摘要
为掌握池塘圈养条件下大口黑鲈养殖周期的生长特征变化规律,测定体质量为(16.3±4.9)～(424.9±27.2)g生长周期内大口黑鲈的体长、全长、吻长、眼径、头长、尾柄长、头高、体高、尾柄高、体宽和体质量生长特征参数,分析其生长特征参数之间的相关性,分别建立基于支持向量回归(SVR)、径向基神经网络(RBF)和随机森林回归(RF)的体质量预测模型,将预测值与实测值拟合确定最佳模型;并运用模型拟合的方法建立各个生长特征参数的最佳生长模型.结果显示:体质量与生长特征参数均呈极显著相关性;基于支持向量回归(SVR)的体质量预测模型预测效果最佳,预测模型的决定系数R2为0.996,均方根误差为9.004,平均绝对误差为6.598;体质量与体长呈幂函数关系W=0.0127×L3.224,决定系数R2为0.977;全长、体长、吻长和头长的最佳生长模型为Logistic模型,头高、体高、眼径和体宽最佳生长模型为Von Bertalanffy模型,体质量、尾柄长和尾柄高最佳生长模型为Gompertz模型;在养殖周期内大口黑鲈肥满度在2.26%～2.93%波动.以上结果表明,可以利用生长模型和体质量预测模型预测掌握圈养条件下大口黑鲈的生长过程,并通过精准投喂达最佳养殖效果.
Abstract
To investigate the growth characteristics and patterns of largemouth bass(Micropterus salmoides)during pond cultivation,various growth parameters including length,total length,snout length,eye diameter,head length,caudal peduncle length,head height,body height,caudal peduncle height,body width,and body mass were measured from individuals ranging from(16.3±4.9)g to(424.9±27.2)g.The correlations among these growth parameters were analyzed,and the predictive mod-els for body mass were constructed using support vector regression(SVR),radial basis function neural net-work(RBF),and random forest regression(RF).The best-fit model was determined by comparing the predicted values with the actual measured values.Optimal growth models were also developed for each growth parameter using model-fitting approach.The results revealed a highly significant correlation between body mass and growth parameters.The SVR-based predictive model exhibited the highest accuracy,with a coefficient of determination(R2)of 0.996,a root mean square error(RMSE)of 9.004,and a mean abso-lute error(MAE)of 6.598.A power function relationship was observed between body mass and body length,with an equation of W=0.0127×L3.224 and a R2 of 0.977.The Logistic models were the best for to-tal length,body length,snout length,and head length.Von Bertalanffy models were the best models for head height,body height,eye diameter,and body width,while Gompertz models were most suitable for body mass,caudal peduncle length,and caudal peduncle height.The condition factor of largemouth bass fluctuated from 2.26%to 2.93%during the cultivation period.These findings suggest that growth models and body mass predictive models can be utilized to understand the growth process of largemouth bass under pond-cultured conditions.Accurate feeding based on these models can lead to optimal cultivation outcomes.
NETL
NSTL
万方数据