Hurricane Impacts to Tree Islands in Arthur R. Marshall Loxahatchee National Wildlife Refuge, Florida
Cristina A. Ugarte, Laura A. Brandt, Stefani Melvin, Frank J. Mazzotti, and Kenneth G. Rice
Southeastern Naturalist, Volume 5, Number 4 (2006): 737–746
Full-text pdf (Accessible only to subscribers.To subscribe click here.)
Site by Bennett Web & Design Co.
2006 SOUTHEASTERN NATURALIST 5(4):737–746
Hurricane Impacts to Tree Islands in Arthur R. Marshall
Loxahatchee National Wildlife Refuge, Florida
Cristina A. Ugarte1,*, Laura A. Brandt2, Stefani Melvin2, Frank J. Mazzotti3,
and Kenneth G. Rice4
Abstract - Tree islands—small, wetland forest communities imbedded in a matrix of
freshwater marsh—characterize Arthur R. Marshall Loxahatchee National Wildlife
Refuge in South Florida. The establishment and spread of invasive exotic plant
species were hypothesized to alter tree-island communities and prolong recovery
times from hurricane disturbances. During the fall of 2004, two hurricanes, Frances
and Jeanne, caused damage to these tree islands. We examined the spatial extent of
damage to tree islands and tree species across the Refuge by sampling 74 islands.
Each tree island was assigned an overall damage rating based on both the openness of
the canopy and the type and quantity of damage received. Distance from the eye-wall
of the hurricanes, tree-island size, average tree height on the island, and relative
abundance of invasive exotic plants were examined as predictors of damage. Over
85% of the sampled tree islands had damage, with most of the damage occurring in
the center of the Refuge. Most tree islands were found to have moderate damage,
(i.e., snapped large branches and less than 50% canopy cover removed). Persea
palustris (swamp bays) had more snapped trunks than Ilex cassine (dahoon holly)
and Myrica cerifera (wax myrtle). Islands with larger trees had heavier damage than
islands with shrubs or smaller trees. Fifty-eight percent of the tree islands sampled
had either Lygodium microphyllum (lygodium) and/or Melaleuca quinquenervia
(melaleuca). The only island with severe damage had abundant lygodium that appeared
to have caused the entire canopy to collapse. These hurricanes present a
unique opportunity to investigate recovery patterns of tree islands in an ecosystem
impacted by invasive exotics. They also provide an opportunity to examine patterns
of spread and recruitment of lygodium and melaleuca.
Hurricanes are a natural process in the Everglades landscape. Under
natural conditions, most communities recover from disturbances; however,
fragmentation and other human induced stresses such as changes in water
management and introduction of exotic species may change how a community
responds to disturbances such as hurricanes. A first step in documenting
1School of Natural Resources and the Environment, University of Florida, Everglades
National Park Field Station, 40001 State Road 9336, Homestead, FL, 33034.
2US Fish and Wildlife Service, Arthur R. Marshall Loxahatchee National Wildlife
Refuge 10216 Lee Road, Boynton Beach, FL, 33437. 3Department of Wildlife
Ecology and Conservation, FLREC, University of Florida, 3205 College Avenue,
Fort Lauderdale, FL 33314-7799. 4United States Geological Survey, Florida Integrated
Science Center, Center for Water and Restoration Studies, University of
Florida Field Station, 3205 College Avenue, Fort Lauderdale, FL. 33314-7799.
*Corresponding author: email@example.com.
738 Southeastern Naturalist Vol. 5, No. 4
responses to disturbances is evaluating immediate impacts to the vegetation.
In fall 2004, we had an opportunity to conduct an evaluation of hurricane
damage to tree islands in the Arthur R. Marshall Loxahatchee National
Wildlife Refuge, FL.
On September 5 and 26, 2004, the eye-walls of hurricane Frances, a
Category-2 storm, and hurricane Jeanne, a Category-3 storm, passed through
northern Palm Beach County, FL within 12 km of each other. Hurricane
Frances produced hurricane-force winds that extended 85 miles from the
center of the eye, affecting the Arthur R. Marshall Loxahatchee National
Wildlife Refuge (Refuge), whose center was approximately 77 miles from
the center of the storm. Tropical-force and hurricane-force winds from
Jeanne affected mostly the northern portion of the refuge. Hurricane Jeanne
was the first Category-3 storm to make landfall on the east coast of Florida,
north of Palm Beach since 1899.
Of particular concern to Refuge staff were immediate and long-term
impacts to tree islands, areas of wetland forest communities embedded
within the marsh. Although these forests have been exposed to repeated
hurricanes over the years (Noel et al. 1995, Roman et al. 1994), the impact of
hurricane disturbance combined with current stressors that include hydrologic
modification and colonization of invasive exotics could exacerbate
wetland forest health and prolong recovery times (Loope et al. 1994, Roman
et al. 1994). Because hurricanes can widely scatter the seeds of winddispersed
plants (Oberbauer et al. 1996), large-scale disturbance from
hurricanes could provide an opportunity for wind dispersing exotics to
colonize new areas (Snitzer et al. 2005). The presence of exotics may
ultimately change a community’s “natural” successional trajectory (Loope
et al. 1994, Ogden 1992, Roman et al. 1994).
Control of exotic plants, particularly Lygodium microphyllum (Cav.) R.
Br. (lygodium) and/or Melaleuca quinquenervia (Cav.) S.T. Blake
(melaleuca) is a management priority at the Refuge. Recent systematic
reconnaissance flights (SRFs), which assess percent of exotics within 1-km
x 1-km grid cells, estimated that over 60% of the cells within the Refuge
contained lygodium or melaleuca (Woodmansee et al. 2005). How the
spread of these exotics will be influenced by the hurricanes is not clear.
However, there is concern that physical damage to tree islands (i.e., removal
of the canopy and increasing light levels) will make them prime colonization
sites for lygodium and melaleuca within these habitats.
The goals of this work were to 1) determine the spatial patterns of
hurricane effects to tree islands throughout the Refuge, 2) classify hurricane
damage to tree-island vegetation, 3) assess species-specific damage, and 4)
assess how certain characteristics (i.e., presence and abundance of exotics
on tree islands, tree height, and distance from the center of the eye of the two
storms) correlate to tree-island vegetation damage.
2006 C.A. Ugarte, L.A. Brandt, S. Melvin, F.J. Mazzotti, and K.G. Rice 739
The Arthur R. Marshall Loxahatchee National Wildlife Refuge (Refuge)
includes approximately 57,000 ha of northern Everglades ridge and slough
habitat which is a mosaic of open sloughs, wet prairies, sawgrass strands,
and tree islands. Tree islands are a signature landscape feature of the Refuge
and provide habitat for a variety of plants and animals (Brandt et al. 2003).
The majority of tree islands are small (< 0.13 ha) islands that formed from
floating peat mats that settled on the peat surface and became colonized with
bayhead vegetation. Tree canopies are composed primarily of Persea
palustris (Raf.) Sarg. (swamp bay), Ilex cassine L. (dahoon holly), and
Myrica cerifera L. (wax myrtle). Ferns (Acrosticum spp., Osmunda spp.),
Chrysobalanus icaco L. (cocoplum), and Cephalanthalus occidentalis L.
(button bush) occur in shrub and ground cover (Brandt et al. 2003).
In late October and early November 2004, tree islands were systematically
sampled to assess hurricane impacts. We used ArcGis 9.0 to place a
2-km x 2-km grid over the Refuge area, and used the vertices of 100 of the
2 x 2 cells within the area of the Refuge accessible by airboat as sample
points. Upon visiting a point, the closest tree island was sampled. If no tree
island occurred at the point, or the tree island was less than 10 m in width,
we sampled the nearest tree island greater than 10 m in width. We were
able to sample tree islands at 74 selected points. Of the other 26 sampling
points, 11 had no tree islands within at least 2 km, 4 were continuous dense
shrubs and inaccessible, and 11 were dominated by melaleuca, with no
typical tree island flora.
Tree islands were circled by airboat to assess damage and to collect
information on tree-island area using a global positioning system (Garmin
MAP 76™). Two scales of damage were assessed at each tree island: the
entire island and individual trees by species.
Each tree island was assigned an overall damage rating based on both the
openness of the canopy and the type and extent of damage received
Severe damage = uprooted trees/leaning over 90 degrees, with 75% open
Heavy damage = major trunk snap off, leaving 50 and < 75% open canopy.
Moderate damage = large side branches snapped and leaning trees, leaving
25 and < 50% open canopy.
Light damage = small branches down, leaves stripped, and leaning trees,
leaving 10 < 25% open canopy.
No damage = no visible damage and less than 10% open canopy.
Since tree height is often related to damage severity in other South
Florida wetland forests (Noel et al. 1995), each tree island was assigned a
canopy height category based on the majority of trees present: trees
740 Southeastern Naturalist Vol. 5, No. 4
(larger than 3 m), small trees (less than 3 m) and shrubs (primarily willow
heads with an open canopy). In addition, we recorded the exotic plant
species present and estimated their cover as: abundant (greater than 50%
cover of the tree island), moderate (between 10–50%), or few (less than
10%). Individually damaged trees were counted and categorized into one
of three damage classes by species: uprooted/leaning; main trunk snapped
off; and side branch off. Surface maps of damage and exotics were created
using the spatial analyst module in ArcGIS 9.0 with an inverse
Because there were so few islands in either the no damage or severe
damage category, we collapsed damage types into two categories for statistical
analyses: damage—which included the severe, heavy, and moderate
categories; and no damage—which included the no damage and light categories.
The hurricane tracks were obtained from NOAA, and a distance
measurement from each island to the eye-wall was calculated using ArcGIS.
We used a forward stepwise logistic regression analysis (SPSS ver. 11.5), to
assess if damage or no damage was a function of distance to the eye-wall of
hurricanes Jeanne or Frances, tree height, island size, and abundance of
melaleuca and lygodium separately.
We sampled 74 tree islands throughout the refuge, ranging in size from
0.03 ha to 1.4 ha (Fig. 1) and tallied 364 damaged trees. A little over one
third of the islands had moderate damage (38.5%), while 24% had heavy
damage, 23% had light damage, 14% had no damage, and only one had
severe damage (Fig. 2).
Of the 364 trees we tallied, the most common damage type across these
islands was major trunk snapping (39%), followed by uprooted/leaning trees
(34%; Table 1). A greater proportion (61%) of swamp bay, which was the
tallest of the tree species sampled, had major trunk snaps compared to
dahoon holly which was usually leaning/uprooted (Fig. 3). Wax myrtle was
most often leaning/uprooted. Melaleuca had very little damage relative to
Of the 74 islands sampled, 58% had either melaleuca or lygodium
present, and 19% had both species. Melaleuca was present on 41% of the
islands while lygodium was present on 38% (Table 2). The most severely
Table 1. A.R.M. Loxahatchee National Wildlife Refuge 2004 hurricane damage counts to
common tree island tree species. S = swamp bay, D = dahoon holly, W = wax myrtle, M =
melaleuca, C = cypress, W = willow, B = button bush, and Co = cocoplum.
Damage type S D W M C W B Co Total %
Uprooted/leaning 25 47 54 0 2 3 1 1 34.0
Major trunk snap 73 40 33 3 1 3 0 2 39.0
Side branch snap 21 28 43 1 0 4 0 0 24.0
Total % stems damaged 32 29 33 0.01 0.8 3 0.3 0.8
2006 C.A. Ugarte, L.A. Brandt, S. Melvin, F.J. Mazzotti, and K.G. Rice 741
damaged island was heavily infested with lygodium. There was no damage
gradient throughout the Refuge, and distance from the hurricane center was
not a significant predictor in our model. It was apparent, however, that the
most damage occurred in the central part of the Refuge (Fig. 2).
The only significant explanatory variable in the forward stepwise logistic
regression model was tree height, which explained 32% of the variation in
the model. Islands with larger trees were five times more likely to obtain
heavy damage than islands with smaller trees or shrubs (Table 3).
Figure 1. Distribution of sizes (ha) of tree islands in A.R.M. Loxahatchee National
Wildlife Refuge sampled for hurricane impacts in October/November 2004. Mean
size of 74 tree islands = 0.18 with st. dev.= 0.21 ha.
Table 2. A.R.M. Loxahatchee National Wildlife Refuge 2004 counts and relative abundance of
most common exotics, melaleuca and lygodium, on 74 sampled tree islands.
Abundant Moderate Few
Species Presence (< 50% canopy) (50–10%) (< 10%)
Lygodium 38% (n = 28) 12.0% (n = 9) 9.0% (n = 7) 16% (n = 12)
Melaleuca 41% (n = 30) 7.0% (n = 5) 7.0% (n = 5) 27% (n = 20)
Presence of either exotic 58% (n = 43)
No exotics present on island 31% (n = 31)
742 Southeastern Naturalist Vol. 5, No. 4
Figure 2. Landscape-level hurricane damage to tree islands in A.R.M. Loxahatchee
National Wildlife Refuge 2004 in reference to hurricane tracks. Black = severe
damage, dark grey = heavy damage, moderate grey = moderate damage, light grey =
light damage, and white = no damage. Each open circle represents a sampled island.
Over 85% of the tree islands sampled after hurricanes Frances and Jeanne
had damage. Most islands sustained moderate impacts to their canopy with the
snapping of large branches and between 25–50% of the canopy removed. We
found that most of the heavier damage occurred in the center of the Refuge;
2006 C.A. Ugarte, L.A. Brandt, S. Melvin, F.J. Mazzotti, and K.G. Rice 743
however, there were patchy areas of heavy damage throughout the Refuge.
This patchiness often occurs in other forms of natural disturbances, and may
be due to a combination of microhabitat variation and wind-field microbursts,
a phenomenon reported to occur with hurricanes (Powell et al. 1996). Islands
with taller trees were five times more likely to suffer damage than tree islands
with shorter trees or shrubs. Shrubbier islands, dominated by wax myrtle,
were found more often in the northern and southern parts of the Refuge, and
may partially explain why we saw more damage to islands in the central part
of the Refuge. The types of damage most often sustained, major-trunk snaps
and uprooted or leaning trees, create openings in the canopy and can result in
tree mortality. The extent of damage varied by species. The greater proportion
of main-trunk snaps for swamp bay compared to dahoon holly was not
surprising as other researchers have noted similar species-specific damage
Table 3. Forward stepwise logistic regression of 2004 hurricane damage in A.R.M.
Loxahatchee National Wildlife Refuge as a function of tree height on sampled tree islands (r2 =
0.324, G = -47.000).
Variable B S.E. Wald df Sig. odds ratio
Step 1(a) Tree height 1.637 0.466 12.341 1 0.0001 5.138
Constant -3.787 1.310 8.363 1 0.0040 0.023
Figure 3. Percent damage to tree islands in A.R.M. Loxahatchee NWR from 2004
hurricanes based upon tree height; shrub, small-tree, and large-tree dominated islands.
Seventy-four tree islands were sampled.
744 Southeastern Naturalist Vol. 5, No. 4
(Duever 2005, Gresham et al. 1991, Putz et al. 1983, Slater et al. 1995, Walker
et al. 1992). Suggested reasons for species-specific damage include habit,
wood strength, and specific density. Putz et al. (1989) found that growth rates
and probability of snapping is inversely correlated to wood strength, and that
larger trees tended to resprout less than smaller trees. The results of this study
suggest that the larger swamp bays, which form a significant component of the
canopy on Refuge tree islands, have weaker wood strength than other treeisland
species, and may not regenerate as quickly as the smaller swamp bays.
The level of damage to a tree species does not necessarily mean that trees will
die or that the structure of communities will be altered in the long term. For
example, short-term mortality was low, 12%, for trees in hardwood hammocks
of Everglades National Park after Hurricane Andrew, despite up to
85% severe damage to upland trees (Slater et al. 1995).
We found very little damage to melaleuca relative to native species.
Although we did not detect relationships between the relative abundance of
exotics and hurricane damage, we do feel that they may cause additional
stresses to these islands. Lygodium, in particular, may severely impact these
islands because its large, heavy, rhizomatous mats may cause larger trees to
become “top heavy” and increase the probability of snapping or uprooting.
We found one case of severe damage in the Refuge where the entire tree
island canopy had collapsed. It was probable that under heavy winds, the
weight of the lygodium mat in addition to weak stems of trees toppled the
trees, leaving almost no structure whatsoever.
In a general review of hurricanes in the Everglades, Duever et al. (1994)
suggested that there are few long-term effects on community characteristics
and that canopies fill in after a few years. Walker et al. (1992) also noted that
trees rapidly regained pre-hurricane foliage and suggested that a species’
ability to re-sprout may be more important in structuring post-hurricane
communities than its ability to survive hurricane-force winds.
Long-term changes in community structure may occur in unaltered systems
from natural disturbances (Snitzer et al. 2005). However, additional
human-caused stress may exacerbate recovery patterns that influence community
structure. At the Refuge, hydrology and invasive exotic species are
two human-caused stressors that may affect recovery of tree islands. Altered
hydrology may limit regeneration of swamp bay and dahoon holly if water
levels are such that tree islands are flooded for extended periods. Particularly
in the south-central part of the Refuge, tree islands may be flooded
above their median height for 350 days, which may not allow for germination
and survival of seedlings (Brandt et al. 2006). A greater threat to
short-term tree-island recovery is the threat of colonization by lygodium and
melaleuca. Fifty-eight percent of the sampled islands had either lygodium or
melaleuca or both. Both of these exotic species have high germination rates
and high growth rates and are likely to colonize openings on islands before
native species. Snitzer et al. (2005) found greater recruitment of exotics in
gaps of high light after hurricane disturbance. We are currently monitoring
2006 C.A. Ugarte, L.A. Brandt, S. Melvin, F.J. Mazzotti, and K.G. Rice 745
recruitment of lygodium on islands that were lightly, moderately, and
heavily affected by hurricanes. Understanding how exotics colonize hurricane-
disturbed tree islands will provide information that can be used to
target priority areas for exotic control efforts.
Hurricanes present a unique opportunity to investigate recovery patterns
of tree islands in a human-altered ecosystem. In particular, hurricanes in
south Florida provide an opportunity to examine patterns of spread and
recruitment of lygodium and melaleuca. If damage sustained to tree islands
and location of islands within the Refuge are important predictors of invasiveness,
then managers can use this information to help prioritize efforts to
This study was supported by the Critical Ecosystem Studies Initiative and the
Priority Ecosystems Studies program, US Department of Interior, and the School for
Natural Resources and the Environment, University of Florida. The Fish and Wildlife
Service and O.L. Bass provided logistic support. We thank C. Bugbee for assistance
in the field, and K.R.T. Whelan for a review of the manuscript.
Brandt, L.A., D. Ecker, I. Gomez Rivera, A. Traut, and F.J. Mazzotti. 2003. Wildlife
and vegetation of bayhead islands in Arthur R. Marshall Loxahatchee National
Wildlife Refuge. Southeastern Naturalist 2:179–194.
Brandt, L.A., G.L. Martin, and F.J. Mazzotti. 2006. Topography of pop-up bayhead
tree islands in Arthur R. Marshall Loxahatchee National Wildlife Refuge Florida
Duever, M. 2005. Hurricane Hugo: Impacts on natural ecosystems and their subsequent
recovery. Presentation at the 2005 annual meeting of the Society of Wetland
Scientists, Charleston, SC.
Duever M.J., J.F. Meeder, L.C.Meeder, and J.M. McCollom. 1994. The climate of
south Florida and its role in shaping the Everglades ecosystem. Age, origin, and
landscape evolution of the everglades peatland. Pp. 225–248, In S. Davis and J.
Ogden (Eds.). Everglades, The Ecosystem, and Its Restoration. St. Lucie Press,
Delray Beach, FL.
Gresham, C.A., T.M. Williams, and D.J. Lipscomb. 1991. Hurricane Hugo wind
damage to southeastern US coastal forest tree species. Biotropica 23:420–426.
Lawrence, L.R. 1995. Timing of post-hurricane tree mortality in Puerto Rico. Journal
of Tropical Ecology 11:315–320.
Loope, L., M. Duever, A. Herndon, J. Snyder, and D. Jansen. 1994. Hurricane
impacts on uplands and freshwater swamp forest. Bioscience 44:238–246.
Noel, J.M., A. Maxwell, W.J. Platt, and L. Pace. 1995. Effects of Hurricane Andrew
on cypress (Taxodium distichum var. nutans) in South Florida. Journal of Coastal
Oberbauer, S. F., K.VonKleist, K.R.T. Whelan, and S. Koptur. 1996. Effects of
Hurricane Andrew on epiphyte communities within cypress domes of Everglades
National Park. Ecology 77:964–967.
746 Southeastern Naturalist Vol. 5, No. 4
Ogden, J.C. 1992. The impact of Hurricane Andrew on ecosystems of South Florida.
Conservation Biology 6:488–490.
Powell, M.D., S.H. Houston, and T.A. Reinhold, 1996: Hurricane Andrew’s landfall
in South Florida. Part I: Standardizing measurements for documentation of
surface wind fields. Weather Forecasting 11:304–328.
Putz, F.E., P.D. Coley, K. Lu, A. Montalvo, and A. Aiello. 1983. Uprooting and
snapping of trees: Structural determinants and ecological consequences. Canadian
Journal of Forest Research 13:1011–1020.
Putz, F.E., V. Nicholas, and L. Brokaw. 1989. Sprouting of broken trees on Barro
Colorado Island, Panama. Ecology 70:508–512.
Roman, C.T., N.G. Aumen, J.C. Trexler, R.J. Fennema, W.F. Loftus, and M.A.
Soukup. 1994. Hurricane Andrew’s impact on freshwater resources. Bioscience
Slater, H.H., W.J. Platt, D.B. Baker, and H.A. Johnson. 1995. Effects of Hurricane
Andrew on damage and mortality of trees in subtropical hardwood hammocks of
Long Pine Key, Everglades National Park, Florida, USA. Journal of Coastal
Snitzer, J.L., D.H. Boucher, and K.L. Kyde. 2005. Response of exotic invasive plant
species to forest damage caused by Hurricane Isabel. Pp. 209–214, In K.G.
Sellner (Ed.). Hurricane Isabel in Perspective. Chesapeake Research Consortium,
CRC Publication 05-160. Edgewater, MD.
Walker, L.R., J. Voltzow, J.D. Ackerman, D.S. Fernandez, and N. Fetcher. 1992.
Immediate impact of Hurricane Hugo on a Puerto Rican rain forest. Ecology
Woodmansee, S.W., K. Bradley, and S. Hodges. 2005. Systematic reconnaissance
flights and exotic plant species mapping at selected national wildlife refuges in
Florida. Final Report. Submitted to US Fish and Wildlife Service, Sanibel, FL.