BIOTROPICA 38(3): 365?374 2006 10.1111/j.1744-7429.2006.00156.x
Zonation Patterns of Belizean Offshore Mangrove Forests 41 Years
After a Catastrophic Hurricane1
Cyril Piou 2
Center for Tropical Marine Ecology, Fahrenheitstrasse 6, 23859 Bremen, Germany
Ilka C. Feller
Smithsonian Environmental Research Center, P.O. Box 28, Edgewater, MD 21037, U.S.A
Uta Berger
Center for Tropical Marine Ecology, Fahrenheitstrasse 6, 23859 Bremen, Germany
and
Faustino Chi
Institute of Marine Studies, University of Belize, P.O. Box 990, Belize City, Belize
ABSTRACT
Mangroves are prone to bearing frequently the full brunt of hurricanes and tropical storms. The extent of destruction and early regeneration are widely studied. The
purpose of this study was to add a long-term view of mangrove regeneration and assess the potential effects on mangrove horizontal zonation patterns of catastrophic
destruction. Hattie, a category five hurricane, hit the Belizean coast in 1961. It passed directly over the Turneffe Atoll where our study area, Calabash Cay, is located.
At four sites on this island, we analyzed mangrove forest structure along transects parallel to the shoreline within zones delineated by species dominance and tree
height. We propose an index based on the Simpson index of diversity to express changes in the heterogeneity of the species dominance. Physical?chemical parameters
and nutrient availability were also measured. The destruction levels were estimated by analysis of the distribution of diameter at breast heights of the bigger trees in
the inland zones. Variations in species dominance among sites and zones could be explained by interactions of various factors. Further, different levels of destruction
between the two sides of the island had a significant effect on current patterns of species and structural zonation at Calabash. We conclude that disturbance regime in
general should be considered as a factor potentially influencing mangrove horizontal zonation patterns.
Key words: Avicennia germinans; Belize; hurricane disturbances; Laguncularia racemosa; long-term regeneration; mangrove forest dynamics; Rhizophora mangle; species
dominance heterogeneity; Turneffe Atoll; zonation patterns.
MANGROVE FORESTS LINE MOST OF THE WORLD?S tropical and sub-
tropical coastlines. In this coastal position, mangroves bear the full
brunt of hurricanes and tropical storms, which are a frequent form
of disturbance in these latitudes. Tree mortality, tree species resis-
tance, and early forest recovery immediately following hurricane
destruction have been widely studied (e.g., Stoddart 1963, Vermeer
1963, Bardsley 1984, Roth 1992, Smith et al. 1994, Roth 1997,
Imbert et al. 1998, Baldwin et al. 2001, Sherman & Fahey 2001,
Imbert 2002). Rates of hurricane-related mortality in mangrove
forests are significantly higher than for any other tropical forests
(compared in Baldwin et al. 2001; see also Imbert et al. 1998).
Some recent studies have also analyzed the impacts of these destruc-
tions on ecosystem functions such as peat formation (Cahoon et al.
2003) or woody debris accumulation (Krauss et al. 2005) within
the decade following hurricanes. In contrast, few studies have ex-
amined long-term recovery patterns (e.g., Smith & Duke 1987).
In the studies carried out in the Caribbean, results of species re-
sistance to hurricanes are contradictory. For instance, according to
Roth, Smith, and Imbert (Roth 1992, Smith et al. 1994, Imbert
1Received 18 March 2005; revision accepted 1 August 2005.
2Corresponding author; e-mail: cyril.piou@zmt-bremen.de
et al. 1998, Imbert 2002), Rhizophora mangle is the least resistant to
hurricane destruction. However, Sherman and Fahey (2001) found
that Avicennia germinans is the least resistant. Regarding the rel-
ative importance of tree size, Roth (1992), Smith et al. (1994),
and Baldwin et al. (2001) found a higher probability of destruc-
tion for intermediate and higher size classes. Other studies have
shown either an inverse pattern (e.g., Imbert et al. 1998), or no
difference in damage among size classes (e.g., Sherman & Fahey
2001).
The extent of forest damage depends on the intensity of the
hurricane. Category one or two hurricanes on the Saffir?Simpson
Hurricane Scale (www.nhc.noaa.gov) should generally inflict less
damage than categories three?five. Nevertheless, the percentage of
dead trees can also vary depending on site exposure. For example,
Imbert (2002) and Sherman and Fahey (2001) found the highest
levels of disturbance in zones close to the shoreline (see also e.g.,
Roth 1992, Imbert et al. 1998). Baldwin et al. (2001) reported also
that the differences in levels of destruction at two sites in Florida
depended on their relative position within the path of the eye of
Hurricane Andrew. At Turneffe Atoll, Stoddart (1963) described
that the mangrove forests on many small windward islands were
completely destroyed by Hurricane Hattie, a category five storm
C? 2006 The Author(s)
Journal compilation C? 2006 by The Association for Tropical Biology and Conservation
365
366 Piou, Feller, Berger, and Chi
that struck Belize in October 1961, whereas inside the atoll the
vegetation was less damaged.
Delayed mortality is another source of variation in hurricane
damage. Some mangrove trees are able to coppice (e.g., A. germinans
and Laguncularia racemosa) and use their prehurricane reserves to
create new tissues and foliage (Tomlinson 1986). However, if these
trees are severely damaged, they do not live much more than 2 yr
after the hurricane (see e.g., Sherman & Fahey 2001). The delayed
mortality is not as high in mangroves as in other tropical forests
(Imbert et al. 1998). Furthermore, mangrove forests appear less
resistant but much more resilient than other tropical forests. Imbert
et al. (1998) suggested that no Caribbean mangrove species are
pioneers when compared to rainforests. However, mangrove species
apparently present different adaptations for recolonization strategies
and wind resistance (Roth 1992), relating the effects of hurricanes
to the mangrove dynamics and structure.
To our knowledge, no studies have yet documented the long-
term effects of hurricanes on the horizontal zonation patterns of
mangrove forests. Such patterns may refer to structural characteris-
tics (structural zonation) or species dominance (species zonation).
Structural zonation describes bands parallel to the shore differing in
tree density, canopy height, or tree diameters. Species zonation ex-
presses zones successively encountered from offshore to inland with
different monospecific composition or particular associations of few
tree species. Several hypotheses have been expressed to explain the
phenomena driving these patterns: plant succession (Davis 1940),
geomorphological factors (Thom 1967), differential dispersal of
propagules (Rabinovitz 1978), differential predation on propagules
(Smith 1987), interspecific competition (Ball 1980), and species-
specific physiological adaptations to gradients (Macnae 1968, for
review see Smith 1992). In Belize, differential predation on propag-
ules is not responsible for differences in species zonation (McKee
1995a). McKee (1993) found that physiological adaptations best
explain zonation patterns in Belizean mangroves. Together with
the species-specific physiological adaptations to gradients of salinity
and tidal influence, nutrient availability plays an important role in
Caribbean mangrove zonation (Feller et al. 2003).
The main objective of this study was to determine whether
perturbations such as hurricanes interact with other possible ex-
planatory phenomena mentioned above to have an effect on these
horizontal zonation patterns in mangrove ecosystems. The second
aim was to document the long-term effects of large-scale destruction
on forest structure and to propose some recovery pathway scenarios.
The offshore Belizean mangrove islands in Turneffe Atoll provided
an ideal location for these investigations since they were previously
destroyed by Hurricane Hattie more than 40 yr ago, and the oc-
currence of only four tree species provides a particular but relatively
simple forest system.
METHODS
STUDY AREA AND SITES.?This study was conducted at Calabash
Cay (also sometimes referred to as the ?Main Calabash Cay? of
a ?Calabash Cays? group), on the eastern side of Turneffe Atoll,
one of the three Belizean atolls (Fig. 1). It is located approximately
34 km east of the Mesoamerican Barrier Reef and about 50 km
from the mainland. Calabash Cay is almost 2 km long and about
1 km at its widest point. It includes an inner lagoon that is linked
to the Turneffe lagoon waters by a small creek on the north end of
the island (Fig. 1). It is fringed by mangroves on all its inner and
outer coasts with the exception of a part of the eastern side. The
tidal range is microtidal (average range <30 cm) and is classified as
mixed semidiurnal.
The climate is tropical to subtropical, with clearly distinct rainy
and dry seasons. The annual rainfall range is below the 2020 mm/yr
of Belize city but above 1500 mm (C. Piou, pers. obs.). Average
maximum air temperatures vary from 32?C in summer (March?
September) to 28?C in winter (October?February). Average mini-
mum temperatures vary with the same monthly pattern from 20?
24?C. The climate is influenced by cold fronts (or ?Northerns?)
coming from North America during the winter months. Between
July and November, hurricanes in the Caribbean Region increase
the occurrence of heavy rain and strong to catastrophic winds.
On 31 October 1961, Hurricane Hattie, a category five hur-
ricane, hit the Belizean coast. The eye of this hurricane went from
east northeast to west southwest over Turneffe Atoll with winds over
260 km/h. The intensity, wind direction, wave action, and storm
surge destroyed all of the human habitations and most of the man-
grove forests of Calabash Cay. According to Stoddart (1963) who
visited Turneffe in 1962, Big Calabash, a small island ca 300 m
to windward of Calabash Cay, was completely cleared of any man-
groves by Hattie. Based on Stoddart?s description, a gradient in
severity of destruction from east to west was expected for mangrove
forest at Calabash Cay. Specifically, Stoddart (1963) described that
defoliation of mangrove stands was less severe and regeneration was
already occurring in the interior part of Turneffe Atoll by early
1962. Calabash Cay has a large inner lagoon, which is considered
a natural shelter by the local population. The wave action and the
tidal hurricane surge were weaker inside the inner lagoon than out-
side. Since the human settlements were destroyed and not rebuilt
after Hattie, the regeneration of the mangrove forests of Calabash
Cay was not influenced by any anthropogenic activities since 1962.
Although tropical storms and lightning strikes could have induced
the death of individual trees, no other hurricanes or catastrophic
perturbations have affected substantially the Calabash mangroves
since Hattie. The gradient of disturbance hypothesized by Stoddart
(1963) in a relatively short distance around Calabash makes this
area ideal for the main objective of this study.
Four study sites were selected around Calabash along the pro-
posed disturbance gradient. These sites were labeled A?D (Fig. 1).
All sites extended from the fringing mangrove forest along the water?s
edge to the high intertidal margin where buttonwood (Conocarpus
erectus) thickets dominated adjacent to the upland terrestrial for-
est. Preliminary surveys were conducted to determine zones within
these sites depending on both the apparent structural and the species
zonation patterns. Site A was located on the northeast side of Cal-
abash where four zones were defined from the shoreline inland. Site
B was located on the northernmost point of Calabash and four
zones were defined. Site C was situated on the northwest side of
Zonation Patterns of Belizean Offshore Mangrove 367
FIGURE 1. Turneffe Atoll with Hurricane Hattie eye?s path and Calabash Cay with localization of the sites (modified from Garcia & Holtermann 1998).
Calabash and extended from Orchid Creek to the upland area with
six different zones. Site D was located on the east side of the inner
lagoon and had three mixed zones.
FOREST MEASUREMENTS.?Mean diameter at breast height (DBH),
tree height, density, and species frequencies were measured by the
point centered quarter method (PCQM) transects laid within each
zone of each site between November 2002 and February 2003. To
avoid measuring the same tree twice, the distance between points
of the PCQM transects varied between 2 m in the dwarf areas to
12 m in tall basin forests. Each of the PCQM transects consisted
of 21 points for a representative measurement of the studied area
(Cintro?n & Schaeffer-Novelli 1984).
PHYSICAL?CHEMICAL SURVEY.?Porewater salinity and pH were
measured with a YSI C? 556 Multiprobe System at three different pe-
riods: 24?30 November 2002, 6?9 February 2003, and 27 February
2003. Tidal measurements were made 13?22 February 2003. Tide
tables and several overnight measurements in a row were used to esti-
mate a February maximum tidal level for each station. A yearly max-
imum tidal level was computed by extrapolation of these February
maximum tidal levels in conjunction with the yearly tide tables. The
slope of the elevation change within each zone was also calculated
with the following formula: slope = H3 ? H1Dist1?3 ? 100, where H1 and
H3 were the relative elevation of the first and last points of each
zone, respectively, and Dist1?3 was the distance between these two
points.
As a measure of site fertility, the nutrient resorption efficiency
(RE) and leaf biomass production per unit of nutrient were calcu-
lated as an indirect way to determine trends of nutrient availability,
as described in Feller et al. (1999). Leaves were collected 20?22
February 2003. In each zone where present, four samples of three
to five pairs of leaves of R. mangle were harvested. A pair composed
of a mature leaf and a ready-to-fall senescent leaf from the same
twig. Digital images and image analysis software (SigmaScan Pro
4.0) were used to measure leaf area. All samples were oven-dried
at 70?C, weighed and ground to pass through a 0.38 mm mesh
screen. We measured nitrogen (N) and phosphorus (P) concentra-
tions in the green and senescent leaves, and calculated nutrient RE
and biomass production per unit of nutrient invested for each exper-
imental tree. RE was calculated as the percentage of N and P recov-
ered from senescing leaves before leaf fall (Chapin III & Van Cleve
1989): RE =
N or P(mg ? cm?2)green leaves?N or P(mg ? cm
?2)senescent leaves
N or P(mg ? cm?2)green leaves
? 100.
Biomass production per unit of nutrient was calculated as the
368 Piou, Feller, Berger, and Chi
inverse of the nutrient concentration in senescent leaves (nutrient
use efficiency, NUE in g biomass(g N or P).
INDICATORS OF DESTRUCTION LEVEL.?No direct information was
available to estimate differences of mangrove destruction among
sites during Hurricane Hattie. Therefore, an indirect measurement
was necessary to extrapolate from the standing trees considered older
than 41 yr to compare an estimated destruction level among the four
sites. From these estimates, a qualitative variable of ?complete? or
?incomplete destruction? was attributed to each site.
In the most inland mangrove zone, named here hinterland
zone, of the four sites, belt transects of 60 m long and 10 m width
were established. The hinterland zones were chosen to make certain
that the trees measured were in a comparable physical?chemical
situation. Within these belt transects, all trees with DBH more
than 20 cm were measured. The distribution patterns of these large
trees were used to detect extraordinary old or large trees seen as
outliers in the right side of the distribution. These outliers were
then used as proxies of destruction levels of the four different sites,
assuming that they represent the trees older than 41 yr.
In the fringe (i.e., the first zone along the water?s edge) and
dwarf zones of site C, a tree aging technique was adapted from
Duke and Pinzon (1992) based on leaf scar production to estimate
the possible destruction level of this forest. An estimation of leaf
scars produced by dwarf R. mangle of Turneffe Atoll was 4.1 scars/yr
(I.C. Feller, pers. comm.). This number was rounded to 5 to keep
an error range and a robust estimate of tree age. Leaf scars were
counted from the topmost apical leaves to the bottom of the trees
for all stems present in three 1 ? 1 m plots. Stems from an individual
tree having more than 205 scars were considered to be older than
41 yr. The percentage of trees >41 yr was calculated from these
measurements.
DATA ANALYSIS.?The PCQM results were used to calculate an
importance value (IV) for each of the four species per zone as
described in Cintro?n and Schaeffer-Novelli (1984). IV was based
on relative density, relative dominance, and relative frequency of
each species. Consequently, the sum of the IV of all four species
was equal to 300 for each zone. To summarize the IV variation
between zones and sites and to consider all species at the same time,
we developed an index of species dominance heterogeneity (ISDH)
based on the following equation, adapted from the reciprocal index
of Simpson (Hill 1973):ISDH =
300 ? (300?1)
?q
i=1(IVi ? (IVi?1))
, where IVi is the
importance value of the species i and q the number of species. ISDH
is comparable to a species diversity index of second order of entropy
because it would increase with a higher number of species but give
more weight to the evenness in IV. The index was used because it
incorporates the relative importance of species in terms of forest
structure instead of the number of individuals. Thus dominance
change would not only be a question of numbers of individuals, but
also of space occupied by each species.
Nonparametric Spearman rank correlation analyses were con-
ducted to investigate the relationships between the physical?
chemical factors including the nutrient availability indicators and
the different forest structure variables (excluding ISDH). A factor
analysis with principal component extraction and varimax rotation
was carried out to group linear correlations among the four nu-
trient variables. Scores of the two principal factors were used as
explanatory variables for further analyses. To investigate the factors
influencing the variation of the ISDH, a general linear model analysis
using Type V decomposition was used, followed by a general regres-
sion model analysis with backward stepwise selection of variables.
The following continuous independent variables were considered
for both analyses: mean recorded salinity, maximum tidal level,
mean pH, slope, and the two synthetic factors of nutrient use and
resorption efficiencies. The qualitative variable of destruction was
also used as categorical factor in these analyses. The ISDH and slope
values were square root (arcsin(x/4) + constant) transformed to get
closer to normality.
RESULTS
FOREST MEASUREMENTS.?The canopy height and DBH at all sites
showed general increases from the fringe zone to the hinterland
(Fig. 2). However, the resulting structural zonation patterns among
sites were different. For example, trees at site C were dwarf over
a 60-m wide area followed by a 15-m tall forest whereas trees at
site A were relatively homogeneous across the forest. Another main
difference among the sites was seen in the species zonation pattern
(Fig. 2). As a general pattern, R. mangle dominated the fringe zone
of all sites. The basin zones were occupied mainly by A. germinans
in sites B, C, and D. These sites were dominated by L. racemosa
in the hinterland zones. In contrast, R. mangle was dominant in all
zones at site A.
Species dominance heterogeneity (ISDH) varied significantly
among zones and sites (Fig. 2). Sites C and D were more hetero-
geneous than sites A and B. The pure dominance of R. mangle in
zones 1 and 2 of site A was not found in other parts of the island
except in the fringe and the first dwarf zone of site C. Also, the
only pure area of A. germinans was found on site B. An herbaceous
species, Batis maritima, was found on both sides of the island and
in different zone types.
PHYSICAL?CHEMICAL MEASUREMENTS.?The salinity was relatively
homogeneous at site A (Table 1). The maximum salinities were
measured in A. germinans basins at sites B, C, and D. The minimum
values were recorded in hinterland zones of all sites. The pH values of
sites A and B were neutral at the fringe and hinterland, and slightly
lower in the central zones probably linked to anoxic conditions
(Table 1). For sites C and D, pH values generally increased toward
the hinterland zones. In sites A and B, the first and last zones
had higher slope values compared to the central zones, which were
almost completely flat. Site D had in general the same pattern,
whereas the topography was more varying at site C (Table 1).
Nitrogen-use efficiency (NUE-N) varied slightly, leading to
few significant differences among zones and sites (Fig. 3). The
highest values occurred in site C, particularly in the dwarf zones.
Phosphorus-use efficiency (NUE-P) was also very high in these
Zonation Patterns of Belizean Offshore Mangrove 369
FIGURE 2. Mangrove forest structure of the different sites (A?D respectively). Zones are named according to position and type of structure (?indicate presence
of Batis maritima on the ground). ISDH values indicate the corresponding Index of species dominance heterogeneity. Pie charts show the species composition. Line
graphs show trees mean DBH, mean canopy height per zone (error bars represent standard deviations), and profile of maximum height of tidal level at equinoxes
spring high tide (in cm above ground).
dwarf zones of site C. For all sites, NUE-N and NUE-P were
generally higher in zone 1 and decreased going landward. There
were no significant differences in N-resorption efficiency (RE-N)
among the sites (Fig. 4). Similar to patterns of NUE-N and NUE-
P, P-resorption efficiency (RE-P) was highest in the dwarf zones
and with lowest values in the inland zones at sites B, C, and D
(Fig. 4). The variations in NUE and RE among the sites and zones
were synthesized with factor analysis (Table 2), which confirms tight
correlation of both NUE -N and -P and RE-P.
INDICATORS OF DESTRUCTION LEVEL.?In the belt transects of the
hinterland zones of sites A and B, the frequency distribution of
DBH of trees >20 cm decreased continuously going toward bigger
size classes (Fig. 5). However, in the hinterland zones of sites C
and D, the distribution included a few outliers with DBH >55
cm. The largest trees in sites A and B had DBH <37 cm, which
together with the continuous decrease, provided evidence that the
trees in those sites had grown subsequent to Hurricane Hattie. On
the contrary, the frequency distribution of DBH of sites C and D
370 Piou, Feller, Berger, and Chi
FIGURE 2. Continued.
TABLE 1. Physicochemical measurements results per sites and zones. (Values are mean salinity ? SD, expressed in practical salinity units (psu); mean pH ? SD; S = slope
of zone ground, in %.)
Site A Site B Site C Site D
Zones Psu pH S psu pH S psu pH S psu pH S
1 37.9 ? 0.4 6.6 ? 0.2 1.20 38.9 ? 0.9 6.7 ? 0.6 1.95 40.5 ? 2.7 6.6 ? 0.1 0.25 40.6 ? 6.3 6.5 ? 0.2 0.90
2 36.9 ? 1.5 6.4 ? 0.05 0.10 45.1 ? 8 6.8 ? 0.3 0.03 44.9 ? 2.8 6.6 ? 0.3 0.33 44.7 ? 6.5 6.9 ? 0.1 0.70
3 37.7 ? 2.0 6.4 ? 0.1 ?0.06 46.0 ? 10.6 6.5 ? 0.1 ?0.12 46.1 ? 5.0 6.7 ? 0.2 0.17 21.6 ? 5.1 7.3 ? 0.2 2.07
4 37.0 ? 3.3 7.0 ? 0.4 1.80 39.6 ? 10.9 6.9 ? 0.2 0.70 39.7 ? 5.1 6.5 ? 0.1 0.10
5 42.0 ? 9.0 6.8 ? 0.3 ?0.20
6 22.8 ? 9.1 7.1 ? 0.1 1.40
Zonation Patterns of Belizean Offshore Mangrove 371
FIGURE 3. Rhizophora mangle leaves nutrient use efficiency of nitrogen (white bars) and phosphorus (filled bars) in grams of created leave biomass/g of N or P
(error bars = SE). The x-axis represents the different sites and zone numbers. Letters show the homogeneous groups (Posthoc analysis Sheffe? test).
was discontinuous between 45 and 55 cm. The few large trees at
the end of the distribution survived Hattie and were thus >41 yr
old. Additionally, in zones 1 and 2 at site C, more than 61 percent
of dwarf trees were >41 yr.
All these observations suggest that Hattie caused complete de-
struction of the mangrove forest at sites A and B, whereas remnants
of the original forests still existed across the forests at sites C and
D, indicating incomplete destruction. This information was used
to generate a dummy variable of two destruction levels: for sites A
and B, 0 = complete destruction during Hattie, for sites C and D,
1 = incomplete destruction during Hattie.
FIGURE 4. Rhizophora mangle leaves resorption efficiency of nitrogen (white bars) and phosphorus (filled bars) in percentage (error bars = SE). The x-axis represents
the different sites and zone numbers. Letters show the homogeneous groups (Posthoc analysis Sheffe? test).
STATISTICAL ANALYSES.?The Spearman correlation analyses showed
that the IV of R. mangle were negatively correlated to the pH
and positively to NUE-N, NUE-P, and RE-P. IVs of A. germinans
were negatively correlated with NUE-N, NUE-P, and RE-P but
positively to the maximum recorded salinity. IVs of L. racemosa were
negatively correlated with the maximum tidal level and positively
to the slope and pH. IVs of C. erectus were negatively correlated to
the maximum tidal level, the mean salinity, NUE-N, NUE-P, and
RE-P but positively to the slope and pH (Spearman correlation test,
P < 0.05 in all cases). Additionally, NUE-N, NUE-P, and RE-P
were negatively correlated with DBH and tree height, and positively
372 Piou, Feller, Berger, and Chi
TABLE 2. Factor loadings of nutrient use variables factor analysis with principal
component extraction method and varimax rotation (marked loadings
are >0.70).
Factor 1 Factor 2
NUE-N 0.738 0.490
NUE-P 0.958 ?0.098
RE-N 0.023 0.978
RE-P 0.901 0.153
Eigenvalues 2.273 1.230
Prop. of total var. 0.568 0.307
correlated with tree density (Spearman correlation test, P < 0.01 in
all cases).
The general linear model constructed with all the variables did
not explain significant variations of the index of species dominance
heterogeneity (ISDH) (test whole GLM model vs. residual, Multiple
R = 0.812, F = 2.483, P = 0.102). However, within this model,
the destruction indicator dummy variable had a significant effect
on the ISDH (GLM univariate significance; F = 5.158, P = 0.049).
None of the other variables (i.e., salinity, tidal level, pH, slope, and
nutrient availability factors) had a significant influence on the ISDH.
On the contrary, the general regression model constructed with a
backward stepwise selection and selecting the destruction indicator
dummy variable and the first factor of nutrient availability indicator
(Table 3), explained a significant proportion of the variance of ISDH
FIGURE 5. Trees DBH class distribution (only DBH >20 cm) in the last zone of each sites (Site A: black filled; Site B: white; Site C: horizontal stripes; and Site
D: vertical stripes).
(test whole GRM model vs. residual, Multiple R = 0.718, F =
7.450, P = 0.006).
DISCUSSION
Forest measurements results showed clearly different horizontal
zonation patterns at Calabash Cay. Firstly, the structural zonation
pattern indicated by DBH, tree height, and forest density differed
at each site. Correlation analyses suggested that these patterns were
related to N-use efficiency, P-use efficiency, and P-resorption effi-
ciency. According to the findings of Feller (1995) and Feller et al.
(2002, 2003), these values were indicative of nutrient availability.
The species zonation patterns were also clearly different from sites to
sites. Rhizophora mangle was more dominant in low nutrient avail-
ability and acidic areas. Similarly, A. germinans dominated more in
high salinity and high nutrient availability areas, and L. racemosa
in higher intertidal, steeper, and more neutral-to-basic zones. The
restriction of C. erectus only in the hinterland zones demonstrated
the adaptation of this species to rarely inundated areas (Tomlinson
1986). These trends fit well with other studies in the Caribbean.
They suggested that differences in propagules dispersal (Rabinovitz
1978, Jime?nez & Sauter 1991), geomorphological factors (Thom
1967), as well as physiological adaptations to gradients across the
intertidal zone (Macnae 1968, or e.g., McKee 1995a,b) were inter-
acting factors driving zonation patterns in Calabash Cay.
According to Stoddart (1963), the vegetation of inner parts of
Turneffe was less damaged by Hurricane Hattie than the vegetation
of outer sites. Both proxies used to indicate destruction levels in
Zonation Patterns of Belizean Offshore Mangrove 373
TABLE 3. Univariate tests of significance and summary of parameters of the general regression model with backward stepwise selection of explanatory variables on the ISDH
variations (marked factors are significant at P < 0.05).
SS df MS F P Parameter SE
Intercept 0.735 1 0.735 1.515 0.239 ?0.208 0.169
Maximum tidal level Pooled 0
Mean salinity Pooled 0
Mean pH Pooled 0
Slope Pooled 0
Factor 1 Nutrients 3.934 1 3.934 8.113 0.013 ?0.559 0.196
Factor 2 Nutrients Pooled 0
Destruction 4.203 1 4.203 8.667 0.011 ?0.502 0.170
Error 6.789 14 0.485
this study supported this observation 41 yr later: the distributions
of all trees with DBH >20 cm showed that no old trees (right
side outliers) existed in the seaward sites A and B, but some big
hollow, remnant trees older than 41yr were found in the inner sites
C and D. Secondly, the number of scars indicating the age of dwarf
trees also showed incomplete destruction in a sheltered position on
the west side of Calabash. The percent of dwarf trees older than
41 yr suggested that the dwarf mangroves of Calabash Cay were
the trees that best resisted Hurricane Hattie. These findings were
consistent with other observations from the Caribbean (Smith et al.
1994, Lugo 1997) and Australia (Bardsley 1984). During Hattie,
the high water level due to the storm surge would have covered
these small individuals, preventing them from being blown down
by strong winds. If these trees had been exposed to wave action, it is
likely that they would have been killed as the fringing zones of most
hurricane-destroyed sites (e.g., Imbert 2002). Thus, the survivability
of the dwarf trees makes them potential sources of propagules for
recolonization processes after large perturbations.
The index of species dominance heterogeneity (ISDH) showed
differences among zones and sites. One of the general trends for each
site was that the hinterland zone had the maximum ISDH value. This
trend resulted from the higher possibility of finding pure R. mangle
or A. germinans stands in the lower intertidal zones than in the
hinterland zones where L. racemosa was frequently mixed with C.
erectus. Besides these intrasite variations, the ISDH at the east side
of the island (i.e., the exposed part that sustained the most damage
from Hattie) was more homogeneous than the west side (sheltered
part). Our statistical results with the GLM model showed the im-
portance of the destruction indicator to these forest heterogeneity
differences. The GRM model also included the factors indicating
nutrient availability, which could be related to two aspects of forest
heterogeneity. First, P-use efficiency and resorption efficiency in-
creased from the fringe to the upper zones, as did the ISDH. Second,
N-use efficiency was highest at sites where R. mangle dominated,
leading to a low ISDH. These relationships supported the hypothesis
that nutrient availability was important to species dominance. Yet,
only the model including both the nutrient availability and the de-
struction indicator explained the differences of forest heterogeneity
between the two sides of Calabash. These results suggested that
forest dynamics at Calabash were influenced by the remnant forest
that survived Hattie. A possible scenario was that the surviving A.
germinans and L. racemosa, as well as the dwarf R. mangle trees on
the western side of the island served as a reservoir for propagules
on the nearby areas. In combination with external propagules, this
could have resulted in a relatively rapid recolonization of Calabash.
Succession modified by the presence of the larger trees might have
increased the spatial patchiness by competition effects. We observed
that some R. mangle trees developed as an understory in A. germi-
nans basins at the sheltered sites, which indicated that this complex
phenomenon was not finished yet. Rhizophora mangle may eventu-
ally outcompete the two other species in some zones. In contrast,
on the eastern sites, succession and competition probably happened
according to the species characteristics to abiotic conditions. This
would have led to more homogeneous zones with clearer partition-
ing of the species along the intertidal. The example of Calabash
Cay having the most heterogeneous sites on the least-destroyed sites
except for the dwarf areas was contradictory to what Baldwin et al.
(2001) predicted for Biscayne Bay (Florida); however, their obser-
vations of the regeneration of mangrove sites were only 7 yr after
Hurricane Andrew, and illustrated the influence of several herba-
ceous species on the early regeneration pathways. The herbaceous
species present in Calabash Cay were found in both heterogeneous
and homogeneous forests. Our study was consistent with Imbert
(2002), who found that the number of surviving trees influenced
the recovery pathways and succession. All these studies showed that
remnant vegetation plays an important role for the recovery of man-
grove forests after hurricane destruction.
To conclude, this study suggests that disturbance intensity can
influence the recovery pathways and succession in mangrove forests.
Detailed quantitative data on forest structure before and after distur-
bance, following the regeneration processes would provide stronger
evidence of this relationship. The example of Calabash Cay man-
groves showed that the perturbation regime should be considered
when interpreting horizontal zonation patterns of mangrove forests.
374 Piou, Feller, Berger, and Chi
ACKNOWLEDGMENTS
The authors wish to thank the Institute of Marine Studies of the
University of Belize for the facility and moral assistance, as well as
Rachel Borgatti and Anne Chamberlain for precious fieldwork help
and Hanno Hildenbrandt for valuable comments. This work was
partly funded by a grant from the National Science Foundation
(DEB-9981535) to I.C. Feller.
LITERATURE CITED
BALDWIN, A., M. EGNOTOVICH, M. FORD, AND W. PLATT. 2001. Regeneration
in fringe mangrove forests damaged by Hurricane Andrew. Plant Ecol.
157: 149?162.
BALL, M. C. 1980. Patterns of secondary succession in a mangrove forest in
southern Florida. Oecologia (Ber.) 44: 226?235.
BARDSLEY, K. 1984. The effects of Cyclone Kathy on mangrove vegetation. In
K. N. Bardsley, J. D. S. Davie, and C. D. Woodroffe (Eds.). Coastal and
tidal wetlands of the Australian monsoon region, pp. 167?185. North
Australia Research Unit, ANU Press, Darwin, Australia.
CAHOON, D. R., P. HENSEL, J. RYBCZYK, K. L. MCKEE, C. E. PROFFITT, AND B.
C. PEREZ. 2003. Mass tree mortality leads to mangrove peat collapse at
Bay Islands, Honduras after Hurricane Mitch. J. Ecol. 91: 1093?1105.
CHAPIN F. S., III, AND K. VAN CLEVE. 1989. Approaches to studying nutrient
uptake, use and loss in plants. In R. W. Pearcy, J. R. Ehleringer, H. A.
Mooney, and P. W. Rundel (Eds.). Plant physiological ecology, volume
field methods and instrumentation, pp. 185?207. Chapman and Hall,
New York.
CINTRO?N, G., AND Y. SCHAEFFER-NOVELLI. 1984. Methods for studying man-
grove structure. In C. J. Snedaker and G. S. Snedaker (Eds.). The man-
grove ecosystem: Research methods, pp. 91?113. UNESCO Mono-
graphs in Oceanographic Methodology, Paris, France.
DAVIS, J. H. 1940. The ecology and geologic role of mangroves in Florida. Pub-
lications of the Carnegie Institute, Washington, District of Columbia.
DUKE, N. C., AND Z. S. PINZON. 1992. Ageing Rhizophora seedlings from
leaf scar nodes: A technique for studying recruitment and growth in
mangrove forests. Biotropica 24(2a): 173?186.
FELLER, I. C. 1995. Effects of nutrient enrichment on growth and herbivory
of dwarf Rhizophora mangle L. (red mangrove). Biotropica 28(1): 13?
22.
???, K. L. MCKEE, D. F. WHIGHAM, AND J. P. O?NEILL. 2002. Nitrogen vs.
phosphorus limitation across an ecotonal gradient in a mangrove forest.
Biogeochemistry 62: 145?175.
???, D. F. WHIGHAM, J. P. O?NEILL, AND K. L. MCKEE. 1999. Effects
of nutrient enrichment on within-stand cycling in a mangrove forest.
Ecology 80(7): 2193?2205.
???, ???, K. L. MCKEE, AND C. E. LOVELOCK. 2003. Nitrogen limitation
of growth and nutrient dynamics in a disturbed mangrove forest, Indian
River Lagoon, Florida. Oecologia 134(3): 405?414.
GARCIA, E., AND K. HOLTERMANN. 1998. Calabash Caye, Turneffe Islands Atoll,
Belize. CARICOMP?Caribbean coral reef, seagrass and mangrove sites.
Coastal region and small island papers 3. B. Kjerfve. UNESCO, Paris,
France.
HILL, A. V. 1973. Diversity and evenness: A unifying notation and its conse-
quences. J. Ecol. 61: 237?249
IMBERT, D. 2002. Impact des ouragans sur la structure et la dynamique forestie`res
dans les mangroves des Antilles. Bois et Fore?ts des Tropiques 273(3): 69?
78.
???, A. ROUSSEAU, AND P. LABBE?. 1998. Ouragans et diversite? biologiques
dans les fore?ts tropicales. L?exemple de la Guadeloupe. Acta Oecologica
19(3): 251?262.
JIME?NEZ, J. A., AND K. SAUTER. 1991. Structure and dynamics of mangrove
forests along a flooding gradient. Estuaries 14(1): 49?56.
KRAUSS, K. W., T. W. DOYLE, R. R. TWILLEY, T. J. SMITH, III, K. R. T. WHELAN,
AND J. K. SULLIVAN. 2005. Woody debris in the mangrove forests of
South Florida. Biotropica 37(1): 9?15.
LUGO, A. E. 1997. Old-growth mangrove forests in the United States. Conserv.
Biol. 11: 11?20.
MACNAE, W. 1968. A general account of the fauna and flora of mangrove swamps
and forests in the Indo-West-Pacific region. Adv. Mar. Biol. 6: 73?
270.
MCKEE, K. L. 1993. Soil physicochemical patterns and mangrove species distri-
bution: Reciprocal effects? J. Ecol. 81: 477?487.
???. 1995a. Mangrove species distribution and propagule predation in Be-
lize: An exception to the dominance-predation hypothesis. Biotropica
27(3): 334?345.
???. 1995b. Seedling recruitment patterns in a Belizean mangrove forest:
Effects of establishment ability and physico-chemical factors. Oecologia
101: 448?460.
RABINOVITZ, D. 1978. Dispersal properties of mangrove propagules. Biotropica
10: 47?57.
ROTH, L. C. 1992. Hurricanes and mangrove regeneration: Effects of Hurricane
Joan, October 1988, on the vegetation of Isla del Venado, Bluefields,
Nicaragua. Biotropica 24(3): 375?384.
???. 1997. Implications of periodic hurricane disturbance for the sustainable
management of caribbean mangroves. In B. Kjerve, L. D. D. Lacerda,
and E. H. S. Diop (Eds.). Mangrove ecosystem studies in Latin America
and Africa. UNESCO & ISME.
SHERMAN, R. E., AND T. J. FAHEY. 2001. Hurricane impacts on a Mangrove
forest in the Dominican Republic: Damage patterns and early recovery.
Biotropica 33(3): 393?408.
SMITH T. J., III 1987. Seed predation in relation to tree dominance and distri-
bution in mangrove forests. Ecology 68: 266?273.
???. 1992. Forest structure. In A. I. Robertson, and D. M. Alongi (Eds.).
Tropical mangrove ecosystems, Vol. 41. American Geophysical Union,
Washington, District of Columbia.
???, AND N. C. DUKE. 1987. Physical determinants of inter estuary variation
in mangrove species richness around the tropical coastline of Australia.
J. Biogeogr. 14: 9?19.
???, M. B. ROBBLEE, H. R. WANLESS, AND T. W. DOYLE. 1994. Mangroves,
Hurricanes, and Lightning Strikes. Bioscience 44: 256?262.
STODDART, D. R. 1963. Effects of Hurricane Hattie on the British Honduras
reefs and cays, Oct. 30?31, 1961. Atoll Research Bulletin 95.
THOM, B. G. 1967. Mangrove ecology and deltaic geomorphology: Tobasco,
Mexico. J. Ecol. 55: 301?343.
TOMLINSON, P. B. 1986. The botany of mangroves. Cambridge University Press,
Cambridge, England.
VERMEER, D. E. 1963. Effects of hurricane Hattie, 1961, on the cays of British
Honduras. Zeitschrift fu?r Geomorphologie 7: 332?354.