Long-term assessment of soil and water conservation measures (Fanya-juu terraces) on soil organic matter in South Eastern Kenya

Geoderma 274 (2016) 1–9

Contents lists available at ScienceDirect

Geoderma journal homepage: www.elsevier.com/locate/geoderma

Long-term assessment of soil and water conservation measures (Fanya-juu terraces) on soil organic matter in South Eastern Kenya Gustavo Saiz a,⁎, Fredrick M. Wandera b, David E. Pelster b, Wilson Ngetich c, John R. Okalebo c, Mariana C. Rufino d, Klaus Butterbach-Bahl a,b a

Institute Meteorology and Climate Research, Karlsruhe Institute of Technology, Garmisch-Partenkirchen 82467, Germany Livestock Systems and Environment, International Livestock Research Institute, Nairobi 30709-00100, Kenya Department of Soil Science, University of Eldoret, P. O. Box, 1125, Eldoret, Kenya d Centre for International Forestry Research (CIFOR), PO Box 30677, Nairobi, Kenya b c

a r t i c l e

i n f o

Article history: Received 27 January 2016 Received in revised form 22 March 2016 Accepted 23 March 2016 Available online xxxx Keywords: Semi-arid ecosystems Soil water conservation Climate-smart agriculture Small household farms Soil organic carbon Stable isotopes

a b s t r a c t A comprehensive assessment of soil organic matter (SOM) dynamics in semi-arid agrosystems implementing soil and water conservation (SWC) measures is still lacking despite their extent, ecological and economic significance. Therefore, we assessed the long-term impact of a commonly used SWC technique (Fanya-juu terracing) on SOM-related properties in South Eastern Kenya. A soil sampling campaign was conducted in a replicated stratified random manner on three land uses that had been continuously managed for over 30 years. Samples were analyzed for organic carbon and nitrogen contents, δ13C, δ15N, pH and texture. Compared to sites implementing conventional agriculture, the establishment of SWC structures in this erosion-prone landscape resulted in the recovery of SOM levels comparable to those observed in neighboring semi-natural ecosystems. Sites under conventional agriculture practices contained 20 Mg C ha−1 (0.85 m), while sites with SWC measures and those hosting semi-natural vegetation stored above a third more. There were significant differences in soil C/N ratios as well as in δ13C and δ15N values between SWC cultivation practices classified according to the presence or absence of trees. The presence of woody vegetation in sites with SWC structures had a strong impact on the spatial variability of SOM-related properties. There was also a significant negative relationship between δ15N values and C/N ratios across the different land uses. Our findings indicate the existence of contrasting SOM dynamics caused by vegetation-related effects, and provide suggestions for enhancing SOM storage in agricultural sites implementing SWC measures. © 2016 Elsevier B.V. All rights reserved.

1. Introduction Land degradation is a serious cause for concern in sub-Saharan Africa where it affects more than two-thirds of its land (Penny, 2009). This is having detrimental effects on ecological functions and has already led to significant losses of agricultural productivity (Biancalani et al., 2011). Overgrazing, vegetation removal, poor agricultural management and overexploitation are among the main causes of soil degradation in subSaharan Africa (Hammond, 1992; Oldeman et al., 1991). These problems are being exacerbated by the ever-increasing pressure on soil resources from growing population and the heavy reliance on an agriculture that is highly vulnerable to environmental change (Liniger et al., 2011; Marks et al., 2009). The implementation of sustainable land management practices may help to increase agricultural productivity, improve

Abbreviations: (SWC), soil and water conservation; (SOM), soil organic matter; (SOC), soil organic carbon; (CSA), climate smart agriculture. ⁎ Corresponding author. E-mail address: [email protected] (G. Saiz).

http://dx.doi.org/10.1016/j.geoderma.2016.03.022 0016-7061/© 2016 Elsevier B.V. All rights reserved.

ecosystem functions and enhance resilience to adverse environmental impacts. Integrative approaches such as climate–smart agriculture (CSA) advocate for the implementation of agricultural practices and technologies aiming at increasing productivity in a sustainable manner (Lipper et al., 2014; Nyasimi et al., 2014). Indeed, small household farmers could effectively contribute to climate change mitigation through the adoption of agricultural practices that sequester carbon (C) and minimise emissions (Liniger et al., 2011). Soil organic matter (SOM) plays a crucial role on determining soil quality (Brady and Weil, 2007), and its enhancement may generate production, adaptation and mitigation benefits through the regulation of C, oxygen and plant nutrient cycling, thus promoting carbon sequestration and enhanced resilience to drought and flooding (Lipper et al., 2014). Soil degradation and/or high erosion rates have a detrimental effect on the numerous essential functions provided by SOM, posing major adverse economic and ecological consequences to livelihoods and the environment (Brady and Weil, 2007; Lal, 2004; Chappell et al., 2015). Given the significance of environmental degradation in sub-Saharan agricultural systems, efforts to revert the situation should focus on preserving two valuable and scarce natural assets, i.e. soil and water.

2

G. Saiz et al. / Geoderma 274 (2016) 1–9

The implementation of soil and water conservation (SWC) techniques comprising structural, vegetative and agronomic measures have the potential to reduce both runoff water and soil erosion, and improve infiltration and soil fertility (Liniger and Critchley, 2007). An example of such measure is the ‘Fanya-juu’ terracing system, which consists of producing embankments along a slope by digging out ditches following contour lines and depositing the soil uphill to form a mound (Fig. 1). The use of these SWC measures has increased crop yields by about 25% in East Africa (Ellis-Jones and Tengberg, 2000; Liniger et al., 2011). Moreover, SWC measures may hold great potential for increasing SOM levels since the regions where these are implemented are often heavily degraded (Liniger and Critchley, 2007). In fact, the semi-arid regions of Africa have been reported as having the largest potential for soil organic carbon (SOC) sequestration in the World (Batjes and Sombroek, 1997; Marks et al., 2009; Saiz et al., 2012). The increase of SOC stocks at a given site can be achieved either through the reduction of factors promoting SOM mineralization and lateral exports (e.g. erosion, percolation), and/or by increasing SOM inputs and enhancing stabilization mechanisms (e.g. physical protection of SOM through stable aggregates). Sites with SWC measures commonly host agroforestry systems that combine crops, grazing lands and trees. Should these systems be sensibly managed through sustainable agronomic practices (e.g. improved tillage practices, use of cover crops, optimum dosage and timely application of fertilizer, etc.) they will likely result in greater SOC stocks (Liniger et al., 2011). Moreover, this would also contribute to CSA aims, as soils rich in SOM require lower chemical inputs to sustain agricultural productivity and may enhance vital ecosystem functions, such as the hydrological and nutrient cycles (FAO, 2013). The detailed study of OM-related properties such as C/N ratio, δ13C, and δ15N in soil profiles may provide valuable information about the different processes affecting SOM pools on sites with contrasting land use. At the landscape scale, variations in land use, soil and vegetation type are the main factors affecting decomposition processes and the quantity and quality of SOM inputs (Feller and Beare, 1997; Silver et al., 2000; Chiti et al., 2014). While at the site scale, differences in local topography, vegetation cover and land management are among the main drivers

behind potentially contrasting SOM dynamics (Peukert et al., 2012; Saiz et al., 2006). Soil 13C and 15N are natural tracers of C and nitrogen (N) cycling (Bird et al., 2004; Nardoto et al., 2014; Wang et al., 2013) that can be combined with C and N elemental analyses to interpret SOM transformation processes (de Freitas et al., 2015; Stevenson et al., 2010). However, fractionation effects such as those associated with microbial reprocessing and the edaphic-dependent physicochemical protection of SOM makes the judgment of the variation in δ13C and δ15N values of SOM difficult (Silver et al., 2000; Blagodatskaya et al., 2011; Rumpel and Kögel-Knabner, 2011). A comprehensive assessment of SOM dynamics in sub-Saharan agricultural sites implementing combined SWC measures is still lacking despite their extent, ecological and economic significance. Specifically, the effect that long-term established SWC measures may have on SOMrelated properties is not yet known. Moreover, there is no information about the impact that contrasting vegetation (grass vs woody) and highly heterogeneous, purposely-made, ground strata may have on SOM dynamics at sites implementing SWC measures. Therefore, our study takes advantage of long-term established agricultural sites managed under real farming conditions, which follows on a similar strategy adopted in Liu et al. (2015). We hypothesized that, compared to sites under conventional agriculture, the implementation of SWC measures in erosion-prone semi-arid agricultural fields lead to an increase in SOM levels due to the greater topsoil preservation and enhanced organic matter inputs. We also hypothesized that in those sites having SWC measures, the existence of well-defined ground strata may promote significant variations in SOM dynamics between the different locations, largely because of their differential regimes in the transport, accumulation and removal of SOM. Our last hypothesis proposes that the presence of woody vegetation in agricultural fields may have a strong effect on both the elemental and stable isotopic composition of SOC and N because of the reported differential SOM dynamics observed in mixed C3/C4 semi-natural ecosystems (Saiz et al., 2015a). The objectives of this study were to: (i) assess the impact of SWC structures on the spatial variability of SOM-related properties; (ii) evaluate the influence of specific SWC cultivation practices (classified according to the presence

Fig. 1. Schematic representation of the SWC structures depicting the three contrasting ground features (trench, mound and undisturbed ground), and location maps for the Makueni/Wote CCAFS site in Kenya.

G. Saiz et al. / Geoderma 274 (2016) 1–9

or absence of trees in the dugout trenches) on SOM dynamics; and (iii) study the variation of these properties across different land uses in a semi-arid tropical landscape.

3

improves the structural stability of these SWC structures (Tenge and Hella, 2005).

2.2. Selection of sites 2. Materials and methods 2.1. Site characteristics This study was conducted at Makueni/Wote, which is a benchmark research site of the Climate Change, Agriculture and Food Security (CCAFS) program of the Consortium of International Agricultural Research Centers (CGIAR) in South Eastern Kenya (Fig. 1). CCAFS sites are established in tropical and subtropical regions that are already significantly affected by climate change and they and are intended to serve as focal locations to gather expertise that could be applied to other sensitive regions worldwide (Sijmons et al., 2013). The Makueni/Wote site is 10 x 10 km in size (1.900S, 37.630E southwest corner) and is located in a semi-arid region that has an irregular mean annual precipitation (Pa) of 600 mm with a marked bimodal annual distribution. The mean annual temperature is 22 °C. The elevation is 900– 1000 m above sea level, and the region is characterized by having an undulating topography with gentle slopes generally not exceeding 15%. Soils in the region are predominantly arenosols, while ferrasols are also present to a much lesser degree. Soil classification is according to the World Reference Base (IUSS, 2015). The local population consists mainly of small household farmers whose main constraint to crop and livestock production is water scarcity and soil erosion (Mwangangi et al., 2012; Kristjanson et al., 2014). Farms are privately owned and practice subsistence mixed farming (i.e. production of food crops and tree fruits, mainly mango and citrus, as well as keeping livestock). The main crops for consumption in the area are maize (Zea mays), cowpea (Vigna unguiculata) and pigeon pea (Cajanus cajan). Organic fertilizer is usually applied on a yearly basis but the use of mineral fertilizer is largely absent due to its high cost. Anecdotal evidence gathered from the farmers showed comparable practices both in the amount and type of organic fertilizer used for the different cropped systems or indeed between different farms (Mwangangi et al., 2012). Nowadays, the rapidly growing human population and degradation of existing croplands is causing a rapid expansion of agricultural croplands at the expense of remnant native vegetation (Mwangangi et al., 2012). The natural vegetation of the region is a scrub-type formation, composed of a varying mixture of grasses, shrubs and trees subject to varying degrees of grazing and browsing pressure. Large tracts of land were cleared in the 1950s for agricultural production. In view of the widespread degradation caused by agricultural activities SWC measures were introduced in the area in the late 1960s and were widely adopted by many local farmers in subsequent years (Liniger et al., 2011). The most commonly used SWC structures in the area are terraces that are constructed by digging ditches along the contour lines and forming parallel embankments (mounds), with the dugout soil either on the upslope side of each ditch ‘Fanya-juu’ or on its down-slope side ‘Fanyachini’. Terrace beds develop gradually behind soil mounds due to soil movement from the upper to the lower part of the terrace. Spacing between ditches depends on slope and soil depth, but generally consist of a trench about 1.0 m wide by 0.5 m deep, and a consolidated mound 0.5 m high by 1.5 m across at the base. The distance between mounds depends upon the slope and may be from 7 m apart on steep (N 30%) slopes to 40 m apart on more gentle (b 5%) slopes. These structures require regular maintenance to ensure their optimum performance, which are ideally conducted every 2–3 years depending on resources availability and actual maintenance needs. Activities include clearing debris from the trenches and mound stabilization through the plantation of fodder grasses that are regularly trimmed to feed the livestock. The establishment and regular trimming of fodder grass on mounds

A household survey was conducted on 132 local farmers in February 2014 to obtain baseline information about agricultural practices and identify potential sampling sites. This work was conducted using as a reference the baseline household survey carried out at the Makueni/ Wote site in 2012 (Mwangangi et al., 2012). The aim of that work was to gather information on household composition, farming systems, diversity in farming activities, and characterize agricultural production systems and land uses. Moreover, group discussions were also held with extension and livestock officers from the Ministry of Agriculture, local administration and farmer self-help groups. Informed verbal consent was obtained from every household prior to the survey. In the end, three main land uses were selected with sites being classified as: (i) croplands having soil water conservation (SWC) structures, (ii) croplands not having SWC structures (conventional agriculture), and (iii) semi-natural vegetation. All selected sites occurred on arenosols, including those hosting semi-natural vegetation. Specific agricultural sites were selected because they were cultivated for over 40 years. Moreover, those having SWC structures had them established for at least 30 years (Table A1). The survey also confirmed that the use of mineral fertilizer across different farms was nonexistent, which minimized the potential confounding effects posed by its differential use between the different farms. Ultimately, 14 sites implementing soil and water conservation measures were selected and classified according to specific cultivation practices (i.e. woody species planted in trenches ‘Fanya-Tree’ (n = 9) or herbaceous crops planted in trenches ‘FanyaGrass’ (n = 5)). Additionally, 10 sites implementing conventional agriculture, and 10 sites hosting native vegetation, being in direct vicinity of the sites with SWC measures, were also selected to serve as reference land uses.

2.3. Sampling Strategy Soil sampling took place from April to June 2014. Sampling locations were randomly selected in sites implementing conventional cultivation, while sites having SWC structures were sampled in a stratified random manner that covered three contrasting ground features, namely trench, mound and undisturbed ground (Fig. 1; Saiz and Albrecht, 2015). Undisturbed ground is defined as that not directly affected by either the addition or removal of soil resultant from the building and maintenance of trenches and mounds. The area occupied by each stratum was measured at each site. Soil sampling was carried out across four replicate sampling points at each of the strata described above, which resulted in a total of 56 locations being sampled per stratum (4 replicates x 14 sites having SWC structures). In the case of stands hosting seminatural vegetation sampling locations were chosen following a stratified sampling strategy that has proved well suited to best account for the inherent heterogeneity of SOC that is characteristic of these ecosystems (Bird et al., 2004; Wynn et al., 2006; Saiz et al., 2012). This approach consists of taking samples near trees (‘Tree samples’ at half canopy radius from trunks) and away from trees (‘Grass samples’ at half the maximum distance between trees). At each sampling location surface litter was removed when present. Two samples were taken at 0–0.05 m with the aid of a stainless steel corer (40 mm inner diameter), before being bulked together to smooth out the typically large local heterogeneity existing at very shallow depths. Subsequently, deep soil sampling was carried out using the same corer at 0.05–0.30 m, 0.30–0.50 m and up to 0.85 m depth (impenetrable layers permitting), with each sample being individually collected. The total number of soil samples taken was 989.

4

G. Saiz et al. / Geoderma 274 (2016) 1–9

2.4. Analytical methods and calculations Samples were weighed in sealed bags, clumps broken by hand and then oven dried at 40 °C to constant weight. An aliquot of these samples was then oven dried at 105°C for four hours to quantify water content in each sample, which allowed for the calculation of soil bulk density (SBD) (Saiz and Albrecht, 2015). Calculation of SBD included fractions N2 mm. Samples were then dry sieved to 2 mm, and gravel content was quantified. Stable isotope composition and elemental abundances of C and N were determined in duplicate for the 989 powdered samples using a Costech Elemental Analyser fitted with a zero-blank auto-sampler coupled via a ConFloIII to a ThermoFinnigan DeltaPlus-XL using Continuous-Flow Isotope Ratio Mass Spectrometry (CF-IRMS) at the Centre for Stable Isotopes at the Institute of Meteorology and Climate Research IMK-IFU/KIT Garmisch-Partenkirchen (Germany). Precisions (S.D.) on internal standards for elemental C and N abundances and their stable isotopic compositions were better than 0.06% and 0.2‰ respectively. Determinations of pH values were obtained using a digital pH meter in a water solution with a water to soil ratio of 2:1. Aliquots of samples showing pH N 6.5 were pre-treated with 1M HCl to ensure the absence of carbonates and were again analysed in the CF-IRMS. Particle size distributions were determined by the hydrometer method as described by Hinga et al. (1980) in aggregated samples (i.e. pooling proportionally the 0–0.05 and 0.05–0.30 depth intervals, as described in Saiz and Albrecht (2015)). The mean site value of any soil property was calculated in relation to the area covered by each stratum. The average SOC stock for a given depth interval (d) was calculated according to the following formula: μ d ¼ BDd  OC d  D  ð1  grÞ=10

ð1Þ

where μ d is SOC stock (Mg OC ha−1); BDd is soil bulk density (g cm−3); OCd is the concentration of OC in soil (b 2 mm; mg OC g−1 soil); D is soil depth interval (cm); gr is fractional gravel content, the soil fraction N2mm. Table A2 shows the cumulative mass coordinate approach used to calculate SOC stocks across the different land uses. The cumulative mass coordinate approach consists of collection and quantification of all the soil mass in a given depth interval. Sampling by mass instead of volume minimizes potential biases derived from varying bulk density caused by land use change or agricultural practices (Gifford and Roderick, 2003; McBratney and Minasny, 2010; Rovira et al., 2015). We conducted sampling at fixed depth intervals in order to compute for soil bulk density and be able to inter-convert between the spatial coordinate and the cumulative mass coordinate approach (Saiz and Albrecht, 2015). The different datasets were tested for normal distribution by the Kolmogorov-Smirnov test, and where necessary data were logtransformed to conduct parametric statistical tests. Within each site and depth, one-way ANOVA was performed to compare soil variables between different sampling locations. Post hoc comparisons using Tukey HSD test were conducted to find out which of the locations differed. The same statistical procedure was used to compare soil properties between different land uses. Analyses of covariance (ANCOVA) were performed to test for significant different differences between regressions. SPSS 17.0 (SPSS Inc. Chicago, IL, USA) was used for all statistical analyses. 3. Results 3.1. Soil properties at sites implementing SWC measures The abundance of both SOC and TN was relatively small, with values consistently lower than 10 mg C g−1 and 1.0 mg N g−1 (Table 1). Significantly greater values of SOC and TN were observed in ‘trench’ compared to ‘undisturbed’ and ‘mound’ sampling locations. However,

Table 1 Abundance and stable isotope composition of soil OC and total soil nitrogen (TN) at sites implementing soil and water conservation (SWC) agricultural practices. Results are shown for each depth interval at different sampling locations. Variable

Soil depth (m)

Location Undisturbed

Mound

Trench

δ13C (‰)

0.0–0.05 0.05–0.30 0.30–0.50 0.50–0.85 0.0–0.05 0.05–0.30 0.30–0.50 0.50–0.85 0.0–0.05 0.05–0.30 0.30–0.50 0.50–0.85 0.0–0.05 0.05–0.30 0.30–0.50 0.50–0.85 0.0–0.05 0.05–0.30 0.30–0.50 0.50–0.85

−20.5 (1.7) ab −19.1 (2.0) a −18.5 (2.4) −18.4 (2.4) 6.0 (1.8) a 5.8 (1.9) a 7.2 (1.1) a 7.2 (1.4) ab 3.8 (1.6) a 3.3 (1.8) a 3.2 (1.8) 2.7 (1.4) 0.3 (0.1) a 0.3 (0.1) a 0.2 (0.1) a 0.2 (0.1) 11.6 (1.8) 11.5 (2.4) a 11.5 (3.8) ab 11.1 (3.4)

−19.8 (1.7) a −19.2 (1.6) a −18.6 (1.8) −18.2 (2.0) 6.5 (1.6) ab 7.2 (1.8) b 7.5 (1.1) a 6.7 (2.1) a 4.8 (2.7) a 3.7 (2.5) a 3.4 (2.3) 3.0 (1.6) 0.4 (0.2) a 0.3 (0.2) ab 0.3 (0.2) ab 0.3 (0.2) 11.5 (2.5) 10.5 (1.6) b 12.1 (2.0) a 12.1 (2.9)

−21.1 (2.1) b −20.1 (2.2) b −19.1 (2.3) −18.8 (1.8) 7.0 (1.4) b 7.2 (1.7) b 8.8 (0.6) b 7.6 (1.0) b 7.2 (5.3) b 5.2 (3.7) b 3.7 (2.0) 3.0 (2.0) 0.6 (0.4) b 0.4 (0.2) b 0.3 (0.2) b 0.3 (0.1) 11.9 (1.8) 12.7 (3.1) c 10.8 (2.5) b 11.3 (2.5)

δ15N (‰)

OC (mg g−1)

TN (mg g−1)

C/N

Numbers in brackets denote standard deviation from the means (n = 56). Different letters within the same depth interval denote significantly different values (P b 0.05); ANOVA, Tukey post hoc test.

these differences between sampling locations got smaller and nonsignificant at the deepest interval. A similar trend to that of SOC and TN was observed for δ15N values, while the opposite was true for δ13C, which had the lowest values in ‘trench’ locations. Table 2 shows soil properties determined for the 0.0–0.30 m depth interval for the different sampling locations. While no significant differences were observed between sampling locations for any of the studied variables in ‘Fanya-Grass’ sites, the ‘trench’ sampling location generally showed significant differences in ‘Fanya-Tree’ sites. The overall differences between both agricultural practices were significant for both δ13C and δ15N values, as well as for C/N ratios, SBD and pH values (Table 2). 3.2. Soil carbon stocks across different land uses There were large differences in SOC stocks between sites implementing SWC and conventional agricultural practices. These were significant at all sampling depths with the exception of the topmost interval (0 to 0.05 m depth; Fig. 2). On the other hand, semi-natural vegetation sites consistently showed greater amounts of OC in the first 0.05 m of the soil, but these variations were not significant at deeper locations. Conventional agricultural sites contained about 20 Mg C ha−1 in the soil profile under consideration (0.85 m), while sites with SWC measures and semi-natural vegetation stored above a third more. Overall, the average SOC contained in the first 0.3 m of the soil accounted for roughly half of that stored in the studied soil profile (Fig. 2; Table A2). 3.3. Soil properties across different land uses Soil textural analyses showed no significant differences across the different land uses (Table 3). However, SOC contents and C/N ratios under conventional agriculture were lower than those of sites having SWC structures and those sustaining semi-natural vegetation. The same observed differential pattern also applied for SOC stocks, while soil nitrogen stocks were higher in semi-natural ecosystems than in cropped sites (Table 3). There was a similar variation in δ13C values along the soil profile for the three land uses (Fig. A1). These exhibited a noticeable increase in δ13C values within the first 0.3 m depth below which they remained relatively constant. From that depth down, the

G. Saiz et al. / Geoderma 274 (2016) 1–9

5

Table 2 Soil properties determined for the 0.0−0.30 m depth interval at sites implementing soil and water conservation measures. Sites are further classified according to specific cultivation practices (i.e. woody species planted in trenches ‘Fanya-Tree’ (n = 9) and herbaceous crops planted in trenches ‘Fanya-Grass’ (n = 5)). Results are shown for the different sampling locations. Totals for each cultivation strategy correspond to the calculated average for each site, making use of the proportion of area covered for each sampling location. Variable

SWC agricultural practice Fanya-Tree

Fanya-Grass

Location

δ13C (‰) δ15N (‰) OC (mg g−1) TN (mg g−1) C/N BD (g cm−3) Gravel (g g−1) OC Stock (Mg C ha−1) TN Stock (Mg N ha−1) Sand (g g−1) Clay (g g−1) pH

undisturbed

mound

trench

−19.9 (0.8) a 6.5 (1.2) a 3.1 (0.7) a 0.3 (0.1) a 10.7 (1.3) ab 1.2 (0.1) a 0.10 (0.03) 10.1 (2.6) a 0.9 (0.4) a 0.76 (0.05) 0.13 (0.04) 6.8 (0.2)

−19.9 (0.8) ab 7.6 (1.5) b 3.5 (1.1) a 0.3 (0.1) a 10.4 (1.0) a 1.1 (0.1) b 0.10 (0.06) 10.6 (3.6) a 1.1 (0.5) ab 0.77 (0.06) 0.14 (0.05) 6.7 (0.4)

−21.3 (1.4) b 7.8 (1.2) b 4.9 (1.1) b 0.4 (0.1) b 11.6 (1.0) b 1.1 (0.1) b 0.09 (0.03) 15.1 (2.5) b 1.3 (0.5) b 0.77 (0.06) 0.13 (0.04) 6.8 (0.4)

Total

Total

Location undisturbed

mound

trench

−20.1 (1.0) A 6.7 (1.3) A 3.3 (1.0) 0.3 (0.1) 10.8 (1.2) A 1.2 (0.1) A 0.10 (0.04) 10.7 (3.3) 1.0 (0.3) 0.76 (0.04) 0.13 (0.04) 6.7 (0.5) A

−18.3 (2.3) B 5.0 (1.5) B 4.2 (2.8) 0.3 (0.2) 13.0 (2.4) B 1.1 (0.1) B 0.12 (0.05) 12.1 (7.8) 0.9 (0.4) 0.74 (0.06) 0.18 (0.07) 6.1 (0.5) B

−18.3 (2.4) 4.6 (1.3) 3.8 (2.6) 0.3 (0.1) 13.1 (2.3) 1.1 (0.1) 0.12 (0.04) 10.9 (7.1) 0.8 (0.4) a 0.75 (0.06) 0.16 (0.05) 6.1 (0.7)

−18.1 (1.5) 6.2 (1.3) 4.7 (3.4) 0.4 (0.2) 11.1 (1.8) 1.1 (0.1) 0.10 (0.02) 14.0 (9.8) 1.2 (0.7) ab 0.72 (0.07) 0.19 (0.06) 6.2 (0.5)

−18.6 (2.0) 6.0 (0.8) 6.9 (6.0) 0.4 (0.3) 14.4 (3.7) 1.1 (0.1) 0.12 (0.04) 20.0 (16.4) 1.3 (0.8) b 0.74 (0.07) 0.19 (0.11) 6.2 (0.5)

Numbers in brackets denote standard deviation from the means (n = 36 and n = 20 for the ‘Fanya-Tree’ and ‘Fanya-Grass’ cultivation practice respectively). Different lower-case letters denote significantly different values between sampling locations within the same cultivation practice. Different upper-case letters denote significantly different total values between the two cultivation practices (P b 0.05); ANOVA, Tukey post hoc test.

‘Fanya-Grass’ sites exhibited higher average δ13C values than any other land use. The variations in δ15N values along the soil profile were similar to those exhibited by δ13C, with the highest values observed in the 0.3– 0.5 m interval for all land uses except for ‘Fanya-Grass’ sites (Fig. A2). The relationship between clay content and δ15N values for the different land uses is shown in Fig. 3. Sites with higher clay contents also showed high δ15N values. Regressions were calculated for each land use with best-fits being provided by power functions. These regressions exhibited the same increasing pattern and were all significant (P b 0.05). Regression coefficients ‘r2’, and parameters ‘a ± SE’ and ‘b ± SE’ are (0.48, 22.0 ± 11.0, 0.65 ± 0.24), (0.43, 16.3 ± 6.7, 0.43 ± 0.20), and (0.69, 19.0 ± 5.3, 0.59 ± 0.14) for the CON Ag, SWC FanyaTree, and NAT Veg sites respectively. Analyses of covariance (ANCOVA) showed no significant differences between regressions (P N 0.05). The sites implementing SWC measures had slightly higher δ15N values than the other two land uses, but no significant differences were detected between the regressions as demonstrated by ANCOVA analyses (P N 0.05). Nonetheless, the inclusion of ‘Fanya-Grass’ sites in the SWC regression would have altered the observed pattern (see

discussion). There was a negative relationship between δ15N values and C/N ratios across the different land uses, with sites implementing ‘Fanya-Tree’ and ‘Fanya-Grass’ SWC cultivation practices showing the broadest variation (Fig. 4). 4. Discussion 4.1. Influence of SWC structures on the spatial variability of SOM-related properties The implementation of SWC structures in semi-arid agricultural regions has been reported to have beneficial effects not only in the preservation of soil but also in the agronomic performance of the crops (Ellis-Jones and Tengberg, 2000; Liniger et al., 2011). Fanya-juu terraces harvest water and prevent soil from being eroded from the slopes on which they have been established, but in doing so they also promote siltation and the accumulation of plant debris at lower locations, particularly in the dugout trenches. The lack of frequent maintenance of these constructions invariably leads to preferential accumulation of

Fig. 2. Soil carbon stocks (Mg C ha−1) for each land use at different depth intervals. Values correspond to the calculated average for each land use. Abbreviations for each land use are as follows: Conventional agriculture (CON Ag), Soil and Water conservation agriculture (SWC Ag), and semi-natural vegetation sites (NAT Veg). Bars represent standard deviations from the means. Different lower-case letters within a depth interval denote significant differences between the various land uses. Different upper-case letters (right panel) denote significant differences in total carbon stocks between the various land uses (P b 0.05); ANOVA, Tukey post hoc test.

6

G. Saiz et al. / Geoderma 274 (2016) 1–9

Table 3 Soil properties determined for the 0.0–0.30 m depth interval for three different land uses. Cropped agricultural sites are further subdivided between those implementing soil and water conservation (SWC) measures and those not using them (conventional). Non-cropped sites hosting semi-natural vegetation are also shown for reference. Variable

Land Use Cropped

δ13C (‰) δ15N (‰) OC (mg g−1) TN (mg g−1) C/N BD (g cm−3) Gravel (g g−1) OC Stock (Mg C ha−1) TN Stock (Mg N ha−1) Sand (g g−1) Clay (g g−1) pH

Non-cropped

n

Conventional

n

SWC

n

Semi-natural vegetation

40 40 40 40 40 40 40 40 40 10 10 10

−20.2 (1.5) 5.6 (1.6) 2.7 (1.1) a 0.3 (0.1) a 10.6 (1.7) a 1.2 (0.0) a 0.10 (0.04) 9.0 (3.1) a 0.8 (0.3) a 0.77 (0.05) 0.12 (0.03) 6.4 (0.4)

56 56 56 56 56 56 56 56 56 14 14 14

−19.4 (1.8) 6.1 (1.6) 3.6 (1.9) b 0.3 (0.1) a 11.6 (2.0) b 1.2 (0.1) ab 0.11 (0.04) 11.3 (5.3) b 1.0 (0.4) a 0.76 (0.05) 0.15 (0.04) 6.5 (0.5)

40 40 40 40 40 40 40 40 40 10 10 10

−19.4 (2.5) 5.8 (1.7) 4.5 (2.1) b 0.4 (0.2) b 11.8 (0.9) b 1.1 (0.1) b 0.10 (0.04) 13.8 (6.3) b 1.2 (0.5) b 0.77 (0.06) 0.14 (0.05) 6.3 (0.5)

Values correspond to the calculated average for each land use. Numbers in brackets denote standard deviation from the means. Different letters for a given variable denote significant differences between the various land uses (P b 0.05); ANOVA, Tukey post hoc test.

organic-rich material, which may explain the significantly greater values of SOC and TN observed in ‘trenches’ compared to ‘undisturbed’ and ‘mound’ sampling locations (De Blécourt et al., 2014; Table 1). Moreover, due to their lower topographical position, dugout trenches tend to also accumulate runoff water, which leads to comparatively higher soil moisture conditions (Tian et al., 2003; Mati, 2006). These factors may promote a favorable environment to boost SOM microbial decomposition and enhance N-transformation processes (Booth et al., 2005; Bai et al., 2009; Peukert et al., 2012). Indeed, trench or furrow locations built along the contour line of a slope have been reported to have significantly higher soil CO2 emissions (Saiz et al., 2006). Similarly, higher N2O emissions have also been recorded at low locations along a topographic gradient (Fang et al., 2009; Vilain et al., 2010). These dynamics result in preferential losses of 14N via denitrification, and may explain the higher soil δ15N values observed at these depositional localities (Table 1; Garten and Miegroet, 1994; Bedard-Haughn et al., 2003; Liu et al., 2007; Biswas et al., 2008; Peukert et al., 2012). Mound locations show similar δ15N values as those observed in trenches simply because mounds are regularly topped-up with soil previously allocated in trenches. With the notable exception of δ15N values, most of the differences in soil properties between sampling locations were not significant

Fig. 3. Relationship between Clay content (Cc) and δ15N values for the different land uses. Each value represents the calculated mean for each site at the 0.0–0.30 m depth interval. Closed and open red circles correspond to SWC agricultural sites implementing the ‘Fanya-Tree’ and ‘Fanya-Grass’ cultivation practice respectively. Regressions are significant at P b 0.05 level, and for the case of SWC sites do not include the ‘FanyaGrass’ sites. Regressions take the form: δ15N = a*(Cc)b.

below the 0.30 m interval, which shows that within-site topographical effects on SOM dynamics were limited at depth (Table 1). 4.2. Influence of specific SWC cultivation practices on SOM dynamics While there is wide consensus about the multiple benefits provided by trees on a wide range of ecosystem functions in natural ecosystems (Gamfeldt et al., 2013; Brandon, 2014), long-term agroforestry research in semi-arid regions have shown complex interactions on resource sharing between trees and crops (Rao et al., 1998; Ong and Leakey, 1999; Atangana et al., 2014). The establishment of fruit trees along trenches (Fanya-Tree) is customary practice for many farms implementing SWC agriculture in this region (Mwangangi et al., 2012). Trees are typically planted less than 10 m from each other, and their presence may exert significant impacts on both the quantity and quality of organic inputs returning to the soil (Post et al., 1982; Saiz et al., 2012). We purposely chose additional SWC agricultural sites with no woody vegetation on them (Fanya-Grass) in order to assess the role of trees on SOM dynamics. Woody vegetation uses the C3 photosynthetic pathway (δ13C b −24‰), while some of the main crops used in tropical agrosystems (e.g. maize,

Fig. 4. Relationship between δ15N values and C/N ratios for all land uses at the 0.0–0.30 m depth interval. Results are shown for the different SWC sampling locations (undisturbed, mound and trench) with closed and open symbols corresponding to SWC sites implementing ‘Fanya-Tree’ and ‘Fanya-Grass’ cultivation practices respectively. Error bars represent standard errors from the means. Crosses correspond to the calculated average for each land use and are shown here for reference. Semi-natural vegetation is further subdivided between ‘Tree’ and ‘Grass’ sampling locations.

G. Saiz et al. / Geoderma 274 (2016) 1–9

sorghum) as well as many grass species primarily use the C4 photosynthetic pathway (δ13C values N −15‰) (Lloyd et al., 2008; Saiz et al., 2015b). Consequently, the abundance and δ13C values of litter and SOM are quite heterogeneous in mixed C3/C4 systems. This heterogeneity is largely controlled by the distribution of trees, with lower soil δ13C values near the stems compared to areas away from their influence (Bird et al., 2004; Wynn and Bird, 2007). The strong influence exerted by trees on soil properties was noticeable at the local scale, with differences between ‘trenches’ and the rest of sampling locations being generally significant in ‘Fanya-Tree’ sites but not significant in treeless sites (Fanya-Grass) (Table 2). The higher SOC and TN contents observed at trench locations in ‘Fanya–Tree’ sites could be the combination of several factors. These include the preferential accumulation of organic-rich material from higher locations, the potentially greater OM inputs occurring in the vicinity of the trees (Mordelet et al., 1993; Scholes and Archer, 1997), and differential decomposition rates of C3- and C4-derived material (Wynn and Bird, 2007; Saiz et al., 2015a). Furthermore, the influence of woody vegetation on SOM was also noticeable at the site scale, as revealed by the significantly lower soil δ13C values observed in ‘Fanya-Tree’ sites compared with the ‘Fanya-Grass’ sites (Table 2). A potential reason for the low δ15N values observed in ‘Fanya–Grass’ sites may be the generally low foliar δ15N of grasses compared to woody species in semi-arid ecosystems (Wang et al., 2013). Furthermore, the presence of nitrogen-fixing plant species (e.g. Acacia sp, Vigna unguiculata, Cajanus cajan, etc.) may also contribute to distinct soil δ15N values (Aranibar et al., 2004; Nardoto et al., 2014; de Freitas et al., 2015). However, farmers indicate comparable abundances of nitrogen-fixing species in both agricultural practices. Additionally, there have also been reports of vertical uplift of deep N by tree roots as a likely cause of 15N enrichment of surface soil (Bai et al., 2013). Fig. A2 shows the vertical profile of δ15N having similar variations in both SWC cultivation types, with the highest values being observed at deep locations. Soil δ15N values tend to increase with depth as increasingly decomposed material moves down the soil profile and as 14N gets lost to N2, or N trace gases (Silver et al., 2000; Hobbie and Ouimette, 2009). Therefore, we cannot discard this mechanism as a potential contributor to the high δ15N values observed in topsoil layers of ‘Fanya-Tree’ cultivation practices. In addition to the quantity and quality of the precursor biomass, SOM dynamics at a given site are also known to be affected by local environmental conditions (Ong and Leakey, 1999; Van den Heuvel et al., 2009). In this regard, trees growing on coarse-textured soils in semiarid regions may have a strong effect on maintaining sustained soil water conditions suitable for the activities of SOM decomposers. This is achieved through canopy interception with subsequent funnelling of precipitation, and reduction of soil water evaporation by shading (Mordelet et al., 1993; Ong and Leakey, 1999; Abbadie et al., 2006). Therefore, SOM decomposition processes may be comparatively more dynamic in semi-arid sites having a direct influence from trees. This is in agreement with observations by Saiz et al. (2015a) who reported a similar pattern in SOM dynamics in mixed C3/C4 semi-arid ecosystems in West Africa. These authors provide evidence about potentially faster decomposition rates at locations dominated by woody vegetation compared to those dominated by grass species in semi-natural ecosystems occurring on coarse-textured soils. Work by de Graaff et al. (2008) shows that a crucial aspect for a continued adoption of SWC technologies is their perceived profitability. The establishment of SWC structures may be initially perceived as a loss of cultivable area, which can be somewhat compensated with the plantation of fruit trees and/or fodder grasses on the purposely-modified ground (trenches and mounds). However, our study shows that local environmental conditions caused by trees planted on trenches may promote the development of denitrification hotspots, and lead to lower SOC levels as compared to ‘Fanya-Grass’ sites. However, such findings can only be ascribed to the experimental conditions under which this study was conducted and only relate to SOM-related properties.

7

Hence, the ecological or economic benefits derived from trees are not considered here. 4.3. Variation in SOM dynamics across different land uses There were significant differences in SOM-related properties between sites implementing conventional agriculture and those having SWC structures (Table 3). The studied sites only have a moderate potential to store belowground organic C as a result of the sandy nature of the soils (Wynn et al., 2006; Saiz et al., 2012). However, the establishment of SWC structures had a positive effect on SOC stocks, to the extent that these sites stored above a third more SOC than conventional cropped sites (Fig. 2), reaching levels comparable to those observed in semi-natural vegetation stands. The SOC stocks observed in noncropped ecosystems (Table A2) agree fairly well with those obtained using a function driven by climatic and edaphic variables proposed by Saiz et al. (2012) for West Africa (12 Mg C ha− 1 for the 0.0–0.3 m depth interval). Moreover, relatively low soil nitrogen stocks were observed in cropped sites, which agrees with the absence of N-based fertilisers in these farms (Table 3). Conventional agricultural sites exhibited lower SOC contents and C/N ratios than farms under SWC measures likely because of the visibly higher rates of topsoil erosion of the former and the lower SOM inputs resultant from their reportedly lower plant productivity (Tenge and Hella, 2005; Liniger and Critchley, 2007). Similarly, higher SOC and TN contents were observed within the first 0.2 m in agricultural sites with SWC structures in central Ethiopia compared to sites implementing conventional agriculture (Hailu et al., 2012). However, both the lack of sampling stratification, which could account for the inherent spatial variability of those sites, and the relative short time since their establishment (5 and 10 years) makes a direct comparison with our results difficult. The characterisation of SOM dynamics under field conditions is an extremely challenging task because of both the potentially large time lag between production and decomposition processes, and the high spatiotemporal variability in OM inputs and turnover rates (Gignoux et al., 2006; Saiz et al., 2015a). Considering that the studied SWC measures had been established for over 30 years, then recent short-term measurements of soil moisture, temperature or soil respiration would be of very limited use to fulfill the objectives of this work (Liu et al., 2015). Indeed, in this study, a sound assessment of changes in SOM dynamics using some of the above environmental parameters necessarily implies conducting long-term monitoring campaigns at high spatiotemporal resolutions, and not just over a few recent years, but on a quasicontinuous basis since their establishment as it is highly likely that a significant proportion of` SOM transformations impacted by the SWC measures would take place during early stages following their construction. Therefore, the implementation of an isotopic approach in this research is better suited to study the long-term variation in SOM dynamics across the different land uses, which is further justified by the large number of replicate sites used (Liu et al., 2015). Specifically, the significant differences observed in C/N ratios as well as in δ13C and δ15N values offer useful information about the potentially contrasting SOM dynamics existing between the different land uses. Soils with comparatively high clay contents generally have higher δ15N values than coarser soils (Silver et al., 2000; Peukert et al., 2012; Bai et al., 2013). This is partly because the comminuted OM bound to the fine fractions has been exposed longer to microbial processing than the OM associated to coarser particles, which results in preferential 15 N enrichment of the residual substrate (Hobbie and Ouimette, 2009; Bai et al., 2013). It is worth noting that soil textural analyses showed no significant differences between the different land uses (Tables 2 and 3). However, SWC sites with ‘Fanya-Grass’ cultivation practice showed lower δ15N values than their SWC counterparts at comparable clay contents (Fig. 3). Therefore, notwithstanding the influence in soil δ15N values exerted by the precursor biomass, the above fact may suggest that the different δ15N values between both SWC cultivation

8

G. Saiz et al. / Geoderma 274 (2016) 1–9

practices may not be due to soil texture and that other factors could be responsible for their potentially contrasting SOM dynamics. Sites with lower C/N ratios tend to have higher δ15N values (Fig. 4), which suggests a more open N cycle, and thus, potentially greater N losses (Amundson et al., 2003; Stevenson et al., 2010; de Freitas et al., 2015). Indeed, low soil C/N ratios in arid regions have been reported to enhance gaseous losses from the system, which can lead to increased soil δ15N values (Aranibar et al., 2004; Brady and Weil, 2007). However, the significantly lower SOC contents and C/N ratios of conventional cropped sites (Table 3) indicates that C losses could be even proportionally greater than those of N. This compares well with work by Stevenson et al. (2010) who studied the variation of soil δ15N across different land uses. They reported that the decrease in SOC contents in cropping systems suggested that SOM and associated N losses were as responsible for the increase in δ15N values as the N availability itself. The strong negative relationship between δ15N values and C/N ratios for equivalent sampling locations at SWC sites implementing ‘Fanya-Tree’ and ‘Fanya-Grass’ cultivation practices (Fig. 4) is in agreement with the previously discussed differential SOM dynamics observed between both systems. On the other hand, Fig. 4 shows that sites hosting semi-natural vegetation line up between the SWC sites, and it is interesting to note that values from ‘Tree’ sampling locations were closer (more comparable) to ‘Fanya-Tree’ sites, while the reciprocal was true for ‘Grass’ sampling locations. This indicates a similar influence of vegetation type-mediated effects on SOM dynamics at both ecosystems. Besides SOM mineralization and lateral OM exports from the system (e.g. crop harvest and soil erosion), the relative importance of other processes affecting soil δ15N (e.g. denitrification, nitrate leaching, ammonia volatilization) need to be purposely quantified in the different systems and seen in light of the strong influence exerted by SWC structures and vegetation type on SOM dynamics. 5. Conclusions The establishment of SWC structures in this erosion-prone landscape resulted in the recovery of SOM levels comparable to those observed in neighboring semi-natural ecosystems. This adds to the wide array of reported benefits derived from the use of SWC measures, which range from ecological (e.g. reduced soil loss), economic (e.g. increased crop yields), socio-cultural (e.g. community strengthening) to off-site gains (e.g. reduced downstream siltation) (Liniger et al., 2011). Moreover, fewer chemical inputs may be required to sustain yields as these measures help preserving valuable topsoil rich in SOM, which have evident positive environmental and economic effects. In this study, the presence of woody vegetation at sites with SWC structures had a strong impact on the spatial variability of SOM-related properties. This effect was particularly evident at depositional locations (trenches), which showed the highest SOC and N contents. Furthermore, the significant differences observed in C/N ratios as well as in δ13C and δ15N values between SWC agricultural practices (classified according to the presence or absence of trees in the dugout trenches) strongly suggest the existence of contrasting SOM dynamics caused by vegetationrelated effects. Within this context, the local environmental conditions caused by trees planted on trenches may promote the development of denitrification hotspots. This latter fact combined with the potentially lower soil GHG emissions and comparatively high SOC storage of ‘Fanya-Grass’ sites, may lead to consider the convenience of using fodder grasses in trenches instead of trees. However, these findings should be seen in light of the experimental conditions under which this study was conducted (e.g. arenosols on gentle slopes). But most crucially, such conclusion only relates to SOM properties and does not consider the ecological or economic benefits derived from trees. Therefore, it would be desirable to integrate trees into this landscape as recommended by Ong and Leakey (1999), which includes exploitation of underutilized niches (e.g. footpaths and home-compounds), and promotion of N2-fixing and termite-resistant trees on the household boundaries. Ultimately, for a given management to be adopted farmers need to

realize that mitigation benefits and food security interact in a costeffective way (Mann et al., 2009; Liniger et al., 2011), for which more integrated research on the topic is clearly needed. Acknowledgements We are truly grateful to the numerous farmers who generously allowed us to work in their fields and patiently answered all our questions. We thank Mr Victor Braojos for his assistance in the analytical work. The Climate Food and Farming Research Network (CLIFF) provided financial support to Fredrick Wandera to conduct this research. This work was undertaken as part of the CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS), which is a strategic partnership of CGIAR and Future Earth. This research was partly carried out with funding by the European Union (EU) and with technical support from the International Fund for Agricultural Development (IFAD). Appendix A. Supplementary information Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.geoderma.2016.03.022. References Abbadie, L., Gignoux, J., Roux, X., et al., 2006. Lamto: structure, functioning, and dynamics of a savanna ecosystem. Springer, New York, USA. Amundson, R., Austin, A.T., Schuur, E., et al., 2003. Global patterns of the isotopic composition of soil and plant nitrogen. Glob. Biogeochem. Cycles 17. http://dx.doi.org/10. 1029/2002GB001903. Aranibar, J.N., Otter, L., Macko, S.A., et al., 2004. Nitrogen cycling in the soil–plant system along a precipitation gradient in the Kalahari sands. Glob. Chang. Biol. 10, 359–373. Atangana, A., Khasa, D., Chang, S., et al., 2014. Ecological Interactions and Productivity in Agroforestry Systems. Tropical Agroforestry. Springer, pp. 151–172. Bai, E., Boutton, T.W., Liu, F., et al., 2009. Spatial variation of the stable nitrogen isotope ratio of woody plants along a topoedaphic gradient in a subtropical savanna. Oecologia 159, 493–503. Bai, E., Boutton, T.W., Liu, F., et al., 2013. 15N isoscapes in a subtropical savanna parkland: spatial-temporal perspectives. Ecosphere 4 (1), 4. http://dx.doi.org/10.1890/ES1200187.1. Batjes, N.H., Sombroek, W.G., 1997. Possibilities for carbon sequestration in tropical and subtropical soils. Glob. Chang. Biol. 3, 161–173. Bedard-Haughn, A., Van Groenigen, J.W., Van Kessel, C., 2003. Tracing 15N through landscapes: potential uses and precautions. J. Hydrol. 272, 175–190. Biancalani, R., Nachtergaele, F., Petri, M., et al., 2011. Land degradation assessment in drylands: methodology and results. LADA project report. FAO. Bird, M.I., Veenendaal, E.M., Lloyd, J.J., 2004. Soil carbon inventories and δ13C along a moisture gradient in Botswana. Glob. Chang. Biol. 10, 342–349. Biswas, A., Si, B.C., Walley, F.L., 2008. Spatial relationship between δ15N and elevation in agricultural landscapes. Nonlinear Process. Geophys. 15, 397–407. Blagodatskaya, E., Yuyukina, T., Blagodatsky, S., et al., 2011. Turnover of soil organic matter and of microbial biomass under C3–C4 vegetation change: consideration of 13C fractionation and preferential substrate utilization. Soil Biol. Biochem. 43, 159–166. Booth, M.S., Stark, J.M., Rastetter, E., 2005. Controls on nitrogen cycling in terrestrial ecosystems: a synthetic analysis of literature data. Ecol. Monogr. 75, 139–157. Brady, N.C., Weil, R.R., 2007. The nature and properties of soils. 14 ed. Prentice Hall, Inc., New Jersey, USA. Brandon, K., 2014. Ecosystem Services from Tropical Forests: Review of Current Science. Washington DC. CGD Working Paper 380. Center for Global Development, Washington DC. Chappell, A., Baldock, J., Sanderman, J., 2015. The global significance of omitting soil erosion from soil organic carbon cycling schemes. Nat. Clim. Chang. http://dx.doi.org/10. 1038/nclimate2829. Chiti, T., Grieco, E., Perugini, L., et al., 2014. Effect of the replacement of tropical forests with tree plantations on soil organic carbon levels in the Jomoro district, Ghana. Plant Soil 375, 47–59. de Blécourt, M., Hänsel, V.M., Brumme, R., et al., 2014. Soil redistribution by terracing alleviates soil organic carbon losses caused by forest conversion to rubber plantation. For. Ecol. Manag. 313, 26–33. de Freitas, A.D.S., Sampaio, E.V.D.S.B., de Souza Ramos, A.P., et al., 2015. Nitrogen isotopic patterns in tropical forests along a rainfall gradient in Northeast Brazil. Plant Soil 391, 109–122. de Graaff, J., Amsalu, A., Bodnár, F., et al., 2008. Factors influencing adoption and continued use of long-term soil and water conservation measures in five developing countries. Appl. Geogr. 28, 271–280. Ellis-Jones, J., Tengberg, A., 2000. The impact of indigenous soil and water conservation practices on soil productivity: examples from Kenya, Tanzania and Uganda. Land Degrad. Dev. 11, 19–36.

G. Saiz et al. / Geoderma 274 (2016) 1–9 Fang, Y., Gundersen, P., Zhang, W., et al., 2009. Soil–atmosphere exchange of N2O, CO2 and CH4 along a slope of an evergreen broad-leaved forest in southern China. Plant Soil 319, 37–48. FAO, 2013. Climate Smart Agriculture Sourcebook. Module 4: Soils And Their Management For Climate-Smart Agriculture. FAO Rome, Italy. Feller, C., Beare, M., 1997. Physical control of soil organic matter dynamics in the tropics. Geoderma 79, 69–116. Gamfeldt, L., Snäll, T., Bagchi, R., et al., 2013. Higher levels of multiple ecosystem services are found in forests with more tree species. Nat. Commun. 4, 1340. http://dx.doi.org/ 10.1038/ncomms2328. Garten Jr., C.T., Miegroet, H.V., 1994. Relationships between soil nitrogen dynamics and natural 15N abundance in plant foliage from Great Smoky Mountains National Park. Can. J. For. Res. 24, 1636–1645. Gifford, R.M., Roderick, M.L., 2003. Soil carbon stocks and bulk density: spatial or cumulative mass coordinates as a basis of expression? Glob. Chang. Biol. 9, 1507–1514. Gignoux, J., Mordelet, P., Menaut, J.C., 2006. Biomass Cycle And Primary Production. In: Abbadie, L., Gignoux, J., Roux, X., Lepage, M. (Eds.), Lamto. Structure, Functioning, and Dynamics of a Savanna Ecosystem, 2006 ed. Springer, New York, pp. 115–137. Hailu, W., Moges, A., Yimer, F., 2012. The effects of “Fanya juu” soil conservation structure on selected soil physical & chemical properties: the case of Goromti Watershed, Western Ethiopia. Resour. Environ. 2, 132–140. Hammond, A.L., 1992. World Resources: 1992–93; A Guide to the Global Environment. Oxford University Press, New York 0-19-506230-2. Hinga, G., Muchena, F.N., Njihia, C.M., 1980. Physical And Chemical Methods Of Soil Analysis. National Agricultural Laboratories, Nairobi. Hobbie, E.A., Ouimette, A.P., 2009. Controls of nitrogen isotope patterns in soil profiles. Biogeochemistry 95, 355–371. IUSS Working Group WRB, 2015. World Reference Base for Soil Resources 2014, Update 2015 International Soil Classification System For Naming Soils And Creating Legends For Soil Maps. World Soil Resources Reports No. 106. FAO, Rome. Kristjanson, P., Garlick, C., Förch, W., et al., 2014. Global Summary Of Baseline Household Survey Results. CGIAR Research Program on Climate Change. Agriculture and Food Security (CCAFS). Lal, R., 2004. Soil carbon sequestration to mitigate climate change. Geoderma 123, 1–22. Liniger, H.P., Critchley, W., 2007. Where The Land Is Greener: Case Studies And Analysis Of Soil And Water Conservation Initiatives Worldwide. World Overview of Conservation Approaches and Technologies (WOCAT), Wageningen, Netherlands. Liniger, H.P., Studer, R.M., Hauert, C., et al., 2011. Sustainable Land Management In Practice: Guidelines And Best Practices For Sub-Saharan Africa. TerrAfrica, World Overview of Conservation Approaches and Technologies (WOCAT) and Food and Agriculture Organization of the United Nations (FAO), Rome. Lipper, L., Thornton, P., Campbell, B.M., et al., 2014. Climate-smart agriculture for food security. Nat. Clim. Chang. 4, 1068–1072. Liu, X.J., Mosier, A.R., Halvorson, A.D., et al., 2007. Dinitrogen and N2O emissions in arable soils: effect of tillage, N source and soil moisture. Soil Biol. Biochem. 39, 2362–2370. Liu, M., Dannenmann, M., Lin, S., et al., 2015. Ground cover rice production systems increase soil carbon and nitrogen stocks at regional scale. Biogeosciences 12, 4831–4840. http://dx.doi.org/10.5194/bg-12-4831-2015. Lloyd, J., Bird, M.I., Vellen, L., et al., 2008. Contributions of woody and herbaceous vegetation to tropical savanna ecosystem productivity: a quasi-global estimate. Tree Physiol. 28, 451–468. Mann, W., Lipper, L., Tennigkeit, T., et al., 2009. Food Security and Agricultural Mitigation in Developing Countries: Options for Capturing Synergies. Marks, E., Aflakpui, G., Nkem, J., et al., 2009. Conservation of soil organic carbon, biodiversity and the provision of other ecosystem services along climatic gradients in West Africa. Biogeosciences 6, 1825–1838. Mati, B.M., 2006. Overview of wAter And Soil Nutrient Management Under Smallholder Rain-Fed Agriculture In East Africa. Working Paper 105. International Water Management Institute. McBratney, A.B., Minasny, B., 2010. Comment on “Determining soil carbon stock changes: Simple bulk density corrections fail”[Agric. Ecosyst. Environ. 134 (2009) 251–256]. Agric. Ecosyst. Environ. 136, 185–186. Mordelet, P., Abbadie, L., Menaut, J.C., 1993. Effects of tree clumps on soil characteristics in a humid savanna of West Africa (Lamto, Cote d'Ivoire). Plant Soil 153, 103–111. Mwangangi, M., Mutie, I., Mango, J., 2012. Summary of Baseline Household Survey Results: Makueni, Kenya. CGIAR Research Program on Climate Change. Agriculture and Food Security (CCAFS), Copenhagen, Denmark (Available online at: www.ccafs. cgiar.org.).

9

Nardoto, G.B., Quesada, C.A., Patiño, S., et al., 2014. Basin-wide variations in Amazon forest nitrogen-cycling characteristics as inferred from plant and soil 15N:14N measurements. Plant Ecol. Divers. 7, 173–187. Nyasimi, M., Amwata, D., Hove, L., et al., 2014. Evidence of Impact: Climate Smart Agriculture in Africa. CCAFS Working Paper. Oldeman, L.R., Hakkeling, R.T.A., Sombroek, W.G., 1991. World Map of the Status of Human-Induced Soil Degradation. An Explanatory Note. Global Assessment of Soil Degradation (GLASOD), October 1991, Second Revised ed. International Soil Reference and Information Centre (ISRIC) and United Nations Environment Programme (UNEP), Wageningen. Ong, C.K., Leakey, R., 1999. Why tree–crop interactions in agroforestry appear at odds with tree–grass interactions in tropical savannahs. Agrofor. Syst. 45, 109–129. Penny, R., 2009. Desertification and Deforestation in Africa. Land Use, Land Cover and Soil Sciences 5. Peukert, S., Bol, R., Roberts, W., et al., 2012. Understanding spatial variability of soil properties: a key step in establishing field-to farm-scale agro-ecosystem experiments. Rapid Commun. Mass Spectrom. 26, 2413–2421. Post, W.M., Emanuel, W.R., Zinke, P.J., et al., 1982. Soil carbon pools and world life zones. Nature 298, 156–159. http://dx.doi.org/10.1038/298156a0. Rao, M.R., Nair, P., Ong, C.K., 1998. Biophysical interactions in tropical agroforestry systems. Agrofor. Syst. 53, 3–50. Rovira, P., Sauras, T., Salgado, J., et al., 2015. Towards sound comparisons of soil carbon stocks: a proposal based on the cumulative coordinates approach. Catena 133, 420–431. Rumpel, C., Kögel-Knabner, I., 2011. Deep soil organic matter—a key but poorly understood component of terrestrial C cycle. Plant Soil 338, 143–158. Saiz, G., Albrecht, A., 2015. Methods for Smallholder Quantification of Soil Carbon Stocks and Stock Changes. Measurement methods Standard Assessment Of Agricultural Mitigation Potential And Livelihoods (SAMPLES) (http://www.samples.ccafs.cgiar.org/ soil-carbon-stocks.html. Accessed 23 December 2015). Saiz, G., Green, C., Butterbach-Bahl, K., et al., 2006. Seasonal and spatial variability of soil respiration in four Sitka spruce stands. Plant Soil 287, 161–176. Saiz, G., Bird, M.I., Domingues, T., et al., 2012. Variation in soil carbon stocks and their determinants across a precipitation gradient in West Africa. Glob. Chang. Biol. 18, 1670–1683. Saiz, G., Bird, M.I., Wurster, C.M., et al., 2015a. The influence of C3 and C4 vegetation on soil organic matter dynamics in contrasting semi-natural tropical ecosystems. Biogeosciences 12, 5041–5059. http://dx.doi.org/10.5194/bg-12-5041-2015. Saiz, G., Wynn, J.G., Wurster, C.M., et al., 2015b. Pyrogenic carbon from tropical savanna burning: production and stable isotope composition. Biogeosciences 12, 1849–1863. Scholes, R.J., Archer, S.R., 1997. Tree–grass interactions in savannas. Annu. Rev. Ecol. Syst. 28, 517–544. Sijmons, K., Kiplimo, J., Förch, W., et al., 2013. CCAFS site atlas–Makueni/Wote. CCAFS site atlas series. The CGIAR Research Program on Climate Change, Agriculture and Food Security (CCAFS), Copenhagen. Silver, W., Neff, J., McGroddy, M., et al., 2000. Effects of soil texture on belowground carbon and nutrient storage in a lowland Amazonian forest ecosystem. Ecosystems 3, 193–209. Stevenson, B.A., Parfitt, R.L., Schipper, L.A., et al., 2010. Relationship between soil δ15N, C/N and N losses across land uses in New Zealand. Agric. Ecosyst. Environ. 139, 736–741. Tenge, A.J., Hella, J.P., 2005. Financial efficiency of major soil and water conservation measures in West Usambara highlands, Tanzania. Appl. Geogr. 25, 348–366. Tian, Y., Su, D., Li, F., et al., 2003. Effect of rainwater harvesting with ridge and furrow on yield of potato in semiarid areas. Field Crop Res. 84, 385–391. Van den Heuvel, R.N., Hefting, M.M., Tan, N., et al., 2009. N2O emission hotspots at different spatial scales and governing factors for small scale hotspots. Sci. Total Environ. 407, 2325–2332. Vilain, G., Garnier, J., Tallec, G., et al., 2010. Effect of slope position and land use on nitrous oxide (N2O) emissions (Seine Basin, France). Agric. For. Meteorol. 150, 1192–1202. Wang, L., Okin, G.S., D’Odorico, P., et al., 2013. Ecosystem-scale spatial heterogeneity of stable isotopes of soil nitrogen in African savannas. Landsc. Ecol. 28, 685–698. Wynn, J.G., Bird, M.I., 2007. C4-derived soil organic carbon decomposes faster than its C3 counterpart in mixed C3/C4 soils. Glob. Chang. Biol. 13, 2206–2217. http://dx.doi.org/ 10.1111/j.1365-2486.2007.01435.x. Wynn, J.G., Bird, M.I., Vellen, L., et al., 2006. Continental-scale measurement of the soil organic carbon pool with climatic, edaphic, and biotic controls. Global Biogeochem. Cycles http://dx.doi.org/10.1029/2005GB002576.