Fire injury to the pine stand became evident within 

 a few weeks of the initial burns with the appearance of 

 variable crown scorch. Crown scorch is defined as the 

 percentage of live crown length that experienced foli- 

 age kill. An evaluation of this injury was made at that 

 time. In 18 units of 2.5 acres each (six for each of the 

 autumn, spring, and summer treatments), an attempt 

 was made to locate and tag two trees in each of 20 size 

 class/scorch class categories. The approximate mid- 

 points and range of the four d.b.h. size classes were 3 

 (1.5 to 4.5), 6 (4.6 to 7.5), 9 (7.6 to 10.5), and 12 (10.6 to 

 15.0) inches. The approximate midpoints and range of 

 the five crown scorch classes were 30 (20 to 40), 50 (40 

 to 60), 70 (60 to 80), 90 (80 to 99), and 100 percent. Burn 

 treatments at 2- and 4-year intervals after the initial 

 treatment caused little or no additional scorch or bole 

 damage because the only fuel consumed was the litter 

 deposited since the last burn. Therefore, virtually all 

 of the evaluated crown damage resulted from the 

 initial fire treatments. 



A total of 526 trees were tagged and checked annu- 

 ally for vitality and, following death, to see if they had 

 fallen. In the autumn treatment, 180 trees were cho- 

 sen, 162 in the spring treatment, and 184 in the 

 summer treatment. This report evaluates tree fall 

 subsequent to mortality over that decade. 



Analyses 



Cumulative tree fall percentages were calculated by 

 dividing the total number of trees that had fallen up to 

 the end of a particular year by the number of dead 

 trees. Not only did tree fall numbers increase over the 

 10-year study, but the number of dead trees also 

 increased. This represented an increasing pool from 

 which trees could fall. 



Annual fall rates were calculated by dividing the 

 number of trees that came down in a year by the 

 number of standing dead trees at the beginning of that 

 year. For each year, the number of trees that could 

 possibly fall (standing dead) was reduced by the num- 

 ber that previously fell and was increased by the 

 number that recently died. Average annual fall rates 

 for a particular seasonal treatment, scorch class, or 

 size class were calculated by averaging annual tree 

 fall rates over the length of the study. 



In the figures, mortality rates are shown along a 

 time scale that is "Year After Fire," whereas the 

 cumulative tree fall percentages are shown along a 

 more appropriate time scale, 'Tear After Death." Be- 

 cause trees fell as a consequence of mortality, the 

 'Tear After Death" scale starts all trees at a point 

 when they can begin falling. 



A logistic regression analysis was run for these 

 binomially distributed data to determine which of 



several factors were significant in explaining differ- 

 ences in total tree fall. Factors tested were season, tree 

 size, scorch class, and number of years between fire 

 injury and mortality. The Statistical Analysis System/ 

 Insight software (SAS 1993) was used to identify the 

 significance of individual factors and appropriate in- 

 teractions. Each dead tree was classified by season of 

 treatment, d.b.h. class, scorch class, years alive after 

 the fire, and a dummy dependent variable to indicate 

 a standing or fallen condition. Tree fall probabilities 

 were computed for each level of all significant factors 

 and factor interactions. A 10 percent significance level 

 was chosen. 



Results 



Harrington (1993) reported the 10-year fire mortal- 

 ity results on these research units. Because rates of 

 tree fall are dependent on mortality rates, pertinent 

 mortality results will be reviewed. 



For combined seasonal treatments, 23 percent of the 

 526 tagged trees were dead by the 10th postfire year. 

 Trees scorched in the active growing season were more 

 likely to die than those in the dormant season. The 

 spring treatment lost 28 percent and the summer 

 treatment lost 30 percent compared to a 12 percent 

 loss of the autumn-damaged trees. Even though most 

 mortality occurred during the first 3 or 4 years, addi- 

 tional trees died over the next several years (fig. la), 

 which is typical following underburning. 



Besides season, other variables that affected mor- 

 tality and could influence tree fall were tree size and 

 degree of crown scorch. A greater percentage of smaller 

 trees died than larger trees, with 40 to 50 percent 

 mortality of trees less than 7.5 inches d.b.h. compared 

 to less than 10 percent mortality for larger trees. 



Increased crown scorch did not result in higher 

 mortality until scorch exceeded 80 percent. Below this 

 threshold, mortality averaged about 10 percent, but 

 increased threefold to sevenfold above it. Again, after 

 high first-year mortality, trees continued to die at a 

 decreasing rate for several years (fig. lb). 



During the third postburn year, the first fire-killed 

 tree fell; it was in the summer treatment. In year 4, 

 eight more trees in that treatment were down. In the 

 spring treatment, nine trees were first found down in 

 year 4. No trees fell in the autumn treatment until 

 year 5 when one tree was down. Five additional trees 

 fell in year 6. Tree fall continued intermittently in all 

 treatments to the end of the study. 



Fallen trees had broken off within 1 ft of the ground 

 where decay was prevalent. 



Of the 123 tagged, dead trees, 92 (75 percent) fell 

 during the 10-year study. Fallen trees had sustained 

 an average crown scorch of 86 percent, which was 



2 



