Aquatic Botany 134 (2016) 10–17

Contents lists available at ScienceDirect

Aquatic Botany
j o ur na l h o me pa g e : w w w . e l s e v i e r . c o m / l o c a t e / aq u ab o t

Mangrove’s response to cyclone Eline (2000): What is happening 14 years later

a,b,∗ c d a b
C.C.F. Macamo , E. Massuanganhe , D.K. Nicolau , S.O. Bandeira , J.B. Adams
aDepartamento de Ciências Biológicas, Universidade Eduardo Mondlane, P.O. Box 257, Maputo, Mozambique b Nelson Mandela Metropolitan University, PO Box 77000, Port Elizabeth 6031, South
Africa, South Africa
c d
Departmento de Geologia, Universidade Eduardo Mondlane, PO Box 257, Maputo, Mozambique WWF
Mozambique, Keneth Kaunda Av. Maputo, Mozambique

article info abstract

Article history: This study assesses the impact of severe cyclones on a mangrove forest and the response over a 14 year period. SPOT images
Received 25 June 2015 from pre- and post-cyclone years were used to assess changes in area and in the Normalized Difference Vegetation Index
Received in revised form 4 May 2016 (NDVI) in a cyclone impacted mangrove forest in central Mozam-bique. Forest structure and condition were evaluated in the
Accepted 5 May 2016 Available online 13
field 11 years after the cyclone, sampling both the protected creek and exposed seaward mangroves. Transects perpendicular
June 2016
to the coast line were set across impact zones, covering high-, mid- and less-impacted zones. Quadrats (10 × 10 m) were set
along each transect, and adult trees within were identified to species, Diameter at Breast Height (DBH) mea-sured, height
Keywords:
estimated, and life condition (living or dead) indicated. Regeneration was assessed in 5 × 5 m quadrats. Cyclone Eline
Hurricane
impacted on 47.8% of the mangrove forest area, as shown by reduced NDVI in 2000 after the cyclone. Field sampling results
Mangrove resilience
Natural disaster indicate substantial recovery in creek forests; while sea-ward mangroves had 100% mortality in Rhizophora mucronata
East Africa dominated areas. This species is sensitive to defoliation and sedimentation and unable to sprout. Only mid- and less-impacted
Disturbance zones had high den-sities of juveniles. Species with high regenerating success were R. mucronata, Ceriops tagal and
Forest dynamic Avicennia marina. Sheltered creek mangroves were able to recover but changes in sedimentation prevented recov-ery in the
exposed seaward sites. This study highlights the need to understand the response of mangroves to cyclones, considering
increased frequency with climate changes. Higher frequency of cyclones might prevent recovery, threatening the forest and
associated coastal protection. However increased forest resilience can be gained with proper management.

© 2016 Elsevier B.V. All rights reserved.

1. Introduction
Globally, there are 7 main tropical cyclone areas, which are the North
Atlantic Ocean, Eastern and Western parts of the North Pacific, Southwestern
Pacific, Southwestern and Southeast-ern Indian Ocean, and the Northern
Indian Ocean (Gray, 1968). The most active areas are the Western North
Pacific with an average of 26 tropical storms each year and the North Atlantic
Basin, where on average 6 of the 10 annual tropical storms reach the
hurricane strength (Henderson-Sellers et al., 1998). In the southern hemi-
sphere, the Southeastern Indian Ocean is the most active area, with an average
of 9 cyclones per year (Henderson-Sellers et al., 1998).
In South Eastern Africa and Western Indian Ocean, the long coastlines of
Mozambique and Madagascar are amongst the most affected by cyclones.
Particularly to Mozambique, in the last 65 years its coast has been affected by
at least 20 cyclones (Mavume et al., 2014), three of which were very severe
and made land-fall in the south-central region between 2000 and 2007:
Tropical cyclone Eline in 2000, which was the longest-lasting (29 days) trop-
ical cyclone in the Indian Ocean (Reason and Keibel, 2004); Tropical cyclone
Japhet in 2003 and Tropical Cyclone Favio in 2007 (Mavume et al., 2014). All
cyclones were very damaging in terms of human lives and infrastructure.
Impacts on natural habitats are also likely to have occurred (Furze and Preble,
2003; Ohara and Falvey, 2007; Massuanganhe et al., 2015).

Understanding the impacts of cyclones on natural habitats, and the

∗ habitat’s response and recovery mechanisms is particularly important in the
Corresponding author at: Departamento de Ciências Biológicas, Universidade Eduardo Mondlane,
P.O. Box 257, Maputo, Mozambique. current scenario of climatic changes exacerbated by an increasing trend in the
E-mail addresses: celia.macamo@uem.mz, celiamacamo@yahoo.com (C.C.F. frequency and intensity of cyclones
Macamo).

http://dx.doi.org/10.1016/j.aquabot.2016.05.004 0304-
3770/© 2016 Elsevier B.V. All rights reserved.
 C.C.F. Macamo et al. / Aquatic Botany 134 (2016) 10–17
worldwide (Webster et al., 2005). Mangroves in particular are key ecosystems
in this matter, considering their role in coastal protection, carbon
sequestration and the high vulnerability and susceptibility to impacts during
such events (Alongi, 2008; Gilman et al., 2008).
Mangroves can be impacted in several ways, such as land-use changes,
conversion to bare soils, physical damages to plants, defo-liation and
uprooting, changes in population structure (including in regeneration),
sedimentation, erosion (Smith et al., 2009) and changes in soil characteristics
(Cahoon et al., 2003). The duration and intensity of such impacts will depend
on numerous factors (Sherman et al., 2001), including site-specific
biophysical char-acteristics, species composition and species ability to cope
with inflicted damages (Baldwin et al., 2001; Sherman et al., 2001). Healthy
resilient forests will be in a better condition to recover after a major
disturbance, and the least frequent the disturbance, the greater the chance for
a full recovery (Alongi, 2008).
Studies on the impact of cyclones (and associated floods) and climate
change on mangroves were documented in parts of the world, including
Australia and Micronesia, United States of Amer-ica, Nicaragua, India,
Bangladesh, Myanmar, and many more (Paling et al., 2008; Kauffman and
Cole, 2010; Aung et al., 2011). Most stud-ies rely on remote sense techniques
to asses land use changes and changes in forest area (Paling et al., 2008;
Bhowmik and Cabral, 2013), while others focus on physical damages caused
by winds and storm surges on the mangrove trees (Doyle et al., 1995;
Houston, 1999), changes in the forest structure and regeneration patterns
(Roth, 1992; Kauffman and Cole, 2010) and forest recovery after cyclone
impact with and without human intervention (Harun-or-Rashid et al., 2009;
Aung et al., 2011, 2013).
In Africa, and the Eastern African region in particular, studies on the
impacts of cyclones and climate change related events are uncommon, and
rarely isolate climate-related events from human interference (Bosire et al.,
2014). To our knowledge, documented studies of cyclone impact on
mangrove structural dynamics in the Western Indian Ocean are scarce, except
for the study that describes die back of Heritiera littoralis in Rufiji Delta
(Tanzania) after El Nino˜ related floods (Erftemeijer and Hamerlynck, 2005)
and the qualita-tive description of the impacts of the recurrent catastrophic
events (cyclones and floods) on human and natural systems in central
Mozambique (Massuanganhe et al., 2015).
This study assesses the condition of a mangrove forest in East-ern Africa
that has been impacted by three major cyclones and associated floods.
Furthermore, it focuses on how impacts and recovery might differ in different
mangrove locations setup in the same cyclone impacted estuary (from Save
River, a major one in eastern Africa). By combining field sampling and
remote sensing techniques (NDVI), the study also documents the forest
dynam-ics, recovery and condition after cyclone impact, highlighting the
regeneration process and change detection.
2. Material and methods
2.1. Description of study site
The study was implemented in Save River Delta, the fourth largest river in
south-central Mozambique, and one of the largest in Eastern Africa.
Biophysically its river mouth area falls within the muddy coast section, a low
lying area with numerous streams dis-charging into the Indian Ocean. The
area is suitable for the growth of well-established mangrove formations,
which extend to the Zam-bezi Delta further north. Two main villages were
built along this section of the river (Machanga and Nova Mambone Villages),
and according to the 2007 census (http://www.ine.gov.mz), the main
economic activities of the local coastal communities are agriculture,
cattle and fishing. The climate of the area is tropical humid (annual
precipitation 1500 mm), with two seasons. The wet season extends from
November to March (peak in January/February), with average temperatures of
◦
33 C; while the dry cooler season extends from April to September, with
◦
average temperature around 18 C. The wet season is also cyclone prone, and
since 2000 the area has been hit by three major cyclones: Eline in 2000 (wind
speed between 168 km/h and 212 km/h when crossed the study area); Japhet
in 2003 (wind speed of 157 km/h in the study area) and Favio in 2007 (wind
speed of 222 km/h in the study area) (Fuze and Preble, 2003; Reason and
Keibel, 2004; Ohara and Falvey, 2007). Cyclone Eline was the most
devastating to the mangroves in the study area due to the severity of the
cyclone when passing the Mozambican coast and proximity of the mangroves
to the landfall site. Key ecosys-tems represented in the area include seagrass
beds and mangrove forests, which grow in most river mouths covering the
adjacent districts of Govuro and Machanga (Massuanganhe et al., 2015).
2.2. Mapping
SPOT data were used to map and analyze the mangrove changes over the
time. The images were orthorectified and projected to UTM zone 36 S. For
the mapping process, the principles of remote sens-ing were used by
extracteing the Normalized Difference Vegetation Index (NDVI) for further
comparison over the time period in analy-sis. The SPOT images used in this
study were taken in 1999 (before the cyclone), 2000 (months after cyclone
Eline), 2007 (months after Cyclone Japhet), 2011(field sampling period) and
2014 (present time) covering a time span of 15 years. The SPOT images were
aligned one to another as some of them were the same scale as the others. This
registration was undertaken as a prerequisite for further comparison by
overlapping that data.
A geomorphological approach was used to separate mangrove wetland
areas from the upper deltaic plain, which included vege-tated dune ridges. The
wetland areas assessed were those where mangrove forest has high chances to
develop. From this stage, the NDVI was extracted from each image and
compared the changes to the subsequent image with the purpose of finding a
pattern that may be correlated with the cyclones that affected this region.
There-fore, each NDVI map was resample to a 10 m cell size for further
comparison and all data were exported to a table containing the attributes of
each cell. This allowed the extraction of cell values and to perform the
arithmetical operation to find the differences between the different years.
The comparisons between the different years were performed by
calculating the difference between the NDVI with the subse-quent year. With
this operation all the positive values would reflect areas of greenness where
the mangrove may have increased allow-ing in this way to record even smaller
changes that could not be accounted when using a visual approach.
2.3. Mangrove sampling
Sampling areas were identified based on local knowledge of impacted
sites after cyclone Eline in 2000 (the first and most devas-tating). The
sampling took place in 2010 and 2011, and at each time a different impacted
area was visited. Two sub-areas were identi-fied in the field: sub-area A with
creek protected mangroves and sub-area B with seaward exposed mangroves.
In both sub-areas, three blocks were identified: Block I for the mangroves
growing along the creeks or sea, highly impacted by the cyclone; Block II of
mid-section mangroves, with medium impacts; and Block III for the inner
mangrove, with few or no visible impacts of the cyclone (Fig. 1). A variable
number of 10 × 10 m quadrats were set in each sub area and blocks along the
transect, separated by at least 25 m. The number of quadrats varied with the
length of each Block. In
12 C.C.F. Macamo et al. / Aquatic Botany 134 (2016) 10–17

Fig. 1. Sampling profile showing creek (protected) and seaward (exposed) mangroves (above), and the linear transect going through Blocks I–III (below).

each quadrat, every tree was identified to species level and had DBH and Multivariate Analysis for non-parametric data, as provided by the
measured and height estimated with a graduated pole (Bandeira et al., 2009). statistical software StatSoft Statistica 12.
Tree condition (alive or dead) was also registered. Dead and cut trees were
identified based on characteristics such as tree physiognomy, type of roots,
3. Results
and bark and wood character-istics. These data were used to describe
structural characteristics of the forest such as mean diameter, mean height,
3.1. Change detection
stand density, and% of dead trees in each block. Other qualitative data were
also collected in the field, including: dominant species in the quadrat, soil
The results from this study show variability in mangrove cover in the
characteristics, tree health condition or other signs of stress, block inundation
study area over the last 15 years, including pre-cyclone (1999) and post-
class, and general visual description of the site. This information was useful
cyclone periods (2000–2014). The total area of the man-grove forest was
for further analysis of the results.
estimated as 14,500 ha in 2014. The changes from the map reflect the canopy
cover of the area as expressed by NDVI. In the series of the maps produced
The regeneration condition of the mangroves was also assessed in 5 × 5 m
(Fig. 2) major changes were identified between 1999 and 2000 (Fig. 2a and b).
sub- quadrats, considering as juveniles those individu-als with DBH below
From the dif-ferent cyclone events, major changes recorded in the map can be
2.5 cm and height below 3 m. Juveniles were identified to species and
associated with Cyclone Eline which took place in February/March 2000
classified as Regeneration Class I (RCI or seedlings) those with height until
months after the second image was taken. It was estimated that 6342 ha of
40 cm; Regeneration Class II (RCII or saplings) with height between 40.1 and
mangrove forest were damaged by cyclone Eline (NDVI reduced). Changes in
150 cm; and Regenera-tion Class III (RCIII or young plant) with height
NDVI were particularly visible along major channels and creeks in the forest,
between 150.1 and 300 cm (Kairo et al., 2002). These data was used to
as well as along the coast line at exposed sites (Fig. 2f). In these areas the
estimate the density of the regenerating community.
damage seemed to be permanent, while in other areas of the forest there was
recovery 14 years after the cyclone.

2.4. Statistical analysis

All data were subject to statistical analysis on Microsoft Excel 2007 and
StatSoft Statistica 12. Comparisons (mean DBH, mean height and stand
density) were made using t-test or one way ANOVA for parametric data; and
Mann-Whitney or Kruskal Wal-lis ANOVA were performed for non-
parametric data. Post-hoc tests were performed to identify homogenous
groups when statistically significant differences were found: Tukey-HSD for
parametric data
3.2. Changes in forest structure
Two main types of impact were identified on site, as a result of the 3
cyclones and associated floods. The Creek protected man-groves were
impacted by prolonged submergence periods due to the combined effect of
river flooding and heavy rains asso-ciated with the cyclone. The main
characteristics of these sites
 C.C.F. Macamo et al. / Aquatic Botany 134 (2016) 10–17
Fig. 2. Changes in mangrove coverage of the study area expressed in NDVI from a to e. Each image referring to a different year as seen in the figure: a – 1999; b – 2000; c – 2007; d – 2011; e – 2014.
The map f shows the difference In NDVI between 1999 and 2000 (after cyclone Eline).
were moderate tree mortality, canopy gaps and a great amount of woody
debris. On the other side, seaward exposed mangroves were subjected to
stronger wave and wind action. These areas were previously dominated by
Rhizophora mucronata as evidenced by the presence of dead trees. Visual
characteristics of this site were: heavy sedimentation of tree roots in some
areas, peat exposure on others and high tree mortality (see Supplementary
Figs. 1 and 2). Many trees were in advanced stages of decomposition (Stage 3
or 4, as described by Kauffman and Donato, 2012) and in many cases it was
visible that dead trees were cut by local communities. Dif-ferences between
these two types of impact were also visible from the percentage of live trees.
Fig. 3 shows that in the creek forest the percentage of living adult trees was
71% in Block I and 97% in Block III, while in the seaward forests, 100% of
the trees were dead in Block I, and living trees made 79% of total trees in
Block III.
Out of the total of 39 quadrats, corresponding to 1060 indi-viduals, six
species were identified: Avicennia marina, Bruguiera gymnorhiza, Ceriops
tagal, Rhizophora mucronata, Sonneratia alba and Xylocarpus granatum.
Rhizophora mucronata was largely domi-
nant in both creek and seaward forests, while B. gymnorhiza and S.
alba were less common, and occurred in creek only.
In general, the structure of the forest is markedly different, as Table 1
shows. In the creek forest, the number of species ranges between 4 and 6
across the Blocks. In Block I, A. marina, R. mucronata, S. alba and X.
granatum are more common. Bruguiera gymnnorhiza and C. tagal also occur
in Block II, while S. alba does not occur in Block III. At the seaward forest, R.
mucronata dominates Block I, although A. marina might occur sometimes.
The number of species rises to 4 in Blocks II and III, other species being C.
tagal and X. granatum.
Trees of the creek forest tend to grow wider than in the seaward forest,
where the vast majority of individuals had DBH less than 6 cm diameter (Fig.
3). Average DBH was 7.6 ± 0.26 cm compared to 5.11 ± 0.29 cm in the
seaward forest, differences being signifi-cant (Student’s t-test: t = 1.405, df =
1003, p < 0.001). The average mean height indicated that creek mangroves
tend to grow taller (3.8 ± 0.01 m) than seaward mangroves (3.4 ± 0.01 m),
differences
14 C.C.F. Macamo et al. / Aquatic Botany 134 (2016) 10–17

Fig. 3. Structural aspect of the creek and seaward forests. Above: Percentage of living trees at
creek (protected) and seaward (exposed) sites in the mangrove forests of the Save river delta
after being hit by cyclones and floods. Below: DBH size classes distribution of the trees on
protected and exposed sites. Frequency indicates how often each size class was repeated in
2
every 100 m . y axis converted to logarithmic format.

Table 1
Structural attributes of the mangrove forest in protected and exposed sites on the Save river Fig. 4. Above: Regeneration pattern of creek and seaward mangroves in the study site. y axis
estuary. Mean ± S.D. Code letters indicate homogenous groups at p < 0.05 following post-hoc presented in logarithmic format. Below: Percentage of regenerating species on creek and
tests Tukey HSD (for parametric data) and Multivariate analysis (for non-parametric data). seaward mangrove forest.

Block I Block II Block III F = 5.05, df = 2, p = 0.006). Similar heights were found when com-paring
Creek forest N of species 4 6 5 Blocks I vs. II; and Blocks I vs. III (Tukey HSD: p < 0.05).
a a
Density (ind/ha) 101 ± 63 289 ± 89 488 ± 111 In the seaward mangroves, there were no significant differences when
b,c b c
Mean height (m) 4.2 ± 2.7 3.7 ± 1.5 4.3 ± 1.5 comparing DBH in Blocks II and III (Man-Whitney U test: U = 1835.7, p =
Mean DBH 12.8 ± 0.9 8.6 ± 1.3 5.9 ± 1.3
2
Basal area (m /ha) 5.2 2.5 2.4
0.441, n1 = 178, n2 = 216), while Block III was sig-nificantly taller than Block
Complexity index 0.9 1.5 2.0 II (Mann-Whitney U test: U = 14762.5, n1 = 178, n2 = 216, p < 0.001).
Seaward forest N of species 2 4 4
Density (ind/0.1 ha) – 413 ± 68 817.7 ± 130 There were very clear differences when comparing seaward and creek
Mean height (m) – 3.2 ± 1.1 3.6 ± 0.8
d d sites with regards to regeneration success. All blocks of the creek forest had
DBH (cm) – 4.5 ± 1.6 5.6 ± 3.7
2
Basal area (m /ha) – 1.1 1.7 some regenerating plants, however densities var-ied, increasing from Block I
Complexity index – 0.58 1.99 to III; whereas in the seaward forests, Block I was stripped of seedlings and
a Similar densities on Block II and Block III Creek forest. saplings, Block II had some, and Block III had more, especially seedlings of
b c
Similar mean heights on Block I and Block II Creek Forest. Similar mean heights on
RCI. In both creek and seaward sites the trend was to have more individuals
Block I and Clock III Creek forest. of RCI and gradually less of the other two regeneration classes (Fig. 4).
d
Similar diameter at breast height in Block II and III Seaward forest.
Only 3 of the 6 species identified had enough seedlings, saplings and
being statistically significant (Student’s t-test: t = 5.346, df = 1003, p < juveniles to analyze regeneration condition. These species were A. marina, C.
0.001). tagal and R. mucronata. In creek forests regeneration was predominantly of A.
Significant differences in mean stand density existed between the three marina and R. mucronata, and R. mucronata was dominant for all regeneration
blocks on the creek forest (Kruskal-Wallis ANOVA: n = 25; h = 14.20; p = classes. In the seaward forests the species were R. mucronata and C. tagal,
0.008). Block I had the lowest stand density (Multi-variate Test: p < 0.05), again R. mucronata was dominant (Fig. 4).
while Blocks II and III had similar densities (Multivariate Test: p > 0.05). On
the seaward forest, the stand den-sity was also significantly different when
comparing Blocks II and 4. Discussion
III (Mann-Whitney U test: U = 3; n1 = n2 = 6, p = 0.02), and Block III had the
highest stand density. 4.1. Changes in NDVI
In the creek forest both height and DBH decreased towards inner parts of
the forest. The differences were statistically sig-nificant when comparing The impact of cyclones and hurricanes on mangroves has been shown in
DBH (one way ANOVA: F = 34.49, df = 2, p < 0.0001). Block I had the many places to cause massive tree mortality and ecosys-tem changes. In
widest trees (most of them being S. alba), followed by Block II and Block III Australia, for example, cyclone Vance caused a loss of 44% of mangrove
(Tukey-HSD: p < 0.05). Differ-ences in height were also statistically area, while in Micronesia 6–32% of man-grove area was lost 4 months after a
significant (one way ANOVA: major cyclone (Paling et al.,
 C.C.F. Macamo et al. / Aquatic Botany 134 (2016) 10–17
2008; Kauffman and Cole, 2010). Most changes are conversion of mangrove
to bare soils or saltmarsh area, and the process can be continuous for several
years after the cyclone (Paling et al., 2008). In the Save Delta, the overall
condition of the forest was affected, as indicated by visible changes in NDVI,
which reflected severe tree mortality as documented in the field. The affected
area in 2000 accounted for 47.8% of the total forest. A substantial recovery is
vis-ible after 2011, four years after the last cyclone, indicating that it is very
likely that cyclones Japhet and Favio slowed down the recovery process, and
possibly inflicted further damages to the forest.
While most areas showed good recovery some years after the cyclone
(increases in NDVI can be seen from 2007), the areas around the main creeks
and along the coast line had slower recovery or failed to recover at all, as also
found in the field. These areas had larger canopy gaps and little to no
recovery, reflected as areas with lower NDVI. A lack of recovery could be
attributed to two main reasons: (1) the cyclones and floods introduced major
changes in the forest structure and biophysical conditions, such as heavy sed-
imentation, changes in soil characteristics, or tree mortality was so high that
recovery was not possible; (2) impacts of cyclones and floods exacerbated the
natural processes of erosion and sedimen-tation in the area, which is common
to most deltas and estuaries (Geyer et al., 2001; Massuanganhe et al., 2015).
This area in par-ticular is very dynamic and this can be shown by the story of
the Nova Mambone village itself (its name meaning “New” Mambone). The
“old” Mambone village was abandoned in 1957 due to severe erosion and
flooding and a new village was built in the current location.
Examples of permanently changed habitats after a cyclone or floods have
been documented in other impacted mangrove forests across the globe (Smith
et al., 2009), however, unless these pro-cesses are verified, a quick natural
recovery of an impacted forest is always expected, including a complete
recovery of pre-cyclone mangrove area (Paling et al., 2008). The mangroves
of the Save Delta have not fully recovered as shown by the maps.
4.2. Changes in forest structure
In the Save River delta mangrove forest, there were two distinct patterns
of impact. The creek forests seemed to have suffered from mechanic stressors
(wind and waves), and from a prolonged sub-mergence period as a result of
heavy rains related to the cyclone (Jiménez et al., 1985). In this particular
case, flooding must also have been exacerbated by the fact that the cyclone
occurred dur-ing the rainy season, and a major flood (frequent in this season)
was already eminent. Prolonged submergence can reduce oxygen intake by
the roots, stress plants and cause massive death (Jiménez et al., 1985). The
outer seaward parts of the forest seemed to have suffered to a greater extent as
these areas were exposed to the sea, and thus more vulnerable to storm surge
and winds. These sites were also more vulnerable to sedimentation as sandy
soils were brought in and deposited from nearby sandy areas. Overall in the
forest, seaward and creek mangroves were most affected and this was also
described by other authors (Swiadek, 1997).
Mortality in the Save estuary mangrove forest might be explained by 3
main factors: (1) natural mortality not related to either cyclones or floods; (2)
mortality right after the cyclone due to physical damages on trees, defoliation
and prolonged submer-gence; and (3) mortality after the cyclone (months or
years) due to changes in biophysical characteristics of the site, such as sedi-
mentation and erosion, changes in soil properties and changes in
geomorphology. In practice it is difficult to precisely distinguish the
contribution of each factor, but it was clear during field visits that massive
mortality occurred only on creek and seaward man-groves, while inner forests
were kept almost intact, in a pattern
that is compatible with transformations caused by other cyclones in other
mangrove forests across the globe (Swiadek, 1997).
The level of transformation which occurred particularly in Block I can be
assessed by comparing structural attributes of the 3 Blocks, assuming that
Block III is the closest representative of the for-est structure before the
impacts. This is particularly true for stand density, basal area and density of
seedlings, saplings and juveniles.
The massive death of adult trees can impair a normal regenerat-ing
process in many ways, such as shortage of propagules, opening of large
canopy gaps and increased vulnerability of seedlings to pre-dation attacks
(Duke, 2001; Walters, 2005). Also in Block I (seaward forest), visual analyses
of the soil indicated profound changes such as heavy sedimentation and peat
exposure, which do not favor a successful settlement and growth of new
plants (Ellison, 1999).
Rhizophora mucronata is present across the forest and strongly dominates
the outer seaward sites. This species is more sensitive to the impacts of
cyclones, by for example, showing higher rates of defoliation and a lower
ability to refoliate the canopy. In Micronesia, for example, 4 months after a
category 4 cyclone hit an area of mangrove forests, the places with the higher
tree mortality were dominated by R. mucronata. Surviving individuals of this
species showed very high levels of defoliation and low canopy refoliation and
sprout (Kauffman and Cole, 2010), while species like S. alba and A. marina
were more resilient, as the canopy refoliated quicker and the trees sprouted
(Woodroffe and Grime, 1999; Kauffman and Cole, 2010). Seaward Block I
was also composed of 2 species only, what might also have influenced its
ability to withstand the cyclone (Sandilyan and Kathiresan, 2015).
These two factors (exposure of seaward sites and dominance by a sensitive
species) can help understand the 100% mortality on Block I, and higher
percentage of live trees of Blocks II and III.
Cyclone Eline was the longest lasting cyclone in the eastern Ocean
Region, and the most devastating in Mozambique in terms of human losses
and infrastructure damages (Reason and Keibel, 2004). Eline was also
mentioned by the locals as the most damaging to the mangroves and other
natural habitats, such as seagrass beds (Community leaders, personal
communication). Hence, it is possi-ble that much of the transformation of the
forest can be attributed to this first cyclone, while subsequent cyclones Favio
and Japhet might have exacerbated such impacts. Determining in the field
how much of the transformation in the forests can be attributed to which
cyclone or flood proved to be difficult, thus we attribute the changes to the 3
events.
Frequent cyclones will have a strong influence on mangrove for-est
structure, changing diameter size-class distributions, species composition, and
zonation patterns. Frequently impacted forests may experience changes in
species relative abundance and density, and tend to have trees with smaller
diameter and shorter canopy (Smith et al., 2009). In the Sundabarns forest the
absence of Rhi-zophoraceae species was attributed to the fact that the area is
on average impacted by 30–40 cyclones a year (Smith and Duke, 1987; Smith
et al., 2009). Collected data on the Save mangroves might help understand
how the process happens, especially when one com-pares the structural data
between the 3 blocks at both creek and seaward sites. Stand density clearly
increases from Block I to III in both cases, and it would be expected that the
highest trees would be found in the sheltered river channel sites (since at the
exposed sites they were more vulnerable to breakage). The dominance of
Rhizophoraceae juveniles at the seaward forests is probably due to the
dominance of this species in the whole forest, but it is impor-tant to note the
absence of seedlings and saplings at very impacted sites (especially Block I).
An additional factor to consider for forest recovery is the pre-cyclone
conservation condition of the forest. The communities at the Nova Mambone
and Machanga villages are highly depen-dent on coastal resources for their
survival, and fish and shrimp
16 C.C.F. Macamo et al. / Aquatic Botany 134 (2016) 10–17

Table 2
Major extreme events that affected the region (Inhambane province and southern Sofala Province) in the last 50 years.

Extreme event Year Category (at landfall) Area of impact in Major impacts
the region
Cyclone Claude 1966 Tropical depression Inhambane Bay –
Cyclone 1972 Tropical storm Vilanculos, –
Caroline Inhambane Bay
Cyclone Danae 1976 Intense Tropical cyclone Vilanculos 50 people killed
Cyclone 1984 Tropical storm Inhambane Bay Destroyed over 50 small dams,
Domoina widespread crop damage
Cyclone Eline 2000 Intense Tropical cyclone Nova Mambone, 700 people killed;
Machanga, infrastructure destroyed
Vilanculos, Beira
(further north).
Cyclone Japhet 2003 Tropical cyclone Vilanculos, Nova 2 people killed; infrastructure
Mambone, destroyed
Machanga
Cyclone Favio 2007 Tropical cyclone Vilanculos, Nova 4 people killed (93.000
Mambone, affected); infrastructure
Machanga. destroyed
Floods Annual Severe in: 2000,2003, Nova Mambone, Crops destroyed, infrastructure
2008,2009, 2011 Machanga (roads, bridges, etc.)
Source: Furze and Preble (2003); Ohara and Falvey (2007); Mavume et al. (2014).

fishing are major sources of income. Mangroves are also an impor-tant source
of woody resources, but harvesting seems to be mostly for domestic
consumption, and not for commercial purposes. Also the community has
relatively limited access to most of the areas, giving the low number of
powerful motorboats. This has con-tributed to the better condition of the
forests, and ultimately to its resilience and ability to recover after the
cyclones. However some severely impacted areas were not able to recover as
shown in both satellite images (where the NDVI did not increase until 2014,
depicting poor forest condition) and forest structural data. Human
consumption of mangrove firewood has reportedly increased after the
cyclone, for people would go to the forest and collect dead dried poles. The
possible impact of such collection on the recovery of the forest has not been
assessed.
4.3. Links to climate change
Historical records show that this area is particularly prone to cyclones and
cyclic floods (Table 2). Climate change predictions for the country indicate an
increase in the frequency and intensity of cyclones in this area, and changes in
rainfall patterns, reducing the extent of the rainy season, but increasing the
number of heavy rain-fall events (Menomussanga and Matavel, 2011). This
means that the probability and severity of floods will also increase, and a sub-
sequent dry period will be more prolonged. These changes will impact the
mangrove forest. If cyclones become more frequent and more severe, the
forest might not have enough time to recover from damages, and gradual
accumulation of impacts will cause profound changes in the forest structure
(Gilman et al., 2008; Smith et al., 2009). One expected change would be a
shift in species composi-tion, R. mucronata becoming far less frequent and
non-dominant in the forest. Mean height of the forest is about 3–4 m, which is
over-all not considered a tall forest (Kairo et al., 2002; Bandeira et al., 2009),
so very pronounced changes in canopy height might not occur. However,
extensive sedimentation, erosion of river channels and massive tree mortality
can end up reducing the total forest area.
There were 3 major cyclones between 2000 and 2007, which is a
relatively short period, when compared to the frequency of cyclones before
2000 (Table 2). It is impossible to attribute these events directly to climate
change; however they can give an insight of what is expected in a scenario
where cyclones become this fre-quent. Mangroves play an important role in
coastal protection, and their efficiency for this function during extreme
climatic events, storm surges and tsunamis has already been documented (Das
and
Vincent, 2009; Harun-or-Rashid et al., 2009; Smith et al., 2009; Kamthonkiat
et al., 2011; Sandylian and Kathiresan, 2015). Key aspects found to be
determinant on how well this service is deliv-ered include width of the forest
belt, forest density, structural complexity, species composition and dominant
species, height and diameter of trees, size and complexity of the roots
systems, and other biophysical aspects, such as elevation of the habitat, nature
of the channels and conservation condition of the forest (Das and Vincent,
2009; Sandilyan and Kathiresan, 2015).
The Nova Mambone Village was built right behind the mangrove forest,
and certainly this was a protecting factor during the cyclone (Fig. 2). The fact
was also acknowledged by many villagers (Local leaders and fisherman,
personal communication), and proved by the profile of the impact on the
forest, where one can clearly see the damage to the forest decreasing towards
inside the forest from the coast. The fact that this forest was not exposed to
very high human pressures was certainly a positive aspect that helped both on
the protection role, and on its recovery. This is however a situation that might
change with an increase in the number of mangrove collec-tors, from the
village as well as from nearby areas. After the cyclone many fisherman
complained that fish catches had decreased con-siderably, mostly due to the
sedimentation of a nearby sea grass bed, which never managed to recover.
This led to many fishers turning to mangrove logging as a secondary activity.
Intensive exploitation of forest resources can compromise the resilience of the
forest, and slow down recovery. In that case, and consider-ing the climate
change scenarios, both the forest and village might be endangered. We
recommend to design and implement a sus-tainable management plan for the
forest, and monitor its recovery after any catastrophic event, in order to
intervene if necessary by replantation or opening of closed channels.
The forest and community can benefit from simple management measures,
such as determining a minimum size for a tree to be cut; identifying no-cut
zones, specifically on seaward, creek forests and other vulnerable areas;
implementing rotating cutting in identi-fied areas and restoring degraded
areas. Current community based organizations (such as the Fishing
Community Council) could also increase community participation in
designing and implementing such measures.
5. Conclusion
Frequent and intense cyclones can change the structure of a mangrove
forest, and sea-exposed forests are more vulnerable then
 C.C.F. Macamo et al. / Aquatic Botany 134 (2016) 10–17
creek and inner forests. The forest condition plays an important role in the
ability of impacted forests to recover after such disturbance. In the current
scenario of climate change, forest resilience should be increased by expanding
the seaward mangrove belt, thus provid-ing protection to both inner mangrove
forest and possible human infrastructure that might occur.
Acknowledgements
We are grateful to WIOMSA for its financial support of the project
Resilience and Adaptation of Mangroves and Dependent Communities in The
WIO Region to the Impacts of Climate Change, and to our project partners of
Kenya Marine and Fisheries Research Institute. Logistical support in the field
was provided by UEM-Dept of Biological Sciences and by the Local
Administration and maritime authorities.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in the online
version, at http://dx.doi.org/10.1016/j.aquabot.2016.05. 004.
References
Alongi, D.M., 2008. Mangrove forests: resilience, protection from tsunamis, and responses
to global climate change. Estuar. Coast. Shelf Sci. 76, 1–13.
Aung, T.T., Than, M.M., Katsuhiro, O., Moshida, Y., 2011. Assessing the status of three
mangrove species restored by the local community in the cyclone-affected area of the
Ayeyarwady Delta, Myanmar. Wetlands Ecol. Manage. 19, 195–208.
Aung, T.T., Moshida, Y., Than, M.M., 2013. Prediction of recovery of cyclone-disturbed
mangroves in mega delta Myanmar. For. Ecol. Manage. 293, 103–113.
Baldwin, A., Egnotovich, M., Ford, M., Platt, W., 2001. Regeneration in fringe mangrove
forests damaged by Hurricane Andrew. Plant Ecol. 157, 149–162.
Bandeira, S.O., Macamo, C.C.F., Kairo, J.G., Amade, F., Jidawi, N., Paula, J., 2009. Evaluation
of mangrove structure and condition in two trans-boundary areas in the Western Indian
Ocean. Aquat. Conserv. Mar. Freshwater Ecosyst. 19, 46–55.
Bhowmik, A.K., Cabral, P., 2013. Cyclone Sidr impacts on the Sundarbans floristic diversity.
Earth Sci. Res. 2, 62–79.
Bosire, J.O., Kaino, J.J., Olagoke, O., Mwihaki, L.M., Ogendi, G.M., Kairo, J.G., Berger, U.,
Macharia, D., 2014. Mangroves in peril: unprecedented degradation rates of peri-urban
mangroves in Kenya. Biogeoscience 11, 2623–2634.
Cahoon, D., Hensel, P., Bczyk, J., McKee, K., Proffitt, C., Peres, B., 2003. Mass tree
mortality leads to mangrove peat collapse at Bay Islands, Honduras after Hurricane
Mitch. J. Ecol. 91, 1093–1105.
Das, S., Vincent, J.R., 2009. Mangroves protected villages and reduced death toll during
Indian super cyclone. Proc. Natl. Acad. Sci. U. S. A. 106, 7357–7360.
Doyle, T.W., Smith III, T.J., Robblee, M., 1995. Wind damage effects of Hurricane Andrew
on mangrove communities along the southwest coast of Florida, USA. J. Coast. Res. 21,
159–168.
Duke, N., 2001. Gap creation and regeneration processes driving density and structure of
mangroves ecosystems. Wetlands Ecol. Manage. 9, 257–269.
Ellison, J., 1999. Impacts of sediment burial on mangroves. Mar. Pollut. Bull. 37, 420–426.
Erftemeijer, P.L.A., Hamerlynck, O., 2005. Die-back of the mangrove Heritiera littoralis
dryand, in the Rufiji delta (Tanzania) following el Nino˜ floods. J. Coast. Res. 42, 228–
235.
Furze, P., Preble, A., 2003. Annual Tropical Cyclone Report. Annual Tropical Cyclone Report.
U.S. Naval Maritime Forecast Center/Joint Typhoon Warning Center Pearl Harbor, Hawaii,
USA.
Geyer, R., WoodaUff, J., Taaykovski, P., 2001. Sediment transport and trapping in the
Hudson River Estuary. Estuaries 24, 670–679.
Gilman, E., Ellison, J., Duke, N., Field, C., 2008. Threats to mangroves from climate change
and adaptation options: a review. Aquat. Bot. 89, 237–250.
Gray, W.M., 1968. Global view of the origin of tropical disturbances and storms.
Mon. Weather Rev. 96, 669–700.
Harun-or-Rashid, S., Biswas, S., Böcker, R., Kruse, M., 2009. Mangrove community recovery
potential after catastrophic disturbances in Bangladesh. For. Ecol. Manage. 257, 923–
930.
Henderson-Sellers Zhang, H., Berz, G., Emanuel, K., Gray, W., Landsea, C., Holland, G.,
Lighthill, J., Shieh, S.-L., Webster, P., McGuffie, K., 1998. Tropical cyclones and global
climate change: a post-IPCC assessment. Bull. Am. Meteorol. Soc. 79, 19–38.
Houston, W.A., 1999. Severe hail damage to mangroves at Port Curtis, Australia.
Mangroves Saltmarshes 3, 29–40.
Jiménez, J., Lugo, A., Cintrón, G., 1985. Tree mortality in mangrove forests.
Biotropica 17, 77–85.
Kairo, J., Dahdouh-Guebas, F., Gwada, P., Ochieng, C., Koedam, N., 2002. Regeneration status
of mangrove forests in Mida Creek, Kenya: a compromised or secured future? Ambio 31,
562–568.
Kamthonkiat, D., Rodfai, C., Saiwanrungkul, A., Koshimura, S., Matsuoka, M., 2011.
Geoinformatics in mangrove monitoring: damage and recovery after the 2004 Indian
Ocean tsunami in Phang Nga, Thailand. Nat. Hazards Earth Syst. Sci. 11, 1851–1862.
Kauffman, B., Cole, T., 2010. Micronesian mangrove forest structure and tree responses
to a severe typhoon. Wetlands 30, 1077–1084.
Kauffman, J.B., Donato, D.C., 2012. Protocols for the Measurement, Monitoring and Reporting
of Structure, Biomass and Carbon Stocks in Mangrove Forests Working Paper 86. CIFOR,
Bogor, Indonesia.
Massuanganhe, E.A., Macamo, C., Westerberg, L.-O., Bandeira, S., Mavume, A., Ribeiro, E.,
2015. Deltaic coasts under climate-related catastrophic events-insights from the Save River
Delta, Mozambique. Ocean Coast. Manage. 116, 331–340.
Mavume, A., Pinto, I., Massuanganhe, E., 2014. Potential climate change impacts on Maputo
Bay. In: Bandeira, S., Paula, J. (Eds.), The Maputo Bay Ecosystem. WIOMSA, Zanzibar
Town, pp. 383–397.
Menomussanga, M., Matavel, A., (2011). Catalogue of climate change and natural resource
management in Govuro District (Nova Mambone Administrative Post). DCS-Zonas
Costeiras, Maputo.
Ohara, J.F., Falvey, R.J., (2007). Annual Tropical Cyclone Report. U.S. Naval Maritime
Forecast Center/Joint Typhoon Warning Center Pearl Harbor, Hawaii. USA.
Paling, E.I., Kobryn, H.T., Humphreys, G., 2008. Assessing the extent of mangrove change
caused by cyclone Vance in the Eastern Exmouth Gulf, northwestern Australia. Estuar.
Coast. Shelf Sci. 77, 603–613.
Reason, C.J.C., Keibel, A., 2004. Tropical cyclone eline and its unusual penetration and
impacts over the Southern African mainland. Weather Forecasting 19, 789–805.
Roth, L.C., 1992. Hurricanes and mangrove regeneration: effects if hurricane joan
october 1988, on the vegetation of Isla Del Venado, Bluefields, Nicaragua.
Biotropica 24, 375–384.
Sandilyan, S., Kathiresan, K., 2015. Mangroves as bioshield: an undisputable fact.
Ocean Coast. Manage. 103, 94–96.
Sherman, R.E., Fahey, T.J., Martinez, P., 2001. Hurricane impacts on a mangrove
forest in the Dominican Republic: damage patterns and early recovery.
Biotropica 33, 393–408.
Smith III, T.J., Duke, N.C., 1987. Physical determinants of inter-estuary variation in mangrove
species richness around the tropical coastline of Australia. J. Biogeogr. 14, 9–19.
Smith III, T.J., Anderson, G., Balentine, K., Tiling, G., Ward, G., Whelan, K., 2009.
Cumulative impacts of hurricane on Florida mangrove ecosystems: sediment deposition,
storm surges and vegetation. Wetlands 29, 24–34.
Swiadek, J.W., 1997. The impacts of hurricane Andrew on mangrove coasts of Southern
Florida: a review. J. Coast. Res. 13, 242–245.
Walters, B., 2005. Ecological effects of small-scale cutting of Philippine mangrove forests.
For. Ecol. Manage. 1–3, 331–348.
Webster, P.J., Holland, G.J., Curry, J.A., Chang, H.R., 2005. Changes in tropical Cyclone
Number, duration, and intensity in a warming environment. Science 309, 1844–1847.
Woodroffe, C., Grime, D., 1999. Storm impact and evolution of a mangrove-fringed chenier
plain Shoal Bay, Darwin, Australia. Mar. Geol. 159, 303–321.