The effect of weighting predictor variables on the performance of habitat suitability index (HSI) models (case study: Caspian Kutum, Rutilus frisii)

Moëzzi, Fateh; Eagderi, Soheil

Volume 14, Issue 4 (winter 2025) J. Aqua. Eco 2025, 14(4): 12-24 | Back to browse issues page

Mendeley

Zotero

RefWorks

Moëzzi F, Eagderi S. The effect of weighting predictor variables on the performance of habitat suitability index (HSI) models (case study: Caspian Kutum, Rutilus frisii). J. Aqua. Eco 2025; 14 (4) :12-24
URL: http://jae.hormozgan.ac.ir/article-1-1123-en.html

The effect of weighting predictor variables on the performance of habitat suitability index (HSI) models (case study: Caspian Kutum, Rutilus frisii)

Fateh Moëzzi

, Soheil Eagderi ^*

Abstract: (344 Views)

This study investigates the effect of weighting predictors on the performance of habitat suitability index (HSI) models using arithmetic mean (AMM) and geometric mean (GMM) methods. The case study focused on the catch data of Caspian Kutum (Rutilus frisii), with habitat parameters including day-time sea surface temperature, near-surface chlorophyll-a concentration, particulate organic carbon concentration, distance from fishing points to the river mouth, and bottom slope. Relative weights of environmental variables and suitability index (SI) fitting were determined using the support vector machine (SVM) technique. Sea surface temperature and particulate organic carbon concentration were identified as the most influential variables, with weights of 0.315 and 0.231, respectively, in explaining fish catch variance. Model performance evaluation revealed that the HSIWGMM model outperformed others, showing the lowest RMSE (training: 0.1818, testing: 0.2540) and the highest correlation coefficient (training: 0.4693, testing: 0.1953). In contrast, the HSIWAMM model showed the weakest performance (training: RMSE = 0.4023, r = 0.3843; testing: RMSE = 0.3858, r = 0.1360).

Keywords: Caspian Kutum, Habitat suitability index (HSI), Modeling, Support vector machine (SVM), Weighting

Full-Text [PDF 1744 kb] (154 Downloads)

Type of Study: Research | Subject: Special
Published: 2025/02/17