Spatial prediction of demersal fish distributions: enhancing our understanding of species–environment relationships
Spatial prediction of demersal fish distributions: enhancing our understanding of species–environment relationships
复制标题
DOI：
10.1093/icesjms/fsp205
复制
发表时间：
2009-10
影响因子：
3.3
通讯作者：
Cordelia Moore;E. Harvey;K. V. Niel
中科院分区：
文献类型：
--
作者：
Cordelia Moore;E. Harvey;K. V. Niel
Moore, C. H.、Harvey, E. S. 和 Van Niel, K. P. 2009。底层鱼类分布的空间预测：增强我们对物种与环境关系的理解。 - ICES 海洋科学杂志，66：2068-2075。我们使用物种分布模型来识别影响底层鱼类空间分布的关键环境变量，并评估这些物种与环境关系的潜力，以准确预测鱼类分布。过去，鱼类分布的预测模型受到限制，因为无法获得更深水域（>10 m）的详细栖息地地图。然而，最近在利用水声勘测绘制更深海洋环境地图方面取得的进展弥补了这一限制。在澳大利亚东南部的豪角海洋国家公园，之前基于水声和拖曳视频数据模拟的底栖栖息地被用来调查底层鱼类的空间生态。为了确定环境变量对这一重要海洋鱼类分布的影响，我们为四种底层鱼类开发了分类树（CT）和广义加性模型（GAM）。两种方法之间观察到了对比优势。 CT 为四个物种中的三个提供了更好的解释变异，并揭示了更好地模拟具有复杂环境相互作用的物种分布的能力。然而，GAM 对这四个物种中的三个物种的预测准确性更高。这两种建模技术都提供了对底层鱼类分布和景观联系的详细了解，以及预测可获得连续空间底栖数据的未采样位置的物种分布的准确方法。此类信息将有助于更有针对性的渔业管理以及更有效的海洋保护区规划和监测。
Moore, C. H., Harvey, E. S., and Van Niel, K. P. 2009. Spatial prediction of demersal fish distributions: enhancing our understanding of species-environment relationships. - ICES Journal of Marine Science, 66: 2068-2075.We used species distribution modelling to identify key environmental variables influencing the spatial distribution of demersal fish and to assess the potential of these species-environment relationships to predict fish distributions accurately. In the past, predictive modelling of fish distributions has been limited, because detailed habitat maps of deeper water (>10 m) have not been available. However, recent advances in mapping deeper marine environments using hydroacoustic surveys have redressed this limitation. At Cape Howe Marine National Park in southeastern Australia, previously modelled benthic habitats based on hydroacoustic and towed video data were used to investigate the spatial ecology of demersal fish. To establish the influence of environmental variables on the distributions of this important group of marine fish, classification trees (CTs) and generalized additive models (GAMs) were developed for four demersal fish species. Contrasting advantages were observed between the two approaches. CTs provided greater explained variation for three of the four species and revealed a better ability to model species distributions with complex environmental interactions. However, the predictive accuracy of the GAMs was greater for three of the four species. Both these modelling techniques provided a detailed understanding of demersal fish distributions and landscape linkages and an accurate method for predicting species distributions across unsampled locations where continuous spatial benthic data are available. Information of this nature will permit more-targeted fisheries management and more-effective planning and monitoring of marine protected areas.