PREDICTION OF GONADAL DEVELOPMENT LEVELS IN THE CHINESE MITTEN CRAB (ERIOCHEIR SINENSIS) BASED ON MULTIPLE LINEAR REGRESSION

The gonadal development of the Chinese mitten crab (Eriocheir sinensis) is a key determinant of commercial quality and market grading. However, rapid and nondestructive methods for evaluating gonadal development are still lacking in aquaculture production. Current assessment methods mainly include destructive anatomical weighing, which provides high accuracy but cannot be applied to live crabs; near-infrared spectroscopy, which enables nondestructive prediction but requires expensive equipment and complex operation; and phenotypic trait-based methods, which are simple and low-cost but have mostly focused on individual traits without establishing systematic quantitative grading models. This study aimed to establish a rapid, nondestructive, and low-cost prediction model for gonadal development based on external phenotypic traits using multiple linear regression analysis. A total of 1000 adult crabs (500 females and 500 males) were collected from major aquaculture regions in eastern China, including Jiangsu, Shanghai, and Zhejiang. Morphometric traits, including body weight/carapace length³ (BW/CL³), body weight/carapace width³ (BW/CW³), body height/carapace length (H/CL), and body height/carapace width (H/CW), together with carapace coloration parameters (red, green, and blue values), were measured. Gonadosomatic index (GSI) and hepatosomatic index (HSI) were determined after dissection and used as indicators of gonadal development. Pearson correlation analysis was first employed to identify key phenotypic traits associated with gonadal development, followed by simple linear regression and multiple linear regression analyses to develop predictive models. Model performance was evaluated using the coefficient of determination (R²), root mean square error (RMSE), and an independent test dataset. The results demonstrated clear sex-specific relationships between morphometric traits and gonadal development. In females, BW/CL³ and BW/CW³ exhibited significant positive correlations with both GSI and GSI+HSI (P<0.01), whereas H/CL and H/CW showed the strongest correlations in males (P<0.01). In contrast, carapace color parameters showed weak and inconsistent relationships with gonadal development, suggesting limited predictive value. Simple linear regression models provided useful thresholds for rapid screening of gonadal development, with the highest R² among single-trait models reaching 0.761. Threshold analysis further indicated that female crabs with BW/CL³ or BW/CW³ ≥ 0.55 generally possessed significantly higher gonadal development, while male crabs with H/CL ≥ 0.55 or H/CW ≥ 0.50 exhibited significantly increased GSI. Compared with simple regression, multiple linear regression improved prediction performance by integrating multiple morphometric variables. The highest R² among the target trait models reached 0.825. Predicted values from the multiple linear regression models were generally consistent with measured values, with the majority of prediction errors falling within ±10%. Prediction accuracy was slightly lower for some male crabs. In conclusion, specific body proportion indices can serve as reliable phenotypic indicators of gonadal development in E. sinensis. Female gonadal development is best characterized by BW/CL³ and BW/CW³, whereas H/CL and H/CW are the most informative predictors for males. The multiple linear regression models developed in this study provide a simple, low-cost, and nondestructive approach for predicting gonadal development. This method has considerable potential for nondestructive assessment of gonadal maturity in adult crabs during the late aquaculture stage and may provide a practical technical reference for on-site evaluation of crab maturity and quality-based commercial grading.