Ecological niche modelling is used in deep-sea research to investigate the environmental preferences and potential distribution of data-poor species. We present a mesoscale assessment of Mediterranean seascape suitability for the cold-water coral Lophelia pertusa (= Desmophyllum pertusum, Linnaeus, 1758). We estimated seascape suitability and uncertainty maps using an ensemble approach of three machine-learning algorithms (Generalized Boosting Model, Random Forest, Maximum Entropy) based on environmental predictors. Bathymetry, bathymetric slope and pH were the most important predictors for the models. Overall the models reached good to excellent performance, with a very reliable prediction of the most suitable areas. In the Mediterranean Sea, L. pertusa encounters environmental settings close to its physiological limits but, despite the highly variable quality of the Mediterranean seascape, we identified high suitability areas mostly along the upper slope and at submarine canyons of the Western and Central margins. The existing MPAs do not overlap with high suitability areas, and therefore L. pertusa is only protected at the deepest fringe of its potential distribution by the implementation of the bottom trawling exclusion beyond 1000 m depth. This seascape suitability assessment may assist future research, including high-resolution modelling targeting high-suitability areas, investigation on the resilience of L. pertusa populations and development of conservation actions.

生态位建模在深海研究中用于调查环境偏好和数据贫乏物种的潜在分布。我们提出了地中海景观对冷水珊瑚Lophelia pertusa（= Desmophyllum pertusum，Linnaeus，1758）的中尺度评估。我们使用了三种基于环境预测器的机器学习算法（广义提升模型，随机森林，最大熵）的集成方法来估算海景的适宜性和不确定性图。测深，测深斜率和pH是该模型最重要的预测指标。总体而言，模型达到了良好的性能，对最合适的区域进行了非常可靠的预测。在地中海，百日草遇到的环境环境接近其生理极限，但是尽管地中海海景的质量变化很大，我们还是确定了高适应性区域，主要是沿着上坡以及西部和中部边缘的海底峡谷。现有的MPA不会与高适应性区域重叠，因此，通过实施超过1000 m深度的底部拖网排除，只能在百日咳潜蝇的潜在分布的最深处对其进行保护。这项海景适宜性评估可能会有助于将来的研究，包括针对高适应性区域的高分辨率建模，百日咳沙门氏菌种群适应力的调查以及保护行动的开展。