Background: There is a clearly increasing trend of wildfires that become catastrophic in some countries such as the United States, Australia, Russia, Portugal, Greece, and Spain. Fuel is one of the key components that influences fire behavior and its effects. Assessing the fuel load and distribution of its components in the landscape provides effective fire management treatments in terms of fire prevention campaigns on a scientific basis. This study aims to evaluate the litter, duff, and herb fuels in highly flammable coniferous forest types in Ukrainian Polissia.

To estimate relationships between forestry variables that reflect the characteristics of the pine stand (DBH, height of the stand, age, relative density, stock of the plantation etc.) and the load of litter, duff, and herb fuels (CWD, FWD, litter, live grass, etc.), correlation analysis was used. To analyze difference between groups of sampling plots that have different forests, we use generalized linear mixed models including random effects of sampling plot type. Cluster analysis was performed using k-means partitioning method and Calinski-Harabasz criterion. To assess the significance of individual variables on which the variation of forest fuel depends, the random forest algorithm was used; for variable selection, we used two parameters: the percent increase in mean squared error and the Gini impurity index.

Results: The research revealed that in the pine forest stands, the stock of litter and duff varies from 15.5 (15 years) to 140 ton/ha (139 years). When modeling, the humidity level of the forest site (soil) significantly affects the dynamics of forest fuel accumulation. In fresh types of forest-growing conditions, the forest litter stock increases to the age of 80–90 years; then, it strongly decreases, while in wet forest types, continuous forest fuel stock accumulation is established during the entire growth period. Moreover, the results showed that the forest fuel load was influenced by the soil fertility. The stock of live and dead herbaceous fuel in fresh and wet conditions is not statistically different, and soil moisture has not had a significant impact. Fine woody debris stocks were more dependent on stand productivity and practically does not depend on the soil fertility index, site moisture content, and its age and ranged from 0.4 to 1.9 t/ha (1 h), from 0.1 to 2.2 t/ha (10 h), and from 0 to 1.6 t/ha (100 h).

Conclusions: The obtained results enabled to develop mathematical models for estimating litter and duff stocks in the Polissia forest stands based on stand characteristic and the soil humidity level. Moreover, the results will serve as basis to develop local forest fuel models as well as to determine potential fire hazards and a fire behavior modeling process in coniferous forests of that region. These models constitute the basis for the national set of fuel model development for each nature zone of Ukraine.