The importance of history in understanding large tree mortality in African savannas
|Title||The importance of history in understanding large tree mortality in African savannas|
|Publication Type||Journal Article|
|Year of Publication||2021|
|Authors||Das A, Thaker M, Coetsee C, Slotow R, Vanak AT|
Fire and elephant herbivory are major drivers of large tree mortality in savanna ecosystems. Although the spatial variation of these agents is well studied, less attention has been paid to how disturbance history influences tree mortality over time. In a long-term cohort study, we examined how the sequence of fire- and elephant-induced damage influences mortality of trees and determined whether risk of mortality is compounded over time. Data on over 2500 large trees were collected from 22 transects in Kruger National Park, South Africa, in 2006, and trees were resampled in 2008, 2011, 2015 and 2018. Over the 12-year period, we recorded a cumulative mortality of 47.6% with an estimated annual mortality of 3-5% from 2006 through 2015 and an increase to 8.8% in 2018. The main attributed agent of tree mortality was elephant damage, occurring either once or across multiple census periods. A classification tree analysis partitioned over different census periods showed that the probability of mortality of the 10 most common tree species depended not only on the type and intensity of elephant-induced damage and fire but also on the historical sequence of damage by these agents. Elephant damage to the main stem incurred up to 12 years earlier increased risk of mortality for large trees, especially in combination with fire damage. As expected, vulnerability to damage and risk of mortality varied among species, resulting in the potential for long-term changes in species composition over large areas. Overall, this study highlights how multiple interacting agents cause emergent and lagging patterns of mortality risk for large trees in savanna ecosystems, a result that only becomes apparent through fine resolution and long-term tracking of cohorts.