Land Use and Soil Texture Effects on Organic Carbon Change in Dryland Soils, Senegal ()
1. Introduction
The establishment of the Kyoto Protocol has resulted in a focus on biomass and soils for C sequestration. In order to make C credited meaningful, sustained C sequestration for decades or longer period is required.
Soils contain 1500 - 2000 Pg C, representing the third reservoir of C after ocean and fossil fuels [1]. Soil organic carbon (SOC) storage is chiefly controlled by primary production input and organic matter evolution. Changes in land use affect the amount of carbon stored in vegetation and soils and, hence, the flux of carbon between land and the atmosphere [1]. Increase of primary production will result in an increase of C storage, whereas increasing decomposition (i.e., reduce C turnover time) will have an opposite effect [2]. The organic matter input to arable soils is lower than to forest soils because a greater proportion of the produced biomass is harvested in the former.
Many factors (climate, land use, soil moisture, soil texture, clay content, mineral, erosion, leaching, fire, etc.) play a role in regulating the accumulation and loss of SOC storage [3,4]. The dynamic of soil carbon storage and release is complex and still not well understood [2]. Delta 13C (δ13C) values have been utilized to document vegetation change and to quantify soil organic matter turnover [5]. The majority of tropical fodder and grass of savanna or steppe are C4-plants with δ13C values in the range of −8% to −18%. Most notable are corn, sorghum, millet and sugar cane [6,7]. C3-plants are more characteristic of the tropical forests where the δ13C varies between −22% and −35% [7], including groundnut plants. Delta 13C values are not altered significantly during decomposition and soil organic matter formation, essentially into the fine-silt-sized fraction [6,8,9]. Consequently, δ13C values of SOC reflect the relative contribution of plant species to net primary production.
In the West Africa region, it has been established that intensive cropping regimes lead to a reduction in the vegetation productivity and an impoverishment of the soil (acidification, reduction in structural stability, etc.)
[10]. This degradation is often related to a reduction in organic carbon stocks [11]. Many studies recently outlined information regarding changes in SOC distribution and the total mass stored for a dryland cropping system in semiarid regions [10-14]. The specific importance for the Sahel is the high vulnerability to climate change in already impoverished rural societies. A carbon market is now operational but without substantial inclusion or participation of Sahelian projects or countries. Many activities are now promoted to determine developing countries’s biophysical and socio-economic potential for carbon sequestration in soils with the ultimate objective to promote sustainable agriculture, restore degraded environments, and enhance food security and livelihoods among people [12,13]. SOC quality studies are very scarce [13,14]. Up-to-date regional, integrated soil and woodland inventories are required for planning improved land-use management and restoration [13]. This study aims to determine SOC stocks in the semiarid central Senegal and to link SOC quality variations to the effects of soil texture (sandy versus sandy clayey) and land use (savanna versus groundnut cultivation).
2. Materials and Methods
2.1. General Aspects of the Study Area in Kaffrine
The Division of Kaffrine (total area around 1.1 × 106 ha) is located in central Senegal (13˚49'N - 15˚19'W). The climate is dry tropical, and rainfall occurs between June and October, averaging 600 - 700 mm per annum (yr 1960-2000). The average annual temperature is 26˚C. The relief is flat and situated at 20 - 50 m above sea level. The geological substratum consists of sandy-clayey to clayey-sandy sandstone, with ferruginous cement, sometimes capped of duricrust [15]. The plateau, on which slash-and-burn agriculture is practised, is still covered with shrubby undergrowth, savanna and old fallows. This vegetation is semi-natural due to human exploitation for fire wood. Resprouting Combretaceae (mostly Combretum geitonophyllum Diels, C. glutinosum Perr. and Terminalia macroptera G. et Perr.) are the major component of the woody vegetation. Bush fields at the edge of the plateau are devoted to the cropping field. Groundnut (Arachis hypogaea L.) is the main crop. Tillage is advised for groundnut cultivation: soil quite ventilated to support the fixing of nitrogen, well-drained to avoid the rot of the seeds before the lifting and well structured to facilitate the development of the gynophores [16]. The farmers always practise at least one ploughing with draught animals to 10 - 15 cm depth. The tree cover on the groundnut fields of Kaffrine is less than 3%. After harvest, the entire groundnut plant, including roots, is removed from soil and stored for animals forage. Land use management consists of clearing savanna vegetation to establish groundnut fields, occasionally alternating with other crops (millet, rice, sorghum and cotton) and fallow. Most land is cropped, so the (semi-) natural savanna soils were very scarce.
The dominant soil type belongs to the order of ferric Luvisols with luvic Arenosols and gleyic Luvisols [17] and is sandy with a clay content of less than 20%. Iron accumulations are often observed in the form of well delimited spots or hardened concretions below the horizon of argillaceous accumulation, under 20 cm depth [18]. Chemical analysis of similar soils reveals clay mineral dominated by kaolinite [15], lesser CEC clay mineral.
2.2. Study Sites
Eleven villages of the division were selected based on soil characteristics (occurrence, area, type of soil) and distribution from previous studies within the division. In the selected villages, the soil surveyed was targeted at two types of land use: groundnut field and savanna. Study sites were selected according to age (years after land clearing) and soil texture. The soils were classified as ferric Luvisols [17]. The soil profile was homogeneous, with massive structure and without coarse elements. They are formed on sandy material of dunes or sandstone deposits. Two textural classes were selected, sandy soils (SAS) and sandy clayey soils (SCS). Furthermore, these two soil texture classes represented the range of soils in terms of land use and texture found within the study area (Table 1). 26 sites were surveyed, with 5 per cultivation duration and soil texture type and 6 sites in wooded savanna.
2.3. Leaves, Stems and Roots Biomass Estimation
The leaf biomass of ligneous plants was evaluated in each of the selected villages, in four circular plots along a 200 m transects, located far from human constructions. The size of each plot was selected depending on tree density and it varied between 1/16 and 1 ha. Ten small branches of each woody species among the most dominant in each plot were defoliated. The total fresh leaf biomass was weighted, and approximately 200 g of this leaf mass from each species were dried at 110˚C over 48 h. Stems samples were treated similarly. The analyses were carried out on 70 plant samples (Table 1).
The leaf biomass, in grams per individual tree, was estimated based on the formula:
(1)
where a and b are constants depending on species (proposed by [19] and used by [20]; a is between 0.34 and 14, and b is between 1.18 and 2.71; C = girth of the considered trunk in centimetres at 1.3 metres height.
Leaves and stems biomass were measured in groundnut fields after harvest and plant uprooting.
Below-ground biomass was not quantified, but it is assumed to be 70% of total biomass, representing the average value of those reported by [20] for a similar Senegalese area. On this basis, the proportion of root biomass and the equivalent mass of carbon were calculated for each soil and land use type.
The rate of annual C refund was estimated according to [21], 1% and 20% of the total biomass for cultivated dryland and savanna fields respectively.
2.4. Soil Sampling and Analysis
Twenty-six soil profiles were dug on groundnut fields and areas representing semi-natural savanna (20 and 6 respectively). Eleven profiles were on SAS soil and fifteen on SCS soil (Table 1), with at least two soil profiles per selected village. Soil samples were collected at 0 - 10, 10 - 20, 20 - 50, 50 - 80 and 80 - 100 cm soil depths. Samples were air-dried and sieved through a 2-mm stainless steel sieve before they were analysed for organic C and N content, pH, phosphorous as well as particle-size distribution.
Soil organic carbon was analysed by conductimetric measurement of CO2 after combustion (apparatus Casumat 8-Adge Wöstoff) and nitrogen according to the Kjeldahl method (Technicon apparatus). Soil pH was measured in a 1:2.5 soil/water solution. Phosphorous was analysed by Bay II method. Non-perturbed soil samples per 10 cm layer, down to 1 m depth, were taken using a cylindrical metal sampler (100 cm3 in volume). Six replicates were done. The bulk densities were determined based on the mass of dry soil/specific volume of samples. C concentrations were converted to total content for each of the 5 selected depths as the product of concentration, depth of sample and bulk density, and expressed in units of kg C∙m−2. Total SOC stock in the upper one metre was obtained as the sum of the SOC stock of the 5 depth intervals, as used by Chen et al. [4].
. (2)
Soil organic-C stocks in the top 20 cm were calculated by summing SOC stocks in the 0 - 10 and 10 - 20 and those in the 100 cm by summing SOC stocks in the top 20 cm, 20 - 50, 50 - 80, and 80 - 100 cm depth intervals.
Soil samples for isotopic analysis were priory freezedried, and were crushed (<200 µm). Organic carbon content was measured by autoanalysor Carlo Erba NA 1500. The isotopic ratios were measured on CO2 obtained by oxidation at 850˚C in a quartz tube sealed in the presence of CuO. CO2 was analyzed on a mass spectrometer VG SIRA by collecting triple and double introduction. The natural abundances are expressed in units of d13C.
(3)
The reference used is the International Pee Dee Belemnite (PDB) standard.
2.5. Statistics Analysis and Error Estimation
Statistical analyses were conducted using the SPSS software (version 10.0) for Windows. Non-parametric Pearson test and t-test at 5% level were performed to test differences between mean C content of each soil depth.
3. Results
3.1. Evaluation of the Leaves, Stems and Roots Biomass
The leaves, stems and roots biomass as well as the proportions of organic carbon per land use are summarized in Table 2. The total leaves, stems and roots biomass in Kaffrine is 4.7 Mg dry matter (dm) ha−1 for the seminatural savanna. Biomass was smaller under groundnut field (3.0 Mg dm∙ha−1).
The calculated rate of annual C refund was 0.42 Mg ha−1∙yr−1 in semi-natural savanna (Table 2) while the groundnut soil gained 0.01 Mg C∙ha−1∙yr−1.
3.2. Soil Characterization
The ferric Luvisols appear similar in terms of colour, pH and bulk density. The moist colour at near-surface layers varied from dull yellowish brown (10YR 4/3) to dark

Table 1. Different sites surveyed in Kaffrine and surface area.

Table 2. Tree-leaves, stems and roots biomass in Kaffrine (mean ± SD).
brown (7.5YR 3/4), yellowish brown (10YR 5/8) to bright brown (7.5YR 5/8) at greater depths. The pH (water) values were 4.3 to 5.8 and increased with depth. Values of pH (KCl) were 0.1 to 0.2 pH units lower than pH (water). Bulk density ranged from 1.40 to 1.53 g∙cm−3.
Table 3 shows the particle-size distribution of the soils studied. The sandy soils (SAS) were slightly structured; near-surface concentrations of nitrogen and phosphorus were between 0.113 - 0.325 g∙kg−1 and 0.029 - 0.087 g∙kg−1, respectively. Those of sandy clayey soils (SCS) were between 0.162 - 0.429 g∙kg−1 and between 0.044 - 0.072 g∙kg−1 near the surface, respectively. At 20 - 100 cm depth, SCS soils had higher clay contents (18% - 35%) than SAS soils (11% - 15%). Statistical analyses of these results show that only the content of silt and silt + clay distinctly defined the two soil texture classes (<12 wt%-silt and <19-wt%-silt + clay for SAS). For the same ferric Luvisols, the sand proportion in 0 - 20 cm depth was evidently higher in groundnut fields than in seminatural savanna.