Discussion

Closed canopy forest stand structure and growth

The intact closed canopy forest structure and growth results were comparable with the other reported values for the mangrove forest along the lower Shark River drainage, Everglades. In the intact forest the three species (based on adult biomass) were co-dominants (39, 26, and 35%, A. germinans, L. racemosa, and R. mangle, respectively). If I combine the adult and sapling life stages (to make them comparable to Chen and Twilley 1999), the density (3,240 ha^-1) and aboveground biomass (156 Mg ha^-1) were in the same range as values reported by Chen and Twilley (site 1.8, 1999). I report approximately 100 Mg ha^-1 less total aboveground biomass compared to Chen and Twilley (1999); however, this is most likely due to the difference in the allometric relationships used to calculate biomass. It appears that mangroves allometric relationships are area specific (see Smith and Whelan, in review). The allometric equations used in this study were determined in this forest (Smith and Whelan, in review) and Chen and Twilley (1999) used relationships generated for mangroves found in Puerto Rico and Mexico. Additionally, the intact forest sites in this study had a greater proportion of A. germinans, both as stem density and biomass than was found by Chen and Twilley (1999, 33 and 39 compared to 8 and 21 %, respectively).

Rates of mangrove seedling survival in closed canopy forest have rarely been reported (Rabinowitz 1978b [Rhizophora mangle, Avicennia germinans, Laguncularia racemosa], Clarke 1995 [Avicennia marina], Ha et al 2003 [Kandelia candel], Padilla et al 2004 [Rhizophora sp.]). Generally, seedling survivorship is reported in reference to disturbance or in experimental manipulations (e.g. Smith 1987, Osborne and Smith 1990, Clarke and Allaway 1993, Sousa et al. 2003). I found that seedling survival differed by species in the closed forest. A. germinans had the greatest seedling survival (91 %), R. mangle survival was moderate (59 %) and no L. racemosa seedlings survived. In Panama, Rabinowitz (1978b) reported 70 % survival for R. mangle and 10 % survival of A. germinans after one year and no survival of L. racemosa after 100 days. In the Rabinowitz study, the three species were located in closed canopy forest, but at different locations within the forest. For example R. mangle seedlings were at the water's edge and the A. germinans seedlings were 0.5 km from the waters edge. In my study all seedlings were co-located in the same plots. Sousa et al (2003) recorded low R. mangle seedling survivorship (< 38 %) in closed forest in Panama, but a stem boring scolytid beetle, Coccotrypes rhizophorae, reduced survivorship. Even though Coccotrypes rhizophorae has been reported in Florida (Atkinson and Peck 1994), I found no evidence (rust-colored frass emitted from entrance holes, Sousa et al 2003) of seedlings being attacked.

To the best of my knowledge, recruitment rates of seedlings in intact canopy mangrove forest have only been reported three times prior to this study (Clarke 1995 [Avicennia marina], Ha et al 2003 [Kandelia candel], Padilla et al 2004 [Rhizophora sp.]). Clearly, more work needs to be done in this area of mangrove research. Seedling recruitment rate in the forest was 0.11 yr^-1 (per capita) and dominated by A. germinans and L. racemosa (0.87 and 0.26 yr^-1, respectively, Table 11). My values are similar to the values reported for Rhizophora sp. in a mangrove forest in the Philippines (0.07 compared to 0.05 yr^-1, for a site with relatively low recruitment). Recruitment was three times greater than mortality for A. germinans seedlings. For the other species, the seedling recruitment rate was less than half the mortality rate (Table 11), even though R. mangle seedlings were very abundant in both the initial survey and at the second survey. These results suggest that A. germinans seedling population was increasing whereas R. mangle seedling population was decreasing. Additionally, the lack of L. racemosa survivorship (1.0 yr^-1 mortality rate) supports the observation that L. racemosa needs higher light conditions than closed canopy forest to survive (Ball 1980, Mckee 1995).

The seedling stem elongation rate of R. mangle (0.12 mm yr^-1) was lower but similar to that reported by Koch (1997, 0.2 mm yr^-1) for closed canopy sites on the nearby Little Shark River (approximately 3 km from my study sites, size class 25-40 cm stems). R. mangle stem elongation rate (0.36 cm mo^-1) from this study was lower than the average monthly reported value (0.8 cm mo^-1) from a study of seedlings growing in a closed canopy forest in Columbia (Elster et al. 1999). The annual growth of seedlings in this study is also lower than that in a year-long study of Rhizophora sp. seedling in the Philippines (4.28 cm yr^-1 compared to 5.6 to 10.6 cm yr^-1). The seedling stem elongation rate of A. germinans in the closed canopy forest decreased during my study period (-0.08 mm d^-1) mainly due to a single individual, in which the stem length deceased from 87 to 61.5 cm from unknown causes. However, the total leaves on the seedling increased from 39 to 45 during this same period of time. Removing this one individual, I found that A. germinans essentially had no stem elongation during this period of time (-0.01 mm d^-1). A. germinans ability to translocate energies to other meristems (having a shrub like appearance) has been noted by others, but in reference to hurricane damage of adult trees (Baldwin et al. 2001). Elster et al (1999) also reported negative stem elongation growth of A. germinans but it was attributed to attacks by Junonia evarte caterpillars.

Sapling survival was high (97 %) overall but lower for L. racemosa (83 %). Sapling recruitment was 0.3 and 1.3 saplings per 500 m², (a per capita recruitment rate of 0.04 and 0.10) for A. germinans and R. mangle, respectively, compared to estimates by Chen and Twilley (5 and 30 saplings per 500 m², A. germinans and R. mangle, respectively, 1998). A. germinans recruitment rate was four times greater than mortality, whereas, R. mangle recruitment was 2.5 times the morality rate (Table 11). This would suggest that at the sapling stage A. germinans is slowly increasing in population size in the intact forest whereas R. mangle sapling population while expanding is at a slower per capita rate once corrected for mortality within the sapling size class.

The change in sapling biomass and relative growth rate was not different among the three species within the forest sites. Average change in adult biomass (1.94 ± 0.34 SE kg tree^-1 yr^-1) for the forest is comparable but lower than values reported by Chen and Twilley (1999). Unlike the high values reported by Chen and Twilley, for L. racemosa (9.17 kg tree^-1 yr^-1), in my study, all three species had similar value for change in biomass for the forest sites (Table 10). Additionally, there was no difference in the relative growth rate between the species. Survival of adults was high (100 % during our study period) and there was no recruitment to the adult stage. These results suggest that future changes in forest structure are due to replacement of adult trees by non-conspecifics (ie. seedling and sapling recruitment and mortality dynamics).

New gaps versus intact forest

Post-disturbance mortality has been found from corals to trees (Knowlton et al. 1981, Platt et al 2000, Sherman et al. 2001), so it is not surprising that I documented post-lightning strike mortality at two of the three new gap sites. The plots were established six months post strike so I may have missed some of the delayed mortality that may have occurred and this may be included in my initial mortality estimates. Knowlton et al. (1981) cautioned against delayed disturbance monitoring indicating that mortality could be underestimated due to decomposition or loss of dead members. However, I feel that due to extremely slow mangrove decomposition (Romero et al. 2005) this was not an issue at my site. I also did not find any evidence of lightning killed seedlings or propagules in my smaller study plots within each gap. Overall, I only found 3 to 5 lightning killed seedlings within the new lightning initiated gaps. These were always located at the center of the gap associated with the tree assumed to be the initial strike tree. Within the new gaps the majority of the lightning damaged but surviving trees were located on the edge of the gap. The dead trees and saplings were concentrated in the center of the gap opening.

The propagules were assumed to be transitory in nature since they were not attached to the substrate when censused (i.e. dispersal phase). Tomlinson (1986a) reported that in Florida, L. racemosa and A. germinans fruit drop occurs during the late summer and fall (July to November and September to October, respectively). R. mangle fruit drop can occur through the year (Tomlinson 1986a). The objective of my January to March vegetation sampling was to mainly capture the newly recruiting seedlings and I feel that the three-month period allowed adequate time for the propagules to attach to the substrate. I found significantly more L. racemosa propagules in the new gaps compared to the intact forest for the first survey, however this trend was not significant at the second survey (Table 11). For both surveys L. racemosa comprised the majority of the all the propagules censused in the new gaps. For both surveys there were at least half as many R. mangle propagules in the new gaps compared to the forest (Table 11). The low number of R. mangle propagules within the new gaps may be due to the reduction in nearby propagule source trees. The gaps are small in size and my assumption was that the surrounding R. mangle trees would be a source of propagules to disperse into the new gaps. Even though R. mangle propagules are believe to be a good long distant dispersal structure, it has been reported that R. mangle trees within a forest (i.e. not adjacent to a waterway) have limited dispersal of the propagules (“maximum of 8 m, and on average less than 3 m from the point of release”, Sousa et al. 2003), suggesting that R. mangle propagules may be dispersal limited within the forest. I am unaware of dispersal distance data for L. racemosa propagules, however, the dispersal by a small propagule through the tangle of mangrove prop roots should be greater than the larger R. mangle propagule.

Even though there were twice as many R. mangle propagules in the intact forest compared to the new gaps, I did not find a corresponding increase in R. mangle seedling recruitment (Table 3 and 11). Overall, the greatest number of recruiting seedlings was R. mangle (26 of 52) and there was no difference between the forest and the new gaps based on seedling density; however, the seedling recruitment rate was twice as high in the new gaps compared to the forest (Table 11). New gaps and the forest had similar numbers of A. germinans seedling recruits, however; the per capita recruitment rate was twice as great in the new gaps compared to the forest (Table 11).

I found significantly higher seedling recruitment for L. racemosa within the new gaps compared to the forest (Table 11). My data suggest that there maybe-additional dispersal of L. racemosa propagules into these new opened sites, since I recorded L. racemosa propagules at a new gap (new gap 3) where no canopy trees of L. racemosa were present. Additionally, five of the ten recruiting L. racemosa seedlings that occurred in the study were found at new gap 3. These results would support suggestions by Ball (1980) that canopy openings (i.e. lightning strikes) within a monospecific stand of R. mangle would allow L. racemosa seedlings to recruit.

Contrary to the results of Ellison and Farnsworth (1993), in which there was increased survivorship of R. mangle and A. germinans seedling in an experimentally removed canopy, in my study the survival of seedlings were similar to the intact forest sites (Table 11). Additionally, I did not find a difference in sapling survivorship between new gaps and the intact forest (Table 11). Typically seedlings and saplings predisposed to low light conditions will suffer harmful effects to being exposed to high light (“irreversible photo-oxidation”, Luttge 1997). In my study there was no greater amount of mortality associated with the high light new gaps compared to low light intact forest (16 and 8 % mean canopy openness, respectively). The greatest seedling survival was for A. germinans followed by R. mangle. L. racemosa survival was low in both forest types (mortality rate of -2.76 and -1.00 yr^-1, respectively Table 11). This finding is contrary to the suggestion in the literature that L. racemosa should have increased survivorship in high light environments (i.e. new gaps) due to its higher light needs for establishment (Ball 1980). However, some species do not have the ability to survive dramatic changes in light conditions as well as others. Increased mortality of L. racemosa seedlings in high light new gaps may be more of an indication that once L. racemosa develops under certain light condition, transition to others could be difficult. Transplant experiments of the seedlings of the three species from closed canopy forest to highlight open marsh found (10, 40, 80 % survivorship, A. germinans, L. racemosa, and R. mangle, respectively, K. Whelan and T. Smith, unpublished data). Mckee (1995) also reported 20 % A. germinans and 40 % R. mangle seedling survivorship when transplanted from closed canopy forest to open canopy site within forest dominated by nonconspecifics.

Summed across all species seedlings had greater stem elongation in the high light new gap sites compared to the forest. The few L. racemosa seedlings (n = 2, all in new gaps) had a high growth rate (11.5 cm yr^-1) and the few A. germinans seedlings in the forest had negative growth. The seedling stem elongation rate of R. mangle (0.26 mm d^-1) was greater in new gaps than the forest (Table 11). This is approximately half the elongation rate reported by Koch (1997, 0.6 mm d^-1) for gap sites on the Little Shark River. Seeing that no mortality was reported for any of the tagged seedlings in the Koch study, some selection may have occurred for more vigorously growing seedlings. Additionally, the high rate of seedling growth occurred over the dry season whereas in this study sampling occurred over the majority of the year (11 to 13 month time period between sampling). Similar to my results, Ellison and Fransworth (1993) found greater stem elongation rates of R. mangle seedlings in canopy removal sites. In summary, the seedling stem elongation in the new gaps appears to have been greater than that in the lower light forest sites.

Sapling relative growth rate was three times higher in the new gaps compared to the forest, suggesting a release of resource limitations (presumably light) for surviving stems. Within the new gaps, R. mangle saplings had higher relative growth rate than A. germinans (L. racemosa could not be compared due to low sample size). These results are contrary to those of the Sherman et al. (2000) study in which A. germinans had the highest relative growth rate within lightning-initiated gaps. Additionally, sapling survival in this study was high for all three species within the new gaps. This is contrary to the results of Sherman et al. (2000), who reported < 60 % survivorship of A. germinans saplings within a two-year period. Direct comparison of sapling growth and survivorship is difficult because in the Sherman et al. (2000) study, sites ranging from new gaps to 10-year-old gaps were lumped together, with no reporting of the proportions of the sample sizes for the different gap stages.

I found no difference in adult recruitment, survival, or relative growth between the new gaps and the forest. The results of this study suggest that if adult stems survive the lightning strike disturbance event (New gap adults), in the short term, then there is little difference in the forest dynamic compared to close canopy settings. Not surprisingly the influence of lightning-initiated gaps is greatest at the other life stages.

Growing gaps versus intact forest

I found significantly less propagules in the growing gaps compared to the intact forest, specifically, fewer R. mangle propagules (Table 11). These sites were full of sapling size stems (Table 1), which previously have been reported as capable of seed production (Tomlinson 1986b). However, no observation of reproduction was ever found in the growing gap sites. Presumably these areas are under stress (low light and high density of saplings) and the plants do not have the resources to flower and fruit. Seedling recruitment was lower in growing gaps than the forest sites, specifically; I found no A. germinans recruitment and low (but not significant) R. mangle recruitment compared to the forest (Table 11). The R. mangle seedling population was decreasing in both forest types. Mortality was 10 times greater than recruitment in the growing gaps and 7 times greater in the forest. The survival of R. mangle seedlings and saplings was lower in the growing gaps compared to the forest based on density and the specific mortality rate, and was at least two times greater in the growing gaps. There was no other difference between the forest and the growing gaps based on density. Additionally, A. germinans sapling mortality rate was greater in the growing gaps compared to the forest.

As far as I am aware, this is the first study of gaps in which advanced stages of regeneration have been monitored separately for survival, growth and recruitment in the mangroves. Albeit, the study only covers one field year, I still found distinct differences in recruitment, survivorship and growth by species and life stages, suggesting that this type of approach is appropriate and insightful when trying to determine how these populations parameters change during gap phase succession.

A study of long-term recovery from hurricane disturbance, in terrestrial forest, reports similar findings of minimal sapling growth and increased mortality six years post-hurricane disturbance at the “building phase of forest development” (Vandermeer and Cerda 2004). The results from my study have allowed confirmation of the presumed reduction in seedling recruitment, and increase mortality of saplings as the gaps fill (Brokaw 1985, Duke 2001). Eventually the sites will begin to thin and return to intact forest densities and biomass (Brokaw 1985, Duke 2001).