应用多元回归估测芒果产量性状的研究

Study on Estimating Yield Properties of Sannianmang by Multiple Regression

应用多元回归分析研究了三年芒１２个生物学性状与座果率之间的关系。结果表明（１）与座果率关系最密切的五个性状依次为：座果数，一次梢粗呈显著正相关。总花序，二次梢数量，一次梢长呈显著的负相关。（２）树冠体积等对座果率的作用较小。（３）为了便于生产上的应用，采用逐步回归建立了估测座果率的最优回归方程，Ｒ＝０．９８４６，Ｐ＜０．０１，比用１２元回归的决定系数Ｒ￣２提高了０．８％左右。

Relationship between fruit setting rate and twele sorts of sannianmang’scharacteristics was studied by multiple regression.Results indicated that it was positiverelationship between fruit setting rate and amount of fruit,thick of first branches,andnegative relationship between fruit setting rate and amount of anthotaxy,amount of sec-ond branches,length of first branches;volume of tree crown had little effect on fruit set-ting rate.Optimisation regression equation was established by stepwise regression to es-timate fruit setting rate,with r=0.9846,P<0.01,which determination coefficient was0.8% higher than by linear muItiple regession.