Ecological Applications, 24(3), 2014, pp.

518–527
Ó 2014 by the Ecological Society of America

Carbon stocks of intact mangroves and carbon emissions

arising from their conversion in the Dominican Republic
J. BOONE KAUFFMAN,1,5 CHRIS HEIDER,2 JENNIFER NORFOLK,3 AND FREDERICK PAYTON4
1
Department of Fisheries and Wildlife, Oregon State University, Corvallis, Oregon 97331 USA
2
Watershed Professionals Network, P.O. Box 1641, Philomath, Oregon 97370 USA
3
Counterpart International, 2345 Crystal Drive, Suite 301, Arlington, Virginia 22202 USA
4
AgroFrontera, Calle Ramón Matı´as Mella Number 77, Montecristi, Dominican Republic

Abstract. Mangroves are recognized to possess a variety of ecosystem services including

high rates of carbon sequestration and storage. Deforestation and conversion of these
ecosystems continue to be high and have been predicted to result in significant carbon
emissions to the atmosphere. Yet few studies have quantified the carbon stocks or losses
associated with conversion of these ecosystems. In this study we quantified the ecosystem
carbon stocks of three common mangrove types of the Caribbean as well as those of
abandoned shrimp ponds in areas formerly occupied by mangrove—a common land-use
conversion of mangroves throughout the world. In the mangroves of the Montecristi Province
in Northwest Dominican Republic we found C stocks ranged from 706 to 1131 Mg/ha. The
medium-statured mangroves (3–10 m in height) had the highest C stocks while the tall (.10 m)
mangroves had the lowest ecosystem carbon storage. Carbon stocks of the low mangrove
(shrub) type (,3 m) were relatively high due to the presence of carbon-rich soils as deep as 2
m. Carbon stocks of abandoned shrimp ponds were 95 Mg/ha or ;11% that of the mangroves.
Using a stock-change approach, the potential emissions from the conversion of mangroves to
shrimp ponds ranged from 2244 to 3799 Mg CO2e/ha (CO2 equivalents). This is among the
largest measured C emissions from land use in the tropics. The 6260 ha of mangroves and
converted mangroves in the Montecristi Province are estimated to contain 3 841 490 Mg of C.
Mangroves represented 76% of this area but currently store 97% of the carbon in this coastal
wetland (3 696 722 Mg C). Converted lands store only 4% of the total ecosystem C (144 778
Mg C) while they comprised 24% of the area. By these metrics the replacement of mangroves
with shrimp and salt ponds has resulted in estimated emissions from this region totaling 3.8
million Mg CO2e or ;21% of the total C prior to conversion. Given the high C stocks of
mangroves, the high emissions from their conversion, and the other important functions and
services they provide, their inclusion in climate-change mitigation strategies is warranted.
Key words: blue carbon; carbon stocks; climate-change mitigation; coastal ecosystems; CO2e emissions;
land use; mangroves; REDDþ; shrimp ponds.

INTRODUCTION The net mangrove deforestation rate in the 1990s was

0.72% and was 0.66% during the 2000–2005 period
Mangrove forests occur along the coasts of most
(FAO 2007). This decline may reflect an increased
major oceans in at least 124 countries. Renowned for an
awareness of the value of mangrove ecosystems (FAO
array of ecosystem services including fisheries and fiber
2007) or the complete conversion of the most accessible
production, sediment regulation, and storm/tsunami
sites. In addition to losses of ecosystem services, the
protection, mangroves are nevertheless declining due
global emissions from mangrove degradation have been
to aquaculture expansion, agricultural development, estimated to be as much as 0.12 Pg C/yr (Donato et al.
overharvest, and development (Alongi 2002, Polidoro 2011). This is equivalent to almost 10% of the global
et al. 2010, Giri et al. 2011). A 30–50% decline in land emissions associated with tropical forests, although
area over the past half century has prompted estimates mangroves occupy ,1% of the land area of tropical
that mangroves may functionally disappear in as little as forests.
100 years (Valiela et al. 2001, Duke et al. 2007). About In the insular Caribbean, mangroves were estimated
185 000 ha were lost every year in the 1980s (a net loss of to cover about 762 250 ha in 1990 (Ellison and Farns-
1.0% per year; FAO 2007). Recently, the rate of loss worth 1996). Mangroves have been estimated to occupy
appears to have slowed down, although it is still high. about 21 215 ha in the Dominican Republic. They are
widespread, particularly along the northern coasts
Manuscript received 5 April 2013; revised 16 August 2013;
(Spalding et al. 2010), with large contiguous areas in
accepted 26 August 2013. Corresponding Editor: C. Craft. the northwestern corner within the Province of Mon-
5 E-mail: [email protected] tecristi, the location of our present study.
518
April 2014 MANGROVE CARBON STOCKS AND EMISSIONS 519

While the values and ecosystem services of mangroves classes similar to those defined for other Caribbean
are well known, few data exist for carbon stocks in Latin mangroves by Adame et al. (2013) and Murray et al.
America and even fewer on the declines in C stocks or (2003): (a) tall mangroves with a mean height .10 m
emissions that would arise from land-use conversion of and usually occurring on the margins of rivers and
mangroves. This is especially true in the Caribbean estuaries; (b) medium mangroves that form dense stands
where studies of aboveground biomass have been of trees of 3 to 10 m in height, usually as interior forest,
conducted, but few studies have quantified the total and (c) low mangroves, composed of dense stands of
ecosystem carbon stocks of the mangroves. This is trees whose height is ,3 m and occurring inland of
important as many studies have reported that below- coastal margins (Table 1). In addition to intact
ground C pools account for 49–98% of the total mangroves, we sampled abandoned shrimp ponds. The
ecosystem C stock in mangroves (Donato et al. 2011, shrimp ponds were mangrove forests until they were
Adame et al. 2013). In addition, few measurements have cleared in 1983. The shrimp ponds were actively used for
been made of the C stocks of land uses that replace 10 years until 1993. The ponds were then abandoned
mangroves, and the associated emissions that arise from and not used from 1993 to the time of sampling in 2012.
land-cover change. Given that mangroves have been Levees continued to block freshwater and tidal inputs
characterized as having a high C density coupled with a into these sites at the time of sampling.
high rate of deforestation, there is an increased interest Rainfall in the region averages 672 mm/yr with the
in including mangroves as part of climate-change majority falling during the months of June to October.
mitigation strategies that would reduce the anthropo- The mean annual temperature is 26.38C (Oficina
genic emissions of greenhouse gasses. To effectively Nacional de Meteorologia, Dominican Republic). The
apply these strategies, it is important to quantify and soils in the mangrove were field classified as a fine silt
understand the emission factors associated with changes mixed isohyperthermic Typic Hydraquent.
in land use, so that baselines can be set for countries to
participate in future climate-change mitigation strategies Field sampling
such as reduced emissions from deforestation and We determined the composition, structure, and
degradation (REDDþ; Murdiyarso et al. 2012). ecosystem carbon stocks of 12 different coastal wetlands
The objectives of this study were to quantify the including three sites each of: (1) tall mangroves; (2)
ecosystem carbon stocks of the largest mangrove medium mangroves; (3) low mangroves; and (4)
ecosystem complex in the Dominican Republic and abandoned shrimp ponds. Within each site, ecosystem
identify baseline C stocks in a manner that would C stocks (above- and belowground) were measured
function as part of a country-level REDDþ program. To following methods outlined by Kauffman and Donato
accomplish these objectives, we (1) identified the current (2012). At each tall and medium mangrove site, six 7-m
and converted mangrove extent of the region, (2) fixed-radius plots were established 25 m apart along a
stratified mangroves based upon geomorphology, struc- 125-m transect established in a perpendicular direction
ture, and land use, (3) measured the total ecosystem C from the marine ecotone. In the low mangroves, 2 m
stocks within each type, and (4) scaled the ecosystem C fixed-radius plots were established in a similar manner
stocks to the landscape scale to estimate total stocks and but at 10-m intervals. Only five plots were sampled in
losses due to conversion. Our hypotheses were that C the shrimp ponds that were covered by bare ground or a
stocks would differ among mangrove types and land sparse cover of halophytic vegetation. At each plot we
uses, and forest structure would be indicative of collected data necessary to calculate total C stocks
ecosystem pools (i.e., the largest C stocks would be derived from standing tree biomass, downed wood (dead
found in the tall forested mangroves). Further, we wood on forest floor) and soils to the depths of the
predicted that C losses from land-cover change would be marine sands.
high relative to land use in upland ecosystems because of
large losses of both aboveground and belowground Biomass of trees and shrubs
pools. Four species of mangroves occur throughout the
METHODS Caribbean: Rhizophora mangle L. (Rhizophoraceae),
Avicennia germinans (L.) Stearn (Avicenniaceae), La-
Study site guncularia racemosa (L.) Gaertn., and Conocarpus
The study area is located in the northwestern erectus L. (Combretaceae). We encountered the first
Dominican Republic, following the coastline of the three in the plots that we sampled; C erectus was
province of Montecristi from ;3 km east of the Haiti observed on the upland edges of the mangroves.
border to ;15 km northeast of the town of Montecristi Composition, tree density, and basal area in tall and
(Fig. 1). The area is within and surrounding the medium mangroves were quantified through measure-
Montecristi National Park. The coastal wetlands of the ments of the species and diameter at 1.3 m height (dbh)
area are flooded by a mixture of seawater from tidal of all trees rooted within each plot of each transect. Plot
fluxes and water from rivers that flow into the wetlands. size for tree measurements in the tall and medium
Mangroves in Montecristi were separated into structural mangroves was 154 m2 (7 m radius) for trees .5 cm dbh
 Ecological Applications
520 J. BOONE KAUFFMAN ET AL.
Vol. 24, No. 3

and a nested plot with a radius of 2 m for trees with a the end of the transect (12 m in total). Large downed
dbh of ,5 cm. The diameter of R. mangle trees was wood was separated in two decay categories: sound and
measured above the highest prop root. For the rotten. Wood debris was considered rotten if it visually
mangrove plants that were ,2 m in height in the low appeared decomposed and broke apart when kicked.
mangroves, the diameter of the main branch was We used data of specific gravity of downed wood
measured at 30 cm from the ground (D30). Additionally, determined for Caribbean mangroves of the Yucatan,
in the low mangroves height for each individual was Mexico, and reported by Adame et al. (2013). Using the
measured. specific gravity for each group of wood debris, biomass
Allometric equations were used to calculate tree was calculated using formulas reported in Kauffman et
biomass for each site (Appendix A). In the tall and al. (2011) Downed wood was converted to C using a
medium mangroves we used species specific formulas factor of 0.50 (Kauffman and Donato 2012).
from French Guinea provided in Fromard et al. (1998).
Aboveground biomass of individuals in the low man- Soil carbon and nutrients
groves, was determined using equations developed in At each plot, soil samples were collected for bulk
Puerto Rico by Cintron and Shaeffer-Novelli (1984) that density and nutrient concentration using a peat auger
used mainstem diameter and tree height. Belowground consisting of a semi-cylindrical chamber of 6.4 cm
root biomass for mangrove trees was calculated using radius. This auger is efficient for collecting relatively
the formula by Komiyama et al. (2005). Tree C was undisturbed cores from wet soils under mangroves
calculated from biomass by multiplying by a factor of (Donato et al. 2011). The core was systematically
0.48 for aboveground and 0.39 for belowground divided into depth intervals of 0–15 cm, 15–30 cm, 30–
biomass (Kauffman and Donato 2012). 50 cm, 50–100 cm and .100 cm (if parent materials
Standing dead trees were included in aboveground were not encountered before 100 cm depth). Samples of
biomass calculations. For each dead tree, the stem a known volume were collected in the field, dried to
diameter was measured and assigned to one of three constant mass, and then weighed to determine bulk
decay classes: Status 1, dead trees without leaves; Status density. The depth to the parent materials (marine
2, dead trees without secondary branches; and Status 3, sediments/sands) was measured at three locations near
dead trees without primary or secondary branches the center of each plot using a graduated aluminum
(Kauffman and Donato 2012). The biomass for each probe. The inference limit of the study was 2.5 m for
tree status was calculated using allometric equations of depths exceeding the probe length.
plant components. For dead trees of Status 1, biomass In the laboratory (Oregon State University Central
was calculated as the total dry biomass minus the Analytical Laboratory), the concentrations of C and N
biomass of leaves. The biomass of trees of Status 2 was were determined using the dry combustion method
calculated for R. mangle as the sum of stem, branches (induction furnace) with a LECO CNS-2000 Macro
and prop roots, and for L. racemosa and A. germinans as Analyzer (LECO Corporation, St. Joseph, Michigan,
the sum of stem and branches. Finally, the biomass of USA). Bulk density and carbon concentration were then
trees of Status 3 was calculated as the biomass of the combined with plot-specific soil depth measurements to
mainstem only. This also required estimation of the dead determine the soil C stocks.
tree height in this decay class. Standing dead trees in the Differences among biomass and carbon stocks among
low forests were very rare and, if present, were included vegetation types (tall, medium, and dwarf mangroves,
with live trees. and shrimp ponds) were tested with analysis of variance
(ANOVA). If the ANOVA was significant, a Tukey
Downed wood HSD test was performed to determine which means were
We used the planar intersect technique adapted for significantly different.
mangroves to calculate mass of dead and downed wood The study-area boundaries were delineated using
(Van Wagner 1968, Kauffman and Donato 2012). At aerial photo interpretation, extensive ground truthing
the center of each plot, four 14-m transects were and image analysis of Landsat imagery (USGS 2011).
established. The first was established in a direction that The wetlands of Montecristi included mangroves,
was offset 458 from the azimuth of the main transect. mudflats, open water, and areas that were converted
The other three were established 908 clockwise from the from mangroves to other land uses (e.g., salt and shrimp
first transect. At each transect, the diameter of any ponds). A rapid, supervised maximum-likelihood classi-
downed, dead woody material (fallen/detached twigs, fication was conducted in the GIS to develop a range of
branches, prop roots, or stems of trees and shrubs) image-band responses for tall, medium, and low
intersecting the transect was measured. Wood debris  mangroves, mud flats, converted areas, and open water
2.5 cm but , 7.5 cm in diameter (hereafter ‘‘small’’ (ESRI 2012). We calculated the total C stock of the
debris) at the point of intersection was measured along region using the areal extent of each cover type
the last 5 m of the transect. Wood debris  7.5 cm in multiplied by the mean C stock derived from the field
diameter (hereafter ‘‘large’’ debris) at the point of data. Changes in ecosystem C stocks (emissions
intersection was measured from the second meter to estimates) from land conversion were calculated as the
April 2014 MANGROVE CARBON STOCKS AND EMISSIONS 521

FIG. 1. Vegetation cover of the Montecristi wetland study area, northwestern Dominican Republic. The black star in the inset
map is the approximate location of the study area (19.7666 N, 71.6919 W, UTM). Exact coordinates or all sampled plots can be
found in Appendix B.

TABLE 1. Generalized descriptions of tall, medium, and low mangroves (mangle alta, media, and enano, respectively, in Spanish)
of the mangroves of Montecristi, Dominican Republic.

Feature
Canopy Canopy tree Canopy tree
Mangrove height diameter density Downed Geomorphic
type (m) (cm) (no./ha) wood position Salinity (ppt) Composition
Tall .10 .5 dense abundant fringing, riverine 25–26 Rhizophora,
(747–1039) edge Laguncularia,
Avicennia
Medium 3–10 most ,5 extremely dense scarce basin, interior zones 33–36 Rhizophora,
(32 000–72 000) Laguncularia
Low ,3 most ,2 extremely dense practically basin, tidal flats, 42–51 Rhizophora,
(13 000–92 000) absent interior zones Avicennia
 Ecological Applications
522 J. BOONE KAUFFMAN ET AL.
Vol. 24, No. 3

open water. Rizophora mangle was the most abundant

mangrove species at all sampled sites with Lagunularia
racemosa and Avicennia germinans present in lower
abundances. While composition did not vary greatly,
structure among the three mangrove types was dramat-
ically different (Fig. 2, Tables 1 and 2). The tall
mangroves exceeded 10 m while few individuals in low
mangroves exceeded 3 m in height. Viable propagules
were observed on individuals in all communities.
Density varied greatly among the mangrove commu-
nity types with a mean tree density of ;900 trees/ha in
the tall mangrove and .42 000 trees/ha in medium and
low mangrove (P ¼ 0.03; Table 2). R. mangle comprised
.99% of the individuals in the sampled low mangroves.
In the medium mangroves, R. mangle averaged 88% of
the live trees while L. racemosa comprised 12% of the
density, though because of its larger stature contributed
40% of the tree mass of a site. Similarly in the tall
mangrove, R. mangle consisted of 90% of the trees
measured and L. racemosa accounted for about 8% of
the trees measured. While the plots we sampled were
dominated by R. mangle, there were communities
apparent from both remote-sensing imagery and from
FIG. 2. Plant carbon stocks of tall, medium, and low observation in the field that were dominated by A.
mangroves of Montecristi, Dominican Republic, by sampling germinans (both tall and low mangrove). Both L.
site. Error bars show þSE of the total aboveground stock. racemosa and A. germinans tended to be more abundant
in areas away from the marine ecotone.
carbon stocks in intact mangroves minus the carbon Among all of the sampled stands, there was almost a
stocks of abandoned shrimp ponds. 40-fold difference in the total plant mass consisting of
live and dead materials and above- and belowground
RESULTS biomass. Total plant mass ranged from a mean of 14
Aboveground biomass and structure Mg/ha in the low mangrove at the Final site to 414 Mg/
ha in the tall mangrove at the Alta Tres site (Table 2).
Tall mangroves tended to occupy the land–water Live trees comprised the vast majority of this mass
ecotone, while medium and low mangroves were most especially in low and medium mangroves. Downed
common in the interior and upland edges away from wood and dead trees were a significant component only

TABLE 2. Density and mass of mangroves of the Montecristi National Park, Dominican Republic.

Mangrove biomass, by component (Mg/ha)

Plot sample Density Total
site (no./ha) Live tree Dead tree Dead wood aboveground
Low mangrove
El Morro 21 619 6 6 554 9.0 6 2.8 0 – 9.03
Granja 13 396 6 1 322 5.5 6 1.1 0 – 5.48
Final 92 178 6 15 196 21.1 6 10.3 0 – 21.12
Mean 42 397 6 25 003 11.9 6 4.7 060 – 11.9 6 4.7
Medium mangrove
Mediano Sur 71 692 6 14 157 62.8 6 11.5 0.1 6 0.1 3.24 6 1.63 66.14
Laguna Media 54 102 6 13 948 58.5 6 12.8 2.6 6 1.3 2.89 6 1.05 63.99
Guinea Media 32 264 6 8 271 41.4 6 7.3 7.2 6 6.6 2.82 6 1.33 51.42
Mean 52 686 6 11 404 54.2 6 6.5 3.3 6 2.1 3.0 6 0.1 60.5 6 4.6
Tall mangrove
Tapion 1 039 6 120 173.2 6 12.8 18.6 6 9.2 19.92 6 3.90 211.72
Balsa Alta 747 6 219 277.2 6 55.9 4.2 6 1.8 26.37 6 4.82 307.77
Alta Tres 909 6 442 271.7 6 39.2 20.8 6 11.7 12.89 6 2.80 305.39
Mean 898 6 85 240.7 6 33.8 14.5 6 5.2 19.7 6 3.9 275.0 6 31.6
Note: The data are plot means 6 SE.
Large wood debris was largely absent from the low mangrove and therefore not measured.
April 2014 MANGROVE CARBON STOCKS AND EMISSIONS 523

in the tall mangrove comprising about 12.5% of the total 26% in the medium mangroves and .13% for low and
aboveground biomass. tall mangroves (Appendix B). Relatively high carbon
In the low mangrove stands the ratio of the concentrations existed throughout the soil profile. For
aboveground to belowground biomass was 0.81. In example, in the Guinea Media and Medianao Sur
contrast, the aboveground : belowground ratio of the tall medium-mangrove sample sites the soils in the top
mangrove was 3.04 where mean total aboveground meter exceeded 20% C concentration. The deeper
biomass was 275 Mg/ha and belowground was 91 Mg/ horizons were substantially lower yet still relatively
ha (Table 2). high at ;9% C (Appendix B). Nitrogen concentration
Mean plant carbon stocks were 161, 47, and 10 Mg/ha tended to be highest in those samples with the highest
for tall, medium and low mangroves, respectively, and concentration of C. Bulk density was inversely propor-
were significantly different among the plant communi- tional to C concentration; samples with the lowest C
ties (P  0.01). The plant carbon of abandoned shrimp concentration had generally higher bulk densities and
ponds was negligible and not measured. In these sites, tended to be associated with the deepest samples. There
vegetation ranged from a sparse cover of regenerating A. were some apparent differences in the C concentration
germinans, to an open cover of salt-tolerant herbs, to among the mangrove types. For example, the surface
bare ground (salt flats). soil concentration of the low mangrove was always
,17% while those of the medium mangroves exceeded
Soils 25%. Throughout the soil profile, soil C concentration
Mean depths of the organic-rich soils to the marine of the low mangrove was lower at the same depths than
sands ranged from 175 to 230 cm for the mangroves the medium mangroves.
(Appendix B). In contrast the soil depth of abandoned The abandoned shrimp ponds were surrounded by
shrimp ponds was significantly lower, with a mean of 71 mangroves and were of a similar geomorphology.
cm. Variation was quite high among stands within the However, soils in abandoned shrimp ponds were quite
tall and low mangroves. In the tall mangroves the soil different than those of the mangroves (Appendix B).
Carbon concentrations of the surface soils of the shrimp
depth ranged from 120 to 211 cm. Similarly, in the low
ponds averaged an order of magnitude lower than those
mangroves soil depth ranged from 66 to at least 250 cm
of mangroves (2.0%; Appendix B). While the bulk
(the inference limit of the soil probe), which was the
density of mangrove soil surfaces rarely exceeded 0.30 g/
widest range in depths among the mangroves. There was
cm3, those of the shrimp ponds were .1.27 g/cm3 . Even
less variation in the medium mangroves, with the
the deepest of soil samples of most of the mangroves
deepest mean soil depths, and in the highly disturbed
were higher in C and N and lower in bulk density than
abandoned shrimp ponds with the shallowest soils
the surface samples of the shrimp ponds. While the soils
(Appendix B).
of the shrimp ponds were moist and some had standing
We found relatively high soil-carbon concentrations
water, tidal and freshwater inputs were limited by the
in mangroves above the marine sands. Carbon con-
presence of abandoned levees surrounding the ponds.
centration of the surface soils (0–15 cm) of the
The relatively high concentration of C throughout the
mangroves was quite high at all mangroves, exceeding profile of the mangroves resulted in high carbon stocks.
Carbon mass of the soils of the mangroves ranged from
TABLE 2. Extended. 262 Mg/ha in the Morro Bajo low-mangrove sample site
to 1136 Mg/ha in the Laguna Media medium mangrove
Mangrove biomass, by component (Mg/ha)
site. The mean soil C pools for the tall, medium, and low
Belowground, Belowground, Total mangroves were 546, 1084, and 713 Mg C/ha, respec-
live dead plant
tively. The relatively deeper soils coupled with higher C
concentration resulted in the higher C pools in medium
14.85 6 4.40 0 23.88 compared to tall mangroves (P  0.001). These
9.05 6 1.51 0 14.53
19.98 6 3.85 0 41.10 differences are likely due to the geomorphic conditions
14.6 6 3.2 060 26.5 6 7.8 of the sites, with tall mangroves being on the water edge
and likely younger and subject to greater disturbances
58.81 6 3.35 0.29 6 0.03 125.24 by storm surges and flooding. There was a high degree
44.34 6 6.19 1.66 6 0.94 109.99 of variation within the low mangroves with a range in
31.22 6 3.34 3.65 6 3.31 86.29 soil C pools from 262 to 919 Mg C/ha. Even at greater
44.8 6 8.0 1.9 6 1.0 107.2 6 11.3
depths for all of the sampled mangroves, the C density
(carbon mass per unit of soil) remained high due to the
72.34 6 8.09 10.37 6 4.28 294.43
100.28 6 18.93 4.04 6 2.02 412.09 increased bulk density.
99.11 6 10.31 9.11 6 3.30 413.61
90.6 6 9.1 7.8 6 1.9 373.4 6 39.5 Ecosystem carbon stocks
The mean total ecosystem carbon stock of the
Montecristi mangroves was 853 Mg/ha (Fig. 3). In
 Ecological Applications
524 J. BOONE KAUFFMAN ET AL.
Vol. 24, No. 3

the area (1678 ha) and comprised 32% of the ecosystem

C stock.
Converted lands were estimated to be 24% of the
vegetated area (1516 ha) but stored only 4% of the total
ecosystem C (144 768 Mg C). This contrasts with
medium mangroves, which covered about half as much
land but stored more than five times the mass of carbon.
Determining the loss of C from land-cover change is
difficult due to an inexact knowledge of the historic
cover and C stocks of areas under land use. Using the
difference of 683.9 Mg/ha from the weighted average of
all sampled mangroves and the mean of sampled
abandoned shrimp ponds, we calculated a conservative
estimate of 1 036 971 Mg C lost due to conversion of
mangrove to cultivated salt ponds, shrimp ponds, and
other uses. This loss translates to a potential decline
from ;4.9 million Mg C to the current 3.8 million Mg
C, or a loss of 21% of the potential ecosystem carbon
FIG. 3. Total ecosystem carbon stocks (mean þ SE) of stocks of the Montecristi wetlands since conversion.
mangroves and abandoned shrimp ponds of Montecristi,
Dominican Republic. Error bars are þSE of the total ecosystem DISCUSSION
stock (aboveground and belowground). Differences between Aboveground biomass
the ecosystem C stocks of shrimp ponds and all mangroves are
highly significant (P ¼ 0.001). There was a tremendous range and variability in the
aboveground forest structure and biomass of the
mangroves sampled in this study. Mangrove density
contrast, the mean total ecosystem C stock of the
ranged from 747 trees/ha in the Balsa Alta tall-mangrove
abandoned shrimp ponds was significantly lower at 95
site to .92 000 trees/ha in the Final low-mangrove site.
Mg/ha (P ¼ 0.001; Fig. 3). The C stocks of shrimp ponds The tree density of tall mangroves of this study (mean of
were only about ;11% of that of the mangroves. This 898 trees/ha) was similar to tree densities reported for
represents a potential loss of 661–1135 Mg C/ha when tall mangroves of the Samana Bay, on the northcentral
mangroves are converted to shrimp ponds. coast of the Dominican Republic (Sherman et al. 2003);
Forest stature was not necessarily a good indicator of they reported the aboveground biomass of 50-year old
ecosystem C stocks. Unlike our predictions, tall tall mangroves ranged from 124 to 384 Mg/ha with a
mangroves were lower in ecosystem C stocks (707 Mg/ mean of 234 Mg/ha. This is similar to the range of tall
ha) than the lower statured medium mangroves (1131 mangroves in our present study (212–305 Mg/ha; Table
Mg/ha; P , 0.01). Low mangroves were similar to 2). In the Samana Bay mangroves, Sherman et al. (2003)
ecosystem C stocks of the tall mangroves (733 Mg/ha). also found that aboveground biomass of the forests was
In all sites soil carbon comprised the overwhelming higher in the inland edge than on coastal fringe
majority of the ecosystem C stock, accounting for 77% suggesting that our estimates of C stocks for tall
of the total in tall mangrove and .96–99% in medium mangrove that were sampled only at the fringe would
and low mangroves (Fig. 3). be conservative or low for the stocks of tall mangroves
further inland.
Landscape-scale carbon stocks Forest structural differences among stands were also
The 9171-ha Montecristi wetland complex is a mosaic apparent by the tremendous range in aboveground
biomass of the mangroves of Montecristi. The total
of mangroves, areas converted for land use, open water,
aboveground biomass of different mangrove stands
and mud flats (Fig. 1). Approximately 6260 ha is
ranged from 5.5 to 305 Mg/ha in our present study. In
vegetated land, with a mangrove cover of 4743 ha
mangroves of Bocas del Toro, Panama, Lovelock et al.
(76%) and 1516 ha (24%) in converted lands. We
(2005) reported that mangrove biomass varied over 20-
estimated the total C stock of the Montecristi wetland is fold (8–194 Mg/ha). These mangroves were also
3 841 490 Mg C (excluding both the unsampled C in dominated by Rhizophora mangle and were similar in
mudflats and open water). Mangroves currently store an density and height ranges to those measured in our
estimated 97% of the carbon of this area (3 696 722 Mg present study.
C). Medium mangroves accounted for only 12% of the Few studies on ecosystem C stocks have been
land area (748 ha), but stored 22% of the total ecosystem conducted in the Caribbean. Adame et al. (2013)
C. Tall mangroves covered 37% of the area (2317 ha) reported ecosystem carbon stocks for Caribbean man-
and stored 43% of the total ecosystem C for the groves of Mexico to be 287 Mg/ha for low mangrove,
Montecristi wetland. Low mangroves covered 27% of 546 Mg/ha for medium mangrove, and 914 Mg/ha for
April 2014 MANGROVE CARBON STOCKS AND EMISSIONS 525

tall mangrove. Unlike the mangroves of this study, the DCLU ¼ DCAB þ DCBB þ DCDW þ DCSOC
tall mangroves in Mexico had the largest C stocks. The
where CLU means change in carbon stocks (or total C
low mangroves on karstic substrates in Mexico were
emissions or sequestration) due to land use; DCAB,
similar in structure and composition to the mangroves
change in aboveground biomass C pool; DCBB, change
of our present study but ecosystem carbon stocks were
in belowground biomass C pool; DCDW, change in dead
much lower due to shallower organic soils. Carbon
wood C pool; and DCSOC, change in soil organic carbon.
concentrations of the surface soil horizons were also
generally lower in the Mexican Caribbean mangroves. To facilitate comparison among most other assess-
Soil C in the low mangroves of Mexico ranged from 205 ments, ecosystem C losses are expressed as potential
to 362 Mg/ha. This is similar to the soil C pool of the CO2 emissions, or CO2 equivalents (CO2e)—obtained by
Granja Bajo site (262 Mg/ha; Appendix B) but much multiplying C stocks by 3.67, the molecular ratio of CO2
lower than the other low mangroves (845 Mg/ha). to C (Fig. 4).
We found that the C stocks of the abandoned shrimp Emissions factors arising from the conversion of
ponds were dramatically lower than that of the mangrove to shrimp ponds would vary by the type of
mangroves (P ¼ 0.001). Assuming that these sites were mangrove being converted. Our emissions estimates,
formerly mangrove, conversion resulted in large losses based upon our actual measurements, ranged from 2165
of C stocks. Vegetation cover of the abandoned sites Mg CO2e/ha in low mangrove to 3554 Mg CO2e/ha in
ranged from being devoid of cover to a sparse cover of medium mangrove. The emissions calculated as the
halophytic species or Avicennia germinans. The aban- difference between the mean C stock of mangroves and
doned levees surrounding the shrimp ponds still that of the abandoned shrimp ponds was 2637 Mg
eliminated tidal flow into the sites and prevented runoff CO2e/ha (Fig. 4). If the mean C stock of the mangroves
offsite. Soil C in abandoned shrimp ponds was lower is representative of the wetlands of Montecristi that have
throughout the soil profile. Losses of the carbon could been converted to salt or shrimp ponds, then the
be from the actual construction of the ponds that cumulative emissions from the 1516 ha that have been
resulted in loss of the aboveground vegetation and soils converted are 3.8 million Mg CO2e .
as well as from the loss of tidal influences and Elsewhere in the tropics the emissions from conver-
inundation altering the formerly suboxic state of the sion of upland tropical forests to cattle pasture are
soils to an oxidized state. comparatively lower (Fig. 4). Using the same approach-
The effects of shrimp pond conversion were apparent es as we did for mangroves, the emissions estimates from
throughout the soil profile. In their estimations of the tropical forest conversion to pastures was 538 Mg CO2e/
emissions arising from land use/land cover change in ha in evergreen forests of the Amazon and 230 Mg
mangroves, Pendleton et al. (2012) limited losses to C CO2e/ha from conversion of primary tropical dry forest
above the soil depth of 1 m. However, the actual field in Mexico (Kauffman et al. 2003, 2009). The emissions
measurements of the study clearly indicates that C losses estimates from forests assume little loss of C from soils
occurred at depths of .1 m on the soil horizon. This is (Kauffman et al. 2009). The emissions arising from
similar to observations in Malaysia by Ong (1993) who conversion of tropical evergreen and dry forest to
reported the losses at depths as great as 2 m upon pasture are about 20% and 9% of the predicted
conversion of mangroves to shrimp ponds could be as emissions arising from the conversion of mangrove to
high as 750 Mg C/ha. The losses predicted by Ong shrimp ponds, respectively. On a comparative basis, the
(1993) are similar to the losses reported here. emissions from 1 ha of mangrove converted to shrimp
pond is equivalent to the emissions of about 5 ha of
Emissions tropical evergreen conversion and 11.5 ha of tropical dry
Given the role of forests and wetlands as sources and forest.
sinks of C and their potential role in climate change Our estimates are based on empirical measurements of
mitigation strategies, it is necessary to predict emissions the C stocks of shrimp ponds and mangroves, and are
from land-use and land-cover change (GOFC-Gold substantially higher than recent estimates given by
2011). One common IPCC (Intergovernmental Panel Donato et al. (2011) and Pendleton et al. (2012).
on Climate Change) protocol for tracking changes in C Emissions estimates from Donato et al. (2011) are based
stocks and predicting emissions from land-cover change upon assumptions of a 50% loss of vegetation biomass
in forestry is the stock-change approach (IPCC 2003). and 25% loss of soil C from the top 30 cm to calculate
Using this approach, we calculated potential emissions their ‘‘low ‘‘estimates (411 Mg CO2e/ha), and a 100%
that occurred over the life of the abandoned shrimp loss of aboveground biomass, 75% loss of the below-
ponds in this study. We compared them with examples ground C in the top 30 cm, and 25% loss in deeper layers
of the emissions that arise from the conversion of to 1 m to calculate their ‘‘high’’ estimates (1439 Mg
tropical upland forests to cattle pasture (Kauffman et al. CO2e/ha). Similarly, Pendleton et al. (2012) provided an
2003, Guild et al. 2004). Differences in carbon stocks emissions estimate of 935 Mg CO2e/ha based upon the
were converted to emissions using the formula assumption that 63% of the C in vegetation and soils
 Ecological Applications
526 J. BOONE KAUFFMAN ET AL.
Vol. 24, No. 3

FIG. 4. Potential C emissions arising from the conversion of tropical ecosystems (Mg CO2e/ha; CO2 equivalents are obtained
by multiplying C emissions by 3.67, the molecular ratio of CO2 to C). Data for the Amazon evergreen rain forest in Brazil are from
Kauffman et al. (2009), and data for the Mexican tropical dry forest are from Kauffman et al. (2003). The estimated emissions
arising from conversion of mangroves are from Donato et al. (2011; two estimates based on different assumptions) and Pendleton
et al. (2012; one mid-point estimate). The emissions for low, medium, and tall mangroves and the mean of all mangroves combined
are from the actual measurements reported in this study.

above 1 m in depth would be lost due to mangrove mangroves provide to dependent coastal societies as well
conversion. as to natural ecosystems.
The estimates that we provide are based upon the
ACKNOWLEDGMENTS
actual measures of C stocks in shrimp ponds and
We thank Counterpart International and the Paul and
mangroves. These are over two-fold greater than the Maxine Frohring Foundation for funds for this study. We also
prediction by Pendleton et al. (2012) and are ;83% thank Tommy Betts, Jesse McCarter, Aristoteles Ponserrate,
greater than the high estimate of Donato et al. (2011). Edwin Fernandez, Diego Valentin Rivas, Gladys A. Rosado,
This is significant, as Donato et al. (2011) suggested that Leoncio de js Pimentel, Yira Rodriguez Jerez, Francisco Santos
Mella, Cesar Rafael Eva, Laura Perdomo, Eridania Hernandez,
worldwide mangrove deforestation generates emissions
Ramon Moran, Cesar A Cruz, Alexis Sakas, and Enrique
of 0.02–0.12 Pg C/yr—as much as 10% of the emissions Pugibet for assistance in the field. We also thank two
associated with forest conversion—despite accounting anonymous reviewers for their thorough and helpful reviews
for just 0.7% of the tropical forest area. Our emissions of earlier drafts. This study was conducted in close collabora-
estimates from mangroves that are similar to or even tion with CIBIMA of the Autonomous University of Santo
Domingo. We especially thank E. Pugibet for advice and inputs
lower in C stocks than regions undergoing extensive into this project.
mangrove conversion, suggest that the global estimates
of emissions provided by Donato et al. (2011) may be LITERATURE CITED
very conservative. Adame, M. F., J. B. Kauffman, I. Medina, J. N. Gamboa, O.
Mangroves have the relevant characteristics of deep Torres, J. Caamal, M. Reza, and J. A. Herrera-Silveira. 2013.
Carbon stocks of tropical coastal wetlands within the karstic
organic-rich soils with exceptionally high C storage. oligotrophic landscape of the Mexican Caribbean. PLoS
When converted, the C losses are high because large ONE 8(2):e56569.
quantities of C formerly stored in suboxic soils are lost, Alongi, D. M. 2002. Present state and future of the world’s
resulting in potentially large greenhouse-gas emissions. mangrove forests. Environmental. Conservation 29:331–349.
Cintrón, G., and Y. Schaeffer-Novelli. 1984. Methods for
Rates of deforestation of mangroves are among the studying mangrove structures. Pages 91–113 in S. C. Saenger
highest of all ecosystems in the tropics (Polidoro et al. and J. G. Snedaker, editors. The mangrove ecosystem:
2010). This is at the expense of a large number of research methods. Monographs in Oceanographic Method-
ecosystem services and the security of coastal inhabi- ology, 8. UNESCO Publishing, Paris, France.
Donato, D. C., J. B. Kauffman, D. Murdiyarso, S. Kurnianto,
tants. Clearly these factors suggest the value of inclusion M. Stidham, and M. Kanninen. 2011. Mangroves among the
of mangroves in climate-change mitigation activities that most carbon-rich forests in the tropics. Nature GeoScience
would maintain the carbon stocks as well as values that 4:293–297.
April 2014 MANGROVE CARBON STOCKS AND EMISSIONS 527

Duke, N. C., et al. 2007. A world without mangroves? Science abandonment in Neotropical landscapes. Ecological Appli-
317:41–42. cations 19:1211–1222.
Ellison, A. M., and E. J. Farnsworth. 1996. Anthropogenic Kauffman, J. B., M. D. Steele, D. L. Cummings, and V. J.
disturbance of Caribbean mangrove ecosystems: past im- Jaramillo. 2003. Biomass dynamics associated with defores-
pacts, present trends, and future. Biotropica 28:549–565. tation, fire, and, conversion to cattle pasture in a Mexican
ESRI [Environmental Systems Research Institute]. 2012. tropical dry forest. Forest Ecology and Management 176:1–
Cluster analysis & maximum likelihood classification. Arc- 12.
GIS Desktop: Release 10.1. ESRI, Redlands, California, Komiyama, A., S. Poungparn, and S. Kato. 2005. Common
USA. allometric equations for estimating the tree weight of
FAO [Food and Agriculture Organization of the United mangroves. Journal of Tropical Ecology 21:471–477.
Nations]. 2007. The world’s mangroves 1980–2005. FAO Lovelock, C. E., I. K. Feller, K. L. Mckee, and R. Thompson.
Forestry Paper 53. FAO, Rome, Italy. http://www.fao.org/ 2005. Variation in mangrove forest structure and sediment
docrep/010/a1427e/a1427e00.htm characteristics in Bocas del Toro, Panama. Caribbean
Fromard, F., H. Puig, E. Mougin, G. Marty, J. L. Betoulle, and Journal of Science 3:456–464.
L. Cadamuro. 1998. Structure, above-ground biomass and Murdiyarso, D., J. B. Kauffman, M. Warren, E. Pramova, and
dynamics of mangrove ecosystems: new data from French K. Hergoualc’h. 2012. Tropical wetlands for climate change
Guiana. Oecologia 115:39–53. adaptation and mitigation: science and policy imperatives
Giri, C. E., L. L. Ochieng, Z. Tieszen. A. Zhu, T. Singh, J. with special reference to Indonesia. CIFOR Working Paper
Loveland, J. Masek, and N. Duke. 2011. Status and 91. Center for International Forest Research, Bogor,
distribution of mangrove forests of the world using earth Indonesia.
observation satellite data. Global. Ecology and Biogeogra- Murray, M. R., S. A. Zisman P. A, Furley, D. M. Munro J.
phy 20:154–159. Gibson, J, Ratter, S. Bridgewater, C. D. Mity, and C. J.
GOFC-GOLD [Global Observation of Forest and Land Cover
Place. 2003. The mangroves of Belize. Part 1: distribution,
Dynamics]. 2011. A sourcebook of methods and procedures
composition and classification. Forest Ecology and Manage-
for monitoring and reporting anthropogenic greenhouse gas
ment 174:265–279.
emissions and removals caused by deforestation, gains and
Ong, J. E. 1993. Mangroves—a carbon source and sink.
losses of carbon stocks in forests remaining forests, and
Chemosphere 27:1097–1107.
forestation. GOFC-GOLD Report version COP17-1.
GOFC-GOLD Project Office, Natural Resources Canada, Pendleton, L., et al. 2012. Estimating global ‘‘blue carbon’’
Alberta, Canada. emissions from conversion and degradation of vegetated
Guild, L. S., J. B. Kauffman, W. B. Cohen, C. A. Hlavka, and coastal ecosystems. PLoS ONE 7(9):e43542.
D. Ward. 2004. Modeling biomass burning emissions from Polidoro, B. A., K. E. Carpenter, L. Collins, N. C. Duke, A. M.
Amazon forest and pastures in Rondonia, Brazil. Ecological Ellison, et al. 2010. The loss of species: Mangrove extinction
Applications 14:232–236. risk and geographic areas of global concern. PLoS ONE
IPCC [Intergovernmental Panel on Climate Change]. 2003. 5(4):e10095.
Good practice guidance for land use, land-use change, and Sherman, R. E., T. J. Fahey, and P. Martinez. 2003. Spatial
forestry. J. Penman, et al., editors. Institute for Global patterns of biomass and aboveground net primary produc-
Environmental Strategies, Kanagawa, Japan. tivity in a mangrove ecosystem in the Dominican Republic.
Kauffman, J. B., and D. C. Donato. 2012. Protocols for the Ecosystems 6:384–398.
measurement, monitoring, and reporting of structure, bio- Spalding, M. D., M. Kainuma, and L. Collins. 2010. World
mass and carbon stocks in mangrove forests. CIFOR atlas of mangroves. Earthscan, London, UK.
Working Paper 86. Center for International Forest Research, USGS [United States Geological Survey]. 2011. Landsat
Bogor, Indonesia. imagery record package LT500804620100122CHM00. http://
Kauffman, J. B., C. Heider, T. Cole, K. A. Dwire, and D. C. earthexplorer.usgs.gov/metadata/3119/LT50080462010022CHM00/
Donato. 2011. Ecosystem C pools of Micronesian mangrove Valiela, I., J. L. Bowen, and J. K. York. 2001. Mangrove
forests: implications of land use and climate change. forests: one of the world’s threatened major tropical
Wetlands 31:343–352. environments. BioScience 51:807.
Kauffman, J. B., R. F. Hughes, and C. Heider. 2009. Dynamics Van Wagner, C. E. 1968. The line intersect method in forest fuel
of C and nutrient pools associated with land conversion and sampling. Forest Science 14:20–26.

SUPPLEMENTAL MATERIAL
Appendix A
Equations used to determine biomass of mangrove trees and downed wood at Montecristi National Park (Ecological Archives
A024-030-A1).

Appendix B
Plot locations (latitude and longitude of the first plot of each site), soil depth, soil carbon and nitrogen concentration, bulk
density and the soil C and N stocks of mangroves and abandoned shrimp ponds in and around Montecristi National Park
(Ecological Archives A024-030-A2).