Tree species controls on soil carbon sequestration and carbon stability following 20 years of afforestation in a valley-type savanna

Forest Ecology and Management 291 (2013) 13–19

Contents lists available at SciVerse ScienceDirect

Forest Ecology and Management journal homepage: www.elsevier.com/locate/foreco

Tree species controls on soil carbon sequestration and carbon stability following 20 years of afforestation in a valley-type savanna Guoyong Tang ⇑, Kun Li Research Institute of Resource Insects, Chinese Academy of Forestry, Kunming 650224, China Yuanmou Desertification Ecosystem Research Station, State Forestry Administration of China, Kunming 650224, China

a r t i c l e

i n f o

Article history: Received 26 July 2012 Received in revised form 26 November 2012 Accepted 1 December 2012 Available online 5 January 2013 Keywords: Soil organic carbon Carbon sequestration Carbon stability Afforestation Soil fraction Tree species

a b s t r a c t In the farmland/forest ecotone of southwestern China, many areas are experiencing afforestation and subsequent shift in the ecosystem carbon (C) stocks. The effects of five tree species on soil organic C (SOC) accumulation and C stability following 20 years of afforestation in a valley-type savanna since 1991 were investigated in fractions below 0.25 mm. The bulk soils were fractionated with a combination of density fractionation and acid hydrolysis techniques. The results showed that SOC densities in the afforested stands have accumulated since the changes in the land uses. The surface soil C sequestration rates varied substantially among the five stands and ranged from 0.13 tC ha1 year1 to 0.47 tC ha1 year1 during the two decades of afforestation. The percentage of organic C in the heavy fraction (via density fractionation) relative to the total SOC was 61.9–68.0% under the five stands after 20 years of afforestation. The overall biochemical recalcitrant C density accounted for 37.6–49.9% of the total SOC in these stands in 2011. The tree species controls on the soil C sequestration and on the C in the separated fractions following 20 years of afforestation in a valley-type savanna. However, the biochemical stability of the physically protected C remained lower than that of the unprotected C, irrespective of the tree species. Among the five studied tree species, Leucaena leucacephala was the most suitable tree species for afforestation in the valley-type savanna for the higher C sequestration rates in the soils and for the moderate stability of SOC, which potentially, at least theoretically, lead to the high biological utilisation (e.g. nutrient release) of soil organic matter when SOC decays. Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction The soil organic C (SOC) pool is the largest C store in terrestrial ecosystems, containing a value of approximately 1500 Gt C (Batjes, 1996). Consequently, this C pool has the potential to greatly impact the global climate by acting as either a source or a sink of atmospheric CO2. The effects of afforestation on soil C pool vary. Previous studies have reported accumulation (Trouve et al., 1994; Lopez-Ulloa et al., 2005; Niu and Duiker, 2006; Grünzweig et al., 2007; Laik et al., 2009), losses (Lopez-Ulloa et al., 2005; Richards et al., 2007) and no net change (Epron et al., 2009; Marin-Spiotta et al., 2009) resulting from the establishment of trees. Changes in total SOC with change in land use can be primarily attributed to the amount of plant C transferred to the C-unsaturated soils compared with the previous land uses and can be partly explained by the way in ⇑ Corresponding author at: Research Institute of Resource Insects, Chinese Academy of Forestry, Kunming 650224, China. Tel.: +86 871 63860039; fax: +86 871 63860027. E-mail address: [email protected] (G. Tang). 0378-1127/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.foreco.2012.12.001

which C is allocated to different fractions of soil organic matter (Six et al., 2002; Billings, 2006; Grünzweig et al., 2007; Epron et al., 2009; Laik et al., 2009; Stewart et al., 2009). SOC is protected by a variety of physical and biochemical mechanisms, which produce soil C pools with different residence times. A number of physical and chemical methods have been developed to separate the bulk pool into fractions with different chemical compositions and/or locations in soil matrix (e.g. Janzen et al., 1992; Rovira and Vallejo, 2002; Six et al., 2002; Silveira et al., 2008). Organic C connects with soil mineral particles (clays and fine silts in particular) to form organ-mineral complexes or is concealed within micro-aggregates to make these C partly unavailable for soil microorganisms, which generally causes the organic C in this fraction, designated as the heavy fraction, to become more stable. Some organic C is in a free state, designated as the light fraction, which is characterised as chemically and visually more plant- or litter-like and is commonly considered more sensitive to changes in climate and land use due to the relatively rapid turnover (Janzen et al., 1992; Wang et al., 2005, 2009; Silveira et al., 2008). Soil organic matter is a complex of organic materials with different C components, such as polysaccharides, lipids, lignin, humus, and

14

G. Tang, K. Li / Forest Ecology and Management 291 (2013) 13–19

the native woody plants are mainly Dodonaea viscose, Phyllanthus emblica, Sophora viciifolia and Acacia farnesiana. The typical soils are classified as Ferralic Arenosols according to the FAO Taxonomy (FAO-UNESCO, 1988). 2.2. Experimental details

charcoal. Some C components present in soil, such as polysaccharides, are considered labile because they are easily utilised and then degraded by microorganisms when they are not physically protected, which is the main reason that the labile fraction is expected to respond most rapidly to environmental changes. Biochemical recalcitrant compounds accumulate in soil as soil organic matter decays, especially the readily decomposed matter, and are of important for C sequestration due to their long residence times (Rovira and Vallejo, 2002; Six et al., 2002; Swanston et al., 2005; Sollins et al., 2006; Silveira et al., 2008; Wang et al., 2009). A moderate SOC stability may take into account both C sequestration and utilisation to serve an ecosystem, such as nutrient pools, when SOC decays just because soil organic matter is much more than a potential tank for impounding excess CO2 (Janzen, 2006). Valley-type savanna is a unique type of savanna that develops in deep-incised river valleys, such as those of the dry-hot valleys in the Hengduan Mountains (Fig. 1), southwest China, and the total area of this type in China is approximately 3.0  106 ha (Jin, 2002). In the dry-hot valleys of China, many valley-type savanna areas are experiencing afforestation by a number of tree species. The quantity of SOC under various land uses in these dry-hot valleys has been investigated (Guo et al., 2007; Tang et al., 2010), but there has been no study of the soil C accumulation and its stability following afforestation by a variety of tree species in these valleys. The hypothesis tested was: SOC and C stability in afforested stands are influenced with respect to the planted tree species. Our aim was to evaluate the effects of four specific tree species on: (1) soil C sequestration and (2) the physical and biochemical stability of SOC in fractions below 0.25 mm following 20 years of afforestation in a valley-type savanna.

The stands investigated in this study were located in the forest region of the Yuanmou Desertification Ecosystem Research Station, State Forestry Administration of China (25°400 N and 101°510 E, 1100–1120 m a.s.l.) in Yuanmou County. In 1991, we selected a 25-ha area of wasteland on a barren hill, which is representative of the valley-type savanna. The wasteland had not been cultivated or planted as planting industry for at least 20 years before the beginning of the experiment, and the site conditions of the land were relatively uniform with homogenous soils. The wasteland was located at a mid-slope landscape position with a slope gradient of approximately 5–7°, and it had undergone light erosion due to water flow (Tang et al., 2010). The native vegetation was H. contortus with sparse D. viscose and the vegetation coverage was approximately 35%. The vegetation was disturbed by human activities, such as free grazing and mowing for livestock, which are local practices. Four multipurpose tree species (Leucaena leucacephala cv. Salvador, Acacia auriculiformis A. Cunn., Azadirachta indica A. Juss. and Eucalyptus camaldulensis Dehuh) were introduced and planted in this selected wasteland at a stocking density of 1667 trees ha1 in May–June 1991. A natural succession treatment without tree establishment was also maintained since the beginning of the afforestation. The experiment was arranged in a randomised block design with five treatments (four tree species and one natural succession savanna) replicated five times. Each replicated plot covered an area of 0.81 ha. These plots were protected from anthropogenic disturbances by barbed-wire fences and did not show any replacement of existing vegetation by other tree species during the trees’ growth period. L. leucacephala and A. auriculiformis could naturally regenerate, in part, after 6 and 9 years of afforestation, respectively. The vegetation coverage of the natural succession plots averaged 77% in 2011. The average tree height and diameter at breast height (1.3 m) of these planted trees at different growth stages are shown in Table 1. Five control plots without any protection were established near (approximately 1.2 km) the afforested plots in 2011. The vegetation and edaphic conditions and the anthropic pressure of these control plots highly resembled the savanna prior to afforestation. The area of each control plot approximated 1.0 ha, and the vegetation coverage averaged 37%. One permanent plot with an area of 400 m2 (20 m  20 m) was established in each replicated plot in May 1996 for long-term observation and sampling. In total, 25 permanent plots (five for each individual stand) were established in the five stands.

2. Materials and methods

2.3. Investigation and sample collection

2.1. Site description

The litter horizon was removed prior to the soil sampling. Soil samples from the 25 permanent plots of the five stands were collected randomly using a stainless cylinder from 0 to 15 cm soil depths. One soil composite sample representing each replication was prepared by mixing 12–15 undisturbed soil cores within each permanent plot. In total, five composite samples (one from each permanent plot) were collected in each individual stand. One bulk density sample was taken using a soil core sampler for 0–15 cm depths in each permanent plot. The soils were sampled in April 1991, May 1997, April 2003 and April 2011. Soil samples from the control plots were collected in April 2011. After the removal of the visible plant residues by hand, the soil samples were airdried, sieved through a 2-mm mesh, and then stored at 4 °C. The

Fig. 1. Landscape of a valley-type savanna in reach of the Jinsha River, southwest China.

The study was conducted in Yuanmou County, Yunnan Province, southwest China. The average annual mean temperature of the study area is 21.6 °C with a mean maximum of 27.1 °C in May and a mean minimum of 14.5 °C in December. The average annual precipitation is 634.3 mm, with approximately 92% falling from May through October. The annual potential evaporation is 3911.2 mm, approximately 6.2 times the precipitation. The relative average annual humidity is 53%. The vegetation in the county (below 1600 m a.s.l.) consists of grasses and shrubs with dispersed single trees or clusters of trees. The dominant grass species are Heteropogon contortus, Bothriochloa pertusa and Imperata cylindica, and

15

G. Tang, K. Li / Forest Ecology and Management 291 (2013) 13–19 Table 1 Basic information about the investigated stands at different sampling times. Sampling time April 1991

c

Tree species Natural succession savanna

d

Average tree height (m)

Average DBHa (cm)

Soil bulk density (g cm3)

Soil pH

C:N ratio of litterb

nd

nd

1.68

5.99

nd

May 1997

Leucaena leucacephala Acaia auriculiformis Azadirachta indica Eucalyptus camaldulensis Natural succession savanna

3.8 3.6 2.3 3.3 nd

6.4 10.0 9.1 12.4 nd

1.64 1.67 1.66 1.66 1.66

6.09 6.14 6.21 6.21 6.01

nd nd nd nd nd

April 2003

L. leucacephala A. auriculiformis A. indica E. camaldulensis Natural succession savanna

5.8 5.2 3.2 5.9 nd

11.1 17.8 15.7 20.5 nd

1.58 1.62 1.61 1.60 1.60

6.21 6.25 6.36 6.23 6.14

nd nd nd nd nd

April 2011

L. leucacephala A. auriculiformis A. indica E. camaldulensis Natural succession savanna

7.5 7.8 4.3 9.7 nd

14.2 22.4 18.8 25.8 nd

1.53 1.59 1.58 1.59 1.55

6.20 6.24 6.34 6.31 6.17

21.1 28.5 23.8 32.2 20.7

b a b a b

a

DBH: diameter at breast height (1.3 m). Litter for the measurement of C:N ratio were collected during 2011–2012. c Soils were sampled before the experiment. d Main vegetation in the natural succession savanna are Heteropogon contortus and Dodonaea viscose. nd: no determination. Common letters within a column indicate no significant difference according to the LSD test at a = 0.05. b

subsamples were ground through a 0.25-mm mesh prior to the determination of the SOC. In the four afforested stands, litter samples were collected biweekly from five randomly located baskets over a period of 12 months every 3 years in each individual permanent plot beginning in May 1996. Above-ground H. contortus (grass) withers during the dry season (November to May of the next year) and is then almost completely replaced by new growth in the next wet season. Therefore, the above-ground biomass of H. contortus could be regarded as litter fall under the natural succession treatment because more than 95% of the total litter fall originated from H. contortus in this savanna (Tang et al., 2010). Five subplots, with each area of 9 m2, were settled at the centre and four corners of each permanent plot in the natural succession stand. The above-ground biomass of H. contortus within these subplots were collected by the clip-harvesting method in Nov. (beginning of the dry season) every year since 1996. All litter were returned to the corresponding plots after being weighed. Until to May 2012, the litter fall in the four afforested stands was investigated six times. To compare the afforested stands, the litter fall of the natural succession plot was shown every 3 years rather than yearly (Fig. 2). Because of the grazing and mowing, the litter was not collected from the control

Litter fall (t ha -1 yr -1 )

4

L. leucojphala E. camaldulensis

A. auriculiformis NS stand

A. indica

3

2

1

0 19961997

19992000

200220052003 2006 Investigation time

20082009

20112012

Fig. 2. Litter fall of the studied stands within 12 months every 3 years during different investigation periods. NS stand: natural succession stand.

plots. The litter collected during 2011–2012 was used to determine the C and N concentration.

2.4. Laboratory analysis The physical and biochemical fractions of the soil samples collected in 2011 were isolated with a combination of density fractionation and acid hydrolysis techniques. Two density fractions, the light fraction (LF) and heavy fraction (HF), were separated from the bulk soil following a modified version of the Janzen et al. (1992) protocol. Briefly, 25 g of air-dried soil was placed in a pre-weighed 250-mL centrifuge bottle with 80 mL of NaI solution adjusted to a density of 1.7 g mL1 and gently shaken by hand for approximately 30 s to form a suspension. The suspension was sonicated for 10 min at 200 J mL1 and then centrifuged for 10 min at 4200 r min1. Floating material was siphoned from the centrifuge bottle and then rinsed with deionised water on Whatman G/F filter paper. This procedure was repeated three times. The material collected on the filter paper, designated as the LF, was washed with deionised water into a pre-weighed container. The residue remaining in the centrifuge bottle (HF) was rinsed repeatedly by adding deionised water, followed by sonication, and centrifugation. To quantify the biochemical stability of the C in the separated density fractions (LF and HF), an acid hydrolysis approach modified from Rovira and Vallejo (2002) was used. In brief, an aliquot of the separated density fractions (1000 mg for the HF, 200 mg for the LF) was hydrolysed in a 50-mL sealed Pyrex centrifuge tube with 1 mL of 2.5 mol L1 H2SO4 in a boiling water bath for 30 min. The hydrolysate was recovered by centrifugation and siphonation. The residue was rinsed with 20 mL of deionised water. The material in the rinse water was recovered by centrifugation and added to the hydrolysate. The unhydrolysed residue was dried at 60 °C. An aliquot of 13 mol L1 H2SO4 (2 mL) was added to the tube and the tube was placed in an end-over-end shaker overnight. After the acid was diluted to 1 mol L1 with deionised water, the residue was hydrolysed for 3 h in a boiling water bath. The hydrolysate was recovered as described above. The residue, taken as the recalcitrant fraction (RF), was washed. Then, the recalcitrant C index (IRC, Rovira and Vallejo, 2002) was calculated and the formula of the IRC is shown in Table 2.

16

G. Tang, K. Li / Forest Ecology and Management 291 (2013) 13–19

Table 2 Formulae of soil parameters used in this text. Parameter

Calculation

Dry mass ratio of HF to soil (%) Dry mass ratio of LF to soil (%) HFC content (g kg1) LFC content (g kg1) HFC/SOC ratio (%) LFC/SOC ratio (%) Dry mass ratio of RF to HF (%) Dry mass ratio of RF to LF (%) HF-RC content (g kg1) LF-RC content (g kg1) Overall RC content (g kg1) IRC of HF IRC of LF C density in soil (or in fractions) (tC ha1)

=dry mass of HF (g)/mass of tested soil (g)  100 =dry mass of LF (g)/mass of tested soil (g)  100 =dry mass ratio of HF to soil (%)  C concentration of HF (g kg1)/100 =dry mass ratio of LF to soil (%)  C concentration of LF (g kg1)/100 =HF-C density (tC ha1)/SOC density (tC ha1)  100 =LF-C density (tC ha1)/SOC density (tC ha1)  100 =dry mass of RF in HF (g)/mass of tested HF (g)  100 =dry mass of RF in LF (g)/mass of tested LF (g)  100 =day mass ratio of RF to HF (%)  C concentration of RF in HF (g kg1)/100 =dry mass ratio of RF to LF (%)  C concentration of RF in LF (g kg1)/100 =HF-RC content (g kg1) + LF-RC content (g kg1) =HF-RC density (tC ha1)/HFC density (tC ha1)  100 =LF-RC density (tC ha1)/LFC density (tC ha1)  100 =d  h  c/10, where d is soil depth (15 cm), h is soil bulk density (g cm3), c is C content in soil (or in soil fractions) (g kg1).

HF: heavy fraction; LF: light fraction; HFC: organic C in heavy fraction; LFC: organic C in light fraction; SOC: soil organic C; RF: recalcitrant fraction; RC: recalcitrant organic C; HF-RC: recalcitrant organic C in heavy fraction; LF-RC: recalcitrant organic C in light fraction; IRC: recalcitrant C index.

All soil fractions (LF, HF and RFs) from each separation method were oven-dried at 60 °C to a constant weight and then weighed. These fractions were pulverised to pass through a 0.25-mm mesh and analysed for the C concentration using a VarioEL elemental analyser (Vario-MAX C/N, Elemental Co., Germany). The soil pH was measured with a combination electrode (soil-towater ratio of 1:2). The SOC and C/N in the litter were determined by a dry combustion method using a VarioEL elemental analyser. 2.5. Statistical analyses The formulae of the soil parameters and relative abbreviations are shown in Table 2. The five composite samples were used as replicates to analyse the variations of these soil parameters. All data were tested for normality and homogeneity of variance. The effects of the tree species or sampling times on the soil parameters were tested by one-way ANOVA (LSD test). Paired-samples t-test was applied to analyse the difference in the IRC between the LF and HF. Linear regression was used to test the relationship between the sequestration rates and litter fall and between the C:N ratio of the litter and the C stability. A statistical software package (SPSS 16.0 vesion) was used. 3. Results 3.1. Total soil organic carbon density in the surface soils The surface SOC density in the five stands ranged from 9.38 tC ha1 to 16.18 tC ha1 following two decades of afforesta-

20 17

Ll stand Ai stand NS stand

B

Aa stand Ea stand Control

C accumulation rate (tC ha -1 yr -1 )

Soil organic C density (tC ha -1)

A

14 11 8 5 1991

1997

2003

Investigation time

tion (Fig. 3A). The tree species had significant effects on the SOC density in 2011, except for the differences between the A. indica and E. camaldulensis stands, which were insignificant (P = 0.088). The afforested soils maintained a high SOC density under the L. leucacephala stand (16.18 tC ha1) and low SOC under the natural succession stand (9.38 tC ha1) in 2011. After 20 years of afforestation, the SOC density across all the afforested stands increased by 5.62 tC ha1 compared with the control plot (6.94 tC ha1). The stand age had a significant (P < 0.01) effect on the SOC density under the investigated stands irrespective of the tree species (Fig. 3A). The surface SOC density decreased negligibly by 1.76% and 0.29% under the E. camaldulensis and A. indica stands, respectively, during the first 6 years after the afforestation (i.e., 1991–1997). With these exceptions, an appreciable enhancement of the SOC was measured in all stands after the change in the land use since 1991. On average, the SOC density across all the stands in 2011 increased by 5.75 tC ha1 compared with the initial value of the previous savanna in 1991. For each individual stand, the SOC density in 2011 was significantly (P < 0.01) higher than the previous savanna. The largest impact of afforestation on the surface SOC density was found under the L. leucacephala stand where the density was 2.38 times that in 1991, whereas the least impact was obtained under the E. camaldulensis stand after 20 years of afforestation. Under the natural succession treatment, the surface SOC density had a rather large increase during the last 20 years with an increase extent of approximately 40%. However, the SOC density under the control treatment was negligibly (P = 0.737) increased compared with the former savanna in 1991.

2011

0.7

Ll stand

Aa stand

Ea stand

NS stand

Ai stand

0.5

0.3

0.1

-0.1

1991-1997 1997-2003 2003-2011 1991-2011 Investigation time

Fig. 3. Surface soil organic C density (A) and C accumulation rate (B) under the selected stands at different sampling times. The error bars represent the standard error of the replicates (n = 5). Ll stand: Leucaena leucacephala stand; Aa stand: Acacia auriculiformis stand; Ai stand: Azadirachta indica stand; Ea stand: Eucalyptus camaldulensis stand; NS stand: Natural succession stand. The soil samples from the control plots were only collected in 2011.

17

G. Tang, K. Li / Forest Ecology and Management 291 (2013) 13–19 Table 3 Distribution of surface soil organic C and recalcitrant C density in the density fractions in 2011. Stand type

LFC density (tC ha1)

HFC density (tC ha1)

RC in the LF (tC ha1)

RC in the HF (tC ha1)

Overall RC (tC ha1)

L. leucacephala stand A. auriculiformis stand A. indica stand E. camaldulensis stand Natural succession stand Control All stands (except control)

6.04 ± 0.53a

10.02 ± 0.83a 37.34 ± 1.42a 61.93 ± 1.86b 2.81 ± 0.18a

3.61 ± 0.20b

6.43 ± 0.33b 39.80 ± 1.21b 46.72 ± 2.61bA 36.18 ± 2.18bB

4.20 ± 0.31b

9.50 ± 0.76a 30.06 ± 0.72c 67.95 ± 1.23a 2.28 ± 0.14b

4.69 ± 0.32a

6.97 ± 0.46a

49.93 ± 2.60a 54.37 ± 2.48aA 49.45 ± 3.04aB

3.84 ± 0.23bc 3.38 ± 0.26c

8.22 ± 0.76b 31.28 ± 1.91c 66.75 ± 2.18a 1.79 ± 0.12c 7.28 ± 0.81b 30.96 ± 0.98c 66.49 ± 0.99a 1.87 ± 0.18c

3.13 ± 0.19c 3.49 ± 0.18b

4.92 ± 0.31c 5.36 ± 0.28c

40.03 ± 1.75b 46.67 ± 1.72bA 38.16 ± 2.28bB 49.19 ± 2.28a 55.23 ± 2.10aA 48.30 ± 3.88aB

3.40 ± 0.42c

5.84 ± 0.83c 36.30 ± 1.97a 62.19 ± 1.75b 1.49 ± 0.16d

2.02 ± 0.19d

3.51 ± 0.33d 37.55 ± 1.37b 44.04 ± 3.13bA 34.78 ± 2.54bB

2.36 ± 0.30d 4.17 ± 1.06

4.40 ± 0.47d 34.00 ± 1.83b 63.44 ± 1.77b 1.09 ± 0.16e 8.17 ± 1.70 33.18 ± 3.36 65.06 ± 2.97 2.05 ± 0.49

1.68 ± 0.19e 3.39 ± 0.90

2.77 ± 0.33e 5.44 ± 1.27

LFC/SOC ratio (%)

HFC/SOC ratio (%)

RC/SOC ratio (%)

IRC of LF

IRC of HF

39.91 ± 1.47b 46.24 ± 2.98bA 38.14 ± 1.18bB 43.30 ± 5.58 49.40 ± 5.12A 41.38 ± 6.87B

LFC: organic C in light fraction; HFC: organic C in heavy fraction; SOC: soil organic C. Data are mean ± standard deviation of five replicates, and common letters within a column indicate no significant difference according to the LSD test at a = 0.05. The significant differences in the recalcitrant C index between LF and HF were labelled by different capital letters according to a paired-samples t-test procedure (a = 0.05).

During the entire research period (i.e., 1991–2011), the surface soil C sequestration rates of the five stands ranged from 0.13 tC ha1 year1 under the natural succession treatment to 0.47 tC ha1 year1 under the L. leucacephala stand (Fig. 3B). During the initial 6 years of the research (i.e. 1991–1997), the surface SOC density under the L. leucacephala stand was enhanced at an average rate of 0.32 tC ha1 year1, and then the increases accelerated to 0.53 tC ha1 year1 and 0.54 tC ha1 year1 during 1997–2003 and 2003–2011, respectively. The dynamics of the surface SOC density under the other four stands during 1991–2011 were broadly similar to these under the L. leucacephala stand. 3.2. Soil organic carbon in the density fractions The organic C in the light fraction (LFC) and in the heavy fraction (HFC) differed significantly (P < 0.01) with respect to the tree species in 2011 (Table 3). The highest LFC density was found under the L. leucacephala stand, followed by the A. auriculiformis and A. indica stands, whereas the lowest was found under the natural succession and E. camaldulensis stands. The LFC densities under the five stands were 1.4–2.6 times higher than that under the control treatment. The HFC densities under the five stands ranged from 5.84 tC ha1 to 10.02 tC ha1, which were significantly (P < 0.01) higher than that under the control plot. Regardless of the stand type, the LFC density was significantly (P < 0.01) lower than the HFC density. On average, the LFC and HFC densities across all investigated stands (except the control treatment) in 2011 were 4.17 tC ha1 and 8.17 tC ha1, respectively, which accounted for approximately one-third and two-thirds of the average SOC, respectively (Table 3). The percentage of the LFC relative to the SOC (LFC/SOC ratio) under the control plots was significantly (P < 0.01) lower than those under the L. leucacephala and natural succession stands, but was significantly (P < 0.01) higher than those under the other three stands. In contrast, the ratio of the HFC to the SOC (HFC/SOC ratio) under the L. leucacephala, natural succession stand and control plots were significantly (P < 0.01) lower than those of the other stands. 3.3. Soil organic carbon in the acid hydrolysis fractions The biochemical recalcitrant C (RC) was significantly (P < 0.01) different with respect to the stand type (Table 3). The RC density ranged from 1.49 tC ha1 to 2.81 tC ha1 in the LF and from 2.02 tC ha1 to 4.69 tC ha1 in the HF under the five stands in 2011, which were both significantly (P < 0.01) higher than those

under the control plot. The overall RC density (sum of RC in both density fractions) varied from 3.51 tC ha1 under the natural succession stand to 6.97 tC ha1 under the A. auriculiformis stand in 2011, which accounted for 37.6–49.9% of the corresponding total SOC. A significantly higher portion of the overall RC density was observed under the five stands compared with the control plot (2.77 tC ha1). The percentage of the overall RC to the total SOC (RC/SOC ratio) was higher under the A. auriculiformis and E. camaldulensis stands than under the other treatments. The recalcitrant C index (IRC) of HF was significantly (P < 0.01) lower than that of LF under each individual stand and across all stands taken together after 20 years of afforestation (Table 3). The IRCs of LF and of HF across all stands (except the control treatment) in 2011 were 49.40 and 41.38, respectively. There were no significant differences in the IRC of LF between the A. auriculiformis and E. camaldulensis stands. However, these indices were significantly (P < 0.01) higher than the values under other treatments. Similarly, the IRC of HF under the control plot was equivalent to those under the natural succession, A. indica and L. leucacephala stands, but was significantly (P < 0.01) lower than those under the A. auriculiformis and E. camaldulensis stands. 4. Discussion 4.1. Soil C sequestration potential by afforestation of valley-type savanna The soil C sequestration potential or C stocks varied widely with the climate zones, ecosystems and their succession stages, vegetation types, management practices and soil C saturation deficits (Six et al., 2002; Lal, 2004; Niu and Duiker, 2006; Tan et al., 2007; Stewart et al., 2009). The C sequestration rates of the topsoil approximated 0.06–0.10 tC ha1 year1 for principal terrestrial ecosystems (Fang et al., 2007) and 0.03 tC ha1 year1 for forestland (Piao et al., 2009) during the last two decades in China. Niu and Duiker (2006) predicted the soil C sequestration potential by afforestation of marginal agricultural land in the Midwestern U.S. and reported the sequestration rate as 0.50 tC ha1 year1 during the first 20 years. In this research, we obtained the C sequestration rates as 0.21–0.47 tC ha1 year1 under the stands afforested by four tree species in a valley-type savanna over 20 years (Fig. 3B). These results showed that afforestation had a large potential to act as a sink for additional C in C-depleted soils in this dry-hot valley. In addition to live, dead and under-decomposed roots as well as root exudates, plant litter often is a primary contributor to the accumulation of

18

G. Tang, K. Li / Forest Ecology and Management 291 (2013) 13–19

organic C in forest soils (Lal, 2004; Niu and Duiker, 2006; Tan et al., 2007). The tremendous differences in the soil C sequestration rates among the five stands might be largely attributed to the difference in the litter fall (Fig. 2). There were significant correlations (r = 0.890, P = 0.043) between the C sequestration rates and the litter fall in this study. In general, SOC is decreased after afforestation as a result of the cessation or reduction of litter input from the previous ecosystem while decomposition still occurs, and then the organic C increases gradually with increase in the litter fall (Trouve et al., 1994; Epron et al., 2009). We have observed slight reduction in the SOC under the A. indica and E. camaldulensis stands during the first 6 years after the afforestation (Fig. 3). The valley-type savanna (i.e. the natural succession stand) has also a large potential for soil C sequestration at an average rate of 0.13 tC ha1 year1 over the last two decades (Fig. 3B). This outcome was most likely due to the freedom from human disturbance by barbed-wire fences. The SOC density under the control plots was close to that of the savanna prior to afforestation (Fig. 3A), implying the negative effects of anthropic disturbances (such as grazing and mowing) on soil C sequestration. 4.2. Effect of tree species on the soil C stability protected by physical mechanism The organic C in the HF is physically protected C and plays an important role in the SOC composition and turnover. In contrast, the organic C in the LF is not protected, being located neither within a stable aggregate structure nor in association with silt and clay particles (Janzen et al., 1992). In general, organic C in the LF accounted for 17–47% of the total SOC in the upper 10- to 15-cm layer of ten forested sites in a temperate zone (Christensen, 1992). The LFC/SOC ratios ranged from 9.61% to 11.71% in a calciorthent after 18 years of afforestation by different tree species in a tropical region (Laik et al., 2009). In this study, 30.1–37.3% of the SOC in the fractions below 0.25 mm was unprotected by a physical mechanism after two decades of afforestation (Table 3). The relatively large shares of the LFC in the surface soils was possibly because of the exogenous organic materials that preferentially entered the LF (Billings, 2006; Richards et al., 2007; Tan et al., 2007; Marin-Spiotta et al., 2009) and that subsequently quickly replenished the LFC under the particular environment in this region, such as under water shortages, high air temperatures, low soil fertility, and low litter fall (Guo et al., 2007; Tang et al., 2010). Although the HFC density under the L. leucacephala stand was the highest among the five stands, the HFC/SOC ratio under the L. leucacephala stand, along with that under the natural succession stand, was markedly lower than those of the other stands (Table 3). The low shares of the HFC in soils of the L. leucacephala and natural succession stands might be partly attributed to the low C:N ratios of these litters compared with the litters from the other stands (Table 1), although the HFC/SOC ratios were not significantly correlated (r = 0.772, P = 0.126) with the C:N ratios of the litters in the five stands. More importantly, the leaf morphology of the selected four tree species is obviously different. The leaves of L. leucacephala are microphyllous and thin, whereas the leaves of the three other tree species are mesophyllous and thick except for the thin leaf of A. indica. Microphyllous, thin leaves are usually readily decayed and subsequently transformed to SOC, especially LFC (Lowman, 1988, 1992; Marin-Spiotta et al., 2009). 4.3. Effect of tree species on the soil C stability due to biochemical recalcitrance Swanston et al. (2005) observed a rapid incorporation of 14C label into otherwise depleted HFC and suggested that HF comprises

highly stable organic materials and more recent inputs. The presence of recalcitrant organic C materials in the LF was observed widely (e.g., Swanston et al., 2005; Richards et al., 2007; Marin-Spiotta et al., 2009; Wang et al., 2009). Wang et al. (2005) hypothesised that the high turnover rates of the LFC are not due to the low biochemical recalcitrance of LFC but rather to the free state C that is more readily available for soil microorganisms due to the shortage of physical protection. They further hypothesised that the high stability of the HFC was not a consequence of the higher recalcitrance of HFC but rather the result of the inaccessibility of the HFC. In this stduy, we observed that the biochemical recalcitrance of HFC (i.e. IRC of HF) was significantly lower than that of LFC (i.e. IRC of LF) in each individual stand and across all stands taken together (Table 3). This finding suggested that the biochemical stability of the physically protected C was lower than that of the unprotected C irrespective of the stand types. One probable reason for this outcome is that the rapid turnover of labile C (non-recalcitrant) with respect to RC results in the relative accumulation of RC in the LF (Billings, 2006; Grünzweig et al., 2007; Marin-Spiotta et al., 2009). The second reason is that the soil microorganisms have limited opportunities to access and then utilise organic C in the HF because the physical protection makes the C inaccessible. The presence of these physical barriers potentially stimulate soil microorganisms and consequently increase the utilisation of all organic C in the HF once they have the chance in a situation involving deficiency of energy and C sources (Billings, 2006; Grünzweig et al., 2007; Stewart et al., 2011). The higher concentrations of recalcitrant organic compounds in litter from trees compared with that from pasture always lead to a higher stability of the SOC in forestlands over pastures or savannas (Trouve et al., 1994; Swanston et al., 2005; Billings, 2006; Epron et al., 2009; Marin-Spiotta et al., 2009; Stewart et al., 2011). In this study, we found that the biochemical stability of the SOC under the natural succession plot, which was comparable with those under the A. indica and L. leucacephala stands, was markedly lower than the E. camaldulensis and A. auriculiformis stands (Table 3). The C:N ratios in the litters from the natural succession plot and the planted A. indica and L. leucacephala plots were markedly lower than those from the A. auriculiformis and E. camaldulensis plots. This result might primarily cause the differences in the SOC stability among the five stands. Our data showed that the C:N ratios of the litters were highly correlated with the RC/SOC ratios (r = 0.941, P = 0.017), with the IRC of LF (r = 0.958, P = 0.010) and with the IRC of HF (r = 0.943, P = 0.016) in this research.

5. Conclusions Afforestation resulted in evident enhancements in the SOC in fractions below 0.25 mm with different increases in a valley-type savanna after 20 years. The C sequestration rates of the afforested soils were 1.6–3.6 times to soils under the natural succession condition during the two decades of afforestation. The tree species has significant effects on the total SOC density and on the organic C in soil fractions separated by density fractionation and acid hydrolysis techniques. 61.9–68.0% of the SOC was protected by physical mechanisms, and 39.8–49.9% achieved biochemical stability after two decades of afforestation. The biochemical stability of the physically protected C was less than that of the unprotected, irrespective of the tree species. Among the five studied tree species, L. leucacephala might be the most suitable tree species for afforestation in the valley-type savanna for the higher C sequestration rates in the soils and for the moderate stability of the SOC, which might potentially lead to a high biological utilisation (e.g. nutrient pool) of soil organic matter when SOC decays.

G. Tang, K. Li / Forest Ecology and Management 291 (2013) 13–19

Acknowledgments This work is supported by the National Natural Science Foundation of China (31100462), the Research Institute of Resource Insects, Chinese Academy of Forestry (riricaf201001M, riricaf2012007M) and the Special Fund for Forestry Research in the Public Interest (201104002-3-2). We appreciate the two anonymous reviewers for their valuable suggestions in improving this manuscript. References Batjes, N.H., 1996. Total carbon and nitrogen in the soils of the world. Eur. J. Soil. Sci. 47, 151–163. Billings, S.A., 2006. Soil organic matter dynamics and land use change at a grassland/forest ecotone. Soil Biol. Biochem. 38, 2934–2943. Christensen, B.T., 1992. Physical fractionation of soil and organic matter in primary particle size and density separates. Adv. Soil Sci. 20, 1–90. Epron, D., Marsden, C., Thongo, M’Bou, A., Saint-André, L., d’Annunzio, R., Nouvellon, Y., 2009. Soil carbon dynamics following afforestation of a tropical savannah with Eucalyptus in Congo. Plant Soil 323, 309–322. Fang, J.Y., Guo, Z.D., Piao, S.L., Chen, A.P., 2007. Terrestrial vegetation carbon sinks in China, 1981–2000. Sci. China Ser. D. 50, 1341–1350. FAO-UNESCO, 1988. Soil map of the world: Revised legend, World Soil Resources Report No. 60. FAO, Rome. Grünzweig, J.M., Gelfand, I., Fried, Y., Yakir, D., 2007. Biogeochemical factors contributing to enhanced carbon storage following afforestation of a semi-arid shrubland. Biogeosciences 4, 891–904. Guo, H.Y., Lang, N.J., He, L.P., Zheng, K., Zhang, L.X., Wen, S.L., Jiang, Q.C., 2007. A study on soil characteristics of eight vegetation types in Yuanmou dry and hot valley. J. West China For. Sci. 36, 56–64 (in Chinese with English abstract). Janzen, H.H., 2006. The soil carbon dilemma: shall we hoard it or use it? Soil Biol. Biochem. 38, 419–424. Janzen, H.H., Campbell, C.A., Brandt, S.A., 1992. Light-fraction organic matter in soils from long-term crop rotations. Soil Sci. Soc. Am. J. 56, 1799–1806. Jin, Z.Z., 2002. Floristic features of dry-hot and dry-warm valleys. In: Yunnan, Sichuan (Eds.), Yunnan Science and Technology Press, Kunming, pp. 1–226. (in Chinese). Laik, R., Kumar, K., Das, D.K., Chaturvedi, O.P., 2009. Labile soil organic matter pools in a calciorthent after 18 years of afforestation by different plantations. Appl. Soil Ecol. 42, 71–78. Lal, R., 2004. Soil carbon sequestration impacts on global climate change and food security. Science 304, 1623–1627. Lopez-Ulloa, M., Veldkamp, E., de Koning, H.G.J., 2005. Soil carbon stabilization in converted tropical pastures and forests depends on soil type. Soil Sci. Soc. Am. J. 69, 1110–1117. Lowman, M.D., 1988. Litterfall and leaf decay in three Australian rainforest formations. J. Ecol. 76, 451–465.

19

Lowman, M.D., 1992. Leaf growth dynamics and herbivory in five species of Australian rain-forest canopy trees. J. Ecol. 80, 433–447. Marin-Spiotta, E., Silver, W.L., Swanston, C.W., Ostertag, R., 2009. Soil organic matter dynamics during 80 years of reforestation of tropical pastures. Global Change Biol. 15, 1584–1597. Niu, X., Duiker, S.W., 2006. Carbon sequestration potential by afforestation of marginal agricultural land in the Midwestern US. For. Ecol. Manage 223, 415– 427. Piao, S., Fang, J., Ciais, P., Peylin, P., Huang, Y., Sitch, S., Wang, T., 2009. The carbon balance of terrestrial ecosystems in China. Nature 458, 1009–1013. Richards, A.E., Dalal, R.C., Schmidt, S., 2007. Soil carbon turnover and sequestration in native subtropical tree plantations. Soil Biol. Biochem. 39, 2078–2090. Rovira, P., Vallejo, V.R., 2002. Labile and recalcitrant pools of carbon and nitrogen in organic matter decomposing at different depths in soil: an acid hydrolysis approach. Geoderma 107, 109–141. Silveira, M.L., Comerford, N.B., Reddy, K.R., Cooper, W.T., El-Rifai, H., 2008. Characterization of soil organic carbon pools by acid hydrolysis. Geoderma 144, 405–414. Six, J., Conant, R.T., Paul, E.A., Paustian, K., 2002. Stabilization mechanisms of soil organic matter: implications for C-saturation of soils. Soil Biol. Biochem. 241, 155–176. Sollins, P., Swanston, C., Kleber, M., Filley, T., Kramer, M., Crow, S., Caldwell, B.C., Lajtha, K., Bowden, R., 2006. Organic C and N stabilization in a forest soil: Evidence from sequential density fractionation. Soil Biol. Biochem. 38, 3313– 3324. Stewart, C.E., Paustian, K., Conant, R.T., Plante, A.F., Six, J., 2009. Soil carbon saturation: implications for measurable carbon pool dynamics in long-term incubations. Soil Biol. Biochem. 41, 357–366. Stewart, C.E., Neff, J.C., Amatangelo, K.L., Vitousek, P.M., 2011. Vegetation effects on soil organic matter chemistry of aggregate fractions in a Hawaiian forest. Ecosystems 14 (3), 382–397. Swanston, C.W., Torn, M.S., Hanson, P.J., Southon, J.R., Garten, C.T., Hanlon, E.M., Ganio, L., 2005. Initial characterization of processes of soil carbon stabilization using forest stand-level radiocarbon enrichment. Geoderma 128, 52–62. Tan, Z., Lal, R., Owens, L., Izaurralde, R.C., 2007. Distribution of light and heavy fractions of soil organic carbon as related to land use and tillage practice. Soil Till. Res. 92, 53–59. Tang, G.Y., Li, K., Sun, Y.Y., Zhang, C.H., 2010. Soil labile organic carbon contents and their allocation characteristics under different land uses at dry-hot valley. Environ. Sci. 31, 1365–1371 (in Chinese with English abstract). Trouve, C., Mariotti, A., Schwartz, D., Guillet, B., 1994. Soil organic carbon dynamics under Eucalyptus and Pinus planted on savannas in the Congo. Soil Biol. Biochem. 26, 287–295. Wang, G., Wang, C.Y., Wang, W.Y., Wang, Q.J., 2005. Capacity of soil to protect organic carbon and biochemical characteristics of density fractions in Ziwulin Haplic Greyxems soil. Chinese Sci. Bull. 50, 27–32. Wang, W.Y., Wang, Q.J., Lu, Z.Y., 2009. Soil organic carbon and nitrogen content of density fractions and effect of meadow degradation to soil carbon and nitrogen of fractions in alpine Kobresia meadow. Sci. China Ser. D. 52, 660–668.