6.

Cover change and regeneration

6. Cover changes and regeneration status of a peri-urban

mangrove

R. mucronata sapling covered with silt

Publication
Mohamed, M. O. S., Neukermans, G., Kairo, J. G., Dahdouh-Guebas, F., and Koedam, N.
2008. A glance at the cover changes and regeneration status of a peri-urban mangrove.
Submitted to Forest Ecology and Management.

102
 6. Cover change and regeneration

Summary

Stability of an ecosystem is determined by its resilience, regenerative capacity and

numerous weak trophic links, amongst other natural and human induced factors. The
Tudor creek mangroves, a typical peri-urban mangrove, are exposed to both episodic
natural and recurrent human disturbances, including decades‟ long exposure to raw
domestic sewage, sporadic unregulated-harvesting and episodic siltation. This study
evaluates the regeneration patterns within extended gaps and the understorey. An
evaluation on species mix and regeneration patterns is also done. Preliminary analysis
of aerial photographs (1969 and 1992) and a satellite image (2005) indicate a 12.5%
decline in closed canopy mangrove between 1969 and 1992, and a 55% decline
between 1992 and 2005. Distribution of adult trees was variable, with mixed stands
and large canopy openings in the mid intertidal range. Species composition of
seedlings and saplings did not always reflect the overstorey species composition and
varied with gap sizes. Gap sizes range between 10 - 50m² have higher or mostly
adequate regeneration, while gaps smaller than 10m² and bigger than 60m² have lower
regeneration levels. R. mucronata seedlings and saplings occurred in the understorey
under all cover types and inundation regime, conferring advantages to this species
under the current disturbance regime. This may favour its establishment in relation to
other species. A. marina and C. tagal saplings and seedlings are restricted to the forest
edges and gaps. The current status of the forest is reminiscent of a recovery phase, a
multiphase succession stage, after a major disturbance event, accompanied by
recurrent anthropogenic pressure. This study shows that species composition depends
in part on the balance between natural large-scale and recurrent small-scale human
disturbances.

Keywords: mangrove, canopy, gap, regeneration, disturbance

103
 6. Cover change and regeneration

6.1 Introduction
The persistent undervaluing and apparent importance of mangroves to humans
(Dahdouh-Guebas et al., 2000; Walters, 2000; Rönnbäck et al., 2003; Walters, 2003;
Bosire et al., 2004; Dahdouh-Guebas et al., 2004; Walters, 2004; Rönnbäck et al.,
2007), wildlife (Nagelkerken et al., 2008), and the global carbon balance (Kristiansen
et al., 2008), have been adequately elaborated, and the call for realisation of the
extrinsic and intrinsic values of mangroves seems more urgent than ever as human
activities continue to threaten the mangrove ecosystems at both local and global scales
(Ewel, 2001; Duke et al., 2007; Nagelkerken et al., 2008), with 35% of the world‟s
mangrove forests lost over the last two decades (Valiela et al., 2001), making the
prospect of a world deprived of the services offered by mangrove ecosystems, perhaps
within the next 100 years, real (Duke et al., 2007). Human induced impacts coupled
with a changing global climate calls for an evaluation and understanding of the
regenerative characteristics of the mangrove ecosystems. This will facilitate the
appreciation of future implications and devise appropriate interventions (Robertson et
al., 1991; Kathiresan and Bingham, 2001).

With the high population growth particularly in tropical coastal areas, coupled with
rapid urbanisation, the peri-urban context of mangroves, seen as exceptional
currently, is fast becoming a recurrent feature for mangroves as observed in
Singapore, Hong Kong, Philippines, Tanzania (Daressalam), Mozambique (Maputo),
India (Calcutta, Goa), Colombia, Mexico etc….(Lacerda et al., 1993; Kathiresan and
Rajendran, 2005b; Tam et al., 1997; Holguin et al., 2001; Holguin et al., 2006;
PUMPSEA, 2007a, b; Walters et al., 2008). The cumulative effects of disturbances on
mangrove range from reduced adult tree density, reduced canopy height, increased
canopy gaps, to altered edaphic conditions (Walters, 2005a, b; López-Hoffman et al.,
2006; Alongi and de Carvalho, 2008). Disturbed stands are characterised by few
species of widely dispersed, dwarf like trees, with a distinctly bushy appearance.
Disturbances that cause nutrients loading, eutrophication or domestic sewage
pollution are widely reported to have no apparent negative effects (Wong et al., 1997;
Vaiphasa et al., 2007), and impact positively on tree growth (Feller et al., 2003b;
Boonsong et al., 2003), but may also enhance herbivory (Feller et al., 1995). Siltation
on the other hand, necessary to some extent (Victor et al., 2003; Kitheka et al., 2003),

104
 6. Cover change and regeneration

has potentially destructive effects by causing tree mortalities (Ellison, 1998;

Thampanya et al., 2002), while over-harvesting or clear-felling can potentially lead to
species shifts, lowering the intrinsic values (Kairo et al., 2002a, b; Dahdouh-Guebas
et al., 2004) and may lead to stand collapse. However, the apparent confounding
effects of these disturbances remain largely unknown and speculative, with no
specific studies. Scenarios of such combinations may also be rare or have avoided
attention.

The peri-urban mangroves of Tudor creek, Mombasa, have variously been described
as structurally stressed (Chapter 3; Mohamed et al., 2008), grossly over-exploited,
undervalued and poorly managed (Chapter 4; Mohamed et al., under review). They
present a typical forest impacted by multiple disturbances at variable temporal and
spatial scales, with a notable resilience to disturbances, maintaining high productivity
(Chapter 5). However, these disturbances can induce changes in forest structure,
composition and micro-climate, including tidal regime, seed/propagule availability
and dispersal, establishment and survival, in addition to lowering the health and
vigour of both seedlings and adult trees and mortalities through suffocation of
breathing roots (Ellison, 1998), significantly altering habitat conditions for
recruitment and establishment of seedlings (Bosire et al., 2003; Walters, 2005a, b;
Alongi and de Carvalho, 2008).

In this study, we assess the changes in mangrove canopy cover over time and the
regeneration patterns in relation to the current canopy cover and species mix. Aspects
of regeneration assessed include the distribution patterns of seedlings and saplings in
relation to adult trees, considering the understorey and gaps. The survival and growth
performance of replanted seedlings was also done at Mikindani in Tudor creek,
considering single and mixed species plots. The outcome of the study will give
important information on the recovery potential of the forest and the need for
restoration. This will guide on the development of management strategies to ensure
ecological sustainability. This study will thus provide useful insights into the
resilience of the mangrove forests of Tudor creek.

105
 6. Cover change and regeneration

6.2 Materials and methods

6.2.1 Study area
Tudor creek (Figure 6.1) bounds Mombasa Island on the northwest and extends some
10 km inland. The creek has two main seasonal rivers, Kombeni and Tsalu, draining
an area of 550 km² (450 and 100 km² respectively) with average freshwater discharge
estimated at 0.9 m³s-1 during the inter-monsoon long rains (cited in Nguli, 2006). It
has a single narrow sinuous inlet with a mean depth of 20 m, that broadens out further
inland to a central relatively shallow basin (5m) fringed by a well developed
mangrove forest mainly composed of R. mucronata, A. marina and S. alba. The basin
has an area of 6.37 km² at low water spring and 22.35 km² at high water spring.
Mangrove forests occupy 8 km² of the creek. Detailed structural attributes of the
Tudor creek mangroves are described by Mohamed et al. (2008; Chapter 3).

The Tudor creek mangrove system has been exposed to raw sewage intensively for
more than a decade. The sewage runs through the mangrove forest in canals and is
discharged into the Tudor creek waters mainly from Mikindani, Tudor and the Old
Town settlements. The mangroves are periodically dozed with sewage every tidal
cycle, with the loading exponentially reducing with distance from source (PUMPSEA,
2007a). It is estimated that about 1,200 kg of nitrogen and 5.5 kg of phosphorus are
discharged via sewage into the Mikindani system every day. The Mikindani Township
with about 67,164 people is located in the west mainland. Sewage from this estate was
according to plans intended to be pumped for processing at the Kipevu treatment
plant. But due to the overloading at the Kipevu plant, sewage from the township is
now discharged directly into the nearby mangrove forest at Mikindani in Tudor creek.
Sediments of Tudor creek are predominantly mud and some parts are covered with
sand. The land surrounding the creek beyond Mombasa Island is mainly agricultural,
largely small-holdings and coconut plantations with rough grazing land further inland,
while the immediate slopes bordering the mangrove creek are being intensively
cleared of vegetation to create space for informal settlements and subsistence farming.

106
 6. Cover change and regeneration

6.2.2 Canopy change detection

Mapping was done using panchromatic aerial photographs of 1969 and 1992
respectively. It was envisaged that the multi-date data sets will generate sufficient
information necessary for detecting changes that have occurred in Tudor creek
mangrove forests over the years. The approach was also meant to produce data that
could also inform management on status of the forest. The method used was a multi-
stage approach starting with; processing and printing of aerial photographs of 1969
(Survey of Kenya) and 1992 (DRSRS) at scale 1:25000; stereoscopic examination;
photo interpretation; and maps generation showing the entire Tudor creek with
distinct areas under mangrove vegetation and the surrounding land cover classes
separately delineated. The map was then digitized and geo-referenced using
ARCVIEW 3.1 software. Comparisons in cover classes were made with satellite
images (IKONOS, 2005) analysed by ERDAS under the PUMPSEA project
(Neukermans, Vrije Universiteit Brussel, unpublished data).

6.2.3 Forest assemblage and regeneration sampling

A stratified sampling technique was used to sample mangroves of Tudor creek. The
location of transect lines were determined by an initial reconnaissance and
examination of medium-scale (1:25,000) panchromatic aerial photographs of the area.
To sample small gap sizes, belt transects of 10 m width were established both
perpendicular and parallel to the creeks across the entire forest in such a way that they
represented optimally the general mangrove formation of Tudor creek (Figure 6.1).
Vegetation sampling was carried out within 10 by 10 m2 quadrats, established along
transects. A total of 196 plots were sampled under dense mangroves, and a total of 41
additional 400m² randomly selected plots were sampled under extended or enlarged
canopy gaps.

107
 6. Cover change and regeneration

Figure 6.1: Map of Mombasa showing the study area (Tudor creek) in relation to the
other creeks Notice the proximity of settlements to the mangroves and the differences in
density between the urban Kombeni and Rural Tsalu. The aerial photographs are to
scale 1:25,000 km (Source, DRSRS aerial photographs 1992).

In each sampling plot, adult trees species were identified and their height and the D130
were measured. This was used to identify species mix or zone within the forest.
Information on the composition and pattern of natural regeneration was obtained
using the method of Linear Regeneration Sampling (Sukardjo, 1987). In 5 x 5 m2
subplots (within the main 10 x 10 m2 quadrats), occurrence of seedlings and saplings
of different species was recorded in each plot. Seedlings were identified as below 100
cm in height, and mostly included new recruits normally termed as potential or/and
established regeneration, while saplings were young trees, with heights above 100 cm,
with D130 below 2.5 cm. Careful evaluation was done before assigning classes based
on site conditions.

108
 6. Cover change and regeneration

The analysis of spatial pattern of adults and juveniles in the field was carried out
inside 10 x 10 m2 plots along transects. The measure of dispersion used was
Morisita‟s dispersion Index (Morisita, 1959), the application of which is described in
Greig-Smith (1983). Morisita‟s Index (Iδ) is:
q
n(n  1)
I  q
i 1 N ( N  1)
Where, q is the number of quadrats, ni is the number of individuals per species in the
ith plot, and N is the total number of individuals in all q quadrats. If Iδ > 1, the
population is clustered, if Iδ = 1, the population is randomly dispersed and if Iδ < 1,
the population is evenly dispersed.

Inundation classes for each plot was estimated according to Watson (1928) and
classified as: (1) inundated by all high tides, (2) inundated by medium high tides, (3)
inundated by normal high tides, (4) inundated by spring tides, and (5) occasionally
inundated by exceptional or equinoctial tides.

6.2.4 Physicochemical parameters

Sediment cores were sampled randomly in twenty plots along four transects laid from
the landward to the seaward fringe, two transects in sewage receiving and two in non-
sewage receiving sites. In each plot, three sediment cores of diameter 4 cm and 15 cm
depth were taken by hand using corers. Laboratory analysis included porewater
analysis extracted with KCl. The sediment cores were sectioned at 0-5 cm and 9-11
cm. The pore water was extracted with KCl under centrifugation (3000 x g). The
resulting pore water was then filtered through prewashed 0.45 µ Whatman GF/F
microfiber filter and analysed for ammonium (NH4+), nitrate and nitrite (NOx-) and
phosphates (PO43-). The pore water collecting in the holes after sediment extraction
was used to measure salinity with a hand held refractometre. For subsequent nutrient
analysis after extraction, the conventional seawater analysis method by Parsons et al.
(1984) was used. For estimation of organic carbon (%TOM), a pre-weighed sampled
was dried to a constant weight at 70°C and then combusted at 450°C for 4 hours
(ashing). The weight loss was then estimated as he %TOM.

109
 6. Cover change and regeneration

6.2.5 Growth
Three common species of mangroves were identified for replanting. These include R.
mucronata, C. tagal and A. marina. For R. mucronata and C. tagal viable propagules
were collected directly from the plants and stored under the shade for five days before
being planted in 10m by 10m plots of mono and mixed species. A. marina seeds were
collected from the forest floor, under shades, and raised in nurseries. Transplanting of
A. marina seedlings was done after three months. Survival of the planted seedlings
and propagules were monitored regularly for a period of two years. The increment in
height, diameter, internodes and leaves were also monitored.

6.2.6 Data analysis

Data was analysed by non-parametric statistics using STATISTICA 8. Graphical
representation were used to display distribution patterns of both adults and juveniles
across the intertidal zone. Correlation between regeneration and canopy gap sizes,
regeneration and cutting densities was done by Spearman R. To assess the occurrence
or absence of regeneration by species under different vegetation zone, the Cochran Q
test, a dichotomous nominal-scale test based on repeated measures as described by
Zar (1999) was used, with the vegetation zone as the groups and the species as blocks.
Diversity indices, including the Shannon Wiener, species richness and species
evenness were also estimated per vegetation zone.

110
 6. Cover change and regeneration

6.3 Results
6.3.1 Forest characteristics
Five cover classes were identified through photo interpretation (Figure 6.2; Table
6.1). In 1969, area under closed canopy mangrove forest in Tudor creek was estimated
at 1,742 ha, 1,525 ha in 1992, and 681 ha in 2005. Human settlements, agricultural
land occupied only 5.9 ha in 1969 and increased to over 50 ha by 1992 around
Mikindani, Jomvu and Mirarani areas. Differences in open sandy areas or mudflats
may arise from (i) differences in tide levels at the time of photography and (ii) plant
succession brought about by ecological changes such as sedimentation along the
coastline or cutting. Changes in terrestrial vegetation (bushland) cover with human
settlements and agricultural activities over the period indicate a significant increase in
demand for wood products and land for farming and settlements. Depletion of
bushlands will eventually expose the mangrove to further exploitation. Land use
changes are also quite notable in the area with more agriculture activities in 1992
compared to 1969. The population of Mombasa has increased from 247,073 in 1969
to 492,024 in 1993. Currently, the population is estimated at 917,864 (GoK, 2005).

Changes between 1969 and 1992 were moderate and may be associated with
population growth reflecting a direct increase in demand on mangrove wood. Changes
between 1992 and 2005 were pervasive and reflect not only the population growth,
but other possible factors. Information from locals indicate massive mortalities of
trees during and after the 1997-1998 El-Niño event mainly due to siltation, which
gave rise to the numerous open areas within the forest, which little recovery 11 years
later.
Table 6.1: Estimated area of cover types in ha for the period
1969, 1992 and 2005.
Class 1969 1992 2005
Closed canopy mangroves 1,741.83 1,525.97 681.99
Open canopy mangroves 203.40 475.16 519.62
Open sandy areas/mudflats 684.62 554.24 692.74
Water 1,498.41 1,273.95 1,057.28
Bare dark mud 0.00 0.00 169.55

111
 6. Cover change and regeneration

Figure 6.2: Maps displaying the changes in mangrove cover for 1969, 1992 and 2005.

112
 6. Cover change and regeneration

6.3.2 Physicochemical characteristics

The 15 cm sediment cores revealed variations in grain sizes between sites in the landward
fringe proper and the intertidal gradient, with the landward fringe having high fractions of
larger grain sizes (sandy). The intertidal fringe was almost predominantly composed of
fine sediments. However, site conditions along transects did not vary significantly for
sediment types (Kruskal Wallis test, H (4, N = 200) = 2.004, p = 0.73) for the low and
mid tidal fringe. However, we observed that C. tagal occupied sites with sandy soils
(coarse sediments ≈ 355µm grain size; 75% ± 5% composition) compared to R.
mucronata which occupied sites with clay to silty sediments (68µm grain size; 64% ± 6%
composition).

Figure 6.3: Sediment grain size distribution across the intertidal gradient.

The sediment pore-water concentrations for NH4+, NOx-, PO43- and total organic matter
(%TOM) for complete and partial sewage exposed sites are shown in table 5.2. There
were significant differences in NH4+ (Kruskal-Wallis test: H (3, N= 124) = 45.36 p <
0.00001), NOx- (H ( 3, N= 124) = 18.48 p = 0.0004), PO43- (H ( 3, N= 124) =29.54084 p
< 0.00001) and %TOM (Z = -2.83, p = 0.005) concentrations between sites, with
completely exposed sites being characterised by higher NH4+ and NOx- levels, with
higher %TOM content. Higher nutrients levels were observed in the landward and
seaward fringe for completely exposed sites and decreasing nutrients levels from
landward to seaward fringe for partially exposed sites and an increase in %TOM from
landward to seaward (3.8% ± 3.31 – 7.79% ± 1.11). Salinity increased from landwards,

113
 6. Cover change and regeneration

attaining values in the range 28‰ – 40‰ in the seaward fringe and 28‰ – 55‰ in the
landward fringe during the dry and wet seasons respectively.

6.3.3 Forest composition and structure

The actual density and dispersal of adult trees was highly variable. A total of 16
vegetation zones or associations were observed, depicting large canopy openings, with an
obscure zonation pattern. Clustered distribution pattern was common for single species
stands and random to uniform dispersion under mixed species (Morisita index; table 6.2).
The distribution of species across the intertidal gradient is summarised graphically in
figure 6.4. The dominant species, R. mucronata, occurred in almost all vegetation zones,
X. granatum and B. gymnorrhiza, seldom occur in pure stands, and were always
associated with R. mucronata. C. tagal is restricted mainly landward and sparsely
seaward, where sediments were sandy (coarse) coupled with high elevation. C. tagal was
characterised by low basal area, D130 and height. S. alba and A. marina were composed of
big and tall trees, S. alba occupying the seaward fringe, while A. marina occurred under
mixed stands with R. mucronata and C. tagal, extending the entire length from landward,
mid- to seaward fringes. A. marina and R. mucronata occupied all zones across the
intertidal gradients, unusual for A. marina, which was described as displaying a disjunct
zonation, occupying only the landward and seaward fringes in Gazi (Ca. 47 km south;
Dahdouh-Guebas et al., 2004). However, the presence of channels criss-crossing the
intertidal mangrove mudflat also influences species distribution, unlike in Gazi Bay.

The height and D130 was defined by species and location across the intertidal gradient,
with A. marina and S. alba attaining higher heights and bigger D130, while C. tagal and R.
mucronata had relatively lower values, and characteristically lower basal areas. In terms
of location across the intertidal gradient, landward or the high tide fringe is dominated by
short and smaller trees for all species, with the landward fringe proper being devoid of
any vegetation and characteristically occupied by stumps, while the mid tidal range had
medium sized trees of each species. The low tide fringe and along edges of channels were
occupied by bigger and taller trees of A. marina, S. alba or R. mucronata.

114
 6. Cover change and regeneration

Table 6.2: Density of trees per ha and basal area (in brackets), average D 130 (± standard
deviation), average height (± standard deviation) and the Morisita Index of dispersion (q =
196) by vegetation assemblage.
Vegetation zone Density/ha D130 (cm) Height (m) Frequency Iδ
(BA m²/ha) (%)
A. marina 863 ± 318 (30.00) 19.53 ± 10.26 6.16 ± 2.90 9 27.44
C. tagal 867 ± 779 (2.71) 5.39 ± 1.42 1.87 ± 0.27 1 3.92
A. marina/ 571 ± 229 (10.94) 14.37 ± 9.49 4.19 ± 2.13 1 0.784
C. tagal 557 ± 532 (2.73) 7.91 ± 3.55 1.89 ± 0.39 2.35
R. mucronata 1,148 ± 561 (5.31) 6.86 ±4.95 3.48 ± 1.34 28 88.2
A. marina / 614 ± 311 (37.18) 24.54 ± 12.27 9.16 ± 3.11 10 19.01
R. mucronata 957 ± 911 (4.33) 8.46 ± 5.78 3.93 ± 1.60 11.76
A. marina/ 200 ± 100 (21.02) 18.67 ± 7.84 5.51 ± 2.00 2 0.80
R. mucronata/ 800 ± 200 (0.145) 7.18 ± 3.34 2.44 ± 0.54 3.90
C. tagal 600 ± 156 (0.30) 5.05 ± 0.93 2.63 ± 0.75 0.02
A. marina/ 400 ± 122 (13.91) 18.62 ± 8.68 6.53 ± 1.87 2 0.80
R. mucronata/ 600 ± 346 (4.83) 8.73 ± 7.24 4.62 ± 1.89 0.04
S. alba 533 ± 289 (9.47) 15.49 ± 7.80 5.97 ± 2.52 4.84
R. mucronata/ 613 ± 421 (2.00) 5.67 ± 2.36 2.53 ± 0.83 23 65.70
C. tagal 1,060 ± 638 (3.96) 5.78 ± 3.21 2.83 ± 0.92 6.21
R. mucronata/ 333 ± 321 (2.96) 6.41 ± 3.19 7.60 ± 11.81 4 0.02
S. alba 1,000 ± 361 (27.36) 14.74 ±12.61 6.88 ± 3.17 11.90
S. alba 1,657 ± 1,437 (26.6) 10.66 ± 6.58 4.39 ± 1.44 20 62.96
S. alba/ 950 ± 71 (18.06) 16.24 ± 6.22 8.00 ± 3.22 0.05 0.16
A. marina 100 ± 141 (4.84) 17.95 ± 3.16 4.00 ± 1.15 0.014

6.3.4 Regeneration patterns

As a general observation, regeneration is higher for the urbanised Kombeni creek, and
lower for the rural Tsalu creek. However, regardless of location, regeneration patterns
under different vegetation associations or mix and across the intertidal gradient displayed
similar distribution patterns. The regeneration patterns are succinctly summarised in table
6.3 and figure 6.5. Regeneration composition by species under different vegetation
associations did not differ significantly for both seedlings and saplings (Cochran’s Q test;
χ² = 17.667, p = 0.064 and χ² = 11.163, p = 0.36 respectively), with species mix within an
association not always restricted to overstorey species composition. This observation
implies a seed/propagule dispersal and recruitment pattern that promotes colonisation of
distant stands by species other than the overstorey species. This has allowed a diverse
range of species mix, indicating the presence of suitable conditions for establishment of
propagules in most locations.

115
 6. Cover change and regeneration

Figure 6.4: The distribution of adult trees per 100m² plots in relation to distance from the
creek, with the corresponding height and diameter distributions. Data presented was
obtained from five transects of ten plots in each transect.

R. mucronata, the most widely distributed species had high seedling and sapling
densities, and like C. tagal, saplings were in higher density than seedlings. R. mucronata
seedlings and saplings occurred in the understorey under all cover types (species mix)
and inundation, conferring advantages to this species under the current disturbance
regime (Table 6.3). This distribution favours R. mucronata establishment in relation to

116
 6. Cover change and regeneration

other species, such as A. marina and C. tagal, which are restricted in distribution.
Seedlings of A. marina occurred in higher densities than saplings (Figure 6.5), a
phenomenon brought about by either high seedling mortality or the effects of „mast
production‟, characterised by variable seed production over years, ranging from years of
seedlings paucity to years of seedlings abundance. This is an adaptation common in some
terrestrial plants to control seed predation or cushioning against adverse conditions
(Zlotin and Parmenter, 2008).

Table 6.3: Seedlings and saplings density ha-1 in the understorey (Ust.) and gaps. Values in
brackets indicate the Morisita Index of dispersion (q = 196).
Vegetation R. mucronata A. marina C. tagal
zone Seedling Saplings Seedlings Saplings Seedlings Saplings
A. marina Ust. 5 (0.0014) 8 (0.00038) 27 (0.19) 392 (170.28) - -
Gap 51 68 495 49 - -
A. marina/ Ust. 2 (-) 12 (0.00033) - 12 (0.12058) 4 (0.0027) 38 (0.14)
C. tagal Gap - - 71 17 85 41
C. tagal Ust - - - - 14 (0.096) 65 (0.60)
Gap - - - - - -
R. mucronata Ust. 1,378 (110.57) 4,303 (111.27) 6 (0.0078) - 7 (0.025) 16 (0.037)
Gap 1,251 2,263 327 824 37 17
R. mucronata/ Ust. 188 (1.51) 360 (0.665) 784 (165.42) 16 (0.27) 3 (0.0099) 22 (0.067)
A. marina Gap 129 322 302 202 - -
R. mucronata/ Ust. 16 (0.015) 45 (0.012) - - 27 (0.38) 44 (0.28)
A. marina/ Gap 176 578 161 - 200 185
C. tagal
R. mucronata/ Ust. 5 (0.0011) 11 (0.00066) 37 (0.37) - - -
A. marina/
S. alba Gap - - - - - -
R. mucronata / Ust. 66 (0.254) 356 (0.76) - - - -
B. gymnorrhiza Gap - - - - - -
R. mucronata/ Ust. 153 (1.35) 546 (1.79) - - 555 (161.89) 984 (138.68)
C. tagal Gap 46 305 15 - 254 417
R. mucronata/ Ust. 21 (0.026) 44 (0.012) - 1 (0.00058) - -
S. alba Gap - - - - - -

The distribution of seedlings and saplings varied in densities with canopy cover and
species composition with location across the intertidal area (Figure 6.5). Regeneration
densities correlated significantly with cutting intensity (Spearman-R = 0.448 (N = 196),
t(N-2) = 6.98, p < 0.001), and subsequent canopy cover (Spearman-R = 0.228 (N = 196),
t(N-2) = 3.26, p < 0.001), medium sized gaps (10 – 50% cover ≈ 10 – 50m²), with lower
to medium cutting intensities having higher seedling and sapling densities compared to
smaller (less than 10% cover ≈ less than 10m²) and larger gap sizes (greater than 50%

117
 6. Cover change and regeneration

cover ≈ greater than 50m²; Figures 6.5 and 6.6). R. mucronata and C. tagal seedlings
occupied small to medium gaps (Figure 6.5), where cutting is low to moderate, while
more A. marina seedlings and saplings occupied extended gaps.

Figure 6.5: The distribution of seedlings and saplings per 100m² plots in relation to
distance from the creek. Data presented was obtained from five transects of ten plots in
each transect.

Across the intertidal gradient, thus inundation, seedlings and sapling densities and species
composition varied (Figure 6.5). A. marina and R. mucronata, with low densities of C.
tagal occurred near the creek (the low-tide range). The mid tidal range was dominated by
R. mucronata and low densities of C. tagal. Densities of C. tagal seedlings and saplings
were higher landwards (the high-tide range), and where elevation was high. The landward
fringe proper lacked any form of regeneration for any species. R. mucronata seedlings
were present regardless of tidal position or vegetation zone. However, as a general
observation, A. marina seedlings colonised enlarged or extended canopy gaps regardless

118
 6. Cover change and regeneration

of vegetation or tidal inundation. Sampled extended or enlarged gaps were mainly

composed of low densities or no adult trees, with the presence of colonising seedlings of
mainly A. marina, R. mucronata, C. tagal, and only in one gap, S. alba.

1600 1600

1400 1400
R. mucronata R. mucronata
Seedlings Counts per 100m²

Seedlings Counts per 100m²

1200 A. marina 1200 A. marina
C. tagal
C. tagal
1000 1000

800 800

600 600

400 400

200 200

0 0

-200 -200
-20 0 20 40 60 80 100 120 140 160 180 0 10 20 30 40 50 60 70 80 90 100
Cuttings per 100m² Percentage Canopy Cover

Figure 6.6: Scattergrams showing the distribution of seedlings and saplings by species per
100m² plots in relation to canopy cover (Spearman-R = 0.228 (N = 196), t(N-2) = 3.26, p <
0.001) and cutting densities (Spearman-R = 0.448 (N = 196), t(N-2) = 6.98, p < 0.001).

6.3.5 Seedling growth and mortality

The mortality rates for replanted plots is summarised in table 6.4 and the growth rates in
table 6.5. Mortality was high in the first year (60% – 79% for single species plots, and
87% – 91% for mixed species plots) and lower for the second year (35% – 87% for single
species and 38% – 81% for mixed species). There were significant differences in
mortalities between species (Kruskal-Wallis test: H (28, N = 84) = 57.22, p < 0.001), R.
mucronata recording the highest (79 – 91%) for both single and mixed species plots in
the first year and lower rates in the second year (35 – 43%). However, in both years, C.
tagal and A. marina recorded high mortality rates, mostly occurring in the rainy season as
a direct impact of siltation. During the rainy season, A. marina seedlings stems were
either dislodged or broken as a result of run-off loaded with sediments, while C. tagal
stems were completely buried owing to their characteristic height (Figure 6.7c).

119
 6. Cover change and regeneration

Table 6.4: Summary of replanted seedlings mortalities over a two year period for pure
stands and mixed plots.
Species Initial after 1 year after 2 years
counts Counts % Mortality Counts % mortality
Mono species A. marina 800 256 68 33 87
C. tagal 800 322 60 95 70
R. mucronata 800 170 79 110 35
Mixed species R. mucronata 800 72 91 41 43
C. tagal 800 101 87 25 78
R. mucronata 800 107 87 66 38
C. tagal 800 105 87 20 81

Table 6.5: Summary of average annual increments (average ± standard deviation) in

growth of replanted seedlings. Values in brackets are the relative growth rates (increment
as percentage of initial measure).
Species Diameter (cm) Height (cm) Internodes Branches
(numbers/yr) (numbers/yr)
A. marina 0.48 ± 0.15 29 ± 9 10 ± 2 13 ± 5
(57 ± 21) (65 ± 8)
C. tagal 0.47 ± 0.14 3.10 ± 2.81 3±1 7±1
(9.97 ± 7.37) (72 ± 7)
R. mucronata 0.96 ± 0.01 24 ± 3 6±2 6±1
(59 ± 17) (50 ± 9)

During the NE–monsoon season, many seedlings, particularly A. marina wilted, probably
the impact of high-temperature injury, displaying necrotic lesions, particularly on stems
and hypocotyls, followed by subsequent death of the seedling. While the SE – monsoon
season was prone to levels of siltation, which in some cases burial of A. marina and C.
tagal seedlings or the entire stems of R. mucronata and C. tagal occurred (Figure 6.7a
and 6.7c). During the two years, no prop roots or pneumatophores were observed
forming, while R. mucronata produced buds by the second year, which never developed
(Figure 6.7b). Diameter and height increments were significantly different between
species (p > 0.05). R. mucronata and A. marina were observed to have high growth rates
compared to C. tagal. However, in terms of relative growth rates, C. tagal displayed a
high relative height increment. Within the forest, C. tagal height ranged between 2m –
3m, and D130 between 2.5cm – 10cm, compared to 24.54 ± 12.27cm D130 and 6 m – 15m
height for A. marina and 8.46 ± 5.78 D130 and 5 – 12m height for R. mucronata,
indicating a general slow growth rate for C. tagal, which is mostly located in the
landward fringe, where inundation is infrequent and tidal heights lower. However, height
measurements were confounded by the effects of accreting sediments during the rainy
season.

120
 6. Cover change and regeneration

Figure 6.7: (a) R. mucronata buried by silt (b) R. mucronata producing flowering buds by
the second year (c) a buried C. tagal, only leaves are visible, observe the crab hole nearby
(d) measuring growth in a R. mucronata plot.

121
 6. Cover change and regeneration

6.4 Discussion
Regeneration, thus recruitment of seedlings, is not restricted to overstorey conspecifics or
by vegetation zones/species mix. This implies good seed dispersal ranges, good site
conditions within the ranges of species tolerance for establishment, and thus a good
capacity for natural regeneration. These are typical traits reflecting the dynamic and
resilient nature of the mangrove environment that has withstood disturbances over time,
promoting species mixtures (Ball et al., 1988). The patterns of tree regeneration further
implies that it is unlikely the current tree species distribution pattern across the intertidal
gradient will be maintained, and in the absence of a major disturbance event, R.
mucronata will be expected to increase in abundance throughout the forest given the
current status and prevailing conditions. The species mix and composition in the forests is
important in enhancing ecosystem productivity and stability, and confer the forest with a
broader niche differentiation, promoting increased total resource use, providing a variety
of trophic pathways likely to support richer faunal communities as compared to single
species stands (Bosire et al., 2006; Bosire et al., 2008). This species mix further elevates
the perturbation threshold for the mangroves of Tudor creek, increasing its resilience and
survival.

The distribution of trees across the intertidal mudflats reflects the general trend of
environmental physicochemical parameters reported by studies in other parts of the world
(Sherman et al., 2000; Ball, 2002; Lovelock and Feller, 2003; Lovelock et al., 2003).
Mangroves align themselves along edaphic physicochemical and hydrological gradients
such as salinity and tidal submergence (Bosire et al., 2008; Krauss et al., 2008) in
addition to nutrients limitations (Lovelock et al., 2003). Landward areas of higher
salinities and higher nitrates being characterized by shorter and smaller trees, compared
to the taller and bigger seaward fringe trees. However, the proximity to human
settlements, coupled with ease of access, the landward fringe may be more exposed to
human pressure, resulting in the observed patterns in size distribution.

122
 6. Cover change and regeneration

6.4.1 Forest cover changes

The major factors that seem to fuel the forest cover changes in Tudor creek range from
unregulated harvesting, a high population and demand for wood products and siltation
(Chapter 3, 4; Mohamed et al., 2008). The effect of unregulated harvesting and high
wood demands were tackled separately in chapter 4 (Mohamed et al., 2008). Siltation is
considered an important factor in mangrove ecology (Hutchings and Saenger, 1987;
Ellison, 1998). This is mainly due to their location on low-lying continental coastlines,
where sediments supply is regarded as „essential substrate accretion‟, and is important in
buffering against erosion and the perceived sea level rise (Wolanski and Ridd, 1986;
Ellison, 1998; Wolanski and Chappell, 1996). Under normal circumstances, sediments
may deposit at an estimate rate of 0.5 cm yr-1 – 1.5 cm yr-1 allowing mangroves to
acclimate (Hutchings and Saenger, 1987; Ellison, 1998). Sediment starvation of this zone
as a result of reduced sediment supply may lead to erosion in some susceptible areas
(Bird et al., 2004. Thampanya et al., 2006). However, impacts of sedimentation may be
positive or negative. Positive impacts include creation of new areas for colonisation
(Thampanya et al., 2002), while negative impacts arise out of sudden sedimentation after
heavy rains, that lowers the growth and vigour of mangroves if aerial roots are blocked,
causing mortality (Terrados et al., 1997; Ellison, 1998)

The sedimentation rates that occurred during the single ENSO event are unknown, but
studies from the neighbouring Port Reitz creek (Mwache creek), estimates a 1.4m
deposition of terrigenous sediments in the middle section of the creek, and 0.2m
deposition in the outer edges during the same period, causing significant mortalities of
mangroves (Kitheka et al., 2003). No recovery has occurred in some of these areas 11
years after the siltation event (Mohamed‟s personal observation). We speculate that a
similar scenario might have occurred in Tudor creek, with variable spatial patterns. This
may have caused high mangrove mortalities, resulting in the emergence of the extended
open areas. We hypothesize that, these openings altered edaphic conditions, including
higher soil salinities and temperatures, particularly in the less frequently inundated areas,
altering species distributions. Prevailing conditions after the changes may have favoured
the establishment of A. marina, characteristically the most salt tolerant of all the

123
 6. Cover change and regeneration

mangrove species, identified as a pioneering species (Ball et al., 1987; Osborne and
Berjak, 1997), producing the highest number of propagules and high success rates in
colonising open substrates (Thampanya et al., 2006), with root initiation and subsequent
establishment hardly hampered by extreme saline conditions (Ye et al., 2005).

A. marina stabilised the loose sediments, creating opportunities for other species to
colonise, which later out-compete A. marina as the canopy closes and sesarmid crabs
colonised the understorey. Under full canopy cover, A. marina saplings were virtually
absent possibly due to light intolerance. It is also possible that crab predation for A.
marina seedlings intensified as the canopy cover increased, as reported in other
mangroves globally, mainly in the mid-tidal ranges (Smith III, 1987a, b; Osborne and
Smith III, 1990; Farnsworth and Ellison, 1997a). The PUMPSEA study also reported a
higher biomass of sesarmid crabs under A. marina cover in Mikindani (Tudor creek site
impacted by sewage) than in Gazi Bay and Shirazi mangroves (PUMPSEA, 2007c). Thus
A. marina is gradually replaced by other species in the canopy as growth creates cover, a
typical multi-phase vegetation regeneration characteristic of mangrove forests (Ferwerda
et al., 2007). This pattern can be observed under enlarged canopy openings, where high
densities of A. marina and R. mucronata seedlings and saplings were present under all
vegetation mixes (Table 6.3; Figure 6.8c). It was also under these gaps that the only
seedlings and saplings of S. alba were encountered in a R. mucronata zone in the mid-
tidal range (Figure 6.8a), a peculiar observation due to location. Table 6.6 gives a
summary of diversity indices for the vegetation zones for adults and juveniles, showing
higher species richness under zones with A. marina, and greater diversity or species mix
for juveniles than for adults under most zones.

6.4.2 Regeneration
With exception of R. mucronata saplings (4,303 ha-1), seedling and sapling densities were
on average low for all species. The seaward fringe, S. alba zone and the landward fringe
proper had particularly low regeneration levels. Bosire et al. (2006) also observed low
regeneration for S. alba zone for Gazi Bay mangroves (ca. 47 km south). The variability
in regeneration across the intertidal gradient may reflect the diverse factors involved

124
 6. Cover change and regeneration

including: the hydrological regime with comparatively stronger mechanical effects of

tides and wave currents in the low tidal fringe (S. alba zone); accretion and siltation (the
landward proper); absence of adequate seed bearing trees (mainly in large gaps and the
high tidal fringe including cutting intensity; Figure 6.8b); influence of gaps and gap sizes
(lack of root masses to allow for stranding of propagules) (Duke, 2001; Sherman et al.,
2000; Di Nitto et al., 2008; Mohamed et al., 2008); and importantly the species
adaptation or/and the seedling/propagule predation rates (Dahdouh-Guebas et al., 1999;
Clarke and Kerrigan, 2002) and dispersal ranges of the species involved (Osborne and
Berjak, 1997; Tomlinson and Cox, 2000; Delgado et al., 2001). In addition, the stand age
is an important factor in determining levels of regeneration, and in particular, recruitment
rates of saplings, which increase with stand age (Bosire et al., 2006; Bosire et al., 2008).
This may be a further indication of a recovery phase within Tudor creek, characterised by
lower saplings for all species except for R. mucronata. However, R. mucronata is also
quite tolerant to siltation (Thampanya et al., 2002) compared to the other species, and
possesses a longer propagule, with ample nutrient storage that allows longer survival for
the species propagule even in the understorey (McKee, 1995).

Table 6.6: Species diversity indices, evenness and Ninf for adult and juveniles under the
different vegetation zones.
Zone Adults Juveniles
Species J' H'(loge) Ninf Species J' H'(loge) Ninf
A. marina 1 **** 0 1 4 0.39 0.54 1.22
A. marina/C. Tagal 2 0.97 0.67 1.67 3 0.91 1 2
A. marina/C. tagal/X. Granatum 3 0.83 0.92 1.83 1 **** 0 1
A. marina/S. alba 2 0.65 0.45 1.2 0 **** 0 ****
B. gymnorrhiza 1 **** 0 1 2 0.98 0.68 1.69
C. tagal 1 **** 0 1 0 **** 0 ****
R. mucronata 1 **** 0 1 4 0.23 0.32 1.09
R. mucronata/A. marina 2 0.99 0.69 1.91 5 0.45 0.72 1.64
R. mucronata/A. marina/C. Tagal 3 0.97 1.06 2.18 2 0.52 0.36 1.13
R. mucronata/A. marina/S. alba 3 0.98 1.08 2.37 3 0.57 0.63 1.26
R. mucronata/B. Gymnorrhiza 3 0.63 0.69 1.35 3 0.42 0.46 1.19
R. mucronata/C. tagal 2 0.91 0.63 1.49 2 0.99 0.69 1.90
R. mucronata/S. alba 2 0.76 0.53 1.29 3 0.58 0.64 1.41
R. mucronata/X. granatum 2 0.95 0.66 1.6 3 0.78 0.86 1.49
S. alba 1 **** 0 1 1 **** 0 1

125
 6. Cover change and regeneration

Figure 6.8: (a) S. alba seedlings and saplings under extended gaps (b) the landward fringe
with snags indicating mortality of adult trees and no regeneration (c) High density of A.
marina seedlings in an extended gap receiving raw sewage (d) replanted A. marina, observe
the variable growth rates, with taller and more bush saplings in the back, also observe the
terracing and farming on the cliff in the back, with houses are also built on the cliff.

Low regeneration along the seaward and landward fringes (Figures 6.7, 6.8b, d) has
important implications on the forest with the prospects of global climate change and
associated sea level rise (Duke, 2001; Gilman et al., 2007; Alongi, 2008; Gilman et al.,
2008). Rising sea-level, necessitates that mangrove forests migrate inland with
subsequent die off along seaward margins (Duke, 2001; Gilman et al., 2008). In Tudor
creek this may not be an option due to the steep cliff that borders the creek on the
landward fringe. However, annual sea level rise is estimated at 0.2 cm yr -1 (Bates et al.,
2008), quite tolerable by mangroves considering established sedimentation rates of
between 0.1 - 0.2 cm yr -1 (Ellison, 1998; Kitheka et al., 2003). However, even
conservative predictions indicate acceleration in the rate of sea level rise, which may

126
 6. Cover change and regeneration

potentially result in imminent exhaustion of the replenishment capabilities of seaward

fringe stands, which may lead to a general collapse of mangrove stands in relatively
exposed locations, once the rate of gap replenishment is exceeded by tree death from sea
level encroachment (Duke, 2001). Modelling the impacts of sea level rise in Gazi Bay, Di
Nitto et al. (2008) predicts that a relatively modest rise in sea level within a time span of
20 years could affect the distribution pattern and the specific proportion of the juvenile
vegetation layer, leading to notable regional floristic modifications.

127
 6. Cover change and regeneration

6.5 Conclusion
The species composition of mangrove forests is dependent at least in part on a balance
between large scale and small scale disturbances. This potentially determines species
mix. Therefore zonation, is not strictly species exclusive within a mangrove forest, and
may depend on, or reflect a stage in the stand growth dynamics under a specific
disturbance regime. Seed production within the mangrove stand was found to be
abundant and natural regeneration in moderately exploited or disturbed sites (small gaps)
was generally observed to be good, especially in the absence of persistent stressors
mostly from human factors as reported by Hamilton and Snedaker (1984) and Bosire et
al. (2008). Natural regeneration may confer higher diversity to the stand in terms of
species selection and more functionality to the emergent forest in the long run by
enforcing and diversifying trophic links through fauna flora interactions, nutrient cycling
etc...... (Bosire et al., 2008). However, seedling dispersal, establishment and survival may
be a major limiting factor slowing recovery in a disturbed mangrove, especially when
disturbance is an ongoing process. Under these circumstances, seedlings of more adapted
species will establish easily, replacing less tolerant species. Furthermore, the
characteristics nature of poorly managed peri-urban ecosystems display a progressively
degrading character, lowering diversity, diminishing ecosystem goods and services,
limiting, if not inhibit the system‟s capacity to recover. Thus a lowered resilience for
peri-urban mangrove forests.

128