Ensemble modeling of black pomfret (Parastromateus niger) habitat in the Taiwan Strait based on oceanographic variables

Abstract

The location, effort, number of captures, and time of fishing were all used in this study to assess the geographic distribution of Parastromateus niger in the Taiwan Strait. Other species distribution models performed worse than generalized linear models (GLMs) based on six oceanographic parameters. The sea surface temperature (SST) was between 26.5 °C and 29.5 °C, the sea surface chlorophyll (SSC) level was between 0.3–0.44 mg/m3, the sea surface salinity (SSS) was between 33.4 °C and 34.4 °C, the mixed layer depth was between 10 °C and 14 °C, the sea surface height was between 0.57 °C and 0.77 °C, and the eddy kinetic energy (EKE) was between 0.603 °C. According to the statistical findings, SST is merely a small effect compared to SSS, SSC level, and EKE in terms of impacting species distribution. By combining four effective single-algorithm models with no obvious bias, an ensemble habitat model was created. The ranges of 117°E–119°E and 22°N–24°N have the highest annual distributions of S.CPUE and nominal CPUE.

Keywords

• Boosted regression tree
• Classification and regression tree
• Ensemble model
• Generalized additive model
• Generalized linear model
• Single algorithm