Geostatistical modelling of spatial distribution of Balaenoptera physalus in the Northwestern Mediterranean Sea from sparse count data and heterogeneous observation efforts
Geostatistical modelling of spatial distribution of Balaenoptera physalus in the Northwestern Mediterranean Sea from sparse count data and heterogeneous observation efforts
复制标题
DOI：
10.1016/j.ecolmodel.2005.08.042
复制
发表时间：
2006-03-15
影响因子：
3.1
通讯作者：
Guinet, C
中科院分区：
文献类型：
--
作者：
Monestiez, P;Dubroca, L;Guinet, C
由于空间上的观测工作不均匀，而且很少看到动物，因此可能难以实现获得相对丰度的准确地图这一目标。然而，描述长须鲸等野生动物的空间分布是保护这些种群和研究它们与环境相互作用的主要优先事项。我们已经将地质统计模型与泊松分布相关联，以模拟空间变化和离散观测过程。假设几个弱的假设丰度的分布，我们已经改进了实验变异函数估计使用的权重，来自预期的方差，并提出了一个偏差校正，占泊松观测过程中增加的变异性。同样，克里格系统经过修改，可以直接内插理论基础动物丰度，比计数数据的噪音观察结果更好。对于1992-2001年夏季长须鲸的累积计数数据，该方法给出了一个相对丰度的地图，这是信息的空间格局。克里格插值方差显着降低-从0.015到0.26 -相比，通常的普通克里格的原始数据。添加随时间平稳性的假设，累积数据估计的变差函数可以用于更稀疏的年度数据。(C)2005 Elsevier B. V.保留所有权利。
Obtaining accurate maps of relative abundance is an objective that may be difficult to achieve on the basis of spatially heterogeneous observation efforts and infrequent and sparse animal sightings. However, characterizing spatial distribution of wild animals such as fin whales is a major priority to protect these populations and to study their interactions with their environment. We have associated a geostatistical model with the Poisson distribution to model both spatial variation and discrete observation process. Assuming few weak hypotheses on the distribution of abundance, we have improved the experimental variogram estimate using weights that are derived from expected variances and proposed a bias correction that accounts for the variability added by the Poisson observation process. In the same way the kriging system was modified to interpolate directly the theoretical underlying animal abundance better than noisy observations from count data. For cumulative count data of fin whales over the summers 1992-2001, the method gave a map of the relative abundance which is informative on the spatial patterns. Kriging interpolation variances were dramatically reduced - ratio from 0.015 to 0.26 - compared to usual Ordinary Kriging on raw data. Adding the hypothesis of stationarity over time the variogram estimated on cumulative data can be then used with more sparser annual data. (C) 2005 Elsevier B.V. All rights reserved.