History of Sustainability Initiatives at University of Canterbury
Deflecting the wave FINAL REPORT June 2015
1. 1
Final Report
Deflecting the Wave: Using Coastal Vegetation to
Mitigate Tsunami and Storm Surge
By
Dr. Andrew Kaufman, Dr. Timothy Gallaher, and
Dr. Alberto H. Ricordi
Tropical Landscape and Human Interaction Lab
Department of Tropical Plant and Soil Sciences
College of Tropical Agriculture and Human Resources
University of Hawai‘i at Mānoa
June 2015
Funded by a grant from:
Kaulunani Urban and Community Forestry Program, USDA Forest Service,
and the DLNR Division of Forestry and Wildlife
2. 2
Deflecting the Wave: Using Coastal Vegetation to Mitigate
Tsunami and Storm Surge
Andy Kaufman1
Timothy Gallaher2
Alberto Henrique Ricordi3
1
Tropical Landscape and Human Interaction Lab,
Department of Tropical Plant and Soil Sciences
University of Hawaii at Manoa, 3190 Maile Way, Honolulu, HI 96822
kaufmana@hawaii.edu
2
Botany Department University of Hawaii at Manoa
gallahert@ctahr.hawaii.edu
3
School of Architecture,
University of Hawaii at Manoa
albertoh@hawaii.edu
KEYWORDS
Tsunami, Bioshield, Pacific Islands, Hawaii, Coastal vegetation. Coastal restoration
CONTACT
For inquiries please send an email to: kaufmana@hawaii.edu
CITATION
Kaufman, A., T. Gallaher, and Alberto H. Ricordi. 2015. Deflecting the Wave: Using Coastal Vegetation
to Mitigate Tsunami and Storm Surge. University of Hawaii at Manoa, Department of Tropical Plant and
Soil Sciences.
June 2015
3. 3
Table of Contents
Executive Summary.................................................................................................................... 5
Introduction –Tsunamis, Storm surge and Coastal Vegetation ...................................................... 8
Tsunami terminology and interaction with coastal vegetation ................................................... 9
Case study: Coastal re-vegetation at Tonga.............................................................................. 21
Summary of Coastal Vegetation, Tsunami, and Storm Surge Studies. .................................... 23
Observations from Upolu Samoa.................................................................................................. 26
Methods..................................................................................................................................... 26
Results....................................................................................................................................... 27
Site 1: Coconut Beach Resort ............................................................................................... 27
Site 2: Saleilua village .......................................................................................................... 29
Site 3: Tafatafa...................................................................................................................... 37
Site 4: Utulaelae-Sapoe villages ........................................................................................... 39
Site 5: Saleapaga village....................................................................................................... 42
Combined Vegetation Analysis ............................................................................................ 43
The Coastal Flora of Hawaii......................................................................................................... 47
Introduction............................................................................................................................... 47
Methods..................................................................................................................................... 49
Results....................................................................................................................................... 54
Discussion................................................................................................................................. 61
Conclusions................................................................................................................................... 83
Coastal Vegetation Restoration in Waimanalo, Hawaii ............................................................... 89
Beach erosion in Hawaii ........................................................................................................... 90
History of Bellows Beach......................................................................................................... 91
Beach Processes........................................................................................................................ 92
Beach Erosion ........................................................................................................................... 93
Beach hardening........................................................................................................................ 96
Invasive Plant Species............................................................................................................... 97
Ironwood Management ......................................................................................................... 97
Man Made Alterations of Bellows Beach............................................................................... 100
Sea-Level Rise ........................................................................................................................ 100
Effects of Sea-Level Rise in Hawaiʻi...................................................................................... 100
Experimental Site at Belows Air Force Base.............................................................................. 101
Objective................................................................................................................................. 101
Site location ............................................................................................................................ 102
5. 5
Executive Summary
Tsunami and storm surge due to tropical cyclones regularly affect the coasts of Hawaii
and other Pacific Islands, in many cases causing severe property damage, injuries, and deaths.
Following the Indian Ocean tsunami of December 26, 2004, the anecdotal accounts of survivors
and the observations of researchers pointed to a potential role of coastal vegetation in mitigating
damage and in some cases reducing the death toll due to a tsunami. This deliberate use of
vegetation as a buffer against ocean waves has been termed bio-shields, shelterbelts, and
green/living sea walls.
The goals of the research described in this report was to focus on Hawaii and other
Pacific islands to: (1) conduct research on the type of vegetation that has survived past tsunami
and storm surge events, (2) gather information on vegetation that grows near the shore in Hawaii
given different environmental factors; (3) examine whether past or existing vegetation has had an
effect on mitigating beach erosion due to wave impact; and, (4) establishment of experimental
coastal reforestation plots for evaluation of coastal re-vegetation planting strategies.
The investigation began with a search for historical documents that might shed light on
the interactions between tsunami waves or storm surge and coastal vegetation in Hawaii and the
Pacific however; little research on this topic has been published along with little more than
anecdotal accounts.
On September 29, 2009 a tsunami inundated the southern coast of Upolu Samoa killing
over 140 people and causing extensive property damage. In January 2010, a team for the
Tropical Landscape and Human Interaction Lab at the University of Hawaii sent a team to make
observations in Upolu to search for interactions between the tsunami and coastal vegetation. Also
conducted, was vegetation surveys on the Islands of Oahu, Hawaii and Kauai to characterize
existing coastal vegetation patterns.
The observations in Samoa lend support to the hypotheses that coastal vegetation
mitigates the effects of a tsunami through several mechanisms: Coastal vegetation forms a
physical barrier to an incoming wave which may result in reduced damage to structures and
reduced erosion. Additionally, coastal vegetation builds elevation at the coast by trapping
organic matter and sand, and coastal vegetation provides a vertical escape for people trapped in
the wave. Finally, coastal vegetation acts as a filter which holds back coral, ships and debris,
6. 6
carried by the wave from being moved inland where it can be destructive to people and property
and from being carried out to sea and onto sensitive reefs.
Conversely, the coastal forests in Hawaii are reduced in species diversity, complexity and
stem density relative to their Samoan counterparts. This will seriously impact the ability of these
forests to provide an effective barrier for tsunami or storm surge waves. In addition, many
coastal areas in Hawaii have been completely deforested in favor of park-like landscapes and
direct development at the coast. Hawaii‟s coastal forests are dominated by a few widespread
invasive species including Prosopis pallida (Mesquite), Rhizophora mangle (American Red
Mangrove) Terminalia catappa (Tropical Almond), Casuarina equisetifolia (Ironwood).
Prosopis pallida was introduced to Hawaii in 1827 and has naturalized largely due to the action
of cattle and feral animals. The mangrove, R. mangle is widely established on Oahu in coastal
areas that are well protected from high energy waves. Terminalia catappa and C. equisetifolia
were planted in the early part of the 20th century to reforest coastal areas. All three dominant
coastal species form largely monotypic stands and disperse through floating propagules. Native
species are not completely absent from Hawaii‟s coastal areas. Several of the transects
encountered native coastal forest including forests of Pandanus tectorius (hala), often mixed
with Metrosideros polymorpha („ohia lehua), and forests where Thespesia populnea (milo) is
dominant. Little work has been done on identifying best practices of native coastal reforestation
in Hawaii or the Pacific. The combined observations from Samoa and Hawaii form the basis for
specific recommendations as to how such bio-shields could be most effectively designed and
implemented in Hawaii and other Pacific Islands however additional research is urgently needed.
The third phase was to develop a method for restoration of native coastal vegetation using
primarily native Hawaiian species and evaluate the method effectiveness, and its effects on wave
power and erosion. The effects of vegetation on wave power has been observed by post-event
surveys after the tsunami in Samoa and through visual documentation of storm water runoff at
Bellows Air Force Station (BAFS) in Waimanalo, Hawaii. Beach erosion as much as two feet
per year has been documented at BAFS, which is mostly attributed to hardened shorelines, but it
is also associated with invasive species such as Casuarina equisitifolia which inhibits growth of
native shrubs and ground covers. This research project tested a planting method for
establishment of native plants after removal of C. equisitifolia, and verified the effectiveness of
temporary windscreens for protection against wind and salt spray. Temporary windscreens
proved beneficial to speed-up the establishment of the plants, especially in the foredune zone
7. 7
(ocean side). However, the windscreens were knocked down by a storm event three months after
planting and there was no visual difference between the plots with or without windscreens one
year after planting. Therefore, the use of windscreens may not be necessary and cost effective
since it only has short term benefits and results in extra cost and potential debris in the beach if
the wind screens and its supports are not completely removed, which also adds cost. A modular
irrigation system was designed for easy removal and reassembly, so it can be re-used in
additional restoration areas. The irrigation was gradually reduced and totally removed eight
months after planting. Data revealed irrigation lines on the windward side of the plots were
buried up to 6 ¼” (six and a quarter inches), and sand accretion was visually evident in the
perimeter of the plots. Additionally, very clear plant zones corresponding to the beach berm,
foredune, dune crest, and backdune zones were present. Sporobolus virginicus (ʻakiʻaki grass)
and the beach morning glory vine Ipomea pes-caprae subsp. brasiliensis (pohuehue) were very
successful to cover the ground throughout all zones, with I. pes-caprae growing up to fourteen
feet beyond the irrigated areas. This report includes the detailed irrigation system used in this
project, visual photographs with a timeline of the planting establishment, ground coverage and
dry matter data collected one year after planting, and recommendations of native plants and their
planting zones for coastal planting and landscaping in Hawaii.
8. 8
Introduction –Tsunamis, Storm surge and Coastal
Vegetation
Tsunami and storm surge due to tropical cyclones regularly affect the coasts of Hawaii
and other Pacific Islands, in many cases causing severe property damage, injuries, and deaths.
Following the Indian Ocean tsunami of December 26, 2004, the anecdotal accounts of survivors
and the observations of researchers pointed to a potential role of coastal vegetation in mitigating
damage and in some cases reducing the death toll due to a tsunami. This deliberate use of
vegetation as a buffer against ocean waves has been termed coastal bio-shields, shelterbelts, and
green/living sea walls.
Tsunami and storm surge due to tropical cyclones regularly affect the coasts of Hawaii
and other Pacific Islands, in many cases causing severe property damage, injuries, and deaths.
Hawaii has experienced destructive tsunami quite regularly with recent events in 1952, 1960,
1975, and 2011. Recent destructive tropical cyclones in Hawaii include Hurricanes Iniki (1992),
Iwa (1982), Dot (1959), and Nina (1957) (Committee on Natural Disasters 1983, Chiu et al.
1995).
The goals of the research described in this report was to focus on Hawaii and other
Pacific Islands to: (1) conduct research on the type of vegetation that has survived past tsunami
and storm surge events, (2) gather information on vegetation that grows near the shore in Hawaii
given different environmental factors and (3) examine whether past or existing vegetation has
had an effect on mitigating beach erosion due to wave impact. The first phase of this
investigation began with a search for historical documents that might shed light on the
interactions between tsunami waves or storm surge and coastal vegetation in Hawaii and the
Pacific however; little research on this topic has been published along with little more than
anecdotal accounts.
On February 27, 2010 a tsunami in the Pacific Ocean was triggered by an 8.8 magnitude
earthquake off the coast of Chile. This prompted officials of the Pacific Tsunami Warning Center
to issue a Pacific-wide tsunami warning. In Hawaii, an orderly evacuation of coastal areas
followed and much of the state waited and watched as the tsunami raced across the Pacific.
Although the tsunami had a devastating impact on coastal areas in Chile, little effect was felt
elsewhere in the Pacific. Had a destructive wave reached Hawaii’s shores, the evacuation
prompted by the early warning system would clearly have been responsible for saving lives.
9. 9
Tsunamis are regular natural disasters in the Pacific. Pacific Islands have experienced
destructive tsunami throughout history with recent events in 1952, 1957, 1960, 1975, and 2009
(NOAA 2011).The most significant natural disaster in terms of loss of human life in Hawaii
continues to be the April, 1946 tsunami which cost 159 lives and $150 million (adjusted to 1982
dollars) in Hawaii and took additional lives across the Pacific. On September 29, 2009, a tsunami
generated near Tonga inundated areas of Tonga, American Samoa and Independent Samoa. Due
to the close proximity of the source event to these islands, there was little time for warning or
evacuation and the tsunami claimed over 150 lives and caused extensive property damage.
While an early warning system along with an educated public is the centerpiece of an
effective tsunami/storm surge mitigation strategy, early warning systems can fail and if a tsunami
is generated locally, there may not be sufficient time for an effective alert or evacuation. In such
cases, and particularly when early warning systems and education of the public are already well
established, additional defensive measures can be taken that may protect people and property.
Sea walls are one type of defensive measure that has been successfully employed. Sea walls
however are expensive to construct and maintain, are not suitable for all coastal types, and they
have in some cases been associated with accelerated sand loss from beaches (Pilkey and Wright
III 1988, Kraus and McDougal 1996, Defeo et al. 2009, DLNR 2010, Fletcher et al. 2012). For
these reasons, seawalls are often restricted to highly populated areas with suitable coastal
geomorphology. Another type of defensive strategy employs coastal vegetation as a barrier to
ocean waves. The use of vegetation as a buffer against ocean waves has been termed coastal bio-
shields, shelterbelts, and green/living sea walls. Throughout this report the term “bio-shields”
will be used due to the greater level of usage of this term in the scientific and popular literature.
Tsunami terminology and interaction with coastal vegetation
On December 26, 2004 a 9.3 magnitude earthquake occurred 100 km west of Sumatra.
This earthquake generated a tsunami which was detectable in all ocean basins around the world
(Titov et al. 2005).The immediate death toll from the tsunami was over 350,000. Over 5 million
people were displaced and the cost of recovery efforts was estimated to be over eight billion US
dollars (Athukorala and Resosudarmo 2005). In the aftermath of the tsunami, several anecdotal
reports emerged that coastal vegetation, particularly mangroves, had mitigated loss of life and
property (Padma 2004, Kremmer 2005).Although there had been some research on the
interactions between coastal vegetation and tsunami waves prior to the 2004 Indian Ocean
10. 10
tsunami (see Hiraishi and Harada 2003; and for a summary of other early studies see Tanaka et
al. 2009), the 2004 event produced renewed interest and provided a number of potential study
sites to investigate these interactions.
In order to understand the existing research it is important to first be familiar with
standard tsunami terms. The inundation distance is the distance from the shoreline to the inland
limit of tsunami inundation. Run-up elevation is the elevation above sea level of a tsunami at
the limit of penetration. Tsunami height is the height above ground level of a tsunami wave in a
given point up to the limit of tsunami inundation (the tsunami height is 0 at the limit of tsunami
inundation) and the Run-up elevation is the estimated elevation above sea level at the limit of
tsunami inundation (USGS 2005). In this report, also defined is an inundation point as a
geographic reference point (recorded with a GPS and/or displayed on a map) which is the best
estimate of the limit of tsunami inundation based on, field reconnaissance or remotely sensed
data and an inundation line as the straight-line interpolated connection of multiple assessed
inundation points.
Figure 1. Terminology related to tsunami measurements. (USGS 2005)
(http://walrus.wr.usgs.gov/tsunami/srilanka05/images/run_up_height_inundation.jpg)
Many studies have attempted to correlate various measures of vegetation fronting
developed areas with measures of the inundation such as run-up elevation, inundation distance,
some measure of damage, or levels of human casualties. Each of these measures is open to
reasonable criticism. For example, damage to structures should take into consideration variation
in construction type (Dahdouh-Guebas and Koedam 2006) and orientation to the oncoming wave
(Dall’osso and Dominey-Howes 2009). Death toll measures must take into account pre-event
11. 11
human population density, and site specific human activity patterns. Vegetation densities can be
estimated or measured in a wide variety of ways that can influence the conclusions (Bhalla 2007,
Kaplan et al. 2009). Finally, all studies should specifically take into account those factors that are
known to be associated with inundation effects such as near shore bathymetry, elevation, and
distance from shore. Studies have taken on a number of different methodologies including
analysis based on, remotely sensed (satellite) data with GIS modeling, wave tank simulations,
mathematical modeling and on the ground field studies of vegetation and direct observations of
wave-induced damage.
In one of the earliest studies, Danielson et al. (2005) used pre-tsunami (May 4, 2003)
satellite imagery of Cuddalore District, Tamil Nadu, India to classify coastal vegetation into
three classes: dense tree vegetation, open tree vegetation, and no trees vegetation. They used
post-tsunami (December 31, 2004) satellite imagery to assess damage into four categories,
damaged, partially damaged, undamaged, and inundated but not damaged. Preliminary statistical
tests indicated that, "dense tree vegetation was associated with undamaged areas and
disassociated with damaged areas". Dahdouh-Guebas and Koedam (2006) pointed out that the
preliminary analysis by Danielsen et al. does not take into consideration variation in construction
type in determining damage classes for buildings, also they point out that results are presented
without considering distance from the shore to villages. They suggest that future studies should
compare villages that are at similar distance from shore and differ only in the level of protection
conferred by coastal vegetation. Danielsen et al. (2006) responded that house construction was
homogenous in the area.
In a follow up study of the same area investigated by Danielson et al. (2005), Olwig et al.
(2007) inspected pre and post tsunami satellite imagery and made a map of areas with dense
woody vegetation and open woody vegetation. In defining their site selection methodology, the
authors identified a number of factors which should ideally be held constant to address in
isolation the effect of coastal forest on tsunami inundation (run-up or damage). These include the
presence of both vegetated and non-vegetated coastline, homogenous bathymetry and
homogenous topography. A fourth criterion, substantial damage reports in the area, has been
criticized since it may serve to bias the results against areas where coastal vegetation provided
significant protective effects. The authors measured widths of (1) dense and (2) open vegetation
along a GIS transect and areas with (3) no woody vegetation with widths of damage behind the
12. 12
vegetation. Damage was classified into four categories: severely damaged, (most of the physical
structures destroyed; partially damaged (some damage but most physical structures intact),
undamaged, and inundated (areas undamaged but inundated) following the criteria of Danielsen
et al. (2005). One hundred transects separated by 200m were laid out over the area following the
same direction as the incoming tsunami wave. The authors have not yet published the statistical
analysis of the data although based on visual inspection of the map, however based on visual
inspection of the map, they concluded that dense vegetation had much less damage behind it than
open vegetation in most cases with outliers of this trend possibly occurring as the result of edge
effects where neighboring gaps allowed the tsunami wave to run farther inland.
Another study from the same area specifically looked at human mortality, and a
socioeconomic indicator “per-capita loss of wealth” to determine if vegetation may have had a
protective effect. Kathiresan and Rajendran (2005) correlated human death toll and per-capita
loss of wealth with distance from shore, elevation from mean sea level, and type of coastal
vegetation in 18 coastal hamlets along the Parangipettai coast of Tamil Nadu State, India. They
reported a negative correlation between death toll and distance of human inhabitation from sea (r
= -0.61, P < 0.01), the elevation from mean sea level (r = -0.63, P < 0.01;), and the area of
mangrove and other coastal vegetation (r = -0.58, P < 0.01). They also observed that many
deaths were caused by the thorns of a single species, Prosopis spicifera (syn.= Prosopis
cineraria (L.) Druce), indicating that some species may actually increase the risk from tsunami
events. Although Prosopis was implicated as a major cause of mortality, presence or absence or
abundance of this species was not analyzed as an independent variable to predict mortality.
Kerr et al. (2006) reanalyzed the data presented by Kathiresan and Rajendran (2005)
using stepwise regression with tsunami mortality as the dependent variable and hamlet elevation,
distance from sea and the area of coastal frontage given to vegetation as independent variables.
Their reanalysis found that with the more appropriate statistical tests (stepwise regression vs.
simple linear regression) differences in coastal vegetation area did not explain variation in
human mortality which was mainly explained (87%) in their analysis by hamlet elevation, and
distance from the sea. This prompted Kerr et al. to conclude that, "given hamlets of equal
elevation and distance from the sea, differences in vegetation area did not mitigate human
mortality caused by the tsunami." Vermaat and Thampanya also reanalyzed the data and after
initially reporting a protective effect of vegetation (Vermaat and Thampanya 2006) retracted
13. 13
their findings when errors in their statistical analysis were revealed (Vermaat and Thampanyab
2007).
Another study, also from Tamil Nadu, India used a remote sensing technique to estimate
the amount of vegetation between known inundation points and the coast along inundated areas
(Bhalla 2007). Their methodology calculated the Normalized Difference Vegetation Index from
satellite imagery (NDVI) [NDVI = (Near infrared – red)/(near infrared + red)] along a straight
line transect between the known tsunami inundation points and the coast . This calculation,
according to the authors, is an estimate of "the amount of chlorophyll present in a given pixel on
a scale from 0 to 1, although this method does not indicate the type of vegetation present. Their
analysis found no statistical significance (P=0.45) between tsunami inundation distance and
NDVI.
Satellite imagery has provided the most often used data for investigations of coastal
vegetation-tsunami interactions. In addition to the work in Tamil Nadu, India, researches have
investigated effects in Phuket, Thailand and Banda Aceh, Sumatra with similar results. Chang et
al. (2006) employed pre and post-tsunami 7-band satellite imagery, spanning the Thailand coast
northward from Phuket Island. Tsunami damage in urban and forested areas was assessed and
classified using post tsunami satellite imagery to detect scouring and debris. Damage maps were
produced in urban areas based on the percent of collapsed buildings and field observations of
damage were made on site eight months following the tsunami. A damage scale for buildings
was assessed with five damage classes and three building types. Pre-tsunami land-use was
classified using ENVI software. The following data sets were analyzed: Land use, change in land
use, reports of damage to property and high loss of life, bathymetry, topography. Four sites were
selected and within each site, “location pairs” defined as communities with similar bathymetric
details and coastline exposure but which had “potentially different protection levels by
mangroves”, were identified. Initial results indicate that lower levels of damage were observed in
three villages situated behind mangroves, with an intermediate level of damage in one village
that was "partially exposed" with the highest level of damage observed in four villages that were
“completely exposed”.
Iverson and Prasad (2006) used satellite imagery of coastal Banda Aceh, Sumatra which
they classified into forested and developed and compared with images of the same area which
had been classified previously by "the US Government" into “damage” and “no damage” areas.
14. 14
They calculated damage: undamaged ratios and in both areas found 2.0 and 2.5 more damage in
developed as compared to forested lands. A model of damage was generated in Random Forests
(RF) modeling software for use in R statistical software, using the following predictors:
classified vegetation types, coastal exposure level, distance to shore, and elevation, was
generated and compared with actual damage. In their model, elevation and distance to shore were
the most important variables to predict tsunami damage followed closely by vegetation and then
by exposure level. This model was able to correctly classify 93.9 percent of the study area. They
also used the same data to produce predictive tsunami risk models for the larger area. The
authors conclude that, "developed land was much more susceptible to tsunami damage than
forested land" and that these results “provide further evidence of the protective power of coastal
forests." Baird and Kerr (2008) have criticized Iverson and Prasad (2006) pointing out that their
experimental design did not specifically test the protective role of coastal forest and that claims
that this constitutes evidence for the protective role of coastal forests is unwarranted.
In addition to the work on coastal vegetation – tsunami interactions, various methods
including satellite imagery has been used to assess interactions between coastal vegetation and
the effects of storm surge. Das and Vincent (2009) analyzed death toll in villages in Kendrapada
District, from a super cyclone that struck the state of Orissa India in 1999. They used pre-storm
satellite imagery to assess the extent of mangroves and restricted the study to 409 villages that
have historically had mangroves so that the absence of mangroves today is likely attributable to
human removal of mangroves rather than some other factor which may have excluded them. This
study found a significant negative correlation between mangrove width and deaths. The average
mangrove width was 1.2km.
Although satellite imagery can be a powerful tool for assessing coastal vegetation-
tsunami interactions, these studies are limited in that they are able only to show presence, and
general shape of coastal vegetation. Studies based on satellite imagery alone are not able to
assess qualities of coastal forest such as density, structure of the understory and branching and
rooting patterns that might vary greatly between forest types and which will directly interact with
an oncoming tsunami wave. The studies by Danielson et al. (2005) and Olwig et al. (2007) find a
protective effect of dense vegetation however density of the vegetation is not directly measured
and represents a categorical determination based on visual inspection of imagery. The study by
15. 15
Bhalla (2007) which uses NDVI likewise does not directly address the physical structure of the
vegetation sensed by satellites.
Another suite of tools used by researchers to assess coastal vegetation – tsunami/storm
surge interactions includes wave tank simulations and mathematical modeling. In a wave tank
simulation, Irtem et al. (2009) used a glass-walled wave channel 22.5m in length, 1.00m in
width, and 0.50m in depth along with sand and artificial pine trees (4.6 cm in diameter and 9 cm
in height) and wooden dowels (to simulate trees without leaves) to model a coastal forest in three
and two configuration respectively. Wave run-up height behind the simulated vegetation was
measured. They found that a dense configuration with leaves had the greatest reduction effect on
run-up height. Thuy et al. (2009) also modeled vegetation using wooden cylinders with a
diameter of 5mm mounted in a staggered arrangement and assessed the effect of gaps through
the simulated vegetation. They found that as the gap width increases, the flow velocity at the gap
exit increases at first, reaches the maximum value, and then decreases. For a forest with a width
of 200m perpendicular to shore, the flow velocity at the end of a 15m wide gap located in the
middle of the forest will reach a maximum value of 2.5 times the velocity without a gap and 1.7
times the velocity of an un-vegetated coast.
Mathematical modeling of the protective effect of coastal forests during a tsunami have
been carried out by Harada and associates (Hiraishi and Harada 2003, Harada and Imamura
2005, Harada and Yoshiaki 2005) and also by (Nandasena et al. 2008). Harada and
Imamura(2005) used forest parameters in a numerical modeling experiment. Their model was
limited in that it could not model for the breaking of trees. Their model accounted for the effect
of forest density(10, 30, 50 trees / m2
), trunk diameter (0.3, 0.15, 0.1m), forest width (50, 100,
200, 400), tree height (10m), branch height (2m), and the "projected area rate" of leaves (0.65).
The model included the effect of coastal forest in the as the resistance force in the momentum
equation. Resistance coefficients of coastal forests were taken from modeled hydraulic
experiments (Harada and Imamura 2000).Tsunami heights of 1, 2, and 3m were tested with wave
period of 10, 20, 30, 40, 50 and 60 minutes. The effect of vegetation was measured as:
(maximum values with forest /max values without forest. = r).In their numerical simulation, the
coastal forest reflected wave energy reducing run-up elevation behind the forest. When the
tsunami reached the level of leaves and branches, a larger effect was observed. An increase in
forest width from 50 to 400 m significantly reduced, maximum inundation depth, hydraulic
16. 16
force, and maximum current velocity. An increase in forest density from 10-50 trees / 100m2
resulted in only small decreases in these variables.
Following up on work by Harada and Imamura (2003) (summarized in Harada and
Imamura 2005), Harada and Yoshiaki (2005) calculated the tsunami resistance as a function of
stand age using forest density, DBH, and branch height parameters for pine forests. They found
that higher densities and lower branch heights contributed to a larger roughness coefficient and
thus a greater effect of the coastal vegetation of the reduction of tsunami run-up elevation.
The simplified systems replicated in mathematical modeling and wave tank simulations
in general tend to predict an attenuating effect of vegetation on tsunami and storm surge. Results
from field studies are typically inconclusive and sometimes contradictory. This is likely due to
the highly complex nature of modeling interactions in the natural and anthropogenically
influenced environment.
As knowledge about coastal vegetation-tsunami/storm surge interactions has grown,
models to explain these interactions and their subsequent effects on people and property have
become more complex. Chatenoux and Peduzzi (2007) used a large data set covering 62 sites
located in Indonesia, Thailand, continental India, Sri Lanka, and the Maldives. A set of
parameters were investigated that might best explain the inundation distance (measured as the
width of flooded land strip = D). For each site, maximal D was used as the dependent variable in
the analysis. Each site represented a single data point. Maximal D was estimated from satellite
images and from data available from other studies. Independent variables assessed included:
bathymetry, location of epicenter coordinates, fault lines, elevation level, information on
coastlines, land cover (in seven classes), distribution of coral, seagrass beds, and mangrove
forests. Combinations of the following parameters were most predictive in the resulting
regression model: the distance from the tectonic origin (distance from subduction fault line), the
near-shore geomorphology, and also environmental features (percentage of coral and percentage
of seagrass beds) (R2
= 0.655). Their results indicate that (1) a steep slope blocks tsunami energy
while a flatter slope builds a higher wave leading to a larger inundation distance (2) in inundated
areas fronted by areas inhabited by seagrass, the distance of impact was less than other areas
without seagrasses, (3) there was a positive correlation between the presence of corals and
inundation distance. The authors found that most sites assessed did not have mangroves directly
fronting exposed coast since mangroves are often present only in protected estuaries. This study
17. 17
therefore was not able to assess the role of mangroves however the authors concluded that, "In
such case it is suspected that areas covered by mangroves forests were less impacted by tsunami
just because mangroves forests communities tend to be located within sheltered coastal areas."
The authors could not rule out if the interactions observed with seagrassess and corals were not
due to unmeasured environmental variables stating, “A mechanism to explain the observations
that the presence of coral reefs positively affected D remains unexplained” and “it is impossible
to differentiate if the presence of seagrass beds has a mechanical influence that absorbs the
energy of the waves or if the area that seagrass usually colonize is already protected from the
wave.”
Kaplan (2009) found significant differences between three vegetation classes, which
differed in overstory and understory, with regard to their effects on inundation depth (as
determined by interview with home owners) and damage to surveyed houses in Sri Lanka. Their
results indicate an effect of vegetation type on water height and damage levels. They report that
the water level was significantly higher at houses behind the vegetation class consisting of dense
undergrowth and coconut and Pandanus overstory) than vegetation classes consisting of (1) a
belt of Pandanus backed by a loose coconut plantation with more or less no undergrowth and (2)
vegetation consisting of only very few trees, but with a dense undergrowth of different shrubs.
The researchers did not formally quantify the vegetation structure and completely unvegetated
areas were not included in the analysis. From their analysis, it cannot be sure if the observed
effect (if related at all to vegetation) was due to total vegetation, tree density, or the density of
undergrowth.
Tanaka et al (2007) investigated several vegetation types in Sri Lanka following the
Indian Ocean tsunami using both field surveys and subsequent modeling. They concluded that
the ability of coastal forest to attenuate wave energy was related to both horizontal and vertical
forest structure. They predict that greater stem density and greater above ground complexity in
terms of branches, leaves, and prop roots, would produce greater drag forces on tsunami waves.
They suggested that a forest with both small and large diameter trees may be particularly
effective as the dense smaller trees and greater amount of above ground structures within the
wave inundation height would reduce wave velocity while large diameter trees would be able to
stop debris and would be less likely to break during the tsunami event. Using Casuarina
equisetifolia as a single species example, they suggest that when the diameter was larger than 0.1
18. 18
m, trunks were not broken by the tsunami and had sufficient stem density to be effective at wave
attenuation, however with an average diameter greater than 0.5 m, stem density was low due to
self thinning of the stand and they presumed that this density had little effect in reducing wave
velocity. Likewise, they found that C. nucifera likely had little effect on the wave because it was
growing in stands with very wide spacing and had a simple above ground structure within the
inundation height. In contrast, their observations suggest that a two layer arrangement of
vegetation in the vertical direction with P. odoratissimus in the understory and C. equisetifolia in
the overstory seems to have provided the greatest level of protection from tsunami waves.
Feagin (2008) questioned whether coastal mangrove forests directly reduced the effect of
large waves or if coastal forests indirectly affect waves by changing or engineer coastal
topography through the formation of dunes. If this is the case, an engineered coastal forest
should take into consideration those attributes of natural forests which allow it to build elevation.
It is well established that vegetation, through a combined effect of above ground and
below ground dynamics, effects soil erosion. This occurs through physical intercepting
raindrops, increasing infiltration through the soil, allowing for transpiration of soil water,
increasing surface roughness, and by adding organic matter to soil. Through these mechanisms,
there is a well established exponential decrease of soil erosion rates with increasing vegetation
cover (Gyssels et al. 2005). The dynamic interactions between waves and coastal erosion is less
well understood and few studies have specifically addressed the role of coastal forests in
influencing patterns of erosion during a storm surge or tsunami event. Coastal vegetation
provides erosion protection through the same mechanisms as other vegetation types. In addition,
coastal vegetation: (1) increases the durability of the sediment root matrix; (2) forms dunes
through the interception of sand, organic material and other particles while reducing wind
erosion; (3) reduces wave heights leading to reduced offshore transport; and, (4) reduces wave
velocity resulting in deposition from waves (Dean 1978, Lancaster and Baas 1998). While there
has been some efforts to quantify these mechanisms, particularly for wetland species (Knutson et
al. 1982, Fonseca and Cahalan 1992), and seagrasses (Fonseca 1996), there is still very poor
understanding of how below and above ground parts of terrestrial coastal vegetation interacts
with coastal erosional processes (Dean 1978).
The effects of erosion may have immediate impact on recovery efforts by undercutting
roads and destroying utilities and may have longer impacts on coastal geomorphology.
19. 19
Accelerated coastal erosion has been linked to development of coastal areas (with accompanying
deforestation) (Dean 1978). For example, Mimura and Nunn (1998) attributed increased coastal
erosion and beach loss in Fiji to increased clearing of coastal vegetation since the 1960’s.Erosion
resulting from the removal of vegetation in coastal areas may result in a longer and more gradual
slope between the ocean and inland areas. This change in coastal geomorphology would present
a reduced barrier to incoming tsunami or storm surge waves. In response, planting vegetation or
encouraging natural vegetation at the coast has been employed for many years as a strategy to
protect against coastal erosion (French 2002).
Some of the work related to tsunami or storm surge bioshields has focused on what
species or vegetation types might best withstand the force of incoming waves as well as survive
the inundation. Jayatissa and Hettiarachi (2006) assessed coastal vegetation in 15 sites to cover
all the major climatic zones in Sri Lanka, 14, 44, and 134 days following the tsunami. Species
were assessed for damage following the tsunami and classified into three groups: (1) Species
unaffected, (2) species affected and recovered over time and (3) species affected and not
recovered. Many of these species are common coastal species in the Pacific and Hawaii. A list of
47 species was compiled, 26 of these are also found in Hawaii (Table 1).
Table 1. Survivorship of species in inundated areas of Sri Lanka following the 2004 Indian Ocean tsunami. Only
species also found in Hawaii are listed (Jayatissa and Hettiarachi 2006).
Species unaffected (12 species)
Barringtonia asiatica (P), Calophyllum inophyllum (N), Clerodendrum inerme (C), Hibiscus
tiliaceus (I or N), Ipomoea pes-caprae (I), Opuntia sp. (N), Pandanus tectorius (I), Prosopis
juliflora (N), Terminalia catappa (N), Thespesia populnea (I), Casuarina equisetifolia (N),
Cocos nucifera (N,C), Opuntia sp.(N)
Species affected and recovered over time (12 species)
Artocarpus altilis (C), Artocarpus heterophyllus (C), Citrus spp. (C), Ficus benghalensis (N),
Hernandia ovigera, (C), Morinda citrifolia (N, C), Parkinsonia aculeata (N), Tamarindus
indica (P), Mangifera indica (N, C), Tamarindus indica (C), Moringa oleifera (C),
Anacardium occidentale (C)
Species affected and not recovered (2 species)
Psidium guajava (N), Macaranga sp. (N)
Status in Hawaii (per Wagner et al. 1999) is given in parenthesis.
(I = indigenous), (N = Naturalized), (P = present but not naturalized), (C = Cultivated).
20. 20
Similar resilience of coastal assemblages has been reported by a number of post storm
studies, however the ability of species and entire vegetation assemblages to survive these events
are highly dependent upon the intensity and duration of the storm surge and. Post storm
vegetation assessments were carried out on Jaluit atoll, Marshall Islands, after Typhoon Ophelia
passed over the atoll in 1958 (Blumenstock 1961), in Tonga following Cyclone Isaac in 1982
(Woodroffe 1983), and following back-to-back cyclones Alix and Carol which struck Mauritius
in January and February 1960 (Sauer 1962). These and other assessments indicate significant
levels of damage due to wind, storm surge and salt spray. For example in Tonga, in the worst hit
areas, coastal vegetation was destroyed up to 30 meters from the coast up to 6m above high tide
due to storm surge (Woodroffe 1983). In this case the coasts were inundated for several hours
with high waves. In Jaluit, researchers found that the greatest damage to vegetation occurred
where there was a combination of strong winds and ocean inundation (Blumenstock 1961). The
storm did not have as great of an effect on understory species except in inundated areas where
they were almost completely wiped out (Blumenstock 1961). In both cases, strong winds were
responsible for very high levels of mortality to trees, particularly Pandanus and coconut, that
were emergent from the canopy. Sauer (1962) reported that following the second cyclone, in
Mauritius, most of the common coastal species were recovering, however Casuarina , which
survived the first cyclone well, showed significant levels of mortality. Sauer also noted that
storm drift was stopped by mangroves (Rhizophora mucronata) which survived well and
appeared to attenuate the effect of wave energy on the vegetation behind them. All of these
accounts indicate that coastal vegetation was highly resistant to persistent effects of salt, spray
and periods of inundation. Even when trees were blown over, and roots undercut by waves, most
retained the capacity to re-sprout. For trees that suffered significant mortality, regeneration by
propagule was rapid.
The specific assemblage of species able to grow at any particular coastal location on
tropical Pacific islands is strongly influenced by climate and the type of coastal ecosystem most
importantly whether the site is a sandy beach or rocky coastline or whether the site is exposed to
wave action, wind, and salt spray or whether the site is protected such as in a bay or harbor
(Richmond and Mueller-Dombois 1972, Mueller-Dombois and Fosberg 1998). Most studies have
focused on particular species assessing species and population parameters (Tanaka et al. 2007).
Few studies have specifically addressed the dynamics of a diverse native coastal vegetation
21. 21
community such as the interaction between species of various forms of both below and above
ground structures.
While most research to date has focused on natural (although undoubtedly
anthropogenically influenced) systems and simulated models of those systems, few studies have
applied the theories generated by coastal green barrier research to specific design
recommendations for denovo coastal bioshield construction (Tanaka et al. 2009). Further, other
studies or coastal projects which have incorporated the findings of coastal bioshield research into
the restoration of native coastal ecosystems were not found. The de-novo design of a bioshield
should consider not only the performance of the green barrier during a tsunami or storm surge
event but also how the event may change the structure of forest following the event (Hayasaka et
al. 2009). Finally, many have pointed out that coastal bio-shield designs must take into account
expected changes in sea-level rise which are likely to alter coastal vegetation assemblages
(Greaver and Sternberg 2007).
The 2004 Indian Ocean tsunami also prompted research on coastal forest rehabilitation
(restoration) and site-specific manuals for coastal re-forestation have been developed (Hanley et
al. 2008).The majority of coastal revegetation efforts and related research has focused on the
reestablishment of mangroves (Chan and Ong 2008).Most of the published reports on non-
mangrove coastal forest rehabilitation or re-vegetation are general guidelines for the
implementation of coastal reforestation projects rather than technical reports based on completed
projects. One technical report following the successful implementation of a coastal reforestation
project in Tonga in the mid 1990’s although not implemented specifically with the idea of
producing a tsunami or storm surge bio-shield, provides good technical information useful for
the planning of similar projects on tropical Indo-Pacific Islands (Thaman et al. 1995). A
summary of the findings from that report is given below. In addition, the report provides species
specific propagation and performance information.
Case study: Coastal re-vegetation at Tonga
In response to the negative effects of coastal deforestation in Tonga, including salt spray
damage to crop plants and structures and the loss of species of cultural importance, a coastal
reforestation program was launched at Houma on the South west coast of Tongatapu (Thaman et
al. 1995). The re-vegetation zone ranged from about 5 to 25 m (15 to 75 ft), and averaged 12 m
22. 22
in width and 2 km long (36ft in width and 1.3 miles long). The project began in 1993. Alien
undesirable species were removed from the re-vegetation area from 1993-1994. Plantings,
fencing and signs were used to demarcate the re-vegetation area. Coastal species were collected
from natural populations and grown at a nursery until large enough for out-planting to the re-
vegetation site. The project concluded in 1995. Total direct costs were USD 12,000. Estimated
man-hours over the two year project was 11,858 with an average of 12.5 days worked per month.
The project relied upon involvement and volunteer from nearby communities. Planting was done
in three phases involving the initial planting of highly salt tolerant pioneer species, the
subsequent planting of salt tolerant non-pioneer species and finally the enrichment planting of
key species. Site maintenance included regular weeding, periodic watering during times of
drought and the addition of soil amendments. Plantings were done in sections to ensure that: 1)
there are enough trees and associated vegetation to establish a good windbreak and a substantial
zone of salt-tolerant and fire-resistant vegetation; 2) to facilitate care and maintenance in the
early stages of reforestation; and, 3) to monitor and learn from the performance of the trees in
initial plantings. Pioneer species planted as part of the study included: Pandanus tectorius,
Hibiscus tiliaceus, Excoecaria agallocha, Calophyllum inophyllum, Scaevola taccada,
Terminalia catappa, Terminalia litoralis, Casuarina equisetifolia, and Tournefortia argentea.
These species were planted to provide a protective buffer for the establishment of the non-
pioneer species. Non-pioneer species were planted beginning six months after the planting of
Pioneer species. These included: Neisosperma oppositifolium, Hernandia nymphaeifolia,
Barringtonia asiatica, Vitex trifolia, Cerbera manghas, Cocos nucifera, Pisonia grandis,
Guettarda speciosa, Acacia simplex, and Cerbera odollam. Following the establishment of non-
pioneer species, enrichment plantings involving species that were harder to propagate or which
required even greater levels of protection from exposure begun to "give the resultant forest
greater species diversity and greater cultural utility; and to enrich the species composition of the
original forest”.
23. 23
Summary of Coastal Vegetation, Tsunami, and Storm Surge Studies.
Following the Indian Ocean tsunami of December 26, 2004, the anecdotal accounts of
survivors and the observations of researchers pointed to a potential role of coastal vegetation in
mitigating damage and in some cases reducing the death toll due to a tsunami. The hypothesis
generated by these observations was that coastal vegetation could provide a protective barrier
against tsunami inundation. By extension, research has also turned to the role of coastal
vegetation to mitigate against risks related to storm surge, erosion and the effects of projected sea
level rise which is expected to exacerbate the effects of all of the above named natural
phenomenon (IPCC 2007).
It has been observed that coastal vegetation can stop rocks, debris, ships and other
material carried by the tsunami wave from reaching land and causing destruction. It can act as a
safety net and vertical escape for people who might be trapped in a wave and who would
otherwise be washed out to sea. Coastal vegetation also traps sand forming sand dunes, reduces
erosion, and traps organic matter which together act to build elevation, increasing the beach
slope and therefore reducing the ability of some waves to inundate the land. Vegetation has been
found to slow down an oncoming tsunami wave, reducing the force of the wave and its
destructive potential. Vegetation may also prevent debris and soil from land from being washed
into the ocean providing a protective effect for coral reef and other near shore ecosystems which
are doubly affected by the direct effect of tsunami and storm surge and the subsequent input of
harmful materials from land. In addition to protective effects, coastal vegetation provides other
important services including providing habitat for seabirds, turtles and other animals and a
potential resource base for people who may use the products of the coastal forest for food,
recreation, materials, medicines, and many other uses (Thaman 1992).
24. 24
Figure 2. An illustration of the proposed interactions between dense coastal vegetation and tsunami or storm surge
waves. A bio-shield may reduce the velocity of an incoming wave, build elevation by trapping sand and organic
matter, reduce foreshore erosion, reduce damage to structures through a reduction in wave energy due to hydraulic
resistance and reflection, hold back coral, debris, & ships carried by the wave which may cause damage inland or
may cause damage offshore to sensitive reef ecosystems., and provide a vertical escape for people, Coastal
vegetation may however be a source of floating woody debris which may cause damage inland. The configuration of
coastal vegetation can alter the tsunami flow direction and flow speed
Evidence from wave-tank studies and mathematical modeling provides evidence to
support the hypothesis that coastal vegetation should be able to attenuate the energy of tsunami
or storm surge waves. Further, these studies suggest that greater vegetation density and greater
surface areas (in terms of leaves, branches, roots and stems) within the inundation depth of a
wave, should increase the resistance of the coastal vegetation on an incoming wave. In addition
the specific configuration of coastal vegetation, like any other barrier, may change the flow of an
incoming wave. In the case of a straight channel through the vegetation perpendicular to the front
of the incoming wave, the configuration may channel water, increasing its velocity potentially
resulting in increased damage inland of the gap. A major criticism of these highly controlled
studies is that they may fail to adequately simulate complex natural-system parameters.
Evidence from post inundation studies, including field studies and studies based on
remotely sensed data are equivocal and all studies performed to date have been subjected to valid
criticisms. Critics and proponents alike conclude that variations in bathymetry, increased
distance from shore, and increased elevation reduces risk from a tsunami (Cochard et al. 2008).
With bioshield
Without bioshield
25. 25
Once these factors are taken into consideration, many studies point to, but have not conclusively
demonstrated, some protective role of coastal vegetation. Given the equivocal results of bio-
shield research, Baird and Kerr (2008) concluded that, "There is, in fact, no empirical data
published to date to suggest that forests provided any meaningful protection from the Indian
Ocean tsunami and much to refute it."
The conflicting results and interpretations of the data from field studies is likely due to
the highly uncontrolled, extremely complex, and temporally rare and ephemeral situations which
characterize these natural events. It is important to note that many of the studies that have
addressed these questions were undertaken in areas inundated by the 2004 Indian Ocean tsunami
which was larger in magnitude than most tsunami events. In addition, most studies have taken
place in a very limited number of localities. One reason for this limited sample is simply that
field researchers must wait for a tsunami or major storm surge event in order to study its affects,
in addition, since there are so many environmental factors that may affect patterns of inundation
and wave-induced damage, it is important to compare sites that share many characteristics yet
vary, along its coastal extent, in certain variables of interest such as vegetation structure or
density, the presence/absence of sea grass, or abundance of coral reefs.
26. 26
Observations from Upolu Samoa
Methods
In January 2010, The University of Hawaii team made observations in six areas on the
south shore of Upolu Samoa which had been inundated by the September 29, 2009 tsunami.
Maximum inundation points were recorded with a handheld Garmin Rino 530 GPS unit. All
position points were averaged for 60-70 seconds to improve precision. Ground scour and ferns or
herbaceous plants killed by salt-water inundation were consistent indicators of the maximum
inundation extent. At randomly selected points in areas where coastal vegetation fronted the
shore, the vegetation structure was assessed using the variable area transect method along 2-3
transects at each site (Sheil et al. 2003). The transects were set to run from the beginning of the
woody coastal vegetation at the top of the beach perpendicular to the shore for 20, 30, or 40
meters inland. The outer boundaries of coastal forest were mapped with a GPS. All points were
projected in ArcMap for analysis. The GPS boundaries of coastal forests were converted to forest
polygons and conformed well to satellite imagery. For inundation and damage assessment points,
the following were measured using measurement functions in ARCMap: distance to shore for
each inundation point, proportion forested along a straight line transect from each point to shore,
reef distance from the closest coastal point, for each point, to the closest point on the fringing
reef. Elevation for all points was interpolated using a 2m contour layer supplies by the
government of Samoa. Slope of the foreshore was recorded using a clinometer.
Structures within the inundation zone were assessed for damage on a 3 category scale (1)
undamaged or damage to contents only, (2) moderate damage including significant damage to
doors, windows or partial collapse of attached structures such as cook houses, (3) destroyed
wood frame or cement house. Species survivorship of woody plants encountered along transects
within inundated areas was assessed on a three point scale (1) unaffected, (2) recovering, (3)
dead/not recovering). In a few areas with extensive erosion along the coast, the volume of sand
remaining behind isolated trees was measured.
27. 27
Results
Observations are presented separately for each of the five sites visited (figure 3).
Figure 3. Observations were made at five sites along the coast of South Upolu. From west to east these are Coconuts
beach Resort, Saleilua, Tafatafa, Utulaelae/Sapoe, and Saleapaga.
Site 1: Coconut Beach Resort
At the Coconuts resort site (figure 4), there was substantial damage to the resort itself and
the numerous structures likely shielded a few houses just behind the resort. The tsunami travelled
through a wetland of the sedge Scirpodendron ghaeri and the fern Acrostichum aureum,
bordered by very dense H. tiliaceus to reach a maximum run-up elevation of approximately 4
meters (mean 3.243, standard error 0.178, standard deviation 0.472). The maximum inundation
distance (max 372.5 mean 293.9, standard error 26.5, standard deviation 70.2) was clearly
dependent upon elevation (see image). UNESCO reported a maximum run-up elevation of 5
meters with an inundation distance of 95 meters to the east of this location.
28. 28
Figure 4. Inundated area near the Coconuts Beach Resort, Upolu, Samoa. The red line, and marked inundation
points indicates the inundation limit of the tsunami wave. Elevation (in meters) courtesy of the Government of
Samoa and James Atherton. Projection: WGS 1984.
The owner of the resort reported that most of the planted ornamental vegetation was
destroyed. Other than uprooted trees within a few meters of the coast, there was little to no
apparent damage due to inundation to trees and woody shrubs in the natural area west of the
resort. Herbaceous plants and ferns were missing from the understory due to ground scour. No
transects were carried out in this area due to presence of the dense (nearly impenetrable) coastal
wetland dominated by H. tiliaceus and Scirpodendron ghaeri. Although it was reported that the
vegetation near this resort may have had a protective effect, our team could find no evidence of
that claim, as all undamaged structures within the inundated areas were near the edge of the
inundation zone or were blocked by the physical structures of the resort including several large
retaining walls. Species encountered at this site included Cocos nucifera, Hibiscus tiliaceus,
Terminalia cattappa. Hernandia nymphaeifolia, Ardesia eliptica and, Scirpodendron ghaeri.
29. 29
Site 2: Saleilua village
West of Saleilua village the team recorded observations in two locations. The first
location was near a peninsula along the road leading to the Ili’ili Beach Resort (figure 5) which
had been completely destroyed by the tsunami wave and has subsequently been abandoned. At
this location a sea wall (see image) made up of large boulders had been destroyed and the
boulders moved inland, in some cases over 100 meters (see point 216). A similar boulder field
originating from a seawall was recorded in the village of Satitoa by the Unesco team (Dominey-
Howes and Thaman 2009). The resort development was situated on land with an elevation of less
than 2 meters above sea level. The coastal road leading to the resort passed through a coastal
forest that extended to the shore surrounded on both sides along the coast by areas cleared of all
but a few trees. The vegetated area was approximately 20 meters wide and 50m long with the
long side running parallel to the coast. Although this area had been inundated, there was no
visible damage to trees and shrubs within or behind this thin vegetated strand and there was little
damage to the road that ran just behind the vegetated buffer. Considerable erosion and road
damage and damage to trees in the relatively cleared areas on either side of this vegetated area
was observed (Figure 5), however it cannot be determined if this damage was due to the
proximity of the road to shore or due to a buffering effect of the vegetation.
At this site, the maximum inundation distance was approximately 83 meters. And the
average run-up height was 1.6 Meters (Stdev = 0.56m). The team recorded inundation points
along this coast to determine if there was any detectable affect of that vegetation on inundation
distance or run-up. The analysis of these points is inconclusive since there was no evidence as to
the direction of the tsunami flow, and it appears likely that the tsunami wave(s) past through the
peninsula and struck the vegetated area from a direction nearly parallel to the coast. The slope of
the shore in the cleared area to the south of the vegetated area was 7% (4 degrees) (see point 230
which was approximately 18 meters from approximate sea level with a measured elevation of
approximately 1.3 meters). In the gap, (see point 228), 88 meters from shore, there was moderate
damage to a house (now abandoned) and the team was informed by a local resident that the
damage was due to the tsunami wave. The gap to the north of the vegetated area had a foreshore
slope of 8% or 4.5 degrees and point 231 was 13.5 meters from approximate sea level, for an
approximate elevation of 1.1 meter). The vegetated area had a somewhat higher elevation and
greater foreshore slope of 14% for a distance of 6.2 meters to the high tide mark.
30. 30
Figure 5. Inundated areas at the Iliili Beach resort area in Saleilua Samoa.
The vegetated area near Ili’ili resort was predominantly composed of dense Hibiscus
tiliaceus and Dendrolobium umbellatum. With very high densities (3167 stems per hectare) of D.
umbellatum in the first 10 meters from shore and 1146 stems per hectare of H. tiliaceus
throughout the area (Table 2). Morinda citrifolia was also a co-dominant species (1129 stems/ha)
along with H. tiliaceus in the inland 10 meters of the area. Other species encountered at low
densities in this site included: Asplenium nidus, Cocos nucifera, Premna serratifolia,
Baringtonia asiatica, Terminalia samoensis, Glochideon ramiflorum, and Tacca leontepaloides.
All of these species appeared unaffected by both the force of the tsunami wave and by salt water
inundation however two of two individuals of Macaranga sp. found within this area were dead
likely due to salt water inundation. The trees in this area would only have experienced inundation
of about 1 meter above ground at maximum. Average estimated height of the trees in this
vegetated area was 5 meters (SE = 0.474) and the base of the canopy was at approximately 3
meters (SE = 0.382), mostly above the inundation depth.
Significant erosion and
undercutting of the road
31. 31
Table 2. Stem density and mean basal area of dominant shrubs and trees in vegetated area on the road to Iliili
Saleilua.
(Distance
from Shore)
Avg. Plot
Area
Relative density stems/ha Mean basal
area
(cm2/m2 or
m2/ha)
D.
umbellatum
H. tiliaceus M.
citrifolia
Total
(0-10m) 13.4 3166.6 0.0 0.0 3589.8 11.0
(10-20m) 10.4 0.0 2292.8 1128.7 4056.4 45.4
Average 11.9 3166.6 1146.4 1128.7 3823.1 28.2
On the peninsula at Saleilua the following species were observed: Cocos nucifera,
Pandanus tectorius, Terminalia cattappa, Baringtonia asiatica, and Crinum asiatica. All
appeared to have survived the tsunami well. The Pandanus in this case were tall and mature,
growing in full sun with numerous above ground roots. A pre-tsunami image of the peninsula
area was found on the internet and a comparison with a post tsunami image shows that most trees
in this very sparsely planted area in fact survived the force of the wave while all of the buildings
were destroyed or severely damaged (figure 10).
At the second location in Saleilua, witnesses reported that the wave came from nearly
perpendicular to shore. The maximum inundation distance at this site was 175 meters and the
maximum run-up elevation was 7m. This high elevation measurement was likely due to a low
slope in this particular area. The average run-up elevation was 6.2 m SE 0.24 Stdev 0.59.) and
the average inundation distance was 144.83m (SE 9.37 Stdev 22.94.) The measured foreshore
slope in this was 8%. The tsunami wave pasted through approximately 20 meters of coastal
vegetation and then through a wetland of Erythrina fusca. The only house in this area (which
was not fronted by substantial vegetation) was completely destroyed and the owners were
rebuilding approximately 200 meters inland.
The team completed one 30 meters long vegetation transect at the second Saleilua
location. This vegetated area was composed of a mix of species without any clear dominant. This
appears to be due to plantings and clearing by the owners of the land. The most commonly
encountered species along this transect was the ornamental Ixora findlaysoniana. Hibiscus
tiliaceus and M. citrifolia were also relatively abundant. Other species encountered at this site
included Leucaena leucocephala, Barringtonia asiatica, Inocarpus fagifer, Mangifera indica,
Adenanthera pavonina, Metroxylon sp., Artocarpus altilis, Terminalia cattappa, and D.
umbellatum. There was little physical damage to vegetation in this area with some exceptions. A
32. 32
single individual each of Cananga odorata, Dysoxylum samoense, and Psidium guajava were
dead, apparently killed by defoliation following inundation. The wetland stand of Erythrina
fusca was completely defoliated by the salt water inundation. According to the land owner, this
occurred several days after the tsunami and he did not know if the trees would survive. Although
there was some indication of regrowth near the base of the stems on many individuals there was
significant top kill and it is unclear if the stand will recover. The herbaceous weed Physalis
angulata (wiwao) was a prominent feature of the understory, likely re-growing following the
near complete removal of herbaceous vegetation by the tsunami.
A single house in this area which was not fronted by substantial vegetation was partially
destroyed. Fronting this vegetation was a low to moderate level of erosion with indication that
some of the initial trees were destroyed by the wave. The village proper of Saleilua, just east of
this vegetated area was not inundated by the tsunami given its elevation mostly above 6 meters.
The UNESCO report found a maximum run-up elevation of nearly 4 meters and an inundation
distance of approximately 23 meters near this village.
Figure 6. Inundated area at Saleilua, Samoa. West of village.
33. 33
Table 3. Stem density and mean basal area of dominant shrubs and trees in vegetated area west of Saleilua village,
Upolu, Samoa.
(Distance
from Shore)
Avg. Plot
Area
Relative density stems/ha Mean basal
area
(cm2/m2 or
m2/ha)
I.
findlaysonian
a
M.
citrifolia
H.
tiliaceus
Total
(0-10m) 27.4 479.8 533.3 0 1510.0 13.4
(10-20m) 36.0 327.6 223.7 223.7 1118.3 36.1
(20-30m) 114.4 0 0 169.4 771.9 32.1
Averages 59.2 403.7 378.5 196.5 1133.4 27.2
A Unosat Image of the area, taken immediately following the tsunami, shows damaged
areas which are similar to the findings in the field. In the area west of the village (figure 6), no
damage is visible from the Unosat image probably due to the high cover of trees. The Erythrina
fusca began to defoliate following the recording of post tsunami aerial imagery and no detectable
damage was visible in those images.
Figure 7: Extensive Damage to a resorts structures at the Peninsula West of Saleilua
34. 34
Figure 8: Trees remain standing at the Iliili resort although buildings in this area were completely destroyed. This
area was partially cleared and represents a low density cleared forest.
Figure 9: Large boulders from a coastal seawall, including the one shown here, were carried over 100 meters inland
by the wave.
35. 35
Figure 10. Before (top) and after (bottom) images at the Iliili resort in Saleilua. Nearly all the buildings at the resort
were destroyed however most of the trees remained except several of the coconut trees closest to the coast. The rock
wall facing the ocean was destroyed and rocks were carried inland up to 100 meters or more. Previous to the
tsunami, this area had most of its coastal vegetation removed, the remaining trees did not provide a significant
barrier to the tsunami wave.
36. 36
Figure 11. Coconut tree roots planted near to a sea wall likely helped to prevent these rocks from being moved
inland
Figure 12. This image shows the south edge of the densely forested buffer between the road to the Iliili beach resort
and the coast. The vegetation is situated on (or may have contributed to the slight increase in elevation visible along
the shore front.
37. 37
Site 3: Tafatafa
At the Tafatafa Village site (figure 13), a coastal area 734 meters long was surveyed.
Along this area there were areas with dense trees (a), areas with trees partially cleared (medium
density) (b), and completely cleared portions (c). The elevation of inundation points were
interpolated from the elevation contour GIS layer. In these areas there was very little erosion
observed (even to un-vegetated areas) indicating that the effect from the tsunami was low. There
appears to be very little difference in run-up elevation along this site. Four of the inundation
points (176-179) are minimum estimates as the wave seems to have inundated the wetlands
situated just behind these points, however there was no further sign of the inundation extent. In a
simple regression using 1= forested and 0 = cleared on the inundation distance, there was no
relationship between the variables (F = 0.21 P = 0.664) a regression of forested / cleared on the
elevation at inundation points showed a statistically significant positive relationship between
forested and the elevation. A positive affect between forest and inundation is an unlikely
outcome and other variables are likely confounding the results in this case. The interpolation of
elevation values are not able to detect fine scale differences in elevation which are not featured
on the 2m interval GIS layer. It is a more plausible conclusion that the presence / absence or
density of vegetation had no effect on inundation distance and run-up elevation at this site. One
other possibility is that the team may have failed to detect the true inundation distance. This is
possible if substantial regrowth of the understory had occurred. The elevation values (n = 9) were
normally distributed (Anderson-Darling Normality Test A squared = 0.41 P = 0.269 while
inundation distance was non-normally distributed (A squared = 0.79, P = 0.025) indicating that
inundation distance was mostly dependent upon the run-up elevation.
38. 38
Figure 13. Inundated area at Tafatafa, Upoly, Samoa. This area consisted of dense forest (A), partially cleared areas
of low density forest (B) and completely cleared areas (C). The presence, absence and density of forest did not
appear to affect inundation distance or run-up height in this area.
The team completed two transects at the site. Tafatafa transect #1 was a partially cleared
site likely part of a nearby beach fale establishment. Several beach fale showed moderate
damage. At this site, most of the understory vegetation had been cleared resulting in a lower
density of total stems (mean = 481 stems per hectare) compared with 2148 stems per hectare at
transect #2.
Tafatafa transect #1 included what appeared to be predominantly planted vegetation. The
most common species encountered in this transect was Fluggea flexuosa of which 4 out of 9
were dead likely due to inundation. Morinda citrifolia and Flacourtia rukam were the second
most abundant species along the transect. Tafatafa transect #2 was a much more densely
vegetated site. There was little to no observable damage to vegetation at both sites with the
exception of the death due to salt water inundation of most individuals of Fluggea flexuosa.
Species encountered at this site include D. umbellatum, Cerbera manghas, Barringtonia
asiatica, Hernandia nymphaeifolia, Hibiscus tiliaceus, Callophyllum inophyllum, Morinda
a
b
c
39. 39
citrifolia, Leucaena leucocephala, Cocos nucifera, Ficus tinctoria, Fluggea flexuosa, Flacourtia
Rukam, Glochideon ramiflorum, Premna serratifolia, Scirpodendron ghaeri, Sophora tomentosa.
Several Fluggea flexuosa in this area were dead.
Site 4: Utulaelae-Sapoe villages
The neighboring villages of Utulaelae and Sapoe (figure 14) represent an interesting case
where vegetation may have provided significant protection from the full damaging effect of the
tsunami wave. Utulaelae had previously cleared the vegetation fronting the village whereas the
village of Sapoe maintained an approximately 30-50 meters wide strip of vegetation between the
village and the shore.
To the west of Utulaelae the coastline shift toward the North and leads to a river
approximately 0.5 kilometers away. Between the village and the river is a wetland known as Fusi
pu which is primarily composed of the sedge and Pandanus look-a-like Scirpodendron ghaeri.
The tsunami wave(s) moved through this wetland causing significant uprooting of this sedge
however many were recovering. The leaves of this sedge presented a very useful indicator of the
inundation distance of the tsunami. At the edge of the wetland these leaves were found up to 2.5
meters in the dense H. tiliaceus trees which bordered the wetland. (See points 82, 83, 84, 85, 87).
The wave swept through the dense H. tiliaceus up to a run-up elevation of 4.5 meters, consistent
with the run-up elevation in Utulaelae, Sapoe, and the forested area East of Sapoe.
Between the wetland and Utulaelae there were several low walls which seems to have
held back the total inundation distance of the wave. Banana leaves from Utulaelae that were
planted along the southwest corner of the village were swept into this area (see map Green
Circles) and these met with leaves from the wetland sedge (see map Red Circles).
At the Utulaelae-Sapoe Site, 30 inundation points were recorded over a distance of 1.2
kilometers. The average inundation distance was 117.13 (stdev.=69.24) meters while the average
run-up height was 4.2 +/- stdev. = 0.5981) meters above sea level. Eight of these points (72-79)
were omitted from subsequent analysis since the wave in this area was obstructed by a number of
low walls and the values recorded at these points do not represent the maximum inundation
extent, another point (100) was omitted due to its close proximity to point 119. With the omitted
points the mean inundation distance was 141.68 m stdev = 37.34 and for elevation 4.35 with a
40. 40
stdev of 0.6014. Both values were approximately normally distributed (Anderson-Darling
Normality test Distance to shore p = 0.351, Elevation p = 0.083).
Figure 14. Inundated areas at Utulaelae and Sapoe villages. These villages primarily differed in the presence (Sapoe)
and absence (Utulaelae) of a coastal forest between houses and the beach.
A correlation matrix was used to investigate the data, there were no apparent correlations
between Elevation, Percent forested, distance to shore, and distance to reef. A regression of
percent forested on the inundation run-up heights of the various points found no significant
relationship (r2
= 0% and p=0.922).
For the 29 houses assessed for damage in the area (figure 15), 19 were found to have little
to no observable damage, (damage to contents was reported but not assessed as part of this
study). Moderate to severe damage was observed to 10 structures (assessed damage value of 1).
These included 4 post houses that were knocked over during the tsunami (assessed damage value
of 2), 3 wooden frame houses that were destroyed (assessed damage value of 3), and 3 cement
houses that sustained damage, two of which were completely destroyed (assessed damage value
Utulaelae
Sapoe
41. 41
of 4), and one which sustained moderate damage to doors, and windows (assessed damage value
of 2). Data was first explored using a correlation matrix. Significant p values were observed
between the Damage value and Elevation (p = 0.011), Distance to shore (p=0.001) percent
forested (p=0.005) and Reef length (p=0.016).
The elevation of structures in both villages were significantly different: T-Test of
difference = 0 (vs. not = 0): T-Value = 3.75 P-Value = 0.003 DF = 12. The distance to shore
between structures in the two villages were not significantly different T-Test of difference = 0
(vs. not =): T-Value = 0.73 P-Value = 0.479 DF = 16
Elevation Distance to shore
N Mean St Dev. SE
Mean
Mean St Dev. SE
Mean
Sapoe 17 3.765 0.193 0.047 98.9 15.3 3.7
Utulaelae 12 3.125 0.567 0.16 92.7 26.8 7.7
Figure 15. Damage assessments at Utulaelae and Sapoe villages, Upolu, Samoa. Damage was assessed on a three
point scale, Minimal damage (Cosmetic Damage to structure or damage to contents only), Moderate damage
(Structural damage requiring repairs), and Major damage (Damage to structure requiring rebuild).
42. 42
Ordinal logistic regression analysis, using percentage forested, distance to shore, distance
to reef and elevation on the response variable “Damage classification” was utilized. Only
distance to shore and Percentage forested were significant variables and the ordinal regression
was repeated with only these two predictive variables. Both variables was a significant predictor
of damage Distance to shore (p=0.008 and Percentage forested (p = .02). Therefore, greater
damage was associated with a closer distance to shore and with lower percentage forested.
Site 5: Saleapaga village
The tsunami was very powerful destructive in the area of Saleapaga (figure 16) as
indicated by the highest wave heights and greatest levels of damage (Cite the UNESCO report).
At this site, most of the coastal area had been cleared for villages and for the tourism industry.
There were very few areas that had natural vegetation and nearly all houses in this area were
destroyed. Satellite imagery shows vegetation behind the houses, most of this was destroyed by
the tsunami wave and much of the debris had already been cleared in this area preventing us
from making a clear assessment of the area (figure 16). A line of trees which were present at the
coast remained standing allowing us to investigate the relationship between tree roots and
erosion (See erosion section). This area had significant levels of erosion which removed a large
amount of sand and altered the coast line.
The wave inundated the entire coastal plain and travelled several meters up the wall of
the mountain. Inundation points were obtained however elevation contours were not obtained for
this site and their analysis would not be appropriate given the very steep slope in this area and
the restricted resolution of the teams GPS unit. The UNESCO study found a maximum run-up
elevation of between 5 and 6.5 meters.
43. 43
Figure 16. Inundated coastal plain showing debris field at Saleapaga.
Combined Vegetation Analysis
Nine variable area transects were carried out at 4 sites within the inundation zone, stem
density and basal area were calculated based on the methodology of (Sheil et al. 2003). Using a
General Linear Model with site and 4 plot distance categories, differences in stems per hectare
among plots were found to be related to distance from shore and site. Differences between sites
were highly significant (F = 10.03 P < 0.001) and with a general near-significant trend towards a
greater number of stems closer to shore (F = 2.94 P = 0.059). Using a General linear model, there
were no significant differences in mean basal area per plot either among sites or in distance to
shore (F = 0.80 P = 0.609, F = 0.51 P = 0.683) (figure 17). These results include the partially
cleared Tafatafa_1 site which had been cleared of most small trees and understory vegetation.
Table 4. Comparison of the number of stems per hectare by distance from the beach in 10 meter intervals. Statistical
grouping uses the Tukey Method and 90.0% Confidence interval. Means that do not share a letter are significantly
different.
Distance
from Shore
Sample
Size
Stems per
hectare
Mean (Stdev)
Grouping
(Stems per
hectare)
Basal area (cm2
/m2
or m2
/ha)
Mean (Stdev)
(0-10m) 9 1931 (1061) A 51.4 (54.1)
(10-20m) 9 1550 (1098) A B 37.86(18.56)
(20-30m) 7 1070 (455) B 30.64(9.28)
(30-40m) 6 1211 (670) A B 30.76(20.85)
44. 44
(30-40m)(20-30m)(10-20m)(0-10m)
3000
2500
2000
1500
1000
500
Plot
Stemsperhectare Interval Plot of Stems per hectare
95% CI for the Mean
Figure 17. Statistical analysis of stems per hectare.
Survivorship of species encountered in inundated areas of all sites on the south coast of Upolu
January 2010.
Overall there was little apparent damage to coastal species with the exception of the
wetland mangrove Bruguiera gymnorhizza. Although this species was reported to survive well
by the UNESCO report, the team observed many standing dead trees at several sites. Mortality
was observed in eight additional species including Cananga orodata, Dysoxylum sp., Macaranga
sp., Musa sp, Psidium guajava. Some individuals of Artocarpus altilis, Fluggea flexuosa, and
Pandanus tectorius were found to have suffered some mortality however the team also found
individuals of these species that had survived inundation. Identification was made of eleven
species that were initially defoliated and which were slowly recovering. In some cases these
species experienced top kill and were resprouting at the base. Twenty seven species appeared to
be unaffected 110-130 days following the September 29, 2009 tsunami (Table 5).
45. 45
Table 5. Species in inundated areas of Upolu Samoa 110-130 days following the September 29, 2009 tsunami
Species unaffected (27 species)
Adenanthera pavonina (n =3), Araucaria sp.(n = 2), Barringtonia asiatica (n = 7), Bischofia
javanica (n = 1), Calophyllum inophyllum (numerous), Ceiba pentandra (n = 1), Cerbera
manghas(n = 17), Cocos nucifera (numerous), Dendrolobium umbellatum (numerous),
Erythrina variegata, Ficus elastica, Ficus tinctoria (n = 4), Geniostoma rupestre (n = 2),
Glochidion ramiflorum (n = 3), Hernandia nymphaeifolia ( n = 6), Hibiscus tiliaceus
(numerous), Intsia bijuga, Ixora finlaysoniana (n = 7), Leucaena leucocephala (n = 7),
Metroxylon warpurgii (n = 2), Morinda citrifolia (numerous), Pandanus tectorius(n = 8)*,
Premna serratifolia (n = 5), Psychotria insularum (n = 2), Terminalia catappa ( n = 10),
Terminalia samoensis (n = 1), Thespesia populnea (n = 2).
Species affected and recovered over time (11 species)
Artocarpus altilis (n = 2), Asplenium nidus (Numerous), Bruguiera gymnorhizza (Numerous),
Erythrina fusca, Flacourtia rukam (n = 7), Flueggea flexuosa (n = 28 regrowing from base),
Inocarpus fagifer (n = 4), Mangifera indica (n = 3), Plumeria rubra (n =2), Scaevola taccada
(n = 5), Scirpodendron ghaeri (numerous), Tournefortia argentea (n = 4)
Species affected and not recovered (9 species)
Artocarpus altilis (C)**, Bruguiera gymnorhizza (numerous), Cananga orodata (n = 1),
Dysoxylum sp. (n=2), Macaranga sp. (n = 2), Musa sp.(numerous) (C), Pandanus tectorius(n =
2)*, (N), Psidium guajava (n=2) (N), Fluggea flexuosa (n = 7) (approximately 20%),
Species encountered in inundated areas of Upolu Samoa 110-130 days following the tsunami.
* Pandanus in open areas appeared to be unaffected by inundation whereas in areas with a
dense over storey found instances of dead Pandanus.
** Observed was both dead and recovering Breadfruit trees. (The UNESCO report suggested
that some varieties appear to have a higher level of tolerance of saltwater inundation.)
Erosion
At Saleapaga, tree roots held back soil and sand and the amount of soil and sand that was
not removed due to the presence of the tree’s roots was estimates by measuring the length / width
and height of remaining sand behind the tree and calculated the volume of soil/sand held back by
the roots. The team measured to the back of the adjacent groove, and it cannot rule out however
that some of the groove area of the erosion front may have been exacerbated by an increase in
flow velocity of the retreating water as it washed around the trees. Clear estimates measures from
13 coconut trees and 5 broad canopy coastal trees were made. The volume of soil/sand behind
coconut trees was normally distributed (Anderson-Darling Normality Test: A squared = 0.22, P =
0.776.) with a mean of 5.248 m3
(figure 18). This volume of sand is related to the width and
depth of coconut roots, these are for the most part vertically oriented and extending 1-2 meters
46. 46
into the ground. It was not possible to get a significant sample size of the other species sampled.
It appears from the limited data that other species would show a wide range of values depending
upon the total size of the tree. Species and age may also be important factors that determine the
volume of sand that these trees have the capacity to retain. Canopy extent may be a good
indicator of total root extent which may estimate the potential volume of sand held back during a
sudden coastal erosion event.
Table 6. Average volume of sand/soil remaining behind isolated trees at the coast in Saleapaga, Upolu, Samoa.
Species
Length
(m)
Width
(m) Height (m)
Volume
(m3) StDev
Sample
Size
C. nucifera 2.724545 2.127273 0.927272727 5.248 2.649 13
B. asiatica 3.245 2.35 1.1 9.79 11.93 2
H.nymph 2.18 5.25 0.53 6.0659 na 1
T.populnea 4.98 2.46 0.74 8.94 2.08 2
Figure 18. Histogram of the volume of sand/soil held back by isolated trees in Saleapaga village following the 2009
tsunami. Trees were classified into coconut tree and ‘other’ for this analysis.
201612840-4
4
3
2
1
0
Volume
Frequency
5.248 2.649 13
8.708 6.247 5
Mean StDev N
Coconut
Other
Species
Histogram of Volume
Normal
47. 47
The Coastal Flora of Hawaii
Introduction
The native coastal flora of Hawai`i has been significantly altered beginning with the
arrival of the Polynesians (Kirch 1994). Coastal areas were among the first cleared for human
use and extensive development at the coast for habitation, recreation, and tourism continues
through the present day. Coastal ecosystems have also been impacted by clearing for agriculture
and by the introduction of alien ungulates, rodents, and invasive plant species. As a result, the
coastal ecosystems of the main Hawaiian Islands are highly altered from their pre-human state
and in fact very little remains to provide evidence for what the coasts of Hawai`i looked like
prior to humans arrived.
There have been several efforts to characterize the coastal vegetation communities in
Hawai`i and to define the environmental factors that structure them. The most recent treatments
include a detailed analysis of coastal ecosystem on Oahu (Richmond and Mueller-Dombois
1972) as well as descriptive summaries of coastal ecosystem from the entire Hawaiian
archipelago (Gagne and Cuddihy 1990, Mueller-Dombois and Fosberg 1998). A more recent
work provides an up to date assessment of the remaining native coastal vegetation communities
(Warshauer et al. 2009).
Richmond and Mueller-Dumbois (1972) conducted transects and vegetation releves at 22
locations on Oahu. They documented 13 ecosystem types characterized by the following
dominant species or species combinations, Hibiscus tiliaceus, Scaevola taccada, Chlois
barbata/Sida fallax, Chloris barbata/Prosopis pallida, Prosopis pallida, Batis maritima,
Rhizophora mangle, and Scirpus californicus/Eichornia crassipes. Of these, Prosopis pallida
(mesquite, kiawe) and Rhizophora mangle (Red Mangrove) are the only true tree species,
Hibiscus tiliaceus is a large “megashrub” and Scaevola taccada is a medium shrub. The
remaining ecosystem types represent coastal grasslands or wetlands. The authors argue that
coastal ecosystem on Oahu are primarily structures by wind exposure, rainfall, and substrate
salinity. The later factor is somewhat dependent upon soil characteristics and hydrological
properties of the area. Further the authors define the coastal floristic zone as the inland extent of:
saltwater inundation, effects of salt-laden wind, and development of coastal geomorphic
formations such as dunes.
48. 48
In their detailed description of vegetation communities throughout the island, Gagne and
Cuddihy (Gagne and Cuddihy 1990) classified vegetation into three climate zones based on
annual rainfall: Dry (<1,200 mm), Mesic (1,200-2,500 mm), and Wet(>2,500 mm) and into 5
physiognomic classes based on vegetation characteristics: Herblands, Grasslands, Mixed
communities, Shrublands, and Forest (Table 7). In addition to rainfall, Gagne and Cuddihy list
wind/wave exposure, substrate type, human disturbance, and the unique history of evolution and
introduction of species as factors shaping the composition of coastal forests in Hawaii today. It is
important to note that of the 25 vegetation types listed, 17 represent native species dominant
ecosystems. However, only two of the coastal forest ecosystem, Pandanus and Pritchardia
forests, are dominated by native species and both ecosystem types are very rare.
Table 7. Coastal vegetation communities by rainfall zone and physiognomic character as described by Gagne and
Cuddihy (1990). Each community is listed by dominant species or dominant species combination.
Physiognomic
class
Dry (<1200 mm) Mesic (1200-2500 mm) Wet (>2500 mm)
Herblands Nama
Sesuvium
Batis
Grasslands/
Sedgelands
Sporobolus
Eragrostis
Lepturus
Schoenoplectus/Bolbos
choenus/Cyperus
Mixed
Communities
Sida
Sida/Chloris
Shrublands Scaevola
Sida
Gossypium
Heliotropium
Santalum
Coastal cliff community
Chenopodium
Myoporum
Leucaena
Hibiscus
Pluchea
Forests Prosopis Pandanus
Pritchardia
Casuarina
Bruguiera/Rhizophora
(Mangroves)
49. 49
Methods
From November 2009 through August 2010 researchers from the Tropical Landscape and
Human Interaction Lab “the team” conducted transects along 41 sites at coastal areas on Kauai,
Oahu, and Hawaii (the Big Island) (Figure 19-22). Three separate methods were used to assess
vegetation depending upon site conditions. At 41 sites the variable area transect method was
used to a maximum of 50m from the start of woody vegetation. At three sites, the team recorded
only a list of species present and at two sites used 10x5m or 10x10 m plots.
Figure 19. Coastal vegetation assessments were carried out at 41 sites on three islands in Hawaii.
50. 50
Figure 20. The variable area transect method was used at all 17 sites on the Big Island Hawaii.
Figure 21. The variable area transect method was used at 13 sites on Kauai while a list of species only was taken at
Princeville
51. 51
Figure 22. The variable area transect was used at eight sites on Oahu. 10x5 and 10x10 m plots were used at the Boat
Harbor site while species lists only were taken at Pearl Harbor and Diamond Head.
Sites were selected based on accessibility; as the team was restricted to sites that could be
accessed by public roads or right of ways. In addition it was attempted to represent as many
coastal vegetation types as possible. A third criterion of selection attempted to represent a great
range of climate variability. Sampling was conducted from sites that represented each of the
recognized moisture zones present on the three islands (Price et al. 2007) (Table 8). In this
classification, zone 1 represents the most arid zone and zone 6 represents the wettest zone. The
sites represent a range of average annual rainfall from 244 mm at Puako (Big Island) to 3465 mm
of annual rainfall at Laupahoehoe (Daly and Halbleib 2006).
Table 8. Moisture zones for each assessment site by island. Moisture zones based on Price et al. (2007).
Island Moisture
Zone 1
Moisture
Zone 2
Moisture
Zone 3
Moisture
Zone 4
Moisture
Zone 5
Moisture
Zone 6
Total
Big
Island
2 2 3 4 1 5 17
Kauai 3 1 1 5 1 0 11
Oahu 3 3 3 4 0 0 13
Total 8 6 7 13 2 5 41
