Is drought‐induced forest dieback globally increasing?

• Published in: The Journal of ecology 2015, Vol.103 (1), p.31-43
• Main Authors: Steinkamp, Jörg, Hickler, Thomas, Gibson, David
• Format: Article
• Language: English
• Subjects: aridity index; Climate; climate impacts; data collection; dieback; Drought; dry forests; drying; dynamic global vegetation model; forest dieback; Forests; heat stress; LPJ‐GUESS; Mortality; palmer drought severity index; plant–climate interactions; soil; Special Feature – Standard Papers; tree mortality; Vegetation; water stress
• ISSN: 0022-0477; 1365-2745
• DOI: 10.1111/1365-2745.12335

Recently, it has been suggested that forest mortality has been generally increasing because of increasing drought and heat stress. But it is unclear if the observations at the investigated forest sites and regions are representative of forests globally and it has not been tested whether forest models are capable of reproducing these observations. We analysed historical climate data and used a dynamic global vegetation model (LPJ‐GUESS) to assess (i) Which forests globally might have been affected by drought, (ii) If the field observations are representative for all forests and (iii) If the model can reproduce the reported mortality events. Using two climate data sets and three drought indices, we identified no general global drying trend across all forests, but a large spatial variability. We neither detected a general increase in extreme drought events. A weak drying trend and an increase in extreme drought events were only apparent for forests in already dry climates and the locations or regions for which drought‐induced mortality trends have been reported are predominantly in these dry climates. LPJ‐GUESS reproduced 66% of the reported mortality events and in 49% of the reported drought‐induced mortality events drought was apparent in any of the climatologically derived drought indices. However, only in 30% of the cases simulated increased mortality coincided with drought events. Synthesis. Our results indeed suggest that dry forests have been experiencing increasing drought‐induced mortality. However, this does not apply to forests in general and the spatial variability has been large. The poor correspondence between the simulated and reported mortality events indicates that models like LPJ‐GUESS driven by standard climatologies, and soil input data do not represent drought‐induced mortality well. But the poor detection of the reported drought events in our climate indices also suggests that drought stress might not be the main driver of all the reported drought‐mortality events.