (PERMANOVA) with the package vegan v2.5-6 in v3.6.3 (functions and adonis, respectively) [40, 41]. C, N, and C:N ratio were not included in the multivariate dataset due to a high amount of missing data, and moisture and SOM were log-transformed prior to analysis. Due to the limited number of samples from barren sites, these were not included in the ANOVA and PERMANOVA procedures.
通过使用一元方差分析（ANOVA）评估了土壤生态系统和深度之间的物理化学组成差异和温室气体通量速率，随后使用 R v3.6.3 中的 和 TukeyHSD 函数进行图基的 HSD 检验[40]。还使用了多元方法验证了土壤生态系统、深度和物理化学性质之间的关系，包括主成分分析（PCA）和包含 vegan v2.5-6 的 permutational ANOVA（PERMANOVA）[40, 41]。由于缺失数据量较大，C、N 和 C:N 比率未包含在多元数据集中，而湿度和 SOM 在分析之前进行了对数转换。由于贫瘠地点样本数量有限，因此未将其纳入 ANOVA 和 PERMANOVA 程序中。

Total DNA and RNA were co-extracted as previously described [42]. Briefly, extraction was performed on of soil using a hexadecyltrimethyl ammonium bromide (CTAB), phenol-chloroform, and bead-beating protocol. DNA was purified using the AllPrep DNA Mini Kit (QIAGEN, Hilden, Germany) and quantified using the Qubit dsDNA BR Assay Kit (ThermoFisher Scientific, Waltham, MA, USA). Library preparation for Illumina metagenome sequencing was performed using the Nextera XT DNA Library Preparation Kit (Illumina, San Diego, CA, USA). Metagenomes were obtained for the 69 samples across two paired-end NextSeq (132-170 bp) and one NovaSeq ( runs. Two samples were additionally sequenced with Nanopore MinION. For this, libraries were prepared using the SQK-LSK109 Ligation Sequencing Kit with the long fragment buffer (Oxford Nanopore Technologies, Oxford, UK) and the NEBNext Companion Module for Oxford Nanopore Technologies Ligation Sequencing Kit (New England Biolabs). Each sample was sequenced for on one R9.4 flow cell.
总 DNA 和 RNA 如前所述一起提取[42]。简而言之，使用十六烷基三甲基溴化铵（CTAB）、苯酚-氯仿和珠破碎方案在 土壤上进行提取。DNA 使用 AllPrep DNA Mini Kit（QIAGEN，德国希尔登）纯化，并使用 Qubit dsDNA BR Assay Kit（ThermoFisher Scientific，美国马萨诸塞州沃尔瑟姆）进行定量。 Illumina 宏基因组测序的文库制备使用 Nextera XT DNA 文库制备套件（Illumina，美国加利福尼亚州圣地亚哥）。69 个样本的宏基因组分别通过两个配对末端 NextSeq（132-170 bp）和一个 NovaSeq（ 运行获得。另外两个样本还使用 Nanopore MinION 进行测序。为此，使用 SQK-LSK109 连接测序套件与长片段缓冲液（Oxford Nanopore Technologies，英国牛津）和 NEBNext Companion Module for Oxford Nanopore Technologies 连接测序套件（New England Biolabs）制备文库。每个样本在一个 R9.4 流式细胞上测序 。

We obtained more than 9 billion Illumina (1.4 Tb) and 7 million Nanopore ( ) reads from the 69 soil metagenomes (mean: , minimum: , maximum: ) (Additional file 1: Table S1). The quality of the raw Illumina data was verified with fastQC v0.11.9 [43] and multiQC v1.8 [44]. Cutadapt v1.16 [45] was then used to trim sequencing adapters and low-quality base calls and to filter out short reads ( ). Nanopore data were basecalled with GPU guppy v4.0.11 using the high-accuracy model and applying a minimum quality score of 7. The quality of the basecalled Nanopore data was assessed with pycoQC v2.5.0.21 [46] and adapters were trimmed with Porechop v0.2.4 [47].
我们从 69 个土壤宏基因组中获得了超过 90 亿个 Illumina（1.4 Tb）和 700 万个 Nanopore（ ）reads（平均值： ，最小值： ，最大值： ）（附加文件 1：表 S1）。原始 Illumina 数据的质量经过了 fastQC v0.11.9 [43]和 multiQC v1.8 [44]的验证。然后使用 Cutadapt v1.16 [45]修剪测序适配器和低质量碱基呼叫 ，并过滤掉短 reads（ ）。Nanopore 数据使用 GPU guppy v4.0.11 进行碱基呼叫，使用高准确度模型，并应用最小质量分数为 7。通过 pycoQC v2.5.0.21 [46]评估了碱基呼叫的 Nanopore 数据质量，并使用 Porechop v0.2.4 [47]修剪了适配器。

c

Fig. 1 (See legend on previous page.)
图 1（请参见前一页的图例。）

Taxonomic profiles of the microbial communities were obtained using a read-based approach, i.e., based on unassembled Illumina data. Due to differences in sequencing depth across the samples, the dataset was resampled to reads per sample with seqtk v1.3 [48]. Reads matching the SSU rRNA gene were identified with METAXA v2.2 [49] and classified against the SILVA database release 138.1 [50] in mothur v1.44.3 [51] using the Wang's Naïve Bayesian Classifier [52] and a confidence cut-off. Differences in community structure across soil ecosystems and depths were assessed using non-metric multidimensional scaling (NMDS) and PERMANOVA with the package vegan v2.5-6 in v3.6.3 (functions metaMDS and adonis, respectively) [40, 41]. The relationship between community structure, soil physicochemical properties, and elevation was also assessed using PERMANOVA and distance-based redundancy analysis (db-RDA) with forward selection with the package vegan v2.5-6 in v3.6.3 (functions adonis and capscale/ordistep, respectively) [40, 41]. The physicochemical dataset included only , moisture, and SOM due to a high amount of missing data for the other variables, and barren sites were not included in the PERMANOVA procedure due to the limited number of samples. Moisture and SOM were log-transformed prior to analysis. Relationships between the abundance of individual genera and flux rates were assessed using linear regression in v3.6.3 [40].
使用基于读取的方法获得了微生物群落的分类学概况，即基于未组装的 Illumina 数据。由于样本间测序深度的差异，数据集使用 seqtk v1.3 [48]重新采样至每个样本 个读取。使用 METAXA v2.2 [49]识别与 SSU rRNA 基因匹配的读取，并在 mothur v1.44.3 [51]中使用 Wang 的朴素贝叶斯分类器 [52]和 置信度截断值对其进行分类，参考 SILVA 数据库发布版本 138.1 [50]。使用 vegan v2.5-6 软件包在 v3.6.3 中（分别使用 metaMDS 和 adonis 函数）[40, 41]进行非度量多维尺度分析（NMDS）和 PERMANOVA 评估土壤生态系统和深度间的群落结构差异。还使用 vegan v2.5-6 软件包在 v3.6.3 中（分别使用 adonis 和 capscale/ordistep 函数）[40, 41]进行 PERMANOVA 和基于距离的冗余分析（db-RDA）以及前向选择评估群落结构、土壤理化性质和海拔之间的关系。 物理化学数据集仅包括 、湿度和 SOM，因为其他变量的数据缺失较多，贫瘠地点未包括在 PERMANOVA 程序中，因为样本数量有限。在分析之前，湿度和 SOM 进行了对数转换。使用线性回归在 v3.6.3 [40]中评估个体属的丰度与 通量率之间的关系。

Metagenome assembling of the Illumina data was performed as two co-assemblies. One co-assembly comprised the upland soils (barren, heathland, and meadow; and the other the fen samples . For each co-assembly, reads from the respective samples were pooled and assembled with MEGAHIT v1.1.1.2 [53]. Assembling of the Nanopore data was done for each sample individually with metaFlye v2.7.1 [54], and contigs were corrected based on Illumina data from the respective sample with bowtie v2.3.5 [55], SAMtools v1.9 [56], and pilon v1.23 [57]. Quality assessment of the (co-) assemblies was obtained with metaQUAST v5.0.2 [58].
Illumina 数据的宏基因组组装分为两个共同组装。一个共同组装包括高地土壤（贫瘠、荒地和草地； ，另一个是沼泽样本 。对于每个共同组装，来自相应样本的 reads 被汇集并使用 MEGAHIT v1.1.1.2 [53]进行组装。Nanopore 数据的组装是针对每个样本单独进行的，使用 metaFlye v2.7.1 [54]，并根据相应样本的 Illumina 数据使用 bowtie v2.3.5 [55]、SAMtools v1.9 [56]和 pilon v1.23 [57]进行校正 contigs。使用 metaQUAST v5.0.2 [58]对（共同）组装的质量进行评估。

Binning of metagenome-assembled genomes (MAGs) was done separately for each Illumina and Nanopore (co-)assembly with anvi'o v6.2 [59] after discarding contigs shorter than . The two Illumina co-assemblies and the two individual Nanopore assemblies yielded more than 4 million contigs longer than , with a total assembly size of . Gene calls were predicted with prodigal v2.6.3 [60]. Single-copy genes were identified with HMMER v.3.2.1 [61] and classified with DIAMOND v0.9.14 [62] against the Genome Taxonomy
对于每个 Illumina 和 Nanopore（共同）组装，使用 anvi'o v6.2 [59] 分别对宏基因组组装的基因组（MAGs）进行分箱，丢弃长度小于 的 contigs。两个 Illumina 共同组装和两个单独的 Nanopore 组装产生了超过 4 百万个长度大于 的 contigs，总组装大小为 。基因调用使用 prodigal v2.6.3 [60] 进行预测。单拷贝基因使用 HMMER v.3.2.1 [61] 进行识别，并使用 DIAMOND v0.9.14 [62] 对 Genome Taxonomy 进行分类。
Database (GTDB) release 04-RS89 [63, 64]. Illumina reads were mapped to the contigs with bowtie v2.3.5 [55] and SAM files were sorted and indexed using SAMtools v1.9 [56]. The co-assemblies covered a significant fraction of the original metagenomic data, with an average read recruitment rate of across samples (minimum: , maximum: ). Due to their large sizes, Illumina co-assemblies were split into 100 smaller clusters based on differential coverage and tetranucleotide frequency with CONCOCT v1.0.0 [65]. Contigs were then manually sorted into bins based on the same composition and coverage metrics using the anvi-interactive interface in anvi'o v6.2 [59]. Nanopore contigs were binned directly without pre-clustering. Bins that were complete according to the presence of single-copy genes were further refined using the anvi-refine interface in anvi'o v6.2 [59]. In addition to taxonomic signal (based on singlecopy genes classified against GTDB), either differential coverage or tetranucleotide frequency was used to identify and remove outlying contigs. The former was used for bins with a large variation in contig coverage across samples, and the latter for those with marked differences in GC content across contigs. Medium- and high-quality bins ( complete and redundant according to the MIMAG standard [66]) were renamed as MAGs and kept for downstream analyses.
数据库（GTDB）发布 04-RS89 [63, 64]。 Illumina 读取与 bowtie v2.3.5 [55]映射到 contigs，并使用 SAMtools v1.9 [56]对 SAM 文件进行排序和索引。共同组装覆盖了原始宏基因组数据的显著部分，平均读取招募率为 （最小值： ，最大值： ）。由于其较大的大小，Illumina 共同组装根据差异覆盖和四核苷酸频率被分成 100 个较小的簇，使用 CONCOCT v1.0.0 [65]。然后，根据相同的组成和覆盖度指标，使用 anvi'o v6.2 [59]中的 anvi-interactive 界面手动将 contigs 分类到不同的垃圾箱中。 Nanopore contigs 直接进行分类，无需预先聚类。根据单拷贝基因的存在来进一步优化完整的垃圾箱，使用 anvi'o v6.2 [59]中的 anvi-refine 界面。除了基于单拷贝基因对 GTDB 进行分类的分类信号外，还使用差异覆盖或四核苷酸频率来识别和删除异常的 contigs。 前者用于在样本之间具有较大的 contig 覆盖变化的垃圾箱，后者用于在 contig 之间具有明显 GC 含量差异的垃圾箱。根据 MIMAG 标准[66]，中等和高质量的垃圾箱（ 完整和 冗余）被重新命名为 MAGs，并保留用于下游分析。

Functional profiles of the microbial communities were obtained using a gene-centric approach based on assembled data. For each (co-)assembly, gene calls were translated to amino acid sequences and searched against the KOfam hidden Markov model (HMM) database with KofamScan v1.3.0 [67]. Only matches with scores above the pre-computed family-specific thresholds were kept. Genes putatively identified as denitrification genes (nirK, nirS, norB, and nosZ) were submitted to further analyses to identify false positives consisting of distant homologues that are not involved in denitrification. Amino acid sequences were aligned with MAFFT v7.429 [68] and alignments were visualized with Unipro UGENE v38.1 [69]. Sequences were then inspected for the presence of conserved residues at positions associated with the binding of co-factors and active sites: nirK, Cu-binding and active sites [70]; nirS, c-heme and -heme binding sites [71]; norB, binding of the catalytic centres cyt , , and [72]; nos : binding of the and centres [72]. Sequences which did not contain the correct amino acid at these positions were removed. Finally, resulting amino acid sequences were aligned with MAFFT v7.429 [68] along with reference sequences from the genome of cultured denitrifiers [14] and a maximum-likelihood tree was computed with FastTree v2.1.11 [73] using
微生物群落的功能概况是通过基于组装数据的基因中心方法获得的。对于每个（共同）组装，基因调用被翻译成氨基酸序列，并与 KOfam 隐藏马尔可夫模型（HMM）数据库进行搜索，使用 KofamScan v1.3.0 [67]。只保留得分高于预先计算的特定家族阈值的匹配项。被推断为反硝化基因（nirK、nirS、norB 和 nosZ）的基因被提交进行进一步分析，以识别不参与反硝化的远缘同源基因。氨基酸序列与 MAFFT v7.429 [68]进行对齐，并使用 Unipro UGENE v38.1 [69]可视化对齐。然后检查序列中与辅因子结合和活性位点相关位置上的保守残基的存在：nirK，Cu 结合和活性位点[70]；nirS，c-血红素和 -血红素结合位点[71]；norB，催化中心细胞 ， 和 的结合[72]；nos ： 和 中心的结合[72]。不包含这些位置上正确氨基酸的序列将被移除。 最终，得到的氨基酸序列与 MAFFT v7.429 [68]中的参考序列以及培养的反硝化细菌基因组中的参考序列进行了比对，并使用 FastTree v2.1 计算了最大似然树。11 [73] 使用
the LG+GAMMA model. Annotation of denitrification genes was also performed for previously published genomes retrieved from GenBank. These included a set of 1529 MAGs obtained from soils in Stordalen Mire, northern Sweden [37], and all genomes of Acidobacteriota strains and candidate taxa (accessed on 9 October 2020).
LG+GAMMA 模型。还对从 GenBank 检索的先前发表的基因组进行了反硝化基因的注释。这些基因组包括从瑞典北部 Stordalen Mire 土壤中获得的 1529 个 MAGs 集合[37]，以及所有酸杆菌门菌株和候选分类群的 基因组（于 2020 年 10 月 9 日访问）。

The abundance of functional genes was computed based on read coverage with CoverM v0.6.1 [74]. For this, Illumina reads were mapped to the contigs with minimap v2.17 [75] and coverage was normalized to reads per kilobase million (RPKM). Differences in functional community structure were assessed using NMDS, PERMANOVA, and db-RDA as described above for the taxonomic profiles. Differences in the abundance of individual genes across soil ecosystems were assessed using ANOVA followed by Tukey's HSD test with the and TukeyHSD functions in R v3.6.3 [40]. Due to the limited number of samples from barren sites, these were not included in the ANOVA and PERMANOVA procedures. Relationships between the abundance of denitrification genes and flux rates were assessed using linear regression in R v3.6.3 [40].
基于 CoverM v0.6.1 [74]的读取覆盖率计算了功能基因的丰度。为此，使用 minimap v2.17 [75]将 Illumina 读取映射到 contigs，并将覆盖率标准化为每千碱基百万读取（RPKM）。使用 NMDS、PERMANOVA 和 db-RDA 评估了功能群落结构的差异，方法与对分类学文件进行描述时相同。使用 ANOVA 后跟 R v3.6.3 [40]中的 Tukey's HSD 测试和 TukeyHSD 函数评估了土壤生态系统中个体基因丰度的差异。由于贫瘠地点样本数量有限，因此未包括在 ANOVA 和 PERMANOVA 程序中。使用 R v3.6.3 [40]中的线性回归评估了反硝化基因丰度与 通量率之间的关系。

Phylogenetic placement of MAGs was done based on 122 archaeal and 120 bacterial single-copy genes with GTDB-Tk v1.3.0 [76] and the GTDB release 05-RS95 [63, 64]. Acidobacteriota MAGs containing denitrification genes were submitted to further phylogenomic analyses alongside all genomes of Acidobacteriota strains and candidate taxa available on GenBank ( ; accessed on 9 October 2020). For this, the amino acid sequence of 23 ribosomal proteins was retrieved for each genome with anvi'o v6.2 [59] and aligned with MUSCLE v3.8.1551 [77]. A maximum likelihood tree was then computed based on the concatenated alignments with FastTree v2.1.11 using the LG+GAMMA model [73]. Escherichia coli ATCC 11,775 was used to root the tree.
MAGs 的系统发育定位是基于 122 个古菌和 120 个细菌的单拷贝基因，使用 GTDB-Tk v1.3.0 [76] 和 GTDB release 05-RS95 [63, 64]。含有反硝化基因的酸杆菌 MAGs 被提交进行进一步的系统发育分析，与 GenBank 上所有酸杆菌菌株和候选分类单元的基因组一起进行分析（ ；于 2020 年 10 月 9 日访问）。为此，使用 anvi'o v6.2 [59] 检索了每个基因组的 23 个核糖体蛋白质的氨基酸序列，并使用 MUSCLE v3.8.1551 [77] 进行了比对。然后，基于连接的比对使用 FastTree v2.1.11 和 LG+GAMMA 模型计算了最大似然树[73]。大肠杆菌 ATCC 11,775 用于根据树。

For metabolic reconstruction, MAGs were annotated against the KOfam HMM database [67] with HMMER v.3.2.1 [61] using the pre-computed score thresholds of each HMM profile. The anvi-estimate-metabolism program in anvi'o v6.2 [59] was then used to predict the metabolic capabilities of the MAGs. A metabolic pathway was considered present in MAGs containing at least of the genes involved in the pathway. Carbohydrateactive enzymes (CAZymes) were annotated with dbCAN v.2.0 based on the dbCAN v7 HMM database [78]. Only hits with an e-value and coverage were considered.
对于代谢重建，MAGs 使用 HMMER v.3.2.1 [61] 对 KOfam HMM 数据库 [67] 进行了注释，使用每个 HMM 概要文件的预先计算的得分阈值。然后使用 anvi'o v6.2 [59] 中的 anvi-estimate-metabolism 程序来预测 MAGs 的代谢能力。包含至少 条参与通路的基因的 MAGs 被认为存在代谢通路。碳水化合物活性酶（CAZymes）根据 dbCAN v7 HMM 数据库 [78] 使用 dbCAN v.2.0 进行注释。仅考虑具有 e-值 和覆盖率 的匹配。

Prior to read recruitment analyses, Illumina and Nanopore MAGs were dereplicated based on a average nucleotide identity (ANI) threshold with fastANI v1.3 [79] to remove redundancy (i.e., MAGs that were recovered multiple times across the different assemblies). Read recruitment analyses were then performed with CoverM v0.6.1 [74]. For this, Illumina reads were mapped to the set of non-redundant MAGs with minimap v2.17 [75] and relative abundances were calculated as a proportion of the reads mapping to each MAG.
在进行招聘分析之前，使用 fastANI v1.3 [79]根据 平均核酸同源性（ANI）阈值对 Illumina 和 Nanopore MAGs 进行去重，以消除冗余（即，在不同组装中多次恢复的 MAGs）。然后使用 CoverM v0.6.1 [74]进行读取招聘分析。为此，使用 minimap v2.17 [75]将 Illumina 读取映射到非冗余 MAG 组，并计算相对丰度，作为映射到每个 MAG 的读取的比例。

Our sampling design in Kilpisjärvi included two soil depths across four ecosystems that are characteristic of the tundra biome (barren soils, heathlands, meadows, and fens) (Fig. 1a-c). In previous studies, we have established in the area a systematic fine-scale sampling of microclimate, soil conditions, and vegetation in topographically distinct environments [42, 80, 81]. Local variation in topography and soil properties creates a mosaic of habitats characterized by contrasting ecological conditions. This makes the study setting ideal to investigate species-environment relationships and ecosystem functioning in the tundra [42, 82, 83].
我们在基尔皮斯亚尔维的采样设计包括了四种典型的苔原生态系统中的两种土壤深度（贫瘠土壤、荒地、草地和沼泽）（图 1a-c）。在先前的研究中，我们在该地区建立了一个系统性的微气候、土壤条件和植被的精细尺度采样，覆盖了地形明显不同的环境[42, 80, 81]。地形和土壤性质的局部变化形成了一系列栖息地，其生态条件迥异。这使得研究环境成为研究苔原中物种与环境关系和生态系统功能的理想场所[42, 82, 83]。

Physicochemical composition varied across samples (Additional file 1: Table S1). Soil ecosystem and depth explained a significant fraction of the variation in , gravimetric soil moisture, and SOM across samples (PERMANOVA, ) (Additional file 2: Fig. S1). Samples from the organic layer had higher moisture and SOM than samples from the mineral layer (one-way ANOVA, ), while did not vary significantly between soil layers (oneway ANOVA, ). Soil physicochemical properties did not differ across soil ecosystems in the mineral layer (one-way ANOVA, ). In the organic layer, however, fens were characterized by higher , moisture, and content (one-way ANOVA, , and, together with the meadows, lower C:N ratio (one-way ANOVA, ) (Fig. 1d).
不同样本之间的物理化学组成有所变化（附加文件 1：表 S1）。土壤生态系统和深度解释了样本之间 、重量土壤湿度和 SOM 变化的显著部分（PERMANOVA， ）（附加文件 2：图 S1）。有机层样本的湿度和 SOM 高于矿质层样本（单因素方差分析， ），而 在土壤层间没有显著变化（单因素方差分析， ）。矿质层土壤物理化学性质在土壤生态系统间没有差异（单因素方差分析， ）。然而，在有机层，沼泽地的 、湿度和 含量较高（单因素方差分析， ， ，以及与草地一起，较低的 C:N 比（单因素方差分析， ）（图 1d）。

In situ measurements showed a high sink-source variability in net fluxes across the ecosystems (Fig. 1e). Although the average flux across all sites was small (net consumption of day ), high emission at rates of up to day was observed at the meadow sites. Likewise, strong consumption (up to day ) was observed particularly at the heathland and meadow sites. Net emissions were observed exclusively at the fen sites.
现场测量显示生态系统中净 通量存在很高的汇-源变异性（图 1e）。尽管所有站点的平均 通量较小（ 天净消耗），但在草地站点观察到高达 天 的高 排放。同样，在荒地和草地站点尤其观察到强烈的 消耗（高达 天 ）。净 排放仅在沼泽站点观察到。

Read-based analyses of unassembled SSU rRNA gene sequences showed that microbial community composition differed across the ecosystems, with fen soils harbouring contrasting microbial communities compared to the other ecosystems (PERMANOVA, , ) (Additional file 2: Fig. S2a). No differences in community structure were observed between soil depths or the interaction between soil ecosystem and depth (PERMANOVA, ). A significant relationship was observed between community structure and soil physicochemical properties , gravimetric soil moisture, and SOM; PERMANOVA, ), but not elevation (PERMANOVA, ). Due to the significant overlap between soil ecosystem and physicochemical composition (Additional file 2: Fig. S1), we used db-RDA with forward selection to investigate in more detail the links between community structure and the environment. The best model explaining community structure comprised soil ecosystem and , . Addition of elevation did not improve the model (db-RDA, ).
基于未组装的 SSU rRNA 基因序列的阅读分析显示，微生物群落组成在生态系统之间存在差异，与其他生态系统相比，沼泽土壤中存在截然不同的微生物群落（PERMANOVA， ， ）（附加文件 2：图 S2a）。在土壤深度或土壤生态系统与深度之间的相互作用方面没有观察到群落结构的差异（PERMANOVA， ）。观察到群落结构与土壤理化性质 、重量土壤湿度和 SOM 之间存在显著关系；PERMANOVA， ），但海拔高度没有（PERMANOVA， ）。由于土壤生态系统和理化组成之间存在显著重叠（附加文件 2：图 S1），我们使用前向选择的 db-RDA 更详细地研究群落结构与环境之间的联系。最佳模型解释群落结构包括土壤生态系统和 ， 。增加海拔并没有改善模型（db-RDA， ）。

Among previously described (i.e., not unclassified) taxa, microbial communities in barren, heathland, and meadow soils were dominated by aerobic and facultative anaerobic heterotrophs such as Acidipila/Silvibacterium, Bryobacter, Granulicella, Acidothermus, Conexibacter, Mycobacterium, Mucilaginibacter, Bradyrhizobium, and Roseiarcus (Additional file 2: Fig. S2b). On the other hand, fen soils were dominated by methanogenic archaea from the genera Methanobacterium and Methanosaeta and anaerobic bacteria such as Thermoanaerobaculum, Desulfobacca, and Smithella, but also the putative aerobic heterotroph Candidatus Koribacter. We did not observe a significant relationship between the abundance of individual microbial genera and flux rates (linear regression, ).
在先前描述的（即非未分类的）分类群中，贫瘠、荒地和草地土壤中的微生物群落主要由需氧和兼性厌氧异养营养者主导，如酸泡菌/银杆菌、苔藓细菌、颗粒细菌、酸热菌、连结细菌、分枝杆菌、黏液细菌、根瘤菌和罗斯亚克斯（附加文件 2：图 S2b）。另一方面，沼泽土壤主要由产甲烷古细菌属的产甲烷菌和甲烷古细菌以及厌氧细菌如热厌氧杆菌、硫酸还原杆菌和史密斯氏菌主导，但也有假定的需氧异养营养者 Koribacter 候选菌。我们没有观察到个别微生物属的丰度与 通量率之间的显著关系（线性回归， ）。

Communities from different ecosystems also differed in their functional potential (Additional file 2: Fig. S2c). Denitrification genes (nirK, nirS, norB, and nosZ) were in general more abundant in the meadows and fens (oneway ANOVA, ) (Additional file 2: Fig. S2d). Fen soils, which had the highest gravimetric soil moisture content, also had a higher abundance of genes involved in sulfate reduction ( and ) and methanogenesis ( and mcrB) (one-way ANOVA, ), indicating the prevalence of anoxic and reductive soil conditions in these wet sites. We did not observe a significant relationship between flux rates and neither the abundance of individual denitrification genes nor the ratio between nos and nirK+nirS abundances (linear regression, ).
不同生态系统的社区在功能潜力上也存在差异（附加文件 2：图 S2c）。总体上，草甸和沼泽中的反硝化基因（nirK、nirS、norB 和 nosZ）更丰富（单向 ANOVA， ）（附加文件 2：图 S2d）。沼泽土壤具有最高的重量比土壤含水量，同时也具有更多参与硫酸盐还原（ 和 ）和产甲烷（ 和 mcrB）的基因（单向 ANOVA， ），表明这些湿地具有缺氧和还原性土壤条件的普遍存在。我们没有观察到 通量速率与反硝化基因的个体丰度或 nos 与 nirK+nirS 丰度比之间的显著关系（线性回归， ）。
However, the ratio between nos and nirK + nirS abundances was higher in the meadows (one-way ANOVA, ) (Additional file 2: Fig. S2d), which indicates a higher potential for consumption in this ecosystem.
然而，草地中 nos 和 nirK + nirS 丰度之间的比率较高（单向 ANOVA， ）（附加文件 2：图 S2d），这表明该生态系统中对 消耗的潜力较高。

Using anvi'o [59], we obtained 8,043 genomic bins and manually curated these to a set of 796 medium- and high-quality MAGs ( complete and redundant according to the MIMAG standard [66]) (Additional file 1: Table S2, Additional file 2: Fig. S3). According to estimates based on domain-specific single-copy genes, the obtained MAGs were on average complete (minimum: , maximum: ) and redundant (minimum: 0.0%, maximum: 9.9%) (Additional file 1: Table S2). Phylogenomic analyses based on 122 archaeal and 120 bacterial single-copy genes placed the MAGs across 35 bacterial and archaeal phyla according to the GTDB classification (Additional file 2: Fig. S3). The most represented phyla were Acidobacteriota , Actinobacteriota , Proteobacteria (Alphaproteobacteria, ; Gammaproteobacteria, , Chloroflexota , and Verrucomicrobiota . Most MAGs belonged to genera that do not comprise formally described species, including 303 MAGs that were placed outside genus-level lineages currently described in GTDB and thus likely represent novel genera (Additional file 1: Table S2).
使用 anvi'o [59]，我们获得了 8,043 个基因组桶，并手动筛选出了一组 796 个中高质量的 MAGs（根据 MIMAG 标准[66]， 完整且 冗余）（附加文件 1：表 S2，附加文件 2：图 S3）。根据基于领域特异性单拷贝基因的估计，获得的 MAGs 平均 完整（最小值： ，最大值： ）且 冗余（最小值：0.0%，最大值：9.9%）（附加文件 1：表 S2）。基于 122 个古菌和 120 个细菌单拷贝基因的系统发育分析，根据 GTDB 分类，将 MAGs 分布在 35 个细菌和古菌门（附加文件 2：图 S3）。最常见的门包括酸杆菌门 ，放线菌门 ，变形菌门（α-变形菌 ；γ-变形菌 ），绿弯菌门 和隐球菌门 。大多数 MAGs 属于不包括正式描述的属，包括 303 个 MAGs，这些 MAGs 被放置在 GTDB 中目前描述的属级别线性之外，因此可能代表新属（附加文件 1：表 S2）。

To investigate their distribution across the different soil ecosystems, MAGs were dereplicated based on a ANI threshold, yielding a set of 761 non-redundant MAGs (Fig. 2). On average, of the reads from each sample were recruited by the set of non-redundant MAGs (minimum: , maximum: ). In agreement with the read-based assessment, we observed differences in MAG composition across the soil ecosystems, with only 50 MAGs shared between the heathland, meadow, and fen soils (Additional file 2: Fig. S4a). Fen soils harboured the highest number of MAGs, with an average of 155 MAGs per sample (Additional file 2: Fig. S4b). Although barren and fen soils had similar taxonomic richness according to the read-based estimates, only a small number of MAGs was detected in the barren soils (average of four MAGs per sample). This is likely a result of limited sampling and sequencing of this ecosystem, which consisted of four samples and a total of of metagenomic data (Additional file 1: Table S1). The number of MAGs in heathland and meadow soils was similar (average of 47 and 63 MAGs per sample, respectively) (Additional file 2: Fig. S4b). In general, barren, heathland, and meadow soils were dominated by the same
为了调查它们在不同土壤生态系统中的分布，基于 ANI 阈值对 MAGs 进行了去冗余处理，得到了一组 761 个非冗余 MAGs（图 2）。平均每个样本中有 的 reads 被非冗余 MAGs 招募（最小值： ，最大值： ）。与基于 reads 的评估一致，我们观察到不同土壤生态系统中 MAG 组成的差异，只有 50 个 MAGs 在荒地、草地和沼泽土壤之间共享（附加文件 2：图 S4a）。沼泽土壤中寄存了最多数量的 MAGs，每个样本平均有 155 个 MAGs（附加文件 2：图 S4b）。尽管根据基于 reads 的估计，荒地和沼泽土壤的分类丰富度相似，但在荒地土壤中只检测到少量 MAGs（每个样本平均四个 MAGs）。这可能是由于对这一生态系统的采样和测序有限，该生态系统由四个样本和 的宏基因组数据组成（附加文件 1：表 S1）。荒地和草地土壤中的 MAGs 数量相似（每个样本平均 47 和 63 个 MAGs）（附加文件 2：图 S4b）。通常，贫瘠的荒地和草地土壤主要由相同的土壤类型主导

Fig. 2 Microbial community composition across different soil ecosystems in the tundra. Relative abundance of 761 non-redundant metagenome-assembled genomes (MAGs) recovered from soils in Kilpisjärvi, northern Finland. Relative abundances were computed as a proportion of the reads mapping to each MAG. Phylum-level taxonomic assignments are shown for the major groups found. More information about the MAGs can be found in Additional file 1: Table S2. The scheme on the top of the figure represents ecosystem-level nitrous oxide fluxes based on in situ measurements (Fig. 1) and the abundance of denitrification genes based on a gene-centric analysis (Additional file 2: Fig. S2). The font size of denitrification genes represents their abundance across the different ecosystems
图 2 凝固带不同土壤生态系统中的微生物群落组成。来自芬兰北部 Kilpisjärvi 土壤中恢复的 761 个非冗余的基因组组装基因组（MAGs）的相对丰度。相对丰度被计算为映射到每个 MAG 的读数的比例。显示了主要群体的门水平分类。有关 MAGs 的更多信息，请参阅附加文件 1：表 S2。图上方的方案代表基于原位测量（图 1）和基于基因为中心的分析（附加文件 2：图 S2）的反硝化基因丰度的生态系统级氧化亚氮通量。反硝化基因的字体大小表示它们在不同生态系统中的丰度。

set of MAGs (Additional file 2: Fig. S4c). These included members of the Acidobacteriota (Sulfotelmatobacter and unclassified genera in the class Acidobacteriae), Actinobacteriota (Mycobacterium and unclassified genera in the family Streptosporangiaceae), and Proteobacteria (Alphaproteobacteria: Reyranella, Bradyrhizobium, and unclassified Xanthobacteraceae; Gammaproteobacteria: unclassified Steroidobacteraceae). On the other hand, fen soils were dominated by MAGs that were not assigned to formally described genera, including lineages of Acidobacteriota (family Koribacteraceae), Actinobacteriota (family Solirubrobacteraceae), Chloroflexota (class Ellin6529), Desulfobacterota (order Desulfobaccales), and Halobacterota.
一组 MAGs（附加文件 2：图 S4c）。其中包括酸杆菌门成员（Sulfotelmatobacter 和酸杆菌纲未分类属），放线菌门（分枝孢链菌科的分枝孢链菌属和未分类属），以及变形菌门（α-变形菌纲：Reyranella、Bradyrhizobium 和未分类的黄杆菌科；γ-变形菌纲：未分类的类固醇杆菌科）。另一方面，沼泽土壤主要由未分配到正式描述的属的 MAGs 主导，包括酸杆菌门（科 Koribacteraceae）、放线菌门（科 Solirubrobacteraceae）、绿弯菌门（纲 Ellin6529）、脱硫菌门（目 Desulfobaccales）和卤菌门。

To gain insights into the microorganisms involved with the cycling of in tundra soils, we traced the curated denitrification genes to the set of recovered MAGs. Denitrification genes were found in 110 of the 796 MAGs (13.8%) (Additional file 1: Table S2). These were affiliated with the archaeal phylum Thermoproteota and many bacterial phyla such as Proteobacteria (classes Gamma- and Alphaproteobacteria), Acidobacteriota, Bacteroidota, Actinobacteriota, Chloroflexota, and Verrucomicrobiota (Fig. 3a). However, only 17 MAGs were assigned to a validly described genera (Additional file 1: Table S2). These included members of the Acidobacteriota (Solibacter, Sulfotelmatobacter, Terracidiphilus, and Gaiella), Myxococcota (Anaeromyxobacter), Planctomycetota (Singulisphaera), Proteobacteria (Alphaproteobacteria: Bauldia, Bradyrhizobium, Methylocella, and Reyranella; Gammaproteobacteria: Gallionella and Rhizobacter), and Verrucomicrobiota (Lacunisphaera and Opitutus). On average, of the reads in each sample were recruited by all denitrifiers combined (minimum: , maximum: ). In general,
为了深入了解参与苔原土壤中 循环的微生物，我们将经过整理的反硝化基因追溯到一组恢复的 MAGs。在 796 个 MAGs 中发现了 110 个（13.8%）含有反硝化基因（附加文件 1：表 S2）。这些基因与古菌门热蛋白菌门以及许多细菌门相关，如变形菌门（Gamma-和 Alphaproteobacteria 类）、酸杆菌门、杆菌门、放线菌门、绿弯菌门和隐杆菌门（图 3a）。然而，只有 17 个 MAGs 被分配到一个有效描述的属（附加文件 1：表 S2）。这些包括酸杆菌门的成员（Solibacter、Sulfotelmatobacter、Terracidiphilus 和 Gaiella）、黏液球菌门（Anaeromyxobacter）、浮游球菌门（Singulisphaera）、变形菌门（Alphaproteobacteria：Bauldia、Bradyrhizobium、Methylocella 和 Reyranella；Gamma-蛋白菌门：Gallionella 和 Rhizobacter）以及隐杆菌门（Lacunisphaera 和 Opitutus）。平均每个样本中 的 reads 被所有反硝化细菌共同招募（最小值： ，最大值： ）。总的来说，

denitrifiers were most abundant in the fens (1.0-6.1%) and least abundant in the heathlands (0.4-2.1%).
脱氮菌在沼泽地最丰富（1.0-6.1%），在荒地最稀少（0.4-2.1%）。

Genes involved in denitrification were found exclusively in MAGs with truncated denitrification pathways, i.e., MAGs missing one or more genes involved in the complete denitrification process (Fig. 3a). Of the 110 MAGs harbouring denitrification genes, the vast majority encoded only one of the Nir, Nor, and Nos enzymes and no MAG encoded all the three enzymes required for the reduction of to . Unsurprisingly, co-occurrence of genes encoding the three enzymes was also not observed in any of the other genomic bins of lower quality that were discarded from the final MAG dataset (i.e., bins that were complete and/or redundant). To verify if microorganisms with truncated denitrification pathways are common in other tundra systems, we expanded our analysis to 1529 MAGs recovered from permafrost peatland, bog, and fen soils in Stordalen Mire, northern Sweden [37]. Among these, 225 MAGs (14.7%) contained denitrification genes (Additional file 2: Fig. S5). MAGs encompassed a similar taxonomic profile as observed in the Kilpisjärvi dataset, and MAGs with truncated denitrification pathways were also the norm in Stordalen Mire soils. Only one MAG, assigned to the Gammaproteobacteria genus Janthinobacterium, encoded all the Nir, Nor, and Nos enzymes required for the reduction of to .
参与反硝化的基因仅在具有截断反硝化途径的 MAGs 中发现，即缺少完整反硝化过程中一个或多个基因的 MAGs（图 3a）。在 110 个携带反硝化基因的 MAGs 中，绝大多数只编码 Nir、Nor 和 Nos 酶中的一个，没有任何 MAG 编码所需的三种酶来还原氮气到氮气。毫不奇怪，编码这三种酶的基因的共存也没有观察到任何其他低质量的基因组桶中，这些基因组桶被从最终 MAG 数据集中丢弃（即被认为是不完整和/或冗余的桶）。为了验证截断反硝化途径的微生物在其他苔原系统中是否普遍存在，我们将分析扩展到从瑞典北部 Stordalen Mire 的永冻泥炭地、泥沼和沼泽土壤中恢复的 1529 个 MAGs [37]。在这些 MAGs 中，225 个（14.7%）含有反硝化基因（附加文件 2：图 S5）。MAGs 具有与 Kilpisjärvi 数据集中观察到的类似的分类学特征，并且在 Stordalen Mire 土壤中，具有截断反硝化途径的 MAGs 也是常见的。 只有一个 MAG，分配给了伽马变形菌属紫色细菌属，编码了所有需要将 还原为 的 Nir、Nor 和 Nos 酶。

The reduction of to , performed by microorganisms harbouring the nirK or nirS genes, is the hallmark step of denitrification and is often referred to as denitrification stricto sensu as it involves the conversion of a soluble substrate to a gaseous product thus leading to the removal of from the system [12]. Of the 110 Kilpisjärvi MAGs harbouring genes involved in denitrification, 46 contained nirK/nirS genes and are thus potential denitrifiers stricto sensu (Fig. 3a). These belonged mainly to the bacterial phyla Chloroflexota, Actinobacteriota, and Proteobacteria (classes Alpha- and Gammaproteobacteria). Most MAGs contained the nirK gene, which encodes the copper-containing form of Nir (Additional file 2: Fig. S6a). The nirS gene encoding the cytochrome -containing form of Nir was present in four Gammaproteobacteria MAGs (Additional file 2: Fig. S6b), including one MAG that contained both genes.
的还原到 ，由携带 nirK 或 nirS 基因的微生物执行，是反硝化的标志性步骤，通常被称为反硝化严格意义上，因为它涉及将可溶性底物转化为气态产物，从而导致 从系统中去除[12]。在 110 个携带参与反硝化基因的 Kilpisjärvi MAGs 中，有 46 个含有 nirK/nirS 基因，因此可能是反硝化者严格意义上（图 3a）。这些主要属于细菌门 Chloroflexota、Actinobacteriota 和 Proteobacteria（阿尔法-和伽马-蛋白菌纲）。大多数 MAGs 含有 nirK 基因，该基因编码铜含量形式的 Nir（附加文件 2：图 S6a）。编码细胞色素 -含量形式的 Nir 的 nirS 基因存在于四个伽马-蛋白菌 MAGs 中（附加文件 2：图 S6b），包括一个同时含有两种基因的 MAG。

The composition of potential denitrifier stricto sensu communities differed across the ecosystems (Fig. 3b). MAGs belonging to the Alphaproteobacteria class of the Proteobacteria were the most abundant in the barren, heathland, and meadow soils, particularly the MAG KUL-0154 assigned to the genera Bradyrhizobium (Fig. 4). Two other Alphaproteobacteria MAGs that do not correspond to formally described genera in the families Acetobacteraceae and Beijerinckiaceae (KUL-0057 and KUL-0056, respectively) were also found at high abundances. In addition, one Actinobacteriota MAG assigned to an uncharacterized genus in the family Gaiellaceae (KWL-0073), was abundant in the meadow soils. On the other hand, fen communities were dominated by MAGs belonging to the phylum Chloroflexota (Fig. 3b), which included seven MAGs assigned to the class-level lineage Ellin6529 (Fig. 4).
潜在反硝化菌群的组成在不同生态系统中有所不同（图 3b）。属于变形菌门中的 Alphaproteobacteria 纲的 MAGs 在贫瘠、荒地和草地土壤中最为丰富，特别是属于 Bradyrhizobium 属的 MAG KUL-0154（图 4）。另外，还发现了两个属于 Acetobacteraceae 和 Beijerinckiaceae 家族中未正式描述的属的 Alphaproteobacteria MAGs（分别为 KUL-0057 和 KUL-0056），且丰度较高。此外，一种属于 Gaiellaceae 家族中未描述的属的放线菌门 MAG（KWL-0073）在草地土壤中丰富。另一方面，沼泽群落主要由属于绿弯菌门的 MAGs 主导（图 3b），其中包括七个属于 Ellin6529 级别的 MAGs（图 4）。

None of the Ellin6529 MAGs that were dominant in the fen communities contained the key genes involved in autotrophic carbon fixation, dissimilatory sulfate reduction, dissimilatory nitrate reduction to ammonia, and nitrogen fixation (Additional file 1: Table S3). Analysis of genes encoding terminal oxidases involved in the aerobic respiratory electron chain revealed that all seven Ellin6529 MAGs harboured the coxABC genes encoding the aa3-type cytochrome c oxidase. Four MAGs also contained the cydAB genes encoding the cytochrome ubiquinol oxidase, a terminal oxidase with high affinity for oxygen that also plays a role in preventing the inactivation of oxygen-sensitive enzymes and protecting against oxidative and nitrosative stress, toxic compounds such as cyanide, and other stress conditions such as high temperature and high . The dominant MAGs in the barren, heathland, and meadow soils encoded a different set of aerobic terminal oxidases. In addition to the cydAB genes, the MAGs KUL-0057 and KUL0154 also contained the cyoABCD genes encoding the cytochrome o ubiquinol oxidase, which is the main terminal oxidase under highly aerobic conditions [86], and KUL-0057 also contained genes encoding the cbb3-type cytochrome c oxidase, a terminal oxidase with high affinity for oxygen [87]. Genes involved in the Calvin cycle (e.g., rbcL, rbcS, and ) were found in the Bradyrhizobium MAG (KUL-0154), and none of the key genes for autotrophic carbon fixation pathways were present in the other Alphaproteobacteria MAGs that were dominant in the barren, heathland, and meadow soils.
在沼泽社区中占主导地位的 Ellin6529 MAGs 中，没有包含参与自养碳固定、异化硫酸盐还原、异化硝酸盐还原为氨和固氮的关键基因（附加文件 1：表 S3）。对涉及有氧呼吸电子链中的终端氧化酶基因进行分析显示，所有七个 Ellin6529 MAGs 都携带了编码 aa3 型细胞色素 c 氧化酶的 coxABC 基因。四个 MAGs 还包含编码终端氧化酶 cytochrome ubiquinol 氧化酶的 cydAB 基因，这是一种对氧气具有高亲和力的终端氧化酶，还在预防对氧敏感酶的失活和保护免受氧化和亚硝化应激、氰化物等有毒化合物以及高温和高压等其他应激条件方面发挥作用。在贫瘠、荒地和草地土壤中占主导地位的 MAGs 编码了一组不同的有氧终端氧化酶。 除了 cydAB 基因外，MAGs KUL-0057 和 KUL0154 还包含编码细胞色素 o 泛醌氧化酶的 cyoABCD 基因，这是在高氧条件下的主要末端氧化酶[86]，而 KUL-0057 还包含编码 cbb3 型细胞色素 c 氧化酶的基因，这是一种对氧气具有高亲和力的末端氧化酶[87]。在 Bradyrhizobium MAG（KUL-0154）中发现了参与 Calvin 循环的基因（例如 rbcL、rbcS 和 ），而在贫瘠、荒地和草地土壤中占主导地位的其他 Alphaproteobacteria MAGs 中没有发现参与自养碳固定途径的关键基因。

The stepwise reduction of to and carried out by microorganisms containing the norB and nos genes, respectively, represents the final step of denitrification and the main biotic control on emissions. Soil denitrification rates depend on multiple environmental conditions such as adequate moisture and inorganic availability, but whether it results in the emission of or is ultimately linked to a balance between the activity of and reducers [11, 15]. norB and genes were identified in 62 and 9 Kilpisjärvi MAGs, respectively, belonging mostly to the phyla Actinobacteriota, Bacteroidota, Acidobacteriota, and Proteobacteria (class Alphaproteobacteria) (Fig. 3a). Apart from one Gemmatimonadota and one Acidobacteriota MAG, norB- and nosZ-containing MAGs were almost exclusively non-denitrifiers stricto sensu, i.e., they did not harbour the nirK/nirS genes involved in the reduction of to . Most MAGs harboured a norB gene encoding the monomeric, quinol-dependent form of Nor ( ), while the remaining MAGs encoded the cytochrome c-dependent Nor (cNor) (Additional file 2: Fig. S6c). In regards to the nos gene, most MAGs ( ) contained sequences affiliated with the clade II (also known as atypical) NosZ [14, 15, 27] (Additional file 2: Fig. S6d). Only four MAGs contained both the norB and nos genes and thus have the potential to reduce completely to (Fig. 3a).
微生物中含有 norB 和 nos 基因的逐步还原 至 和 代表了反硝化的最后一步和 排放的主要生物控制。土壤反硝化速率取决于诸如充足湿度和无机 可用性等多种环境条件，但它是否导致 或 的排放最终取决于 和 还原剂之间活性平衡 [11, 15]。在 62 个和 9 个 Kilpisjärvi MAGs 中分别鉴定出 norB 和 基因，它们主要属于放线菌门、杆菌门、酸杆菌门和变形菌门（α-变形菌纲）（图 3a）。除了一个宝石藻门和一个酸杆菌门 MAG 外，几乎所有含有 norB 和 nosZ 的 MAGs 严格来说都不是反硝化菌，即它们不含有涉及将 还原为 的 nirK/nirS 基因。大多数 MAGs 含有编码单体、依赖喹诺醇的 Nor 形式（ ），而其余 MAGs 编码细胞色素 c 依赖的 Nor（cNor）（附加文件 2：图 S6c）。 关于 nos 基因，大多数 MAGs（ ）包含与 clade II（也称为非典型）NosZ [14, 15, 27]相关的序列（附加文件 2：图 S6d）。只有四个 MAGs 同时包含 norB 和 nos 基因，因此有潜力将 完全还原为 （图 3a）。

As observed for the denitrifier stricto sensu communities, the communities of potential and reducers also differed between the ecosystems (Fig. 3b). MAGs assigned to the Alphaproteobacteria class of the Proteobacteria were the most abundant in the barren, heathland, and meadow soils. In particular, the MAG KWL-0112 assigned to the genera Reyranella was the dominant norB-containing MAG, while KUL-0116 (belonging to an uncharacterized genus in the family Acetobacteraceae) was the dominant MAG harbouring the
观察到，与反硝化细菌严格意义上的群落一样，潜在 和 还原剂的群落在生态系统之间也有所不同（图 3b）。被归类为变形菌门中的阿尔法变形菌纲的 MAG 在贫瘠、荒地和草地土壤中最为丰富。特别是，属于 Reyranella 属的 MAG KWL-0112 是主导的 norB 含有 MAG，而属于醋酸杆菌科未分类属的 MAG KUL-0116 是主导的 MAG，其中含有

KWL-0567 Acidobacteriota, Thermoanaerobaculia, Thermoanaerobaculales, Thermoanaerobaculaceae KWL-0108 KWL-0248 KWLL-O349
KWL-0567 酸杆菌门，热厌氧菌纲，热厌氧菌目，热厌氧菌科 KWL-0108 KWL-0248 KWLL-O349
KWL-0242
KWL-0133 KWL-007 KWL.-05
KWL-O510 KWL.-510
KWL-0381 KWL-0276 KWL-025
KWL-0150 KWL-0330
KWL-0246 KWL-0246
KWL-024 KWL-0243
KWL-0143 KWL-0143
KWL-0141 KWL-0141
KWL-0123 KUL-0057
KUL-0154
KUL-0056 KUL-0056
KUL-0047 KUL-0047
KWL--185
KWL-132
KWL-361 KWL-0240
KUL-0171
KUL-0170 KUL-0170
KWL-0261 KWL-0261
KWL-0232 KWL-0310
KWL.02 KWL-0240
KWL-0226 KUL-0148
KUL-0198 KWL-0480 KWL-0111
KWL-0516 KWLL-0099
KWL-0292 KWL-0097
KWL-0103 KWLLO103
KWLL-0265 KWL-0184
KWL-0385 KWL-0326
KWL-021 KWL.O494
KWL-0052 KWL-0052
KWL-0229 KWL-0241
KWL-0347 KWL-0560
KWL-0356 KWL-0331
KWL-0472
KWL.-0277 KWL-0277
KWL-0459 KWL-0456 KWL-0446
KWL-0437, KWL-0437 KWL-0437，KWL-0437
KWL-0018
KWL-0014 KWL-0001 KWL-0274
KWL-0019 KWL-0012 KUL-0116 KUL-0116
KWL-0408 KWL-0171
KWL-0112 KWL-0112
KWL-0095 KWL-0289
KWL-0288 KWL-023 KWL-0022 L-0116 Proteobacteria, Alphaproteobacteria, Acetobacterales, Acetobacteraceae, BOG-930
KWL-0288 KWL-023 KWL-0022 L-0116 草酸杆菌门, 甲基杆菌目, 醋酸杆菌科, 醋酸杆菌属, BOG-930

Fig Relative abundance of metagenome assembled genomes (MAGS) harbouring de tundra. MAGs were recovered from soils in Kilpisjärvi, northern Finland, and annotated for genes encoding the nitrite (nirK/nirS), nitric oxide (norB), and nitrous oxide (nosZ) reductases using a three-step approach (see methods). Relative abundances were computed as the proportion of reads mapping to each MAG. MAG taxonomy is based on the Genome Taxonomy Database (GTDB) release 05-RS95
图 2.在芬兰北部 Kilpisjärvi 的土壤中恢复的含有苔原生基因组组装基因组（MAGS）的相对丰度。使用三步方法（见方法）对 MAG 进行了注释，以编码亚硝酸盐（nirK/nirS）、一氧化氮（norB）和氧化亚氮（nosZ）还原酶的基因。相对丰度被计算为映射到每个 MAG 的 reads 的比例。MAG 分类基于基因组分类数据库（GTDB）发布 05-RS95。

nos gene (Fig. 4). On the other hand, fen communities were dominated by Acidobacteriota MAGs (Fig. 3b), particularly the norB- and nosZ-containing MAG KWL-0326 affiliated with the class Thermoanaerobaculia (Fig. 4). This MAG contained the same set of genes encoding aerobic terminal oxidases as found in the nirK-containing
在另一方面，酚类群落主要由酸杆菌门 MAGs（图 3b）主导，特别是与热厌氧菌纲（图 4）相关的 norB-和 nosZ 含有 MAG KWL-0326。这个 MAG 包含了与 nirK 含有的相同一组编码有氧末端氧化酶的基因。

Ellin6529 MAGs that were dominant in the fen sites, namely coxABC and cydAB (Additional file 1: Table S3). No genes involved in carbon fixation, dissimilatory sulfate reduction, dissimilatory nitrate reduction to ammonia, and nitrogen fixation were found in any of the dominant norB- and nos -containing MAGs.
Ellin6529 在沼泽地点中占主导地位的 MAGs，即 coxABC 和 cydAB（附加文件 1：表 S3）。在任何主导的 norB-和 nos -含有的 MAGs 中都没有发现参与碳固定、异化硫酸盐还原、异化硝酸盐还原为氨和固氮的基因。

To elucidate the phylogenetic placement of the Acidobacteriota MAGs and to verify if the potential for NO and reduction is present in other members of this phylum, we analysed all available genomes of Acidobacteriota strains and candidate taxa available on GenBank. This revealed that genes encoding the Nir and Nos enzymes are widespread across the phylum Acidobacteriota (Fig. 5). Genes encoding the Nor enzyme were present in all but one of the six Acidobacteriota subdivisions with genomes from cultured representatives. This included the strains Acidobacterium ailaaui PMMR2 (subdivision Gp1), Acidipila sp. 4G-K13 (Gp1), Silvibacterium bohemicum DSM 103,733 and S. bohemicum S15 (Gp1), Acidobacteriaceae bacterium URHE0068 (Gp1), Edaphobacter aggregans DSM 19,364 (Gp1), Luteitalea pratensis DSM 100,886 (Gp6), Geothrix fermentans DSM 14,018 (Gp8), and Thermoanaerobaculum aquaticum MP-01 (Gp23), as well as the candidate taxa Candidatus Koribacter versatilis Ellin345 (Gp1), Candidatus Sulfotelmatomonas gaucii SbA5 (Gp1), and Candidatus Solibacter usitatus Ellin6076 (Gp3). On the other hand, genes encoding the Nos enzyme were found only in members of the subdivisions Gp6 and Gp23.
为了阐明酸杆菌 MAGs 的系统发育位置，并验证该门其他成员是否具有还原 NO 和 的潜力，我们分析了 GenBank 上所有可用的酸杆菌菌株和候选分类单元的基因组。结果显示，编码 Nir 和 Nos 酶的基因在酸杆菌门中广泛分布（图 5）。编码 Nor 酶的基因存在于所有六个具有培养代表基因组的酸杆菌亚门中，唯一例外。这包括酸杆菌属细菌 URHE0068（Gp1）、土壤酸杆菌 DSM 19,364（Gp1）、草原黄色细菌 DSM 100,886（Gp6）、地球硫酸盐还原细菌 DSM 14,018（Gp8）和水热厌氧杆菌 MP-01（Gp23），以及候选分类单元多功能变形杆菌 Ellin345（Gp1）、硫泉酸杆菌 SbA5（Gp1）和土壤酸杆菌 Ellin6076（Gp3）。另一方面，编码 Nos 酶的基因仅在 Gp6 和 Gp23 亚门的成员中发现。

The 796 MAGs obtained in the present study by a manual binning and curation effort represent one of the largest genomic catalogues of microorganisms from tundra soils to date. Earlier gene-centric investigations have revealed the potential for complete denitrification in tundra soils [22, 88], however, these approaches fail to reveal the wider genomic context of the genes involved in this pathway. By applying the genome-resolved metagenomics approach, we traced denitrification genes to specific microbial populations, thereby allowing a detailed investigation of the genomic makeup of potential denitrifiers in tundra soils. This approach also enabled us to access the genomes of uncultured, poorly characterized taxa, which comprise the majority of the microorganisms in soils and other complex ecosystems .
本研究通过手动分箱和整理工作获得的 796 个 MAGs 代表迄今为止来自苔原土壤的微生物的最大基因组目录之一。早期基因为中心的研究揭示了苔原土壤中完全反硝化的潜力，然而，这些方法未能揭示参与该途径的基因的更广泛的基因组背景。通过应用基因组解析的宏基因组学方法，我们将反硝化基因追溯到特定的微生物群体，从而允许对苔原土壤中潜在反硝化细菌的基因组构成进行详细调查。这种方法还使我们能够访问未培养的、鲜为人知的分类群的基因组，这些分类群占据了土壤和其他复杂生态系统中大多数微生物的比例。

Our genome-resolved survey revealed that denitrification across different tundra soil ecosystems is dominated by microorganisms with truncated denitrification pathways (i.e., harbouring only a subset of the genes required for complete denitrification), most of which represent poorly characterized taxa without cultured representatives. The congruence of these findings in both our original dataset of northern Finland soils and a re-analysis of a comprehensive metagenomic dataset from soils in northern Sweden [37] suggests that truncated denitrification pathways are not a methodological artifact arising from the metabolic reconstruction of fragmented genomes. Indeed, recent genome-resolved investigations have shown that cross-feeding between microorganisms with truncated metabolic pathways, also known as metabolic handoffs, are the norm across a wide range of ecosystems such as grassland soil, aquifer sediment, groundwater, and the ocean, and not only in relation to denitrification but other redox transformations as well [30, 89, 90]. Although it has been established that denitrification is a community effort performed by different microbial populations [12-15], these genome-resolved metagenomic studies are beginning to reveal a more indepth, ecosystem-centric representation of the denitrification pathway. In addition to their predominance in genomic databases [14], it appears that truncated denitrifiers are also dominant within defined ecosystems across various terrestrial and aquatic biomes, including the tundra. It has been suggested that the partitioning of metabolic pathways across different populations via metabolic handoffs is advantageous as it eliminates competition between enzymes accelerating substrate consumption and provides flexibility and resilience to the communities in face of environmental disturbances [30]. We further hypothesize that the predominance of denitrification pathways characterized mostly by metabolic handoffs in tundra soils could be related to limitation. If metabolic handoffs enable a more effective substrate consumption as previously suggested , truncated denitrification pathways would be favoured in tundra soils which are mostly limited but undergo rapid surges in availability, e.g., during the spring melting season [91].
我们的基因组解析调查揭示，不同苔原土壤生态系统中的反硝化主要由具有截断反硝化途径（即仅携带完整反硝化所需基因的子集）的微生物主导，其中大多数代表着未经良好表征的分类群，没有培养代表。我们在北芬兰土壤的原始数据集和对瑞典北部土壤的一项全面宏基因组数据集的重新分析中发现的这些结果的一致性表明，截断反硝化途径并非源自于碎片化基因组的代谢重建的方法论偏差。事实上，最近的基因组解析调查表明，具有截断代谢途径的微生物之间的交叉供给，也被称为代谢交接，是各种生态系统的常态，如草地土壤、含水层沉积物、地下水和海洋，不仅涉及反硝化，还涉及其他氧化还原转化[30, 89, 90]。 尽管已经确定反硝化是由不同微生物群体共同完成的社区努力[12-15]，但这些基因组解析的宏基因组研究开始揭示反硝化途径更深入、生态系统为中心的表征。除了它们在基因组数据库中的主导地位[14]，截短的反硝化者似乎也在各种陆地和水生生物群落中占主导地位，包括苔原。已经提出通过代谢交接在不同群体之间分配代谢途径是有利的，因为它消除了酶之间的竞争，加速底物消耗 ，并为面对环境干扰的社区提供了灵活性和韧性[30]。我们进一步假设，在苔原土壤中主要由代谢交接特征化的反硝化途径的主导地位可能与 限制有关。 如果代谢交接能够像之前建议的那样实现更有效的底物消耗 ，那么在主要受 限制但在可用性迅速增加的苔原土壤中，截短的反硝化途径将受到青睐，例如在春季融化季节期间[91]。

Tundra ecosystems are typically heterogeneous. Previous studies in the Kilpisjärvi region have shown that soil properties such as and moisture do not have any strong relationship with the macrotopography of the area (50-500 m scale). Instead, environmental variation is controlled by the fine-scale mesotopographic variation of the relief ( scale), resulting in a mosaic of different soil ecosystems with contrasting vegetation [42, 80-83]. Our results agreed with this observation and showed that denitrifier communities in the tundra differ between drier upland ecosystems (barren, heathland, and meadow soils) and waterlogged fens. This is likely related to differences in soil moisture affecting oxygen availability in these ecosystems. The dominant denitrifier populations in the oxic dry upland soils, related to the genera Bradyrhizobium, Reyranella, and other uncharacterized genera in the class Alphaproteobacteria, encoded aerobic terminal
苔原生态系统通常是异质的。基尔皮斯亚尔维地区的先前研究表明，土壤性质，如 和湿度与该地区的宏观地形（50-500 米尺度）没有明显的关系。相反，环境变化受到地形细尺度（ 尺度）的控制，导致不同土壤生态系统的马赛克，具有对比鲜明的植被[42, 80-83]。我们的结果与这一观察一致，并显示苔原中的反硝化菌群在干燥的高地生态系统（贫瘠、荒地和草地土壤）与水浸泽地之间存在差异。这可能与土壤湿度的差异有关，影响了这些生态系统中氧气的可用性。氧化性干燥高地土壤中的优势反硝化菌群，与 Bradyrhizobium 属、Reyranella 属和其他未经表征的 Alphaproteobacteria 纲属编码的有氧终端

Fig. 5 Metabolic potential for denitrification among members of the phylum Acidobacteriota. Phylogenomic analysis of 85 Acidobacteriota metagenome-assembled genomes (MAGs) containing denitrification genes recovered from tundra soils in Kilpisjärvi (northern Finland) and Stordalen Mire (northern Sweden), and 69 genomes of Acidobacteriota strains and candidate taxa. Maximum likelihood tree based on concatenated alignments of 23 ribosomal proteins and rooted with Escherichia coli ATCC 11775 (not shown). Genes encoding the nitrite (nirk), nitric oxide (norB), and nitrous oxide (nos reductases were annotated using a three-step approach (see Methods)
图 5 酸杆菌门成员中反硝化代谢潜力。对来自芬兰北部 Kilpisjärvi 和瑞典北部 Stordalen Mire 苔原土壤中含有反硝化基因的 85 个酸杆菌门代谢组装基因组（MAGs）进行系统发育分析，以及 69 个酸杆菌门菌株和候选分类群的基因组。基于 23 个核糖体蛋白的串联比对构建的最大似然树，并以大肠杆菌 ATCC 11775 为根（未显示）。使用三步法注释编码亚硝酸盐（nirk）、一氧化氮（norB）和氧化亚氮（nos）还原酶的基因（详见方法）。

oxidases that are active under highly aerobic conditions as well as oxidases with high oxygen affinity [86, 87]. The former likely provides an adaptive advantage in these soils by allowing rapid aerobic growth under standard conditions of high oxygen availability, and the latter would sustain growth in microoxic niches within the soil matrix and during periods of reduced oxygen availability (e.g., during the spring melting season).
在高度好氧条件下活跃的氧化酶以及具有高氧亲和力的氧化酶[86, 87]。前者可能通过在高氧可用性的标准条件下快速进行好氧生长来为这些土壤提供适应性优势，后者将在土壤基质中的微氧缺口以及氧供应减少时（例如，在春季融化季节）维持生长。

On the other hand, fen soils are continuously inundated because they are located at lower topographic positions where the water table is permanently at or near the soil surface. The result is a mostly anoxic environment due
另一方面，沉积土壤不断被淹没，因为它们位于较低的地形位置，地下水位永久处于或接近土壤表面。结果是由于缺氧环境主要是
to the slow rate at which oxygen diffuses into the waterlogged soil, favouring reduced rather than oxidized soil chemistry. In line with this, we found a predominance of anaerobic processes in the fens, including a higher abundance of genes involved in denitrification, sulfate reduction, and methanogenesis, the latter supported by in situ measurements showing net emission at the fen sites. Communities of potential denitrifiers in the fen soils were dominated by somewhat enigmatic taxa, namely potential reducers affiliated with the class Ellin6529 of the Chloroflexota and reducers assigned to the subdivision Gp23 of the Acidobacteriota. Both groups are major members of microbial communities in soils worldwide [92], and RNA-based investigations have shown that they are active in tundra soils during both summer and winter seasons [42, 93]. Thermoanaerobaculum aquaticum MP-01, the only cultivated member of the Acidobacteriota subdivision Gp23, is a strictly anaerobic bacterium that has been shown to use Fe and , but not nor , as electron acceptors in anaerobic respiration [94]. However, studies investigating the use of nitrogen oxides in anaerobic respiration usually provide soluble or as electron acceptors, not the gases and , which bias against truncated denitrifiers that do not contain the narG and nirK/nirS genes [95]. Ellin6529-formerly G04-were first detected by culture-independent methods in alpine tundra wet meadow soil in the Colorado Rocky Mountains, USA [96], and later isolated in a study targeting slow-growing and mini-colony forming bacteria from Australian agricultural soil [97]. However, their ecological, physiological, and metabolic preferences remain largely unknown. Their genomic composition and high abundance in the water-logged, anoxic fen soils suggest that the Ellin6529 and Gp23 populations found in this study are likely able to grow anaerobically with the use of and as electron acceptors. However, it is known that in addition to their role in anaerobic respiration, and reduction can be used as a detoxification mechanism or as electron sink for metabolism. For example, the aerobe Gemmatimonas aurantica T-27 is not able to grow on alone, but can use as electron acceptor transiently when oxygen is depleted [98].
由于氧气以缓慢的速率扩散进入积水的土壤，有利于还原而不是氧化的土壤化学反应。与此一致，我们发现沼泽地主要是厌氧过程，包括参与反硝化、硫酸盐还原和产甲烷的基因丰度更高，后者得到了现场测量的支持，显示沼泽地点净排放甲烷。沼泽土壤中潜在反硝化菌群以某种神秘的类群为主导，即与叶绿菌门 Ellin6529 类相关的潜在还原菌和酸杆菌门 Gp23 亚门的还原菌。这两个群体是全球土壤微生物群落的主要成员[92]，基于 RNA 的研究表明它们在夏季和冬季的苔原土壤中活跃[42, 93]。酸杆菌门 Gp23 亚门唯一培养的成员 Thermoanaerobaculum aquaticum MP-01 是一种严格厌氧细菌，已被证明在厌氧呼吸中使用 Fe 和，但不使用或作为电子受体[94]。 然而，研究调查厌氧呼吸中氮氧化物的使用通常提供可溶性 或 作为电子受体，而不是气体 和 ，这会偏向不含 narG 和 nirK/nirS 基因的截短反硝化细菌[95]。Ellin6529-前 G04-首次通过无培养方法在美国科罗拉多洛矶山脉的高山苔原湿地土壤中被检测到[96]，后来在一项针对澳大利亚农业土壤中生长缓慢和微小菌落形成细菌的研究中被分离[97]。然而，它们的生态、生理和代谢偏好仍然大部分未知。它们在水浸、缺氧的沼泽土壤中的基因组成和高丰度表明，本研究中发现的 Ellin6529 和 Gp23 种群可能能够利用 和 作为电子受体进行厌氧生长。然而，除了在厌氧呼吸中的作用外， 和 还可以用作解毒机制或作为代谢的电子汇。 例如，需氧菌 Gemmatimonas aurantica T-27 无法单独在 上生长，但当氧气耗尽时，可以短暂地使用 作为电子受体[98]。

In addition to microbial community structure, differences in fluxes observed between upland and fen soils also appear to be linked to soil moisture. Some of the drier upland sites investigated were hotspots of consumption. This is particularly interesting for the acidic heathland soils, as low is known to impair the expression of the NosZ enzyme thus promoting emission . On the other hand, fens had close to net-zero fluxes, which is in line with previous observations for water-saturated soils both in the tundra [7] and worldwide [11, 13]. This has been linked to lower rates of mineralization and nitrification in anoxic ecosystems, which limit the availability of and and promote complete denitrification, resulting in as end product rather than . Indeed, supplementing fen soils in the tundra with and has shown to promote emissions [101]. Moreover, climate change models predict lowering of the water table in high-latitude wetlands, which could lead to increased emissions from these ecosystems which contain substantial amounts of both and bound to the soil organic matter .
除了微生物群落结构之外，高地和沼泽土壤之间观察到的 通量差异似乎也与土壤湿度有关。一些研究的较干燥的高地站点是 消耗的热点。这对于酸性荒地土壤尤其有趣，因为已知低 会影响 NosZ 酶的表达，从而促进 排放 。另一方面，沼泽土壤的 通量接近净零，这与之前在苔原[7]和全球范围内[11, 13]对水饱和土壤的观察一致。这与缺氧生态系统中 矿化和硝化速率较低有关，这限制了 和 的可用性，并促进了完全的反硝化，导致 成为最终产物，而不是 。事实上，在苔原的沼泽土壤中添加 和 已经显示出促进 排放[101]。此外，气候变化模型预测高纬度湿地的地下水位下降，这可能导致这些含有大量 和 的生态系统排放增加，这些物质与土壤有机物 结合。