Akademik Veri Yönetim Sistemi
2nd International Conference on Engineering, Natural and Social Sciences, Konya, Türkiye, 4 - 06 Nisan 2023, ss.129
Seed germination is critical for seedling establishment and subsequent plant health and vigor.
Rapeseed germination depends on numerous environmental and stress factors. The seedling vigor index
(SVI) can be used to compare the negative effects of various treatments and stress factors on seed
germination. Therefore, this study proposed a mathematical model to estimate SVI values to quantitatively
explain the relative effects of different salt and ascorbic acid concentrations on seedling germination and
growth. To this end, two AI-based models were compared to predict the effects of five salinity levels (0.20,
5.0, 10.0, 15.0 and 20.0 dS m-1 NaCl) and four priming doses (0, 0.5, 1.0 and 2.0 mM ascorbic acid (AsA))
on rapeseed germination. These models were artificial neural networks (ANN) and support vector machines
(SVM). All the developed models were evaluated according to the following performance indices:
coefficient of determination (R2), root mean square error (RMSE) and mean absolute error (MAE). The
results showed that all the implemented models had high performance in modelling SVI. However, the best
results were obtained in the single layer neural network (R2=0.996, MAE=23.89, and RMSE=30.76, testing
subset) using 3 neurons in the hidden layer. The SVI prediction model developed in this study can be
successfully used to predict the SVI value when salinity stress and ascorbic acid priming doses are within
the concentrations (doses investigated in this study) at which rapeseed seeds can germinate. Thus, SVI
values for rapeseed can be predicted in advance by salinity and AsA concentrations without germination
tests.