Postfire recruitment failure in Scots pine forests of southern Siberia

Wildfire disturbances effect changes in vegetation communities that in turn influence climate. Such changes in boreal forest ecosystems can persist over decadal time scales or longer. In the ecotone between boreal forest and steppe in the region southeast of Lake Baikal in southern Siberia, shifts between the two vegetation types may be precipitated by variations in site specific conditions, as well as disturbance characteristics such as fire frequency and severity. Warmer, drier conditions in the region have been associated with a decrease in fire return intervals and greater burn severity that may, in turn, drive conversion of forests to steppe vegetation at a greater rate than has occurred prior to the onset of warming and drying. Stand-replacing fires in Pinus sylvestris stands in southern Siberia may lead to recruitment failure postfire, particularly on southwest to west-facing slopes, which are more often dominated by grasses. This study uses a combination of field data and remotely sensed indices of vegetation and moisture to distinguish between recruitment pathways in southern Siberia, and to study the influence of factors related to soils, topography, fire severity and winter snow cover on these. We expected that recruitment success would be associated with lower burn severity (higher NBR), higher greenness (NDVI) and moisture (NDMI), and winter snow (NDSI) postfire. We also expected phenological characteristics to differ among recruitment paths. Prior to burning, our sites are broadly similar in terms of remotely sensed indices of moisture (NDMI), vegetation (NDVI), and winter fractional snow cover (NDSI), but recruitment failure sites are generally drier and less green postfire. Initial differences in greenness and moisture among sites characterized by abundant recruitment (AR), intermediate recruitment (IR) and recruitment failure (RF) become more pronounced over the initial decades postfire. The earliest separability of AR and RF sites using remotely sensed indices occurs in the winter months 3–4 years postfire, during which time NDSI is highest for AR sites and lowest for RF. Although seasonality was important with regard to distinguishing among AR, IR and RF index values, the timing of phenological events such as start and end of season did not differ significantly among the sites.