Table 2. Linear regression models for estimating PMFI/FR-total scarred trees/plots, based on the 96-case calibration dataset.

All slopes (ß) were significant (p < 0.001) at α = 0.05.

Estimator	ß †	Outliers‡	n	R2adj	RMSE§
Mean CFI—all fires	2.440	25, 89	82	0.721	18.14
Median CFI—all fires	2.450	25, 89	74	0.675	18.52
Weibull Scale CFI—all fires	2.655	25, 93	56	0.755	19.05
Weibull Mean CFI—all fires	2.915	25, 93	56	0.762	18.63
Weibull Median CFI—all fires	3.294	25, 93	56	0.730	20.12
Mean CFI—10% scarred	2.467	25, 89	59	0.837	15.65
Median CFI—10% scarred	2.783	25, 89	60	0.812	16.34
Weibull Scale CFI—10% scarred	2.423	25, 93	55	0.856	16.09
Weibull Mean CFI—10% scarred	2.666	25, 93	55	0.865	15.39
Weibull Median CFI—10% scarred	2.992	25, 93	55	0.826	17.66
Mean CFI—25% scarred	1.715	2, 89	69	0.923	11.00
Median CFI—25% scarred	1.834	26, 89	63	0.870	13.67
Weibull Scale CFI—25% scarred	1.597	2	55	0.925	11.96
Weibull Mean CFI—25% scarred	1.749	2	55	0.929	11.36
Weibull Median CFI—25% scarred	1.867	2	55	0.906	13.00
Mean ITFI	1.121	2, 70	65	0.944	10.30
Median ITFI	1.366	24, 26	64	0.896	12.57
Weibull Scale ITFI	1.108	2	55	0.970	8.04
Weibull Mean ITFI	1.216	2	55	0.972	7.52
Weibull Median ITFI	1.361	2	55	0.958	9.46
PMFI/FR-recorders	1.337	None	52	0.961	10.39

† All models have the form: PMFI/FR-total scarred trees/plots = ß * predictor

‡ Numbers represent row numbers in the 96-case calibration dataset (S1 Table)

§ RMSE = root mean square error, the prediction error, in years, from the 10-fold cross validation