Landslides are triggered by factors such as heavy rainfall, seismic activity, and construction on hill-slopes. The leading cause of landslides in Puerto Rico is intense and/or prolonged rainfall. A rainfall threshold for rainfall-triggered landsliding is delimited by 256 storms that occurred between 1959 and 1991 in the central mountains of Puerto Rico, where mean annual rainfall is close to or in excess of 2,000 mm. Forty one of the 256 storms produced intense and/or prolonged rainfall that resulted in tens to hundreds of landslides. A threshold fitted to the lower boundary of the field defined by landslide-triggering storms is expressed as I = 91.46 D-0.82 where I is rainfall intensity in millimeters per hour, and D is duration in hours. Landslide-producing storms occurred at an average rate of 1.2 per year. In general the landslides triggered by short-duration, high-intensity rainfall events were mainly shallow soil slips and debris flows, while the long-duration, low-intensity rainfall produced larger, deeper debris avalanches and slumps. For storms that had durations of up to 10 h, landsliding did not occur until rainfall intensity was as much as three times as high as the rainfall intensity reported as sufficient to trigger landsliding in temperate regions. As storm durations approach 100 h, the rainfall conditions necessary to initiate landsliding in Puerto Rico converge with those defined for temperate regions. A comparison of the Puerto Rico threshold with rainfall data from other humid-tropical regions suggests that the threshold developed for Puerto Rico may be applicable to other similar environments throughout the world.

The Luquillo Mountains of northeastern Puerto Rico harbours important fractions of tropical montane cloud forests. Although it is well known that the frequent occurrence of dense fog is a common climatic characteristic of cloud forests around the world, it is poorly understood how fog processes shape and influence these ecosystems. Our study focuses on the physical characteristics of fog and quantifies the fogwater input to elfin cloud forest using direct eddy covariance net flux measurements during a 43-day period in 2002.We used an ultrasonic anemometer–thermometer in combination with a size-resolving cloud droplet spectrometer capable of providing number counts in 40 droplet size classes at a rate of 12.5 times per second. Fog occurred during 85% of the time, and dense fog with a visibility <200 m persisted during 74% of the period. Fog droplet size depended linearly on liquid water content(r2 ¼ 0:89) with a volume-weighted mean diameter of 13.8 mm. Due to the high frequency of occurrence of fog the total fogwater deposition measured with the eddy covariance method and corrected for condensation and advection effects in the persistent upslope air flow, averaged 4.36 mm day1, rainfall during the same period was 28 mm day1. Thus, our estimates of the contribution of fogwater to the hydrological budget of elfin cloud forests is considerable and higher than in any other location for which comparable data exist but still not a very large component in the hydrological budget. For estimating fogwater fluxes for locations without detailed information about fog droplet distributions we provide simple empirical relationships using visibility data.