浙江塘栖枇杷黄毛虫种群数量特征及预测模型

Quantitative population characteristics and a prediction model for Melanographia flexilineata from Tangqi, Zhejiang

采用灯下诱蛾和田间幼虫系统调查相结合的方法,研究枇杷黄毛虫Melanographia flexilineata种群数量特征和消长动态,并建立其发生趋势预测模型。结果表明:2008-2013年灯下诱蛾越冬代成虫始见期在4月中旬至下旬(4月14日至29日),其中2010-2012年相对较迟,峰期持续时间、蛾量受当年气候等条件的影响呈相应的变化,2008年和2011年为最高峰期虫量,2013年明显低于历年同期;田间幼虫发生量也呈现出相似的趋势,年度间均以第2代幼虫危害最重,峰期主要出现在6月下旬,少数年份推迟至7月上旬,第4代灯下成虫与田间幼虫量均下降较快。在此基础上,以枇杷黄毛虫的田间系统监测资料及气温、相对湿度、降水量等气象因素作为预测因子,采用逐步回归分析法,筛选出了具有显著回归影响的24个因子进入回归模型,建立了第1代至第3代枇杷黄毛虫发生期和发生量预测模型,其中影响枇杷黄毛虫种群数量消长的关键因子为种群基数和降水量、气温和相对湿度。经检验,各代次发生期、发生量预测模型均达到99分以上的历史符合率和预测准确度,模型拟合值与实测值相符,能准确地预测出其发生量和发生高峰期。

To improve the forecasting and sustained prevention level of Melanographia flexilineata, population dynamics and quantitative characteristics were observed, then a forecast model was established by method of lamplight moth-trapping and a systematic larvae investigation from 2008-2013. Then, based on monitoring data including insect population and weather factors, such as temperature, humidity, and precipitation, a mathematical prediction model for quantity of occurrence and period of the 1st to 3rd generation of M. flexilineata was established using stepwise regression. Results by lamplight moth-trapping showed that the start for overwintering adult trappings was from middle to late April(from April 14 th to 29th), but it was relatively late from2010-2012 in contrast to other years. The peak period and quantity were affected by environmental factors such as climate, which led to peak quantities in 2008 and 2011 but a low ebb in 2013. Population dynamics of the larvae were similar trends with the 2nd generation causing the extremely serious harm; larval peak period occurred in late June but seldom in early July. The fourth-generation quantity of moth and larvae decreased rapidly. Based on the monitoring data collected from 2008 to 2013 including insect quantity and weather factors, the prediction mathematical models of occurrence quantity and period for the pest were established using the method of stepwise regression. Twenty-four factors which had a significant effect(P<0.05 or 0.01) on the regression forecasting model were screened out. Among them, population cardinal number, precipitation, humidity, and temperature were the key factors influencing pest population dynamics. Predicted values agreed well with measured values by inspection and application with the scores which could reflect the historical coincident rate and prediction accuracy of these models being over 99. Thus, this model could accurately predict the quantity and period of occurrence for M. flexilineata. The established prediction technique in this paper was signality to the accurate prediction and scientific prevention of M. flexilineata, which provided scientific basis for supervisors in production.