Mangrove forests occupy the intertidal zone at the land-sea interface; at landfall, hurricanes are often most intense and inflict destructive physical forces of wind and surge. Because hurricanes are fueled by warm sea temperatures, they are usually at their greatest overland strength at initial landfall. Mangroves are a prominent and ubiquitous ecosystem throughout the world's tropical coastlines, yet little has been published on their response and resiliency to tropical cyclones. Recent hurricanes, Hugo in 1989 and Andrew in 1992, have spawned numerous scientific studies of storm effects on plant and animal communities (FINKL and PILKEY, 1991; LOOPE et al., 1994; PIMM et al., 1994; WALKER et al., 1991) establishing a baseline of research for comparisons among regions, forest types, and species.

Some of the earliest observations of hurricane effects on south Florida mangroves were reported by CRAIGHEAD (1964), DAVIS (1940), and EGLER (1952). Craighead, in particular, described and photographed hurricane-damaged vegetation, including mangroves, throughout Everglades National Park following the 1947,1960, and 1965 storms. He also documented a range of damage responses from complete blowdowns, canopy defoliation, and root suffocation following hurricanes Donna in 1960 and Betsy in 1965 (CRAIGHEAD and GILBERT, 1962; CRAIGHEAD, 1971). Follow-up studies by Craighead showed that mangrove trees in some areas were still undergoing delayed mortality as long as ten years after initial impact (CRAIGHEAD, 1971).

More recently, SMITH et al. (1994) reported on initial storm effects of Hurricane Andrew for east and west coast mangroves of south Florida. Their work combined aerial and ground surveys showing the distribution and pattern of mangrove destruction and mortality on a local and regional basis. Six months after initial surveys, SMITH et al. (1994) found that tree death was not a static process with the passing of Hurricane Andrew, but rather a dynamic struggle of survivorship (resprouting) and delayed mortality akin to Craighead's findings. Their results also showed that young recruits in pre-existing light gaps proved resilient to impact, indicating a possible interaction between small-scale lightning gap dynamics with less frequent, large-scale hurricane effects.

Structural impacts from hurricanes have been reported for mangrove systems elsewhere. STODDART (1963) documented the local effects of wind and surge from Hurricane Hattie in 1961 on mangrove and reef structure of Belize. Stoddart described and mapped the general extent and compass orientation of forest damage attributed to direct wind and wave action. In a report on damage caused by cyclone Kathy to mangrove forests in Australia, BARDSLEY (1985) determined that dwarf mangrove species and forests sustained little impact in contrast to tall emergent mangroves. WUNDERLE et al. (1992) censused a number of forest types in Jamaica following Hurricane Gilbert in 1988 and found the greatest structural damage in their mangrove plots. They reported that average foliage height was significantly reduced by 4 m or more due to the severe loss of dominant overstory trees. Following Hurricanes Domoina and Imboa in 1984, STEINKE and WARD (1989) conducted mortality counts of mangrove trees by species found in the St. Lucia estuary on the east coast of southern Africa. Their work estimated 33% mortality of white mangrove, Bruguiera gymnorrhiza, which made up 92% of the woody stems in near shoreline sites. Observations by ROTH (1992) of Hurricane Joan in 1988 on the Nicaraguan coast also affirm the susceptibility of emergent mangroves to windthrow under hurricane conditions.

Collectively, these studies document the susceptibility and vulnerability of mangrove species and systems to hurricane disturbance. However, it is yet unclear how the physical and biological elements interact to explain the varying degrees of windthrow and mortality relative to hurricane intensity and spatial dynamics. In this study, we investigated wind damage relationships of Hurricane Andrew on mangrove forest structure and sites along the southwest coast of Florida, USA. Important questions that were addressed included the degree and expression of impact in relation to the path and intensity of Andrew as it moved across and exited the peninsula. We also examined differences in tree size and species susceptibility within and between sites under different projected wind conditions. A hurricane simulation model was used to reconstruct predicted wind speeds and vectors for each study site for every 15 minutes of Andrew's course across south Florida. A chronological overview of Andrew's storm track and general meteorology are given in STONE et al. (1993).