Population Dynamics of Major Rice Insect Pests and Selected Natural Enemies: Multiple Regression, Lag Analysis and Temporal Validation of Light Trap Catches
K. Vasanta Bhanu *
Department of Entomology, ANGRAU- Regional Agricultural Research Station, Maruteru, West Godavari District, Andhra Pradesh, India.
*Author to whom correspondence should be addressed.
Abstract
Background: Rice insect pests and their natural enemies exhibit considerable seasonal variation influenced by meteorological conditions. Understanding the relationships between insect population dynamics, weather variables and temporal fluctuations is essential for effective pest monitoring and forecasting. Integrating light-trap observations with statistical modelling and independent temporal validation may improve the understanding of insect population dynamics and support the development of reliable pest-forewarning systems.
Aim: The study aimed to examine the temporal dynamics, seasonal patterns and weather associations of major rice insect pests and selected natural enemies using light-trap observations, and to evaluate the predictive performance of statistical models for insect catches.
Study Design: An observational time-series study was conducted using weekly light-trap catches and corresponding meteorological observations.
Place and Duration of Study: The study was carried out at the Regional Agricultural Research Station, Maruteru, Andhra Pradesh, India, over the three-year period 2012–2014.
Methodology: Weekly catches of major rice insect pests and selected natural enemies were analysed alongside maximum and minimum temperatures, rainfall, relative humidity and sunshine hours. Pearson correlation analysis assessed associations between insect catches and current and lagged weather variables. Autocorrelation and seasonal decomposition were used to characterize temporal persistence and seasonal patterns. Principal component analysis (PCA) summarized variation among current and lagged meteorological variables. Negative-binomial generalized linear models and lagged regression models were developed to identify weather-related associations and predict insect catches. Predictive performance was assessed through independent temporal validation using 2014 observations.
Results: Insect catches showed strong temporal persistence, with first-order autocorrelation coefficients ranging from 0.774 to 0.830, and distinct seasonal patterns among insect taxa. White-backed planthopper catches were positively associated with rainfall, whereas yellow stem borer catches were negatively associated with evening relative humidity and positively associated with sunshine. Five rotated principal components explained 77.43% of the variation in meteorological variables. Previous-week catches had positive coefficients in all nine lagged regression models. During independent validation, the models for yellow stem borer females and brown planthopper achieved R² values of 0.504 and 0.551, respectively, whereas four insect series showed negative validation R² values.
Conclusion: Rice insect pests and natural enemies exhibited species-specific temporal patterns and weather associations. Although the statistical models identified useful patterns, their predictive performance varied among taxa. Longer time series, explicit consideration of temporal dependence and repeated independent validation are required before these models can be reliably applied in operational pest-forewarning programmes.
Keywords: Rice, light trap catches, insect population dynamics, weather variables, principal component analysis, lagged regression