We conducted our field studies in Tripura, which is the second smallest state of Northeast India with total 10,491 km² geographical area. This sanctuary covers total 194.704 km² geographical area located between 23˚05'N - 23˚25'N, and 91˚20'E - 91˚35'E (Figure 1), which was notified in November 1988, with total 27 revenue Mouza of Belonia, Udaipur and Sonamura Civil sub-division of South Tripura District. As per the Champion and Seth (1968) [22] classification system forest types of the sanctuary mainly consists of: 1) Cachar Tropical Semi Evergreen Forest 2) East Himayan lower Bhabhar

Moist Deciduous Sal Forest, 3) Moist Mixed Deciduous Forest, 4) Moist Bamboo Brakes and 5) Savanah Wood Land [23] . The elevation of the study sites ranges between 17 m to 83 m above mean sea level. The climate of this area is generally moist and humid. The minimum and maximum temperatures in summer are 21˚C and 38˚C. In winter it ranges between 4˚C and 33˚C. Humidity is generally high throughout the year. In the summer season the relative humidity differs between 50% - 74% whereas in the rainy season it is over 85%. The mean wind speed is 7.1 km/hr, with maximum of 13.1 km/h in May and minimum of 3 km/h in December. The mean annual rainfall varies between 192 - 285 cm and increased from Southwest to Northeast. The soil type of the study area is mainly red loam and sandy loam soils.

Vegetation inventories data were collected during 2010 to 2011 by 20 line transects. Line transects were placed randomly in different forest patches in Trishna Wildlife Sanctuary. The successional gradients were identified visually following the characteristics drawn by Champion and Seth (1968) [22] . Out of 20 transects, 5 transects represents Moist Bamboo Brakes (MBB) dominated by Bambusa tulda, 6 transects from Moist Deciduous Sal (MDS) patch dominated by Shorea robusta, 5 transects represents Moist Mixed Deciduous (MMD) patch dominated by Schima wallichii and 4 transects from Semi Evergreen Dipterocarpus (SED) patches dominated by Dipterocarpus turbinatus. All woody individuals at ≥10 cm girth at breast height (gbh), at 1.3 m height were measured in 10 m wide and 500 m length transects. Thus, each line transect represented an area of 0.5 ha (10 × 500 m) and encompassed 5 (10 × 100 m) contiguous sub-plots. Specimens were identified with the help of The Flora of Tripura State [24] . The reference herbarium was deposited in herbarium of Botany Department; Tripura University. Information about individual species, their status including their herbarium voucher number recorded from the study site is available in a separate publication [23] .

Field data were quantitatively analysed on per hectare basis for density and basal area [25] . For estimation of woody Above Ground Biomass (AGB), regression model was used proposed by Brown et al. (1989) [26] due to relatively similar climatic condition of the study area (rainfall 150 - 400 cm range per year): AGB per woody species in kg (Y) = 42.69 − 12.800 (dbh) + 1.242 (dbh²) was and presented into S I unit (AGB Mg ha⁻¹). Where, DBH ranges between 5 - 148 cm, number of sampled tree was 170 and regression coefficient (R²) = 0.84 [26] . Moist deciduous forest frequently dominated several bamboo species. Since, present study area was found to be dominate by Bambusa tulda; we used separate equation for biomass estimation of bamboo species [27] , W = a dbh^b. Where, W= weight of bamboo culm in kg, dbh = diameter at breast height, allometric values for a = 0.141 and b = 2.48 with regression coefficient (R²) = 0.973 [27] . The estimation of Cstocks (C Mg ha⁻¹) was calculated as 50% of the total biomass; since, C content in plant tissue is approximately half of the dry weight of aboveground live biomass [4] [6] . The significance of differences in forest structural variables among the patches was statistically tested using one?way analysis of variance (ANOVA) and Tukey’s test.

To compare the size-class distributions of biomass storage in each forest patches, we plotted AGB and C Mg ha⁻¹ along ten gbh classes (>10, 10.1 - 20, 20.1 - 30, 30.1 - 40, 40.1 - 50, 50.1 - 60, 60.1 - 70, 70.1 - 80, 80.1 - 90 and >90 cm). The relationship between AGB Mg ha⁻¹ (y) and density ha⁻¹ (x) along teng bh classes in four different forest patches were examined and compared by linear curve fitting. The statistical analysis was performed by PAST version 1.89 [28] .

Stem density was recorded highest for MBB (1088.4 ± 96.15 ha⁻¹) and lowest in case of SED (701 ± 45.79 ha⁻¹). However, density of stems was found varied significantly (F = 4.20, df = 3.19; p = 0.02) within the forest patches. Basal area at the study sites ranged between 8.91 ± 1.39 m² ha⁻¹ (MMD)to 33.69 ± 8.76 m² ha⁻¹ (SED), followed by 9.17 ± 1.24 m² ha⁻¹ in MBB and 9.05 ± 1.60 m² ha⁻¹ in MDS. Basal area also significantly varied (F = 10.04, df = 3.19; p < 0.001) among the patches. Mean stem dbh was significantly high in SED (50.56 ± 2.77 cm) than other patches (F = 25.88, df = 3.19; p < 0.001). Even, mean canopy height (m) was also recorded high in SED (9.28 ± 0.97 m) and found significantly differed among the forest patches (F = 12.71, df = 3.19; p < 0.001) (Table 1).

Highest stem density in MBB might be due to high abundance of Bambusa tulda (density ha⁻¹); which is locally adaptive and ecologically dominated species, facilitated

with high regeneration abilities and greater growth rate [29] . Even, high growth trait of bamboo can fix atmospheric C faster than similar dbh sized of a tree. In fact, Bambusa tulda have the potentiality to shift the present forest under bamboo controlled retrogression stage. Usually, there is a high density of stems with low dbh in early and intermediate stages (e.g. MMD and MBB patches), and as dbh increases with successional trend, stem density decreases in late stage (SED patch). Negi et al. (2003) [30] observed that the tree types have maximum C stored in the order conifers > deciduous > evergreen > bamboos. Instead of low stem density in SED patch (701 ± 45.79 ha⁻¹), basal area was significantly high (33.69 ± 8.76 m² ha⁻¹) due to presence of voluminous Dipterocarpus turbinatus as the representative of late successional stage. Nevertheless, forest structure changed along the successional gradient according to the general pattern of secondary succession with a gradual increase in height and basal area as described for tropical forests. In addition, Shorea robusta and Dipterocarpus turbinatus locally possess good natural regeneration trait and fast growing ability [31] . Therefore, these species came out as significant C sequester in this region and long term monitoring of C dynamics in those forests are possible through timescale observation on these two key dominated species. We also predicted that, significant differences in species composition and dominant trends may influence the biomass and C stock in each forest patch through control over water availability, litter and debris deposition, composition and quantity of root exudates and the distribution of C in the soil profile.

The value of AGB in the whole study area, ranged between 20.86 Mg ha⁻¹ (MDS) to 126.37 Mg ha⁻¹ (SED). However, mean AGB was significantly greater (F = 7.01, df = 3.19; p < 0.01) in SED (85.59 ± 17.76 Mg ha⁻¹) than recorded minimum value (37.85 ± 4.02 Mg ha⁻¹) in case of MMD. Highest contributor of AGC stock was SED patch (35.13 ± 8.88 Mg ha⁻¹) followed by MBB (22.08 ± 2.60 Mg ha⁻¹), MDS (19.11 ± 2.39 Mg ha⁻¹) and MMD (18.92 ± 2.01 Mg ha⁻¹), which also found significantly varied among the different patches (F = 7.01, df = 3.19; p < 0.01) (Table 2). In MBB patch, Bambusa tulda contributed about 35.5% (14.46 Mg ha⁻¹) of total biomass. Other dominant tree shared less than 10% of total biomass i.e. Schima wallichii (9.21% or 3.86 Mg ha⁻¹); Terminalia bellirica (6.55% or 2.72 Mg ha⁻¹); Microcos peniculata (4% or 1.68 Mg ha⁻¹) and Lannea coromandellica (2.62% or 1.10 Mg ha⁻¹). In MDS patch, Shorea robusta was the highest contributor and shared about 56.33% (23.81 Mg ha⁻¹) of total biomass followed by Dipterocarpus turbunatus 6.09% (2.58 Mg ha⁻¹), Schima wallichii 3.33% (1.41% Mg ha⁻¹), Terminalia bellirica 3.13% (1.33 Mg ha⁻¹) and Microcos peniculata 2.14% (0.91 Mg ha⁻¹). Among the present studied forest patches, most common dominant species was Schima wallichii and it contributed about 8.1% (16.63 Mg ha⁻¹) of the total biomass of the study area. Overall, the highest contributor was Dipterocarpus turbinatus, it shared about 22.34% (46.38 Mg ha⁻¹) of total biomass (Table 2). In the present estimation of total biomass, 70% was stocked at <30 cm dbh class; which indicated sufficient amount of young stems or maximum number of trees could not attained full maturity in those

forest patches, possibly these forest patches were recovering from significant historic disturbances. Being as early and intermediated stages MMD and MBB patches (37.85 and 41.84 Mg ha⁻¹) recover much biomass compared to MDS patch (42.26 ± 5.91 Mg ha⁻¹). This may due to the changes in forest composition and structure during succession, occurred at very different rates; and biomass generally recovers more rapidly than species richness [8] [11] [12] . About 40% tree species contributed >70% of total biomass in dry deciduous forest [26] [27] . In the present moist deciduous forest patches, five most dominant species contributed 57.41% of biomass in MBB, 70.06% in MDS, 21.02% in MMD and 70.08% in SED. Out of 5 most dominated species in all forest patches, the top dominant species contributed 34.85% biomass (Bambusa tulda) in MBB, Shorea robusta had 56.34% biomass in MDS, Terminalia bellirica had 5.92% in MMD and Dipterocarpus turbinatus contributed 51.18% biomass in SED (Table 2). Both SED and MSD forests were came out as more mature or late successional phase than other patches and might preserve much more C and also responsible for greater C management. Further, AGB recorded in the present study ranged from 20.86 Mg ha⁻¹ to 126.37 Mg ha⁻¹ also fall within the range of the earlier study from the area [15] - [17] , which was found less than AGB value of Central Himalaya (171.9 - 380.3 Mg ha⁻¹) [32] , Western Ghats (468 - 607.7 Mg ha⁻¹) [33] and in Eastern Ghats (15.61 - 597.13 Mg ha⁻¹) [34] . Present AGB value was found close to other reported value of biomass in northeast India viz. Meghalaya, Assam and Manipur [10] - [12] .

Estimation of live tree biomass is very crucial for an ecosystem, especially to understand overall ecosystem health and services including the hydrological cycle, soil erosion, nutrient cycling and dynamics of terrestrial C [3] [5] [6] . Although, our biomass and C stock means were statistically different, which suggested there was considerable variation in stand structure across the forest patches. The biomass contrast of five top most dominant trees was consistent with diversity and structural complexity. For instance, basal area, density of voluminous trees and diameter were found greater in late successional forest patch (SED) than the patches at early stages, and thereby repre- sented higher quantity of biomass and C stocks (Table 1). Results also suggest that key dominant trees make a proportionately greater contribution to total biomass as stands undergo late-successional development [4] [5] [20] [21] . Our rationale was that local ecological factors (patch size, isolation, species composition, soils, productivity, and disturbance regimes) were accounted for variability in biomass C stock levels in those forest patches. And, succession has the potential influences to biomass development and C cycling in those forest patches.

Mean maximum AGC stock was recorded from 42.80 Mg ha⁻¹ (SED) to18.93 Mg ha⁻¹ (MMD). The value is quite comparable with other previous estimates of biomass in different forest areas of Tripura. In India, the estimated Forest phytomass carbon density pool for the recent period are mostly in the range of 50 - 68 Mg ha⁻¹. Tripura was having phytomass C density between 0 - 25 Mg ha⁻¹ in 1988 [15] [16] . The cumulative net “C” flux from Indian Forests during 1888-1996, due to land use change (deforestation, afforestation and phytomass degradation) was estimated at 4.54 PgC. Using Biomass expansion factor total estimated biomass of Tripura was about 40 Mt. In Tripura average biomass density ranged from 66.7 to 83.6 Mg ha⁻¹ for open forest and dense forest respectively [16] [17] . While, stocks of C was substantially declined in larger dbh classes; except in case of SED patch, most stock of AGC was restricted in smaller dbh classes (<30 cm) (Figure 2). Comparison of allometric relationships between total AGB and density distribution in different dbh classes in four tropical moist deciduous forest patches implied that there were significant differences among the AGB distribution (Figure 3). Overall, 69.38% (35.73 Mg ha⁻¹) of total biomass was concentrated in <30 cm dbh class and 23% at >70 cm dbh class (12.08 Mg ha⁻¹). The intermediated dbh classes between >30 to <70 cm contributed only 7% of biomass (3.69 Mg ha⁻¹). However in MBB, 90% (37.55 Mg ha⁻¹) of biomass was represented by <40 cm girth class, which indicated that the higher proportion of AGB was contributed by lower dbh classes, particularly in the form of dominated bamboo culms. Biomass distribution in 10 dbh classes was significantly varied (F = 74.13, df = 1.9; p < 0.001), which showed relatively highest coefficient for allometric relationships between density and AGB in different dbh classes (equation: Y = 0.56X − 0.23, R² = 0.90; Figure 3 MBB). In MDS patch, about 90% (36.62 Mg ha⁻¹) of AGB was concentrated in <20 cm dbh class and 8% (3.35 Mg ha⁻¹) between <30 to <60 cm dbh class; but only 1.45% (0.59 Mg ha⁻¹) of ABG represented by higher dbh class (>90 cm). Analysis showed that the allometric equation of total AGB and density distribution highly correlated (Y = 0.56X − 0.20, R² = 0.81; Figure 3 MDS). AGB distribution was significantly higher (F = 34.59, df = 1.9; p < 0.001) among the lower dbh classes which indicated that forest is either immature or recovering its maturity from historic disturbances. About 75% (28.28 Mg ha⁻¹) of AGB was contributed by <20 cm dbh and about 90% of biomass (33.22 Mg ha⁻¹) by <30 cm dbh class in MMD patch. In addition, analysis showed low correlation for the allometric equation (Y = 0.49X − 0.08, R² = 0.76; Figure 3 MMD) of total AGB in MMD and AGB distribution was significantly less in the higher dbh classes (F = 25.25, df = 1.9; p < 0.001). Whether in case of SED patch, only 51% (44.33 Mg ha⁻¹) of AGB was shared by <60 cm dbh class and 39% (33.55 Mg ha⁻¹) contributed by due to >90 cm dbh; whereas total AGB as a function of density distribution in dbh classes showed low correlation (Y = 0.35X − 0.31, R² = 0.48; Figure 3 SED), significantly high biomass storage among the larger dbh classes (F = 7.41, df = 1.9; p < 0.05). Among the tree species, Dipterocarpus turbinatus (43.80 Mg ha⁻¹) accounted highest AGB in SED patch followed by Shorea robusta (23.81 Mg ha⁻¹) for MDS patch, Bambusa tulda (14.46 Mg ha⁻¹) for MBB patch and Terminalia bellirica (2.24 Mg ha⁻¹) for MMD patch (Table 2).

Likewise, Day et al. (2014) also reported high variability in biomass distribution, but in general more diverse forest tends to be accumulating more biomass [35] . In spite of having highest species richness, MMD had lowest value of biomass (37.85 ± 4.02); even, there was also relatively low coefficient (R² = 0.76; p < 0.01) for allometric relationships between density distribution and AGB (Figure 3 MMD). Similarly tree density was

highest for MBB (1088.4 ± 96.15), but it had lower value of biomass (41.84 ± 2.94) showed relatively highest coefficient value (R² = 0.90; p < 0.001) for allometric relationships between density distribution and AGB (Figure 3 MBB). Inverse to that, SED

had highest amount of biomass stock (85.59 ± 17.76) but was relatively lowest correlation coefficient (R² = 0.48; p < 0.05) for equation relationships (Figure 3 SED); which may due to low density and irregular distribution of trees along the dbh classes. This may further be able to explain by the fact that differences in tree composition and dominance of the species along the patches, and more importantly this may related to the forest successional age [10] [20] [21] . On the basis of biomass stocked in all dbh classes, SED patch found different from rest other three types; where trees in higher dbh class >90 cm stocked maximum biomass (40%) and might represented that this forest community has comparatively reached to its maximum potential in terms of sequestration of CO₂. Surprisingly, Bambusa tulda in MBB patch contributed about 35% of biomass, this shows that bamboo patch could capture considerable amount of CO₂ within very early stage. C pool in the above ground biomass of village bamboo in Assam increased from 21.69 Mg ha⁻¹ to 76.55 Mg ha⁻¹ within three years [14] . The recovery of biodiversity and biomass in tropical moist deciduous forest patches provides hope that they are helping to sequester considerable amount of atmospheric C. It is predicted that present trends will change over time when the early stages will grow into more mature layers and subsequently increase total biomass and C stock in those patches in course of secondary succession. In addition, sapling or regenerating woody stems density was high in MMD and MBB patches; which suggested that these forest patches can provide important ecosystem services as C sequestration. Since the regenerating stage of most woody species can sequester considerable amount of CO₂ by reducing it into biomass [18] . Our result also supporting the potentiality of fragmented village bamboo patches for quick C sequestration; and, this is due to bamboo rapid growth rate, easy multiplication ability and high biomass production. In fact, C assimilation in bamboo plantation is 16% - 20% more compared to other planted tree species i.e. Dalbergia sissoo (11.11%), Terminalia arjuna (12.07%) and natural forest of Shorea robusta (3.34%) [14] .