---

作 者 ：杨靖一;G.A.Wadsworth;D.J.Greenwood

期 刊 ：植物营养与肥料学报 1995年 1卷 1期 页码：71-78

关键词：肥料效应曲线；市场产量；回归方差贡献率(％v.a.f.)；线性加指数模型；

Keywords:Fertilizer response curves．Marketable yield．Percentage variance accountted for(%V.a.f.), Linear plus exponential model．,

摘 要 ：本文选取英国ADAS.1987—1989年的三种蔬菜(花茎甘蓝、大白菜、韭葱)氮肥试验为材料,分别选取总产量、干物重和市场产量为因变量,以施N量为自变量,对具有代表性的三种肥料效应函数:二次型Y=b₀+b₁X+b₂X²(M1)、线性加指数Y=a+br^x+cX(M2)、倒线性Y=dbX/(8+bX)+cX(M3)进行了系统的比较研究。研究结果指出:用三种模型所配置的60个肥料效应方程均有55个达到了F_0.1以上显著水准,其中M2具有最大的平均回归方差贡献率(93%)和最小的平均残差方差(15.9)。进一步t检验指出:用M1所预报的最高平均产量显著高于M2和M3的预报结果。但M2所计算的平均最高施肥量显著低于M1和M3的计算结果。从模型的极值点考虑,M2较适合于模拟花茎计蓝、大白菜的肥料效应曲线;M3较适合于模拟花茎甘蓝、韭葱的肥料效应的曲线;M1的通用性较强。

Abstract:This study was based on the experimntal data from ADAS．U．K．(1987-1989)．Total yield．dry wieght and marketable yield were selected as Y variables, and N fertilizer applied as X variable for each of eight experiments．Three types of response curves：quadratic Y-b₀+b₁X+b₂X² (M1),linear plus exponential Y-a+br^x+cX(M2) and inverse linear Y-abX/(a+bX)+cX(M3), have been fitted for each of the three variables in order to give a comparison as to which one is the best fit at each site．F test shown that among 60 N response equations fitted by three models．55 passed F_0.1 and above significant levels．but linear plus exponential has a maximum average percentage variance accountted for (%v.a.f.-93%) and a minimum average residual variance (15.9). The further t test shown that average maximum yield predicted by M1 was significantly higher than that predicted by M 2 & M3．But average maximum N calculated by M 2 was significantly lower than that calculated bY M 1 ＆M 3．Considering model’S maximum value．M2 is suited for calabrese and Chinese cabbage．M3 is suited for calabrese and 1eek．while M1 is suited for all three crops．

---