基于BRT模型的北太平洋长鳍金枪鱼栖息水层分析

Habitat depths of North Pacific Thunnus alalunga based on the BRT model

  • 摘要:
    目的 揭示北太平洋长鳍金枪鱼栖息水层差异及其与环境因子的响应规律。
    方法 整合文献和2023年海上试验获得的渔获深度数据,利用Gamma分布拟合长鳍金枪鱼栖息水层概率分布,通过1964—2019年历史CPUE数据来表征北太平洋区域的种群强度,联合哥白尼海洋环境监测服务(CMEMS)的海洋环境数据(经度、纬度、水温、盐度、溶解氧含量、叶绿素a)和加权重采样方法生成与环境网格匹配的长鳍金枪鱼栖息深度样本,构建梯度提升回归树(BRT)模型,定量分析长鳍金枪鱼栖息水层与环境因子的响应关系,以及预测不同纬度(0°、10°、20°、30°N)剖面长鳍金枪鱼的栖息水层。
    结果 Gamma分布拟合结果显示,长鳍金枪鱼栖息水层可分为浅层型与深层型两类,其中浅层型平均栖息深度为88.3~117.9 m,深层型则为193.5~259.9 m。BRT模型结果表明纬度为最主要影响因子(重要性约0.47),其次为溶解氧含量(0.15)、经度(0.12)、水温(0.10)、盐度(0.10)和叶绿素a(0.09)。偏依赖关系表明水温与溶解氧含量对长鳍金枪鱼栖息水层具有显著的非线性影响,且随着叶绿素a升高,长鳍金枪鱼栖息水层深度逐渐变浅。纬度剖面下的预测结果发现0°N附近长鳍金枪鱼栖息深度整体较深且相对稳定(约200 m),30°N高纬海域则明显变浅(小于150 m)。
    结论 北太平洋长鳍金枪鱼栖息深度存在明显的纬度梯度差异,主要受水温、溶解氧含量、盐度、叶绿素a等环境因子共同驱动。以Gamma分布概率模型和CPUE加权重采样为基础的BRT建模方法可在渔获数据相对有限的条件下,有效揭示长鳍金枪鱼栖息深度分布与环境因子的响应特征,为延绳钓捕捞效率提升与远洋渔业资源可持续利用提供方法支撑。

     

    Abstract: This study aims to reveal the differences in habitat depth layers of North Pacific albacore tuna and their response patterns to environmental factors. By integrating published literature and catch depth data obtained from offshore experiments conducted in 2023, a Gamma distribution was used to fit the probability distribution of albacore tuna habitat depths. Historical CPUE data from 1964 to 2019 were employed to represent population intensity in the North Pacific region. Combined with marine environmental data from the Copernicus Marine Environment Monitoring Service (CMEMS), including longitude, latitude, temperature, salinity, dissolved oxygen, and chlorophyll-a concentration, and using a weighted resampling approach, habitat depth samples of albacore tuna matched to environmental grids were generated. A boosted regression tree (BRT) model was then constructed to quantitatively analyze the response relationships between habitat depth layers and environmental factors, and to predict habitat depths of North Pacific albacore tuna along latitudinal transects at 0°, 10°, 20°, and 30° N. The Gamma distribution fitting results indicate that the habitat depths of albacore tuna can be classified into two types: a shallow-layer type and a deep-layer type. The mean habitat depth of the shallow type ranged from 88.3 to 117.9 m, whereas that of the deep type ranged from 193.5 to 259.9 m. The BRT model results show that latitude was the most influential factor (relative importance ≈ 0.47), followed by dissolved oxygen (0.15), longitude (0.12), temperature (0.10), salinity (0.10), and chlorophyll-a (0.09). Partial dependence analyses demonstrate that temperature and dissolved oxygen exert significant nonlinear effects on albacore tuna habitat depths, and that habitat depth gradually becomes shallower with increasing chlorophyll-a concentration. Predictions along latitudinal transects reveal that habitat depths near 0° N are generally deeper and relatively stable (approximately 200 m), whereas they become markedly shallower at higher latitudes around 30° N (less than 150 m). Pronounced latitudinal gradients exist in the habitat depths of North Pacific albacore tuna, primarily driven by the combined effects of environmental factors such as temperature, dissolved oxygen, salinity, and chlorophyll-a. The BRT modeling framework based on a Gamma probability distribution and CPUE-weighted resampling effectively characterizes the relationships between albacore tuna habitat depth distributions and environmental variables under conditions of limited catch data, providing methodological support for improving longline fishing efficiency and promoting the sustainable utilization of distant-water fishery resources.

     

/

返回文章
返回