Estuaries and Coasts

DOI 10.1007/s12237-017-0210-9

SHORT COMMUNICATION

Deforested Mangroves Affect the Potential for Carbon Linkages

between Connected Ecosystems
L. G. Gillis 1 & E. F. Belshe 1 & G. R. Narayan 1

Received: 16 September 2016 / Revised: 29 December 2016 / Accepted: 3 January 2017

# Coastal and Estuarine Research Federation 2017

Abstract Mangrove forests are important sinks and sources seascape can be comprised of mangrove forests, seagrass beds,
of carbon especially for connections to coral reefs and and coral reefs (from land to ocean). The connectivity between
seagrass beds. However, they are increasing under threat from these ecosystems are functionally influenced by dynamic car-
anthropogenic influences. We investigated correlations be- bon and nutrient exchange and fluxes. Positive interactions
tween carbon fluxes from the sediment and water column in within each ecosystem are thought to support the establishment
deforested and intact mangroves. Our findings show that de- and expansion of connected ecosystems (Gillis et al. 2014;
forestation has a negative effect on sediment organic carbon Ogden and Gladfelter 1983). For instance, high tree density
storage and CO2 fluxes. However, species richness and den- helps to entrap upland sediment runoff and nutrients in man-
sity showed a positive correlation with sediment organic car- groves, which can have positive benefits for the growth of
bon storage and CO2 fluxes. An increased density of saplings seaward seagrass and coral reefs. Much work has been com-
showed a positive relationship with dissolved inorganic and pleted on mangrove forests as sinks for organic carbon (OC)
organic carbon draining the mangrove forest at high tide. This and their potential to benefit both seagrass beds and coral reefs
research offers insights into the importance of the key forest (Adame et al. 2010; Granek et al. 2009; Lee 1995), but there is
characteristics influencing the storage and fluxes of carbon. little research regarding the transfer of inorganic carbon.
Alterations in mangrove carbon stocks and retention may af- The mangrove outwelling hypothesis was initially pro-
fect connected ecosystems. posed by Odum (1968) and Odum and Heald (1975) and
has been discussed in other studies (Bouillon 2009;
Keywords Connectivity . Dissolved organic carbon . Hemminga et al. 1994). This hypothesis postulates that
Dissolved inorganic carbon . Carbon dioxide fluxes . the export of mangrove-derived material supports adjacent
Outwelling . Tropical coastal seascape ecosystems and food webs. This export can occur as par-
ticulate material (Adame et al. 2010; Boto and Wellington
1988), which upon decaying releases both dissolved or-
Introduction ganic (DOC) and inorganic carbon (DIC). The importance
of OC derived from mangrove material exported into ad-
Coastal seascapes contain some of Earth’s most productive jacent ecosystems, such as seagrass beds (Adame et al.
ecosystems such as mangrove forests, seagrass beds, 2010) and its availability to a variety of heterotrophic ma-
saltmarshes, and peat swamps (Barbier et al. 2008). The tropical rine organisms such as zooplankton, has been highlighted
in other studies (Granek et al. 2009). However, despite the
Communicated by Alberto Vieira Borges important role that mangroves play in exporting carbon to
adjacent waters, a recent reevaluation of existing data sug-
* L. G. Gillis gests that the majority of the carbon tidally exported out of
[Link]@[Link] mangroves occurs as DIC (Bouillon et al. 2008). DIC is a
product of autotrophic and heterotrophic respiration de-
1
Leibniz Center for Tropical Marine Ecology (ZMT), rived from decomposing organic matter (OM) (Bouillon
Fahrenheitstraße 6, 28359 Bremen, Germany et al. 2008; Maher et al. 2013).
 Estuaries and Coasts

Mangroves are deforested at significantly higher rates than between 11 and 15 h (air temperature = 27–31 °C). Soil efflux
any other tropical forest (Ahmed and Glaser 2016; Lovelock collars (20-cm-diameter and 7-cm-height beveled PVC) were
et al. 2011; Valiela et al. 2001). When the trees are cut down, installed 1 hour prior to taking measurements to a depth of
the top layer of the sediment is exposed to physical forces approximately 3 cm at locations at least 3 m apart. At each of
resulting in greater erosion, chemical weathering of the sedi- these six spatial replicates (within each plot at each location),
ment, and increased rates of decomposition (Lovelock et al. three sub-replicate 2-min measurements were taken,
2011). The clearing of trees also reduces the above ground consisting of a 30-s pre-purge, 10 s for chamber closing, 10-
biomass, especially aerial and prop roots, which causes a re- s dead band, and 70-s observation period. CO2 data from these
duction in OC sequestration (Furukawa and Wolanski 1996; 70 s were used to fit an exponential equation:
Mackenzie et al. 2016). Therefore, accumulation and erosion
rates of OM will likely affect the strength and presence of
dCc
dissolved carbon fluxes (Lovelock et al. 2011). ¼ a½Cs−Ccð0Þe−at ð1Þ
Here, the authors propose important linkages between dt
mangrove forest characteristics including species richness, where Cc is the chamber CO2 concentration corrected for water
basal area, density of trees and saplings, above and below vapor dilution (μmol mol−1), Cs is the CO2 concentration in the
ground biomass, and components of the carbon cycle such soil surface layer communicating with the chamber (μmol
as sediment organic carbon (SOC, DIC, DOC (draining from mol−1), also corrected for water vapor dilution, and a is the rate
the mangrove forest), and CO2 flux from the sediment). We constant (s−1). All model fits had a % coefficient of variation of 1
aim to answer two key questions: (i) Do mangrove forest and an R2 of 0.98 or higher. The minimum CO2 detection limit of
characteristics correlate with carbon storage or sediment the LiCor 8100A set-up is approximately 0.01 μmol m−2 s−2
CO2 flux? and (ii) Can we show relationships among SOC, using the 20-cm survey chamber, and the nominal flow rate
DIC, DOC, and CO2 within intact mangrove forest? Because through the sampling loop was 1.5 SLPM, to maintain chamber
mangroves are undergoing intense deforestation, we also conditions near atmospheric pressure.
aimed to understand how deforestation affects carbon storage In each 10 m2 intact plot (I) at all locations (L1, L2, and
and CO2 flux from the sediment. Using this information, we L3), the forest characteristics including mangrove tree species,
aim to indicate how deforestation might affect carbon diameter at breast height (DBH), and density of trees and
outwelling potential. saplings were measured. In each 10 m2 intact plot, we counted
the number of trees (height <1 m) and saplings (height >1 m)
(Gross et al. 2014); each tree and sampling were identified to
Materials and Methods species level (Ngoile and Shunula 1992; Shunula and Unesco
1995). The DBH (1.3 m) was measured using a diameter tape
This study was conducted in the Chwaka Bay mangrove (Gross et al. 2014); from this, we calculated basal area (BA)
forest, in Eastern Zanzibar (March 2015) (Fig. 1). This for- (Gross et al. 2014):
est (area ≈32.4 km2) is situated in a bay characterized by
seagrass and algal habitats and fringing reefs (Fig. 1). We  . 2
chose three locations in the mangroves, L1 was located next π DBH 2
to the ocean (0 m), L2 was in the middle of the forest BA ¼ ð2Þ
2
(500 m), and L3 was high in the mangrove close to terres-
trial influence (1000 m) (Fig. 1). At each location and over
From this information, we calculated species richness by
the duration of an entire month, we monitored two 10 m2
counting the number of trees within the different species. We
plots, one that was intact (I) and another that was deforested
then calculate above ground biomass (AGB) and below
(clear-cut) ≈6 months (D).
ground biomass (BGB) using algometric equations (Fromard
At all locations (L1, L2, and L3) and plots (I and D), we
et al. 1998; Komiyama et al. 2005):
measured SOC and CO2 fluxes. Three 15 cm sediment cores
were divided into three segments: 0–5, 5–10, and 10–15 cm. BAG ¼ 0:251ρD2:46 ð3Þ
The samples were transported to the marine research station
and dried at 60 °C for ≈24 h. Midday ecosystem respiration BBG ¼ 0:199ρ0:899 D2:22 ð4Þ
(Reco: autotrophic + heterotrophic CO2 emissions) was mea-
sured using a LICOR-8100A infrared gas analyzer (LiCor where ρ equates to wood density (gcm−3) and D is DBH (cm).
Biosciences Inc., Lincoln, NE) at six stations in each location We used general equations as regional equations for AGB and
(L1, L2, and L3) in both intact and deforested plots (I and D) BGB have not been developed (Stringer et al. 2015). We used a
on sequential days during a 2-week period. CO2 fluxes were average of all species present for ρ (0.86 g cm−3) (Stringer et al.
measured within a 20-cm-diameter darkened survey chamber 2015).
Estuaries and Coasts

Fig. 1 Location of study area a in

Zanzibar, Tanzania on Unguja
Island with b the study sites (L1,
L2, and L3) within the mangrove
forest (in green) adjacent to
Chwaka Village

We sampled DIC and DOC when the high tide started to analysis isotope ratio mass spectrometry (EuroEA). The DIC
drain from the mangrove forest. We took three samples samples were measured using the Versatile Instrument for the
every hour until there was no more water draining from Determination of Total Inorganic Carbon and the Titration
the forest in that location. The samples were taken directly Alkalinity System, using the coulometric principle. DOC sam-
from an adjacent channel from the intact plots (I) at each ples were analyzed using an infra-red gas analyzer via a Skaler
location (L1, L2, L3) during three separate tidal cycles. SAN**System.
Sample bottles were transported to the marine station in a For the intact sites (I), landscape-level trends of carbon
cool box, and then stabilized with mercury chloride (DIC) variables (SOC, DIC, DOC, Reco) were assessed using lin-
and hydrochloric acid (DOC). ear regression, with a Tukey’s post-hoc test for differences
All samples were transported to the Leibniz-Center for among all locations (L1, L2 , and L3). Using Pearson’s
Tropical Marine Ecology, Germany for further analysis. correlations, we explored the relationship between carbon
Sediment samples were homogenized, then acidified to re- variables and forest characteristics. For the deforested plots
move carbonates and analyzed for OC by means of elemental (D), we compared differences in SOC storage and Reco

Fig. 2 Trends in mangrove forest

characteristics a tree density,
individuals m−2, b sapling
density, individuals m−2, c: total
basal area, m2 ha−1, and d: species
richness and components of the
carbon cycle, e SOC, %, F
midday Reco, μmol m−2 s−1, g
DOC, μmol and h DIC, μmol
kg−1) with distance from site L1
(oceanic) towards L3 (terrestrial)
 Estuaries and Coasts

from the intact plots (I) using ANOVA for all locations (L1, R2 = 0.28; P < 0.0001) towards L3 (Fig. 2). We found no
L2, and L3). However, because L1 appeared to be very recent- statistical difference between DIC and DOC concentrations
ly deforested ≈1 week, we excluded it from the deforested vs. within samples taken at each hour starting at high tide; there-
intact forest analysis. All statistical analyses were done in R (R fore, we pooled these data. Unlike CO2 fluxes, there were no
Core Team 2012). clear linear decreasing trends in DIC and DOC draining from
the mangrove forest from L1 to L3. DIC and DOC concentra-
tions in the water column at L1 and L3 were significantly lower
Results than DIC (Tukey’s posthoc P = 0.002) and DOC (Tukey’s
posthoc P < 0.0001) in the water column at L2 (Fig. 2).
Both tree and sapling density decreased from L1 (oceanic) to L3 Across the landscape, we found that changes in SOC and
(terrestrial) (Figs. 1 and 2). This corresponded with a decrease in Reco were positively correlated with species richness, total
total basal area from L1 to L3 (Fig. 2), which was driven by basal area, density of trees, AGB, and BGB; in addition,
decreases in tree abundance because tree DBH was significantly SOC was correlated with Reco (Table 1). Reco was also pos-
greater at L3 (mean = 20.4 cm) when compared to L2 (mean itively correlated with sapling density. There was significantly
DBH = 5.9 cm, P = 0.001), or L1 (mean = 7.7 cm, P = 0.004); less SOC in the deforested (D) plots compared to intact plots
however, DBH at sites L1 and L2 did not differ (P = 0.78). (I) (F = 25.5, P < 0.0001). DIC and DOC concentrations in the
Mangrove species composition shifted from a mixed species water column were positively correlated with sapling density,
stand (Rhizophora mucronata, Avicennia marina, Ceriops tagal, and DIC was correlated with DOC (Table 1). However, DIC
Bruguiera gymnorrhiza) at L1, to Rhizophora mucronata dom- and DOC concentrations were negatively correlated with
inated (L2) to Avicennia marina and Bruguiera gymnorrhiza AGB, BGB, and total basal area (Table 1). Species richness,
dominated (L3). A trend in decreasing species richness from basal area, density of trees and saplings, AGB, and BGB were
L1 (species richness = 4), L2 (species richness = 3) and L3 positively correlated with one another except for total basal
(species richness = 3) was recorded (Fig. 1). area versus sapling density, which were not correlated, and
SOC significantly decreased from L1 towards L3 (adj. AGB and BGB versus sapling density, which were negatively
R2 = 0.59; P < 0.0001) (Fig. 2). This corresponded to a mar- correlated (Table 1). Across locations (L2 and L3), the
ginally significant decrease in mean SOC storage in the top deforested sites had significantly lower surface (top 15 cm)
15 cm of sediment with distance (adj. R2 = 0.10; P = 0.06). SOC (F = 32.3, P < 0.0001) (Fig. 3). There was significantly
Midday Reco also significantly decreased from L1 (adj. lower midday Reco in deforested sites (F = 32.4, P < 0.0001;

Table 1 Pearson’s correlations among components of the carbon cycles (%SOC, midday Reco, DIC, DOC) and mangrove forest characteristics

%SOC Reco DIC DOC TD SD TBA AGB BGB Richness

%SOC – 0.83(***)a −0.27(ns)a −0.29(ns)a 0.71(***) 0.37(.) 0.76(***) 0.60(**) 0.55(**) 0.82(***)
Reco 0.53(***)b – −0.17(*)b −0.03(ns)b 0.47(***) 0.25(**) 0.47(***) 0.37(***) 0.33(***) 0.52(***)
DIC −0.30(**)c −0.22(ns)c – 0.54(***)c 0.20(ns) 0.64(***) −0.63(***) −0.82(***) −0.85(***) 0.43(.)
DOC −0.18(ns)d −0.01(ns)d 0.54(***)d – 0.10(ns) 0.35(**) −0.36(**) −0.46(***) −0.48(***) −0.25(.)
TD 0.71(***) 0.47(***) 0.20(ns) 0.10(ns) – 0.83(***) 0.56(***) 0.26(**) 0.18(*) 0.75(***)
SD 0.37(.) 0.25(**) 0.64(***) 0.35(**) 0.83(***) – 0.01(ns) −0.32(***) −0.39 (***) 0.26(*)
TBA 0.76(***) 0.47(***) −0.63(***) −0.36(**) 0.56(***) 0.01(ns) – 0.94(***) 0.92(***) 0.97(***)
AGB 0.60(**) 0.37(***) −0.82(***) −0.46(***) 0.26(**) −0.32(***) 0.94(***) – 0.99(***) 0.83(***)
BGB 0.55(**) 0.33(***) −0.85(***) −0.48(***) 0.18(*) −0.39 (***) 0.92(***) 0.99(***) – ()
richness 0.82(***) 0.52(***) 0.43(.) −0.25(.) 0.75(***) 0.26(*) 0.97(***) 0.83(***) 0.79(***) 0.79(***)

Due to unequal sample sizes among carbon variables, all correlations on the horizontal (rows) are for the full variable vs. the mean of the other carbon
variables.
Units for all variables are as follows: SOC, %; midday Reco, μmol m−2 s−1 ; DOC, μmol; DIC, μmol kg−1 ; TD, individuals m−2 ; SD, individuals m−2 ;
AGB, Kg; BGB, Kg; TBA, m2 ha−1 )
TD tree density, SD seedling density, TBA total basal area, AGB total above ground biomass, BGB total below ground biomass, species richness
Statistical significance is noted in parentheses (*** p < 0.0001, **p = 0.001–0.01, *p = 0.01–0.05, p = 0.1–0.5, ns = not significant >0.05).
a
Pearson’s correlations among %SOC and mean Reco, mean DIC or mean DOC
b
Pearson’s correlations among Reco and mean %SOC, mean DIC or mean DOC
c
Pearson’s correlations among DIC and mean %SOC, mean Reco or mean DOC
d
Pearson’s correlations among DOC and mean %SOC, mean Reco or mean DOC
Estuaries and Coasts

high productivity, amplifying OC inputs (Clough 2013). Forest

basal area and both above- and below-ground biomass were
found to be negatively correlated with DIC and DOC in the
water column draining from the mangrove at high tide. This
potentially indicates that the complex aboveground biomass
hinders lateral flow and subsequent export of dissolved carbon.
Sapling density however was positively correlated with DIC
and DOC in the water column. Smaller, thinner trees with no
aboveground root biomass such as saplings, do not substantial-
ly reduce hydrodynamics (Mazda et al. 1997). This in turn can
potentially lead to the increased DIC and DOC draining from
the mangrove forest seen in the plots that contained a higher
density of saplings. However, sampling during other seasons is
needed to rule out an ephemeral phenology-induced peak of
saplings at the L2 site, which showed the highest DIC and
Fig. 3 Differences in a SOC storage (g C m−2) and b midday Reco (μmol
DOC and negated any clear linear landscape trends in DOC
m−2 s−1) between deforested and intact plots, with solid line denoting or DIC concentration (Fig. 2). It must be noted that our land-
median, dashed line denoting mean, box denotes the 25th and 75th scape patterns of carbon storage and export are only correlative;
quartile, whiskers denote the 5th and 95th quartile, and points are data the effects of forest characteristics on carbon sequestration
outside of the 95th quartile
needs further investigation.
The reduction in SOC storage in deforested sites supports
Fig. 3) across locations (L2 and L3) compared to the intact sites.
our findings of the importance of forest cover for carbon se-
The loss of carbon via respiration within the deforested sites
questration. We found 26% less surface SOC storage on aver-
was on average 1.8 ± 1.4 μmol m−2 s−1 (6600 ± 5164 μmol
age in deforested plots (Fig. 3), which could be due to increased
m−2 h−1). Our overall values reflect midday maximum fluxes,
erosional losses of both particulate and dissolved OC after de-
therefore by using a conservative estimate of 10% of the max-
forestation (Pendleton et al. 2012). Additionally, changes within
imum flux, we estimate a daily emission of 0.96 g C m−2. In the
the sediment environment following a disturbance can lead to
intact forest plots, midday maximum Reco was on average
OC destabilization if the effectiveness of the initial stabilization
4.2 ± 2.5 μmol m−2 s−1 (15,240 ± 8903 μmol m−2 h−1); how-
mechanism(s) are reduced or eliminated (Torn et al. 2009). For
ever, estimates of gross primary production (which are assumed
example, sediment un-consolidation after deforestation could
to be zero in the clear-cut deforested sites) would be required to
lead to OC losses if the dominant mechanism stabilizing the
compare the daily net carbon balance between intact and
OC was inaccessibility of microbes to the substrate within com-
deforested sites.
plex sediment structures or aggregates (Torn et al. 2009).
Alternatively, drying and subsequent oxygenation of the sedi-
ment can lead to increased decomposition if microbial commu-
Discussion nity composition or abundance was supressed by the lack of
oxygen (Torn et al. 2009; Trumbore 2009). Our correlative
In Chwaka Bay, both SOC and gaseous carbon losses (Reco) results do not allow us to identify the mechanisms responsible
increased across the landscape in tandem with each forest char- for the reduction of SOC storage but do generate a number of
acteristic (Table 1); the values were comparable to previous hypotheses that require further investigation.
results (Leopold et al. 2015; Leopold et al. 2013). The high Deforested plots had significantly lower midday Reco
forest species richness and density resulted in increased carbon (Fig. 3), which is at least partly due to the loss of trees (that
sequestration, likely due to the presence of more complex root are large contributors to autotrophic respiration), but may also
structures (Mackenzie et al. 2016) but also could depend on the be due to alterations in the availability of substrate for mi-
OC content of the leaves and the type of wood (hard or soft crobes. When mangroves are initially cleared, the CO2 efflux
wood) (Ray et al. 2012). Large above ground biomass has been primarily reflects the oxidation of labile fractions of OM new-
shown to reduce water velocity facilitating sedimentation or ly exposed to oxic conditions (Bulmer et al. 2015; Lovelock
trapping of OM (Furukawa and Wolanski 1996; Jennerjahn et al. 2011). However, this fraction is rapidly depleted
and Ittekkot 2004; Mackenzie et al. 2016). In addition, resulting in slower decomposition of refractory pools, which
diversity-mediated stimulation of microbial activity has been then dominate the CO2 efflux (Lovelock et al. 2011).
shown to result in an increase in SOC due to proliferation of Although, the exact time since deforestation occurred in our
microbial products occurring in the slow-cycling carbon pool plots is unknown, the remaining heterotrophic respiration
(Lange et al. 2015). A high density of trees will also equate to from the sediment was likely dominated by this slower pool,
 Estuaries and Coasts

resulting in a continued estimated loss of 0.96 g C m−2 day. References

Our CO2 efflux is likely to be an underestimation because we
did not make measurements over pneumatophores or crab Adame, M.F., D. Neil, S.F. Wright, and C.E. Lovelock. 2010.
burrows which both are thought to be major contributors to Sedimentation within and among mangrove forests along a gradient
of geomorphological settings. Estuarine Coastal and Shelf Science
CO2 efflux (Bouillon et al. 2008; Kristensen et al. 2008). In
86: 21–30.
addition, another pathway of CO2 to adjacent ecosystems Ahmed, N., and M. Glaser. 2016. Coastal aquaculture, mangrove defor-
could be at the surface-water interface (Borges et al. 2003; estation and blue carbon emissions: is REDD+ a solution? Marine
Bouillon et al. 2008). We did not measure this, but a future Policy 66: 58–66.
Barbier, E.B., E.W. Koch, B.R. Silliman, S.D. Hacker, E. Wolanski, J.
research direction could be quantifying this flux from
Primavera, E.F. Granek, S. Polasky, S. Aswani, L.A. Cramer, D.M.
deforested mangrove forests. Stoms, C.J. Kennedy, D. Bael, C.V. Kappel, G.M. Perillo, and D.J.
While DIC and DOC in the water column draining from the Reed. 2008. Coastal ecosystem-based management with nonlinear
deforested plots were not measured in this study, other studies ecological functions and values. Science 319: 321–323.
have reported reduced SOC by 50% over 8 years after the re- Borges, A.V., S. Djenidi, G. Lacroix, J. Theate, B. Delille, and M.
Frankignoulle. 2003. Atmospheric CO2 flux from mangrove sur-
moval of trees (Pendleton et al. 2012), which likely leads to rounding waters. Geophysical Research Letters 30.
increased export of both particulate and dissolved carbon. Boto, K.G., and J.T. Wellington. 1988. Seasonal variations in concentra-
Deforestation leads to the exposure of the sediment surface to tions and fluxes of dissolved organic and inorganic materials in a
hydrodynamics and chemical weathering; therefore, the top tropical, tidally-dominated, mangrove waterway. Marine Ecology
Progress Series 50: 151–160.
layers of the sediment will be eroded. These erosional processes Bouillon, S., A.V. Borges, E. Castañeda-Moya, K. Diele, T. Dittmar, N.C.
likely contributed to the reduction of SOC in our deforested plots Duke, E. Kristensen, S.Y. Lee, C. Marchand, J.J. Middelburg, V.H.
and have potentially led to an increased DOC and DIC efflux. Rivera-Monroy, T.J. Smith, and R.R. Twilley. 2008. Mangrove pro-
This would represent an increase in export to adjacent ecosys- duction and carbon sinks: A revision of global budget estimates.
Global Biogeochemical Cycles 22: n/a-n/a.
tems over the short term. Depending on the forests’ function (i.e. Bouillon, S., and R. Connolly. 2009. Carbon Exchange Among Tropical
buffering), increased carbon export could have negative effects Coastal Ecosystems. In Ecological connectivity among tropical
on connected ecosystems. However, even if the forest is a carbon coastal ecosystems, ed. I. Nagelkerken, 45–70: Springer
source, the loss of forest primary production with deforestation Netherlands.
Bulmer, R.H., C.J. Lundquist, and L. Schwendenmann. 2015. Sediment
will ultimately result in a reduction of beneficial connectivity. properties and CO2 efflux from intact and cleared temperate man-
grove forests. Biogeosciences 12: 6169–6180.
Clough, B.F. 2013. Primary Productivity and Growth of Mangrove
Conclusion Forests. In Tropical Mangrove Ecosystems, 225–249: American
Geophysical Union.
Fromard, F., H. Puig, E. Mougin, G. Marty, J.L. Betoulle, and L.
In the mangrove forest surrounding Chwaka Bay, Tanzania, Cadamuro. 1998. Structure, above-ground biomass and dynamics
we show correlative evidence that forest with dense, structur- of mangrove ecosystems: new data from French Guiana.
ally complex, and diverse trees retains more organic carbon in Oecologia 115: 39–53.
Furukawa, K., and E. Wolanski. 1996. Sedimentation in mangrove for-
the sediment and donate less DIC and DOC to surrounding
ests. Mangroves and Salt Marshes 1: 3–10.
waters, as measured by concentrations in water draining from Gillis, L.G., T.J. Bouma, C.G. Jones, M.M. van Katwijk, I. Nagelkerken,
the forest. These correlations are further bolstered by compar- C.J.L. Jeuken, P.M.J. Herman, and A.D. Ziegler. 2014. Potential for
isons of SOC and Reco between intact and deforested sites, landscape–scale positive interactions among tropical marine ecosys-
tems. Marine Ecology Progress Series 503: 289–303.
which show a 26% reduction in SOC storage in deforested
Granek, E.F., J.E. Compton, and D.L. Phillips. 2009. Mangrove-exported
sites along with an estimated loss of 0.96 g C m−2 d−1 via nutrient incorporation by sessile coral reef invertebrates. Ecosystems
ecosystem respiration. While more research is needed across 12: 462–472.
greater temporal and spatial scales, this study offers insight Gross, J., E.E. Flores, and L. Schwendenmann. 2014. Stand structure and
into the importance of the key forest characteristics influenc- aboveground biomass of a Pelliciera rhizophorae mangrove forest, gulf
of Monitjo Ramsar site, Pacific coast, Panama. Wetlands 34: 55–65.
ing the storage and fluxes of carbon. Further research on the Hemminga, M., F. Slim, J. Kazungu, G. Ganssen, J. Nieuwenhuize, and
effects of deforestation on the carbon cycle within mangrove N. Kruyt. 1994. Carbon outwelling from a mangrove forest with
forests and between connected ecosystems is imperative be- adjacent seagrass beds and coral reefs (Gazi Bay, Kenya). Marine
cause mangroves are being deforested at an accelerating rate Ecology Progress Series 106: 291–301.
Jennerjahn, T.C., and V. Ittekkot. 2004. Relevance of mangroves for the
globally (Valiela et al. 2001). production and deposition of organic matter along tropical continen-
tal margins. Naturwissenschaften 89: 23–30.
Acknowledgments We thank the Leibniz Center for Tropical Marine Komiyama, A., S. Poungparn, and S. Kato. 2005. Common allometric
Ecology chemical laboratory for assistance and the Institute or Marine equations for estimating the tree weight of mangroves. Journal of
Sciences, University of Dar es Salaam and Dr. Narriman S. Jiddawi for Tropical Ecology 21: 471–477.
scientific and logistical support in Stone Town, Zanzibar. The authors Kristensen, E., S. Bouillon, T. Dittmar, and C. Marchand. 2008. Organic
would also like to thank Professor Martin Zimmer for his support and carbon dynamics in mangrove ecosystems: a review. Aquatic Botany
help with final edits of the ms. 89: 201–219.
Estuaries and Coasts

Lange, M., N. Eisenhauer, C.A. Sierra, H. Bessler, C. Engels, R.I. Odum E & Heald EJ (1975) The detritus based food web of an estuarine
Griffiths, P.G. Mellado-Vazquez, A.A. Malik, J. Roy, S. Scheu, S. mangrove community. In Estuarine Research, ed. Cronin LE, 1:
Steinbeiss, B.C. Thomson, S.E. Trumbore, and G. Gleixner. 2015. 265–286. Academic Press.
Plant diversity increases soil microbial activity and soil carbon stor- Ogden, J.C., and E.H. Gladfelter. 1983. Coral reefs, seagrass beds and
age. Nat Commun 6. mangroves: their interactions in the coastal zones of the Caribbean.
Lee, K.S. 1995. Mangrove outwelling: a reveiw. Hydrobiologia 295: Pendleton, L., D.C. Donato, B.C. Murray, S. Crooks, W.A. Jenkins, S. Sifleet,
303–212. C. Craft, J.W. Fourqurean, J.B. Kauffman, N. Marba, P. Megonigal, E.
Leopold, A., C. Marchand, J. Deborde, and M. Allenbach. 2015. Temporal Pidgeon, D. Herr, D. Gordon, and A. Baldera. 2012. Estimating global
variability of CO2 fluxes at the sediment-air interface in mangroves Bblue carbon^ emissions from conversion and degradation of vegetated
(New Caledonia). Science of the Total Environment 502: 617–626. coastal ecosystems. PloS One 7: e43542.
Leopold, A., C. Marchand, J. Deborde, C. Chaduteau, and M. Allenbach. Ray, R., N. Majumder, C. Chowdhury, and T.K. Jana. 2012. Wood chem-
2013. Influence of mangrove zonation on CO2 fluxes at the sedi- istry and density: an analog for response to the change of carbon
ment–air interface (New Caledonia). Geoderma 202-203: 62–70. sequestration in mangroves. Carbohydrate Polymers 90: 102–108.
Lovelock, C.E., R.W. Ruess, and I.C. Feller. 2011. CO2 efflux from R Core Team (2012). R: A language and environment for statistical com-
cleared mangrove peat. PloS One 6: e21279. puting. R Foundation for Statistical Computing, Vienna, Austria.
Mackenzie, R.A., P.B. Foulk, J. Van Klump, K. Weckerly, J. ISBN 3–900051–07-0, URL [Link]
Purbopuspito, D. Murdiyarso, D.C. Donato, and N.N. Vien. 2016.
Shunula, J.B., and Unesco. 1995. Human impact on mangrove forests: A
Sedimentation and belowground carbon accumulation rates in man-
case study of Zanzibar.
grove forests that differ in diversity and land use: a tale of two
mangroves. Stringer, C.E., C.C. Trettin, S.J. Zarnoch, and W.W. Tang. 2015. Carbon
Maher, D.T., I.R. Santos, L. Golsby-Smith, J. Gleeson, and B.D. Eyre. stocks of mangroves within the Zambezi River Delta, Mozambique.
2013. Groundwater-derived dissolved inorganic and organic carbon Forest Ecology and Management 354:139–148.
exports from a mangrove tidal creek: the missing mangrove carbon Torn, M. S., C. W. Swanston, C. Castanha, and S. E. Trumbore. 2009.
sink? Limnology and Oceanography 58: 475–488. Storage and turnover of organic matter in soil. In biophysico-
Mazda, Y., M. Magi, M. Kogo, and P.N. Hong. 1997. Mangroves as a chemical processes involving natural nonliving organic matter in
coastal protection from waves in the Tong King delta, Vietnam. environmental systems, eds. N. Senesi, B. Xing, and P. M. Huang.
Mangroves and Salt Marshes 1: 127–135. John Wiley & Sons, Inc., Hoboken, NJ, USA. doi:10.1002
Ngoile, M.A.K., and J.P. Shunula. 1992. Status and exploitation of the /9780470494950.ch6
mangrove and associated fishery resources in Zanzibar. Trumbore Susan. 2009. Radiocarbon and soil carbon dynamics. Annual
Hydrobiologia 247: 229–234. Review of Earth and Planetary Sciences 37(1):47–66
Odum, E.P .1968. A research challenge: evaluating the productivity of Valiela, Ivan, Jennifer L. Bowen, and Joanna K. York. 2001. Mangrove
coastal and estuarine water. Proc 2nd Sea Grant Conf, University of forests: one of the world’s threatened major tropical environments.
Rhode Island: 63–64. BioScience 51(10):807