Species Distribution Modelling of Corals and Sponges from Research Vessel Survey Data in the Newfoundland and Labrador Region for Use in the Identification of Significant Benthic Areas

We used a species distribution modelling approach called random forest (RF) to predict the probability of occurrence and biomass of sponges, sea pens, and large and small gorgonian corals across the entire spatial extent of Fisheries and Oceans, Canada's (DFO) Newfoundland and Labrador Region. A suite of 66 environmental variables from different data sources were used. Models utilized catch records from the DFO multispecies trawl survey, DFO/industry northern shrimp surveys, and Spanish trawl surveys. Most models had excellent predictive capacity with cross-validated Area Under the Receiver Operating Characteristic Curve (AUC) values ranging from 0.786 to 0.926. Areas of suitable habitat were identified for each taxon and were contrasted against their known distribution. Generalized additive models (GAMs) were developed to predict the biomass distribution of each taxonomic group and serve as a comparison to the RF models. The RF and GAM models provided similar results, although GAMs provided superior predictions of biomass along the slopes of Newfoundland and Labrador for some taxonomic groups. Aside from providing continuous prediction maps of significant benthic taxa for the entire Newfoundland and Labrador Region that will be useful in ecosystem management decision-making processes, these results could be used to refine the outer boundaries of significant concentrations of these organisms identified by kernel density analyses and identify new suitable habitat not sampled by the trawl surveys.本研究采用名为随机森林（Random Forest, RF）的物种分布建模方法，对加拿大渔业与海洋部（Fisheries and Oceans Canada, DFO）所辖纽芬兰与拉布拉多海域全域内的海绵、海笔以及大小型柳珊瑚的出现概率与生物量进行预测。本研究纳入多源数据共计66项环境变量，模型训练所用的渔获记录源自加拿大渔业与海洋部多物种拖网调查、加拿大渔业与海洋部/行业北部虾类调查以及西班牙拖网调查。多数模型具备优异的预测性能，经交叉验证的受试者工作特征曲线下面积（Area Under the Receiver Operating Characteristic Curve, AUC）值介于0.786至0.926之间。研究为各分类单元划定了适宜栖息区域，并将其与已知分布范围开展对比验证。本研究同时构建了广义可加模型（Generalized Additive Models, GAMs），用于预测各分类类群的生物量分布，以作为随机森林模型的对照方案。尽管广义可加模型在纽芬兰与拉布拉多海域的部分分类类群斜坡区域的生物量预测精度更优，但随机森林与广义可加模型的整体预测结果趋于一致。本研究不仅为整个纽芬兰与拉布拉多海域生成了重要底栖类群的连续分布预测图谱，可为生态系统管理决策提供有力支撑；同时还可用于修正通过核密度分析划定的该类生物集中分布区的外边界，并识别出拖网调查未采样到的新适宜栖息海域。