Hong Liu and Eric S. Menges. 2005. Winter fires promote greater vital rates in the Florida keys than summer fires. Ecology 86:1483–1495.

Appendix C. Results of Kaplan-Meier survival curves analysis and logistic regression on survival using plant cohort, fire treatment, and block as predictors.

Three years survival curves analysis. Three years survival curves (with blocks pooled) were significantly different among fire treatments (Log Rank = 93.94, df = 2, P < 0.001; Fig. C1). Post hoc tests indicated that plants in both summer- and winter-burn treatments had significant lower survival than control treatment (Log Rank = 96.00, df = 1, P < 0.001; and Log Rank = 20.15, df = 1, P < 0.001, respectively). Plants in the summer-burn treatment had significant lower survival than those in winter-burn treatment (Log Rank = 27.55, df = 1, P < 0.001).

Survival year 2 post fire. Plant cohort was the only main significant predictor for survival two years post fire (Wald = 40.27, df = 1, P < 0.001; Fig. C2i). Plants that existed before fires were 3.5 times more likely to survive than plants that were first noted after fires.  Effects of fire treatment were only marginally significant (Wald = 5.38, df = 2, P = 0.068). None of the two way or three way interactions had significant effects on plant survival two years post fire. These results are consistent with analyses that did not include plant cohort as a predictor (Table B1), except that block effects were significant in the latter analysis. 

Survival year 3 post fire. Plant cohort was again a significant predictor (Wald = 72.29, df = 1, P < 0.001; Fig. C2ii). Plants that existed before fires were 4 times more likely to survive than plants that were first seen after fires. In addition, block (Wald = 6.80, df = 1, P = 0.009), fire treatment (Wald = 9.98, df = 2, P = 0.007), and the interaction between block and fire treatment (Wald = 6.91, df = 2, P = 0.032) were significant. These results and those that did not include plant cohort as a predictor (Table B1) are similar except that the interaction between block and fire treatment was marginal.

In the Orchid block, survival in summer-burn treatment was significantly higher than in the control (Wald = 8.03, df = 1, P = 0.005) and winter-burn treatment (Wald = 7.210, df = 1, P = 0.007). In Poisonwood block, the winter-burn treatment had significantly higher survival than the control treatment (Wald = 7.06, df = 1, P = 0.008).

FIG. C1. Survival curves over three years post fire for plants that existed before the burns. Different letters indicate significant differences among fire treatments.

FIG. C2. Annual survival by plant cohort (plants that existed before vs. only after the burns) and fire treatment. (i) year two post fire, (ii) year three post fire.  Different letters indicate a significant difference between plant cohorts.