Methods

Study area

This study examined the mangroves on the southwest coast of Florida, in Everglades National Park, (25°N - 26°N). This area encompasses approximately 60,000 ha of mangrove forest (Figure 1). These mangroves form a continuous band along the coast, varying in width from 0.1 to 15 km (Smith 1994). Tree height generally declines with distance from the coast (Chen & Twilley 1999). The climate is subtropical with an average maximum temperature above 27 °C. Rainfall has a distinct dry and wet season and varied from 86 to 224 cm over a 10-year period (Duever 1994). Tidal amplitude fluctuated from 10 to 60 cm. Three mangrove species [Rhizophora mangle L. (red mangrove), Laguncularia racemosa (L.) Gaertn. (white mangrove), and Avicennia germinans (L.) Stearn (black mangrove)] are found in this area varying from heterogeneous mixed stands to single species dominated forest. Hurricane disturbance to this region is common, with catastrophic hurricanes occurring approximately every thirty years (Doyle 1997). The 1935 “Labor Day” Hurricane, Hurricane Donna (1960) and Hurricane Andrew (1992) all strongly impacted the Everglades mangrove region. Additionally, lightning-initiated canopy gaps are common in this forest (Chapter I, Houston and Powell 2003, Smith et al. 1994).

Figure 1 Location of three sites on the lower Shark River, Everglades National Park, Florida, USA. Open circles represent new gaps, gray circles growing gaps and dark circles are intact forest locations. More detailed view is shown in insert of Site 2. [larger image]

I grouped gaps into two ages: (1) three new gaps that were known to have been initiated by lighting between July to September of 2002; and (2) growing gaps, approximately seven to 15 yrs old, generally with a very dense sapling layer. These categories correspond to the following stages within Duke's (2000) small gap mangrove conceptual model: gap initiation combined with gap opening and gap filling, respectively. The growing gap category was assigned based on a subset of gaps of known approximate age (pers. obs. K. Whelan). To compare the community attributes of the lightning-initiated gaps to the surrounding intact forest, I established three intact forest sites. Each of the three groups were comprised of one new gap, one growing gap and one intact forest location (Table 1). At the group location, all gaps of the time series were within 300 m of each other. In this paper, “forest stage” includes new, growing gaps and intact forest sites whereas the phrase “gap phase” only refers to comparisons between new and growing gaps.

Table 1. Site description. Size of gap, percent canopy openness, distance to main river, distance to rivulet, density of adults (4 cm dbh), density of saplings (< 4 m dbh), density of seedlings, biomass of trees and saplings combined (kg). All densities and biomass values are from 2004 survey year and are live stems standardized to per 500 m2. Site Stage Gap Size (m2) Percent canopy openness River (m) Rivulet (m) Canopy height (m) Adult Sapling Seedling Biomass (kg) 1 New Gap 265 17.16 22 NA 2.87 47 219 1687 1362 Growing 491 7.63 40 NA 14.11 87 259 406 3376 Forest NA 7.67 27 NA 17.19 45 109 969 4899 2 New Gap 437 19.34 14 20 NA 60 32 1187 4129 Growing 486 7.59 37   13.34 97 271 312 3212 Forest NA 7.66 20 25 18.30 109 40 1625 7634 3 New Gap 132 11.08 30 10 NA 85 15 844 5384 Growing 393 7.6 32 5 5.83 30 933 156 3122 Forest NA 8.47 40 20 21.01 57 52 1125 6888

Additionally, at each gap site the gap size (expanded gap size sensu Runkle 1982), canopy openness (determined by hemispherical photography), distance to river and rivulet and canopy height (mean of six dominant stems) were determined (Table 1). A circular plot (radius eight-m) was established in the center of each gap (site). The plot size was chosen to confine sampling within the canopy gap area. I used the size class definition of Koch (1987) and Chen and Twilley (1998) to ensure comparability of this work with previous studies in this forest. Propagules were not established or rooted in the substrate (i.e. still in a dispersal phase). Seedlings were defined as all individuals attached to the substrate and < 1.4 m in height. Saplings were defined as all stems > 1.4 m and < 4 cm in diameter at breast height (dbh). Adults were defined as all stems greater than 1.4 m in height and 4 cm dbh. Seedlings and propagules were counted in four 4 m² plots; nested within the circular plot. All stems were permanently and uniquely tagged, identified to species, and had a condition status assigned. Vegetation plots were established from February to May 2003 (from five to nine months post-strike) and resurveyed in March to April 2004. During the 2004 resurvey, all new stems were permanently and uniquely tagged, identified to species, and had a condition status assigned. Recruitment in this work was defined as the number of individuals that were new to the life stage at the second survey. For example, five seedlings that had sufficient stem growth to be included in the sapling stages (> 1.4 m height) were considered recruiting saplings. The same was true for seedlings and adult stems.

Seedlings were tagged and total stem length (soil surface to the dominant meristem) was measured. For R. mangle seedling, the stem length was measured from the top of the propagule scar to the bottom of the leaf sheath of the main meristem. Seedling elongation rates (E-Rate, mm/d) were calculated as (sensu Koch 1997):

E-Rate (mm d^-1) = (Stem length _(t1) - Stem length _(t0)) / (_(t1) - _(t0))

Seedling stem growth (cm yr^-1) was E-Rate * 365 days. Sapling and adult tree growth was determined by two successive measurements of dbh. I used the species-specific allometric formulas of Smith and Whelan (in review) to convert from dbh to living biomass. Growth was determined by converting dbh to biomass and determining the relative growth rate (sensu Evens 1977) as:

Relative Growth Rate (RGR) = Log (Biomass₂) - Log (Biomass₁) expressed on an annual basis.

Specific per capita recruitment (R) and mortality (M) rates for the sample period were calculated using the following equations (sensu Padilla et al. 2004):

R _t+1 = Ln{[(N_t + NR_t+1)/N_t]/t}
M _t+1 = Ln{[(N_t - D _t+1)/N_t]/t}

where R _t+1 is the specific recruitment rate annualized, and M _t+1 is the specific mortality rate annualized, N_t is the total population at initial survey, NR_t+1 is the number of new recruits at the second survey; D_t+1 is the number of dead stems at the second survey; t is the number of days between first and second survey divided by 365.

Data analysis

Abundance within the propagule class (count data) was transformed and followed with a parametric one-way way Analysis of Variance (ANOVA) (Quinn and Keough 2002, Zar 1999). Differences in initial seedling height, survival of seedling, and elongation rates were analyzed with a one-way and two-way ANOVA. Due to low sample size L. racemosa seedlings were removed from the elongation rate comparison. Logistic regression was used to determine the relationship between the probability of survival and species, initial seedling height, and forest stage.

For saplings and adult trees the initial dbh, change in dbh, annual change in biomass and relative growth rates were log₁₀ (x+0.5) transformed to increase normality and meet homogeneity of variance requirements for the two-way ANOVA test of forest stage and species. L. racemosa saplings were removed from the analysis of change in dbh, annual change in biomass and relative growth rates due to low sample size. Half normality probability plots were used to assess normality for the linear regressions. Normality plots were used to assess normality for parametric tests. Unless otherwise noted, I used Tukey's Honestly Significantly Difference test for unequal sample sizes for post- hoc comparison. I used a Arcsin transformation for proportional data. Analyses were performed using STATISTICA 5.0 (Statsoft, Inc., 1996), SPSS 11.0.1 (SPSS, Inc., 2001) and Statistix for Windows (96 Analytical Software, Inc).