The study was conducted at an abandoned shrimp pond located at Khanom, Nakorn Sri Thammarat province, Thailand (9°17’ N, 99°49’ E) (Figure 1). According to the local people, the study area had been a mangrove plantation before the land was converted to a shrimp pond. After abandonment in the 1980s, sedimentation formed within the pond. Since a number of active shrimp ponds are still operating in the surrounding area (Figure 2), effluents from these ponds flowed via creeks to deposit in the study site. On the eastern coast of the southern Thai Peninsula, 18.5 km² of the coastal areas altered significantly between 1975 and 2006, with 17.3 km² of erosion and 1.2 km² of accretion. [23]. Besides the effluents from the surrounding active shrimp ponds, the study site is likely to receive sedimentation from offshore. Before land treatment, the area was covered with Suaeda maritima and Sesuvium portulacastrum (Figure 3), which are indicators of dry and saline soil, and with the pioneer mangrove species Avicennia marina (AM).

The size of the study site is 4.5 ha, being 150 m wide and 300 m long (Figure 4). The elevated land was excavated using heavy machinery in August 2001, resulting in reducing the elevation by 25 cm, which is 10 cm lower than the mean sea water level. This enabled the access of tidal water into the area, where water flow was previously prevented by the height of the elevated land. The increased frequency of water inundation brought by the excavation was expected to improve forestry productivity and land quality in the area. Half of the site was left unexcavated for a comparison with the excavated area. Ground level was measured using an auto-level (AFL 320; Pentax Industrial Instrument Co., Ltd., Tokyo, Japan) in December 2007. Water levels in the study site were recorded during the ground level measurement and inundation frequency was calculated by referring to the 2007 tidal records of the nearest tide recording station (Ko Prap, Surat Thani) to the study site (Figure 5).

We planted two mangrove species, Rhizophora mucronata (RM) and Bruguiera cylindrica (BC), in September 2001 at a spacing of 1.5 × 1.5 m, which is the standard spacing designated by the Thai government. RM is the species that local people prefer to plant, while BC is the species that grows naturally in areas adjacent to the study site. Seedlings with an average height of 50 cm for RM and 20 cm for BC were prepared for a half year prior to replanting at the nursery belonging to the Department of Marine and Coastal Resources located in Khanom.

RM was planted in five different blocks and BC in four blocks (Figure 4). Total plantation area for RM and BC were 2.5 and 2.0 ha, respectively. The number of RM planted was 4,488 for the excavated area and 6,732 for the unexcavated area while 4,488 BC seedlings were planted on each of the excavated area and unexcavated area. In total, 20,196 trees were planted during the course of the study.

After the mangroves were planted in September 2001, the tree height and survival rate were measured for approximately 17.6% of all planted trees on six occasions: February 2002, March 2003, November 2003, November 2004, October 2005 and August 2011, representing intervals from the planting of approximately 0.4, 1.5, 2.3, 3.3, 4.2 and 10 years, respectively. Since survival rate of planted trees is affected by various factors such as land treatment (excavation in this case), and species, key-factor/key-stage analysis was performed by MANOVA test (JMP 4.0, SAS Institute Inc.) in order to identify which stage and factor contributed most to the survival rate [24].

The diameter at breast height (DBH) was measured during the measurement period on two occasions (in October 2005 and in August 2011) to calculate the standing biomass of the planted mangroves. The biomass of RM was calculated using the equation developed by Meepol [25] while the biomasses of BC and AM were determined using the equation proposed by Pranchai [26]. These equations were originally proposed for the region where this study site is located. Thereby it was thought most appropriate to use them for the biomass calculation in this study.

Prior to land treatment, soil C content was measured from surface soils collected at three different places. To assess changes in soil C after the excavation, soils were sampled/analyzed from around 10 points of both the excavated area and the unexcavated area on four occasions: two occasions before excavation (August 2001) and at 2.3 years after excavation (July 2003) from bulk samples and another two occasions 4.1 years after planting (September 2005) and at 10 years after planting (August 2011) from soil core samples. Soil core samples were randomly collected from surface soils (0–5 cm) with 100 cc volumetric cylinders (DIK-5561; Daiki Rika Kogyo Co., Ltd., Kounosu, Japan). Total C was determined using the dry combustion method (NC-analyzer 1000; Sumigraph, Shimadzu Co., Ltd., Kyoto, Japan). Soil organic carbon (OC) was estimated using bulk density calculated from drying core samples.

Statistical analysis was carried out for total C using Tukey’s test (JMP 4.0, SAS Institute Inc., Cary, NC, USA) and differences at the p < 0.05 level were considered to be significant.

In general, survival rates remained largely invariant within the excavated area during the measurement period; however, the survival rate of BC was remarkably lower in the unexcavated area even from the early measurements (Figure 6). The survival rate of RM in the excavated area decreased slightly one year after planting (March 2003), but became constant thereafter. Survival rates of RM in the unexcavated area decreased gradually and dropped to 33% at the 10th year after planting (August 2011).

Key-factor/key-stage analysis showed that 4.2 years after planting (November 2004) was the most determinant stage for survival of the planted trees (Table 1). Land treatment (excavation in this study) influenced the survival rates as strongly as species, though a third contribution to the survival rate is still unknown. This result suggests that when planting mangroves in an abandoned area, attention should be paid not only to species selection but also to the treatment of land in order to attain favorable survival rates. Tree growth was highest for RM in the excavated area (Figure 7) with tree heights exceeding 500 cm after 10 years (August 2011). Although RM tree growth resembled that of BC in the early stages, significant differences were recorded by 4.2 years after planting (November 2004). In the unexcavated area, RM showed only slight tree growth and BC also grew slightly more with just over 100 cm growth in 10 years.

In accordance with different survival and growth rates, biomass accumulation differed markedly between the unexcavated and excavated area and between the RM and BC species (Table 2). Highest biomass was recorded for RM in the excavated area with 98.7 ton/ha, and the lowest value was for BC in the unexcavated area with 0.53 ton/ha. A decrease of the standing biomass in RM over time is due to its decreased survival rate and low growth rate.

A higher standing biomass, in the 6 years from October 2005 to August 2011, was recorded in the excavated area. The species RM and BC accumulated 5 times and 2.2 times more standing biomass in the excavated area than in the unexcavated area, respectively.

Aerial images of the study site were taken from a radio-controlled helicopter on two different occasions in September 2003 and February 2006 (Figure 8). Comparison of the two images explicitly demonstrated the opposite growth behavior in two different areas, e.g., vigorous growth in the excavated area but poor growth in the unexcavated area. The growth rate of AM, which colonized into the unexcavated area, was rather high in the first two years. However, numbers of AM trees and growing areas of AM were almost the same in September 2003 and February 2006. Moreover, the standing biomass of AM remained similar between October 2005 and August 2011 (Table 2). These facts indicate that the fast growing species AM can capture carbon quickly but can only continue its ability for a short period.

Without land treatment, the abandoned area will remain as unproductive land for a long time being dominated by AM and halophyte. To facilitate improved forest recovery of an abandoned shrimp pond and to enhance forest productivity, hydraulic restoration is inevitable.

Similar amounts of total C were recorded for the unexcavated and excavated areas when measured in July 2003, but these figures increased by contrasting amounts when measured in September 2005 and they differed greatly in August 2011 (Figure 9). The excavated area showed higher total C contents than the unexcavated area, which may reflect the contrasting rates of tree growth in the two areas. Moreover, carbon from external sources can be imported through inflow and trapped in sediment in the excavated area. The unexcavated area also recorded a slight increase in total C. This would be caused by biomass accumulation of AM which intruded and grew quickly for a short period.

Prior to the land excavation (August 2001), the surface soil to a depth of 5 cm contained 71.8 tC/ha of soil OC (Figure 10). This amount is low compared with those of natural mangrove forests which represented 370–553 tC/ha [27]. Significant amounts of soil OC may have been lost during the construction of shrimp ponds or after abandonment of the pond in this area. C stocks that were once resistant to decay (under anaerobic conditions) can be lost through aerobic respiration [28]. Clearing of mangrove forests results in statistically significant reductions of C stocks in sediments [29]. Up to 50% of soil C was lost over an 8-year period [30,31] and this study estimated 13%–20% loss in nearly 20 years from the 1980s to 2001.

Soil OC increased from 113.3 tC/ha (September 2005) to 154 tC/ha (August 2011) in the excavated area while in the unexcavated area soil OC decreased from 97.4 tC/ha (September 2005) to 68.3 tC/ha (August 2011) in 6 years (Figure 10). While AM grew vigorously from September 2001 for some years, soil OC could also increase so that soil OC in September 2005 was higher than in August 2001. However, soon after AM growth slowed down, a degree of soil OC decomposition intensified resulting in lower soil OC content measured in August 2011. Unless hydraulic conditions are improved in the unexcavated area, soil OC decomposition is likely to be continuing.

Cohen et al. [30] reported that an increase in tidal height of 20 cm at the mangrove stand corresponds to a twofold increase in inundation frequency. In this study it was revealed that excavating 25 cm increased the annual inundation time from 463 hours to 7,597 hours (Figure 5). With this improvement of hydraulic conditions, soil OC accumulation rate is likely to have increased drastically. We adopted one uniform depth for excavation in this study. Our study examining effects of excavation on mangrove growths was therefore limited by a lack of growth data at other excavation depths. Before a certain mangrove species is planted, adequate inundation frequency shall be determined. Since optimum inundation frequency is different among mangrove species, a depth to be excavated has to be sought for species by species.

The unexcavated area has no inundation for 4 months (April, August, September, October) (Figure 5). Severe dry conditions without inundation may accelerate soil OC decomposition, which is the main reason for carbon loss during the 6 year period from September 2005 to August 2011 in the unexcavated area.

IPCC deals with carbon changes in the chapter of Land-use change and forestry (LULUCF), however its consideration is mainly given to land conversions from inland forest and/or grassland. Since mangrove ecosystems contain a huge amount of carbon underground [31], land use change in a mangrove ecosystem should be paid more attention since a significant amount of carbon will be lost due to the change from mangrove to shrimp pond at many coastal areas in the tropics.

The REDD+ program for the valuation of carbon stored forests includes conservation, enhancement of carbon stocks, and the sustainable management of forests. Mangroves are eligible for the REDD+ program. However, large uncertainties still exist in the carbon sequestration potential of mangroves [32]. Hydraulic conditions significantly determine mangrove forest development and the rate of carbon sequestration as well. Therefore, appropriate hydraulic conditions have to be implemented for enhancement of mangrove carbon stocks. Hydrology restoration should be completed, especially if mangrove rehabilitation is intended to sequester atmospheric CO₂.