Science of the Total Environment 706 (2020) 135416

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

The impacts of degradation, deforestation and restoration on mangrove ecosystem carbon stocks across Cambodia ⇑ Sahadev Sharma a,1, Richard A. MacKenzie b, , Thida Tieng c, Kim Soben d, Natcha Tulyasuwan e, Amomwan Resanond e, Geoffrey Blate f, Creighton M. Litton a a Department of Natural Resources and Environmental Management, University of Hawaii at Manoa, 1910 East-West Rd., Honolulu, HI, USA b USDA Forest Service, Institute of Pacific Islands Forestry, 60 Nowelo St., Hilo, HI, USA c Asian Institute of Technology, Klong Luang, Pathumthani 12120, Bangkok, Thailand d Royal University of Agriculture, Khan Dangkor, Phnom Penh, Cambodia e USAID LEAD RDMA Program, 87 Wireless Road, Bangkok, Thailand f USAID, 1300 Pennsylvania Ave. NW, Washington, DC, USA highlights graphical abstract

 Deforestation reduced total Our results concluded, that deforestation and degradation results in significant losses of TEC stocks from ecosystem carbon (TEC) stocks of mangroves. While the prevention of deforestation and degradation is the most effective strategy for cli- intact mangroves by 60%. mate change mitigation and adaptation, it appears that restoration results in mangroves that can con-  TEC stocks from degraded mangroves tinue to combat climate change after 25–30 years. forests did not differ from intact forests.  TEC stocks from 25-year-old restored mangroves were similar to intact mangroves.  A gridded sampling approach effectively captured TEC variability across an entire country.

article info abstract

Article history: Mangrove forest conservation can help reduce global C emissions. Despite this benefit to climate change Received 1 September 2019 mitigation and adaptation, mangrove forests are being deforested or degraded at an alarming rate, Received in revised form 30 October 2019 though restoration efforts may offset these losses. The impacts of deforestation to C stocks are relatively Accepted 5 November 2019 intuitive and result in significant decreases in C stocks. It remains unclear how degradation from selective Available online 23 November 2019 harvesting of trees affects C stocks or how effective restoration efforts are at restoring C stocks. Editor: Jose Julio Ortega-Calvo Furthermore, total ecosystem C (TEC) stocks of pristine mangroves can significantly vary spatially. To address these issues, we conducted an intensive, national assessment of mangrove forests across Cambodia using a grid approach to: 1) examine how land use land cover (i.e., pristine, deforested, Keywords: Blue carbon degraded, and restored forests) impacts TEC stocks, and 2) how TEC stocks vary spatially across the coun- Carbon inventory try. TEC stocks from deforested mangroves were always lower than pristine forests, resulting in an overall À1 Rhizophora plantation loss of 60% C (480 Mg C ha ). However, TEC stocks from degraded and 25-year-old restored mangroves À Climate change mitigation forests did not differ from pristine forests. Mean TEC in mangroves was 784.7 ± 30.1 Mg C ha 1,

⇑ Corresponding author. E-mail addresses: [email protected] (S. Sharma), [email protected] (R.A. MacKenzie), [email protected] (T. Tieng), [email protected] (K. Soben), [email protected] (N. Tulyasuwan), [email protected] (A. Resanond), [email protected] (G. Blate), [email protected] (C.M. Litton). 1 Present address: Institute of Ocean and Earth Sciences, University of Malaya, C308 Institute of Postgraduate Studies Building, Kuala Lumpur, Malaysia. https://doi.org/10.1016/j.scitotenv.2019.135416 0048-9697/Ó 2019 Published by Elsevier B.V. 2 S. Sharma et al. / Science of the Total Environment 706 (2020) 135416

Gridded sampling decreasing from 957.2 ± 32.8 Mg C haÀ1 in the northern region to 628.9 ± 33.1 Mg C haÀ1 in the central region to 386.2 ± 19.1 Mg C haÀ1 in the southern region of Cambodia. Intensive sampling in mangroves across Cambodia verified impacts of deforestation reported elsewhere, revealed the lack of degradation impacts on TEC stocks, and demonstrated the effectiveness of restoration on TEC stocks after only 25 years. Our gridded sampling approach was able to capture spatial variability across Cambodia and pro- vide a more realistic TEC stock information that can be used for national reporting or participation in C markets. Ó 2019 Published by Elsevier B.V.

1. Introduction As the number of TEC stock assessments have increased over the past decade, it has become clear that the C storage can vary The large carbon (C) stocks in mangroves forests resulting from substantially across global regions and within countries. For exam- the removal and storage of C from atmospheric CO2 have been well ple, across Southeast Asian mangroves, TEC varied from 442 to documented (i.e., Donato et al., 2011; Murdiyarso et al., 2015). 1267 Mg C haÀ1 (Murdiyarso et al., 2015) and from 154 to While this highlights how mangrove conservation can promote cli- 1484 Mg C haÀ1 in West-Central Africa (Kauffman and Bhomia, mate change mitigation and adaptation, deforestation and degra- 2017). Within Indonesia, mangrove TEC stocks varied from 593 dation of mangroves continues to be a major threat to their to 1397 Mg C haÀ1 (Murdiyarso et al., 2015). The wide range of existence. Deforestation, a land use and land cover (LULC) defined TEC stocks reported across global regions is due in part to esti- here as the complete removal of a mangrove forest or stand of mates from most countries being restricted to a very small sample mangrove trees where the land is thereafter converted to a non- size consisting of only a few study plots or transects. Using an ade- forest use, has occurred globally at a rate of 1 to 2% per year. Loss quate number of plots to accurately estimate TEC stocks coupled of mangrove forests has largely been due to conversion of man- with a random and intensive grid approach would cover the grove forests to shrimp and fish aquaculture, rice, oil palm planta- entirety of existing mangrove ecosystems and effectively capture tions, and urban development. In Southeast Asia, where 60% of the spatial variability of TEC stocks at a country level. A grid approach world’s mangroves are found (FAO, 1994; Spalding, 2010), more for national assessments would also allow the development of than 114,000 ha of mangrove forests (2.5%) have been converted higher tier level emission factors that would increase the accuracy to aquaculture ponds, rice or oil palm from 2000 to 2012 and credibility of national reporting and the capacity of participa- (Richards and Friess, 2016). Conversion of mangroves to shrimp tion in voluntary C markets. ponds in Indonesia, Central America, and the Caribbean has The primary goal of this study was to conduct an intensive resulted in C stocks 2–8 times lower than adjacent pristine man- country-wide assessment of TEC stocks in mangrove forests across grove forests (i.e., C loss of 554 ± 230 Mg C haÀ1 assuming that Cambodia to address the following objectives: (1) compare TEC baseline conditions of deforested sites were similar to pristine ones stocks across various mangrove LULC types (pristine, deforested, (Kauffman et al., 2017a)). Similar impacts are expected to occur degraded, and restored); and (2) examine regional patterns in across SE Asia with mangrove deforestation, yet few studies have mangrove TEC stocks across Cambodia. We hypothesized that: quantified this despite the large areas of mangroves there, the high (H1) TEC stocks would be highest in pristine mangroves because rates of deforestation, and the potential loss of large amounts of of C loss from timber extraction and soil degradation in deforested terrestrial C. and degraded plots; (H2) TEC stocks would be greater in degraded Forest degradation is a LULC characterized by a change in the plots than deforested plots due to the forest structure that still original forest condition through disturbances such as illegal cut- remains in degraded plots after the harvest of a few select trees ting for fuelwood, burning, or human structures that compact soils versus complete clear cutting of deforested plots; and (H3) or alter hydrology. Forest degradation can also be caused by natu- restored mangrove plots would have lower TEC than pristine man- ral phenomena such as cyclones, diseases, die back, tsunami etc. grove plots due to the young age and development of the above This leads to a reduction in capacity of the forest to provide goods and belowground forest biomass. or services but does not result in a land cover change (i.e., conver- sion to non-forest land) (Bahamondez and Thompson, 2016; Krauss et al., 2010; Simula, 2009). As a result, degraded sites, 2. Materials and methods unlike deforested sites, still support some forest structure and thus land cover that can be difficult to quantify from remote sensing. 2.1. Study area This suggests that while degraded sites may have lower TEC stocks compared to pristine forests, they will likely be greater than defor- Cambodia lies within the tropics between 10° and 15°N latitude ested mangroves. Furthermore, if degradation is not included as a and 102° and 108°E longitude, with 435 km of coast (Fig. 1)(Rizvi LULC, field or remote sensing assessments may overestimate TEC and Singer, 2011). While mangrove forests are found along the stocks of mangrove forests. entire coastline of Cambodia, we focused our study on the Koh To offset losses from deforestation as well as to reduce C emis- Kong and Preah Sihanouk provinces as more than 90% of Cambo- sions, several countries and non-governmental agencies have dia’s mangroves are found there (FAO, 1994). Within Koh Kong invested heavily in projects to restore mangroves by planting province, we sampled mangroves from Peam Krasoap Wildlife monospecific stands of Rhizophora sp. (Wylie et al., 2016). While Sanctuary (PKWS) that represented northern Cambodia and Srae many of these restoration projects have failed for various reasons Ambel (SA) that represented central Cambodia (Fig. 1). In Preah (Lewis and Gilmore, 2007; Primavera and Esteban, 2008), it is Sihnouk, we sampled mangroves from Prey Nob (PN) that repre- unclear if the successful mono-specific plantations store similar sented southern Cambodia. levels of C as the once diverse pristine systems they replace and Since the 1990s, large areas of mangroves have been deforested thus how effective these projects are at restoring C stocks of in Cambodia for shrimp ponds, salt pans, charcoal production mangroves. (Spalding, 2010) and urbanization and resort development. In S. Sharma et al. / Science of the Total Environment 706 (2020) 135416 3

1993, 23,750 ha of mangroves within Koh Kong were designated as dead trees that only had the main trunk remaining. Biomass of the Peam Krasoap Wildlife Sanctuary (PKWS), a complex habitat Class I dead trees was estimated to be 97.5% of a live tree, Class system of estuaries and islands. Village level management commit- II 80% of a live tree, and Class III 50% of a live tree (Kauffman and tees were also formed in the northern Koh Kong province and Donato, 2012). When available, species-specific regional allometric around Ream Krasop Park along the southwest coast to protect equations were used to convert live and dead tree dbh measure- mangrove forests from illegal harvesting (Marschke and Nong, ments into aboveground biomass values for each species, other- 2003). This also resulted in major restoration activities led by the wise a common mangrove allometric equation was used community and that included planting approximately 3400 ha of (Table 2). Belowground root biomass was determined by using a deforested mangroves with Rhizophora apiculata seedlings in general equation (Komiyama et al., 2008). Regional wood densities 1998 (Tieng et al., 2019). Despite these actions, harvesting of select (g cmÀ3) for each species was obtained from the wood density trees for timber and fuelwood continues by local communities and database (Zanne et al., 2009)(Table 2). Dry biomass of live and has resulted in degraded mangrove forests in many areas. dead trees was converted to C using conversion ratios of 50% C for aboveground biomass and 39% C for belowground biomass 2.2. Field sampling (Kauffman and Donato, 2012).

We sampled a total of 68 plots from pristine (n = 26), degraded 2.2.2. Downed woody debris carbon (n = 12), restored (n = 8) and deforested (n = 22) mangrove forests Downed woody debris was measured using the planar intercept across the northern, central, and southern regions of Cambodia technique (Brown, 1974). At each of the subplots, four 12-m long (Fig. 1). Restoration efforts have largely focused in the north. As a woody debris transects were established along the cardinal direc- result, restored plots could only be found and sampled from this tions. Diameters of large wood debris (7.6 cm diameter) that region. The location of each plot was determined by randomly lay- crossed each transect were measured and counted along the entire ing a sampling grid over a map of the Cambodia coastline. The grid 12-m length of each transect and differentiated as sound or rotten. consisted of 40 cells  48 cells, with each cell being 3 km2. A total Medium (2.5–7.6 cm diameter), small (0.6–2.5 cm diameter), and of 61 nodes (points where grid lines intersected) fell onto areas fine wood debris (<0.6 cm diameter) were counted in the 2–7, 7– classified as mangrove forest. Using a power analysis and prelimi- 10, and 10–12 m length of each transect, respectively. Conversion nary data collected from 8 plots in the northern region, we deter- of diameters and counts for each size class to biomass was then mined that 48 plots would need to be sampled to effectively done as described in Kauffman and Donato (2012) using average quantify the C stocks of Cambodia’s mangroves. This number wood density and quadratic mean diameter values from included 4 additional plots that could be thrown out in the event Murdiyarso et al. (2009). Woody debris biomass was then con- they were not located in mangroves or some other unforeseen cir- verted to C value using a conversion ratio of 50%. cumstance. Forty-eight nodes/plots were then randomly selected (Fig. 1), two of which were not sampled because they were located 2.2.3. Soil carbon stock in non-mangrove areas. Each sample plot was a 0.25 ha square plot Soil core samples were collected using a 5.5 cm diameter open- that consisted of five, clustered 7-m radius circular subplots. Each face peat gouge auger. Soil cores were divided into depth intervals plot was categorized as pristine, degraded, deforested, or restored of 0–15, 15–30, 30–50, 50–100, and greater than 100 cm. (Table 1). Pristine plots had no signs of human activity or habitat The greater than 100 cm typically represented the 100–200 inter- fragmentation, while degraded plots had some forest structure vals, although achieving this depth was not always possible. The but with obvious signs of tree harvesting (e.g., cut tree stumps, total depth of soil cores was recorded to represent the end point cut logs, slash). Deforested plots had no trees and had been con- of that interval and a 5-cm sub-section of soil was collected in verted to aquaculture ponds or salt pans. Restored plots had been the field from each of the five depth intervals. Sediment samples planted ~25 years ago with R. apiculata seedlings (Marschke and were returned to the lab, dried to a constant mass at 60 °C, and Nong, 2003). weighed to the nearest 0.1 g. Sediment samples (including fine Carbon stocks were determined for each subplot using the Sus- roots) were ground into a fine powder using a mortar and pestle tainable Wetland Adaptation and Mitigation Program (SWAMP) or a WileyTM mill and then sieved through a 2-mm mesh sieve to protocol (Kauffman and Donato, 2012). At each subplot, standing remove any large roots that were already estimated using the live and dead trees and downed wood (dead wood debris on the Komiyama belowground allometric equation discussed above. Bulk forest floor) were measured and soil cores collected to determine density was determined for each interval by dividing the total dry bulk density, C concentration, and, from these, soil C densities weight by the total sample volume (96 cm3). A homogenized sam- and stocks. There was no vegetation or dead wood at the defor- ple from each sediment interval was then analyzed for total C using ested sites (deforested), hence only soil cores were collected from a CostechTM elemental analyzer (Costech Analytical Technologies, those plots. Valencia, California) at the University of Hawaii at Hilo Analytical Lab. Soil C stock (Mg C haÀ1) was determined as the sum of the pro- 2.2.1. Above and belowground tree carbon duct of bulk density (g cmÀ3), %C concentration, and total depth All trees >5 cm in diameter at breast height (dbh) were identi- interval (cm) from each interval sampled. fied to species and dbh measured to the nearest 0.1 cm within the entire 7-m radius circular subplot. Trees < 5 cm dbh (e.g., saplings) 2.3. Data analysis were identified to species and dbh measured within a 2-m radius circular plot nested within the larger 7-m radius subplot. For trees Differences in aboveground carbon (AGC), and downed wood with prop roots (Rhizophora spp.), the point of measurement for debris (DWD), soil C, and TEC were compared among different determining dbh was 15 cm above the highest prop root that could mangrove LULC types (pristine, degraded, deforested) and Cambo- be safely measured. Standing dead trees were categorized into dian regions (northern, central, southern) using a two-way analy- classes I, II, or III based on the proportion of attached branches sis of variance (ANOVA), where fixed effects included LULC type, (Kauffman and Donato, 2012). Class I represented a recently dead region, and any interactions between LULC type and region. tree with the majority of its primary and secondary branches still Because restored sites were only present in the north, differences attached. Class II represented a dead tree with some primary in these variables could only be compared across LULC types branches still attached to the main trunk, and Class III represented (pristine, degraded, deforested, and restored) in the northern 4 S. Sharma et al. / Science of the Total Environment 706 (2020) 135416

Fig. 1. Study area and sampling regions across Cambodia: the northern region represented by maps 1 and 2, the central region represented by maps 3 and 4, and the southern region represented by map 5 with four land use land cover types; twenty-six pristine (I1 to I26), ten degraded (D1 to D10), ten restored (R1 to R10) and twenty-two deforested mangrove sites (P1 to P22). S. Sharma et al. / Science of the Total Environment 706 (2020) 135416 5

Table 1 Carbon stocks in the measured mangrove ecosystem components from all sampling sites and land use land cover (LULC) types in Cambodia. Mean (±SE) were compared across sites and LULC using a two-way analysis of variance. Different letters indicate a significant difference at the 5% level of significance using Tukey’s HSD.

Region LULC type Carbon stocks in various ecosystem compartments (Mg C haÀ1) Fraction of total ecosystem C (%) Trees Downed Root Soil Total ecosystem (tree + downed Trees Downed Root Soil wood wood + root + soil) wood Northern Pristine 57.3 ± 8.6 15.3 ± 3.6 43.7 ± 4.5 837.6 ± 46.9 953.9 ± 51.8 6.0 1.6 4.6 87.8 Cambodia Degraded 74.5 ± 22.8 19.0 ± 5.3 44.8 ± 9.0 856.8 ± 173.1 995.1 ± 190.8 7.5 1.9 4.5 86.1 Restored 62.6 ± 13.3 14.0 ± 3.8 34.0 ± 2.8 838.7 ± 68.2 949.4 ± 64.4 6.6 1.5 3.6 88.3 Mean + SE 51.3 ± 6.9 a 13.0 ± 2.3 a 35.4 ± 3.7 a 840.1 ± 31.9 a 957.2 ± 32.8 a 5.6 1.4 3.8 87.4 Deforested 0 0 0 610.0 ± 68.1 610.0 ± 68.1 0 0 0 100 Central Pristine 92.6 ± 13.3 11.9 ± 2.2 39.6 ± 5.7 475.2 ± 44.8 619.4 ± 60.5 15.0 1.9 6.4 76.7 Cambodia Degraded 80.8 ± 16.9 18.3 ± 2.8 54.5 ± 13.3 494.4 ± 75.1 648.1 ± 95.5 12.5 2.8 8.4 76.3 Mean + SE 44.3 ± 10.5 7.0 ± 1.7 a 22.3 ± 5.5 a 481.6 ± 30.9b 628.9 ± 33.1b 10.4 1.7 5.2 76.5 a Deforested 0 0 0 221.3 ± 44.2 221.3 ± 44.2 0 0 0 100 Southern Pristine 25.0 ± 5.6 16.7 ± 9.4 19.5 ± 1.2 349.4 ± 119.6 410.6 ± 103.4 6.1 4.1 4.8 85.1 Cambodia Degraded 39.3 ± 4.5 22.4 ± 12.9 30.5 ± 1.8 284.3 ± 16.7 376.5 ± 11.8 10.4 5.9 8.1 75.5 Mean + SE 20.5 ± 5.8b 12.1 ± 6.0 a 16.0 ± 4.3b 302.9 ± 17.1b 386.2 ± 19.1b 5.9 3.5 4.6 80.3 Deforested 0 0 0 296.5 ± 53.7 296.5 ± 53.7 0 0 0 100 Cambodia Pristine 65.6 ± 9.3 14.4 ± 2.4 40.6 ± 17.7 688.5 ± 49.9 809.2 ± 52.8 8.1 1.8 5.0 85.1 Degraded 61.9 ± 9.3 20.2 ± 5.3 42.1 ± 19.0 497.4 ± 81.1 621.7 ± 89.3 9.9 3.2 6.8 80.0 Deforested 0 0 0 326.7 ± 45.3 326.7 ± 45.3 0 0 0 100.0

Table 2 Allometric equations and wood density values used to determine mangrove tree biomass of the major mangrove species in Cambodia.

Species AGB/Trees (kg) Reference BGB/Root (kg) Reference Wood density* (g cmÀ3) Avicennia alba B = 0.1848 * D2.3524 Dharmawan and Siregar (2008) B = 1.28 * D1.17 Comley and Mcguinness 0.60 Avicennia marina (2005) .59 2.505 2 Bruguiera Bwood = 0.0754 * r * D Cole et al. (1999), Clough and B = 0.0188*D *(D/ Tamai et al. (1983) 0.71 1.4914 0.909 gymnorrhiza Bleaf = 0.0679 * D Scott (1989) (0.025D + 0.583)) .72

Bruguiera B=Bwood +Bleaf cylindrica Ceriops decandra B = 0.251 * r * D2.46 Komiyama et al. (2008) B = 0.199 * r0.899 *D2.22 Komiyama et al. (2008) 0.77 Ceriops tagal .77 Excoecaria B = 0.251 * r * D2.46 Komiyama et al. (2008) B = 0.199 * r0.899 *D2.22 Komiyama et al. (2008) 0.48 agaloccha Heritiera littoralis B = 0.251 * r * D2.46 Komiyama et al. (2008) B = 0.199 * r0.899 *D2.22 Komiyama et al. (2008) 0.87 Lumnitzera littorea B = 0.251 * r * D2.46 Komiyama et al. (2008) B = 0.199 * r0.899 *D2.22 Komiyama et al. (2008) 0.67 Lumnitzera 0.71 racemosa Rhizophora B = 0.043 * D2.63 Amira (2008) B = 0.199 * r0.899 *D2.22 Komiyama et al. (2008) 0.85 apiculata .82 Rhizophora mucronata 2.101 0.899 2.22 Sonneratia alba Bwood = 0.3814 * r * D Cole et al. (1999) B = 0.199 * r *D Komiyama et al. (2008) 0.51 (À1.1679+(1.4914* Sonneratia Bleaf =10 .37 obovata (LOG(D))))

B=Bwood +Bleaf Xylocarpus B = 0.1832 * D2.21 Tarlan (2008) B = 0.199 * r0.899 *D2.22 Komiyama et al. (2008) 0.53 granatum .59 Xylocarpus moluccensis

Note: AGB = Aboveground biomass (kg); BGB = Belowground biomass (kg); D = Diameter at breast height (cm); r = wood density (g cmÀ3) *Zanne et al. (2009). Global wood density database, available at http://hdl.handle.net/10255/dryad.235. region using a one-way ANOVA, where fixed effects included only 3. Results LULC type. AGC and DWD data did not meet assumptions of nor- mality and equal variance and were therefore log transformed 3.1. Impacts of LULC on TEC stocks and other mangrove C pools prior to analysis. Paired comparisons were then conducted using Tukey’s honestly significant difference (HSD) procedure to iden- Deforestation of mangroves resulted in a significant decrease in tify significantly different means between fixed effects. All statis- all C pools measured. Average TEC and soil C stocks were 50% lower tical analyses were conducted using SYSTAT 12.02 (2007. Systat in deforested plots compared to pristine plots (Tukey’s HSD, Software Incorporated, San Jose, California) with an alpha level p < 0.001 and p < 0.05, respectively). Average aboveground tree, of 0.05. belowground root, and downed wood were all 100% lower in defor- 6 S. Sharma et al. / Science of the Total Environment 706 (2020) 135416 ested sites than pristine sites (Tukey’s HSD, p < 0.001, p < 0.001, plots (Tukey’s HSD, p < 0.001 and p < 0.001, respectively). In the and p < 0.001, respectively; Table 1, Fig. 2a). northern region, TEC and soil C stocks from deforested mangrove Average TEC and soil C stocks from degraded mangrove plots plots were 40 and 30% lower, respectively, than pristine ones did not statistically differ from pristine plots (Tukey’s HSD, (Tukey’s HSD, p < 0.01 and p = 0.15, respectively). TEC stocks and p = 0.99 and p = 0.99, respectively), but were significantly and soil C from deforested and pristine mangrove plots were not signif- nearly 50% greater than deforested plots (Tukey’s HSD, p < 0.001 icantly different in the southern region (Tukey’s HSD, p = 1.0). and p < 0.05, respectively). Average aboveground tree, below- Average aboveground tree, belowground roots, and downed wood ground root, and downed wood were all similar between degraded were all significantly and 100% lower in deforested mangrove plots and pristine mangrove plots (Tukey’s HSD, p = 0.41, p = 0.10, and than pristine ones in all three regions (Tukey’s HSD, p < 0.001 for p = 0.41, respectively), but were significantly greater than defor- all C pools in all three regions; Table 1, Fig. 2a). ested plots (Tukey’s HSD, p < 0.001, p < 0.001, p < 0.001, respec- There were no significant differences when C pools were com- tively; Table 1, Fig. 2a). pared between degraded mangrove plots and pristine ones in each Within regions, deforested mangrove plots had significantly region. Degraded mangrove plots from the northern, central, and lower TEC and soil C stocks than pristine mangrove forests in the southern regions all had similar levels of TEC (Tukey’s HSD, northern and central regions, but not the southern region (Table 2). p = 1.0, p = 1.0, p = 1.0), soil C (Tukey’s HSD, p = 1.0, p = 1.0, TEC and soil C stocks from central region deforested mangrove p = 1.0), above ground tree C (Tukey’s HSD, p = 0.97, p = 1.0, plots were significantly and 50% lower than pristine mangrove p = 0.88), belowground root C (Tukey’s HSD, p = 1.0, p = 0.76, p = 0.78), and downed wood C (Tukey’s HSD, p = 0.95, p = 0.88, p = 1.0) as pristine plots (Table 1, Fig. 2a). Degraded mangrove plots from the northern and southern region had similar levels of TEC than deforested plots (Tukey’s HSD, p = 0.09, p = 0.9), but were significantly greater than defor- ested plots in the central region (Tukey’s HSD, p < 0.001). There were no significant differences in soil C among degraded and defor- ested plots of all three regions (Tukey’s HSD, p = 0.5, p = 1.0, p = 0.1). Degraded mangrove plots had significantly greater above ground tree C (Tukey’s HSD, p < 0.001, p < 0.001, p < 0.001), below- ground root C (Tukey’s HSD, p < 0.001, p < 0.001, p < 0.001), soil C (Tukey’s HSD, p = 1.0, p = 1.0, p = 1.0), and downed wood C (Tukey’s HSD, p < 0.001, p < 0.001, p < 0.001) than deforested plots in all three regions (Table 1, Fig. 2a). Comparisons between pristine and restored mangrove plots could only be made in the northern region as this is the only region where restoration has happened. Twenty-five year old restored

mangrove plots had similar levels of TEC (F1, 22 = 0.01, p = 0.96), soil C(F1, 22 = 0.0, p = 0.99), aboveground tree C (F1, 22 = 0.01, p = 0.68), belowground tree C (F1, 22 = 1.64, p = 0.21), and downed wood C stocks (F1, 22 = 0.01, p = 0.94) as pristine ones (Table 1).

3.2. Regional patterns in TEC stocks and other mangrove C pools

The mean TEC stock of mangrove plots in Cambodia was 784. 7 ± 30.1 Mg C haÀ1. Regionally, mean TEC stocks decreased from north to south, with stocks being more than 2 times greater in the northern region (957.2 ± 32.8 Mg C haÀ1) than in the central À1 (628.9 ± 33.1 Mg C ha; Tukey’s HSD, p < 0.001) or southern region (386.2 ± 19.1 Mg C haÀ1; Tukey’s HSD, p < 0.001; Fig. 2b, Table 1). TEC stocks were largely comprised of soil C stocks, which con- tributed to 80–90% of TEC stocks (Table 1). Total soil C stocks in the northern region (840.1 ± 31.9 Mg C haÀ1) were significantly and more than 2 times greater than the central (481.6 ± 30.9 Mg C h aÀ1; Tukey’s HSD, p < 0.001) or southern region (302.9 ± 17.1 Mg C h aÀ1; Tukey’s HSD, p < 0.001; Table 1). Aboveground tree, belowground root, and downed wood C stocks contributed much less to TEC stocks than soils, making up 6–10, 4–5, and 1–4%, respectively, of TEC in the northern, central, and southern regions (Table 1, Fig. 2b). Average aboveground tree and belowground root C stocks were significantly greater in the northern (Tukey’s HSD, p < 0.05 and p < 0.05, respectively) and cen- tral region (Tukey’s HSD, p < 0.001 and p < 0.05, respectively) than the southern region. Aboveground tree and belowground root C stocks were similar between the northern and central region (Tukey’s HSD, p = 0.17 and p = 0.99, respectively). Downed wood Fig. 2. Average (±SE) ecosystem C stocks of mangroves partitioned into dominant pools from (a) four land use land cover types and (b) three regions across Cambodia. did not significantly differ across regions (F2, 51 = 0.01, p = 0.99; The zero value on the y-axis partitions the above and belowground C pools. Table 1). S. Sharma et al. / Science of the Total Environment 706 (2020) 135416 7

4. Discussion immediately and significantly increased compared to pristine forested sites (Lang’at et al., 2014). Sediment CO2 fluxes also signif- Mangrove forests remove C from the atmosphere and store icantly decreased with age of mangrove clearing in Belize, with flux large amounts of C primarily in soils, providing an important rates from 20-year-old clearings being 3 times lower than one- mechanism to mitigate global climate change (Pendleton et al., year-old clearings (Lovelock et al., 2011). Soil TEC stocks are also 2012). As such, the number of mangrove C stock assessments have reduced in deforested sites through the leaching of dissolved inor- significantly increased as countries start to include mangrove C ganic carbon (DIC) and dissolved organic carbon (DOC) produced into their nationally determined contributions. The national from root decomposition (Grellier et al., 2017; Moore et al., assessment that we conducted across mangroves in Cambodia 2013). Increased soil temperatures from loss of tree canopy further revealed the impact of LULC on those C stocks and the need to exacerbates soil C losses as the resulting warmer and drier condi- quantify and include C stocks from those different LULC in national tions that can increase microbial activity, decomposition, and flux reports or participation in C credit markets. Our results also rates of gaseous and dissolved forms of C (Lang’at et al., 2014). revealed the importance of mangrove conservation, the potential Finally, lower soil C stocks can be due to a lack of forest structures value of sustainable harvesting, and the success that mangrove (e.g., roots, trunks) in deforested plots that play an important role plantations have at restoring C stocks. TEC stocks from pristine in trapping allochthonous sources of C from flooding tide or river- mangroves had 60% more C than deforested mangroves, but did ine waters (Furukawa and Wolanski, 1996; Furukawa et al., 1997; not significantly differ from degraded sites. Therefore, we were Krauss et al., 2014). only able to partially accept our 1st hypothesis that TEC stocks Using a stock change approach, we estimated that, on average, would be highest in pristine mangroves compared to deforested TEC loss from converting pristine mangrove forests to deforested À and degraded plots as this only held true for pristine and defor- LULC is 482.5 Mg C ha 1. This is equivalent to an emission of À1 ested plots. TEC stocks were also significantly greater in degraded 1770.8 Mg CO2eha from deforested mangroves in Cambodia. plots than deforested plots, supporting our 2nd hypothesis. TEC These deforestation emissions are similar to other studies that À1 stocks from 25-year-old restored plots did not significantly differ have reported emissions of 1900 ± 335 Mg CO2eha from Central from pristine mangrove forest plots and we rejected our 3rd America and Southeast Asia (Kauffman et al., 2017a) and À1 hypothesis that pristine mangroves would store higher TEC stocks 1786.5 Mg CO2eha from the Philippines (Castillo et al., 2017). than restored ones, at least in the case of studied sites in northern Cambodia. 4.2. Mangrove degradation

Degraded plots had obvious signs of cutting (i.e., cut trunks, 4.1. Mangrove deforestation logs, slash). Slash likely contributed to the greater amounts of downed woody debris observed in degraded plots compared to Nearly 20% of Cambodia’s mangroves have been converted to pristine ones. However, despite the removal of mangrove trees aquaculture ponds and salt pans (Tieng et al., 2019). Areas are first from selective harvesting, TEC stocks did not significantly differ clear cut and then the hydrology is manipulated through the cre- between degraded and pristine mangrove plots within the north- ation of small berms or walls around the perimeter of the pond ern, central, or southern regions. Patterns in tree densities, basal or pan. Assuming that pre-deforestation conditions were similar areas, and DBH values revealed that fewer but larger trees were to pristine plots, deforestation reduced TEC stocks of pristine Cam- generally present in degraded plots compared to pristine ones. This bodian mangroves in the northern and central regions of Cambodia would have resulted in the similar levels of aboveground tree C 40–50%, respectively. TEC stocks from deforested southern plots observed between degraded and pristine plots. Larger trees in did not significantly differ from pristine ones, which was likely degraded plots were likely due to the fact that smaller trees are due to a combination of low replication of pristine sites (n = 2) in generally harvested for house poles and charcoal, respectively (B. that region, high variability, and low C stocks in the sediments of Racy, Fisheries Administration, pers. communication). The harvest- pristine and deforested mangroves. Deforestation has similarly ing of a few trees from degraded plots may function similar to tree reduced TEC stocks of pristine mangroves forests by more than thinning, where thinning can increase tree growth rates (Alongi, 90% in India (Bhomia et al., 2016), 50–90% in Latin America, 50% 2008; Devoe and Cole, 1998; Duke, 2001) that would have resulted in Indonesia (Kauffman et al., 2017b; Kauffman et al., 2014), and in the bigger trees that were observed in the central and southern 50% in Malaysia (Sanderman et al., 2018). Differences in deforesta- regions. This might have also increased belowground C stocks of tion impacts to TEC across these sites are likely due to the different degraded plots such that they were similar to pristine plots. ages of the studied deforested sites, where older deforested sites The lack of differences in TEC between pristine and degraded are likely to have lower TEC than younger deforested sites. The plots suggests two possibilities. The first is that harvesting of man- reduction of TEC stocks in deforested mangroves from Cambodia grove trees in Cambodia is currently at a level that might be con- and these studies was partly due to the cutting and removal of sidered sustainable. Sustainable harvest of mangrove trees can trees that resulted in a 100% reduction of aboveground tree C. occur in systems with high net productivity, such as Matang Man- Lower TEC stocks in deforested plots are also likely influenced by grove Forest in Malaysia (Ewel et al., 1998) and Bintuni Bay in the fact that there are no trees to measure in deforested plots Indonesia (Sasmito et al., 2019; Sillanpää et al., 2017). Timber har- and estimates of belowground root biomass still present after trees vest from both of these mangroves occurs at much larger scales are removed cannot be made. Because belowground root biomass (~1000 ha clear cuts yrÀ1) than was observed in Cambodia and fur- typically represents <5% of TEC stocks, TEC stock differences ther suggests that the harvesting of only a few trees as opposed to between deforested and pristine are still significant, though they clear cutting entire areas may indeed be sustainable. This may be may be overestimated in some cases. especially true as selective harvesting would not impact external Lower TEC stocks in deforested mangrove plots are also due to drivers that drive mangrove productivity such as inputs of fresh- soil C losses that can occur through several mechanisms and can water, nutrients, or sediments. More research is needed to deter- increase over time. First, the cutting of trees initially results in mine what level of harvest is sustainable from these and other the death and decomposition of fine roots that releases C back to mangrove ecosystems and that don’t have negative impacts on the atmosphere through gas fluxes. After trees were initially cut other ecosystem services that mangrove forests provide (e.g., from a mangrove in Kenya, CO2 and CH4 fluxes from sediments storm protection, fish habitat) to coastal human populations. The 8 S. Sharma et al. / Science of the Total Environment 706 (2020) 135416 second possibility is that, while change from deforestation hap- grow in nutrient-poor conditions (Reef et al., 2010). Thus, pens rapidly, the impacts of degradation can be more gradual increased nitrogen and phosphorous availability are important abi- and take many years to decades before any impacts are observed otic factors that can influence productivity and TEC stocks (Adame (Foley et al., 2007, 2005). Thus, any negative impact to TEC stock et al., 2013). Mangrove tree basal area and canopy cover are biotic may still be observed in future sampling events. factors that have also been positively correlated to mangrove TEC values (Bukoski et al., 2017; Rahman et al., 2015). Species diversity 4.3. Mangrove restoration is another biotic factor that may also influence TEC values as it has been positively correlated with belowground C stocks (Atwood Large funded government or NGO projects have attempted to et al., 2017; MacKenzie et al., 2016). Better understanding of rela- offset mangrove losses over the last 30 years through the creation tionships between abiotic and biotic factors and variability in TEC of monospecific Rhizophora spp. plantations. These plantations are in mangrove forests will improve the prediction and modeling of intended to sequester C and reduce or offset national greenhouse TEC under current and future climate and LULC scenarios. gas emissions. In the northern region, we found that TEC stocks of 25-year-old Rhizophora apiculata restored mangroves plots had 5. Conclusion similar TEC stocks compared to pristine plots. Thirty-five and 20 year old R. apiculata restored mangrove plots in Vietnam had Our results support findings from other studies, that deforesta- similar levels of TEC stocks compared to pristine mangrove forests tion results in significant losses of TEC stocks from mangroves. (Nam et al., 2016a; Tue et al., 2014). TEC stock values calculated While the prevention of deforestation is the most effective strategy from DelVecchia et al. (2014) also revealed that both 10 year old for climate change mitigation and adaptation, it appears that restored shrimp ponds and 20 year old afforested sites had similar restoration results in mangroves that can continue to combat cli- values to pristine mangroves in Ecuador. While TEC stocks from mate change after 25–30 years. While it appears that monotypic restored sites < 10 years old were significantly lower than adjacent plantations were successful in Cambodia, this is not always the pristine mangrove forests (Bhomia et al., 2016; DelVecchia et al., case elsewhere in the world. Additional studies or tools are needed 2014; Duncan et al., 2016; Thant et al., 2012), soil C stocks of young that can compare different types of restoration techniques to iden- restored sites are often similar to pristine sites. This is likely due to tify where and what method of restoration will be most effective. refractory C still being present in the soils from the once well- More information is also needed that compares other ecosystem developed mangrove that grew there originally. Younger sites also services between restored and pristine mangroves. Finally, selec- appear to accumulate carbon faster in their roots zones than older tive harvesting of trees from mangroves in Cambodia did not ones. For example, reforested mangroves in Malaysia accumulated appear to significantly alter TEC stocks of degraded mangroves. À1 À1 9.5 Mg C ha yr during the first 10 years after reforestation. While this suggests there is a potential balance between mangrove À1 À1 After 10 years, C accumulation was reduced to 2.8 Mg C ha yr harvest and TEC stocks in Cambodia, caution should be used in (Sanderman et al., 2018). Greater surface elevation increases in interpreting these results as there is no globally accepted definition created mangroves <20 years old compared to older (>20 years) of degradation and our definition of degradation may not have reference sites were due in part to greater root growth at younger been as conservative as those definitions provided by FAO. Under- sites that resulted in a significant expansion of the root zone standing at what point harvesting of mangroves is unsustainable is (Krauss et al., 2017). Results from this and other studies indicate critically needed so that coastal communities can continue to uti- that mono-specific restoration plantations can reach similar levels lize mangrove forests as well as benefit from the many ecosystem ~ of TEC stocks after only 20–30 years. Thus, while protection of services that they provide. pristine forests still remains the best conservation strategy, restoration of deforested and degraded mangroves may still pro- Author Contributions vide an effective way to sequester atmospheric CO2.

4.4. Regional pattern of TEC across Cambodia SS, RAM, AS and GB conceived of and designed the study. SS, RAM, TT, KS, NT, AS and GB performed the research. SS, RAM, TT The mean TEC stocks of Cambodia’s mangrove forests was 784. and NT analyzed the data and performed statistical analyses. All 7 ± 30.1 Mg C haÀ1 that is within the range (355–1385 Mg C haÀ1) authors interpreted results and contributed to the MS, with writing reported in the mangrove literature (Adame et al., 2015; Adame led by SS, RAM and CML. et al., 2013; Donato et al., 2011; Hossain, 2014; Jones et al., 2014; Kauffman et al., 2011; Murdiyarso et al., 2015; Nam et al., Declaration of Competing Interest 2016b; Phang et al., 2015). However, our mean TEC estimates are 18% lower than the global mean TEC value for mangroves The authors declare that they have no known competing finan- (956 Mg C haÀ1;(Alongi, 2014), and 23% lower than the mean cial interests or personal relationships that could have appeared TEC value from the Indo-Pacific region (1023 ± 88 Mg C haÀ1; to influence the work reported in this paper. (Donato et al., 2011). Lower values from our study are likely due to the fact that we sampled soils to only 2 m. Many of the sites Acknowledgments sampled by Alongi and Donato et al. took soil cores to 3 m, which can significantly increase the reported TEC stocks. This study was conducted under the Sustainable Wetlands The decreasing TEC stocks we observed from the northern, to Adaptation and Mitigation Program (SWAMP), a collaborative central, to southern regions was likely due to variation in the abi- effort by the Center for International Forestry Research (CIFOR), otic and biotic factors controlling TEC within each region.. Rainfall, Oregon State University, the United States Forest Service (USFS), salinity, and nutrient availability are abiotic factors that can influ- and the University of Hawaii at Manoa, with financial support from ence TEC values through impacts on productivity. Higher rainfall the United States Agency for International Development (USAID). along decreasing latitudes lowers salinities in mangrove forests We are thankful to Royal Government of Cambodia, Ministry of which releases them from osmotic stress and thus increases their Environment, Ministry of Agriculture, Forestry and Fisheries, Royal productivity and TEC stocks (Adame et al., 2013; Bukoski et al., University of Phnom Penh, Royal University of Agriculture in Cam- 2017; Kauffman and Bhomia, 2017). Mangroves also generally bodia, USAID Cambodia, and USAID Low Emissions Asian Develop- S. Sharma et al. / Science of the Total Environment 706 (2020) 135416 9 ment (LEAD) program for providing logistic support and permis- Kauffman, B.J., Arifanti, V.B., Hernández Trejo, H., Carmen Jesús García, M., Norfolk, sion to study sites. We are also grateful to the team of hardworking J., Cifuentes, M., et al., 2017a. The jumbo carbon footprint of a shrimp: carbon losses from mangrove deforestation. Front. Ecol. Environ. 15, 183–188. and committed participants who enabled field data collection. Kauffman, J.B., Arifanti, V.B., Trejo, H.H., García, M.C.J., Norfolk, J., Cifuentes, M., et al., 2017b. The jumbo carbon footprint of a shrimp: carbon losses from mangrove deforestation. Front. Ecol. Environ. 15, 183–188. Kauffman, J.B., Bhomia, R.K., 2017. Ecosystem carbon stocks of mangroves across References broad environmental gradients in West-Central Africa: Global and regional comparisons. PLoS ONE 12, e0187749. Kauffman JB, Donato D. Protocols for the measurement, monitoring and reporting of Adame, M., Santini, N., Tovilla, C., Vázquez-Lule, A., Castro, L., Guevara, M., 2015. structure, biomass and carbon stocks in mangrove forests. Center for Carbon stocks and soil sequestration rates of tropical riverine wetlands. International Forestry Research (CIFOR), Bogor, Indonesia, 2012. Biogeosciences 12, 3805–3818. Kauffman, J.B., Heider, C., Cole, T.G., Dwire, K.A., Donato, D.C., 2011. Ecosystem Adame, M.F., Kauffman, J.B., Medina, I., Gamboa, J.N., Torres, O., Caamal, J.P., et al., carbon stocks of Micronesian mangrove forests. Wetlands 31, 343–352. 2013. Carbon stocks of tropical coastal wetlands within the karstic landscape of Kauffman, J.B., Heider, C., Norfolk, J., Payton, F., 2014. Carbon stocks of intact the Mexican Caribbean. PLoS ONE 8, e56569. mangroves and carbon emissions arising from their conversion in the Alongi, D.M., 2008. Mangrove forests: Resilience, protection from tsunamis, and Dominican Republic. Ecol. Appl. 24, 518–527. responses to global climate change. Estuarine, Coastal and Shelf Science 76, 1– Komiyama, A., Ong, J.E., Poungparn, S., 2008. Allometry, biomass, and productivity 13. of mangrove forests: a review. Aquat. Bot. 89, 128–137. Alongi, D.M., 2014. Carbon cycling and storage in mangrove forests. Annual Review Krauss, K.W., Cahoon, D.R., Allen, J.A., Ewel, K.C., Lynch, J.C., Cormier, N., 2010. of Marine Science 6, 195–219. Surface elevation change and susceptilibity of different mangrove zones to sea- Amira, S., 2008. Biomass Estimation of Rhizophora apiculata at Batu Ampar Kubu level rise on Pacific high islands of Micronesia. Ecosystems 13, 129– Raya Regency, West Borneo. Bogor Agriculture Institute. 143. Atwood, T.B., Connolly, R.M., Almahasheer, H., Carnell, P.E., Duarte, C.M., Lewis, C.J. Krauss, K.W., Cormier, N., Osland, M.J., Kirwan, M.L., Stagg, C.L., Nestlerode, J.A., E., et al., 2017. Global patterns in mangrove soil carbon stocks and losses. et al., 2017. Created mangrove wetlands store belowground carbon and surface Nature Climate Change 7, 523. elevation change enables them to adjust to sea-level rise. Sci. Rep. 7, 1030. Bahamondez, C., Thompson, I.D., 2016. Determining forest degradation, ecosystem Krauss, K.W., McKee, K.L., Lovelock, C.E., Cahoon, D.R., Saintilan, N., Reef, R., et al., state and resilience using a standard stand stocking measurement diagram: 2014. How mangrove forests adjust to rising sea level. New Phytol. 202, 19–34. theory into practice. Forestry 89, 290–300. Lang’at, J.K.S., Kairo, J.G., Mecuccini, M., Bouillon, S., Skov, M.W., Waldron, S., et al., Bhomia, R.K., MacKenzie, R.A., Murdiyarso, D., Sasmito, S.D., Purbopuspito, J., 2016. 2014. Rapid losses of surface elevation following tree girdling and cutting in Impacts of Land Use on Indian Mangrove Forest Carbon Stocks: Implications for tropical mangroves. PLoS ONE 9. Conservation and Management. Ecological Applications 26, 1396–1408. Lewis, R.R., Gilmore, R.G., 2007. Important considerations to achieve successful Brown JK. Handbook for inventorying downed woody material. Gen. Tech. Rep. INT- mangrove forest restoration with optimum fish habitat. Bull. Mar. Sci. 80, 823– 16. Ogden, UT: US Department of Agriculture, Forest Service, Intermountain 837. Forest and Range Experiment Station. 24 p. 1974; 16. Lovelock, C.E., Ruess, R.W., Feller, I.C., 2011. CO2 efflux from cleared mangrove peat. Bukoski, J.J., Broadhead, J.S., Donato, D.C., Murdiyarso, D., Gregoire, T.G., 2017. The PLoS ONE 6, e21279. use of mixed effects models for obtaining low-cost ecosystem carbon stock MacKenzie, R.A., Foulk, P.B., Klump, J.V., Weckerly, K., Purbospito, J., Murdiyarso, D., estimates in mangroves of the Asia-Pacific. PLoS ONE 12, e0169096. et al., 2016. Sedimentation and belowground carbon accumulation rates in Castillo, J.A.A., Apan, A.A., Maraseni, T.N., Salmo, S.G., 2017. Soil C quantities of mangrove forests that differ in diversity and land use: a tale of two mangroves. mangrove forests, their competing land uses, and their spatial distribution in Wetlands Ecol. Manage. 24, 245–261. the coast of Honda Bay, Philippines. Geoderma 293, 82–90. Marschke, M., Nong, K., 2003. Adaptive co-management: lessons from coastal Clough, B., Scott, K., 1989. Allometric relationships for estimating above-ground Cambodia. Can. J. Dev. Stud. 24, 369–383. biomass in six mangrove species. Forest Ecology and Management 27 (2), 117– Moore, S., Evans, C.D., Page, S.E., Garnett, M.H., Jones, T.G., Freeman, C., et al., 2013. 127. Deep instability of deforested tropical peatlands revealed by fluvial organic Cole, T.G., Ewel, K.C., Devoe, N.N., 1999. Structure of mangrove trees and forests in carbon fluxes. Nature 493, 660. Micronesia. Forest Ecology and Management 117 (1–3), 95–109. Murdiyarso D, Donato D, Kauffman JB, Kurnianto S, Stidham M, Kanninen M. Carbon DelVecchia, A.G., Bruno, J.F., Benninger, L., Alperin, M., Banerjee, O., de Dios, Morales storage in mangrove and peatland ecosystems: A preliminary account from J., 2014. Organic carbon inventories in natural and restored Ecuadorian plots in Indonesia. 2009. mangrove forests. PeerJ 2, e388. Murdiyarso, D., Purbopuspito, J., Kauffman, J.B., Warren, M.W., Sasmito, S.D., Devoe, N.N., Cole, T.G., 1998. Growth and yield in mangrove forests of the Federated Donato, D.C., et al., 2015. The potential of Indonesian mangrove forests for States of Micronesia. For. Ecol. Manage. 103, 33–48. global climate change mitigation. Nat. Clim. Change 5, 1089. Dharmawan, I.W.S., Siregar, C.A., 2008. Karbon tanah dan pendugaan karbon Nam, V.N., Sasmito, S.D., Murdiyarso, D., Purbopuspito, J., MacKenzie, R.A., 2016. tegakan Avicennia marina (Forsk.) Vierh. di Ciasem, Purwakarta. Jurnal Carbon stocks in artificially and naturally regenerated mangrove ecosystems in Penelitian Hutan dan Konservasi Alam 5 (4), 317–328. the Mekong Delta. Wetlands Ecol. Manage. 24, 231–244. Donato, D.C., Kauffman, J.B., Murdiyarso, D., Kurnianto, S., Stidham, M., Kanninen, Pendleton, L., Donato, D.C., Murray, B.C., Crooks, S., Jenkins, W.A., Sifleet, S., et al., M., 2011. Mangroves among the most carbon-rich forests in the tropics. Nat. 2012. Estimating global ‘‘blue carbon” emissions from conversion and Geosci. 4, 293. degradation of vegetated coastal ecosystems. PLoS ONE 7, e43542. Duke, N.C., 2001. Gap creation and regenerative processes driving diversity and Phang, V.X., Chou, L., Friess, D.A., 2015. Ecosystem carbon stocks across a tropical structure of mangrove ecosystems. Wetlands Ecol. Manage. 9, 267–279. intertidal habitat mosaic of mangrove forest, seagrass meadow, mudflat and Duncan, C., Primavera, J.H., Pettorelli, N., Thompson, J.R., Loma, R.J.A., Koldewey, H.J., sandbar. Earth Surf. Process. Landforms 40, 1387–1400. 2016. Rehabilitating mangrove ecosystem services: a case study on the relative Primavera, J.H., Esteban, J., 2008. A review of mangrove rehabilitation in the benefits of abandoned pond reversion from Panay Island, Philippines. Mar. Philippines: successes, failures and future prospects. Wetlands Ecol. Manage. Pollut. Bull. 109, 772–782. 16, 345–358. Ewel, K.C., Twilley, R.R., Ong, J.E., 1998. Different kinds of mangrove forests provide Rahman, M.M., Khan, M.N.I., Hoque, A.F., Ahmed, I., 2015. Carbon stock in the different goods and services. Global Ecol. Biogeogr. Lett. 7, 83–94. Sundarbans mangrove forest: spatial variations in vegetation types and salinity FAO U. UNEP: Land degradation in south Asia: Its severity, causes and effects upon zones. Wetlands Ecol. Manage. 23, 269–283. the people. Food and Agriculture Organization of the United Nations, United Reef, R., Ball, M.C., Feller, I.C., Lovelock, C.E., 2010. Relationships among RNA: DNA Nation Development Programme, United Nation Environmental Programme, ratio, growth and elemental stoichiometry in mangrove trees. Funct. Ecol. 24, Rome, Italy 1994. 1064–1072. Foley, J.A., Asner, G.P., Costa, M.H., Coe, M.T., DeFries, R., Gibbs, H.K., et al., 2007. Richards DR, Friess DA. Rates and drivers of mangrove deforestation in Southeast Amazonia revealed: forest degradation and loss of ecosystem goods and Asia, 2000–2012. Proceedings of the National Academy of Sciences 2016; 113: services in the Amazon Basin. Fronti. Ecol. Environ. 5, 25–32. pp. 344–349. Foley, J.A., DeFries, R., Asner, G.P., Barford, C., Bonan, G., Carpenter, S.R., et al., 2005. Rizvi A, Singer U. Cambodia coastal situation analysis. 2. by: IUCN, Gland, Global consequences of land use. Science 309, 570–574. Switzerland, 2011, pp. 3. Furukawa, K., Wolanski, E., 1996. Sedimentation in mangrove forests. Mangroves Sanderman, J., Hengl, T., Fiske, G., Solvik, K., Adame, M.F., Benson, L., et al., 2018. A and Salt Marshes 1, 3–10. global map of mangrove forest soil carbon at 30 m spatial resolution. Environ. Furukawa, K., Wolanski, E., Mueller, H., 1997. Currents and sediment transport in Res. Lett. 13, 055002. mangrove forests. Estuar. Coastal Shelf Sci. 44, 301–310. Sasmito, S.D., Taillardat, P., Clendenning, J.N., Cameron, C., Friess, D.A., Murdiyarso, Grellier, S., Janeau, J.-L., Hoai, N.D., Kim, C.N.T., Phuong, Q.L.T., Thu, T.P.T., et al., D., et al., 2019. Effect of land-use and land-cover change on mangrove blue 2017. Changes in soil characteristics and C dynamics after mangrove clearing carbon: a systematic review. Global Change Biol. (Vietnam). Sci. Total Environ. 593, 654–663. Sillanpää, M., Vantellingen, J., Friess, D.A., 2017. Vegetation regeneration in a Hossain, M., 2014. Carbon pools and fluxes in Bruguiera parviflora dominated sustainably harvested mangrove forest in West Papua, Indonesia. For. Ecol. naturally growing mangrove forest of Peninsular Malaysia. Wetlands Ecol. Manage. 390, 137–146. Manage. 22, 15–23. Simula M. Towards defining forest degradation: comparative analysis of existing Jones, T., Ratsimba, H., Ravaoarinorotsihoarana, L., Cripps, G., Bey, A., 2014. definitions. Forest Resources Assessment Working Paper 2009; 154. Ecological variability and carbon stock estimates of mangrove ecosystems in Spalding, M., 2010. In: World atlas of mangroves. Routledge. northwestern Madagascar. Forests 5, 177–205. 10 S. Sharma et al. / Science of the Total Environment 706 (2020) 135416

Tamai, S., Nakasuga, T., Tabuchi, R., Ogino, K., 1983. Ecological studies of mangrove Resolution Images and Google Earth Engine algorithm for entire Cambodia. forests in southern Thailand—standing structure and biomass. Mangrove IOP Conference Series: Earth and Environmental Science. 266. IOP Publishing, Ecology in Thailand 3–15. 2019, pp. 012010. Tarlan, M.A., 2008. Biomass estimation of nyirih (Xylocarpus granatum Koenig. 1784) Tue, N.T., Dung, L.V., Nhuan, M.T., Omori, K., 2014. Carbon storage of a tropical in primary mangrove forest in Batu Ampar, West Kalimantan. Bogor mangrove forest in Mui Ca Mau National Park, Vietnam. Catena 121, 119–126. Agricultural University. Wylie, L., Sutton-Grier, A.E., Moore, A., 2016. Keys to successful blue carbon Thant, Y.M., Kanzaki, M., Ohta, S., Than, M.M., 2012. Carbon sequestration by projects: lessons learned from global case studies. Mar. Policy 65, 76–84. mangrove plantations and a natural regeneration stand in the Ayeyarwady Zanne A, Lopez-Gonzalez G, Coomes D, Ilic J, Jansen S, Lewis S, et al. Global wood Delta, Myanmar. Tropics 21, 1–10. density database. 2009. Tieng T, Sharma S, Mackenzie RA, Venkattappa M, Sasaki N, Collin A. Mapping mangrove forest cover using Landsat-8 imagery, Sentinel-2, Very High