Anthropogenic disturbance and the formation of oak savanna in central Kentucky, USA

Aim To deepen understanding of the factors that influenced the formation of oak savanna in central Kentucky, USA. Particular attention was focused on the link between historical disturbance and the formation of savanna ecosystem structure. Location Central Kentucky, USA. Methods We used dendrochronological analysis of tree‐ring samples to understand the historical growth environment of remnant savanna stems. We used release detection and branch‐establishment dates to evaluate changes in tree growth and the establishment of savanna physiognomy. We contrasted our growth chronology with reference chronologies for regional tree growth, climate and human population dynamics. Results Trees growing in Kentucky Inner Bluegrass Region (IBR) savanna remnants exhibited a period of suppression, extending from the establishment date of the tree to release events that occurred c. 1800. This release resulted in a tripling of the annual radial growth rate from levels typical of oaks suppressed under a forest canopy (< 1 mm year^?1) to levels typical of open‐grown stems (3 mm year^?1). The growth releases in savanna trees coincided with low branch establishment. Over the release period, climatic conditions remained relatively constant and growth in regional forest trees was even; however, the growth increase in savanna stems was strongly correlated with a marked increase in Euro‐American population density in the region. Main conclusions Our data suggest that trees growing in savanna remnants originated in the understorey of a closed canopy forest. We hypothesize that Euro‐American land clearing to create pasturelands released these trees from light competition and resulted in the savanna physiognomy that is apparent in remnant stands in the IBR. Although our data suggest that savanna trees originated in a forest understorey, this system structure itself may have been a result of an unprecedented lack of Native American activity in the region due to population loss associated with pandemics brought to North America by Euro‐Americans. We present a hypothetical model that links human population dynamics, land‐use activities and ecosystem structure. Our model focuses on the following three land‐use eras: Native American habitation/utilization; land abandonment; and Euro‐American land clearance. Ecological understanding of historical dynamics in other ecosystems of eastern North America may be enhanced through recognition of these eras.