Data from: Mapping coral and sponge habitats on a shelf-depth environment using multibeam sonar and ROV video observations: Learmonth Bank, northern British Columbia, Canada

Efforts to locate and map deep-water coral and sponge habitats are essential for the effective management and conservation of these vulnerable marine ecosystems. Here we test the applicability of a simple multibeam sonar classification method developed for fjord environments to map the distribution of shelf-depth substrates and gorgonian coral- and sponge-dominated biotopes. The studied area is a shelf-depth feature Learmonth Bank, northern British Columbia, Canada and the method was applied aiming to map primarily non-reef forming coral and sponge biotopes. Aside from producing high-resolution maps (5 m2 raster grid), biotope-substrate associations were also investigated. A multibeam sonar survey yielded bathymetry, acoustic backscatter strength and slope. From benthic video transects recorded by remotely operated vehicles (ROVs) six primary substrate types and twelve biotope categories were identified, defined by the primary sediment and dominant biological structure, respectively. Substrate and biotope maps were produced using a supervised classification mostly based on the inter-quartile range of the acoustic variables for each substrate type and biotope. Twenty-five percent of the video observations were randomly reserved for testing the classification accuracy. The dominant biotope-defining corals were red tree coral Primnoa pacifica and small styasterids, of which Stylaster parageus was common. Demosponges and hexactinellid sponges were frequently observed but no sponge reefs were observed. The substrate classification readily distinguished fine sediment, Sand and Bedrock from the other substrate types, but had greater difficulty distinguishing Bedrock from Boulders and Cobble. The biotope classification accurately identified Gardens (dense aggregations of sponges and corals) and Primnoa-dominated biotopes (67% accuracy), but most other biotopes had lower accuracies. There was a significant correspondence between Learmonth's biotopes and substrate types, with many biotopes strongly associated with only a single substrate type. This strong correspondence allowed substrate types to function as a surrogate for helping to map biotope distributions. Our results add new information on the distribution of corals and sponges at Learmonth Bank, which can be used to guide management at this location.

定位并绘制深水珊瑚与海绵生境的分布图，对于高效管理与保护这些脆弱的海洋生态系统至关重要。本研究旨在检验一种专为峡湾环境开发的简易多波束声呐（multibeam sonar）分类方法的适用性，该方法可用于绘制陆架深度底质与柳珊瑚（gorgonian coral）、海绵主导的生物群落生境（biotope）的分布情况。本研究区域为加拿大不列颠哥伦比亚省北部的利尔蒙特浅滩（Learmonth Bank），该区域属于陆架深度地貌，本研究应用上述方法，主要旨在绘制非造礁珊瑚与海绵主导的生物群落生境分布图。除了生成高分辨率的5平方米栅格网格地图外，本研究还对生物群落生境-底质关联模式展开了调查。本次多波束声呐调查获取了测深数据（bathymetry）、声学背散射强度（acoustic backscatter strength）与坡度。由遥控水下机器人（Remotely Operated Vehicles, ROVs）录制的底栖视频样带（benthic video transects）中，研究人员识别出6种主要底质类型与12类生物群落生境，二者分别以优势沉积物与主导生物结构作为定义依据。底质与生物群落生境地图通过监督分类（supervised classification）方法生成，该方法主要基于各类底质类型与生物群落生境对应的声学变量的四分位距（inter-quartile range）。研究人员随机预留25%的视频观测数据，用于验证分类精度。主导生物群落生境的珊瑚主要为红色树状珊瑚（Primnoa pacifica）与小型柱星螅类（styasterids），其中副美柱星螅（Stylaster parageus）较为常见。研究中频繁观测到寻常海绵（Demosponges）与六放海绵（hexactinellid sponges），但未发现海绵礁。底质分类可轻松区分细沉积物（fine sediment）、砂（Sand）与基岩（Bedrock）与其他底质类型，但在区分基岩与漂砾（Boulders）、砾石（Cobble）时难度较大。生物群落生境分类可准确识别‘花园生境’（海绵与珊瑚的密集聚集区）与以红色树状珊瑚为主导的生物群落生境（分类精度达67%），但其余多数生物群落生境的分类精度较低。利尔蒙特浅滩的生物群落生境与底质类型之间存在显著对应关系，多数生物群落生境仅与单一底质类型紧密关联。这种强对应关系使得底质类型可作为替代指标，辅助绘制生物群落生境的分布地图。本研究的结果为利尔蒙特浅滩的珊瑚与海绵分布提供了新的数据支撑，可用于指导该区域的海洋管理工作。