INTRODUCTION
Forest ecosystems play an important role in regulating atmospheric CO2 concentration (Dixon et al. Reference DIXON, BROWN, HOUGHTON, SOLOMON, TREXLER and WISNIEWSKI1994, Houghton Reference HOUGHTON2002). A forest with high net primary production (NPP) would be a strong carbon sink if the transfer from biomass increment to carbon uptake is efficient at the ecosystem level. Small losses of carbon via soil respiration would strengthen this assumption. For mangroves, which are highly productive in terms of annual biomass increment (Amarasinghe & Balasubramaniam Reference AMARASINGHE and BALASUBRAMANIAM1992, Christensen Reference CHRISTENSEN1978, Day et al. Reference DAY, CONNER, LEY, DAY and NAVARRO1987, Ong et al. Reference ONG, GONG and CLOUGH1995, Sherman et al. Reference SHERMAN, FAHEY and MARTINEZ2003), loss of carbon via soil respiration may theoretically counter high rates of primary productivity in assessment of the source or sink strength for atmospheric carbon of mangrove ecosystems.
While the NPP of mangrove forests has received much attention (Putz & Chan Reference PUTZ and CHAN1986, Robertson et al. Reference ROBERTSON, DANIEL and DIXON1991, Ross et al. Reference ROSS, RUIZ, TELESNICKI and MEEDER2001, Sherman et al. Reference SHERMAN, FAHEY and MARTINEZ2003), only a few studies have considered the soil respiration rate of mangroves (Alongi et al. Reference ALONGI, WATTAYAKORN, PFITZNER, TIRENDI, ZAGORSKIS, BRUNSKILL, DAVIDSON and CLOUGH2001, Kristensen et al. Reference KRISTENSEN, HOLMER, BANTA, JENSEN and HANSEN1995, Lovelock Reference LOVELOCK2008, Mall et al. Reference MALL, SINGH and GARGE1991). Most of these studies reported a low rate of soil respiration, possibly caused by the anaerobic conditions of mangrove soils.
Mangrove forests, periodically inundated by seawater, are formed into vegetation zones with the dominant tree species changing with distance from the seashore (Bunt Reference BUNT1996, Youssef & Saenger Reference YOUSSEF and SAENGER1999). Potential causes for the vegetation zones include plant succession and physiological adaptation to physico-chemical gradients in mangrove forest (Smith Reference SMITH, Robertsons and Alongi1992). The physical condition of mangrove soil differs among vegetation zones due to inundation patterns and changes in elevation. However, few studies have compared the rate of soil respiration among vegetation zones.
In the present study, the magnitude of CO2 emission through soil respiration is compared among four zones dominated respectively by Sonneratia, Avicennia, Rhizophora and Xylocarpus species in a secondary mangrove forest in both the wet and dry seasons. The soil temperature is also compared among them. We hypothesized that rates of soil respiration differ among the vegetation zones and seasons. Moreover, the relationship between soil respiration and temperature is examined. We discussed the zonal variation in soil temperature by causal factors such as elevation and inundation.
METHODS
Site description
The study site was a secondary mangrove forest on an estuary of the Trat River, Trat Province, eastern Thailand (12° 12′N, 120° 33′E, Figure 1a). This forest was formerly used for timber and charcoal production. Since the 1980s, Mangrove Forest Research and Development Station No. 4 (Department of Marine and Coastal Resources, Thailand) has managed this area. The annual precipitation and temperature at Trat were 5214 mm and 27.6 °C from 2003 to 2006 (Department of Meteorology, Thailand). In 2006, the precipitation in the wet season (May–October) was 89% of the total for the year.
Study site and sample plot. Location of the study site (a), sample plot (40 × 110 m) (b). Contour lines show the relative elevation from the datum point. The four survey areas (A, B, C and D) represent the sites where soil respiration and temperature were measured in the respective zones.
A study plot (40 × 110 m) was established in 2001. Tree density (diameter at breast height, dbh > 4.0 cm) was 1682 stems ha−1, and average dbh and height were 11.7 cm and 10.8 m, respectively. The dominant tree species changed with distance from the river, and the forest was divided into four vegetation zones based on dominant tree species (Figure 1b). The river fringe was occupied by a narrow belt of Sonneratia in very soft and deep mud. This zone formed a pure stand of S. caseolaris with regularly distributed conically shaped, 20–30-cm-high pneumatophores. Inland of the Sonneratia zone was the Avicennia zone, with many tall (height = 17 m) A. alba trees that produced a high density of short, slim pneumatophores. The soil was soft mud but shallower than that of the Sonneratia zone. Next was a zone mainly composed of dense stands of Rhizophora apiculata and R. mucronata, which produced many stilt roots in a soft mud substrate approximately 5–10 cm deep. A dense root necromass had formed below that depth. The fourth zone, farthest from the river, was the Xylocarpus zone, comprising X. granatum, Bruguiera gymnorrhiza and Ceriops tagal, with small buttresses. Large amounts of root necromass were densely accumulated in the Xylocarpus zone soils.
The elevation of the study site was relatively low in the Sonneratia zone, with the slope gradually increasing toward the inland zones to a total 1.2-m difference in elevation between the riverside and the most inland zone (Figure 1b). The range of elevation in each plot was 0–0.3 m in the Sonneratia zone, 0.6–0.8 m in Avicennia, 0.6–0.9 m in Rhizophora and 0.9–1.2 m in the Xylocarpus zone.
The study site is generally affected by a single tide, as is most of the mangrove forest area of the Trat River estuary (Royal Thai Navy, Thailand). The highest tide level in the wet season at the various zone elevations is typically around 1.5, 0.9, 0.7 and 0.3 m in the Sonneratia, Avicennia, Rhizophora and Xylocarpus zones, respectively.
Measurement of soil respiration
To measure soil respiration, we established four survey areas in each vegetation zone (c. 50 m2 each; Figure 1b). Measurements in the 2006 wet season for Avicennia, Rhizophora and Xylocarpus were taken on 11–12 August 2006. Measurements in the 2007 wet season for Sonneratia and Avicennia were taken on 8–12 August 2007. Measurements in the 2007 dry season for all vegetation zones were taken on 21–30 March 2007. At least 19 measurements of soil respiration were made in each area for each season (Table 1).
Average rates of CO2 emission via soil respiration in the mangrove forest.
Measurements were done during low-tide periods using a closed system consisting of a chamber and an infrared gas analyser (model LI-840, LI-COR Biosciences, USA). The open area of the chamber was 78.5 cm2. A 10-cmtall PVC collar was carefully inserted about 1 cm into open and non-disturbed soil surfaces with no above-ground roots, such as pneumatophores, stilt roots or buttresses, and remained in place 10 min before the chamber was attached. An elastic ring gasket was inserted between the chamber and collar to prevent air leaks. We used platforms to avoid soil disturbance around the chamber. The CO2 concentration in the chamber was measured at 1-s intervals continuously for 15 min. The data were recorded with a computer connected to the gas analyser. Differential CO2 concentrations were calculated for each second. The soil respiration rate of each measurement was then obtained by averaging the differential CO2 concentrations (60 s × 15 min). During soil respiration measurements, the soil temperature at a 5-cm depth was measured near the chamber, with the temperature recorded at 0, 5, 10 and 15 min, using a temperature sensor and recorder (Model TR-71U, T&D Co., Ltd.). The temperatures were averaged to obtain a mean temperature for each soil respiration measurement.
Measurement of environmental factors
Twelve temperature sensors and loggers (Thermochron G-type, KN Laboratories Inc., Japan, and TidbiT v2 Temp logger, Onset Computer Co., Ltd.) were buried at a depth of 5 cm from the soil surface in each study plot (Figure 1b), and measurements of soil temperature were taken at 30-min intervals from August 2006 to August 2007. Another sensor measured the water temperature of the river in front of the study plot. Unfortunately, the data loggers recording the August 2006–February 2007 temperature of the Xylocarpus zone were lost. The temperature sensors and loggers for all zones were reset in March 2007. Therefore, no temperature data were available for any zone in early March.
The inundation period between the low and high tides was investigated on 15 August 2006 (wet season) and 22 March 2007 (dry season). We recorded the time when the tide started to move inland from the river edge, and then periodically recorded as it reached each part of the study plot. The water table was assessed during low tides on 22–23 August 2006. Two-cm-diameter plastic pipes were placed in the soil 0, 35, 80 and 105 m from the river. The distance between the soil surface and the water in the pipes was measured using a ruler.
Statistical analysis
All statistical analyses were performed using SPSS 13.0 for window (SPSS Inc., USA). The difference in soil temperature across the vegetation zones was tested by one-way ANOVA and Duncan's multiple range test. Two-way ANOVA was used to test for effects of vegetation zone and season on rate of soil respiration. The analysis of significant differences in rate of soil respiration was tested by Games-Howell test (Day & Quinn Reference DAY and QUINN1989). The relationship between soil temperature and respiration was tested by regression analysis.
RESULTS
Inundation pattern and water table of vegetation zones
On 15 August 2007 (wet season), the single-tide inundation of seawater started at 12h25 in the Sonneratia zone. The Avicennia zone was completely submerged 20 min later. The seawater flooded the end of the Xylocarpus zone at 14h40. On 22 March 2007 (dry season), inundation began at 09h00 and slowly reached the Xylocarpus zone at 14h10. On this day, a section of the Xylocarpus zone was never submerged. Thus, the daily inundation period of the Sonneratia zone was longer than that of the innermost Xylocarpus zone, totalling at least 2 h 15 min in the wet season and 5 h 10 min in the dry season.
The water table (Figure 2) at the riverside in the Sonneratia zone had a relatively wide range of 5–15 cm depth below the soil surface. However, 35, 80 and 105 m from the river, the water table was stable, usually varying only slightly (0–3 cm depth).
Relative water table at four points from the river during low tide on 22 August 2006 at 09h20 (◆) and 13h30 (■); and 23 August 2006 at 10h00 (▲), 10h30 (×), 10h45 (◯), 11h30 (●) and 11h45 (+). Negative values indicate that the water level was lower than the soil surface.
Soil temperature
The soil temperatures were calculated by averaging the soil temperature across all four vegetation zones from August 2006 to August 2007, these were 28.19 °C in the dry (November to April) and 27.21 °C in the wet (May to October) seasons. We compared the average March–August 2007 soil temperatures among the four vegetation zones (Figure 3), because of the lost data from Xylocarpus zone mentioned above. Soil temperatures differed among the zones (ANOVA, F4,680 = 83.4, P < 0.01). The average soil temperatures tended to decrease with increasing distance from the river and were 30.14 °C, 29.63 °C, 28.38 °C and 27.71 °C for Sonneratia, Avicennia, Rhizophora and Xylocarpus, respectively. The average water temperature was 30.05 °C, which was not significantly different from the soil temperature in the Sonneratia zone (Duncan's multiple range test, P = 0.985) but significantly higher than soil temperatures of the other three zones (Duncan's multiple range test, P < 0.01) as shown in Figure 4.
Daily soil temperature for each zone and river water temperature during August 2006–August 2007. The Xylocarpus zone data of August 2006–February 2007 were not available because the data loggers were lost. For all vegetation zones, no data were obtained in early March due to the reset and preparation of the temperature sensors and loggers.
Mean CO2 emission (●), soil temperature (■), and elevation of the substrate (▲) among the four vegetation zones. Values with the same letters are not significantly different. The soil temperature of each zone was obtained by averaging the daily soil temperature from March to August 2007.
Soil respiration
The average rates of soil respiration in the wet and dry seasons of each zone are shown in Table 1. The rate of soil respiration in the Avicennia zone did not differ significantly between 2006 and 2007 in the wet season (ANOVA, F1,30 = 1.02, P = 0.320) or across all zones between the wet and dry seasons (two-way ANOVA, F1,160 = 3.35, P = 0.069). There was no significant interaction between zone and season (two-way ANOVA, F3,160 = 0.79, P = 0.053). However, the difference in soil respiration was significant (two-way ANOVA, F3,160 = 14.4, P < 0.001) among the four vegetation zones in both seasons. When the average rate of soil respiration in each zone (Table 1) was calculated regardless of season, the average rate was higher in the Sonneratia zone (0.876 μmol CO2 m−2 s−1) than in either the Rhizophora or Xylocarpus zones (0.525 μmol CO2 m−2 s−1), but did not differ from the zone immediately adjacent (Avicennia: 0.732 μmol CO2 m−2 s−1). The rate of soil respiration for the Rhizophora zone was lower than both the Sonneratia and Avicennia zones, roughly tracking the pattern of soil elevation change (Figure 4) and water table depth (Figure 2).
We checked the overall relationship between soil respiration and the temperature obtained in each measurement (n = 168). Although the relationship was significant in both linear (F1,167 = 17.5, P = 0.00004) and power function (F1,167 = 20.3, P = 0.00001), the coefficient of determination (R2) of regression line fitted a power function rather than that of a linear regression (0.109 vs. 0.0953). Therefore, we adopted the relationship between soil respiration and temperature fitted by a power function (Figure 5).
Response of soil respiration to soil temperature in the mangrove forest. The regression line was fitted to a power function as Y = 0.0004 X 2.204, R 2 = 0.109.
DISCUSSION
Carbon dioxide emission via soil includes CO2 efflux from roots and heterotrophic respiration (Hanson et al. Reference HANSON, EDWARDS, GRATEN and ANDREWS2000, Lee et al. Reference LEE, NAKANE, NAKATSUBO and KOIZUMI2003). Recently, Komiyama et al. (Reference KOMIYAMA, ONG and POUNGPARN2008) suggested a method of balancing net ecosystem production (NEP) without taking root respiration into consideration. Consequently, the NPP and heterotrophic respiration in soil allow estimation of NEP. The mangroves we studied develop peculiar root systems, in which the aerenchymatous tissue of underground roots is connected with lenticels on pneumatophores, prop roots, and buttresses above the ground (Tomlinson Reference TOMLINSON1986). Scholander et al. (Reference SCHOLANDER, VAN DAM and SCHOLANDER1955) clarified the process of gas exchange through these lenticels under submerged conditions. Most metabolic respiration from underground roots is released through the lenticels. The underground roots likely make only a small contribution to soil respiration when measured via respiration chambers placed to avoid above-ground roots. Therefore, the soil CO2 efflux obtained in the present study would be close to the CO2 released from heterotrophic respiration.
The soil temperature at the study site fluctuated in a relatively small range by season, although exceptionally low soil temperatures occurred during two short periods (Figure 3) in mid-December and early February, due to cold air masses from high-latitude regions. Soil temperature in tropical zones usually fluctuates in a narrow range (Hashimoto et al. Reference HASHIMOTO, TANAKA, SUZUKI, INOUE, TAKISAWA, KOSAKA, TANAKA, TANTASIRIN and TANGTHAM2004, Kosugi et al. Reference KOSUGI, MITANI, ITOH, NOGUCHI, TANI, MATSUO, TAKANASHI, OHKUBO and NIK2007) compared with that in temperate zones (Hirata et al. Reference HIRATA, HIRANO, SAIGUSA, FUJINUMA, INUKAI, KITAMORI, TAKAHASHI and YAMAMOTO2007).
The rate of soil respiration basically depends on soil temperature (Davidson et al. Reference DAVIDSON, BELK and BOONE1998, Lee et al. Reference LEE, NAKANE, NAKATSUBO and KOIZUMI2003, Martin et al. Reference MARTIN, BERINGER, HUTLEY and MCHUGH2007, Nakane et al. Reference NAKANE, YAMAMOTO and TSUBOTA1983). While we found a strong relationship between soil temperature and soil respiration (Figure 5), temperature dissimilarity between the wet and dry seasons assessed was apparently not sufficient to drive seasonal difference in soil respiration rate. Hashimoto et al. (Reference HASHIMOTO, TANAKA, SUZUKI, INOUE, TAKISAWA, KOSAKA, TANAKA, TANTASIRIN and TANGTHAM2004) also reported that the difference in soil respiration rate was not significant between dry and rainy seasons in a tropical forest in northern Thailand, and showed a small change in soil temperature throughout the year.
Some studies on soil respiration in tropical forests are shown in Table 2. This provides evidence pointing to generally lower rates of soil respiration in mangrove forests than in tropical upland forests. The water table at our study site was high (Figure 2), and the soil respiration rate is negatively correlated to soil water content (Davidson et al. Reference DAVIDSON, BELK and BOONE1998, Kosugi et al. Reference KOSUGI, MITANI, ITOH, NOGUCHI, TANI, MATSUO, TAKANASHI, OHKUBO and NIK2007, Scott-Denton et al. Reference SCOTT-DENTON, SPARKS and MONSON2003). Moreover, mangrove soils often experience anaerobic conditions (Lawton et al. Reference LAWTON, TODD and NAIDOO1981, McKee et al. Reference MCKEE, MENDELSSOHN and HESTER1988), which may restrict the activity and respiration of microbes and benthic organisms in mangrove forests compared to other terrestrial forests.
Published ranges of soil respiration rates in tropical forests.
It is noteworthy that the difference in soil respiration rate was significant among the four vegetation zones. The rate was highest in the Sonneratia zone at the riverside and trended lower in the more inland zones (Figure 4, Table 1). On the other hand, soil temperature tended to decrease with distance from the river (Figure 4). Thus, the spatial difference in soil respiration over the zones can be explained by the variation in soil temperature.
We considered the effect of the temperature of inundating seawater on soil temperature over the mangrove vegetation zones. At the study site, the seawater temperature was usually higher than that of the soil (Figure 3), and therefore the water warmed the soil when it invaded the forest.
The period of inundation depends on microtopography, especially elevation (Figure 1b). The Sonneratia zone near the river had the lowest elevation among the four vegetation zones, and thus was submerged for the longest period of time each day, with the long period of inundation sustaining the higher temperature of the soil. In contrast, Xylocarpus, the most inland zone, was submerged for the shortest period of time. The time spent submerged differed by approximately 2–5 h d−1 between the Sonneratia and Xylocarpus zones. The difference was longest in the dry season, possibly due to the low specific gravity (salt concentration) of the inundating seawater in the wet season.
In conclusion, the magnitude of CO2 emissions through soil respiration is low in mangrove forests compared with other tropical upland forests. Moreover, we found that the rate of soil respiration differed by vegetation zone. Zones at lower elevations, which are submerged for long periods by warm seawater, show higher rates. Therefore, the decomposition process of organic matter may be closely related to the inundation pattern or elevation of mangrove forests. This may help to explain the zonation patterns of mangroves in terms of soil formation and soil properties.
ACKNOWLEDGEMENTS
We thank the SATECO-COE Program of Gifu University, Japan, for providing equipment and financial support. We also thank the staff at Mangrove Research and Development Station No. 4 at Trat, Thailand for assistance in the field.
