NOTE: A later version of this document was published as:

Ross, M.S., P.L. Ruiz, G.J. Telesnicki, and J.F. Meeder. 2001. Aboveground Biomass and Production in Mangrove Communities of Biscayne National Park, Florida (USA). Wetlands Ecology and Management 9:27-37.

Aboveground Biomass and Production in Mangrove Communities of Biscayne National Park, Florida (USA), Following Hurricane Andrew

Report to The South Florida Water Management District
Michael S. Ross, Pablo L Ruiz, Guy J. Telesnicki, & John F. Meeder
June 25, 1998
Southeast Environmental Research Program, Florida International University, Univeristy Park, OE-148, Miami, Fl., 33199
Florida International University, Southeast Environmental Research Program, Copyright © 1998. All rights reserved.



Study Area: 



  • Biomass:
  • Production:
  • Discussion:  Conclusion: 

    Literature Cited:

    Abstract:  The dramatic transition in forest structure that characterized the pre-hurricane mangrove zone in much of southern Biscayne Bay, Florida, USA, was entirely disrupted by Hurricane Andrew (August, 1992).  Tall Fringe forests adjacent to the coast were leveled by intense winds which reduced the total aboveground biomass from 200 to about 1 Mg per hectare, while Dwarf forest communities a few hundred meters inland were relatively unaffected. Five years later, the initial structural gradient was becoming reestablished, with biomass in the Fringe forest reaching 56 Mg/ha and the Dwarf forest remaining below 20 Mg/ ha. In 1996-97, annual aboveground production in the Fringe forest was 3.5 times greater than in the Dwarf forest (29.8 and 8.6 Mg/ha/yr, respectively). This contrast in production between the two forest types was not evenly distributed among woody and photosynthetic tissues; Fringe production rates were approximately two, six, seven, and nine times higher for leaves, stems, branches, and proproots, respectively.  Moreover, mean leaf longevity did not differ between forest types (p > 0.50).  The above estimates for woody tissues were calculated using a standard dimensional analysis method.  Because of the difficulty associated with litter collection in a regularly inundated tidal shrubland, a new demographically-based procedure, adapted from Dai and Weigert (1996), was introduced for the purpose of  estimating leaf production. These methods may prove effective in estimating production in other terrestrial ecosystems, especially those in which shrubs are the dominant growth form. Our study calls attention to the dynamic nature of community structure in hurricane- prone mangrove ecosystems, especially the size and distribution of aboveground biomass components.
    Introduction:  Many ideas have been advanced regarding the principal causes of variation in the biomass or productivity of mangrove ecosystems.  A complex of proximate factors which may be involved include salinity (Cintron et al. 1978), salinity variation (Ball 1988; Lin and Sternberg 1992), accumulation of sulfides or other toxic substances (McKee 1993), soil redox conditions (Pezeshki et al. 1997), and nutrient limitation (Feller 1995; Boto and Wellington 1983).  Efforts to test the importance of alternative environmental variables on the functioning of mangrove ecosystems require precise measurements of community response, including biomass and production.

        If mangrove production is controlled by an interrelated complex of belowground resources and/or stresses, as the studies cited above suggest, then the same factors may alter patterns of allocation to leaves, woody tissues or roots.  In this regard, Tilman's (1988) model of competition among size-structured plant populations provides a useful conceptual background for considering differences between forests of fertile and infertile sites. The model predicts that as the availability of belowground resources increases, species or genotypes that allocate more of their production to stem tissues are favored in competition over those that allocate more to leaves or roots. While in situ root production is very difficult to measure accurately in mangrove sediments composed largely of dead root material, estimating the production of leaves and aboveground woody tissues present less intractable problems.

        In their classification of mangrove ecosystems, Lugo and Snedaker (1974) identified five forest types on the basis of physiognomy and physiographic setting.  Locations where these types intergrade are excellent places to examine the nature of variation in mangrove forest production or stature.  In southeast peninsular Florida and the Florida Keys, relatively tall coastal mangroves (Fringe forest) are often found adjacent to interior basins or flats characterized by stunted individuals (Dwarf forest).  Fringe and Dwarf mangrove forests have been examined at a number of sites within and outside of South Florida, and some of these studies have included estimates of both biomass and production (Day et al. 1987; Day et al. 1996; Teas 1979). However, no single study has compared the aboveground biomass and annual productivity of these two forest types by identical methods on adjacent sites.

        The mangrove swamps of Biscayne National Park were among the first ecosystems intercepted by Hurricane Andrew in its passage across the South Florida mainland in August 1992.  Prior to the hurricane, the coastal vegetation assumed a very steep gradient in stature with respect to distance from the coast, with the height of dominant stems decreasing from about 20 meters near the shoreline to less than 1 meter within about 500 meters.  The coastal and interior ends of this structural gradient may therefore be described as Fringe forest and Dwarf forest, respectively.

        In 1993, we initiated a study of these two very different mangrove forests.  In the research described in this paper, our objective was to compare the production and allocation of aboveground biomass in these two forests. We expected that production would be higher in the Fringe forest, and that a smaller proportion of it would go toward leaves than toward structural tissues (stems, branches, and proproots). To assess the leaf production of these two structural types on an equal footing, we developed a method which combined standard dimension analysis with observations of leaf demography, thereby avoiding problems associated with extrapolating from litterfall estimates.  Comparisons between forest types were complicated by  the uneven impacts of the hurricane on Fringe and Dwarf forest, because both biomass and production vary with stand age (Peet 1981; Sprugel 1985).  Better understanding of these successional relationships is especially important for mangrove forests, which are periodically exposed to a variety of partial or stand-initiating disturbances, including tropical storms and freeze events in northern portions of their distribution (e.g., Craighead 1964; Olmsted et al. 1993).

    Study Area:  The study area, located in the southern Biscayne Bay watershed about 1 km north of Convoy Point (Lat 25o27' N, Lon 80o20' W), is a thin coastal strip, about 2 km long and 700 meters in width (Figure 1). It is semi-impounded on the west, north and south by the levees of the L-31E, Military, and Mowry Canals, respectively, and divided by old drainage ditches into five rectangular units of 25-30 hectares each.  Three distinct vegetation units are recognizable as successive bands parallel to Biscayne Bay. The Fringe forest occupies a 100-200 meter zone closest to the coast, the Transition forest a 50-300 meter band further toward the interior, and the Dwarf forest the remainder of the mangrove swamp east of the L-31E levee.  All three units contain a mixed-species assemblage of Rhizophora mangle L. (red mangrove), Laguncularia racemosa (L.) Gaertn. (white mangrove), and Avicennia germinans (L.) Stearn. (black mangrove), but R. mangle is usually dominant in the Fringe and Dwarf forests, while L. racemosa is most abundant in the Transitional forest. The pronounced gradient in structure and productivity which characterized this zonation pattern is described below. The three units differ fundamentally in their hydrology.  The Fringe forest is inundated regularly by the semidiurnal tide,  the Transition forest is inundated less regularly but drained rapidly, and the Dwarf forest is in a basinal setting that is reached less frequently by tides but remains inundated for longer consecutive periods of time. Tidal amplitude in western Biscayne Bay is generally 0.2 - 0.3 meters.

    Methods:  The data reported in this paper were collected from 1993 through 1997, in conjunction with a broader experimental assessment of the ecological effects of fresh water diversion from the L-31E drainage canal into the coastal mangroves.  For the most part, they are representative of conditions during the pre-treatment period.  During September-October 1997, a small amount of water (ca 21 acre-feet) was diverted into Block 1 (Figure 1), but this release was not considered of sufficient magnitude to have elicited a measurable growth response prior to our final census in December 1997.

        Difficulty of access to the Convoy Point mangrove forest limited the intensity of vegetation sampling that was practical immediately after Hurricane Andrew (August 1992), but more intensive methods became feasible after the first year. Thus, we used one sampling method to estimate mangrove aboveground biomass before and soon after Hurricane Andrew, and a second to obtain biomass estimates for 1995, 1996, and 1997. The two methodologies are described separately below, along with the procedures for estimating productivity during 1996 and 1997.

        Biomass prior to Hurricane Andrew, and in 1993.  During July-August of 1993 we established two 120-meter transects in the Fringe forest zone, beginning at the Biscayne Bay shoreline and extending perpendicular to it towards the L-31E levee (Figure 1). Both transects were located within about 100 meters of the permanent plots described below, and were similar to them in post-hurricane vegetation structure. Sampling points were established at 20-meter intervals along each transect. A point-quarter method was used to estimate densities of live R. mangle, L. racemosa, and A. germinans in three height classes (0-60 cm, 60-250 cm, and >250 cm), as well as hurricane-killed trees (>5 cm DBH) of each species. Densities were calculated from distances to the base of the nearest stem of each category in four quadrants (Bonham 1989), to a maximum distance of 5 meters. Stem diameter (at 30 cm) and crown diameter of all sampled live stems were measured, and means for each species-size category at each point were calculated. Aboveground biomass for these mean individuals were estimated from regression equations developed as described below, and multiplied by density to provide point estimates of biomass per unit area. The DBH of dead stems was measured, and a mean for each point calculated. Pre- hurricane biomass of these large stems was estimated from diameter-based regression equations provided by Fromard (pers. comm.) for R. mangle, A. germinans, and L. racemosa in French Guiana (Fromard et al.1998). Live aboveground biomass in 1993 was calculated as the mean of twelve point estimates, and pre-hurricane (1992) tree biomass from the mean of point estimates based on dead stems plus living stems > 2.5 meters height.

        Biomass, 1995-1997.   In April-August 1995, permanent plots were established for the estimation of mangrove biomass in two of the five blocks which divide the study area (Figure 1). Seven plots were in Dwarf forest (four in Block 1, three in Block 4), and eight were in the Fringe forest type (four in each block).  No plots were established in the Transitional forest.  Construction of a network of boardwalks allowed us to approach and work in the rectangular plots without serious disturbance to the fragile mangrove sediments. Plots were 0.5 meter wide, and ranged from 3 to 10 meters in length, with smaller plots in areas of higher stem density. Each mangrove individual rooted within the plots was tagged and given a grid coordinate, and the following measurements were recorded: total height (TOTHT), height to base of crown (CRWNHT), crown length and width (CRWNL and CRWNW), and diameter at 30 cm (D30) and 140 cm (D140) (for stems taller than 40 and 150 cm, respectively). The product [CRWNHT * CRWNL * CRWNW] provided an index of crown volume (CRWNV). Measurements were repeated during the periods September - December 1996 and November - December 1997.

        Biomass regressions were developed for R. mangle, L. racemosa, and A. germinans in the summer of 1997. Regressions were based on relationships between the structural measures listed above and biomass components (trunk, branch, proproot, leaf, and total aboveground biomass) of individuals harvested from adjacent areas which resembled the study plots in vegetation structure. The regression data set included 53 R. mangle, 44 L. racemosa, and 21 A. germinans stems.  These individuals encompassed a range of sizes from 40 cm to 6.5 meters in height, which were as tall as the largest trees in the permanent plots.  5-14 individuals < 40 cm height of each species were also sampled and weighed. All possible regression models of the form

     ln (biomass component) = b0 + ba*ln (independent variable a) + ... bz*ln (ind. var z)
    were examined, where the independent variables were (D30), TOTHT, CRWND (i.e., mean of CRWNL and CRWNW), CRWNV, and [(D30)2*TOTHT]. A single model form was chosen for each species-tissue combination, based on variance explained and the distribution of residuals. 2- parameter models were selected over more heavily-parameterized ones when the latter did not substantially improve model performance.

        For trees in the permanent plots, biomass of stems taller than 40 cm was estimated by applying the regression models to annual structural measurements. According to Baskerville (1972), if  µ-hat = the estimate of Ln(Y), and s2-hat = the sample variance of the logarithmic equation, then the estimated value of Y in arithmentic units is:

         Individuals 40 cm or less in height were assigned stem, branch, proproot, and leaf biomass equal to the mean of the sample of small stems of each species.  Biomass components were calculated on an areal basis for each plot, and Dwarf and Fringe forest data were grouped without regard to Block.

        A localized freeze in January 1996 caused extensive mortality and dieback in two plots in the Dwarf forest portion of Block 4. Damage in the other plots was negligible.  To avoid obscuring more general trends evident in the undisturbed vegetation, we eliminated the two freeze-affected plots from the current analysis of biomass. Periodic freezes are ecologically important events in South Florida (Olmsted et al. 1993; Duever et al. 1994); recovery from the January 1996 freeze will be described elsewhere.

        Production in 1996-97.  We used data from twelve permanent plots (four in Dwarf and eight in Fringe forest) to estimate aboveground production of mangroves for the year beginning July 1, 1996. In forest communities, aboveground production  consists of the growth of woody tissues and leaf production. Stem, branch, and proproot production were calculated as the sum of estimated increases in biomass for surviving individuals, plus component biomass of newly established individuals.   Biomass components for each census date were calculated as described earlier, and a linear interpolation between successive census dates was used to estimate biomass on July 1, 1996 and June 30, 1997.

        Estimation of leaf production relied in part on observations of leaf dynamics at ca 4-month intervals through a two-year period. The demographic study was initiated in December 1995.  In the Dwarf forest, leaf production and survival were monitored on all branches of nine canopy R. mangle individuals (two or three in each of the four permanent plots in Block 1). In the Fringe forest, the fate of leaf cohorts was tracked on five branches representing different strata within the canopy of eight large R. mangle stems (one per permanent plot in both blocks). During the first survey, a loop of thin, colored wire was tied between the distal and second-to-distal leaf pair on all shoot tips, and all leaves including the distal ones were counted on each shoot. At subsequent surveys (March, July, and December 1996; May, September, and December 1997; and April 1998), leaves of previously circumscribed cohorts were counted, and a new cohort marked in similar fashion.  New branches were incorporated in the sampling design as they formed. On Dwarf forest trees, the demography of leaves on all new branches was monitored.  In the Fringe forest, where branch production was much more prolific, one new branch per sampling period was randomly selected for monitoring.  By counting and grouping all branches according to period of initiaion, it was possible to scale up observations on representative branches to the tree as a whole.

        In combination with data from the annual structural censuses, leaf demographic observations were employed to estimate the annual biomass production of mangrove leaves. The method was based on the balance between leaf production, senescence, and standing crop:

    Leaf Standing CropYr (x+1) = Leaf Standing CropYr (x) + Leaf Prod'nyr (x+1) - Leaf MortalityYr (x+1)
    and therefore,
     Equation 1: Leaf Prod'n Yr(x+1) =   Standing Crop Yr (x®x+1) + Leaf Mortality Yr (x+1)

    A major assumption of our method was that the production and mortality rates underlying the first and third terms in Equation 1 were similar whether expressed on a density or biomass basis. In solving Equation 1 for biomass,   Standing CropYr (x®x+1) was calculated by applying our biomass regression equations to structural data from censuses of successive years in each plot. Leaf Mortality Yr (x+1)  was calculated as follows:

     Equation 2: Leaf Mortality Yr (x+1)= Tbiom.Yr (x+1) * Biom Yr (x+1)

     Biomass.  The regressions equations used to calculate biomass components for the three mangrove species are listed in Table 1. For stem biomass, we chose a two-parameter model in which (D30)2*TOTHT was the single independent variable. For proproot biomass, total height was the lone independent variable in the selected model.  For branch, leaf, and total biomass, regression models which included crown volume and diameter as independent variables were selected. Coefficients of determination were highest for stem biomass, and exceeded 0.90 for all species-tissue combinations except R. mangle proproot biomass (R2=0.74). While these models generally provided a better fit than models based on stem diameter alone, models based on the latter may provide useful biomass estimates for studies in which height or crown dimensions were not measured. Regression equations for models predicting biomass components from  (D30)2 are listed in Appendix 1.

        Mortality from Hurricane Andrew in the Fringe sites was near-complete, reducing total aboveground biomass in this tall (18-22 meters) forest from nearly 200 Mg (1 Mg = 1000 Kg) to about 1 Mg per hectare (Figure 2).  Recruitment of new seedlings was patchy during the next few years, but by the winter of 1995 most of the site had been reoccupied by a dense stand of mangrove saplings. Subsequent stand development was rapid. By the winter of 1997, five years after Hurricane Andrew, aboveground biomass averaged more than 55 Mg per hectare (Figure 2). 64% was accounted for by woody tissues, 14% by leaves, and 22% by proproots (Table 2). R. mangle comprised about 90% of stand biomass, but localized areas were dominated by L. racemosa (8%). Canopy height also exhibited wide local variation, ranging from about 3 to 6 meters.

        No quantitative estimates of pre-hurricane biomass or hurricane-related mortality were available from the Dwarf forest, where sampling began in 1995.  However, photographs taken during reconnaissance of the area soon after Hurricane Andrew indicated that structural damage due to the storm was minor, and that most mangroves retained their leaves, probably because they were under water during the period of high winds. Compared to the rapidly developing Fringe forest, the increase in Dwarf forest biomass from 14 to 18 Mg between 1995 and 1997 was modest (Figure 2). Compartmentalization of biomass was similar to the Fringe forest, with slightly higher percentages of total aboveground biomass  in leaves (16%) and stems/branches (66%), and slightly lower percentages (18%) in proproots (Table 2). As in the Fringe forest, Dwarf plots were dominated by R. mangle on a biomass basis.

     Production.  Mean red mangrove leaf longevities ranged from about 139 days (Dwarf forest, April 1996 cohort) to 232 days (Dwarf forest, September 1996 cohort) (Figure 3). Repeated measures analysis of variance indicated that leaf life spans did not differ between forest types (F=0.01, p=0.91), but did differ among cohorts (F=56.52, p < 0.001).  Post-hoc multiple comparison testing demonstrated significant differences between all pairs of cohorts (Scheff  test, p   0.024), with leaf longevities arranged in the order April 1996 < February 1997 < September 1996. The longevities illustrated in Figure 3 equate to turnover rates (Tdens) of 1.7 to 2.9 generations per year.

        Total aboveground production in the Fringe forest (29.8 Mg/ha/yr) was about 3.5 times higher than Dwarf forest production (8.6  Kg/ha/yr) (Table 3).  This difference was accompanied by an equally striking contrast in tissue allocation, i.e., leaves constituted about 76% of aboveground production in the Dwarf community, but only 46% in Fringe. This pattern was a consequence of more rapid production of woody tissues in the Fringe forest, not lower production of leaves. For instance, compared to the Dwarf forest, Fringe production rates were approximately two, six, seven, and nine time higher for leaves, stems, branches, and proproots, respectively (Table 3).  The comparatively modest response of leaf production reflected the site- insensitivity of leaf turnover rate (Figure 3).


    Total aboveground biomass and production.  The results presented above call attention to the dynamic nature of community structure in hurricane-prone mangrove ecosystems, especially the size and distribution of aboveground biomass components.  Successional changes in standing crop biomass are accompanied by changes in measured production, though maximum production is achieved earlier in stand development than is maximum biomass (Daniel et al. 1979; Sprugel 1985). In comparing the hurricane-damaged Convoy Point Fringe forest with the relatively unaffected Dwarf forest, it is important to bear in mind this contrast in the temporal functions of biomass and production.

        Biomass. We are aware of two studies in which the biomass of  mangrove communities similar in height (< 1.5 m) to our Dwarf forest were estimated. Lugo and Snedaker (1974) harvested leaves, stems, and proproots in three 9-m2 plots in another location in the Biscayne Bay watershed, and found a total aboveground biomass of 7.9 Mg/ha. Woodroffe (1985) found even lower aboveground biomass (6.8 Mg/ha) in an Avicennia-dominated Dwarf forest in New Zealand. Our Dwarf forest mean of 18.2 Mg/ha included plot estimates ranging from 9.7 to 31.8 Mg per hectare. Plots with higher biomass were generally located closer to the border with the Transitional forest, while low-biomass plots were farthest from the coast.  This gradient of decreasing biomass with distance from the coast within the Dwarf mangrove zone parallels the interzonal contrast between Fringe and Dwarf forest, and may be a widespread pattern in the Biscayne Bay watershed, if not more generally.  For example, along a transect 15 km south of Convoy Point, in which a continuous band of Dwarf mangrove vegetation < 1.5 meter in height stretched interiorward from the shoreline for several kilometers, total macrophyte cover decreased from about 40% adjacent to the coast to less than 10% within one kilometer (Ross et al. in prep.).

        Among mangrove studies reviewed by Saenger and Snedaker (1993), no research site above 10o north or south latititude exceeded 200 Mg/ha in total aboveground biomass. While mangrove forests of large stature are increasingly rare in Florida, biomass in mature Fringe or Riverine forest in southwest Florida averaged 170-180 Mg/ha, and exceeded 200 Mg/ha in individual plots (Lugo and Snedaker 1975). Based on our pre-hurricane biomass estimate in the Convoy Point Fringe forest (195 Mg/ha), and on the rapid rate of biomass accumulation following Hurricane Andrew (56 Mg/ha in five years), it seems likely that, given no major disturbance, a forest approaching 200 Mg/ha may develop within a few decades on productive sites in South Florida.  Certainly, the current 3:1 ratio in biomass between the Convoy Point Fringe and Dwarf forests greatly underestimates their ratio at maturity.

        Production. Our assessment of mangrove production at the two Convoy Point sites included separate estimates for aboveground woody tissues (including proproots) and leaves, but did not incorporate reproductive parts or belowground tissues. We used standard dimension analysis methods to estimate wood production (e.g., Whittaker and Woodwell 1968), but introduced a new, demographically-based procedure for estimating leaf production.

        Standard methods for estimating leaf production in mangrove or other forest ecosystems often involve litter collection (e.g., Day et al. 1987; Megonigal et al. 1997). However, effective litter sampling methods are difficult to devise in tidal shrublands, where tide waters sometimes rise well into the crowns of dominant stems. In its utilization of demographic parameters such as mean leaf longevity and turnover rate, our method of estimating leaf production resembled a non- destructive method recently applied by Dai and Wiegert (1996) in Spartina alterniflora marsh. In their study, leaf demography of all stems within nine small (0.25 - 1 m2) quadrats was monitored at short intervals over a 21-month period, and annual leaf production was calculated as:

     mean stem density * mean leaf number per stem * mean leaf turnover * mean leaf mass
        In contrast, we calculated leaf turnover rates from a small sample of mangrove individuals, then multiplying this rate by the mean standing crop of leaves in our large (1.5 - 5 m2) plots, assuming that a density-based turnover rate (leaves*leaf-1*year-1) was equal to a biomass-based turnover rate (grams *grams-1*year-1). This assumption should be valid if leaves grow to reach a fixed size range before senescing, and this size distribution does not change substantially within the year. We have made a few observations which suggest these assumptions are roughly true on our sites. Another assumption of our method is that the leaf longevity patterns exhibited by the sampled individuals are representative of the community as a whole.  Since our demographic surveys were restricted to R. mangle, this assumption is unlikely to be true.   However, in light of the dominance of R. mangle in these forests (96 and 90% of total biomass in Fringe and Dwarf forest, respectively), application of these data to L. racemosa and A. germinans shouldn't alter community production estimates substantially.

        Our estimates of  aboveground biomass production were near the high end of the range of values reported from similar mangrove forests.  Convoy Point Fringe production of 29.8 Mg/ha/yr exceeded the 16.0 Mg/ha/yr and 24.6 Mg/ha/yr found by Day et al. (1987) in adjacent Fringe and Riverine forests in Mexico.  In an A. germinans-dominated Basin forest near the above sites,  annual aboveground production over a seven-year period was lowest in mid-basin (3.2 - 4.8 Mg/ha/yr) and highest near the Fringe (6.6 - 7.6 Mg/ha/yr) (Day et al. 1996); measured production in the Convoy Point Dwarf forest (8.6 Mg/ha/yr) was slightly above this range.  Teas (1979) calculated production rates of 32.1 and 3.8 Mg/ha/yr in adjacent Fringe and Dwarf forest sites in South Florida, using an expansion factor of 3x to estimate NPP from litterfall measurements. It is not possible to determine how much of the variation among these studies is attributable to: (1) methodological differences, (2) inter-annual variation associated with climatic or hydrologic conditions, (3) fundamental differences in site potential,  and (4) differences in the developmental stage of the mangrove community. With regard to the last source of variation,  most models of successional change in forest production following disturbance include an early reorganization period during which stand production increases steadily to a maximum level (Peet 1981; Sprugel 1985). Though this recovery period sometimes lasts less than a decade (Marks 1974), at five years old the Convoy Point Fringe forest may not yet have passed entirely through it. If it has not, the contrast in production between the Fringe and Dwarf sites, demonstrated through identical methods and over the same period, may be expected to increase slightly from its current ratio of 3.5:1.

    Allocation and partitioning of aboveground biomass.  The partitioning of annual biomass production among leaf and structural tissues supported our initial expectation: leaf production was less responsive to the difference in site quality between our two stands than was the production of woody tissues, and therefore comprised a smaller proportion of total aboveground production in the Fringe than the Dwarf forest (Table 3). The pattern is further illustrated through a consideration of the 17 red mangrove trees whose leaf demographies were monitored in the two areas. During the period of our study, Fringe and Dwarf forests differed enormously in the initiation and extension growth of stems, branches, and proproots. Between March 1996 and April 1998, leaf demography trees in the Fringe site initiated an average of 96.3 primary, secondary, or tertiary branches, which reached a mean length of 25.7 cm. Comparable figures for the nine trees monitored in the Dwarf were 1.4 branches, with mean length of 8.2 cm. While the total number of leaves produced per branch was also higher in Fringe than Dwarf forest (11.7 and 10.1, respectively), the number produced per cm of branch length was much less (0.5 in Fringe, 1.4 in Dwarf).  Because leaf longevity did not differ between sites (Figure 3), the result was a more diffuse arrangement of leaves in the Fringe forest. The absence of site effects on the longevity of R. mangle leaves has previously been observed in Belizean mangal (Ellison and Farnsworth 1996), though mean longevities in the latter study (ca nine months) were higher than the 4-8 months observed at Convoy Point.

        The site differences in biomass allocation discussed above were not evident in the standing structure of the Fringe and Dwarf forests, which did not differ substantially in the relative proportions of leaf and woody tissues (Table 2).   This discrepancy between allocation and structure is probably due to the contrast in the developmental stages of the two forests.  Because of their rapid turnover rates in comparison to woody tissues, leaves become a smaller proportion of total biomass as stand development proceeds following disturbance (Sprugel 1985).  Fromard et al. (1998) emphasized the effect of stand age on mangrove structural characteristics, including biomass partitioning. In their youngest Guianese mangrove stands (ages 3-5 yrs), leaves accounted for 6-10% of total aboveground biomass, but by the time these forests reached structural maturity, this percentage had decreased to less than 3%, despite an absolute increase in leaf biomass. A similar decrease in the proportion of leaf biomass may already have begun in the five year-old Fringe forest at Convoy Point, where crown closure has in places resulted in the death of all leaves below about 2.5 meters height.  In years to come, the ratio of leaves to woody biomass is likely to decrease more precipitously in the very young Fringe forest than in the more mature Dwarf forest, eventually reflecting the prevailing patterns in annual allocation between the two stands.

        When Rhizophora species are important components of stand composition, proproots may also constitute a substantial proportion of mangrove forest biomass.  Our values of 22 and 18% of aboveground biomass in Fringe and Dwarf forest, respectively, are not exceptionally high in comparison to other reports.  In their early review of mangrove ecology, Lugo and Snedaker (1974) summarized biomass partitioning from thirteen forest sites. Among the ten reports in which proproot biomass was listed separately, it constituted 2-42% (mean 26%) of total aboveground biomass.  Proproots comprised a mean of 16% (range 4-47%) of the biomass of individual R. apiculata stems in Malaysia (Gong and Ong 1990).

    Conclusion:  The contrast in productivity between mangrove forests occupying the ocean fringes vs. neighboring communities inhabiting interior basins has been observed and discussed by many researchers (e.g., Davis 1940, Lugo and Snedaker 1974; Twilley 1998).  In a sense, these analyses are based on an infrequently tested assumption, i.e., that existing variation in forest structure may be equated to variation in the rate at which biomass is produced. Our results generally confirm the validity of this assumption for the two sites we examined, with both biomass and production elevated by 3-4 times in Fringe forest relative to Dwarf forest.  However, based on the trajectories we expect these variables to follow as stand development proceeds, the production rates of these two ecosystem types may not be nearly as different as their respective standing crops at maturity. Moreover, mature Fringe and Dwarf forests are expected to differ more in woody tissue biomass than in leaf biomass, in line with the strong contrast in how production is allocated in the two forest types.

    Literature Cited

    Ball, M.  1988.  Ecophysiology of mangroves.  Trees 2:  129-142.

    Baskerville, G. L. 1972. Use of logarithmic regression in the estimation of plant biomass. Canadian Journal of Forest Research. 2: 49-53.

    Bonham, C. D. 1989.  Measurements for terrestrial vegetation. Wiley, New York, USA. 338 pp.

    Boto, K. G. and J. T. Wellington. 1983. Phosphorus and nitrogen nutritional status of a northern Australian mangrove forest. Marine Ecology and Progress Series 11: 63-69

    Cintron, G., A. E. Lugo, D. J. Pool, and G. Morris.  1978.  Mangroves of arid environments in Puerto Rico and adjacent islands.  Biotropica 10:  110-121.

    Dai, T. and Wiegert, R. G. 1996. Ramet population dynamics and net aerial primary productivity of Spartina alterniflora. Ecology 77(1): 276-288.

    Davis, J. H. Jr.  1940.  The ecology and geologic role of mangroves in Florida.  Carnegie Institute of Washington 517: 303-412.

    Day, J. W., Jr., W. H. Conner, F. Ley-Lou, R. H. Day, and A. M. Navarro. 1987. The productivity and composition of mangrove forests, Laguna de Terminos, Mexico. Aquatic Botany 27:  267-284.

    Day, J. W., Jr., C. Coronado-Molina, F. R. Vera-Herrera, R. Twilley, V. H. Rivera-Monroy, H. Alvarez-Guillen, R. Day, and W. Conner.  1996. A 7-year record of above-ground net primary production in a southeastern Mexican mangrove forest.  Aquatic Botany 55: 39- 60.

    Daniel, T. W., J. A. Helms, and F. S. Baker. 1979. Principles of silviculture. McGraw-Hill, New York. 500 pp.

    Duever, M. J.,  J. F. Meeder, L. C. Meeder, and J. M. McCollom.  1994.  The climate of South Florida and its role in shaping the Everglades ecosystem. pp. 225-248 In:  S. M. Davis and J. C. Ogden (eds.), The Everglades: the ecosystem and its restoration.  St. Lucie Press, Delray Beach, FL.

    Ellison, A. M. and E. J. Farnsworth.  1996.  Spatial and temporal variability in growth of Rhizophora mangle saplings on coral cays:  links with variation in insolation, herbivory, and local sedimentation rate.  Journal of Ecology  84: 717-731.

    Feller, I. C. 1995. Effects of nutrient enrichment on growth and herbivory of dwarf red mangrove (Rhizophora mangle). Ecological Monographs 65: 477-505.

    Fromard, F., H. Puig, E. Mougin, G. Marty, J. L. Betoulle, and L. Cadamuro. 1998. Structure, above-ground biomass and dynamics of mangrove ecosystems: new data from French Guiana. Oecologia 115: 39-53.

    Gong, W. and J. Ong.  1990.  Plant biomass and nutrient flux in a managed mangrove forest in Malaysia.  Estuarine, Coastal and Shelf Science 31: 519-530.

    Lin, G. and L. da S. L. Sternberg.  1992.  Differences in morphology, photosynthesis, and carbon isotope ratios between scrub and fringe mangroves.  Aquatic Botany 42:  303-313.

    Lugo, A. E. and S. C. Snedaker. 1974. The ecology of mangroves. pp. 39-64 in Annual Review of Ecology and Systematics, Volume 5: 39-64.

    Lugo, A. E. and S. C. Snedaker. 1975. Properties of a mangrove forest in southern Florida. pp 170-212 in G. E. Walsh, S. C. Snedaker, and H. J. Teas, eds. Proceedings of the international symposium on the biological management of mangroves.  Institute of Food and Agricultural Sciences, University of Florida, Gainesville, FL, USA.  846 pp.

    Marks, P. L. 1974.  The role of pin cherry (Prunus pennsylvanica L.) in the maintenance of stability in northern hardwood ecosystems.  Ecological Monographs 44: 73-88.

    McArdle, R. E., W. H. Meyer, and D. Bruce. 1949. The yield of Douglas-fir in the Pacific Northwest. USDA Forest Service Technical Bulletin 201.

    McKee, K. L. 1993.  Soil physicochemical patterns and mangrove species distribution --- reciprocal effects?  Journal of Ecology 81:  477-487.

    Megonigal, J. P., W. H. Conner, S. Kroeger, and R. R. Sharitz.  1997.  Aboveground production in southeastern floodplain forests: a test of the subsidy-stress hypothesis.  Ecology 78:  370-384.

    Olmsted, I., H. Dunevitz, and W. J. Platt. 1993. Effects of freezes on tropical trees in Everglades National Park, Florida, USA. Tropical Ecology 34(1): 17-34.

    Peet, R. K. 1981. Changes in biomass and production during secondary forest succession. pp 324- 338 In: Forest succession: concepts and applications.  Springer-Verlag, New York. 517 p.

    Pezeshki, S. R., R. D. DeLaune, and J. F. Meeder.  1997.  Carbon assimilation and biomass partitioning in Avicennia germinans and Rhizophora mangle seedlings in response to soil redox conditions. Environmental and Experimental Botany 37: 161-171.

    Saenger, P. and S. C. Snedaker. 1993. Pantropical trends in mangrove above-ground biomass and annual litterfall. Oecologia 96: 293-299.

    Sprugel, D. G.  1985.  Natural disturbance and ecosystem energetics.  pp. 335-352 In:  S. T. A. Pickett and P. S. White, eds. The ecology of natural disturbance and patch dynamics.  Academic Press, Orlando, USA. 472 p.

    Stewart-Oaten, A., W. W. Murdoch, and K. R. Parker. 1986.  Environmental impact assessment: "pseudoreplication" in time?  Ecology 67: 929-940.

    Teas, H. J.  1979.  Silviculture with saline water.  pp. 117-161 In:  A. Hollaender, J. C. Aller, E. Epstein, A. S. Pietro, and O. R. Zaborsky, eds.  The biosaline concept.  Plenum Press, New York.

    Twilley, R. R.  1998.  Mangrove wetlands.  pp. 445-473 In: M. G. Messina and W. H. Conner, eds. Southern forested wetlands.  Lewis Publishers, Boca Raton, FL.

    Whittaker, R. H. and G. M. Woodwell. 1968. Dimension and production relations of trees and shrubs in the Brookhaven Forest, New York. Journal of Ecology 56: 1-25.

    Woodroffe, C. D. 1985. Studies of a mangrove basin, Tuff Crater, New Zealand: I. Mangrove biomass and production of detritus. Estuarine and Coastal Shelf Science 20: 265-280.