Factors Affecting Tropical Tree Damage and Survival after Catastrophic Wind Disturbance

The structure and dynamics of cyclone‐prone tropical forests are driven in part by variation in tree species resistance to and survival after wind‐induced structural damage. We determined the factors associated with tree damage and 3‐yr survival following Category 5 Cyclone Olaf on the Polynesian island of Ta'u, American Samoa. Despite sustaining a high rate of severe damage (34.6% of all trees snapped, 23.0% uprooted), system resilience was high with 74.3 percent stem survival overall and an annual mortality rate of 7.9 percent compared with 2.1 percent in nearby undisturbed late successional forest. Three‐yr survival rate of trees sustaining severe damage was 63.1 percent, compared to about 89 percent for trees sustaining only branch loss or defoliation. Three‐yr survival differed according to damage type, 78.5 percent after snapping vs. 38.4 percent after uprooting. Species differed widely in resistance to and survival after snapping and uprooting. Several species and individual traits were associated with the probability of snapping or uprooting; however, wood density was the only species trait consistently, and negatively, associated with the probability of sustaining either damage type. Survival after snapping was negatively associated with the proportion of the tree snapped off, which was determined by individual tree architecture. Species growth rate was negatively associated with survival after uprooting, indicating the importance of shade tolerance for survival after uprooting. Thus, whereas species traits seemed to exclusively underpin resistance to and survival after uprooting, they only partly accounted for snapping resistance, and did not determine the intensity of snap damage or survival after snapping. Our results highlight the importance of considering each damage type separately when considering ecological trade‐offs.     

Frequent,low‐amplitude disturbances drive high tree turnover rates on a remote,cyclone‐prone Polynesian island

Aim How important are frequent, low‐intensity disturbances to tree community dynamics of a cyclone‐prone forest? We tested the following hypotheses concerning the ‘inter‐cataclysm’ period on a remote Polynesian island: (1) tree turnover would be high and recruitment rates would be significantly higher than mortality; (2) low‐intensity disturbance would result in a marginal increase in tree mortality in the short term; (3) turnover would vary among species and would be associated with plant traits linked to differences in life history; and (4) mortality and recruitment events would be spatially non‐random. Location Tutuila, a volcanic island in the Samoan Archipelago, Polynesia. Methods We censused the tree (stem diameter ≥ 10 cm) community in 3.9 ha of tropical forest three times over a 10‐year period, 1998–2008. We calculated annual mortality, recruitment and turnover rates for 36 tree species. We tested for non‐random spatial patterns and predictors of mortality, and non‐random spatial patterns of tree recruitment. A 2004 cyclone passing within 400 km allowed us to measure the effects of a non‐cataclysmic disturbance on vital rates. Results Annual turnover was 2.8% and annual recruitment was 3.6%; these are some of the highest rates in the tropics, and likely to be a response to a cyclone that passed < 50 km from Tutuila in 1991. Species turnover rates over 10 years were negatively correlated with wood specific gravity, and positively correlated with annual stem diameter increment. Mortality was spatially aggregated, and a function of site, species and an individual’s growth rate. Recruitment was highest on ground with low slope. The low‐magnitude cyclone disturbance in 2004 defoliated 29% of all trees, but killed only 1.8% of trees immediately and increased annual mortality over 5 years by 0.7%. Main conclusions The inter‐cataclysm period on Tutuila is characterized by frequent, low‐amplitude disturbances that promote high rates of tree recruitment and create a dynamic, non‐equilibrium or disturbed island disequilibrium tree community. Species with low wood density and fast growth rates have enhanced opportunities for recruitment between cataclysms, but also higher probabilities of dying. Our results suggest that increases in the frequency of cyclone activity could shift relative abundances towards disturbance‐specialist species and new forest turnover rates.     

Diversity and structure of tropical rain forest of Tutuila,American Samoa: effects of site age and substrate

We report tree community diversity, guild composition, and forest structure from three 1.2 ha (100 m × 120 m) permanent forest research plots on Tutuila, American Samoa, an isolated volcanic island in the South Pacific Ocean. Plots were established in three habitat types of lowland hill forest: two in mature tracts that differed in substrate type (talus vs non-talus), and a third in a 30–40 year-old abandoned plantation on non-talus soil. We encountered a total of 57 tree species 10 cm dbh. Richness was similar across sites, but composition differed substantially. We were able to classify with confidence 24 tree species into four distinct guilds based on forest-type preference: early successional or persistent successional (disturbed forest), generalist (no preference), or mature-phase. Sample size limitations or interactions between site age and substrate precluded categorization of all species. Thirty-eight percent (9/24) of the tree species were successional, a result which contrasts sharply with data from (formerly) continental forest in Panama. Spatial distributions of 33 species revealed 17 species exhibiting clumping or hyperdispersion (i.e., regular spacing) in at least one site. Possible non-anthropogenic mechanisms promoting clumping in the plots were (a) topography (edaphic), (b) gap affiliation, (c) inefficient or altered patterns of propagule dispersal, and (d) lack of natural seedling predators. Forest structure differed across site type, with stem densities highest in regenerating forest; conversely, regenerating forest had the lowest basal areas. Steep talus forest sequestered the most carbon (344.3 Mg ha^–1), and secondary forest sustained only 42% of levels found in talus forest (145.5 Mg ha^–1). Mature forest on non-talus soil sequestered the majority of carbon in mid-sized trees (30–50 cm dbh). Future assessments of land-use cover and biomass will provide for a complete estimate of the carbon budget of Tutuila. Finally, the results of this study suggest that conservation of the native fauna is essential in retaining the potential for regeneration of native forest after large-scale disturbance.