A regression model to predict seed filling in ruzi grass (Brachiaria ruziziensis)

Abstract

Brachiaria ruziziensis (congo signal or ruzi grass) is an important fodder crop in India. Cultivation of ruzi grass through seed is restraint due to a higher component of chaffy seeds in a seed lot. A number of filled seeds in a lot are an important seed quality parameter in ruzi grass. Currently, X-ray radiography and manual estimation of filled seeds are means to identify filled seeds in a seed lot. X-ray radiography method is an expensive and manual estimation of a number of filled seeds is tedious and time-consuming. Hence, a regression model was developed to estimate the number of filled seeds in a seed lot based on the weight of 100 seeds. Various regression diagnostics like standard residual plot, Normal Q-Q plot, Scale–location plot and Leverage plots were used to validate the model. A third-degree polynomial with 100 seed weight as a predictor was found to be the best fit to predict the number of filled seeds in a seed lot.