Results
H1: Seasonal fuel types are reliable fuel breaks under current climatic conditions
The data underwent a log1p transformation for a multi-factor ANOVA. This was followed up with a pairwise analysis to assess the changes in fuel type observations at the different burn codes. This study focuses on seasonal fuels; despite the large error bars, an interesting trend is still visible (Figure 9). There appears to be no change in seasonal observations in the burnt and unburnt categories. The pairwise analysis also has a high p-value, indicating little difference (Table 3). These statements have little statistical power due to the high standard deviation. Still, seasonal fuels may not be contributing to fire perimeter formation in a meaningful way. These trends do not support hypothesis 1 as there is no statistical difference between unburnt and burnt observations; however, with the very high standard deviation in the dataset, there is a possibility this result is due to random chance. We can not confidently reject the null hypothesis that seasonal fuels are not reliable fuel breaks.
Figure 9. Pairwise comparison of the mean observations of Fuel Category split by Burn Code in the combined dataset. The y-axis is mean counts on a log scale. Error bars represent a 95% confidence interval.
Table 3. Pairwise comparison of Fuel Categories in the combined dataset. Estimate is the estimated distance between Burnt observations compared to unburnt. SE is the standard error. p.value is the statistical significance of each pair.
H2: Seasonal fuel types are not reliable fuel breaks in early-season wildfires
Next, the early-season dataset was analyzed. A multi-factor ANOVA was conducted following a log1p transformation and followed by a pairwise analysis. The trends visually looked similar to the combined dataset, with a slightly larger increase in mean seasonal unburnt and a slightly larger decrease in the mean unburnt conifer observations (Figure 12). Seasonal fuels again saw little change and an even higher p-value (Figure 10, Table 4). Again, large standard deviations reduce statistical confidence, but there is an apparent trend of seasonal fuels not contributing to fire cessation. We can not confidently reject the null hypothesis that seasonal fuels are not effective fuel breaks in early-season wildfires.
Figure 10. Pairwise comparison of the mean observations of Fuel Category split by Burn Code in the early dataset. The y-axis is mean counts on a log scale. Error bars represent a 95% confidence interval.
Table 4. Pairwise comparison of Fuel Categories in the early dataset. Estimate is the estimated distance between Burnt observations compared to unburnt. SE is the standard error. p.value is the statistical significance of each pair.
Additional Finding of Interest
Another trend appeared in the late dataset when analyzing the data. Similar to the other sets, no statistically significant results from the ANOVA or pairwise comparisons were found (Table 9). That said, seasonal fuels also seemed to play a reduced role in fire perimeter formation in the late data set. This shows that seasonal fuels may have played a minimal role in stopping late-season and early-season fires (Figure 11). Once again, this has low statistical confidence due to high standard errors.
Figure 11. Pairwise comparison of the mean observations of Fuel Category split by Burn Code in the late dataset. The y-axis is mean counts on a log scale. Error bars represent a 95% confidence interval.
Table 5. Pairwise comparison of Fuel Categories in the late dataset. Estimate is the estimated distance between Burnt observations compared to unburnt. SE is the standard error. p.value is the statistical significance of each pair.
Discussion
The results of this study were unable to prove that seasonal fuels are effective fuel breaks under current climatic conditions or in spring wildfires. This does not mean that seasonal fuels are no longer effective fuel breaks, but it does point to that possibility. One possible cause for seasonal fuels being less influential could be the extreme conditions in the 2023 fire season (Beverly and Schroeder 2024). The early season fires saw effects such as spring dip, a lack of canopy cover, and dead grasses, which may have allowed the fires to burn through these fuels faster than normal (Jolly et al. 2016). Late-season fires occurred during an extreme drought; drought code values would have been very high, meaning most fuels would be available to burn, potentially even those that typically would not (Beverly and Schroeder 2024). These altered climatic and seasonal conditions likely contribute to fuel type seemingly playing a lessened role in fire perimeter formation. Other factors such as weather (temperature, precipitation, relative humidity, wind) or suppression may have played larger roles in stopping these fires.
The results of this study indicate fire managers may need to take extra precautions when treating seasonal fuels as fuel breaks during early-season fires and times of extreme drought. This may require cautionary steps such as reinforcing fuel breaks with dozer guards or aerial retardant drops.
That said, standard deviations were very high due to the small sample size, and statistical confidence was greatly reduced.
Conclusions
To conclude, seasonal fuel types may have played a reduced role in perimeter formation in the 2023 fire season. This study failed to prove seasonal fuels as adequate fuel breaks in the 2023 fire season. In the future, fire managers should exercise caution when treating seasonal fuel types such as deciduous stands as fuel breaks in early-season and high drought code fires. In modern climate conditions, fire interactions with various fuel types have shifted; previously, non-volatile fuels may be more available.
Limitations and Next Steps
This study was limited to 8 fires from one fire season. Expanding the scope to more fires and years would allow the findings to be more confidently extrapolated for future fire management. The study also relied on transects, which caused sampling error; future studies could use a program such as Terra in r to analyze the full extent of the fire perimeter. This study had high standard deviations, making finding significant fuel type effects difficult. Future research should focus on a larger sample size to broadly apply findings and reduce standard error. This study could not prove that seasonal fuel contributed to perimeter formation; future studies are needed to prove that they do not statistically. Another interesting future field of research stemming from this would be the interaction of drought code and fuel type's influence on perimeter formation.