Electronic versions

Documents

DOI

The effects of the spatiotemporal (> 100 years) range of hurricane disturbance intensity on tree diversity and density patterns are largely unknown, because data on past stand or landscape scale hurricane impacts are sometimes unavailable. We therefore reconstructed and mapped topographic exposure (a proxy to disturbance) to twelve category 2–4 hurricanes that affected the rain forests of the Blue Mountains (BM) and the John Crow Mountains (JCM) in Jamaica, over 155 years. Maps of average topographic exposure and the spatial outputs from a pixel-based polynomial regression of the cardinal directions of the tracks of past hurricanes (predictor) and past exposure (response) were then used to represent the aggregate spatiotemporal range of exposure. Next, we used data collected over the period 1974-2009 from 35, 10 x10 m nested subplots and 1991-2004 from 16, 200 m2 circular plots for the BM and 2006-2012 from 45, 25 x 25 m plots for the JCM, and Bayesian spatiotemporal, Integrated Nested Laplace Approximation (INLA) models to determine whether stand-level (≈ 1 km2) tree Shannon diversity and density patterns were primarily influenced by exposure to a single hurricane, the most severe hurricane or to multiple hurricanes and the duration of hurricane effects on Shannon diversity and tree density. In the BM, long-term diversity peaked at locations with intermediate values of average exposure for six hurricanes (five of which made landfall over the period 1903-1988). Short-term diversity peaked at locations that experienced significantly higher exposure situated to the south or north of the hurricane’s track when the tracks were to the north or south of the island, respectively. Short-term density peaked at locations that were always highly exposed. Moreover, the influence of the most severe hurricane on diversity can last up to 101 years and the influence of the most recent hurricane (Gilbert) on diversity became evident after 16 - 21 years. The JCM was more susceptible to hurricanes and this diminished the influence of past hurricanes. Consequently, density peaked at sites with the highest average exposure to the four most recent hurricanes (1988-2007), only one of which made landfall. If historical hurricane disturbance data are unavailable, reconstructed exposure maps can be used to provide valuable insights into the effects of past hurricanes on stand-level tree diversity and density patterns.

Keywords

  • Bayesian, Cyclone, Forest structure, Integrated nested laplace approximation, Intermediate disturbance hypothesis, Spatiotemporal models, Topographic exposure
Original languageEnglish
Number of pages21
JournalAgricultural and Forest Meteorology
Volume276-277
Early online date19 Jun 2019
DOIs
Publication statusPublished - 15 Oct 2019

Total downloads

No data available
View graph of relations