Scholarlink

Context <p>Habitat suitability models (HSMs) are widely used to predict the distribution of optimal habitat, yet in marine systems, their performance is often constrained by the coarse resolution of ocean-climate predictors and limited representation of fine-scale seabed heterogeneity arising from terrain and sediment variation. Hierarchical model frameworks may help address this limitation by incorporating multiscale environmental data at the spatial scales that influence the distribution of benthic habitats.</p> Objectives <p>We evaluated whether a hierarchical habitat suitability model (H-HSM) improves spatial predictions of snow crab (<i>Chionoecetes opilio</i>, O. Fabricius 1788) habitat relative to single-scale models. Specifically, we tested whether the H-HSM enhances understanding of regional and landscape-scale processes shaping habitat suitability across the distribution range of the Atlantic Canadian population and whether these processes vary among sex and size classes.</p> Methods <p>A two-level H-HSM was developed to represent regional-scale ocean-climate drivers (e.g. temperature, salinity, mixed-layer depth, bottom stress) and landscape-scale seabed terrain and sediment drivers (e.g. depth, slope, rugosity, sediment mobility). Separate Random Forest models with spatial cross-validation were trained for female, small male, and large male snow crab. Model performance was evaluated using&#xa0;the area under the receiver operating characteristic curve (ROC-AUC), Cohen’s Kappa, and the true skill statistic (TSS), and compared with single-scale models fitted using all environmental predictors jointly at one scale.</p> Results <p>The H-HSM outperformed single-scale models, with the landscape level achieving the highest predictive accuracy across sex and size classes (ROC-AUC = 0.92–0.94). Regional predictors, particularly bottom temperature, defined the broad environmental envelope constraining snow crab habitat, while landscape predictors refined local suitability patterns within this broader climatic structure. The H-HSM also produced more spatially refined sex and size-specific habitat suitability patterns, suggesting that behavioural and ontogenetic shifts within the population contributed to spatial habitat partitioning.</p> Conclusions <p>The H-HSM provides an ecologically coherent, scale-aware framework for modelling benthic habitat distribution. By separating broad-scale ocean-climate controls from finer-scale seabed structure, it produced ecologically interpretable predictions of snow crab habitat and provides a useful basis for broad-scale habitat assessment, including benthic monitoring and climate-habitat vulnerability assessments.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Hierarchical habitat suitability models highlight the influence of benthic seascapes on snow crab distribution

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年6月15日

查看全文
  • Emmeline Broad,
  • Craig J. Brown,
  • Nancy Shackell,
  • Katleen Robert

摘要

Context

Habitat suitability models (HSMs) are widely used to predict the distribution of optimal habitat, yet in marine systems, their performance is often constrained by the coarse resolution of ocean-climate predictors and limited representation of fine-scale seabed heterogeneity arising from terrain and sediment variation. Hierarchical model frameworks may help address this limitation by incorporating multiscale environmental data at the spatial scales that influence the distribution of benthic habitats.

Objectives

We evaluated whether a hierarchical habitat suitability model (H-HSM) improves spatial predictions of snow crab (Chionoecetes opilio, O. Fabricius 1788) habitat relative to single-scale models. Specifically, we tested whether the H-HSM enhances understanding of regional and landscape-scale processes shaping habitat suitability across the distribution range of the Atlantic Canadian population and whether these processes vary among sex and size classes.

Methods

A two-level H-HSM was developed to represent regional-scale ocean-climate drivers (e.g. temperature, salinity, mixed-layer depth, bottom stress) and landscape-scale seabed terrain and sediment drivers (e.g. depth, slope, rugosity, sediment mobility). Separate Random Forest models with spatial cross-validation were trained for female, small male, and large male snow crab. Model performance was evaluated using the area under the receiver operating characteristic curve (ROC-AUC), Cohen’s Kappa, and the true skill statistic (TSS), and compared with single-scale models fitted using all environmental predictors jointly at one scale.

Results

The H-HSM outperformed single-scale models, with the landscape level achieving the highest predictive accuracy across sex and size classes (ROC-AUC = 0.92–0.94). Regional predictors, particularly bottom temperature, defined the broad environmental envelope constraining snow crab habitat, while landscape predictors refined local suitability patterns within this broader climatic structure. The H-HSM also produced more spatially refined sex and size-specific habitat suitability patterns, suggesting that behavioural and ontogenetic shifts within the population contributed to spatial habitat partitioning.

Conclusions

The H-HSM provides an ecologically coherent, scale-aware framework for modelling benthic habitat distribution. By separating broad-scale ocean-climate controls from finer-scale seabed structure, it produced ecologically interpretable predictions of snow crab habitat and provides a useful basis for broad-scale habitat assessment, including benthic monitoring and climate-habitat vulnerability assessments.