Table 1.
The meanings of model parameters.
Fig 1.
Comparison of data (symbols) and best-fit predictions (curves) for the standard model which uses both TE and NTE (panels A-B) or the TE-only model which does not use NTE (panels C-D). Best-fit parameter values for both model versions are provided in Table 2. The focus here is the shape of the response to radiation dose rate. Error bars represent 95% CIs calculated using the score confidence interval method for binomial proportions [42].
Table 2.
Best-fit model parameter values and comparison of model fit qualities.
The results for four model versions (standard = TE + NTE, Eq (2); TE-only = NTE parameters b and q were set to zero; NTE-only = TE parameter a was set to zero; TDR = time-dependent dose rate, Eq 14) are arranged in columns. ΔAICc and Akaike weight are information theoretic metrics of relative model performance: the model with the highest support from the data among all tested models has the lowest ΔAICc and the highest weight. Details are discussed in the main text.
Fig 2.
Comparison of data (symbols) and best-fit predictions (curves) for the standard model.
The focus here is the time dependence of radiation effects. In this and the following figures, error bars on the data points are not shown to make visualization of the data more convenient. They are provided in Fig 1.
Fig 3.
Comparison of data (symbols) and best-fit predictions (curves) for the internal Sr model.
In this model the dose rate component from internally-incorporated 90Sr was weighted by a factor of 25. Details are described in the main text.
Fig 4.
Exploration of model parameter uncertainties by MC simulation which included random variability in dose rate estimates and aberration counts.
The bars represent best-fit parameter values for the standard (blue) and internal Sr models (red). Parameter interpretations and units are provided in Table 1. Error bars represent 95% CIs. The best-fit parameter values for the standard model were: bac = 0.0117 (95% CI: 0.0110, 0.0122), a = 0.0080 (0.0013, 0.0143) year/Gy, b = 0.242 (0.228, 0.259), q = 0.0216 (0.0193, 0.0255) Gy/year, c = 0.0506 (0.0440, 0.0570) year-1. For the internal Sr model they were: bac = 0.0117 (0.0114, 0.0130), a = 0.0183 (0, 0.0279) year/Gy, b = 0.197 (0.169, 0.252), q = 0.0234 (0.0205, 0.0551) Gy/year, c = 0.0498 (0.0415, 0.0570) year-1. Details are described in the main text.
Fig 5.
NTE contributions (Eq 11) to the radiation responses predicted by the standard and internal Sr models.
Curves represent best-fit results, and symbols show the lower and upper 95% CIs at selected dose rates. The dose rate shown on the x-axis is excess dose rate (above natural background). Details are described in the main text.
Table 3.
NTE contributions (Eq 11) to the radiation responses predicted by the standard and internal Sr models.
Excess dose rate represents radiation exposure above natural background.
Fig 6.
Application of the best-fit radiation response (red curve) produced using the standard model on chromosomal aberration data in snail embryos to a different data set: The fraction of young amoebocytes in the haemolymph of adult snails (blue symbols).
Details of this calculation are described in the main text. The NTE-only model curve produced by setting the TE parameter a to zero is shown in green.