Molecular data for Davis 14/15 ocean acidification minicosm experiment

Experimental Design A six-level, dose-response ocean acidification experiment was run on a natural microbial community from nearshore Antarctica, between 19th November and 7th December 2014. Seawater was collected from approximately 1 km offshore of Davis Station, Antarctica (68◦ 35’ S, 77◦ 58’ E), pre-filtered (200 μm), and transferred into six 650 L tanks (minicosms) located in a temperature-controlled shipping container. Six CO2 levels were achieved by altering the fugacity of carbon dioxide (ƒCO2) within each minicosms. The ƒCO2 was adjusted stepwise to the target concentrations for each minicosm (343, 506, 634, 953, 1140, 1641 μatm) over a five-day period using 0.2 μm filtered seawater enriched with CO2. This acclimation to CO2 was conducted at low light (0.9 ± 0.2 μmol m−2 s−1) so there was low growth of the phytoplankton. Light levels were then increased over a further two days to 90.52 ± 21.45 μmol m−2 on a 19:5 light/dark non-limiting light cycle. After this acclimation period, the microbial community was allowed to grow for 10 days (days 8-18), during which the ƒCO2 levels within each minicosm was adjusted daily to maintain the target ƒCO2 level for each minicosm, and light levels were kept constant. No nutrients were added during the experiment. For a more detailed description of minicosm set-up, lighting and carbonate chemistry see;Davidson, A. T., McKinlay, J., Westwood, K., Thomson, P. G., van den Enden, R., de Salas, M., Wright, S., Johnson, R., and Berry, K.:Enhanced CO2 concentrations change the structure of Antarctic marine microbial communities, Mar. Ecol. Prog. Ser., 552, 93-113, 2016.Deppeler, S. L., Petrou, K., Westwood, K., Pearce, I., Pascoe, P., Schulz, K. G., and Davidson, A. T. Ocean acidification effects on productivity in a coastal Antarctic marine microbial community, Biogeosciences, 15(1), 2018.Sample Collection Samples of 40-400 L were collected and sequentially size-fractionated filtered onto 293 mm biomass filters with 3.0 and 0.1 μm pore-sized polyethersulfone membrane filters (Pall XE20206 Disc 3.0 μm Versapor 293 mm and 656552 Disc 0.1 μm Supor 293 mm) using the design of the Global Ocean Sampling expedition (Rusch et al., 2007). Samples were collected on days 0 (immediately after seawater collection), 12 (mid-exponential growth) and 18 (end of experiment). On day 0, 400 L of seawater was collected from the reservoir tank (pre-filtered 200 μm), from which all the minicosms were filled, to allow characterisation of the initial community. This sample was collected from the reservoir, and not the minicosms, due to the large volume needed to collect sufficient microbial biomass on the filters. On day 12 and 18, 40 L was collected from each minicosm for filtration. The later samples were of a smaller volume due to the increase in biomass in the minicosms during the experiment, meaning less volume of water was required to gain sufficient material on the filters to perform molecular analysis. The filter membranes containing the concentrated microbial biomass were stored in 15 mL of storage buffer, flash frozen in liquid nitrogen and stored at - 80◦C. The storage buffer was freshly prepared on each sampling day with a mixture of 2.5 mM EGTA, 2.5 mM EDTA, 0.1 mM Tris-EDTA, RNA Later (0.5x house prepared), 1 mM PMSF and Protease Inhibitor Cocktail VI (Ng et al., 2010). Between samples the filtration apparatus was sequentially washed with 2 x 25 L 0.1 M NaOH, 2 x 25 L 0.07% Ca(OCl)2 and 2 x 25 L fresh water. All samples were stored and transported at -80◦C to the Australian Antarctic Division, Hobart, Australia for DNA extraction. DNA Extraction and Sequencing The DNA was extracted from half of each filter (3.0 and 0.1 μatm per sample) via the method described in Rusch et al. (2007). In short, the filters were cut into small pieces and agitated in a lysozyme and sucrose buffer for 60 minutes and underwent three freeze/thaw cycles in a Proteinase K solution. This was followed by a gentler agitation at 55◦C for 2 hrs to remove all contents from the filter membranes. DNA was then separated using buffer saturated phenol, pelleted and washed in alcohol. The final DNA pellet was dissolved and stored in a 3 M sodium acetate (pH 8.0) and 100% ethanol solution and stored at - 80◦C. The DNA was transported and stored at 4◦C to the University of Queensland, St Lucia, Australia for sequencing within two months of extraction. Eukaryotic 18S rRNA genes (V8-V9 regions) were amplified using polymerase chain reaction (PCR) with the primers V8f (5’ - AT AAC AGG TCT GTG ATG CCC T - ’3) and 1510r (5’ - CCT TCY GCA GGT TCA CCT AC - ’3) (Bradley, 2016). The 16S rRNA genes V8 region were amplified using PCR and primers 926F (5’-AAA CTY AAA KGA ATT GAC GG-3’) and 1392wR (5’-ACG GGC GGT GTG RC-3’) (Engelbrektson et al., 2010). PCR was performed using 1 or 1.5 μL of sample DNA, 2.5 μL 1x PCR buffer minus Mg+2 (Invitrogen), 0.75 μL MgCl2, 0.5 μL deoxynucleoside triphosphate (dNTPs, Invitrogen), 0.125 μL U Taq DNA Polymerase (Invitrogen), 0.625 μL of forward/reverse primer and made up to the final volume of 25 μL using molecular biology grade water. Forward and reverse primers were modified at the 5’-end to contain an Illumina overhang adaptor with P5 and i7 Nextera XT indices, respectively. The PCR thermocycling conditions were as follows: 94◦C for 3 min, 35 cycles of 94◦C for 45 sec, 55◦C for 30 sec, 7◦C for 10 min and a final extension of 72◦C for 10 min. Amplifications were performed using a Vertiti®96-well Thermocycler (Applied Biosystems) and success, amplicon size and quality was determined by gel electrophoresis. The resultant amplicons were purified using Agencourt AMPure magnetic beads (Axygen Biosciences), dual indexed using Nextera XT Index Kit (Illumina). The indexed amplicons were purified using Agencourt AMPure XP beads and quantified using PicoGreen dsDNA Quantification Kit (Invitrogen). Equal concentrations of each sample were pooled and sequenced on an Illumina MiSeq at the University of Queensland’s School for Earth and Environmental Science using 30% PhiX Control v3 (Illumina) and a MiSeq Reagent Kit v3 (600 cycle; Illumina). Bioinformatics Sequencing data and runs were merged to produced single FASTQ file for 16S and 18S rDNA per sample and imported in QIIME2 (v2019.9) (Caporaso et al., 2010). A modified version of the UPARSE analysis pipeline was used to analyse the data. Specifically, the primer sequences were removed from forward reads of the 16S rDNA and reverse complement of the 18S rDNA Illumina read pairs, and chimeras removed using UCHIME2 (Edgar, 2016). These were then trimmed to a length of 200 bp and high-quality sequences identified using USEARCH (v10.0.240) (Edgar, 2010). Duplicate sequences were removed and a set of unique operational taxonomic units (OTUs) were generated using USEARCH employing a 97% OTU similarity radius. Mitochondrial and chloroplast OTUs were classified and removed from the 16S rDNA sequence data using the BIOM tool suite (McDonald et al., 2012). Representative OTU sequences were assigned taxonomy using SILVA132 (Quast et al., 2012) and PR2 (Guillou et al., 2012) for the eukaryotic group Bacillariophyceae (diatoms). Taxonomic assignments were validated against microscopy identifications conducted on the same samples (Chapter 3, Hancock et al. 2018) as well as phylogenetic trees built in iTOL (Letunic and Bork, 2006). Residual eukaryotic chloroplast and mitochondrial sequences were removed from the 16S rDNA data. Other obvious contaminants were removed manually including: Escherichia-Shigella (16S rDNA OTU75) and Saccharomycetales (18S rDNA OTU7, 146 and 160). Escherichia-shigella was removed as this group likely represents external contamination, similarly Saccharomycetales are yeast and are obvious skin-driven contaminants. A total of 9448 OTUs were identified from the 16S rDNA reads and 232 OTUs from the 18S rDNA read data. The number of reads were rarefied to 1300 and 1200 reads per sample for the 18S and 16S rDNA datasets respectively. The following samples were removed due to lack of extracted, amplified and/or sequenced DNA, or due to low quality reads and/or low read numbers:18S, 3.0 μm, day 18, 634 μatm ƒCO2 treatment18S, 0.1 μm, day 12, 343 μatm or control ƒCO2 treatment 18S, 0.1 μm, day 18, 343 μatm or control ƒCO2 treatment 16S, 0.1 μm, day 18, 506 μatm ƒCO2 treatment Statistical Analysis The minicosm experiment was based on a repeated measure design, therefore due to being a dose-response experiment with no replication, no formal statistics could be undertaken on the interactions between time and ƒCO2. The richness (number of taxa) and evenness (equivalent to abundances within a sample) of the eukaryotic and prokaryotic microbial communities within each minicosm over time was estimated using three different alpha diversity indexes: observed number of OTUs (Sobs) (DeSantis et al., 2006), the Chao1 estimator of richness (Colwell et al., 2004), and Simpson’s diversity index and Berger-Parker index which account for both richness and evenness (Simpson, 1949; Berger and Parker, 1970) using QIIME2. Clustering and ordinations were performed on Bray-Curtis resemblance matrices of the rarefied, square-root transformed OTU data as per Chapter 3 (Hancock et al., 2018). In brief, hierarchical agglomerative cluster analyses were performed using group-average linkage, and significantly different clusters were determined using similarity profile permutations method (SIMPROF) (Clarke et al., 2008). Both unconstrained (non-metric multidimensional scaling, nMDS) and constrained (canonical analysis of principal coordinates, CAP) ordinations were performed using the Bray-Curtis resemblance matrixes (Kruskal, 1964a,b; Oksanen et al., 2017). The constraining variables in the CAP analysis were ƒCO2, Si, P and NOx. All cluster and ordination analyses were performed using R v.1.1.453 (R Core Team, 2016) and the add-on package Vegan v.2.5-3 (Oksanen et al., 2017). A full description of the statistical methods used for this paper is described in;Hancock, A. M., Davidson, A. T., McKinlay, J., McMinn, A., Schulz, K. G., and van den Enden, R. L. Ocean acidification changes the structure of an Antarctic coastal protistan community, Biogeosciences, 15(1), 2018.

实验设计
本研究于2014年11月19日至12月7日开展了一项针对南极近岸天然微生物群落的六级剂量响应型海洋酸化实验。实验所用海水采集自南极戴维斯站（68°35′ S，77°58′ E）离岸约1km的海域，经200μm预过滤后，转移至位于控温集装箱内的6个650L中型实验生态箱（minicosms）中。通过调节每个中型生态箱内的二氧化碳分压（ƒCO2）设置6个CO₂水平：采用经0.2μm过滤的富CO₂海水，在5天内逐步将ƒCO2调整至目标浓度（343、506、634、953、1140、1641 μatm）。该CO₂驯化阶段在低光照条件（0.9 ± 0.2 μmol·m⁻²·s⁻¹）下进行，此时浮游植物生长速率较低；随后在2天内将光照强度提升至90.52 ± 21.45 μmol·m⁻²·s⁻¹，采用19:5的光暗周期，达到非限制性光照水平。驯化阶段结束后，微生物群落继续培养10天（第8-18天），期间每日调整各生态箱内的ƒCO2以维持目标值，光照强度保持恒定，且实验过程中未添加任何营养盐。关于中型实验生态箱的搭建、光照设置与碳酸盐化学特征的详细描述，请参见：Davidson, A. T., McKinlay, J., Westwood, K., Thomson, P. G., van den Enden, R., de Salas, M., Wright, S., Johnson, R., and Berry, K.:Enhanced CO2 concentrations change the structure of Antarctic marine microbial communities, Mar. Ecol. Prog. Ser., 552, 93-113, 2016；Deppeler, S. L., Petrou, K., Westwood, K., Pearce, I., Pascoe, P., Schulz, K. G., and Davidson, A. T. Ocean acidification effects on productivity in a coastal Antarctic marine microbial community, Biogeosciences, 15(1), 2018。

样本采集
采集40-400L海水样本，按照全球海洋采样考察（Global Ocean Sampling expedition，Rusch等，2007）的设计方案，依次进行分级过滤，将截留的微生物生物质富集于293mm规格的聚醚砜膜滤器上，滤膜孔径分别为3.0μm和0.1μm（型号分别为Pall XE20206 Disc 3.0 μm Versapor 293 mm与656552 Disc 0.1 μm Supor 293 mm）。样本采集时间为第0天（海水采集后立即）、第12天（指数生长中期）与第18天（实验结束）。第0天采集400L储水箱中的预过滤海水（200μm），该储水箱为所有中型生态箱的供水源，此举用于表征初始微生物群落；由于需要大体积海水以获取足够的微生物生物量，该样本直接取自储水箱而非生态箱。第12天与第18天，从每个中型生态箱中采集40L海水进行过滤；由于实验过程中生态箱内微生物生物量增加，仅需较小体积的海水即可获得足够用于分子分析的生物量。将载有富集微生物生物质的滤膜置于15mL储存缓冲液中，经液氮速冻后于-80℃保存。储存缓冲液于每个采样日新鲜配制，成分为：2.5mM EGTA、2.5mM EDTA、0.1mM Tris-EDTA、0.5×体积自制RNA Later、1mM PMSF与蛋白酶抑制剂混合物VI（Ng等，2010）。样本采集间隙，过滤装置依次用2×25L 0.1M NaOH、2×25L 0.07% Ca(OCl)2与2×25L新鲜水进行清洗。所有样本于-80℃条件下保存并运输至澳大利亚霍巴特的澳大利亚南极分部进行DNA提取。

DNA提取与测序
采用Rusch等（2007）的方法，从每个滤膜的一半（对应3.0μm与0.1μm孔径滤膜的各样本）中提取DNA。简要步骤如下：将滤膜剪碎，置于溶菌酶蔗糖缓冲液中振荡60分钟，随后在蛋白酶K溶液中经历三次冻融循环；接着在55℃下温和振荡2小时，以去除滤膜上的所有内容物；随后用饱和酚缓冲液分离DNA，经酒精沉淀与洗涤后，将最终的DNA沉淀溶解于3M乙酸钠（pH 8.0）与100%乙醇的混合溶液中，于-80℃保存。提取完成后的两个月内，将DNA置于4℃条件下运输至澳大利亚圣卢西亚的昆士兰大学进行测序。
采用聚合酶链式反应（polymerase chain reaction，PCR）扩增真核18S rRNA基因的V8-V9区，所用引物为V8f（5’-AT AAC AGG TCT GTG ATG CCC T -3’）与1510r（5’-CCT TCY GCA GGT TCA CCT AC -3’）（Bradley，2016）；扩增16S rRNA基因的V8区所用引物为926F（5’-AAA CTY AAA KGA ATT GAC GG-3’）与1392wR（5’-ACG GGC GGT GTG RC-3’）（Engelbrektson等，2010）。PCR反应体系如下：1或1.5μL样本DNA、2.5μL 1×不含Mg²+的PCR缓冲液（Invitrogen）、0.75μL MgCl2、0.5μL脱氧核苷三磷酸（dNTPs，Invitrogen）、0.125μL U Taq DNA聚合酶（Invitrogen）、0.625μL上下游引物，用分子生物学级水补至终体积25μL。上下游引物的5’端分别添加带有P5与i7 Nextera XT索引的Illumina接头。PCR热循环条件为：94℃预变性3分钟，35个循环（94℃变性45秒、55℃退火30秒、7℃延伸10分钟），最后72℃终延伸10分钟。扩增反应采用Vertiti®96孔热循环仪（Applied Biosystems）完成，通过凝胶电泳验证扩增成功与否、扩增产物的大小与质量。所得扩增产物经Agencourt AMPure磁珠（Axygen Biosciences）纯化，采用Nextera XT Index Kit（Illumina）进行双索引标记；标记后的扩增产物经Agencourt AMPure XP磁珠纯化，并用PicoGreen dsDNA定量试剂盒（Invitrogen）进行定量。将各样本按等浓度混合后，在昆士兰大学地球与环境科学学院采用Illumina MiSeq平台进行测序，测序时添加30% PhiX Control v3（Illumina）与MiSeq Reagent Kit v3（600循环；Illumina）。

生物信息学分析
将测序数据合并，为每个样本生成16S与18S rDNA的单FASTQ文件，随后导入QIIME2（v2019.9）（Caporaso等，2010）。采用修改版UPARSE分析流程对数据进行分析：首先去除16S rDNA正向读段与18S rDNA反向互补Illumina读段的引物序列，并用UCHIME2（Edgar，2016）去除嵌合体序列；随后将序列修剪至200bp长度，采用USEARCH（v10.0.240）（Edgar，2010）识别高质量序列；去除重复序列后，采用USEARCH以97%的操作分类单元（operational taxonomic unit，OTU）相似性阈值生成唯一OTU集合。采用BIOM工具套件（McDonald等，2012）对16S rDNA数据中的线粒体与叶绿体OTU进行分类并移除。采用SILVA132（Quast等，2012）与PR2（Guillou等，2012）对代表OTU序列进行分类学注释，针对真核类群硅藻门（Bacillariophyceae，diatoms）。分类学注释通过同一批样本的显微镜鉴定结果（第3章，Hancock等，2018）以及在iTOL中构建的系统发育树（Letunic和Bork，2006）进行验证。从16S rDNA数据中移除残留的真核叶绿体与线粒体序列，并手动移除其他明显污染物：埃希氏菌-志贺氏菌属（16S rDNA OTU75）与酵母菌目（Saccharomycetales，18S rDNA OTU7、146与160）；其中埃希氏菌-志贺氏菌属可能为外部污染物，酵母菌目为酵母类群，属于皮肤来源污染物。最终从16S rDNA读段中鉴定出9448个OTU，从18S rDNA读段中鉴定出232个OTU。对18S与16S rDNA数据集分别进行稀疏化处理，使每个样本的读段数分别达到1300与1200条。移除以下样本：因未成功提取、扩增和/或测序，或读段质量差、读段数过低的样本：18S，3.0μm，第18天，634 μatm ƒCO2处理组；18S，0.1μm，第12天，343 μatm或对照ƒCO2处理组；18S，0.1μm，第18天，343 μatm或对照ƒCO2处理组；16S，0.1μm，第18天，506 μatm ƒCO2处理组。

统计分析
本中型生态箱实验采用重复测量设计，且为无重复的剂量响应实验，因此无法对时间与ƒCO2的交互作用进行正式统计分析。采用三种不同的α多样性指数，估算每个中型生态箱内真核与原核微生物群落的丰富度（类群数量）与均匀度（样本内丰度分布情况）：观测OTU数量（Sobs）（DeSantis等，2006）、Chao1丰富度估计量（Colwell等，2004），以及同时考虑丰富度与均匀度的Simpson多样性指数与Berger-Parker指数（Simpson，1949；Berger和Parker，1970），分析过程采用QIIME2完成。对稀疏化且经平方根转换的OTU数据的Bray-Curtis相似性矩阵进行聚类与排序分析，详细方法参见第3章（Hancock等，2018）。简要步骤如下：采用组平均连锁法进行层次凝聚聚类分析，使用相似性轮廓置换法（SIMPROF）（Clarke等，2008）确定显著差异的聚类；采用Bray-Curtis相似性矩阵分别进行无约束排序（非度量多维标度，non-metric multidimensional scaling，nMDS）与约束排序（主坐标典范分析，canonical analysis of principal coordinates，CAP）（Kruskal，1964a,b；Oksanen等，2017）。CAP分析中的约束变量为ƒCO2、Si、P与NOx。所有聚类与排序分析采用R v.1.1.453（R核心团队，2016）及其扩展包Vegan v.2.5-3（Oksanen等，2017）完成。本研究所用统计方法的完整描述参见：Hancock, A. M., Davidson, A. T., McKinlay, J., McMinn, A., Schulz, K. G., and van den Enden, R. L. Ocean acidification changes the structure of an Antarctic coastal protistan community, Biogeosciences, 15(1), 2018。