nvestigating the influences of tidal inundation and surface elevation on the establishment and early development of mangroves for application in understanding mangrove rehabilitation techniques 1 5

Chapter 4 – Interspecific variations in survival and growth
responses of mangrove seedlings to three contrasting
inundation durations

4.1 Introduction
Mangroves generally exhibit multiple adaptations to anaerobic substrate conditions
and periodic inundation, primarily through having modified roots that comprise mostly
of aerenchyma tissue with air spaces that allow rapid diffusion of oxygen through the
root lenticels to the rest of the submerged root system (Lovelock et al., 2006, 2006b).
Yet, studies have reported that there exist species-specific responses to duration of
inundation (Luzhen et al., 2005). Experimental treatments simulating natural tidal
amplitudes showed that under different treatments, species grew at different rates
(Ellison & Farnsworth, 1997; Chen et al., 2005; He et al., 2007). He et al. (2007)
reported species-specific differences in biomass, morality and carbon allocation to
roots, stems and leaves, and that experimentally determined inundation tolerance of
mangrove species paralleled the pattern of species distribution for that location. Kitaya
et al., (2002) similarly reported differing growth responses and mortality to elevation
(i.e. inundation duration) for seven species. However, little variation in mortality,
establishment and growth rates was found among five species planted at low- and
high-elevation sites (Clark, 2004).
Focussing on life stage, mangrove seedlings are more vulnerable to prolonged
inundation when compared to saplings and trees as they are physically smaller (i.e.
more prone to whole-plant submergence) and less established (i.e. developmentally
immature). The ability of each individual to respond to the stresses of prolonged
inundation thus becomes crucial towards it survival and growth. Hence, to achieve

47


ecological mangrove rehabilitation success, it is imperative to gain an understanding of
species-specific responses of mangrove seedlings to prolonged inundation periods. In

Chapter 4, the field study examined the influence of surface elevations on the
establishment of mangroves. Given that surface elevation inherently controls
inundation period, whereby lower elevations are inundated more frequently, for longer
durations and to a greater depth, a second mesocosm experiment was designed to
specifically examine the complementary knowledge of inundation durations on
survival and early development of mangroves. Moreover, the mesocosm experiment
further serve as a control for potential confounding factors in the field study (Chapter
3) that may affect observed field results. The objectives of this controlled mesocosm
experiment was to determine if Avicennia alba and Rhizophora mucronata seedlings
exhibited species-specific variation in i) overall survival and, ii) seedling growth in
response to three different inundation durations.
4.2 Materials and Methods
Mesocosm set-up – 12 fibreglass tanks with inner dimensions of 3 x 1 x 1 m were
housed at 82 Sungei Tengah Road, northwest Singapore. There were six experimental
and six reservoir tanks, where each experimental tank was elevated above a reservoir
tank, and connected by a pump (Sobo WP 9200, 2400 L h-1) (Figures 4.1 and 4.2).
This set-up was adapted from a larger study by Balke et al. (2013). The reservoir tanks
were filled with seawater with a salinity of 20 ppt and pH 11. A timer (SoundTech
MDT-338) on the pump was used to fill and drain the experimental tanks
automatically. To simulate tidal inundation, activating the pump pumped water from
the reservoir tank into the experimental tank (i.e. high tide) while de-activating the
pump resulted in complete drainage of the experimental tank back into the reservoir

48


(i.e. low tide). Inside the experimental tank, an overflow at a height 0.9 m maintained
the water depth when filled (Figure 4.1b).

Figure 4.1: (a) Aerial view of experimental set-up, (b) side and (c) aerial view of each pair of

reservoir and experimental tank.

49


Figure 4.2: Photographs of (a) the actual mesocosm set-up and the experimental pots with (b)
Rhizophora and (c) Avicennia seedlings.

Materials

and

Experimental

Design

–

Fresh

R.

mucronata

propagules

(Rhizophoraceae, R. mucronata Lamk.) propagules were collected from the sediment
surface and trees at Sungei Buloh Wetland Reserve and Chek Jawa Wetlands,
northwest Singapore. Rhizophora propagules exhibiting embryonic leaves or a whitish
collar were

considered fresh.

Similarly, fresh

Avicennia

alba

propagules

(Avicenniaceae, A. alba Blume) were collected from the sediment surface at Sungei
Buloh Wetland Reserve. Avicennia propagules that had just shed or have an intact seed
coat were considered fresh (Balke et al., 2013). Rhizophora and Avicennia propagules
were sown, as per Kitaya et al. (2002) and Balke et al. (2013) respectively, in
polystyrene bags filled with mangrove mud collected from mangroves near Sungei
50


Buloh. To allow for seedling anchoring, Rhizophora propagules were watered with
freshwater for 35 days and Avicennia propagules, 10 days. Anchored seedlings were
then randomly assigned to one of the three experimental treatments: (i) short
inundation of 5 h (ii) medium inundation of 7 h and (iii) long inundation of 9 h, semidiurnal tidal regime. The short and medium inundation treatments were chosen as they
were within the typical inundation durations in Southeast Asian mangroves (Van Loon
& Van Mensvoort, 2007), whereas the long inundation treatment served to test the
upper inundation tolerances of seedlings.
It is recognised that the planting of Rhizophora propagules may be a limitation of the
study design as Rhizophoraceae propagules may not always establish in an upright
position under natural conditions (Tomlinson, 1986). Propagules may establish on their
sides and later, develop lateral roots that exhibit differential elongation to then allow

the rooting and vertical establishment of Rhizophora seedlings (Tomlinson, 1986).
Hence, the planting of Rhizophora seedlings will prematurely raise the propagule and
new plumule above the water level of 0.9 m, exposing it to shorter inundation periods
over time. This may favour photosynthesis and result in an over-representation of
survival rates of Rhizophora seedlings. Nonetheless, given that mangrove propagules
are naturally buoyant, planting was necessary to facilitate the rooting of both
Rhizophora and Avicennia seedlings to ensure that their development and survival
could be followed across time. This experimental design is similar to other studies
investigating the effects of surface elevation on establishment and survival of
mangrove seedlings under field conditions (Kitaya et al., 2002; Chen et al., 2005; Lu et
al., 2013).
A total of 138 Rhizophora seedlings and 222 Avicennia seedlings were used. The
seedlings were monitored throughout the experiment for survival and stem height
51


every week. The mesocosm experiment lasted for 11 weeks, from 26 August 2014 – 2
November 2014. Salinity was monitored weekly and marine salt added when necessary
to maintain constant salinity. Temperature and daylight hours in Singapore are
generally constant throughout the year. A separate one-time final measurement of root
and stem length was conducted from 13 – 15th December 2014 for Rhizophora
seedlings only as all Avicennia seedlings had died by then. Mean root length was
derived from measuring four randomly chosen roots, across four cardinal points at the
base of propagule,
Data analyses – Effects of species and inundation treatments on seedling survival were
analysed with a generalised linear model (GLM); binomial distribution, logit link
function (logistic regression), using R 3.1.2 (R development core team, 2014). Using
only data from the 11th week, the response variable was defined as the “Status” of each
seedling, and assigned a binary code of either 1 (alive) or 0 (dead). The explanatory
variables were “Species” (Avicennia or Rhizophora) and “Treatment” (5, 7 or 9 hours

of inundation).
A separate analysis was conducted to model the relation between stem height and
inundation duration and species using a mixed-effects linear model. The software used
was the nlme library in R 3.1.2 (R development core team, 2014). A mixed-effects
model incorporates a mixture of fixed and random effects. Fixed effects are associated
with the average dynamic that may differ among treatment groups, specifically, the
variability in stem height between different species and inundation treatments.
Random effects are associated with the variability of the dynamic among groups,
specifically, the experimental tanks (Pinheiro & Bates, 2000). This approach allows
the analysis to account for unknown differences in species (given they are of different
genera) and temporal correlation (as repeated height measures were taken from the
52


same individual). The response variable was “Stem height” (log transformed).
“Species” (Avicennia or Rhizophora) and “Treatment” (5, 7 or 9 hours of inundation)
were included as fixed effects. “Block” (where experimental tanks were assigned
numbers ranging from 1 to 6) was modelled as a random effect.
Both analyses started with fitting a global model that contained all explanatory
variables as well as two-way interactions. A final model was subsequently determined
by step-wise exclusion of the least significant terms, starting with non-significant twoway interactions (p > 0.05), and then non-significant main effects not included in the
interactions. The best fitting model was determined to be that with the lowest Akaike’s
Information Criterion (AIC) value (Akaike, 1974). The AIC is a measure of the
parsimony of models, and is based on a trade-off between deviance reduction and the
number of parameters fitted in the model.
Finally, for the remaining survivals (which were Rhizophora seedlings), Analysis of
Variance (ANOVA) was used to analyse if root length differed between inundation
treatments. The data was tested for normality and homogeneous variances before
applying the ANOVA.
4.3 Results

4.3.1 Seedling survival
Across all three inundation treatments, Rhizophora seedlings exhibited 100 % survival
rates (Figure 4.3; R5, R7 and R9 = 1.0). For Avicennia seedlings, the proportion of
seedlings alive exhibited an inverse relationship to inundation period. Treatment A9
with the longest inundation period of 9 hours exhibited the lowest percentage survival
46.1 % whereas Treatment A5, with the shortest inundation period of 5 hours exhibited

53


the highest percentage survival of 94.4 %. Avicennia seedlings experiencing moderate
inundation period had a survival rate of 48.6 %.

Figure 4.3: Proportion of seedlings alive per inundation treatment. (A = Avicennia spp.; R =
Rhizophora spp.; 5, 7 and 9 represent the number of inundation hours).

A total of three GLMs were fitted and a comparison of the AIC values indicated that
the minimum adequate model was that which included both “Species” and
“Treatment” as parameters, with the AIC value of 261.06 (Table 4.1). “Treatment” was
identified as a significant and negative parameter in explaining for the variance in
survival rates between seedlings, thus showing that longer inundation periods lowered
seedling survival rates.

54


Table 4.1: Performance matrix of the GLM models fitted.

Model
Type

Global

Function

AIC Value

Status ~ Species + Treatment + Species:Treatment

263.06

Intercept
Species
Treatment
Species:Treatment

Best
Fitting

Coefficient
Estimate
4.66
14.91
-0.57
0.57

P-value
0.00*
0.99
0.00*
0.99

Status ~ Species + Treatment

Intercept
Species
Treatment

261.06

Coefficient
Estimate
4.66
19.24
-0.57

P-value

Coefficient
Estimate
4.47
-0.44

P-value

0.00*
0.98
0.00*

Status ~ Treatment
Intercept

Treatment

364.16
0.00*
0.00*

4.3.2 Seedling growth responses
Rhizophora seedlings exhibited similar cumulative stem height across inundation
treatments over 11 weeks. At the 11th week, the mean Rhizophora stem height for
treatments R5, R7 and R9 are 17.0 cm, 16.0 cm and 16.6 cm (Figure 4.4). Avicennia
seedlings exhibited more variation in mean cumulative stem height across time and
inundation treatments. The shorter the inundation duration, the longer the final stem
length in the 11th week. Inundation treatment A5 (with the shortest inundation period
of 5 hours day-1, semi-diurnal) had the longest stem (11.0 cm). The mean stem length
for Treatments A7 and A9 were 9.2 cm and 8.4 cm respectively (Figure 4.4).

55


Figure 4.4: Cumulative stem height of both Rhizophora (top row) and Avicennia seedlings
(bottom row), segregated by inundation treatment, across weeks 1 to 11. Standard errors are
indicated by whiskers.

A total of three models were fitted and a comparison of the AIC values (Table 4.2)
indicated that the best fitting model was that which included “Block” as a random
effect and “Species” as a fixed effect (Table 4.2, p-value < 0.05). An examination of
the fixed effect parameters showed that the stem height of Rhizophora seedlings were
on average, 0.528 cm less than that of Avicennia seedlings (Table 4.2).

56

Table 4.2: Performance matrix of the mixed-effects models fitted.

Model
Type
Global

Fixed Effects
LogHt ~ Species + Treatment + Species:Treatment
Intercept
Species2
Treatment
Species: Treatment

Value
-0.454
-0.370
0.00
-0.09

P-value
0.14
0.50
0.99
0.29

LogHt ~ Species + Treatment
Intercept
Species2

Treatment

Value
0.590
-0.611
-0.022

8241.12
P-value
0.00*
0.00*
0.46

Best Fitting LogHt ~ Species
Intercept
Species2

AIC
Value
8248.14

8233.35
Value
0.627
-0.528

P-value
0.00*
0.00*

4.3.3 Root length of Rhizophora seedlings
In the last week, stem length data was similar across inundation treatments (ANOVA
p-value > 0.05). The mean stem length observed for each inundation treatments R5, R7
and R9 respectively are 22.6 cm, 21.7 cm and 23.9 cm (R9). Root length data was
significantly different across inundation treatments (ANOVA p-value < 0.05).
Inundation treatment R7 had the longest mean root length (21.2 cm), followed by R9
(18.9 cm) and R5 (17.9 cm) (Figure 4.5).

57


Figure 4.5: Barplot of (a) mean stem height and (b) mean root length of Rhizophora seedlings
across three Inundation Treatments (R5, R7 and R9). Standard errors are indicated by
whiskers.

4.4 Discussion
4.4.1 Impacts of prolonged inundation on seedling survival
Of the explanatory variables tested, inundation treatment was flagged as most
important explanatory variable in the GLM that influences seedling survival rates
(Table 4.1; p-value < 0.05). Longer inundation treatments lowered seedling survival
rates. This was more applicable to Avicennia seedlings as Rhizophora seedlings
exhibited 100 % survival across all three treatments (Figure 4.3). For Avicennia
seedlings, the lowest percentage survival (46.1 %) was observed in the 9 hour
inundation treatment, with the highest percentage survival (94.4 %) in the 5 hours
inundation treatment. Avicennia seedlings experiencing moderate inundation period
had 48.6 % survival.

58

The trend where prolonged inundation reduces seedling survivorship is expected as
two direct mechanisms exist by which inundation acts to influence seedling survival.
First, prolonged inundation reduces oxygen available to roots, thereby reducing the
rate of aerobic metabolism and water-use efficiency (Naidoo, 1985; McKee, 1996).
Hovenden et al. (1995) demonstrated that Avicennia marina seedlings had sufficient
aerenchyma to supply the oxygen requirements of the root system for a period of
approximately 1.5 – 3.5 hour per tide. When inundation periods exceed this range,
plants become anaerobic, consuming more energy in order to maintain their
metabolisms. Second, short-term inundation depresses photosynthetic capacity of
mangrove seedlings. When tidal inundation depth and duration exceeds the natural
tolerance of mangroves, the leaf photosynthetic capacity becomes strongly limited
(Ellison & Farnsworth, 1997; Chen et al., 2005). These natural tolerances for optimal
inundation periods occur across a spectrum and are species-specific, eliciting differing
responses across species (Kitaya et al., 2002). For example, Sonneratia caseolaris
seedlings function best in the range of 2 – 4 hours with that for Sonneratia apetala to
be 2 – 8 hours (Chen et al., 2013). This species-specific reason was attributed to the
fact that compared to the exotic S. apetala, S. caseolaris was a local species and hence,
have adapted best to local environmental conditions. Kandelia obovata and Bruguiera
gymnorrhiza functioned best in shorter optimal inundation periods of about 4 and 2
hours respectively (Chen et al., 2005; Wang et al., 2007).
Seedling mortality was observed only in Avicennia but not Rhizophora seedlings.
Seedling mortality began in the 3rd week for the long inundation treatment R9, and was
not observed until the 6th week in the short inundation treatment R5 (Figure 4.3).
Effects of prolonged inundation elicit an immediate, extreme response of seedling
mortality as unlike mature trees/saplings, mangrove seedlings have yet to develop
59


aerial roots and stem lenticels as adaptations to lowered soil oxygen. Moreover, the
short time span might prove insufficient for seedlings to adapt via increasing root

porosity, an anatomical adaptive strategy employed to increase root oxygen reserves
and reduce volume of respiring tissue (Youssef & Saenger, 1996). Throughout the
experiment, all Avicennia seedlings were subjected to whole-plant submergence at the
fixed inundation depth of 0.9 metre. Seedlings thus faced reduced gaseous exchange
and depressed light intensity, inhibiting respiration and photosynthetic assimilation (Lu
et al., 2013). While all Rhizophora seedlings were subjected to whole-plant
submergence at the start of the experiment, their taller seedling height resulting from
an extended hypocotyl allowed some individuals to grow and extended their plumule
beyond the water column. This allows the distal portion of the seedling to assume
aerobic metabolism, potentially conferring the advantage of overcoming complete
inundation in a tidal environment. In a field study by Kitaya et al. (2002), similar
results were observed that relates elevation (i.e. inundation period) to seedling survival
– R. mucronata was concluded to have higher tolerance to prolonged inundation
compared to Avicennia officinalis at an early growth stage. At low elevations, R.
mucronata achieved 80% survival whereas all A. officinalis seedlings died. Likewise,
all A. marina seedlings died when fully submerged for 12 hours (Lu et al., 2013).
Pezeshki et al. (1987) argued that the survival and growth of any species exposed to
underwater stress conditions is dependent on its ability to maintain a functional
photochemical and biochemical system that enables maintenance of positive net
photosynthesis. In this sense, Avicennia seedlings possibly have a lower tolerance to
underwater stress compared to Rhizophora seedlings. A study by Mangora et al. (2014)
showed that ability of Avicennia marina to photosynthesis underwater was lower
compared to B. gymnorrhiza (family Rhizophoraceae). Seedling survival in early

60


development stages can also be influenced by the effect of propagule reserves (Smith
& Snedaker, 2000). Seedlings likely survived fairly well until depletion of propagule
reserves, at which point survival shifts to a higher dependence upon photosynthetic

capacity of individual seedlings. Smaller propagule sizes and lower cotyledon reserves
in Avicennia may thus explain for the observed lower survival rates. The critical
amount of time in which growth responses begin to depend largely on photosynthesis
over propagule reserves was probably crossed during the 11-week period – potentially
around the 4th to 6th week where Avicennia seedling mortality rates were the highest
(Figure 4.3; A7 & A9). This was probably not crossed during the 11-week period for
Rhizophora seedlings since survival rates remained at 100 %. However, this might just
be attributed to larger propagule reserves as Rhizophora propagules are of a much
larger size. Once the reserves have been depleted, mortality rates might increase.
4.4.2 Impacts of prolonged inundation on seedling growth
This study also showed differential growth response of Rhizophora and Avicennia
seedlings to inundation treatments as substantiated in the mixed-effects model wherein
“Species” was a significant parameter (Table 4.2). Rhizophora seedlings did not
exhibit significant variation in stem height across treatments in comparison to
Avicennia seedlings (Figure 4.4). The data demonstrates that these treatments represent
optimal inundation periods (5 – 9 hours) for Rhizophora seedlings. This concurs with
findings by Hoppe-Speer et al. (2011) where average weekly increase in height was
similar across the 3, 6 and 9 hours semi-diurnal inundation treatments. In those
treatments, seedlings performed best as they exhibited maximum photosynthetic
performance, high stomatal conductance and had the highest biomass accumulation
and leaf production. Yet, Rhizophora seedlings may be able to tolerate longer
inundation periods as they have not been stressed enough to exhibit a stem elongation
61


response, a typical response of wetland plants to prolonged inundation (Jackson &
Drew, 1984). Stem elongation represents plants adapting to prolonged inundation
conditions through shifting biomass accumulation from roots to shoots, thereby
allowing rapid initial growth to increase biomass above water surface. Alternatively,
the observed stem elongation responses in R. apiculata (Kitaya et al., 2002) and R.

mangle (Ellison & Farnsworth, 1993) could in actuality be driven by a combination of
environmental factors given that these were field studies in which conditions were not
ideally kept constant across time and space. Also, longer inundation periods
suppressed Avicennia seedling growth and height. In the 5 hour inundation treatments,
the greatest Avicennia seedling height documented was 44 cm compared to 26 cm and
24 cm in the 7 and 9 hours inundation treatments. Potential reasons would be that
inundation durations exceeding a threshold reduced respiration and photosynthetic
abilities, thereby resulting in low- or non-functioning photochemical and biochemical
systems in seedlings. Also, the root length of seedlings was significantly longer in
inundation treatments with prolonged inundation durations (R7 and R9; Figure 4.5).
While higher frequency of inundation has been shown to maximise growth and
aboveground productivity (Krauss et al., 2006), these results are coherent in suggesting
that permanent flooding might function to stimulate root biomass allocation as an
adaptation to more soil reducing conditions and sulphide accumulation (CastañedaMoya et al., 2011). In the long term, this translates into a response to changes in
hydroperiod that is reflected as changes in aboveground and belowground biomass and
productivity (Krauss et al., 2008).
An examination of the fixed-effect parameters showed that the height of Rhizophora
seedlings were on average, 0.528 cm less than that of Avicennia seedlings (Table 4.2).
Vivipary is the condition in some seed plants in which the sexually produced embryo
62


of the seed continues its development without dormancy into a seedling, while still
attached to the parent plant (Elmqvist & Cox, 1996). Inherent physiological
differences exist in that Rhizophora exhibits vivipary whereas Avicennia provides a
good example of cryptovivipary (Tomlinson & Cox, 2000). For Rhizophora, this
occurs through elongation of the hypocotyl to produce a cigar-shaped seedling. Thus,
Avicennia being a good example of “cryptovivipary”, its growth strategy encompasses
a lack of dormancy but rapid seedling growth after dispersal from the parent plant
(Tomlinson, 1986).

4.4.3 Interactions between inundation and other physical factors that affect seedling
growth
Under field conditions, physical factors exert a combined influence on seedlings
instead of in isolation as designed in this experiment. Thus, observed responses to
prolonged inundation may be confounded through interactions with other physical
factors such as salinity (Ball, 2002). Experiments on Kandelia candel seedlings
showed that while prolonged inundation stimulated stem elongation, this response
remained unchanged in treatments that combined conditions of prolonged inundation
with increased salinity (Ye et al., 2010). Stem diameter of K. candel seedlings did not
change as well. A separate experiment involving Laguncularia racemosa and
Rhizophora mangle seedlings showed inter-specific responses to combined effects of
inundation and salinity. All growth characteristics of L. racemosa seedlings (e.g.
relative growth rate, leaf area, stem height etc.) were higher than that of R. mangle
across all treatments. Under prolonged inundation, this species growth differentiation
was also more pronounced at low salinity compared to high salinity (Cardona-Olarte et
al., 2006), suggesting that under the same conditions of low to mild stress by

63


inundation and salinity, L. racemosa exhibited the advantage of being competitively
more dominant than R. mangle.
Combined effects of inundation with salinity could also determine spatial distributions
of certain species. Field observations of mature K. obovata trees highlighted that
inundation tolerance thresholds were influenced by a salinity gradient – plants
exhibited longer inundation duration tolerances as salinity decreased (Yang et al.,
2013), and thus preferentially occupied areas with such conditions.
Hence, inundation as a factor in isolation, or in combination with other factors across
varying ranges, will correspond to differences in responses across individual growth
characteristics, species and life stages. Faced with a change in such environmental

conditions, certain species may be conferred a competitive advantage if better growth
is encouraged. This may manifest as some form of spatial restructuring of mangrove
communities (i.e. changes in species composition) and/or mangrove distributions,
depending on spatial and temporal scales of these environmental changes. In the
context of mangrove rehabilitation, specifically the change in environmental
conditions through the re-establishment of tidal connectivity and surface elevation
changes, local inundation frequencies, duration and depth will be altered. This may
translate into landcover changes (e.g. a shift from bare aquaculture ponds to reforested
land) when conditions in degraded site prove favourable for mangrove colonisation.
4.5 Summary
This study highlights that survival and growth of mangrove seedlings vary across both
inundation periods and is species-specific. The threshold of Avicennia seedlings to
prolonged inundation treatments is potentially constrained at 3 hours, as prolonged
inundation past this threshold lowered both the survival and growth rates of Avicennia
64


seedlings. Rhizophora seedlings may have a higher threshold to prolonged inundation
as both survival and growth rates remained unaffected. These results provide
information for use in rehabilitation projects should the restorer resort to planting
mangroves seedlings (as a last resort) in areas prone to prolonged inundation periods.
Yet, results are only representative for early developmental stages and other
environmental factors such as wave energy and salinity, factors contributing to
seedling survival, must be considered before use in mangrove rehabilitation projects.

65