Quantification of safflower (Carthamus tinctorius) seed germination response to water potential and priming: hydro time models on the basis of normal, weibull and gumbel distributions

Introduction Water potential osmotic is one primary environmental regulators of seed germination. Quantification of germination response to water potential is possible using hydro time model, also low seed germination and seedling emergence is one of the main problems in dry areas. In this study, application of hydrothermal time models on the basis of normal, weibull and gumbel distributions for quantification of Carthamus tinctorius germination response to water potential. Material and methods Treatments included drought levels (0, -0.2, -0.4, -0.6 and -0.8 MPa) and priming (Priming with GA 50 ppm for 15 h at 20 °C and control seeds). Three replicates of 25 seeds were used for each temperature. A seed was considered germinated when its radicle protruded through the seed coat at least 2 mm. Cumulative germination response of seeds to differential water potential were quantified using three-parameter sigmoidal model, then, germination response of seeds was quantified using hydro time model (normal, weibull and gumbel distributions). The akaike information criterion (AICc), root mean square of errors (RMSE) and coefficient of determination (R2) and relationship between the observed and the predicted germination percentage and base water potential were used to select the superior model from among the employed methods. Results Germination percentage and time to 50% maximum seed germination of Carthamus tinctorius were calculated for the different treatments seed (Priming with GA 50 ppm for 15 h at 20 °C and control seeds) by fitting 3-parameter sigmoidal functions to cumulative germination data. Results indicated that germination percentage and germination rate reduced as a result of water potential increment but increased with priming. The highest germination percentage was obtained from control and -0.2 MPa osmotic potential and priming seed (97.32 and 97.30 %). The minimum time to 50% maximum seed germination was obtained from control osmotic potential and priming seed (0.66 seed per day). Results indicated that normal (Akaike information criterion for control and priming seed was -240.76 and -241.50 respectively and root mean square of errors for control and priming seed was 0.114 and 0.111 respectively) and weibull (Akaike information criterion for control and priming seed was -232.34 and -240.53 respectively and root mean square of errors for control and priming seed was 0.113 and 0.110 respectively) hydrothermal time models more accurately predicted germination than gumbel (Akaike information criterion for control and priming seed was -254.10 and -247.40 respectively and root mean square of errors for control and priming seed was 0.121 and 0.118 respectively) hydrothermal time model. According to the hydro time models, the hydro time constant (θH) estimated with normal, weibull and gumbel distributions was 1.11, 1.01 and 1.11 respectively for control seed and 0.92, 0.91 and 0.94 MPa d-1 for priming seed. Base water potential estimated with normal (ψb(50)), weibull (median) and gumbel (median) distributions was -0.79, -0.93 and -0.86 MPa respectively for control seed and -0.79, -0.93 and -0.86 MPa for priming seed. The shape parameter (λ) of the Weibull hydrothermal time model for control and priming seed was 1.65 and 1.45 respectively, which implied asymmetry of base water potential data and skewness of distribution to the right. Based on the Weibull hydrothermal time model, water potential threshold for the onset of germination (location parameter of weibull hydro time model (µ) or ψb(0)) for control and priming seed was equal to -1.71 and -1.95 MPa. Conclusion Using hydro time model in order to quantify the germination response of Carthamus tinctorius seeds to different water potential and priming resulted in satisfactory outcomes. Due to the flexibility of the Weibull distribution, this model provides a useful method for predicting germination and weibull distribution may be more suitable than the normal distribution for seed germination modeling.