Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

8/10/2019 Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

1/13

8/10/2019 Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

2/13

D.E. Orihuela-Belmonte et al. / Agriculture, Ecosystems and Environment171 (2013) 7284 73

and can thus play an important role in sequestering atmosphericCO2(Kauffman et al., 1995).

Already in 1980s, tropical secondary forest occupied around600106 ha (35% of total forest; Brown and Lugo, 1990). FAO esti-matedtheareaoccupiedasshiftingcultivationaround360 106 hain the early 1980s (FAO/UNEP, 1982), which represents about 30%of all exploitable soils. In Mexico, the area classified as secondarytropical lowland forest amounted to 19.3106 hain2002(64.5%ofall tropical lowland forest), of which 12.4106 ha are consideredas part of the shifting cultivation cycle (INEGI, 2005).

Assessments of the carbon dynamics in tropical secondaryforests that develop afterabandoning agricultural practices are stillrare (Hughes et al., 1999, 2000; Markewitz et al., 2004; Eaton andLawrence, 2009; De Jong et al., 2010). The potential for tropicalforests to remove C from the atmosphere and store it effectivelydepends on net primary productivity, the allocation of C to the liv-ing plant reservoirs (leaf, branch, stem, root), and the residencetime ofC in theliving anddeadreservoirs(litter, dead woodydebris,soil organic carbon, etc.). A major limitation for our understandingof how forests recover after abandoning agricultural activities, istheshorttime availablefor itsstudyrelative to thegenerallylongertime span over which the process takes place (usually decades).Uncertainties in the rates of secondary forest growth are criticaldata gaps limiting our understanding of the role of tropical forestsassourcesandsinksofatmosphericCO2(Asner,2011;DeJongetal.,2010).

Successful design and implementation of strategies designed toreduce emissions through the forestry sector requires an under-standing of C dynamics including both the size of C pools and therates of C loss or accumulation (Kanninen et al., 2007). More pre-cise data on carbon stocks and carbon stock change over time areof particular interest for countries like Mexico in the light of pos-sible future financial mechanisms that are discussed in relationto reducing emissions from deforestation and forest degradation,or managing and enhancing carbon stocks (REDD+). Currently,five options are considered for national REDD+ approaches. Twooptions require information on the size of carbon stocks: reduced

deforestation and stock conservation. Three options will requireadditional data on carbon stock changes over time: reduced for-est degradation, sustainable forest management and carbon stockenhancements (Herold and Skutsch, 2011). REDD+ is a proposedperformance-based mechanism, in which developed countriescompensate developing countries for forest emissions reductionsthrough a variety of anticipated financial mechanisms, working atdifferent scales (Phelps et al., 2010).

A key question in the international debate concerns the level(scale) at which accounting should be carried out and incen-tives offered for REDD+ activities (see Angelsen et al., 2009). Thescale for accounting of REDD+ mitigation potential has implica-tions for the effectiveness, efficiency and equity of the REDD+financial mechanisms (Wertz-Kanounnikoff and Angelsen, 2009;

Vatn and Angelsen, 2009). It is expected that in most countriesactivitieswillbe implemented at communitylevel. A performance-based system of carbon management through community-levelactivities will require their accountability at the local scale to betransparent and coherent with national accounting (Corbera andSchroeder, 2011) through some type of nested approach (Cortezet al., 2010). Research should therefore focus on how carbonsequestration estimations can be developed thatcan be scaledfromcommunity-based mitigation measurements, through landscape-level assessments to national-scale accounting.

Due to the increased extent of tropical secondary forests,enhancement of forest carbon stocks should be considered as amajor component in the future negotiations of REDD+ (Edwardset al., 2010), for which not only stock data are required but also

estimation of stock change over time (carbon accumulation rates).

At a local scale, carbon accumulation rates in secondary forestsmay vary according to age (time after abandonment; Read andLawrence, 2003), land-use history (number of times converted tocrop land;Van derWal, 1999; Eaton andLawrence, 2009), localsiteconditions (environmental conditions, physical and chemical char-acteristics of soils; Batjes and Sombroek, 1997; Feller and Beare,1997; Lorenz et al., 2009) or a combination of both. To analyzethe rates at the landscape level, it is expected that the main factorinfluencing the carbon accumulation rate will be the area occupiedby secondary vegetation, its age structure and LU-history (Eatonand Lawrence, 2009), whereas for national accounting the amountof secondary forest will be the most important parameter used toestimate carbon accumulation rates (De Jong et al., 2010).

Inthispaperwepresent theresults of a study that collecteddataon C pools in a Mexican tropical secondary forest chronosequence.We used data collected in 87 plots spread to quantify the carbonstocks and rates of accumulation in living and dead biomass andsoil pools, and to evaluate the influence of factors such as age, sitequality and land-use intensity (prior land-use history) on the sizeof the stocks and accumulation rates. We examined the combinedinfluence of these factors on the magnitude and rate of change oftheC pools (vegetation,dead wood, litter, soil) in a secondaryforestchronosequence. Finally we propose a nested approach in whichdetailed studies, such as presented in this paper, can be up-scaledto landscapeand regional estimates of carbonpools in forestfallowand undisturbed forests and carbon accumulation rates in forestfallow.

2. Methods

2.1. Study area

The study was carried out in the El Ocote Biosphere Reserve,located in the north-eastern part of Chiapas, Mexico, between164542 and 170900 North and between 935419 and932120 East. The reserve extends about 100,000ha, of whicharound 40% is completely protected, whereas the remaining 60% is

considered as buffer zone with high conservation importance dueto the high number of endemic species (Fig. 1). The whole area ispartoftheMeso-AmericanBiologicalCorridor(CONANP,2000). Thereserve is located between two neo-tropical provinces, the Pacificand Tehuantepec, characterized by a highly varying topographywith altitudesbetween180 and1300m. a.s.l. The climate is tropicalhumid (20002500mm yr1) to subhumid (15002000mmyr1).The area is characterized by a karstic landscape with its sink-holes, hills, field depressions and caves (Mulleried, 1957). The soilsare predominantly of alluvial origin, sedimentary rocks, limestone,sandstone, mudrock, limonite, intrusive rocks and conglomerates(Leptosols, Luvisols and Regosols). The original vegetation is trop-ical evergreen and semi-evergreen rainforest. Colonization of thearea started in the 1970s and since then part of the original forests

has been converted to secondary forests and agricultural land dueto slash-and-burn agriculture. This type of agricultural produc-tion is a cyclic system that includes a short period of corn andbean cultivation followed by a fallow period that formerly lastedabout 10 years, but which in recent years has often been reducedto two to four years due to land-use intensification (CONANP,2006).

2.2. Sampling

We established 87 monitoring plots of 1000m2 in medium-stature semi-evergreen rainforest in seven communities locatedaround the border of the biosphere reserve. Of these, 63 plots werelocated in forest fallows recovering between 1 and 40 years from

cultivation, and 24 in undisturbed forests (Fig. 1).

8/10/2019 Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

3/13

8/10/2019 Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

4/13

D.E. Orihuela-Belmonte et al. / Agriculture, Ecosystems and Environment171 (2013) 7284 75

Table 1

Parameters and values used to calculate thesite quality indices.

Parameter Value

1 2 3 4 5

Site Quality 1Nitrogen content (%) 0.221Phosphorous content (ppm) 30Cation exchange capacity (cmolkg1) 40Calcium carbonate content (%) 05 5.110pH 8.3 7.48.3 6.17.3Soil texture Sandy Clayey Sandy clay Silt loam Silt loam

6 7 8Sandy clay loam Clay loam Frank

Site Quality 2Percent of area covered by rocks 50100% 2650% 1125% 110% 0%Position on hill side Top Medium Bottom FlatOrigin of fire Intentional Natural No fireSlope (%) 50% 4050% 3039% 2029% 019%Average soil depth

(5cm classes)05 cm 5.110 10.115 15.120 20.1256 725.130 >30

where LU-Intensity is the land-use intensity; t is the number ofcycles; C is the years in cultivation in each cycle; F is the years

under fallow in each cycle.We developed two site quality indices (Table 1), one derivedfrom chemical and physical characteristics of the soil sample (SiteQ1) and one derived from a series of visible parameters recordedin each plot (Site Q2). The chemical and physical characteristicsof the soil samples included in Site Q1 are: nitrogen content (%),phosphorous (ppm), Cation Exchange Capacity (CEC; cmolkg1),considering that CEC values below 5.0cmolkg1 indicates a degreeof infertility and higher CEC indicates better availability of nutri-ents (FAO, 1979), Calcium Carbonate (%), considering that in thepresence of carbonate the clay complex is dominated by exchange-able Ca, which in turn favours soil physical conditions, pH, takinginto account that most plant nutrients are best available in a neu-tralpH(Landon, 1991), and soil texture, where we consider mainlythe water availability and drainage capacity. The Site Q1 was cal-culated separately for each soil depth (010, 1120, 2130). Theenvironmental variables taken into account in Site Q2 include: per-cent of area covered by rocks, position on hill side, slope, origin offire (induced or natural), and average soil depth in 5 cm classes.

The scores of each factor were summed for each plot, withthe final score of Site Q1 theoretically ranging from 6 (all val-ues 1; lowest site quality) to 27 (all values at max; highest sitequality) and with the final score of Site Q2 ranging from 5 (all val-ues 1; lowest site quality) to 19 (all values at max; highest sitequality).

In Table 2 we present the average, maximum andminimum val-ues of LU-intensity, Site Q1 and Site Q2 for each forest fallow age

class and undisturbed forest. As expected, the average LU-intensityindex decreases from the low to high age categories, as older veg-etation is indicative of a less frequent use of land for agriculture,taking into consideration that the area has been colonized for onlyabout 40 years. Site Q1 and Site Q2 do not show a systematic dif-ference between low and high age categories.

2.3. Statistical analysis

The C-pools were analyzed independently and in combination:carbon (C) in living tree biomass (TreeB) is combined with C inliving non-tree biomass into C in total living biomass (TLB); C inTLB combined with C in dead biomass (woody debris and litter)into C in total living and dead biomass (TLDB), C in TLDB withsoil carbon into Total C. Stocks of the pools were calculated foreach plot. Average increments of C in living and dead pools wereestimated for each plot, dividing the pool sizes by the number ofyears of recovery for C in living tree biomass (TreeB Incr), total liv-ing biomass (TLB Incr), and total living and dead biomass (TLDBIncr).

We tested for differences in stocks and accumulation ratesat three levels of plot aggregation: (1) all forest fallow plotsaggregated together versus plots located in undisturbed forest,(2) the forest fallow plots grouped in age class categories and(3) multiple linear regression analysis with age, LU-intensity

Table 2

Average, maximum and minimum values of Land-use intensity and site quality indices.

Age class Undist. forest

05 610 1115 1620 2125 2630 3140

Number of plots 14 15 13 5 6 5 5 24Land-use intensity

Average 1.12 0.73 0.40 0.25 0.19 0.06 0.03 n.a.Max 2.57 1.60 0.64 0.40 0.32 0.12 0.03 n.a.Min 0.40 0.40 0.12 0.11 0.04 0.03 0.03 n.a.

Site Quality 1Average 20.3 20.7 20.1 21.2 19.5 21.8 19.0 18.8Max 23 24 23 24 22 24 22 22Min 18 18 17 19 16 19 16 16

Site Quality 2Average 12.3 9.3 9.5 10.0 9.7 12.7 9.7 10.7Max 18 13 17 13 14 18 13 15Min 7 6 7 9 7 10 6 8

Undist. Forest, undisturbed forest; n.a., not applicable; max, maximum value observed;min, minimum value observed.

8/10/2019 Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

5/13

8/10/2019 Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

6/13

D.E. Orihuela-Belmonte et al. / Agriculture, Ecosystems and Environment171 (2013) 7284 77

Table 4

Average carbon content (MgC ha1) in themain pools of forest fallow and undisturbed forests in theOcote, Mxico.

Secondary forest n= 63 Undisturbed forest n= 24 Sign. Diff.

Tree biomass (TreeB) 59.29 (8.96) 104.03 (14.52) *Total living biomass (TLB) 60.26 (8.99) 105.29 (14.56) *Litter 7.46 (0.84) 10.13 (1.37) *Total dead biomass (TDB) 11.08 (1.83) 16.71 (2.97) *Living + dead biomass (TLDB) 71.34 (9.63) 121.99 (15.61) *Soil organic carbon(030cm) 74.93 (+9.8; 8.7) 74.80 (+12.2; 10.5) N.S.

TotalC 155.37 (12.78) 191.47 (16.80) **p< 0.05; between parenthesis 95% confidence interval. Soil organic carbon derived from log-transformed average and confidence interval.

mean carbon density in the age classes varied between 115.4 MgCin forests of 15 years old, to 211.5 MgC in forests of 2530 yearsold. Thecontribution of tree biomass to total C-contentin theforestecosystems varied between 21% in young secondary forests to 53%in undisturbed forest (Tables 4 and 5). The magnitude of the otherC pools varied little in absolute terms, although their proportion oftotal C decreased along the age gradient, as carbon in tree biomassincreased.

3.5. Correlation between carbon pool size and accumulation rates

in relation to age, LU-intensity and site qualities

All biomass carbon pools were positively correlated with age(Table 6). TreeB, TLB, TLDB, Total C were negatively correlated withLU-intensity, whereas TreeB, TLB, and TLDB were also negativelycorrelatedwith Site Q1 (chemical and physical soil characteristics).A positive correlation was observed between Total C and site Q2(environmentalcharacteristics). Both DWDandlitter C stocks werepositively correlated with age, whereas DWD was negatively cor-related with LU-intensity. Site Q2 was only positively correlatedto all living and soil pools in undisturbed forest (p< 0.05; resultsnot presented). SOC content in the first 20cm is positively corre-lated with both site quality indices, whereas SOCat 2130cm depthwas only positively correlated to Site Q2. All calculated accumula-

tion rates were negatively correlated with age and positively withLU-Intensity, whereas all rates that include tree biomass (TreeB

Incr, TLB Incr, and TLDB Incr) were also positively correlated withSite Q1. None were correlated with site Q2.

3.6. Carbon accumulation rates in secondary forest

We calculated the accumulation rates also at three scales:regional, landscape and local. To estimate the average accumula-tion rate at the regional level, we modeled C in total living biomassagainst age with a ratio estimator (linear regression through ori-gin of total living biomass of each plot against its age). The

forest fallow accumulates on average 3.600.41Mg C ha1 yr1(Fig. 3a). Young fallows (

8/10/2019 Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

7/13

78 D.E. Orihuela-Belmonte et al. / Agriculture, Ecosystems and Environment171 (2013) 7284

Table 6

Correlation coefficients of C pools and accumulationrates of secondary forests with independent variables.

Age LU-Intensity Site Q1 Site Q2

Pool or pool change ra pb r p r p r p

TreeB 0.63

8/10/2019 Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

8/13

D.E. Orihuela-Belmonte et al. / Agriculture, Ecosystems and Environment171 (2013) 7284 79

Table 7

Multiple linear regression between C pools and accumulation rates with indepen-dent variables.

Independent variables

Age LU-Intensity Site Q1 Site Q2

C C C C

Dependent variables (s.e.) (s.e.) (s.e.) (s.e.)

TreeB 2.18 9.70NS

1.63(0.40) (8.63) (0.48)TLB 2.19 9.39NS 1.66

(0.40) (8.21) (0.48)TLDB 2.35 9.20NS 2.08

(0.43) (8.96) (0.52)Litter 0.45 0.17NS

(0.07) (0.13)DWD 0.17 0.06NS

(0.06) (0.05)TDB 0.42 0.15*

(0.07) (0.06)SOC010 6.39* 0.76** 3.11

(3.62) (0.32) (0.61)SOC1120 2.57

(0.15)SOC2130 0.59 2.27

(0.21) (0.36)TreeB Incr 0.10** 0.23 0.18NS

(0.04) (0.08) (0.14)TLB Incr 0.11 0.22 0.21NS

(0.04) (0.08) (0.14)TLDB Incr 0.11* 2.07* 0.24** 0.19NS

(0.06) (1.21) (0.09) (0.15)

Allequationsand coefficients(C) have a probabilityof errorof less than 1%(p

8/10/2019 Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

9/13

8/10/2019 Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

10/13

D.E. Orihuela-Belmonte et al. / Agriculture, Ecosystems and Environment171 (2013) 7284 81

0 1 - 2 3 - 4

Number of cycles

0

5

10

15

20

25

Totaldeadbio

mass(MgC

ha-1)

Median and non-outlier range

Fig. 7. Dead biomass stocks in relation to number of agricultural cycles.

Uhl et al. (1988) reported that secondary forests of sites that had

experienced relatively light intensities of land use had higher ratesof aboveground biomass accumulation than those with high LU-intensity.In our case,the positive correlation between LU-intensityand biomass accumulation rates, combined with the positive cor-relation between the index and Site Q2 generates the idea thatfarmers may be able to select plots with higher quality for moreintensive slash-and-burn agriculture, where logically carbon inbiomass accumulates faster after abandonment, although we didnot test this hypothesis.

4.2. Soil

The transformation of forest land into agricultural land leads tosubstantial losses of the above-ground biomass, but its effect on

soil C stocks remains poorly understood. It is commonly assumedthat after conversion to agricultural use the soil is impoverishedin C which is reversed during the fallow period (Greenland andNye, 1959; Raich, 1983; Cerri et al., 1991; Lugo and Brown, 1992).In contrast with our expectations, soil C storage from 0 to 30cmin secondary forests did not increase with age or decrease due toconversion to secondaryforest. Within each depth class, secondaryforest soil C storage was roughly the same and did not show anyrelationship with increased aboveground C storage. The mean SOC

storage of all fallow sites together was similar to SOC in primaryforest (74.9 t C ha1 in secondary plots, compared to 74.8t C ha1

in the primary forest plots, Table 3) and there was no statisticaldifference in total SOC or in SOC at any of the three depth classes.

Hughes et al. (1999) found that soil C pools (to 1m depth) ofhumid forests ranged from 139 to 249 Mgha1, and that patternsof variation were not related to either forest age or duration of landuse, similar to our findings. Based upon analysis of soil pools inAmazonia,Hughesetal.(2002)and McGrathetal.(2001) concludedthat soil C (and nutrient) pools were most strongly and positivelycorrelated with clay content. Contrary to these findings, our datapresented a negative correlation between C-content (in % of vol-ume) and clay content at all three layers (r=0.37 at 010cm,0.43 at 1120 cm, and 0.33 at 2130 cm;p

8/10/2019 Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

11/13

82 D.E. Orihuela-Belmonte et al. / Agriculture, Ecosystems and Environment171 (2013) 7284

be used successfully to estimate biomass parameters over largerareas. However, if interest is vested in predicting biomass for par-ticular sites,it maybe more fruitful to useadditional data collectionthat account for specific site conditions or disturbance events. Thisis indicated by the large variation observed in Fig. 3, which can-not be explained by a simple linear model. Through the alternativemethods we would obtain propositions in terms of probabilitiesof finding certain values of biomass parameters for site parame-ters independent of age such as land use intensity, soil depth, sitequality, etc. A scale-dependent combination of forest type, land-scape and site-specific models allows adapting local estimationsto landscape and forest type level estimations, while conservinguncertainties in the estimations (Table 9).

5. Conclusions

REDD+ is a proposed performance-based mechanism undernegotiation through the United Nations Framework Conventionon Climate Change (UNFCCC), which will require demonstratedemission reductions through improved forest conservation, sus-tainable forest management, and/or enhancementof carbon stocks.They suggest that research should focus on how carbon seques-tration varies with differing levels of community engagement andautonomy, and across diverse tenure regimes and landscapes. Akey question in the debate concerns the level (scale) at whichaccounting needsto becarried outandincentives offered forREDD+activities. The scale for accounting of REDD+ has profound impli-cations for the effectiveness, efficiency and equity of the REDD+mechanism (Angelsen et al., 2009). As activities will be imple-mented at a local scale, if performance-based financial systemswill be developed for REDD+, the countries will need to be ableto up-scale estimations over time and space from local (com-munity), through subnational landscapes to national accountingsystems.

One of the largest sources of uncertainty in GHG accountingrests in the spatial variation of aboveground carbon stocks and

stock changes in primary forests, secondary forests and on fallowland (Asner, 2011). In countries like Mexico land tenure is suchthat REDD+ activities will be implemented at a local scale wherelandowners will be invited to change their land-use strategies.Particularly if the country implements performance-based stockenhancement activities at this scale, factors that may affect car-bon accumulation rates have to be taken into account. These inturn need to be up-scaledto landscapeand regional estimates withcost-efficient methodologies.

As age of secondary vegetation is the main factor explainingthe variation in C-stock and accumulation rates independent of thescale of analysis, this can be used as the basis to scale the estima-tions from community level, through landscapes up to forest type(Table 9).

The information required at the community level would be theage distribution of secondary vegetation and their LU-intensity(which is correlated with environmental site characteristics of SiteQ1), andsamplescould betakento estimatethe variationin texturaland chemical characteristics of the soils, which in general is costlydue to laboratory procedures that are expensive. At the landscapescale, the age class distribution and LU-intensity would be ade-quate to estimate the C-accumulation potential, these data can bederived from time series analysis of satellite imagery (e.g. Hayesand Sader, 2001; Yang and Lo, 2002; Lu et al., 2004). The totalarea of secondary vegetation would be the source of information,derived from land-use maps or satellite imagery, to estimate therate of C-accumulation in secondary forests within a forest type. Asdatarequirements and modeling complexity increases from region

to community, the uncertainty analysis required to estimate the

confidence intervals of the results becomes more complex, fromsimple 95% confidence intervals of average biomass densities oraccumulation rates at the regional scale to complex Monte Carlosimulations at the community scale (Table 9).

The secondary medium stature rainforest covers about8.7106 ha in Mexico. Of these, an estimated 900,000 h a arelocated within the 15002500 rainfall zone, with an estimatedpotential C-accumulation rate of 3.600.41MgCha1 yr1 (thisstudy) and 7.8106 ha are located within the 9001500mmrainfall zone, with an estimated C-accumulation potential of1.520.24Mg C ha1 yr1 (Read and Lawrence, 2003). If theseareas were to be managed as part of the REDD+ stock enhance-ment activity,the C-accumulationpotential of the wholesecondaryforests is estimatedat 56.78.4TgCO2 yr1, which is equivalent tomore than 65% of thenet annualemissions between 1993 and2002of the whole LULUCF sector of Mexico (De Jong et al., 2010). Sec-ondary forest is considered by farmers as temporal land reservesthat can rapidly and readily be converted to other land uses asa response to demands of certain products or services (e.g. bio-energy, GHG mitigation, food, timber). As such, management ofthese forests for GHG mitigation maybe regarded as a key strategyfor Mexican REDD+ policy.

Acknowledgements

We gratefully thank two anonymous reviewers for their valu-able comments, which improved the paper substantially. All theinstitutions and persons who helped carry out this study, in par-ticular Ing. Roberto Escalante and MC Karla Leal of Conanp, Biol.Froiln Esquinca Cano of SEMAHN, Dr. Cristian Tovilla Hernn-dez of ECOSUR-Tapachula. Financial assistance wasobtained fromCONACyT through a scholarship for DEOB. Financial assistance ofUSAID/ABT through Ambio facilitated the fieldwork, for which weare very grateful. Additional funding was obtained from CONAFORfor complementary fieldwork and laboratory analysis. We are par-ticularly grateful to all the people of the communities Armando

Zebada, Nuevo San Juan Chamula, Tierra Nueva, Veinte Casas,Emiliano Zapata, EmilioRabasa andEncajonado to allow us to workin their plots, to live in their houses and to assist with the field-work, in particular Manuel Ruz, Manuel Daz, Javier Hernndez,Eduardo Hernndez, Jos Antonio Gmez, Sal Argueta and DavidHernndez, and finally we thank Margeli Nuril Acero and Diana L.Rodrguez O. for their help in the laboratory.

References

Anderson, J.M., Ingram, J.S.I., 1993. Tropical Soil Biology and Fertility: A HandbookofMethods. CABI, Wallingford, RU.

Angelsen, A., Brockhaus, M., Kanninen, M., Sills, E., Sunderlin, W.D., Wertz-Kanounnikoff, S. (Eds.), 2009. Realising REDD+: National Strategy and PolicyOptions. CIFOR, Bogor, Indonesia, p. 362.

AOAC, 1980. Official Methods of Analysis of the Association of Official AnalyticalChemist, 15th edition. Association of Official Analytical Chemist, Washington,D.C.

Asner, G.P., 2011. Painting the world REDD: addressing scientific barriersto monitoring emissions from tropical forests. Environ. Res. Lett. 6, 13http://iopscience.iop.org/1748-9326/6/2 (revised 11.02.13).

Batjes,N.H., Sombroek,W.G., 1997. Possibilities for carbon sequestration in tropicaland subtropicalsoils. Global Change Biol. 3, 161173.

Brown, J.K., Roussopoulos, P.J., 1974. Eliminating biases in the planar intersectmethodfor estimating volumes of small fuels. Forest Sci. 20, 350356.

Brown,S., 2002. Measuring carbonin forests: current status and future challenges.Environ. Pollut. 116, 363372.

Brown, S., Gillespie,A.J.R., Lugo,A.E., 1989. Biomass estimationmethodsfor tropicalforests with applications for forest inventorydata. Forest Sci. 55 (4), 881902.

Brown,S.,Lugo,A.E.,1992. Aboveground biomass estimatesfor tropical moistforestoftheBrazilian Amazon. Interciencia 17 (1), 818.

Brown,S., Lugo, A.E., 1982. Thestorage andproduction of organic matterin tropicalforests and their role in theglobal carbon cycle.Biotropica 14, 161187.

Brown, S.,Lugo, A.E., 1990. Tropical secondaryforests. J. Trop. Ecol. 6, 132.

8/10/2019 Carbon Stocks and Accumulation Rates in Tropical Secondary Forests At

12/13