Pan-Atlantic comparison of deep-water macrobenthos diversity collected by epibenthic sledge sampling and analysis of patterns and environmental drivers.

Epibenthic samples were collected during the following research expeditions: ANDEEP-SYSTCO II, BIOPEARL I, DIVA1-3, IceAGE1-3&RR, IceDIVA1,2, JR275 and Vema-TRANSIT (Brandt & Wuerzberg 2014, Linse 2006, DIVA cruise reports, Brix et al. 2014, Devey & Brix 2017 cruise report, Brix & Taylor 2022, 2023, Griffiths 2012, Rhiel et al. 2018). While EBS diversity data at high taxon level were published for ANDEEP_SYSTCO II, DIVA1-3 and Vema-TRANSIT (Brandt et al 2014, Kaiser et al. in review, Brandt et al 2018), unpublished diversity data for BIOPEARL I, IceAGE1-3&RR, IceDIVA1, 2 and JR275 came from sample databases at DZMB Senckenberg and British Antarctic Survey, and are published here for the first time. During all 13 expeditions EBS with an epibenthic and a suprabenthic netsampler following the sampler sizes and height defined by Brandt and Barthel (1995) and Brenke (2005) were used, enabling comparability of samples. This type of EBS was a suitable device for sampling small benthic fauna on and above the seabed, including macrofauna and small-sized megafauna. The EBS holds an epi-and a suprabenthic netsampler (Brenke, 2005). Each of these nets has an opening of 100x33 cm and net mesh size of 500 micrometers. The cod ends are equipped with net-buckets containing 300 micrometer mesh windows. On deployment 1.5 times cable length to water depth were laid out and then EBS was trawled with 1 kn for 10 minutes on the seabed at a 1 knot for deployments in 500 m to 1500 m. Once on the deck, the content of the samplers was immediately fixed in 96 percent undenaturated and pre-cooled (at -20 degrees Celsius) ethanol. Samples were stored in a -20 degrees Celsius freezer for at least 48 h to reduce degradation of DNA for subsequent genetic studies. During this time, samples were gently rolled every three to six hours. Ethanol was changed once for all sub-fractions. In the laboratory, samples were initially sorted under a stereomicroscope to higher taxonomic ranks, lowest were class level for this analysis. Live specimen numbers were counted for abundance data and thanatocoenosis were not taken into account. For comparison between stations abundance data were standardised to 1000 m2 trawled area. The haul distances were calculated based on equation (4) in Brenke (2005). For comparison between stations abundance data were standardised to 1000 m2 trawled area. In this study we included data for 41 higher taxa of the initially separated 50 taxa ranging from phyla to orders. We excluded Foraminifera, Bryozoa, and fish. Brachiopoda, Chaetognatha, Echiura, Hemichordata, Nematoda, Nemertea, Phoronida, Platyhelminthes, Porifera and Priapulida were not identified beyond phylum level. Annelida were separated into Polychaeta, Sipunculida, Oligochaeta and Hirudinea. The phylum Arthropoda was split into the subphylum Chelicerata and Crustacea, with the former comprising Pycnogonida and Acarina and the latter crustacean order levels. Chordata only consisted of Tunicates. Echinodermata and Mollusca were separated into classes. For the DIVA1 & 2 data sets, Cnidaria and Echinodermata were not further discriminated into classes. For the DIVA-1, IceAGE 1-3 and IceDIVA1,2 data sets, Aplacophora were not separated into Caudofoveata and Solenogastres. So, if corresponding data on class assignment were available, these were reported separately, but for all univariate and multivariate analyses, classes within Aplacophora, Cnidaria and Echinodermata were grouped. The environmental parameters for this study were provided by Bio-ORACLE (http://www.bio-oracle.org/; Tyberghein et al., 2012; Assis et al., 2018). Bio-ORACLE identifies average (mean) values for different physical and chemical variables over a 14-year time period through a combination of satellite and in-situ measurements (2000 - 2014; Assis et al., 2018), at a resolution of 5 arcmin. In this study, surface and benthic (maximum depth) values were used for the following variables: salinity (PSS), silicate (mol/m3), iron (mmol/m3; mol/m3), phosphate (mmol/ m3; mol/ m3), nitrate (mmol/ m3; mol/ m3), chlorophyll-a (mg/m3; mg/cm3) and primary production (g/m3/day). Additionally, benthic data for temperature (degrees Celsius) and current velocity (m/s) as well as surface data for calcite (mol/ m3) and dissolved oxygen (mol/ m3) were included. This dataset outlines each EBS event along with temporal and spatial data. Abundances are reported raw and standardised /1000m2. PRIMER v6.0 (Clarke and Gorley, 2005) was used to perform a suite of multivariate statistical analyses: Similarity Percentage (SIMPER), Analysis of Similarities (ANOSIM), Biota-Environment Stepwise Analysis (BioEnv BEST) and Principal Components Analysis (PCA) (Hong and Zhinan, 2003). Analyses use taxonomic classification at the phyla level and at the multitaxon level (highest taxonomic classification achievable for each sample). 2 SIMPER analyses were conducted using standardised abundances to assess the contribution of taxa (at the phyla and 'multitaxon' level) to similarity analyses. A Bray-Curtis similarity matrix was used, and 90 percent was the cut off for low contributions to similarity. They used depth and region as factors. Depth split into the following factors using bins: abyssal (4000m-6000m), hadal (6000m), shelf (200m), bathyal (1000-4000m) and slope (200-1000m). Region is split into Vema Fracture Zone, Puerto Rico Trench, South West Atlantic, South East Atlantic, Southern Ocean, North West Atlantic and North West Atlantic. ANOSIM was conducted using standardised abundances from EBS sample to test whether there was a significant difference between samples based on different factors. 4 ANOSIMs were carried out: one for each depth and region (splitting samples as for SIMPER) at the phyla and 'multitaxon' level. PCA analysis is used to understand the associations between community composition and environmental factors, using standardised abundances. 2 PCA analyses were performed: one including 21 environmental variables (all factors: depth (m), latitude (decimal degrees), longitude (decimal degrees), temperature - mean at max dept (degrees Celsius), chlorophyll-a - mean at max depth (mg/m^3), dissolved oxygen concentration - mean at max depth (mmol/m^3), dissolved iron concentration- mean at max depth (mol/m^3), nitrate - mean at max depth (mmol/m^3), phosphate - mean at max depth (mol/m^3), salinity - mean at max depth (PSS), silicate - mean at max depth (mmol/m^3), current velocity - mean at max depth (m/s) primary productivity - mean at max depth, chlorophyll - mean at surface (mg/cm^3), calacite - mean at surface (mol/cm^3), dissolved iron concentration - mean at surface (mol/m^3), nitrate - mean at surface (mol/m^3), phosphate - mean at surface (mol/m^3), primary productivity - mean at surface, salinity - mean at surface (PSS), silicate - mean at surface (mol/m^3)) and one including 11 environmental variables (main factors: depth (m), latitude (decimal degrees), longitude (decimal degrees), temperature - mean at max depth (degrees Celsius), chlorophyll-a - mean at max depth (mg/m^3), dissolved oxygen concentration - mean at max depth (mmol/m^3), dissolved iron concentration- mean at max depth (mmol/m^3), nitrate - mean at max depth (mmol/m^3), phosphate - mean at max depth (mmol/m^3), salinity - mean at max depth (PSS), silicate - mean at max depth (mmol/m^3), current velocity - mean at max depth (m/s) primary productivity - mean at max depth). BioEnv analysis were conducted to test which environmental variables (up to 5 variables) best explain patterns in abundances. Spearman rank correlation method was used. Euclidean distance was used for resemblance analysis. 2 analyses were performed: one at the phyla level and one at the 'multitaxon' level. The following environmental variables were included in the analyses: depth (m), latitude (decimal degrees), longitude (decimal degrees), temperature - mean at max depth (degrees Celsius), chlorophyll-a - mean at max depth (mg/cm^3), dissolved oxygen concentration - mean at max depth (mol/m^3), dissolved iron concentration- mean at max depth (mol/m^3), nitrate - mean at max depth (mol/m^3), phosphate - mean at max depth (mol/m^3), salinity - mean at max depth (PSS), silicate - mean at max depth (mol/m^3), current velocity - mean at max depth (m/s) primary productivity - mean at max depth, chlorophyll - mean at surface (mg/cm^3), calacite - mean at surface (mol/m^3), dissolved iron concentration - mean at surface (mol/m^3), nitrate - mean at surface (mol/m^3), phosphate - mean at surface (mol/m^3), primary productivity - mean at surface, salinity - mean at surface (PSS), silicate - mean at surface (mol/m^3).

近底生物采样（Epibenthic Sampler, EBS）样品采集自以下科考航次：ANDEEP-SYSTCO II、BIOPEARL I、DIVA1-3、IceAGE1-3&RR、IceDIVA1,2、JR275及Vema-TRANSIT（参考文献：Brandt与Wuerzberg 2014、Linse 2006、DIVA航次报告、Brix et al. 2014、Devey与Brix 2017航次报告、Brix与Taylor 2022、2023、Griffiths 2012、Rhiel et al. 2018）。
此前已有ANDEEP_SYSTCO II、DIVA1-3及Vema-TRANSIT航次的高分类阶元EBS多样性数据发表（Brandt et al 2014、Kaiser et al. 待刊、Brandt et al 2018），而BIOPEARL I、IceAGE1-3&RR、IceDIVA1,2及JR275航次的未发表多样性数据则源自德国森肯伯格海洋生物多样性研究中心（DZMB Senckenberg）与英国南极调查局（British Antarctic Survey）的样品数据库，本次为首次正式发表。
本次13个航次均采用遵循Brandt与Barthel（1995）及Brenke（2005）规定的采样器尺寸与高度参数的近底与超近底联合采样网（EBS），以保障不同样品间的可比性。该类采样器适用于海床及近底层小型底栖生物的采集，涵盖大型底栖生物与中小型巨型底栖生物。EBS包含近底采样网与超近底采样网（Brenke, 2005），每张网的开口尺寸为100×33 cm，网衣孔径为500 μm，囊网则配备300 μm孔径的网桶。
采样时，缆绳放出长度为水深的1.5倍，随后在海床以1节的速度拖网10分钟；对于水深500~1500 m的站位，拖网速度保持1节。样品回收至甲板后，需立即用预冷至-20℃的96%非变性乙醇固定。样品置于-20℃冰箱保存至少48小时，以降低DNA降解程度，便于后续遗传学研究；期间每3~6小时轻轻翻动样品一次，并更换一次乙醇用于所有子组分的保存。
在实验室中，首先通过体视显微镜将样品分选至高分类阶元，本次分析的最低分类阶元为纲级。仅统计活动个体的数量以获取丰度数据，不纳入尸积群落（thanatocoenosis）数据。为便于站位间比较，丰度数据均标准化为每1000 m²拖网面积的个体数。拖网距离通过Brenke（2005）中的公式（4）计算得到。
本研究最终纳入初始分选得到的50个类群中的41个高阶类群，涵盖从门到目的分类单元。排除有孔虫门（Foraminifera）、苔藓动物门（Bryozoa）及鱼纲（Fish）。腕足动物门（Brachiopoda）、毛颚动物门（Chaetognatha）、螠虫动物门（Echiura）、半索动物门（Hemichordata）、线虫动物门（Nematoda）、纽形动物门（Nemertea）、帚虫动物门（Phoronida）、扁形动物门（Platyhelminthes）、多孔动物门（Porifera）及鳃曳动物门（Priapulida）仅鉴定至门级。环节动物门（Annelida）进一步划分为多毛纲（Polychaeta）、星虫纲（Sipunculida）、寡毛纲（Oligochaeta）及蛭纲（Hirudinea）。节肢动物门（Arthropoda）分为螯肢亚门（Chelicerata）和甲壳亚门（Crustacea），前者包含海蜘蛛纲（Pycnogonida）与蜱螨亚纲（Acarina），后者则划分至甲壳动物纲级阶元。脊索动物门（Chordata）仅包含尾索动物亚门（Tunicates）。棘皮动物门（Echinodermata）与软体动物门（Mollusca）均划分至纲级。对于DIVA1和DIVA2数据集，刺胞动物门（Cnidaria）与棘皮动物门未进一步细分至纲级；对于DIVA-1、IceAGE 1-3及IceDIVA1,2数据集，单板纲（Aplacophora）未进一步划分为沟腹纲（Caudofoveata）与无板纲（Solenogastres）。因此，若存在对应的纲级分类数据，则单独报告；但在所有单变量与多变量分析中，单板纲、刺胞动物门与棘皮动物门的类群均合并处理。
本研究使用的环境参数源自Bio-ORACLE数据库（http://www.bio-oracle.org/; Tyberghein et al., 2012; Assis et al., 2018）。该数据库通过卫星与原位观测数据的结合（2000~2014年），生成了14年时间尺度下不同物理化学变量的平均数值，空间分辨率为5弧分。本研究选取的变量包括表层与底栖（最大水深处）的盐度（PSS）、硅酸盐（mol/m³）、铁（mmol/m³; mol/m³）、磷酸盐（mmol/m³; mol/m³）、硝酸盐（mmol/m³; mol/m³）、叶绿素a（mg/m³; mg/cm³）及初级生产力（g/m³/天）；此外还纳入了底栖环境的温度（℃）与流速（m/s），以及表层环境的方解石（calcite）与溶解氧（mol/m³）数据。本数据集包含每一次EBS采样事件的时间与空间信息，丰度数据同时报告原始值与标准化至每1000 m²拖网面积的数值。
本研究使用PRIMER v6.0软件（Clarke and Gorley, 2005）开展一系列多变量统计分析，包括相似性百分比分析（Similarity Percentage, SIMPER）、相似性分析（Analysis of Similarities, ANOSIM）、生物群落-环境逐步回归分析（Biota-Environment Stepwise Analysis, BioEnv BEST）及主成分分析（Principal Components Analysis, PCA）（Hong and Zhinan, 2003）。分析分别采用门级分类单元与多分类单元级别（即每个样品可达到的最高分类阶元）的分类数据。共开展2次SIMPER分析，均使用标准化后的丰度数据，以评估类群（门级与多分类单元级别）对相似性分析的贡献度；分析采用Bray-Curtis相似性矩阵，以90%作为低贡献类群的截断值。分析以水深与区域作为分组因子：水深划分为深渊带（4000~6000 m）、海斗深渊带（>6000 m）、陆架带（<200 m）、半深海带（1000~4000 m）及陆坡带（200~1000 m）；区域划分为维马断裂带（Vema Fracture Zone）、波多黎各海沟（Puerto Rico Trench）、西南大西洋、东南大西洋、南大洋、西北大西洋（原文重复两次西北大西洋，保留原表述）。
共开展4次ANOSIM分析：分别基于门级与多分类单元级别数据，以水深与区域作为分组因子（分组规则与SIMPER分析一致），使用标准化后的EBS样品丰度数据，检验不同分组间样品是否存在显著差异。PCA分析用于揭示群落组成与环境因子间的关联，同样采用标准化后的丰度数据，共开展2次：第一次包含21个环境变量，分别为水深（m）、纬度（十进制度数）、经度（十进制度数）、最大水深处平均温度（℃）、最大水深处平均叶绿素a浓度（mg/m³）、最大水深处平均溶解氧浓度（mmol/m³）、最大水深处平均溶解铁浓度（mol/m³）、最大水深处平均硝酸盐浓度（mmol/m³）、最大水深处平均磷酸盐浓度（mol/m³）、最大水深处平均盐度（PSS）、最大水深处平均硅酸盐浓度（mmol/m³）、最大水深处平均流速（m/s）、最大水深处平均初级生产力、表层平均叶绿素浓度（mg/cm³）、表层平均方解石浓度（mol/cm³）、表层平均溶解铁浓度（mol/m³）、表层平均硝酸盐浓度（mol/m³）、表层平均磷酸盐浓度（mol/m³）、表层平均初级生产力、表层平均盐度（PSS）、表层平均硅酸盐浓度（mol/m³）；第二次包含11个环境变量，分别为水深（m）、纬度（十进制度数）、经度（十进制度数）、最大水深处平均温度（℃）、最大水深处平均叶绿素a浓度（mg/m³）、最大水深处平均溶解氧浓度（mmol/m³）、最大水深处平均溶解铁浓度（mmol/m³）、最大水深处平均硝酸盐浓度（mmol/m³）、最大水深处平均磷酸盐浓度（mmol/m³）、最大水深处平均盐度（PSS）、最大水深处平均硅酸盐浓度（mmol/m³）、最大水深处平均流速（m/s）、最大水深处平均初级生产力。
BioEnv分析用于检验哪些环境变量（最多5个）可最佳解释丰度数据的分布模式，采用斯皮尔曼秩相关分析法，以欧氏距离作为相似性度量标准。分析同样分为2次：分别基于门级分类单元与多分类单元级别数据，纳入的环境变量包括：水深（m）、纬度（十进制度数）、经度（十进制度数）、最大水深处平均温度（℃）、最大水深处平均叶绿素a浓度（mg/cm³）、最大水深处平均溶解氧浓度（mol/m³）、最大水深处平均溶解铁浓度（mol/m³）、最大水深处平均硝酸盐浓度（mol/m³）、最大水深处平均磷酸盐浓度（mol/m³）、最大水深处平均盐度（PSS）、最大水深处平均硅酸盐浓度（mol/m³）、最大水深处平均流速（m/s）、最大水深处平均初级生产力、表层平均叶绿素浓度（mg/cm³）、表层平均方解石浓度（mol/m³）、表层平均溶解铁浓度（mol/m³）、表层平均硝酸盐浓度（mol/m³）、表层平均磷酸盐浓度（mol/m³）、表层平均初级生产力、表层平均盐度（PSS）、表层平均硅酸盐浓度（mol/m³）。