Ecological Applications, 17(6), 2007, pp. 1678?1693
 2007 by the Ecological Society of America
MANGROVE RECRUITMENT AFTER FOREST DISTURBANCE IS
FACILITATED BY HERBACEOUS SPECIES IN THE CARIBBEAN
KAREN L. MCKEE,1,4 JILL E. ROOTH,1,2 AND ILKA C. FELLER3
1National Wetlands Research Center, U.S. Geological Survey, 700 Cajundome Boulevard, Lafayette, Louisiana 70506 USA
2Elkhorn Slough National Estuarine Research Reserve, 1700 Elkhorn Road, Watsonville, California 95076 USA
3Smithsonian Environmental Research Center, 647 Contees Wharf Road, Edgewater, Maryland 21037-0028 USA
Abstract. Plant communities along tropical coastlines are often affected by natural and
human disturbances, but little is known about factors in?uencing recovery. We focused on
mangrove forests, which are among the most threatened ecosystems globally, to examine how
facilitation by herbaceous vegetation might improve forest restoration after disturbance. We
speci?cally investigated whether recovery of mangrove forests in harsh environments is
accelerated by nurse plants and whether the bene?cial effects are species-speci?c. Quanti?cation
of standardized effects allowed comparisons across performance parameters and over time for:
(1) net effect of each herbaceous species on mangrove survival and growth, (2) effects of pre- and
post-establishment factors associated with each herbaceous species, and (3) need for arti?cial
planting to enhance growth or survival of mangrove seedlings. Mangrove recruitment in a clear-
cut forest in Belize was accelerated by the presence of Sesuvium portulacastrum (succulent forb)
and Distichlis spicata (grass), two coastal species common throughout the Caribbean region. The
net effect of herbaceous vegetation was positive, but the magnitude of effects on mangrove
survival and growth differed by species. Because of differences in their vegetative structure and
other features, species effects on mangroves also varied by mechanism: (1) trapping of dispersing
propagules (both species), (2) structural support of the seedling (Distichlis), and/or (3) promotion
of survival (Sesuvium) or growth (Distichlis) through amelioration of soil conditions
(temperature, aeration). Arti?cial planting had a stronger positive effect on mangrove survival
than did edaphic conditions, but planting enhanced mangrove growth more in Sesuvium than in
Distichlis patches. Our study indicates that bene?cial species might be selected based on features
that provide multiple positive effects and that species comparisons may be improved using
standardized effects. Our ?ndings are not only relevant to the coastal environments found in the
Caribbean region, but our assessment methods may be useful for developing site-speci?c
information to restore disturbed mangrove forests worldwide, especially given the large pool of
mangrove associates (.45 genera) available for screening.
Key words: coastal wetland; disturbance; facilitation; forest regeneration; mangrove associate; nurse
plant; patch dynamics; restoration; Rhizophora; seed dispersal; seedling recruitment; stress.
INTRODUCTION
Knowledge of mechanisms in?uencing plant estab-
lishment and growth is not only important to an
understanding of how plant communities are organized,
but also in the enhancement of restoration of important
ecosystems. Rapid reestablishment of native vegetation,
particularly after a large-scale disturbance, can be
critical in preventing soil erosion, invasion by exotics,
and other unwanted outcomes. Recolonization of
disturbed sites may be slow and unpredictable, especially
if seed sources are remote. Recovering plant communi-
ties are also shaped by the positive and negative
interactions among component species. Although com-
petition has been emphasized as an organizing force, a
large body of information supports facilitation as a
mechanism in?uencing plant distribution and success
(see Connell 1983, Hunter and Aarssen 1988, Goldberg
and Barton 1992, Callaway 1995, Bertness and Leonard
1997, Bruno et al. 2003).
Ecological restoration may involve not only arti?cial
reintroduction of the original community dominants, but
also nurse species that improve seed trapping and
establishment (Day andWright 1989), attract seed carriers
(Robinson and Handel 1993), enhance soil conditions
through organic matter or nutrient accumulation (Pugn-
aire et al. 1996, Maestre et al. 2001), or provide protection
of sensitive seedlings (Martinez 2003, Gomez-Aparicio et
al. 2004). Ecological restoration approaches, however,
must be based on a thorough understanding of the natural
successional dynamics of the system as well as the growth
requirements of the dominant plant species. This under-
standing is particularly important in forested systems in
which the age-determined features of ecosystem structure
and function are slow to recover.
Manuscript received 22 September 2006; revised 12 February
2007; accepted 13 February 2007. Corresponding Editor: E.
Cuevas.
4 E-mail: karen_mckee@usgs.gov
1678
The current challenge in ecological restoration is to
manipulate development so that recovery of the entire
suite of structural and functional features is achieved as
quickly as possible (Dobson et al. 1997). However, for
many types of ecosystems the information necessary to
make critical decisions about species introductions is not
currently available. Few studies have experimentally
examined facilitation in the context of restoration
(Maestre et al. 2001, Gomez-Aparicio et al. 2004).
Consequently, the scienti?c and resource community has
been hesitant to apply facilitation to restoration practices.
A better understanding of plant interactions, particularly
facilitation, may lead to better techniques that support
and encourage ecological restoration approaches.
A key question is whether facilitation is species-speci?c
(i.e., involves speci?c traits of the benefactor) or those
positive effects on the bene?ciary are due to environmen-
tal changes that can be generated by any morphologically
similar species or even nonliving objects such as rocks or
woody debris. The answer to this question has clear rele-
vance for ecological restoration. Facilitation may not only
involve amelioration of environmental conditions that
promote growth of a bene?ciary species, but can also arise
from effects on dispersal and establishment, e.g., trapping
of seeds or propagules. A benefactor species that increases
the rate of seedling establishment and generates condi-
tions conducive to growth of established seedlings would
have a greater in?uence on seedling dynamics than a
species that provides a single effect. Studies have generally
investigated facilitation by one of these processes, but not
of both simultaneously (Callaway 1995). Consequently,
the necessary information needed to select bene?ciary
species is often lacking to support restoration efforts.
Facilitation has been studied in extreme environments
such as salt marshes (Bertness 1991, Bertness and Hacker
1994, Egerova et al. 2003), where plants must cope with
stresses such as salinity, ?ooding, and variable sediment
and nutrient supplies. Mangroves are the tropical
equivalent of temperate salt marshes, but in contrast to
marsh grasses that can propagate vegetatively, these tidal
forests are dominated by tree species dependent upon
seedling recruitment for regeneration. Mangroves are
frequently disturbed by hurricanes and human activities,
which severely damage or eliminate the forest community.
Mangroves are considered to be one of the most
endangered ecosystems in the world, with ;35% of the
original area degraded or destroyed since 1980 (Valiela et
al. 2001; Food and Agricultural Organization of the
United Nations database, available online).5 Mangrove
loss is particularly devastating to the Caribbean region
(;28% loss between 1980 and 2000), where they constitute
a critical habitat intimately linked to adjacent seagrass
and coral reef ecosystems (Rivera-Monroy et al. 2004).
Such losses may be reversed through application of the
principles of ecological restoration.
Recent work emphasizes the use of ecological
engineering to achieve cost-effective restoration by
properly preparing soil elevations (Lewis 2005). Com-
plimentary approaches that promote rapid revegetation
and also minimize cost and effort are needed, especially
for developing countries with few resources and
incentives to restore degraded forests (Kaly and Jones
1998, Alongi 2002, Barbier 2006). Here we document the
potential for use of nurse plants to promote mangrove
recruitment in large disturbed sites where harsh condi-
tions limit mangrove recovery.
Mangrove plant communities often contain herba-
ceous species, which are common components of the
tropical beach habitat, salt marshes, or other wet coastal
communities (Tomlinson 1994). Such ??mangrove asso-
ciates?? may occur naturally as understory, inhabit a
back-mangal ecotone, or invade only upon disturbance
of the dominant mangrove vegetation. Their presence
could positively in?uence mangrove reestablishment and
growth and, consequently, be used to accelerate natural
or managed recovery from disturbance. Although
factors in?uencing mangrove recruitment such as seed
and seedling predators (McKee 1995a, Lindquist and
Carroll 2004), ?ooding and salinity (McKee 1995b,
Elster et al. 1999, Sherman et al. 2000), and sedimen-
tation (Elster 2000) have been studied in neotropical
forests, effects of herbaceous associates are relatively
unstudied (Flower 2004, Milbrandt and Tinsley 2006).
Mangroves may be extremely slow to recolonize and
grow, especially in harsh (e.g., arid, hypersaline)
environments of the Caribbean region (Cintron et al.
1978, Feller 1995, McKee 1995b, Imbert et al. 2000,
Feller et al. 2003, Flower 2004, Milbrandt and Tinsley
2006). Mangrove ecosystems thus constitute not only a
critical habitat with important ecological and societal
bene?ts, but are systems in which facilitative interactions
might be applied to improve restoration techniques.
Furthermore, an understanding of how benefactors may
facilitate survival and growth of mangroves will lead to
identi?cation of vegetative characteristics to screen
potential candidates for restoration projects.
In this study, we examined mangrove seedling
establishment and subsequent growth in response to
two herbaceous species, Distichlis spicata (L.) Greene
(Poaceae) and Sesuvium portulacastrum L. (Aizoaceae),
which are common halophytes found along shorelines in
the Caribbean region. We hypothesized that in extreme
environments such as those found on intertidal Carib-
bean islands, these two species would have a positive
effect on mangrove recruitment in large disturbed areas.
Because of differences in vegetative structure and other
features (grass vs. succulent forb), we further hypothe-
sized that effects of herbaceous species on mangroves
would differ. These hypotheses were tested at a long-
term study site (Twin Cays, Belize) that has been the
focus of research by the Smithsonian Institution since
the early 1970s (Ru?tzler and Feller 1996).5 hhttp://www.fao.org/docrep/007/j1533e/J1533E00.htmi
September 2007 1679FACILITATION OF MANGROVES
METHODS
Study site
The study was conducted at Twin Cays, a 75-ha
archipelago located in the shoreward lagoon of the
MesoAmerican barrier reef system, ;12 km from the
mainland of Belize (168500 N, 888060 W; Fig. 1). Twin
Cays is a series of low-relief, intertidal islands that are
peat-based (Ru?tzler and Feller 1996). Mangrove islands
in this part of the reef complex established on a
Pleistocene platform approximately 8000 years ago
and have been mangrove communities throughout the
Holocene (Macintyre et al. 2004, McKee et al. 2007).
The mangrove forest is the oceanic-island type, and
species composition is typical of the Caribbean region
with Rhizophora mangle L. (red mangrove, see Plate 1),
Avicennia germinans (L.) Stearn. (black mangrove), and
Laguncularia racemosa (L.) Gaertn. f. (white mangrove).
The dominant species is R. mangle, which forms
monospeci?c stands along the shoreline and in the
interior forest. The topography is heterogeneous with
vegetated areas interrupted by tidal creeks, open ?ats,
and shallow ponds. The system is intertidal (20?30 cm
range), and the only source of freshwater is rainfall,
which averages ,1000 mm/yr. Salinity consequently
varies from near sea-strength (36?) to hypersaline (80?
90?). Nutrient availability is low, and external inputs
are limited to seawater and rainfall. Detailed informa-
tion on the vegetation, geomorphology, chemistry, and
hydrology is available elsewhere (Wright et al. 1991,
McKee 1993a, Feller 1995, Woodroffe 1995).
Historically, Twin Cays has experienced little human
in?uence, making it a valuable experimental ?eld site.
More recently, human impact on Twin Cays and other
mangrove islands in the vicinity has escalated through
clear-cutting of mangroves for construction of ?shing
camps and tourist resorts and dredging to raise soil
elevations. However, because the substrate is peat and
maintenance of soil elevations relative to sea level is
dependent upon inputs of organic matter from man-
groves (McKee and Faulkner 2000, Macintyre et al.
2004, McKee et al. 2007), these areas rapidly subside
and are ultimately abandoned.
This study focused on a 2-ha area cleared of
mangroves in 1991 on the western island of Twin Cays
(Fig. 1). Colonization by herbaceous vegetation and
mangroves was monitored in permanent plots during the
?rst three years after disturbance (January 1992?
FIG. 1. Map showing location of the study area at Twin Cays, Belize, and the study site that was clear-cut in 1991. Section 1
was monitored until it was back?lled with sand in 1995; section 2 continued to regenerate and was reassessed in 2001.
KAREN L. MCKEE ET AL.1680 Ecological Applications
Vol. 17, No. 6
January 1995). However, in 1995, part of the recovering
area was covered with dredge material (sand and peat)
to raise elevations above the intertidal zone. Because the
majority of permanent plots were destroyed, further
monitoring of mangrove recovery was not possible. Part
of the original clear-cut area (;0.7 ha) was not
impacted, and this site was used in 2000?2001 to
experimentally examine effects of herbaceous vegetation
on mangrove recruitment.
Initial vegetative colonization
In January 1992, 16 plots (434 m) were established in
a strati?ed random design across the 2-ha clear-cut area.
Measurements of edaphic factors were conducted in the
center of each plot as described previously (McKee et al.
1988, McKee 1995b). Percent cover of herbaceous
vegetation by species was estimated visually in each
plot, and all mangrove seedlings were counted and
identi?ed to species. Measurements were repeated at
approximately six-month intervals for three years. In
1995, this effort was suspended when the area was
partially back?lled, and 10 of the 16 plots were
permanently buried.
Changes in herbaceous plant cover and soil variables
for three years after clear-cutting were analyzed with a
repeated-measures ANOVA (n ? 16 plots) (SAS 2002).
A one-way repeated-measures ANOVA with mangrove
species and time as grouping factors was conducted on
mangrove density. Contrasts of interest (1 df) were
conducted when a signi?cant interaction was found.
Relationships between herbaceous plant cover and soil
variables were examined by correlation analysis.
Herbaceous patch characterization
In February 2001, soil and plant measurements were
again undertaken to characterize herbaceous patches in
an area unaffected by dredge deposition (section 2, Fig.
1). Discrete patches dominated by either S. portulacas-
trum or D. spicata (see Plate 1) were identi?ed, and a
subset of three patches per species was randomly
selected for study. Sampling positions were established
with two stations at each of the following positions
along a transect traversing each patch: outside patch in
unvegetated ground (1 m from the patch edge), patch
edge (outer 50 cm), and patch interior (25?50 cm from
the center). The two patch types were compared using
stem density, canopy height, and rooting depth to
characterize above- and belowground vegetative struc-
ture. Total number of stems and maximum stem height
were measured in 0.1-m2 circular plots. Shallow cores
(15 cm depth 3 4 cm diameter) were taken in the same
locations, and rooting depth was measured. Soil redox
potentials and soil temperature (pH/temperature/mV
meter, model 5938-10; Cole-Parmer, Chicago, Illinois,
USA) were measured over depth at each sampling
station. Cores (12 cm depth 3 1.2 cm diameter) were
collected with a piston corer and stored in a watertight
plastic bag for analysis of salinity, bulk density, water
content, and organic content. After obtaining the wet
mass of the core, water was extracted by centrifugation,
and the supernatant was measured for salinity and pH.
The soil was dried at 808C to constant mass and
reweighed to determine bulk density and water content.
The dried soil was ashed at 5508C for 6 h to determine
mineral mass after organic loss on ignition. Additional
soil samples were analyzed for C and N contents
(ThermoQuest Flash EA 1112 Series CHN Analyzer;
CE Elantech, Lakewood, New Jersey, USA). Plant and
soil variables were analyzed using a split-plot model,
with patch species as the main plot and spatial position
as the subplot (SAS 2002).
Mangrove distribution and status in patches
To determine whether distribution and condition of
mangroves differed between patch and non-patch
environments, mangroves growing inside and outside
selected herbaceous patches were examined in detail.
The design was a split-plot replicated three times. Each
main plot contained one of the herbaceous patches
selected previously and an adjacent bare area equivalent
in size to the patch. All mangrove seedlings and saplings
in the subplots were identi?ed to species and counted.
Mangrove condition inside and outside patches was
assessed for the dominant species, R. mangle. The
height, number of nodes, number of growing points
(terminal buds), and reproductive output (number of
fruit or propagules per plant) were determined on every
individual present in patch and bare plots. An index of
plant vigor was calculated by dividing the total number
of growing points by the number of nodes on the main
axis. Total mangrove and individual species densities
and R. mangle condition were examined using a split-
plot model with patch species as the main plot and
spatial position (inside, outside) as the subplot (SAS
2002). Frequency distribution of node range for R.
mangle recruits was examined for each patch type
separately by ?tting a normal distribution to the data
and examining the line of ?t generated from the quantile
plot (SAS 2002).
Manipulative experiment
In August 2001, mature R. mangle propagules were
collected from trees in the undisturbed forest and placed
within patches dominated by either S. portulacastrum or
D. spicata and in adjacent bare areas to examine patch
effects on establishment and early growth of mangrove
seedlings. Propagules were selected based on uniformity
in size and labeled by writing directly on the hypocotyl
with a permanent marker. The propagules of R. mangle
are large (mean wet mass? 10 g) and elongated (;25 cm
in length), and the primary mode of dispersal is water
(McKee 1995a). Vegetated patches and bare-ground
plots containing few or no mangroves initially were
identi?ed, and propagules were randomly assigned to
them. Two types of planting conditions were also used
to separate patch effects on establishment from effects
September 2007 1681FACILITATION OF MANGROVES
on subsequent survival and growth. In each of the
replicate plots, 10 propagules were planted in an upright
position by inserting the radical tip into the soil and
another 10 propagules remained unplanted, i.e., depos-
ited untethered in a horizontal position. Deployment of
unplanted propagules was conducted to simulate depo-
sition by tides in vegetated or unvegetated plots, i.e.,
propagules were laid gently in a horizontal position on
the herbaceous plant canopy (or bare ground) during an
ebb tide. Survival and condition (unrooted or rooted
and upright) of seedlings was determined at 6-, 12-, and
24-month intervals. Cumulative leaf production on
established seedlings was used as an index of growth.
The data were analyzed as a split-plot with patch type
(S. portulacastrum, D. spicata, and bare ground) as the
main plot and treatment (planted, unplanted) as the
subplot (n?3). Net effects of patches and effects of post-
establishment conditions (edaphic) and planting treat-
ment were quanti?ed by making comparisons indicated
in Fig. 2. Net effects of herbaceous patches were
estimated by calculating differences in performance
between unplanted seedlings in each patch type with
unplanted seedlings placed on bare soil. Post-establish-
ment (e.g., edaphic) effects were estimated by calculating
differences in performance between planted seedlings in
each patch type with planted seedlings placed on bare
soil (i.e., without trapping and structural support
effects). Effects of arti?cial planting were estimated by
calculating differences in performance between planted
and unplanted seedlings in each patch type.
The magnitudes of all effects were standardized to
allow comparison across performance parameters and
over time. The standardization procedure followed that
described previously (Holzapfel and Mahall 1999) based
on recommendations for treatment comparisons used in
meta-analysis (Gurevitch and Hedges 1993). Each effect
was calculated by subtracting the means of the
performance parameters and dividing by the pooled
standard deviation:
d ? X1  X2
sp
J ?1?
sp ?
???????????????????????????????????????????????????????????
?N1  1??s1?2 ? ?N2  1??s2?2
N1 ? N2  2
s
?2?
where d is the effect size, X is the mean of group 1 or 2, sp
is the pooled standard deviation of groups 1 and 2, s is
the standard deviation of groups 1 or 2, and N is the
total number of individuals in groups 1 or 2. We
corrected for sample-size bias by multiplying the term J:
J ? 1 3
4?N1 ? N2  2?  1
 
: ?3?
The variance associated with each standardized effect
was then calculated as recommended (Gurevitch and
Hedges 1993):
FIG. 2. Experimental design to assess effects of herbaceous patches dominated by Sesuvium portulacastrum and Distichlis spicata
and planting treatment (planted, unplanted) on establishment, survival, and growth of Rhizophora mangle seedlings. Net, post-
establishment (edaphic), and arti?cial planting effects were calculated as shown adjacent to each patch type?treatment combination.
Net effects of herbaceous patches were estimated by calculating differences in performance between unplanted seedlings in each
patch type and unplanted seedlings placed on bare soil (UD  UB, US  UB). Post-establishment (e.g., edaphic) effects were
estimated by calculating differences in performance between planted seedlings in each patch type and planted seedlings placed on
bare soil (i.e., without trapping and structural support effects; PD PB, PS PB). Effects of arti?cial planting were estimated by
calculating differences in performance between planted and unplanted seedlings in each patch type (PS US, PD UD).
KAREN L. MCKEE ET AL.1682 Ecological Applications
Vol. 17, No. 6
v ? N1 ? N2
N1N2
? d
2
2?N1 ? N2?
?4?
where v is the variance.
RESULTS
Initial condition and vegetative colonization
of the clear-cut area
The clear-cut area was initially completely devoid of
vegetation (see Plate 1), except for three mature A.
germinans trees left by the cutting crew. By January 1992
when the study was initiated, there was ;20% vegetative
cover, and this gradually increased to 40% over the next
three years (Fig. 3). The herbaceous vegetation was
initially dominated by Batis maritima (see Plate 1),
which increased in percent cover until January 1995.
Although D. spicata (6?7% cover) and S. portulacastrum
(,1% cover) were present in some plots, they did not
increase signi?cantly during this observation period.
Mangrove seedling density was low, but increased
signi?cantly at a rate of 0.03 seedlingsm2yr1 (F5,37
? 4.95, P ? 0.001). Seedlings of all mangrove species
recruited at similar rates, although total density of L.
racemosa was lower overall compared to R. mangle and
A. germinans (F2,41? 2.87, P? 0.068; Fig. 4). Mangrove
seedlings established naturally in areas devoid of
vegetation as well as in areas with herbaceous cover.
FIG. 3. Changes from 1992 to 1995 in plant and soil conditions measured in replicate plots across the study site where
mangroves were clear-cut in 1991: percent cover of herbaceous vegetation; density of mangrove seedlings; soil redox potential (Eh)
at two depths (2 and 16 cm); sul?de concentrations; porewater salinity; and pH. measurements were made at intervals until January
1995, when the site was partially buried by dredge material. Values are mean 6 SE (n ? 16 replicate plots per sample date).
September 2007 1683FACILITATION OF MANGROVES
However, no spatial pattern in mangrove establishment
was obvious during early recolonization of the site.
Edaphic conditions in the plots changed signi?cantly
(Fig. 3). Soil redox potential, Eh, which indicates the
degree of soil oxidation/reduction and re?ects soil
waterlogging, remained low at a 16 cm depth, but
?uctuated seasonally near the soil surface. Porewater pH
and concentrations of sul?de gradually declined over
time. Porewater salinity ?uctuated during the ?rst two
years after clear-cutting, mainly between wet/cool
(January) and dry/hot (July) seasons, but showed no
consistent trend over time. Signi?cant correlations
between herbaceous plant cover and soil Eh (R ? 0.66,
P ? 0.0017) and sul?de (R ? 0.55, P ? 0.0126) were
found, with the soil becoming less reducing over time as
herbaceous cover increased.
When part of the clear-cut area was back?lled in 1995,
further monitoring of the original plots was not possible.
The area not impacted by back?lling continued to
develop, however, and in February 2001 it contained
FIG. 4. Structural characteristics of herbaceous patches dominated by either Sesuvium portulacastrum or Distichlis spicata
determined in 2001. Sampling positions occurred across each patch from the edge to interior. Values are mean6 SE (n?3 replicate
patches per patch type [each position associated with each patch is the mean of three replicate measurements]).
KAREN L. MCKEE ET AL.1684 Ecological Applications
Vol. 17, No. 6
numerous monospeci?c patches of herbaceous vegeta-
tion, either D. spicata or S. portulacastrum. Although B.
maritima was still present in some spots, it had decreased
in abundance and did not form distinct patches.
Structure of herbaceous patches and associated
soil physicochemistry
Herbaceous patches were typically circular in shape
and appeared as islands of vegetation surrounded by
bare substrate. Patches ranged in size from 8 to 24 m2
and were distributed throughout the cleared area that
had not been back?lled (Plate 1, Fig. 1). Some patches
had coalesced to form larger vegetated areas, but ;30%
of section 2 remained unvegetated. The transition from
patch to surrounding area was abrupt, with little or no
decrease in vegetative cover near the patch edge.
The morphological traits of the two herbaceous
species differed substantially (Fig. 4). Distichlis spicata,
a perennial grass with spike-like shoots, was ;30 cm tall
and occurred in densities of 600?1200 culms/m2.
Sesuvium portulacastrum, a succulent groundcover, had
stem densities similar to that of D. spicata, but was
considerably shorter in stature (10?15 cm), with a more
prostrate growth form. Stem density of both herbaceous
species was greatest at the patch edge (F2,20 ? 7.43, P 
0.01). Rooting depth of the two species also differed,
with that of D. spicata three times greater than S.
portulacastrum (F1,2? 213, P  0.01). Both shoot height
and rooting depth of each species were relatively
consistent across the patches.
Signi?cant differences in several soil factors were
found between patches and bare ground and between
patch types (Fig. 5). Soil Eh (2 cm depth) was
signi?cantly higher in patches than in unvegetated areas
(F4,30? 3.46, P  0.05). At 16 and 32 cm depths, Eh was
less variable, with no signi?cant differences between
patch species or between patch and bare ground. The
depth pro?les of soil Eh, however, were closely
associated with the rooting depths of the two species
(Fig. 6). Soil temperature also exhibited spatial variation
and was on average 58?88C lower inside patches
compared to bare soil (P  0.0001; Fig. 5). This spatial
pattern was most distinct at a 1-cm depth where
temperature decreased from outside to the interior of
patches (;68C for D. spicata and ;38C for S.
portulacastrum; Fig. 5). Salinity in unvegetated areas
was typically high (45?), but was lower inside D.
spicata patches (,40?; F4,30 ? 5.88, P  0.01; Fig. 5).
PLATE 1. Ground view of study site in (A) July 1992 just after clearcutting and (B) February 2001 showing patches of three
herbaceous species colonizing the clearcut area: (1) Batis maritima, (2) Sesuvium portulacastrum, and (3) Distichlis spicata. Uncut
forest can be seen in the background. (C, D) Rhizophora mangle (red mangrove) seedlings growing in (C) D. spicata and (D) S.
portulacastrum patches. Photo credits: K. McKee.
September 2007 1685FACILITATION OF MANGROVES
In contrast, the salinity in S. portulacastrum patches (47?
51?) was similar to adjacent bare soil. Soil bulk density
also decreased from outside to the interior of patches
(0.16?0.13 g/cm3; F4,30 ? 3.52, P  0.05), but was
generally low overall. Soil moisture content (83?85%)
and organic matter content (71?73%) varied slightly
from outside to interior patches, but showed no
signi?cant differences between patch types or with
position. Soil N content (;20 mg/g soil) was similar
between patches and spatial positions.
Density, size, and vigor of natural mangrove recruits
in patches
By February 2001, three mangrove species had
established naturally in the herbaceous patches and in
adjacent bare areas of the clear-cut site, but the density
of mangroves was several-fold higher inside compared
to outside patches (Fig. 7). Density of R. mangle was not
different between S. portulacastrum and D. spicata
patches (1 df contrast: t ratio? 0.21, P . 0.05), whereas
FIG. 5. Variation in selected soil variables across herbaceous patches dominated by either Sesuvium portulacastrum or Distichlis
spicata determined in 2001. Sampling positions occurred across each patch from ;1 m outside where no vegetation occurred (bare
soil) and at different soil depths (as indicated for each variable). Values are mean 6 SE (n?3 replicate patches per patch type [each
position associated with each patch is the mean of three replicate measurements]).
KAREN L. MCKEE ET AL.1686 Ecological Applications
Vol. 17, No. 6
that of A. germinans was higher in D. spicata patches (1
df contrasts: t ratio ? 13.80, respectively, P  0.0001).
Density of L. racemosa was also higher in D. spicata
patches, although the difference was not signi?cant (1 df
contrast: t ratio? 1.62, P . 0.05). In addition to higher
densities of recruits, the height, vigor, and reproductive
output of mangroves were also greater in patches
compared to bare ground. Maximum height and vigor
of R. mangle in D. spicata patches was greater than
adjacent bare soil (Fig. 8; 1 df contrasts: t ratio?3.00
and 4.62, P  0.05). However, height and vigor of R.
mangle was lower in S. portulacastrum patches and not
signi?cantly different from that in bare soil. Approxi-
mately 20% of R. mangle individuals within patches were
reproductive, whereas none growing outside patches
reproduced during the study (Fig. 8). However, there
were no signi?cant differences in R. mangle reproduction
between S. portulacastrum and D. spicata patches (1 df
contrast: t ratio ? 0.63 and 1.28, P . 0.05, number of
FIG. 6. Variation in soil redox potential with depth in herbaceous patches dominated by either Sesuvium portulacastrum or
Distichlis spicata determined in 2001. Values are mean 6 SE (n is the mean of all within-patch positions [4]3 replicate patches [3]
per patch type ? 12). The shaded area shows rooting depth for each species.
FIG. 7. Density of mangrove recruits (seedlings and saplings) in patches dominated by either Sesuvium portulacastrum or
Distichlis spicata compared to bare soil adjacent to each patch determined in 2001. Mangrove species are Rhizophora mangle (RM),
Avicennia germinans (AG), and Laguncularia racemosa (LR). Values are mean 6 SE (n? 3 replicate patches per patch type [each
position associated with each patch is the mean of three replicate measurements]).
September 2007 1687FACILITATION OF MANGROVES
reproductive trees and number of fruit per tree,
respectively).
Qualitative observations indicated that herbaceous
patches contained self-regenerating colonies of man-
groves with reproductive saplings and numerous seed-
lings (,10 node class). An analysis of node class
distribution of R. mangle growing in patches and bare
ground showed differences in population structure,
which agreed with the observations of reproductive
output (Fig. 9). Node class distribution of mangroves
growing on bare ground followed a normal distribution
with a peak in the 20?24 node class, whereas the pattern
for herbaceous patches was skewed due to high numbers
in the 0?4 or 5?9 node classes in D. spicata (10
individuals) or S. portulacastrum (9 individuals) patches,
respectively.
Effects of patches on trapping, establishment,
and growth of mangroves
Transplantation of R. mangle propagules into patches
and bare areas indicated effects of herbaceous vegeta-
tion on trapping, vertical orientation, rooting rate, and
subsequent growth of established seedlings (Fig. 10a?d).
Arti?cial planting of mangrove propagules not only
prevented removal by tides but also promoted rapid
rooting. All arti?cially planted propagules became
rooted within the ?rst six months and survived for two
years in bare areas and in S. portulacastrum patches;
four planted seedlings in D. spicata patches were killed
by crabs, but the remainder rooted and survived (Fig.
10a). The majority (;75%) of unplanted propagules
placed in bare areas was washed away by tides, and
those few that were retained either never rooted or never
achieved a vertical axis. Many of the unrooted and/or
prostrate seedlings exhibited signs of sunburn and
desiccation of the hypocotyl. Approximately 40?55%
of unplanted propagules deployed into patches were also
removed by tidal action, but many became rooted and
ultimately achieved a vertical orientation (43?57%). All
of the propagules surviving in S. portulacastrum patches
rooted in a prostrate position, but ultimately became
upright through curvature of the hypocotyl. However,
propagules placed in D. spicata patches slipped into a
vertical position (with the radical end touching the soil)
soon after deployment and remained so through
structural support by the stiff shoots of this grass.
Long-term survival was most strongly affected by
FIG. 8. Maximum height and relative growth index
(number of growing points per node on the main shoot axis)
of Rhizophora mangle recruits in patches dominated by either
Sesuvium portulacastrum or Distichlis spicata compared to bare
soil adjacent to each patch determined in 2001. The percentage
of reproductive mangroves in each patch type is indicated by
inset numbers. Values are mean 6 SE (n ? 3 replicate patches
per patch type [each position associated with each patch is the
mean of three replicate measurements]).
FIG. 9. Frequency distribution of node range for Rhizo-
phora mangle recruits in patches dominated by either Sesuvium
portulacastrum or Distichlis spicata compared to bare soil
determined in 2001. Values are the mean of numbers of
individuals in each node class and patch type. Nodes were
counted on all seedling and saplings in each patch and adjacent
bare soil plots (n ? 3 replicate patches per patch type [each
position associated with each patch is the mean of three
replicate measurements]).
KAREN L. MCKEE ET AL.1688 Ecological Applications
Vol. 17, No. 6
arti?cial planting; percentage of survival of planted
propagules was 80?100% regardless of location (Fig.
10a). In contrast, survival of unplanted propagules was
,60%, with lowest values occurring in bare areas.
Comparison of standardized effects indicated that
planting had a stronger positive effect on mangrove
survival in both S. portulacastrum and D. spicata patches
than did post-establishment conditions (Fig. 10c).
Although both herbaceous species had a strong effect
on mangrove survival, the net effect of S. portulacastrum
was slightly higher than D. spicata (Fig. 10c).
Cumulative leaf production showed how growth of
surviving mangrove seedlings was affected by patch type
(Fig. 10b). The seedlings planted in S. portulacastrum
produced more leaves than in any other treatment
combination (1 df contrast: t ratio ? 2.78, P ? 0.0066)
and signi?cantly more than unplanted seedlings (1 df
contrast: t ratio ?3.44, P ? 0.0009), which fared no
better than those deployed to bare sediment (Fig. 10b).
Distichlis spicata also promoted growth of R. mangle
seedlings, regardless of whether they were planted or not
(1 df contrast: t ratio? 2.07, P? 0.041). Bare areas were
FIG. 10. Survival and growth (cumulative leaf production) of Rhizophora mangle seedlings transplanted to patches dominated
by either Sesuvium portulacastrum or Distichlis spicata or to bare soil. Propagules were either planted or placed unplanted and
unrestrained in each position, and the positions were then monitored for two years. Values are means 6 SE (n ? 3 replicates per
treatment [planted, unplanted] 3 location [Sesuvium, Distichlis, bare] combination). Left panels show absolute values, and right
panels show standardized effects of herbaceous patches over time. See Fig. 2 and Methods for calculations of standardized effects
and variances.
September 2007 1689FACILITATION OF MANGROVES
least conducive to the growth of seedlings, regardless of
planting treatment (1 df contrast: t ratio ?2.13, P ?
0.0359). Although both herbaceous species had a
positive net effect on mangrove growth, D. spicata had
a stronger effect than did S. portulacastrum (Fig. 10d).
Planting increased mangrove leaf production in S.
portulacastrum, but had no effect in D. spicata. Post-
establishment effects on growth became more positive
over two years in S. portulacastrum, whereas the positive
effect of D. spicata became weaker over time (Fig. 10d).
DISCUSSION
Facilitative interactions have been identi?ed in a
variety of plant communities, either through spatial
associations of species or ?eld or laboratory experiments
(Callaway 1995). Clear demonstration of facilitation and
its potential value in restoration practices is not always
straightforward, however. An experimental approach is
required to demonstrate the interference type of plant
interaction (Olofsson et al. 1999), and the same is
certainly true of plant facilitation. However, of 179
studies of plant facilitation, only 20% employed a
manipulative ?eld experiment that produced evidence
of the mechanism(s) involved (Callaway 1995). The
objective of most ?eld manipulations has been to
separate effects of the benefactor species from spatial
variation in abiotic factors. This approach typically
involves examination of the bene?ciary species perfor-
mance (1) with and without the natural presence of the
benefactor species and/or (2) after removal of the
benefactor species (e.g., manipulation of canopy or
roots). These types of manipulations are required to
demonstrate a cause and effect relationship and, further,
to determine how the benefactor species is improving the
performance of the bene?ciary species. However,
interpretation of such studies may be problematic due
to failure of the manipulation to completely eliminate
the in?uence of the benefactor (e.g., removal of root
in?uence, but not canopy [Cook and Ratcliff 1984]) or
to unintended disturbance of the soil, which confounds
the effects of benefactor removal (Callaway 1995). A
preferred approach is to introduce the bene?ciary
species as a seed or transplant into areas with and
without the benefactor to examine effects on the early
life history of the target species. This approach assumes
that positive interactions are more prevalent in highly
disturbed landscapes (e.g., during secondary succession)
in which the benefactor is the precursor to the existence
of the bene?ciary (Brooker and Callaghan 1998).
In our investigation, herbaceous vegetation clearly
facilitated recolonization of mangroves in a disturbed
forest. This mangrove ecosystem was characterized by
low sediment supply and low nutrients, low elevations
(allowing overwashing by tides), high temperatures,
hypersalinity, and strongly reducing soils with accumu-
lation of plant phytotoxins, such as sul?de (McKee et al.
1988, McKee 1993b, 1995b) (Fig. 3). Consequently,
facilitation of mangrove recruitment might involve
alteration of one or more of these stressful factors.
Patches of herbaceous vegetation may promote man-
grove recolonization of disturbed areas by (1) trapping
dispersing propagules, (2) promoting establishment and
rooting, and/or (3) enhancing survival and growth of
seedlings through amelioration of physicochemical
conditions. Differences in herbaceous patch structure
and soil physicochemical factors demonstrated the
heterogeneity of conditions at the study site. Although
spatial patterns of mangrove recruitment were not
evident initially (at three years post-clear-cut), they did
appear later after monospeci?c patches of D. spicata and
S. portulacastrum developed.
Where species-speci?c facilitation exists, the differ-
ences among potential bene?cial species may be due to
variation in ability to generate a positive effect, but also
to relative differences in negative effects (e.g., competi-
tion or allelopathy [Mahall and Callaway 1992]). By
combining observations of natural recruitment rates and
patterns with experimental manipulations, we were able
to evaluate the various ways herbaceous vegetation
controlled mangrove recolonization of the disturbed
forest at Twin Cays.
Facilitation of seedling establishment through propagule
trapping and structural support
Although mangrove seedlings were readily dispersed
to the clear-cut site from nearby sources and began
establishing within a year after disturbance, recruitment
was slow for the ?rst three years. Batis maritima, which
was initially dominant, did not appear to have an effect
on early mangrove recruitment rates or patterns, as
reported for degraded forests in south Florida (Mil-
brandt and Tinsley 2006). However, this disagreement
may re?ect differences in mangrove species, environ-
mental conditions, timing, presence of other herbaceous
species, or other factors in the two locations and points
to the need for site-speci?c information. Nine years post-
clear-cut, mangrove recruitment had increased, but
more so in herbaceous patches of D. spicata or S.
portulacastrum. Tidal ?uctuation readily removes any
unrestrained and buoyant object such as propagules
(McKee 1995a). The higher densities of natural recruits
in patches compared to bare ground may re?ect
propagule trapping, combined with the stranding
requirement for propagule rooting (McKee 1995a).
Experimental dispersal of R. mangle propagules showed
that most unplanted R. mangle propagules were washed
away from bare ground, but about half were retained in
the patches.
Greater height and stiffer structural support provided
by the D. spicata canopy promoted greater retention of
both natural and experimental mangrove recruits.
Densities of A. germinans and L. racemosa recruits were
in fact higher in D. spicata compared to S. portulacas-
trum patches, possibly due to trapping of these smaller
and more buoyant propagules (McKee 1995a). The
densities of natural R. mangle recruits were not
KAREN L. MCKEE ET AL.1690 Ecological Applications
Vol. 17, No. 6
signi?cantly different between the two patch types,
however. Also, retention of unplanted R. mangle
propagules was about the same in D. spicata and S.
portulacastrum patches. It is possible that the relative
trapping in?uence of these species is less important for
larger, heavier propagules such as R. mangle. However,
the aboveground structure of D. spicata patches clearly
provided structural support for R. mangle, while
canopies (30?35 cm height) were not tall enough to
shade the seedlings (.30 cm height). Since an upright
posture promotes long-term survival of R. mangle
(McKee 1995b), structural support may be an additional
factor affecting recruitment. In contrast, all unplanted
survivors in S. portulacastrum patches established in a
prostrate position and later curved upward. A shorter,
weaker seedling axis may explain slower growth of
unplanted seedlings in S. portulacastrum (Fig. 10d).
Such species-speci?c differences in nurse plants could be
exploited to assist in seedling success when arti?cial
planting of mangroves is not feasible.
Facilitation through amelioration
of physicochemical conditions
Some species can act as autogenic engineers by
altering the physical state of an environment and
modifying a resource that bene?ts another species (Jones
et al. 1994). The effects may be species-speci?c or
provide a broad range of advantages to several related
species. Since our study was not designed to experimen-
tally show patch effects on physicochemical conditions,
we cannot say with certainty whether the herbaceous
vegetation altered the environmental conditions of the
clear-cut site or plants simply established in areas with
different edaphic features. However, the results are most
consistent with the former explanation. Soil conditions
were stressful across the entire clear-cut area in January
1992 (Fig. 3). Over the next three years, soil Eh increased
and sul?de decreased along with increasing plant cover.
Comparison of patches dominated by D. spicata and
S. portulacastrum provided additional insight into how
herbaceous vegetation might in?uence environmental
conditions. The taller canopy of D. spicata patches
shaded the soil and reduced soil temperatures more than
S. portulacastrum, and these changes were associated
with lower salinities (Fig. 5). Also, the close association
between rooting depth and soil redox potential (Fig. 6)
suggests that the herbaceous vegetation was oxidizing
the sediment through oxygen release from roots, as
demonstrated for salt marsh species (Mendelssohn and
Morris 2000) and mature mangroves (McKee et al. 1988,
McKee 1993b).
Comparison of experimental mangrove recruits also
indicated species-speci?c facilitation by the herbaceous
vegetation. When net effects were compared using
standardized units, S. portulacastrum was found to be
better at promoting mangrove establishment and long-
term survival, but D. spicata enhanced growth of
survivors (Fig. 10c, d). A comparison of planting effects
in the two patch types indicates how much each patch
species promoted mangrove survival and growth
through trapping and structural support. Distichlis
spicata apparently promoted growth of unplanted
mangroves through physical support. Effects of post-
establishment conditions on mangrove growth were not
strong initially (at six months), but became increasingly
positive over time as the plants became larger (Fig. 10d).
This ?nding may re?ect lower temperature and salinity
and less reducing soils in patches compared to bare
ground (Fig. 5).
Facilitation and competition can occur simultaneous-
ly between co-growing species (Holmgren et al. 1997),
and this may be the case for mangroves and herbaceous
associates. Effects of post-establishment conditions on
mangrove growth were more positive for S. portulacas-
trum compared to D. spicata, especially over time (Fig.
10d). Both S. portulacastrum and D. spicata ameliorated
stressful soil conditions (Figs. 4, 5). However, since S.
portulacastrum roots were relatively shallow, they were
not competing for nutrients at lower depths. Distichlis
spicata had deeper root systems, overlapping that of
mangrove recruits (up to 50 cm deep). Thus, even
though D. spicata patches had lower soil temperatures,
lower salinities, and higher redox potentials than bare
soil, competition for nutrients may have partially
counteracted the positive effects.
Establishment of herbaceous vegetation may be an
important ?rst step in site preparation for mangrove
revegetation where stressful conditions prevail. Initial
colonization of the clear-cut site by B. maritima may
have ameliorated soil conditions, but had little direct
effect on trapping and establishment of mangrove
seedlings. Distichlis spicata and S. portulacastrum, which
established later, increased propagule trapping and also
ameliorated stressful conditions. Herbaceous patches
additionally promoted growth and reproduction by
mangrove recruits (Figs. 8?10). Many patches contained
reproductive saplings surrounded by numerous propa-
gules and seedlings; this pattern was not seen in
unvegetated areas (Fig. 9). This pattern likely re?ects
trapping not only of propagules dispersed from sur-
rounding areas, but of patch-borne propagules. Rapid
development of self-regenerating colonies of mangroves
may be important in areas in which dispersal from
mature forests is limited (Imbert et al. 2000).
Implications for mangrove restoration
Although many restoration projects involve planting
of mangroves (Field 1996), little information exists on
factors determining seedling survival and subsequent
growth rates in degraded, restored, or created mangrove
forests (Day et al. 1999, Elster 2000, Bosire et al. 2003,
Milbrandt and Tinsley 2006). The ?nding that some
herbaceous species can have a positive effect on
mangrove recruitment suggests that natural regeneration
of large disturbed areas may occur much faster if
bene?cial species are present (Milbrandt and Tinsley
September 2007 1691FACILITATION OF MANGROVES
2006). However, our work shows that effects of
herbaceous vegetation on mangrove regeneration may
depend upon the species involved as well as the factors
affecting mangrove recruitment in each geographic area.
Consequently, nurse species might be identi?ed based on
positive features such as shoot structure affecting
shading and amelioration of soil temperature and
salinity, propagule trapping and retention, and root
effects on soil aeration and competition for nutrients or
water. Additionally, such comparisons may be improved
using standardized effects as illustrated here. Quanti?-
cation of standardized effects showed: (1) the net effect
of each herbaceous species on mangrove survival and
growth, (2) effects of pre- and post-establishment factors
associated with each patch type, and (3) the need for
arti?cial planting to enhance growth or survival.
Our work provides an example of how facilitation
might promote mangrove regeneration and how such
effects might be assessed. However, application to
restoration projects will require site-speci?c information
as well as an understanding of the mechanisms involved.
Although we directly compared the effects of only two
herbaceous species, extensive lists of mangrove associ-
ates worldwide include more than 25 plant families and
more than 45 genera (Tomlinson 1994). Growth forms
include grasses, rushes, sedges, succulents, forbs, and
ferns. This large pool of mangrove associates, many of
which are cosmopolitan, indicates the potential for
selection of bene?cial species in other geographic
regions. Future work should develop a broader picture
of how herbaceous associates interact with mangroves to
in?uence forest regeneration rates and patterns in a
variety of environmental conditions. This type of
information will be increasingly important as human
pressures on the coastal zone grow and the need for
better management and restoration strategies increases.
ACKNOWLEDGMENTS
We thank Klaus Ru?tzler, Mike Carpenter, and the Caribbe-
an Coral Reef Ecosystems Program (CCRE), Smithsonian
Institution, for logistical support and use of the Carrie Bow Cay
?eld station. We also thank Andrea Anteau, Tara Martinez,
Greg Glowacki, Richard Hines, Tommy McGinnis, and Todd
Herbert for laboratory and ?eld support, and Beth Middleton,
Ed Prof?tt, Darren Johnson, and Jarita Davis for comments on
the manuscript. This work was funded by a NSF Biocomplexity
in the Environment grant, DEB-9981483. This is CCRE
contribution number 788, supported in part by the Hunterdon
Oceanographic Research Fund.
LITERATURE CITED
Alongi, D. M. 2002. Present state and future of the world?s
mangrove forests. Environmental Conservation 29:331?349.
Barbier, E. B. 2006. Natural barriers to natural disasters:
replanting mangroves after the tsunami. Frontiers in Ecology
and the Environment 4:124?131.
Bertness, M. D. 1991. Interspeci?c interactions among high
marsh perennials in a New England salt marsh. Ecology 72:
125?137.
Bertness, M. D., and S. Hacker. 1994. Physical stress and
positive associations among marsh plants. American Natu-
ralist 144:363?372.
Bertness, M. D., and G. H. Leonard. 1997. The role of positive
interactions in communities: lessons from intertidal habitats.
Ecology 78:1976?1989.
Bosire, J. O., F. Dahdouh-Guebas, J. G. Kairo, and N.
Koedam. 2003. Colonization of non-planted mangrove
species into restored mangrove stands in Gazi Bay, Kenya.
Aquatic Botany 76:267?279.
Brooker, R. W., and T. V. Callaghan. 1998. The balance
between positive and negative plant interactions and its
relationship to environmental gradients: a model. Oikos 81:
196?207.
Bruno, J. F., J. J. Stachowicz, and M. D. Bertness. 2003.
Inclusion of facilitation into ecological theory. Trends in
Ecology and Evolution 18:119?125.
Callaway, R. M. 1995. Positive interactions among plants.
Botanical Review 61:306?349.
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.
Connell, J. H. 1983. On the prevalence and relative importance
of interspeci?c competition: evidence from ?eld experiments.
American Naturalist 122:661?696.
Cook, S. J., and D. Ratcliff. 1984. A study of the effects of root
and shoot competition on the growth of green panic
(Panicum maximum Var. trichoglume) seedlings in an existing
grassland using root exclusion tubes. Journal of Applied
Ecology 21:971?982.
Day, S., W. J. Streever, and J. J. Watts. 1999. An experimental
assessment of slag as a substrate for mangrove rehabilitation.
Restoration Ecology 7:139?144.
Day, T. A., and R. G. Wright. 1989. Positive plant spatial
association with Eriogonum ovalifolium in primary succession
on cinder cones?seed-trapping nurse plants. Vegetatio 80:
37?45.
Dobson, A. P., A. D. Bradshaw, and A. J. M. Baker. 1997.
Hopes for the future: restoration ecology and conservation
biology. Science 277:515?522.
Egerova, J., C. E. Prof?tt, and S. E. Travis. 2003. Facilitation
of survival and growth of Baccharis halimifolia L. by Spartina
alterni?ora Loisel. in a created Louisiana salt marsh.
Wetlands 23:250?256.
Elster, C. 2000. Reasons for reforestation success and failure
with three mangrove species in Colombia. Forest Ecology
and Management 131:201?214.
Elster, C., L. Perdomo, and M. L. Schnetter. 1999. Impact of
ecological factors on the regeneration of mangroves in the
Cienaga Grande de Santa Marta, Colombia. Hydrobiologia
413:35?46.
Feller, I. C. 1995. Effects of nutrient enrichment on growth and
herbivory of dwarf red mangrove (Rhizophora mangle).
Ecological Monographs 65:477?505.
Feller, I. C., D. F. Whigham, 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 134:405?414.
Field, C. D. 1996. Restoration of mangrove ecosystems.
International Society for Mangrove Ecosystems, Okinawa,
Japan.
Flower, J. M. 2004. Experimental seedling at two decaying
mangrove sites: advantage for the protected areas of
Guadeloupe. Revue d?Ecologie?La Terre et La Vie 59:171?
180.
Goldberg, D. E., and A. M. Barton. 1992. Patterns and
consequences of interspeci?c competition in natural commu-
nities: a review of ?eld experiments with plants. American
Naturalist 139:771?801.
Gomez-Aparicio, L., R. Zamora, J. M. Gomez, J. A. Hodar, J.
Castro, and E. Baraza. 2004. Applying plant facilitation to
forest restoration: a meta-analysis of the use of shrubs as
nurse plants. Ecological Applications 14:1128?1138.
KAREN L. MCKEE ET AL.1692 Ecological Applications
Vol. 17, No. 6
Gurevitch, J., and L. V. Hedges. 1993. Meta-analysis:
combining the results of independent experiments. Pages
378?398 in S. M. Scheiner and J. Gurevitch, editors. Design
and analysis of ecological experiments. Chapman and Hall,
New York, New York, USA.
Holmgren, M., M. Scheffer, and M. A. Huston. 1997. The
interplay of facilitation and competition in plant communi-
ties. Ecology 78:1966?1975.
Holzapfel, C., and B. E. Mahall. 1999. Bidirectional facilitation
and interference between shrubs and annuals in the Mojave
Desert. Ecology 80:1747?1761.
Hunter, A. F., and L. W. Aarssen. 1988. Plants helping plants:
new evidence indicates that bene?cence is important in
vegetation. BioScience 38:34?40.
Imbert, D., A. Rousteau, and P. Scherrer. 2000. Ecology of
mangrove growth and recovery in the Lesser Antilles: state of
knowledge and basis for restoration projects. Restoration
Ecology 8:230?236.
Jones, C. G., J. H. Lawton, and M. Shachak. 1994. Organisms
as ecosystem engineers. Oikos 69:373?389.
Kaly, U. L., and G. P. Jones. 1998. Mangrove restoration: a
potential tool for coastal management in tropical developing
countries. Ambio 27:656?661.
Lewis, R. R. 2005. Ecological engineering for successful
management and restoration of mangrove forests. Ecological
Engineering 24:403?418.
Lindquist, E. S., and C. R. Carroll. 2004. Differential seed and
seedling predation by crabs: impacts on tropical coastal
forest composition. Oecologia 141:661?671.
Macintyre, I. G., M. A. Toscano, R. G. Lighty, and G. B.
Bond. 2004. Holocene history of the mangrove islands of
Twin Cays, Belize, Central America. Atoll Research Bulletin
510:1?16.
Maestre, F. T., S. Bautista, J. Cortina, and J. Bellot. 2001.
Potential for using facilitation by grasses to establish shrubs
on a semiarid degraded steppe. Ecological Applications 11:
1641?1655.
Mahall, B. E., and R. M. Callaway. 1992. Root communication
mechanisms and intracommunity distributions of two Mo-
jave desert scrubs. Ecology 73:2145?2151.
Martinez, M. L. 2003. Facilitation of seedling establishment by
an endemic shrub in tropical coastal sand dunes. Plant
Ecology 168:333?345.
McKee, K. L. 1993a. Determinants of mangrove species
distribution in neotropical forests: biotic and abiotic factors
affecting seedling survival and growth. Louisiana State
University, Baton Rouge, Louisiana, USA.
McKee, K. L. 1993b. Soil physicochemical patterns and
mangrove species distribution?Reciprocal effects? Journal
of Ecology 84:477?487.
McKee, K. L. 1995a. Mangrove species distribution and
propagule predation in Belize: an exception to the domi-
nance-predation hypothesis. Biotropica 27:334?345.
McKee, K. L. 1995b. Seedling recruitment patterns in a
Belizean mangrove forest: effects of establishment and
physico-chemical factors. Oecologia 101:448?460.
McKee, K. L., D. R. Cahoon, and I. C. Feller. 2007. Caribbean
mangroves adjust to rising sea level through biotic controls
on change in soil elevation. Global Ecology and Biogeogra-
phy, in press.
McKee, K. L., and P. L. Faulkner. 2000. Mangrove peat
analysis and reconstruction of vegetation history at the
Pelican Cays, Belize. Atoll Research Bulletin 468:46?58.
McKee, K. L., I. A. Mendelssohn, and M. W. Hester. 1988.
Reexamination of pore water sul?de concentrations and
redox potentials near the aerial roots of Rhizophora mangle
and Avicennia germinans. American Journal of Botany 75:
1352?1359.
Mendelssohn, I. H., and J. T. Morris. 2000. Eco-physiological
controls on the productivity of Spartina alterni?ora Loisel.
Pages 59?80 in M. P. Weinstein and D. A. Kreeger, editors.
Concepts and controversies in tidal marsh ecology. Kluwer
Academic, Dordrecht, The Netherlands.
Milbrandt, E. C., and M. N. Tinsley. 2006. The role of saltwort
(Batis maritima L.) in regeneration of degraded mangrove
forests. Hydrobiologia 568:369?377.
Olofsson, J., J. Moen, and L. Oksanen. 1999. On the balance
between positive and negative plant interactions in harsh
environments. Oikos 86:539?543.
Pugnaire, F. I., P. Haase, and J. Puigdefabregas. 1996.
Facilitation between higher plant species in a semiarid
environment. Ecology 77:1420?1426.
Rivera-Monroy, V. H., R. R. Twilley, D. Bone, D. L. Childers,
C. Coronado-Molina, I. C. Feller, J. Herrera-Silveira, R.
Jaffe, E. Mancera, E. Rejmankova, J. Salisbury, and E. Weil.
2004. A conceptual framework to develop long-term
ecological research and management objectives in the wider
Caribbean Region. BioScience 54:843?856.
Robinson, G. R., and S. N. Handel. 1993. Forest restoration on
a closed land?ll?rapid addition of new species by bird
dispersal. Conservation Biology 7:271?278.
Ru?tzler, K., and I. C. Feller. 1996. Caribbean mangrove
swamps. Scienti?c American 274:94?99.
SAS. 2002. JMP statistics and graphics guide. Version 5. SAS
Institute, Cary, North Carolina, USA.
Sherman, R. E., T. J. Fahey, and J. J. Battles. 2000. Small-scale
disturbance and regeneration dynamics in a neotropical
mangrove forest. Journal of Ecology 88:165?178.
Tomlinson, P. B. 1994. The botany of mangroves. Cambridge
University Press, Cambridge, UK.
Valiela, I., J. L. Bowen, and J. K. York. 2001. Mangrove
forests: one of the world?s threatened major tropical
environments. BioScience 51:807?815.
Woodroffe, C. D. 1995. Mangrove vegetation of Tobacco
Range and nearby mangrove ranges, central Belize barrier
reef. Atoll Research Bulletin 427:1?35.
Wright, R. M., D. W. Urish, and I. Runge. 1991. The
hydrology of a Caribbean mangrove island. Pages 170?184
in G. Cambers, editor. Coastlines of the Caribbean.
American Society of Civil Engineers, New York, New York,
USA.
September 2007 1693FACILITATION OF MANGROVES