Soil Carbon Pool In Different Agroecosystems

e c o l o g i c a l e n g i n e e r i n g 3 4 ( 2 0 0 8 ) 289–299

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/ecoleng

The soil C pool in different agroecosystems derived from the dry tropical forest of Guanacaste, Costa Rica Juan J. Jiménez a,∗ , Rattan Lal a , Humberto A. Leblanc b , Ricardo O. Russo b , Yogendra Raut a a

Carbon Management and Sequestration Center, School of Environment and Natural Resources, The Ohio State University, 2021 Coffey Road, Columbus, OH 43210, USA b EARTH University Guácimo, Limón, Costa Rica

a r t i c l e

i n f o

a b s t r a c t

Article history:

Inventories of soil C pools are still lacking from tropical sites. Our objective was to assess

Received 22 September 2007

total C and N concentrations in the different mineral soil fractions down to 50 cm depth

Received in revised form

in relation to selected physical and chemical properties of 5 ecosystems at La Flor Sus-

31 March 2008

tainable Center in Guanacaste, Costa Rica. The ecosystems studied were a derived savanna

Accepted 1 April 2008

with scattered trees, a gallery forest, an abandoned Mango indigofera L. plantation, a Citrus sp. plantation, and a Saccharum officinarum L. (sugarcane) plantation. Significant differences were found for the main fixed factor ecosystem for all variables analyzed (ANOVA). The TSC

Keywords:

concentration was significantly higher in the sugarcane plantation compared to the rest of

Soil carbon pool

land use systems. The TSC concentration decreased significantly with increase in depth in

Agroecosystems

all ecosystems and ranged from 20.3–38.3 to 4.3–20.9 g kg−1 in the 0–10 and 40–50 cm depth,

Sustainability

respectively. In all cases, the clay + silt fraction (<50 ␮m) contained the highest C concentra-

Dry tropical forest

tion. N concentration (0–10 cm depth) at La Flor ranged from 0.32 to 0.19%, and decreased

Particle-size fraction

in the order sugarcane > Curatella savanna > Mango and Citrus plantations > gallery forest. A

Costa Rica

principal component analysis (PCA) performed with all variables studied showed that the

Between-within PCA

ordination of land uses (ecosystems) in the factorial plane defined by the first two axes was significant (Monte Carlo permutation test, P < 0.0001). The highest TSC pool down to 50 cm depth was obtained in the sugarcane plantation (160 Mg C ha−1 ) while less C was found in the rest of ecosystems, i.e. from 66 (gallery forest) to 80 Mg C ha−1 (Curatella savanna). The TSC concentration obtained in the sugarcane plot is likely the result of the incorporation of surface residues into the soil that would have otherwise been lost through burning, which is the current practice in the region. Further studies on C stabilization in the clay fraction are thus needed to test the hypothesis of soil C enrichment due to residue management. Finally, trade-offs are to be considered for both preservation of the fragile dTf and the productivity of derived land uses that increases soil C at the same time. © 2008 Elsevier B.V. All rights reserved.

1.

Introduction

The seasonally dry tropical forest (dTf) represents 42% of all tropical forests (Brown and Lugo, 1982). This ecosystem is

∗

among the most endangered ecosystems in the world, with less than 0.1% of the original dry forests of Pacific Mesoamerica under protection (Murphy and Lugo, 1986; Janzen, 1988). All of the dTf of Costa Rica is located in the Guanacaste province and

Corresponding author. Present address: Instituto Pirenaico de Ecología-CSIC, Avda. Regimiento Galicia, s/n. Jaca, E-22700 Huesca, Spain. E-mail address: [email protected] (J.J. Jiménez). 0925-8574/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.ecoleng.2008.04.016

290

e c o l o g i c a l e n g i n e e r i n g 3 4 ( 2 0 0 8 ) 289–299

covers an area of ca. 60,000 ha. A serious threat to the preservation of the dTf is the presence of the introduced pyrophyte grass Hyparrhenia rufa (Ness), which aggressively invaded the dTf in the 1940 s (Parsons, 1972) due to its extremely flammable properties during the dry season (Daubenmire, 1972). Global patterns in soil carbon (C) and its vertical distribution across biomes have been compiled and analyzed by several authors (Post et al., 1982; Jobággy and Jackson, 2000). For instance, 50% of the total soil C (TSC) pool up to 1 m in depth is in the upper 20 cm layer in tropical sites. The loss of soil C by conversion of natural vegetation to cultivated use is well known, and a considerable number of studies have assessed the environmental effects of the process in tropical ecosystems (Cerri et al., 1991; Tiessen et al., 1992; Trumbore et al., 1995; Neill et al., 1997; Werner, 1984; Buschbacher et al., 1988; Veldkamp, 1994; de Moraes et al., 1996; Groffman et al., 2001; Powers, 2004; Powers and Veldkamp, 2005). The soil C pool (0–30 cm) in Central America is estimated at 43.3 Mg C ha−1 (FAO, 2005). In the dTf of South America the TSC pool was estimated at 54.7 Mg C ha−1 up to 40 cm depth under savanna vegetation in the Brazilian Cerrado (da Silva et al., 2004). In Argentina, Abril and Bucher (2001) reported a soil C pool of 64.2, 33.8 and 18.3 Mg C ha−1 for the 0–20 cm in the secondary forest, forest-grassland transition (grazing) and heavily grazed grassland, respectively, in the “Chaco” region. These values were similar to those reported by Houghton (1995) for the open forest, i.e. 64 Mg C ha−1 . In Costa Rica, studies have mainly been conducted on soil C pool after conversion of the tropical humid forest into pastures, and the seasonally dTf from the Pacific region have received insufficient attention. Cleveland et al. (2003) reported a TSC pool of 33.8, 37.5, 44.9 and 55.4 Mg C ha−1 in the 0–10 cm depth in southwestern Costa Rica. Consequently, additional studies are needed to continue reporting the soil C pool under a range of different ecosystems and land uses. The TSC pool is composed of different fractions with different residence times. Physical fractionation methods are used to study the dynamics and turnover of organic matter in the soil (Christensen, 2001) and they are useful to isolate C pools more sensitive to changes in land management or differences between ecosystems in order to elucidate processes and mechanisms involved in the storage of C (Six et al., 2002). Sand-size (20–2000 ␮m) macro-aggregates are important in the short-term dynamics of OM, while clay (<2 ␮m) and silt-size (2–20 ␮m) separates are important in the longer term due to the complex associations of C with the structure of clays (Guggenberger et al., 1995). Research data on the soil C concentration within particle-size fractions across a range of different land uses are also lacking in studies conducted in Costa Rica. In this study we selected a range of ecosystems from La Flor Sustainable Center in Northwestern Guanacaste (Costa Rica). Our main objective was first to provide a baseline information on the amount of TSC pool under different land use systems in order to explore the option for which land use fits better in a future scheme of C credits. Secondary objectives in our study were: (1) to assess the vertical distribution of TSC pool up to 50 cm depth, (2) to quantify C concentration in the different mineral particle-size fractions <200 ␮m and, (3) to determine the relationship of TSC with

some selected by means of multivariate ordination techniques (between- and within-class principal component analysis). By so doing meaningful trends and variations in the soil C pool might be identified that may help to adopt better land management decisions while preserving the unique ecosystem of dTf. The search for more sustainable, socially acceptable, economically profitable and environmental-friendly agricultural practices is thus necessary to reduce land degradation.

2.

Materials and methods

2.1.

Study site

The study was conducted in western Guanacaste province at La Flor Sustainable Center, 15 km south Liberia, during August 2005. Yearly average temperature and rainfall is 28 ◦ C and 1530 mm, respectively, with a marked dry season between November and May. Soils at the study site are Inceptisols defined as Typic Ustropept (USDA). Main vegetation types include: (a) mixed deciduous forest with Calycophyllum candidissimum (Vahl) DC. (Rubiaceae), Bombacopsis quinatum (Jacq.) (Bombacaceae) and Luehea candida (DC.) Mart (Tiliaceae) among the dominants, along with fig trees Ficus sp. and rosewood (Dalbergia retusa Hemsl. [Papilonaceae]) also represented; (b) evergreen gallery forests along streams; (c) savannas with exotic jaragua grass H. rufa (Nees) Stapf. and scattered trees of Byrsonima crassifolia (L.) Kunth (Malpighiaceae) and Curatella americana L. (Dilleniaceae); and (d) oak forests with Quercus oleoides Schltdl. & Cham. (Fagaceae) (“encino”) dominant and Acacia collinsii Saff. (Mimosaceae) (Janzen, 2000; Allen, 2001). Five vegetation types were sampled at La Flor: a derived savanna (Sv) of H. rufa with scattered trees of B. crassifolia and C. Americana, which are fire-resistant savanna specialists indicative of prolonged savannisation; a gallery forest (Gf) associated to the “Santa Isabel” river; a 35 years-old Mango indigofera L. commercial plantation (Mg) that was abandoned for 17 years and being thinned by the time field sampling was carried out; a 28 years-old Citrus sp. commercial plantation (Ci), and a 25 years-old Saccharum officinarum L. (sugarcane) plantation (Ca) with no burning of post-harvest surface residues.

2.2.

Litter and soil sampling

A transect line (300 m long) with four sampling points was used in each land use. We followed this procedure to minimize the possible spatial autocorrelation of soil variables. Prior to excavation of pits (approx. 1 m × 0.5 m) the amount of litter in the soil surface was manually collected from 0.5 m2 metal frames. Litter was later oven-dried at 60 ◦ C for 72 h. In each land use soil samples were collected from each sampling point down to 50 cm depth and split in 10 cm increments. Disturbance for both the soil and the site were reduced as much as possible. Approximately 500 g of soil was transferred into plastic bags and air dried for several days. We gently crumbled the soil manually breaking the aggregates along planes of weakness when at field moisture content. After several days soil

e c o l o g i c a l e n g i n e e r i n g 3 4 ( 2 0 0 8 ) 289–299

was dropped onto a hard surface to ease aggregate separation and sieved at 8 mm to remove roots and stones. This soil was used for C, N, pH determinations, textural analysis, and aggregate size-class distribution. Bulk density (d ) was measured in each soil layer by the core method (Blake and Hartge, 1986). Regarding soil physical properties a standard dry-sieving method was followed to assess the distribution of size-class aggregates (Kemper and Rosenau, 1986). From each repetition, a 60 g sub-sample of air-dried soil was passed through a column sieve of 4.75, 2.0, 1.0, 0.5 and 0.250 mm and mechanically moved in a shaker for 30 min. Data were analyzed to compute the mean weight diameter (MWD) using the Eq. (1)

MWD =

n 

x¯ i mi

(1)

i=1

where x¯ i is the mean diameter of each aggregate fraction (2) (mi ) =

Msieve i Mtotal sample

(2)

where Msieve i is the dry mass of the particles retained in the sieve i, and Mtotal sample is the dry mass of the initial total sample. Another 50 g of dry soil from the four samples collected in each land use or ecosystem were combined for texture analysis following the hydrometer method. No treatment was used to remove organic matter. Soil pH was determined in <2 mm air-dried samples in H2 O (1:1) and CaCl2 . Additionally, sample material was collected from the biogenic structures produced by one plant-feeder and one mound-building termite species and one ant species of the genus Atta. This material was only used for C determination.

2.3. C in particle-size fractions (simplified physical fractionation method) Physical fractionation methods have increased steadily to study the factors involved in the associations between soil mineralogy and C, differing in structure and function (Christensen, 1992, 2001). In this study, a simplified physical fractionation method was used to assess C concentration within particle-size fractions in aggregates <200 ␮m. No chemical methods were used to remove organic debris from the mineral fractions. Owing to the specific objective in our study we dispersed soil aggregates followed by wet-sieving in order to release the primary particles, i.e. clay, silt and sand size-classes (Emerson and Greenland, 1990). Briefly, 50 g of <2 mm air-dried soil were dispersed in 50 ml 0.5 m Na-hexametaphosphate +75 ml deionized water for 18 h and mechanically stirred in a multi-mixer machine for 30 min. Later, soil was passed through a nest of sieves of 200, 105, 53, and 20 ␮m to separate the coarse sand (200–105 ␮m), fine sand (105–53 ␮m), coarse silt (53–20 ␮m) and silt + clay (<20 ␮m) fractions, respectively, in beakers that were oven-dried at 60 ◦ C for 72 h. The fraction >200 ␮m was discarded. The <20 ␮m fraction was flocculated with MgCl2 and allowed to settle before discarding the supernatant.

2.4.

291

Carbon and Nitrogen determinations

A sub-sample of the air-dried soil from each sampling plot was passed through a 2-mm sieve, and ground in a mortar and sieved at 200 ␮m prior to C and N analyses. Concentrations of these elements were determined with a CN Elementar Vario Analyzer for both the ground soil and each particle-size fraction. No test was performed to detect qualitatively the presence of carbonates, thus we refer data as TSC. The TSC pool was computed as follows (Batjes, 1996): TSC pool0–50 (Mg ha−1 ) = 0–50 [TSC concentrationlayer (kg Mg−1 ) × (Bulk density)layer (Mg m−3 ) × Depth (m) × −3 −1 4 2 −1 10 Mg kg × 10 m ha ].

2.5.

Statistical analyses

Normality of the data was determined with the KolmogorovSmirnov test. All data were log transformed when necessary to meet the assumption of normality. A two-way analysis of variance (ANOVA-GLM procedure) with ecosystem and depth as the main fixed factors was performed with data collected for TSC and N concentrations, C:N ratio, d , MWD and aggregate size-class distribution. When significant differences were found, multiple comparisons of means were performed with the Fisher probable least-squares differences (PLSD) test. Significant differences in the TSC pool between land uses were performed with a t-test. The R-package 2.4.0 (R Development Core Team) was used to perform ANOVAs and the Sigmaplot 7.1 software (SPSS Inc.) for graph representation. A multivariate ordination technique was used in order to identify the main sources of variation and significance among land uses: the between-within class principal component analysis (PCA). A first PCA named between-class PCA focuses on the differences between groups (Gf, Sv, Ca, Ci, Mg). See Dolédec and Chessel, 1989 for details on the use and interpretation of this procedure. The significance of the ordination was tested with a Monte Carlo randomisation (10,000 simulations) procedure (Manly, 1991). This analysis was later followed by a within-class PCA in order to focus on the remaining variability after the land use (ecosystem) effect has been removed. The higher the inertia obtained in the between PCA the lower in the within PCA will be. Removing the class effect is achieved by placing all centers of classes at the origin of the factorial maps while the sampling units are scattered with the maximal variance around the origin (Dolédec and Chessel, 1991). The within-class PCA gives very similar results (ordination) to a normal PCA (data not shown) and differences between sites would have been masked by those obtained within each site. The discriminant analysis module included in the ADE4 software package was used (Thioulouse et al., 1997).

3.

Results

Soils from the different land uses were slightly acid (5.2–6.9) and with loamy texture (Table 1). There were highly significant differences for the main fixed factor ecosystem for all variables analyzed (P < 0.001, Table 2). Regarding the main fixed factor depth only significant differences were observed for C

292

e c o l o g i c a l e n g i n e e r i n g 3 4 ( 2 0 0 8 ) 289–299

Table 1 – Soil textural analysis (hydrometer method) and pH under the different ecosystems evaluated at La Flor Sustainable Center System

Soil layer (cm)

Texture (%)

Soil type

pH

Sand

Silt

Clay

(USDA)

H2 O 1:1

CaCl2

Curatella savanna

0–10 10–20 20–30 30–40 40–50

62.7 62.7 52.6 46.6 43.0

26.6 26.8 30.0 30.0 32.6

10.7 10.5 17.4 23.4 24.4

Sandy loam Sandy loam Loam Loam Loam

6.0 6.3 6.1 6.3 5.7

5.3 5.4 5.3 5.3 5.2

Gallery forest

0–10 10–20 20–30 30–40 40–50

68.6 69.0 72.8 62.5 62.7

18.0 18.8 14.9 19.2 20.1

13.4 12.3 12.3 18.3 17.2

Sandy loam Sandy loam Sandy loam Sandy loam Sandy loam

6.9 6.8 6.8 6.8 6.9

6.6 6.4 6.3 6.2 6.4

Mango plantation

0–10 10–20 20–30 30–40 40–50

59.4 56.8 46.0 37.1 40.2

25.1 26.6 29.6 29.4 23.5

15.5 16.6 24.4 33.5 36.3

Sandy loam Sandy loam Loam Clay loam Clay loam

6.7 6.7 6.6 6.3 6.3

6.1 6.0 6.0 5.9 5.9

Citrus plantation

0–10 10–20 20–30 30–40 40–50

48.4 45.4 42.5 34.6 28.3

32.1 28.7 26.9 28.8 28.7

19.5 25.9 30.6 36.6 43.0

Loam Loam Clay loam Clay loam Clay

6.1 6.3 6.2 5.6 5.2

5.8 5.8 5.6 5.1 4.7

Sugarcane

0–10 10–20 20–30 30–40 40–50

49.1 45.0 37.4 31.9 37.7

32.6 34.1 35.8 39.4 35.9

18.3 20.9 26.8 28.7 26.4

Loam Loam Loam Clay loam Loam

5.8 6.2 6.3 6.6 6.5

5.5 5.8 6.1 6.0 6.0

and N concentrations and C:N ratio (P < 0.001) and bulk density (P < 0.05). Significant ecosystem × soil depth interactions were also found for C:N ratio and 1–2 mm aggregates (P < 0.05), and for micro-aggregates (P < 0.01). At La Flor, TSC concentration was significantly higher in the sugarcane plantation (Fig. 1) compared to the rest of land use systems (P < 0.001, PLSD test). TSC concentration decreased significantly with depth (PLSD test) in all ecosystems (Fig. 1).

Fig. 1 – Total soil C concentration in <2 mm ground soil (mean ± 1 S.E.) in all ecosystems studied at La Flor. Fisher PLSD critical significance level is 3.53 (P < 0.05).

The TSC concentration decreased significantly with increase in depth in all ecosystems and ranged from 38.3 to 20.3 g kg−1 in the 0–10 cm layer at La Flor. TSC concentration at 40–50 cm depth was 4.3 g kg−1 in the Citrus plantation and the Curatella savanna. Differences in total soil N among ecosystems (land uses) were significant (ANOVA, P < 0.001; Table 3). At La Flor, N concentrations ranged from 0.32 to 0.19%, and decreased in the order sugarcane > Curatella savanna > Mango and Citrus plantations > gallery forest. Total N concentrations varied significantly between the sugarcane plot and the other land use systems (PLSD test). There was a significant decrease in the C:N ratio in the Curatella savanna (P < 0.001, PLSD test). The fractionation of soil within particle-size classes also revealed that the highest TSC concentration was obtained in the < 20 ␮m fraction for all ecosystems evaluated, whereas the coarse-sand (105–200 ␮m) and fine-sand (53–105 ␮m) fractions contained less C (Fig. 2). The variability of total C concentration in the particle-size fractions analyzed is shown in Fig. 3. Since only composite samples were used by combining the four repetitions to obtain one data, for each depth we included all the data obtained from the five land uses. The coarse and finesand fractions showed less data variation than the coarse silt and the silt + clay fractions, especially at deeper layers (Fig. 3, box-plot). This was the general trend observed in all fractions, except for the <20 ␮m fraction. Regarding the two biogenic structures sampled at La Flor, C concentration in the ant hill was similar to that reported in the soil collected at 40–50 cm depth (Fig. 1). Regarding termites, total C concentrations were higher in the termite mound pro-

5114.40 68.19 2180.77 29.08 868.96 11.59

10.68*** 0.78 NS 1.39 NS 25.88 1.61 NS 1.08 NS 29.19 0.31 NS 2.03*

2.00–4.75

*** ***

1.00–2.00

1395.21 18.60 2164.40 28.86 5392.05 71.89 1.70 0.02 27.78 0.37

Mean weight diameter. Bulk density. ∗ P < 0.05. ∗∗ P < 0.01. ∗∗∗ P < 0.001. b

a

75 75 Error SS Error MS

The F-ratios for each variable are indicated. NS: not significant.

60.92 0.81 0.18 0.002

4 4 16 Ecosystem (A) Depth (B) A×B

23.57 0.31

21.35 0.43 NS 1.31 NS 7.76 0.83 NS 0.78 NS 75.24 0.27 NS 2.68** 11.75 3.11* 0.82 NS 21.83 0.36 NS 1.31 NS 19.94 5.47*** 1.83* 33.83 38.19*** 0.60 NS 42.44 36.31*** 0.61 NS

0.50–1.00

*** ***

0.25–0.50 <0.25

*** *** *** *** ***

d b MWDa C:N ratio Nitrogen Carbon d.f. Source of variation

Table 2 – Two-way ANOVA for all variables studied in the five ecosystems studied at La Flor

293

duced by a soil-feeding termite (94.5 g kg−1 ) and a plant-litter feeding termite (488 g kg−1 ) than in the soil (Fig. 1). Regarding soil physical variables, the d in the sugarcane plantation varied significantly (P < 0.05, Fisher PLSD test) when compared to the other ecosystems (Table 4), whereas the MWD of aggregates in the Curatella savanna was significantly different (P < 0.05, Fisher PLSD test) compared to the other systems evaluated (Table 5). Regarding the distribution of aggregates (Fig. 4) significant differences were observed for all cases for the main fixed factor ecosystem. Only for <0.25 and 1.0–2.0 mm size aggregates the interaction between the two sources of variation was significant (Table 1).

3.1.

***

Aggregate size-class distribution (mm)

>4.75

e c o l o g i c a l e n g i n e e r i n g 3 4 ( 2 0 0 8 ) 289–299

Multivariate ordination analysis

Only two axes were retained from the between-class PCA which explained 92.1% of the total variability (Fig. 5a and b). Axis I (58.5% of variability) discriminated soil samples with a high MWD, medium to large aggregates > 0.5 mm and percentage of silt and clay, in contrast to those samples characterized by small aggregates <0.5 mm, high percentage of sand and pH. This axis distinguishes soil samples on the basis of aggregation and texture and thus can be defined as the soil type effect. Axis II (33.6% of variation explained) clearly separated samples with high C and N concentrations, in opposition to those with high bulk density. The projection of objects (land uses) onto the factorial plane defined by axes I and II (Fig. 5c) clearly separated the gallery forest (less disturbed) from the rest of land uses. The Curatella savanna was placed around the origin of the axes, being also the natural environment in the region from which the rest of land uses derive. Both fruit tree plantations, i.e. the Citrus and Mango, were not characterized by high concentrations of C and N, and only the sugarcane was linked to high concentration of C in all fractions, especially the coarse- and fine-silt fractions. The separation of the 5 land uses in the factorial plane of the between-class PCA was highly significant (P < 0.0001) as indicated by the Monte Carlo permutation test. None of the 10,000 random simulations led to an inertia higher or equal to that of the original data (not shown). With regards to the within-class PCA the first and second axes of the within-class PCA accounted for 71.7% of the withinclass inertia, respectively. Axis I (57.2% of variability explained) discriminated those samples with high bulk density and percentage of clay in opposition to those samples with high C concentrations in all fractions and a high percentage of sand (Fig. 6a). Axis II (14.5%) separated large aggregates and the MWD from smaller aggregates <1 mm. This axis shows the differences among land uses regarding soil aggregation and to a lesser extent pH. The projection of the objects onto the factorial plane defined by both axes in the within-class PCA was given in Fig. 6c. The ecosystem effect was removed by placing all centers of classes at the origin of the factorial maps (Fig. 6c–i), while the sampling units (soil from each land use at different depth) are scattered with the maximal variance around the origin. Fig. 6a-ii and a-iii illustrate the ordination of samples by soil depth and soil texture, respectively. While the between-class PCA analysis showed that differences between land uses were basically determined by soil type and man-

294

e c o l o g i c a l e n g i n e e r i n g 3 4 ( 2 0 0 8 ) 289–299

Table 3 – Average N concentration and C:N ratio in all ecosystems studied at La Flor Depth (cm)

Land use (ecosystem) Curatella savanna

Mango plantation

Citrus plantation

Gallery forest

Sugarcane

0–10

N C:N

0.27 (0.05) 11.5 (0.4)

0.21 (0.08) 10.3 (0.4)

0.21 (0.04) 10.8 (0.1)

0.19 (0.10) 10.0 (0.2)

0.32 (0.06) 11.9 (0.2)

10–20

N C:N

0.14 (0.04) 10.1 (0.2)

0.13 (0.02) 10.2 (0.3)

0.13 (0.05) 10.4 (0.3)

0.11 (0.05) 9.8 (0.3)

0.30 (0.07) 11.4 (0.2)

20–30

N C:N

0.09 (0.02) 9.4 (0.6)

0.08 (0.01) 9.8 (0.2)

0.08 (0.01) 9.4 (0.2)

0.08 (0.02) 9.6 (0.2)

0.25 (0.09) 11.4 (0.6)

30–40

N C:N

0.06 (0.01) 9.0 (0.6)

0.06 (0.01) 10.2 (0.2)

0.06 (0.02) 8.7 (0.2)

0.06 (0.01) 9.1 (0.3)

0.20 (0.09) 11.2 (1.0)

40–50

N C:N

0.05 (0.02) 9.1 (0.6)

0.05 (0.01) 10.6 (0.5)

0.05 (0.01) 9.0 (0.6)

0.05 (0.02) 9.0 (0.3)

0.16 (0.07) 12.8 (0.8)

The Fisher PLSD minimum significant difference for both variables is 1.753 and 0.568 (P < 0.05). Standard error within brackets.

Fig. 2 – Total soil C concentration in the different particle-size fractions at La Flor; clay + silt (<20 ␮m), coarse silt (20–53 ␮m), fine sand (53–105 ␮m) and coarse sand (105–200 ␮m).

295

e c o l o g i c a l e n g i n e e r i n g 3 4 ( 2 0 0 8 ) 289–299

Fig. 3 – Box-plots of C concentration in the different particle-size fractions with depth. Each box-plot represents 5 data (land uses) that were obtained from a composite sample of 4 repetitions. Differences were significant according to Kolmogorov-Smirnov one sample test using logistic distribution.

agement, the within-class variability showed a trend in soil texture, with sandy-loam texture soils containing more C in the upper layers than in the more heavy-texture deep soil layers.

3.2.

Soil C pool and litter

Since litter was only sampled once during the rainy season, we are not accounting for differences in yearly litter production. At La Flor the highest quantity of litter was observed in the sugarcane plot due to the leftover of surface residues after harvest (mulching), followed by the Curatella savanna, the gallery forest, and the Mango plantation (Table 6). The highest TSC pool was obtained in the sugarcane plantation, i.e. 160 Mg C ha−1 (Table 6). These pools were significantly different (t-test, P < 0.01) compared to the other ecosystems studied within each site. From 50 to 65% of the total TSC pool was recorded in the first 20 cm.

4.

Discussion

The comparison of our results is limited due to the number of studies conducted in the tropical dry forest and derived land use systems in the region. Johnson and Wedin (1997) reported TSC concentration in the intact deciduous forest at 52.3 g kg−1 (0–15 cm) and 17% lower in the grassland plots (H. rufa) compared to the intact forest (vegetation effect on TSC was significant at P < 0.01) in soils developed from volcanic tuffs at the Lomas Barbudal biological reserve in southern Guanacaste. The TSC concentration obtained in our study was lower and ranged from 30.5 to 38.3 g kg−1 (0–10 cm) under the Curatella savanna and the sugarcane plot, respectively. In southwestern Costa Rica TSC concentration has been reported at 56–68 g kg−1 under derived 20 years-old pasture in very fertile alluvial mollisols (Cleveland et al., 2003). In another study TSC concentrations were given at 32.5 g kg−1 in the dry tropical forest of the Mexican Pacific coast (García-Oliva et al., 2004).

Table 4 – Soil bulk density (b ) down to 50 cm depth Depth

Curatella savanna

Mango plantation

Gallery forest

Sugarcane

Mg m−3

(cm) 0–10 10–20 20–30 30–40 40–50

Citrus plantation

1.2 (0.06) 1.2 (0.05) 1.3 (0.06) 1.5 (0.14) 1.4 (0.09)

1.3 (0.10) 1.3 (0.09) 1.3 (0.11) 1.4 (0.06) 1.4 (0.04)

1.3 (0.08) 1.3 (0.05) 1.4 (0.10) 1.4 (0.05) 1.6 (0.04)

The Fisher PLSD minimum significant difference is 0.095 (P < 0.05). Standard error within brackets.

1.3 (0.05) 1.3 (0.05) 1.3 (0.04) 1.4 (0.04) 1.3 (0.07)

1.2 (0.07) 1.1 (0.03) 1.0 (0.09) 1.1 (0.11) 1.1 (0.06)

296

e c o l o g i c a l e n g i n e e r i n g 3 4 ( 2 0 0 8 ) 289–299

Table 5 – Mean weight diameter (MWD) of soil aggregates down to 50 cm depth in all ecosystems studied in northwestern Guanacaste Depth

Curatella savanna

Mango plantation

(cm) 0–10 10–20 20–30 30–40 40–50

Citrus plantation

Gallery forest

Sugarcane

1.52 (0.35) 1.32 (028) 0.98 (0.24) 0.77 (0.29) 0.70 (0.22)

1.99 (0.14) 2.32 (0.08) 2.40 (0.17) 2.36 (0.12) 2.64 (0.57)

mm 2.09 (0.07) 1.71 (0.18) 1.58 (0.25) 1.57 (0.16) 1.70 (0.06)

2.36 (0.08) 2.18 (0.11) 2.24 (0.09) 2.49 (0.63) 3.32 (0.56)

2.27 (0.08) 2.26 (0.19) 2.88 (0.50) 2.74 (0.45) 2.48 (0.40)

Mean comparisons are only valid for the same site. The Fisher PLSD minimum significant difference is 0.667 and 0.383 for ecosystems at Santa Rosa and La Flor, respectively (P < 0.05). Standard error within brackets.

In general, values reported in these studies were higher than those obtained in our study, being dependent on land use and soil characteristics. The between-within class PCA (Figs. 5 and 6) is a recommendable and a very useful tool in showing the linkages between C and N concentrations and the rest of soil variables measured in this study, and also to identify the differences regarding land uses even if these differed in soil texture. In

our study, the high percentage of sand in the alluvial soil may have limited the concentration of C compared to the other ecosystems (soil type effect). The ordination of ecosystems in the space of the factorial plane defined by the two first axes of between-within PCA indicated a clear and significant separation of ecosystems (P < 0.0001). The Curatella savanna occupied a central position to the factorial plane and the Citrus and Mango plantations were also close to it. The gallery

Fig. 4 – Aggregate size (mm) distribution in the five land uses studied at La Flor.

e c o l o g i c a l e n g i n e e r i n g 3 4 ( 2 0 0 8 ) 289–299

Fig. 5 – Between-class principal component analysis (PCA). (a) Ordination of variables in the factorial plane; (b) “Eigenvalues” diagram; (c) projection of the soil samples onto the factorial plane defined by axes I and II. Circles indicate the center of gravity for each cloud of points in each land use.

forest and the sugarcane plantation showed a farther separation which is explained by both the soil type (texture effect) and management (surface mulch). In our study, the TSC pool in the gallery forest was 66 Mg ha−1 down to 50 cm depth, 17% lower than that in the Curatella savanna with scattered trees, whereas the TSC pool in the fruit tree plantations was 11–14% lower than in the Curatella savanna. Large differences exist regarding the total soil C pool of the gallery forests in different tropical sites. For example, the soil C pool in the gallery forest of the Yucao watershed in the Colombian “Llanos” is reported at 133 Mg C ha−1 down to 1 m depth (E. Veneklass, pers. comm.) and 183.5 Mg C ha−1 (0–50 cm depth) in a gallery forest of the Atlantic humid Costa Rica (Jiménez et al., 2008), which are much higher than the values obtained in this study. The TSC concentration (and pool) in the sugarcane plantation was

297

Fig. 6 – Within-class PCA of the soil variables. (a) Ordination of variables in the factorial plane; (b) “Eigenvalues” diagram; (c) ordination of soil samples with all centroids (groups or ecosystems) at the origin the factorial plane. Variability of scores among sites after land use effect has been removed (i), ordination of samples by soil depth (ii), and by soil texture (iii).

more than twice as much to the other ecosystems at La Flor. The TSC pool in the Mango plantation was slightly higher than in the Citrus plantation (Table 6). Among the factors that determine the trend and rate of accumulation of C are those increasing the litter input and later incorporation in the soil particle-size fractions. The amount of litter in the Mango plantation was three-fold higher than in the Citrus plantation and, subsequently, may have increased the TSC pool. After harvest of sugarcane plant residues are left in the soil surface, and earthworms were abundant together with optimal soil moisture levels compared to the Citrus and Mango plantations at the time of the study (August 2005). The TSC concentration in the sugarcane plot is likely the result of mulching, as reported in other studies conducted in tropical sites (Graham et al., 2002; Razafimbelo et al., 2006), i.e. the incorporation

298

e c o l o g i c a l e n g i n e e r i n g 3 4 ( 2 0 0 8 ) 289–299

Table 6 – Litter and TSC pool (Mg C ha−1 ) in the 5 ecosystems studied at La Flor Sustainable Center Depth

Curatella savanna

Mango plantation

Citrus plantation

Gallery forest

Sugarcane

Mg C ha−1

(cm) In litter 0–10 10–20 20–30 30–40 40–50

4.9 37.6 17.1 10.5 8.4 6.1

4.5 28.9 17.7 9.2 7.7 6.9

1.5 27.5 16.4 10.1 7.2 6.8

3.8 26.9 14.4 9.6 7.6 7.6

Totala

79.7 a

70.4 a

68.0 a

66.1 a

7.6 44.6 38.5 28.5 26.0 22.4 160.0 b

Values followed by the same letter are statistically different at P < 0.01 (t-test). a

Litter not included.

of surface residues into the soil that would have otherwise been lost through burning, which is the current practice in the region (Moreira, pers. comm.) In our study, plant residues (light fraction) were observed at different stages of decomposition even at 50 cm depth in addition to a high earthworm activity. Razafimbelo et al. (2006) showed C enrichment in the clay fraction (0–2 ␮m) compared to the coarse fractions. At La Flor the light fraction in the coarse and fine-sand fractions (Fig. 3) and, especially in the former, was similar in all soil layers studied (management effect). Further studies are needed to (i) assess the contribution of surface residues to the global soil C budget by analyzing the differences in the C4 and C3 concentrations, and (ii) on C stabilization in the clay fraction to test the hypothesis of soil C enrichment due to residue management. Finally, the dTf is among the most endangered ecosystems worldwide. Historic land use conversion in the dTf may have already caused considerable losses of C. The environmental effects of land conversion in the Pacific region of Costa Rica are to be quantified in order to improve decision-making processes aimed at both the preservation of fragile ecosystems as the dTf and sustainable agricultural practices that enable C increases in the soil.

Acknowledgements Financial support during field work was provided by the US Department of Energy. We would like to acknowledge the invaluable help and assistance provided by D. Moreira at La Flor. Thanks to Jean-Pierre Rossi (INRA) for his help in statistical analyses. The first author thanks Fundación ARAID-GA (Spain) for financial support during manuscript revision and two anonymous reviewers for their helpful comments.

references

Abril, A., Bucher, E.H., 2001. Overgrazing and soil carbon dynamics in the western Chaco of Argentina. Appl. Soil Ecol. 16, 243–249. Allen, W., 2001. Green Phoenix: Restoring the Tropical Forests of Guanacaste, Costa Rica. Oxford University Press, New York, NY, 310 pp.

Batjes, N.H., 1996. Total carbon and nitrogen in soils of the world. Eur. J. Soil Sci. 47, 151–163. Blake, G.R., Hartge, K.H., 1986. Bulk density. In: Klute, A. (Ed.), Methods of Soil Analysis. Part 1. Physical and Mineralogical Methods, 2nd ed. Agron. Monogr. 9. ASA and SSSA, Madison, WI, pp. 363–382. Brown, S., Lugo, A.E., 1982. Storage and production of organic matter in tropical forests and their role in the global carbon cycle. Biotropica 14, 161–187. Buschbacher, R., Uhl, C., Serrão, E.A.S., 1988. Abandoned pastures in eastern Amazonia. II. Nutrient stocks in the soil and vegetation. J. Ecol. 76, 682–699. Cerri, C.C., Volkoff, B., Andreaux, F., 1991. Nature and behaviour of organic matter in soils under natural forest, and after deforestation, burning and cultivation, near Manaus. For. Ecol. Manage. 38, 247–257. Christensen, B.T., 1992. Physical fractionation of soil and organic matter in primary particle size and density separates. Adv. Soil Sci. 20, 1–90. Christensen, B.T., 2001. Physical fractionation of soil and structural and functional complexity in organic matter turnover. Eur. J. Soil Sci. 52, 345–353. Cleveland, C.C., Townsend, A.R., Schmidt, S.K., Constance, B.C., 2003. Soil microbial dynamics and biogeochemistry in tropical forests and pastures, southwestern Costa Rica. Ecol. Appl. 13, 314–326. da Silva, J.E., Resck, D.V.S., Corazza, E.J., Vivaldi, L., 2004. Carbon storage in clayey Oxisol cultivated pastures in the “Cerrado” region: Brazil. Agric. Ecosys. Environ. 103, 357–363. Daubenmire, R., 1972. Ecology of Hyparrhenia rufa (Nees) in derived savanna in north-western Costa Rica. J. Ecol. 9, 11–23. de Moraes, J.L.B., Volkoff, B., Cerri, C.C., Bernoux, M., 1996. Soil properties under Amazon forest and changes due to pasture installation in Rondonia, Brazil. Geoderma 70, 63–81. Dolédec, S., Chessel, D., 1989. Rythmes saisonniers et composantes stationnelles en milieu aquatique II. Prise en compte et élimination d’effets dans un tableau faunistique. Acta Oecol. 10, 207–232. Dolédec, S., Chessel, D., 1991. Recent developments in linear ordination methods for environmental sciences. Adv. Ecol. India 1, 133–155. Emerson, W.W., Greenland, D.J., 1990. Soil aggregates—formation and stability. In: de Boodt, M.F., Hayes, M.H.B., Herbillon, A. (Eds.), Soil Colloids and Their Associations in Aggregates. Plenum Press, New York, NY, pp. 485–511. FAO, 2005. Global Forest Resources Assessment 2005. Progress towards sustainable forest management. FAO Forestry Paper 147, 320 p.

e c o l o g i c a l e n g i n e e r i n g 3 4 ( 2 0 0 8 ) 289–299

García-Oliva, F., Oliva, M., Sveshtarova, B., 2004. Effect of soil macroaggregate crushing on C mineralization in a tropical deciduous forest. Plant Soil 259, 297–305. Graham, M.H., Haynes, R.J., Meyer, J.H., 2002. Soil organic matter content and quality: effects of fertilizer applications, burning and trash retention on a long-term sugarcane experiment in South Africa. Soil Biol. Biochem. 34, 93–102. Groffman, P.M., McDowell, W.H., Meyers, J.C., Merriam, J.L., 2001. Soil microbial biomass and activity in tropical riparian forests. Soil Biol. Biochem. 33, 1339–1348. Guggenberger, G., Zech, W., Thomas, R.J., 1995. Lignin and carbohydrate alteration in particle-size separates of an Oxisol under tropical pastures following native savanna. Soil Biol. Biochem. 27, 1629–1638. Houghton, R.A., 1995. Changes in the storage of terrestrial carbon since 1850. In: Lal, R., Kimble, J.M., Levine, E., Stewart, V.A. (Eds.), Soils and Global Change. Lewis Publisher CRC, Boca Raton, FL, pp. 45–65. Janzen, D.H., 1988. Tropical dry forests: the most endangered major tropical ecosystem. In: Wilson, E.O. (Ed.), Biodiversity. National Academy Press, Washington, DC, pp. 130–137. Janzen, D.H., 2000. Costa Rica’s Area de Conservación Guanacaste: a long march to survival through non-damaging biodevelopment. Biodiversity 1, 7–20. Jiménez, J.J., Lal, R., Russo, R.O., Leblanc, H.A., 2008. Soil organic carbon in particle size separates under secondary forests of northeastern, Costa Rica Ecol. Eng. Jobággy, E.G., Jackson, R.B., 2000. The vertical distribution of soil organic carbon and its relation to climate and vegetation. Ecol. Appl. 10 (2), 423–436. Johnson, N.C., Wedin, D.A., 1997. Soil carbon, nutrients, and mycorrhizae during conversion of dry tropical forest to grassland. Ecol. Appl. 7, 171–182. Kemper, W.D., Rosenau, R.C., 1986. Aggregate stability and size distribution. In: Klute, A. (Ed.), Methods of Soil Analysis: Part 1, Physical and Mineralogical Methods, 2nd ed. American Society of Agronomy, Madison, WI, pp. 425–442. Manly, B.J.F., 1991. Randomization and Monte Carlo Methods in Biology. Chapman and Hall, London, 281 p. Murphy, P.G., Lugo, A.E., 1986. Ecology of tropical dry forest. Ann. Rev. Ecol. Sys. 17, 67–88.

299

Neill, C., Melillo, J., Steudler, P.A., Cerri, C.C., Moraes, J.F.L., Piccolo, M.C., Brito, M., 1997. Soil carbon and nitrogen stocks following forest clearing for pasture in the southwestern Brazilian Amazon. Ecol. Appl. 7, 1216–1225. Parsons, J.J., 1972. Spread of African pasture grasses to the American tropics. J. Range Manage. 25, 12–17. Post, W.M., Emmanuel, W.R., Zinke, P.J., Stangenberger, G., 1982. Soil carbon pools and world life zones. Nature 298, 1586–1589. Powers, J.S., 2004. Soil carbon and nitrogen storage following contrasting land-use transitions in Northeastern Costa Rica. Ecosystems 7, 134–146. Powers, J.S., Veldkamp, E., 2005. Regional variation in soil carbon and ␦13 C in forests and pastures of northeastern Costa Rica. Biogeochemistry 72, 315–336. Razafimbelo, T., Barthés, B., Larré-Larrouy, M.-C., De Luca, E.F., Laurent, J.-Y., Cerri, C.C., Feller, C., 2006. Effect of sugarcane residue management (mulching versus burning) on organic matter in a clayey Oxisol from southern Brazil. Agric. Ecosyst. Environ. 115, 285–289. Six, J., Conant, R.T., Paul, E.A., Paustian, K., 2002. Stabilization mechanisms of soil organic matter: implications for C-saturation in soils. Plant Soil 241, 155–176. Thioulouse, J., Chessel, D., Dolédec, S., Olivier, J.M., 1997. ADE-4: a multivariate analysis and graphical display software. Stat. Comp. 7, 75–83. Tiessen, H., Salcedo, I.H., Sampaio, E.V.S.B., 1992. Nutrient and soil organic matter dynamics under shifting cultivation in semi-arid northeastern Brazil. Agric. Ecosys. Environ. 38, 139–151. Trumbore, S.E., Davidson, E.A., de Camargo, P.B., Nepstad, D.C., Martinelli, L.A., 1995. Belowground cycling of carbon in forests and pastures of Eastern Amazonia. Global Biogeochem. Cycles 9, 515–528. Veldkamp, E., 1994. Organic carbon turnover in three tropical soils under pasture after deforestation. Soil Sci. Soc. Am. J. 58, 175–180. Werner, P., 1984. Changes in soil properties during tropical wet forest succession in Costa Rica. Biotropica 16, 43–50.