DISCUSSION

Forest height confounds a relationship between maximum projected hurricane wind speeds and volume of woody debris (Fig. 3). In ENP-Eye, the reduced forest stature of JMC corresponded to less woody debris and was suggestive of a potential relationship between forest height and downed wood (Table 1). However, only 28 percent of the variation in the volume of woody debris can be explained by the linear relationship of woody debris with forest height in some areas (Fig. 3a). Similarly, forest height and woody debris are not correlated for plots with maximum estimated windspeeds of 200-205 km/h (Fig. 3b).

Sites associated with the greatest Hurricane Andrew storm winds had, as expected, larger amounts of woody debris. Transects that were run perpendicular to the prevailing azimuth of downed wood orientation resulted in large estimates of woody debris, which may have biased surveys for some transects. Within the eyewall, defined azimuths of downed logs were less pronounced (Doyle et al. 1995) and may have been influenced less by sampling with a random azimuth. The few transects that were established in Everglades City, just outside the eyewall, may have been affected slightly by this artifact. It is unlikely, however, that estimates were skewed greatly since wood volume was visibly high on these sites. Everglades City plots were also isolated as a noncontinuous forest patch. Not only were decomposition rates potentially lower on these drier sites, but also hurricane winds were fairly unimpeded by surrounding forests and may have resulted in greater impact than HURASIM-predicted maximum windspeeds of 220 km/h indicated. Overall, however, these results suggest that woody debris, which is associated with tropical storm activity, can be fairly high in South Florida mangrove ecosystems.

Downed wood represents a large carbon and potential nutrient pool in mangrove wetlands (Robertson & Daniel 1989, cf., Harmon et al. 1986). Some 9-10 yr following Hurricane Andrew, woody debris is still prevalent on South Florida sites. Much of the downed wood, especially in the eyewall and areas to the immediate right of the storm path, were large trees and, therefore, had much slower decay rates relative to smaller woody debris. Mangrove trunk wood, for example, was still present in the understory after falling 15 yr previously (Robertson & Daniel 1989). Mangrove trees killed by Hurricane Donna (1960) in South Florida had stumps present in 1993, and many R. mangle boles were not completely decomposed even after 30 yr (T. Doyle, pers. obs.). The distrubution of downed wood in this investigation (Fig. 2) suggests that much of the intersected coarse woody debris, especially in ENP-Eye and ENP-Right, resulted directly from Hurricane Andrew. As a consequence of either the distance from Hurricane Andrew's influence or lower forest stature, respectively, estimates from RB-Right or TAY-Left may provide a better idea of background woody debris levels under a reduced relative hurricane incidence regime in South Florida.

Our survey technique, which differed slightly from Allen et al. (2000) in that we estimated wood that rotted away internally from large downed logs, accounted for overestimating woody debris based upon outside diameter measurements only. Since the majority of debris on our surveys was categorized as rotten (Table 2), and much of what was remaining was of downed logs likely to be around for many additional years, it was important to subtract the debris that had already disappeared. This was not a major consideration for surveys conducted by Allen et al. (2000) in Micronesia, since most downed logs were fully intact (J. Allen, pers. comm.). Long-term woody debris persistence has been noted in other surveys from tropical areas, but with trees remaining erect during the earlier stages of decomposition (cf., Delaney et al. 1998, Santiago 2000). Most mortality in South Florida mangroves was not associated with attrition of large trees with a standing decomposition period, as in Jiménez et al. (1985), but rather from strong winds, with most decomposition occurring on the ground after tree or branch fall. Tree mortality due to lightning strikes (Smith et al. 1994) and single-tree attrition cannot be discounted, especially in RB-Right and TAY-Left plots. Shifts in the relative percentage of coarse woody debris versus fine woody debris by region (Table 2; Fig. 2) and the decomposition state of larger wood further indicates that the majority of the coarse woody debris in ENP-Eye and ENP-Right sites may have been from Hurricane Andrew. In fact, a fine:coarse woody debris ratio of 0.37 and 0.44 for ENP-Eye and ENP-Right, respectively, indicates that lower ratios may indicate a more recent disturbance when compared to sites of similar forest stature. Fine-to-coarse woody debris ratios exceeded 0.75 for RB-Right and ENP-Left; TAY-Left sites (3.0) were not comparable on the basis of a much reduced relative forest stature. A low fine-to-coarse woody debris ratio of 0.11 was found on mangrove sites on Kosrae, Federated States of Micronesia, under an accelerated individual tree harvest regime relative to mangrove forests of Pohnpei (0.56) and Yap (0.61: Allen et al. 2000). Such analyses of components of downed wood may provide a relative index of disturbance among similar mangrove forests in a geographical region; however, the variation in this metric indicates that more research is warranted.

Accordingly, fine woody debris comprised 36 percent of the total debris on a per volume basis (Table 2). At least that percentage of woody debris is probably not associated with direct hurricane influence, although delayed mortality of trees and subsequent woody debris fall cannot be discounted (Whigham et al. 1991, Smith et al. 1994, Sherman et al. 2001). Converting volume of fine woody debris into biomass equates to about 12.0 t/ha over all sites, or approximately one half of the total biomass of combined woody debris, which is estimated at 23.8 t/ha. Roughly, 14.1 t/ha of fine woody debris and 16.3 t/ha of coarse woody debris were recorded from eyewall and immediate right-side impact regions of Hurricane Andrew. Any disparity in percentages of volume and mass is associated with the large amount of coarse woody debris (>7.5 cm) categorized as rotten; decomposed mangrove wood is assumed to weigh 60 percent less than sound wood (Allen et al. 2000). Therefore, woody debris mass is not expected to follow the same patterns depicted for volume (as in Fig. 2). If surveys had been conducted just after the passage of Hurricane Andrew, much greater biomass of woody debris would have been likely on some sites (>40 t/ha). Even those projected values would have been quite low compared to some estimates as high as 550 t/ha from coniferous forests in the U.S. Pacific Northwest (Harmon et al. 1986). Much larger tree size and potentially slower decomposition rates in Harmon et al. (1986) can account for some differences relative to South Florida mangroves.

Volume and biomass of woody debris from South Florida mangroves represent high levels in some regions (i.e., ENP-Eye, ENP-Right) and low to intermediate levels in others relative to published accounts from other mangrove wetlands. Robertson and Daniel (1989) reported 9.5 t/ha of woody debris in Australian mangrove wetlands, considerably less than our estimate from South Florida. However, woody debris volume and biomass from RB-Right, ENP-Left, and TAY-Left regions were less than debris stores from mangrove forests on Kosrae, Micronesia. A measured downed wood volume of 104 m³/ha in Allen et al. (2000) exceeded levels found in areas impacted only moderately by Hurricane Andrew, but was similar to the volume of 98-132 m³/ha reported from ENP-Eye and ENP-Right regions (Table 2). Large amounts of woody debris on Kosrae were attributed to forest stature, with tree DBH ranging from 15 to 130 cm and height ranging from 19 to 27 m (Ewel et al. 1998), and to an accelerated harvest rate (Allen et al. 2000). In South Florida, we can attribute the large volume of woody debris in some locations to a direct hurricane effect.

In general, mangrove wetlands support less woody debris than upland forests (Allen et al. 2000). Hydrological conditions of mangrove wetlands, which include a diversity of tide, precipitation, and river-flow regimes, can complicate direct comparisons with upland forests. Polit and Brown (1996) indicate that lowered stocks of woody debris may be partially explained by the higher decomposition rates of woody debris in wetlands. Also, decay of fallen mangrove wood may be quick at first, relative to most temperate systems, due in part to consistently higher temperatures, a prolonged wet season, and a combined terrestrial and marine fungal community in mangroves (cf., Kathiresan & Bingham 2001). Although we did not have comparable upland sites in this investigation, our sites in Everglades City were much dryer than in other locations and were partially isolated hydrologically by a nearby road. These sites also supported some of the highest estimated levels of woody debris. Yet, a reduced moisture-related decomposition rate is not supported by Harmon et al. (1987), who reported an inverse relationship between woody debris decay and precipitation by geographic region. It may be difficult, however, to compare a trend based upon freshwater moisture in an upland environment with one based upon saltwater moisture from tidal inundation. Plus, soil waterlogging, as measured through soil oxidation-reduction, did not seem to influence the large volumes of woody debris (estimated at 136-428 m³/ha) in a Hawaiian montane cloud forest (Santiago 2000).

Forest height was only a moderate predictor of woody debris in South Florida mangrove forests, even after stratifying forests by maximum sustained Hurricane Andrew windspeeds (Fig. 3). Total stand volume of a mangrove forest can also be a poor indicator of downed wood (Allen et al. 2000), yet the relationship between forest basal area and coarse woody debris can be strong (Santiago 2000). Increases in downed wood with an increase in maximum hurricane windspeed relative to preimpact standing biomass may be a more useful predictor of downed woody debris, since this method may account for three-dimensional aspects of tree and forest structure. Determining a more appropriate metric to predict the volume of downed wood in Florida mangrove forests may involve intensive studies on plot-level variation within forests and should be a focus of future woody debris research efforts on long-term ecological research plots.