The distribution of mangrove forests is limited to the coastal zones of tropical and subtropical regions, and their total area is far smaller than that of upland forests (Spalding et al. Reference SPALDING, KAINUMA and COLLINS2010). Mangrove forests often show unique patterns of biomass allocation and carbon dynamics because they are periodically submerged by tides (Komiyama et al. Reference KOMIYAMA, ONG and POUNGPARN2008). Therefore, the contribution of mangrove forests to the global carbon fixation process should be carefully evaluated even though their distribution area is limited.

Ecologists have recently focused on the net ecosystem productivity (NEP) of forests (Houghton Reference HOUGHTON2002, Luyssaert et al. Reference LUYSSAERT, SCHULZE, BÖRNER, KNOHL, HESSENMÖLLER, LAW, CIAIS and GRACE2008) in relation to climate change. NEP is estimated as the balance between the gross primary productivity (GPP) and total respiration in an ecosystem, taking into account both autotrophic respiration (AR) and heterotrophic respiration (HR). Theoretically, GPP is composed of net primary productivity (NPP) and AR. Originally derived by Kira & Shidei (Reference KIRA and 1967), the summation method expresses NPP as the sum of the rate of biomass increment (Y), litter productivity (L), and the amount grazed by herbivores (G). An extension of the summation method, NEP can be expressed as a product of NPP minus HR from the following formula: NEP = GPP – (AR + HR) = NPP − HR.

Although extensive information is available on the NPP of mangrove forests (Khan et al. Reference KHAN, SUWA and HAGIHARA2009, Komiyama et al. Reference KOMIYAMA, ONG and POUNGPARN2008, Ross et al. Reference ROSS, RUIZ, TELESNICKI and MEEDER2001, Sherman et al. Reference SHERMAN, FAHEY and MARTINEZ2003) and carbon stocks (Donato et al. Reference DONATO, KAUFFMAN, MURDIYARSO, KURNIANTO, STIDHAM and KANNINEN2011, Reference DONATO, KAUFFMAN, MACKENZIE, AINSWORTH and PFLEEGER2012, Kauffman et al. Reference KAUFFMAN, HEIDER, COLE, DWIRE and DONATO2011), relatively few studies have examined the NEP of these ecosystems (Alongi Reference ALONGI2011, Barr et al. Reference BARR, ENGAL, FUENTES, ZIEMAN, O'HALLORAN, SMITH and ANDERSON2010), which is probably due to the difficulty of measuring soil respiration under tidal conditions. In addition, separating HR from AR is difficult in field studies. Using the same mangrove forest examined in the present study, Poungparn et al. Reference POUNGPARN, KOMIYAMA, TANAKA, SANGTIEAN, MAKNUAL, KATO, TANAPERMPOOL and PATANAPONPAIBOON(2009) measured emission rates of CO₂ from the soil surface without any pneumatophores in the soil chamber. They assumed that these CO₂ emission rates were closed to HR, because it is well known that most mangroves have internal conductive tissues connected to roots (Tomlinson Reference TOMLINSON1986) and thus undergo most root AR through pneumatophores to the atmosphere.

Following Poungparn et al. (Reference POUNGPARN, KOMIYAMA, TANAKA, SANGTIEAN, MAKNUAL, KATO, TANAPERMPOOL and PATANAPONPAIBOON2009), the present study was designed to update information on carbon dynamics via the estimation of NEP of distinct vegetation zones. From 2006 to 2009, we measured the rates of NPP and HR in a secondary mangrove forest characterized by Sonneratia–Avicennnia, Rhizophora and Xylocarpus zones. We hypothesized that NEP differed across these zones. The zonal and yearly differences in NEP are discussed, and the magnitude of the NEP in this mangrove forest is compared with those of other forest ecosystems.

The study site is located on an estuary of the Trat River, eastern Thailand (12°12′N, 120°33′E). It is the same mangrove forest examined by Poungparn et al. (Reference POUNGPARN, KOMIYAMA, TANAKA, SANGTIEAN, MAKNUAL, KATO, TANAPERMPOOL and PATANAPONPAIBOON2009). This area of Thailand experiences wet and dry seasons throughout the year. According to the Department of Meteorology, annual precipitation for 2006–2007 and 2008–2009 was comparable (5102 mm vs. 5332 mm). The average mean temperature from 2005 to 2010 was 27.6 °C with a small variation among the years.

A 40-m-wide plot was established on a river edge extending 110 m inland as described by Poungparn et al. (Reference POUNGPARN, KOMIYAMA, TANAKA, SANGTIEAN, MAKNUAL, KATO, TANAPERMPOOL and PATANAPONPAIBOON2009). Based on the dominant tree species, four vegetation zones were recognized from the riverside; the Sonneratia and Avicennia zones (dominated by S. caseoralis (L.) Engl. and A. alba Blume, respectively), the Rhizophora zone (dominated by R. apiculata Blume and R. mucronata Poir.), and the Xylocarpus zone (dominated by X. granatum Koenig). The tree density (diameter of trunk at breast height or 0.3 m above highest prop root (D) more than 4.5 cm), total basal area, and range of D for each zone are shown in Table 1.

Table 1.

Forest structure in three zones of a secondary mangrove forest in Trat Province, eastern Thailand. D and H refer to trunk diameter and total tree height, respectively.

NPP was estimated by using the summation method (Kira & Shidei Reference KIRA and 1967) as a sum of the following rates: the biomass increment (Y), litter productivity (L) and amount grazed by herbivores (G). The rates of G and below-ground litter productivity were not measured in this study. To estimate Y, individual tree biomasses (above-ground and root weight, W _top and W _R) were calculated for each year using the common allometric equations (Komiyama et al. Reference KOMIYAMA, POUNGPARN and KATO2005); W _top = 0.251 ρ D ^2.46 and W _R = 0.199 ρ ^0.899D ^2.22 (where D and ρ are trunk diameter and wood density, respectively). These common equations were developed using 36 sample trees (from a total of 104) cut from the mangrove forest in this study. In the Sonneratia–Avicennia, Rhizophora and Xylocarpus zones, Y was calculated for the two periods of August 2006–July 2007 (Period I) and August 2008–July 2009 (Period II).

To estimate L, a total of 12 litter traps (area 1 ×1 m at 1.3 m above ground) were used to collect monthly litter fall in the three vegetation zones for the two periods. Litterfall was separated into plant organs (wood, leaves and reproductive parts) before obtaining the dried weight. The carbon contents for all components of Y and L were assumed to be 50% of the dry weights (MacDicken Reference MACDICKEN1997). Unit of the Y and L rates was expressed in Mg C ha⁻¹ y⁻¹.

The monthly rates of HR in Period I for the present study were estimated based on the average rate reported in Poungparn et al. (Reference POUNGPARN, KOMIYAMA, TANAKA, SANGTIEAN, MAKNUAL, KATO, TANAPERMPOOL and PATANAPONPAIBOON2009). They previously measured the rates of HR for four vegetation zones during low-tide periods in August 2006 (during the rainy season) and March 2007 (during the dry season). The HR rates during the submerged periods were assumed to be the same as those during dry periods. For the Sonneratia–Avicennia zone, we used the average values of HR between the two zones because no significant differences were observed between them. The monthly HR in Period II was estimated according to the relationship between soil temperature and HR (HR = 0.0004X ^2.204, where HR in μCO₂ m⁻² s⁻¹, X in °C; after Poungparn et al. Reference POUNGPARN, KOMIYAMA, TANAKA, SANGTIEAN, MAKNUAL, KATO, TANAPERMPOOL and PATANAPONPAIBOON2009), and summed values to calculate the total HR. For this calculation, we measured the soil temperature in each zone using temperature sensors and loggers (TidbiT v2 Temp logger, Onset Computer Co., Ltd.). Three sensors and loggers were buried at a depth of 5 cm from the soil surface in the three zones. Measurements of soil temperature were taken at 30-min intervals.

Above-ground and root biomasses gave the rates of Y ranging from 4.43 to 7.24 Mg C ha⁻¹ y⁻¹, based on vegetation zone and study period (Table 2). The value of Y in the Xylocarpus zone tended to be lower than those of the other two zones. The Sonneratia–Avicennia zone showed similar rates of Y between Period I and II. The other two zones exhibited relatively high rates in Period I, although the differences were not large. The rates of annual L for the three vegetative zones had a narrow range of 4.38–6.66 Mg C ha⁻¹ y⁻¹ (Table 2). The highest L occurred in the Rhizophora zone, whereas L was relatively low in the Sonneratia–Avicennia zone during both Periods I and II.

Table 2.

Biomass increment, litter productivity and NEP in three zones of the study; secondary mangrove forest during Period I (August 2006–July 2007) and Period II (August 2008–July 2009).

Based on the estimations of Y and L, calculated NPP varied from 9.35 to 12.9 Mg C ha⁻¹ y⁻¹ (Table 2). NPP was high in the Rhizophora zone and low in the Xylocarpus zone. The range of NPP obtained by the present study was slightly higher than values of NPP of 6.25 and 10.5 Mg C ha⁻¹ y⁻¹ reported for New World mangrove forests with similar dimensions for the above-ground parts (Day et al. Reference DAY, CORONADO-MOLINA, VERA-HERRERA, TWILLEY, RIVERA-MONROY, ALVAREZ-GUILLEN, DAY and CONNER1996, Sherman et al. Reference SHERMAN, FAHEY and MARTINEZ2003: unit of NPP (Mg C ha⁻¹ y⁻¹) was converted assuming 50% carbon content).

The HR rate exhibited a narrow range of 1.72 to 2.63 Mg C ha⁻¹ y⁻¹ during Period I (Table 2). Based on the relationship between soil respiration and soil temperature, values of HR in Period II were estimated to be 2.26, 2.16 and 2.01 Mg C ha⁻¹ y⁻¹ for the Sonneratia–Avicennia, Rhizophora and Xylocarpus zones, respectively. The narrow range of HR examined by this study is comparable to the HR rates of 1.40–1.93 Mg C ha⁻¹ y⁻¹ reported for mangrove forests of Australia and New Zealand (Lovelock Reference LOVELOCK2008) and 1.12–4.33 Mg C ha⁻¹ y⁻¹ in mangrove forests of Malaysia, Thailand and Australia (Alongi Reference ALONGI2011). The reported HR estimates for mangrove forests tend to be low, perhaps due to the anaerobic conditions of the soils. Decomposition of organic matter is considered to be slow in most mangrove forests.

We estimated NEP from NPP and HR. During Periods I and II, the NEP of the study mangrove forest varied from 7.34 to 11.3 Mg C ha⁻¹ y⁻¹ depending on the vegetation zone (Table 2). The highest NEP was found in the Rhizophora zone in Period I. The Xylocarpus zone exhibited relatively low rates of NEP throughout both periods, probably due to the relatively low NPP rates in this zone (Table 2). Although some variation in NEP was observed among vegetation zones, no significant differences were found (one-way ANOVA, F_{(2, 5)} = 4.21, P = 0.135). Furthermore, no temporal differences in NEP were detected between the two periods (t-test, t = 0.179, P = 0.856, df = 4). The absence of extreme climatic events during the study periods may have resulted in these similar estimates of NEP.

Therefore, as determined using the summation method, the NEP of this mangrove forest was estimated to range from 7.34 to 11.3 Mg C ha⁻¹ y⁻¹. Using the same method, Alongi (Reference ALONGI2011) predicted a higher NEP of 15.6 Mg C ha⁻¹ y⁻¹, accounting for rates of fine-root litter and algal productivity. Furthermore, using the eddy-covariance method, Barr et al. (Reference BARR, FUENTES, O'HALLORAN, BARR and ZIEMAN2006, Reference BARR, ENGAL, FUENTES, ZIEMAN, O'HALLORAN, SMITH and ANDERSON2010) estimated that the NEP of mangrove forests in Florida, USA, fall into a similar range of 7.00–11.7 Mg C ha⁻¹ y⁻¹. However, in upland forests of tropical Brazil, Araújo et al. (Reference ARAÚJO, NOBRE, KRUIJT, ELBERS, DALLAROSA, STEFANI, RAINDOW, MANZI, CULF, GASH, VALENTINI and KABAT2002) and Miller et al. (Reference MILLER, GOULDEN, MENTON, ROCHA, FREITAS, FIGUEIRA and SOUSA2004) reported low values of NEP, ranging from 1.00 to 8.00 Mg C ha⁻¹ y⁻¹, estimated using the eddy-covariance method. Finally, Luyssaert et al. (Reference LUYSSAERT, SCHULZE, BÖRNER, KNOHL, HESSENMÖLLER, LAW, CIAIS and GRACE2008) determined low values of NEP for 519 upland forests in the temperate and boreal climates, with maximum rates of approximately 8.0 Mg C ha⁻¹ y⁻¹.

In conclusion, the rates of NEP estimated in the present study are comparable to those reported for mangrove forests studied elsewhere but higher than those reported for other upland forests. Mangrove forests are considered to be effective sinks of atmospheric carbon due to their high NPP and low HR.