|
|
, Dominique SERÇAGreenhouse gas (GHG) emissions from the surface soils and surface water receiving animal excreta may be important components of the GHG balance of terrestrial ecosystems, but the associated processes are poorly documented in tropical environments, especially in tropical arid and semi-arid areas. A typical sylvo-pastoral landscape in the semi-arid zone of Senegal, West Africa, was investigated in this study. The study area (706 km² of managed pastoral land) was a circular zone with a radius of 15 km centered on a borehole used to water livestock. The landscape supports a stocking rate ranging from 0.11 to 0.39 tropical livestock units per hectare depending on the seasonal movements of the livestock. Six landscape units were investigated (land in the vicinity of the borehole, natural ponds, natural rangelands, forest plantations, settlements, and enclosed plots). Carbon dioxide (CO2), nitrous oxide (N2O) and methane (CH4) fluxes were measured with static chambers set up at 13 sites covering the six landscape units, and the 13 sites are assumed to be representative of the spatial heterogeneity of the emissions. A total of 216 fluxes were measured during the one-year study period (May 2014 to April 2015). At the landscape level, soils and surface water emitted an average 19.8 t C-CO2eq/(hm²•a) (CO2: 82%, N2O: 15%, and CH4: 3%), but detailed results revealed notable spatial heterogeneity of GHG emissions. CO2 fluxes ranged from 1148.2 (±91.6) mg/(m²•d) in rangelands to 97,980.2 (±14,861.7) mg/(m²•d) in surface water in the vicinity of the borehole. N2O fluxes ranged from 0.6 (±0.1) mg/(m²•d) in forest plantations to 22.6 (±10.8) mg/(m²•d) in the vicinity of the borehole. CH4 fluxes ranged from -3.2 (±0.3) mg/(m²•d) in forest plantations to 8788.5 (±2295.9) mg/(m²•d) from surface water in the vicinity of the borehole. This study identified GHG emission “hot spots” in the landscape. Emissions from the surface soils were significantly higher in the landscape units most frequently used by the animals, i.e., in the vicinity of the borehole and settlements; and emissions measured from surface water in the vicinity of the borehole and from natural ponds were on average about 10 times higher than soil emissions.
Climate change is among the most pressing challenges facing humanity today. It is now widely accepted that the concentrations of greenhouse gases (GHGs) have risen at unprecedented rates since the industrial revolution and that carbon dioxide (CO2), methane (CH4) and nitrous oxide (N2O) are the three GHGs that have contributed more than 90% of human-inducedglobal warming(IPCC, 2013). The GHGexachanges between terrestrial ecosystems and the atmosphere play a major role in regulating the concentrations of these three greenhouse gases in the atmosphere (Chevallier et al., 2015). However, our knowledge on the magnitude of GHG exchanges between terrestrial ecosystems and the atmosphere isstill rather limited (Wang et al., 2013; Ahlströ m et al., 2015). Particularly, large uncertainties remain regarding soil-air fluxes of GHGs in tropical drylands (Ahlströ m et al., 2015; Borges et al., 2015).
Pastoral systems occupy most of drylands in the world (Nori et al., 2005) and account for a quarter of the total drylands in Africa(Boval and Dixon, 2012). In West Africa (i.e., our study area), pastoral systems were estimated to have expanded considerably in recent past because of the observed global warming and the expansion trend isexpected to continue in the near future under projected warming (Thornton and Herrero, 2015). Pastoral ecosystems are characterized by constraining climatic conditions with limited precipitation falling in a limited time frame (Martí nez et al., 2011). Pastoralism is a way of life that is still very common in world’ s arid and semi-arid zones (Turner et al., 2014). Pastoral communities have succeeded in making best advantage of the natural resources based on their mobility and “ ancestral” knowledge (Manoli et al., 2014) and consequently developedtheir economy-ecology coupled social systems in which a large number of livelihoods are reported to have the highest rates of GHG emissions per unit of animal products (Gerber et al., 2013). Despite the increasingly acknowledged importance of pastoral ecosystems in the global GHG cycles and their potentials in further elevating the GHG concentrations in the atmosphere, in situ measurements are still not known (Tagesson et al., 2015a).
Soils in rangelands, combined with livestock productions, are reported to be responsible for a large share of GHG emissions, but these soils can also act as carbon sinks under certain conditions(Soussana et al., 2010; Valentini et al., 2014). Under arid to semi-arid rangelands, soil moisture is one of the main factors controlling emissions of CO2 (Kuzyakov and Gavrichkova, 2010; Yemadje et al., 2016) and N2O from the soil to the atmosphere (Ussiri and Lal, 2013). It should be noted that although CH4 emissions occur only in hydromorphic conditions, CH4 emissions may also be significant, since there are extensive water bodies (e.g., ponds) and wetlands in certain parts of world’ s drylands (Serrano-Silva et al., 2014). It should also be stressed that large amounts of manure are produced and depositedin pastoral ecosystems, thereby directly or indirectly affecting GHG emissions via the involvement of microbial processes (e.g., methanogenesis, nitrification, and denitrification) and modification of soil characteristics(e.g., texture and moisture)(Thangarajan et al., 2013). In addition, livestock, as a vector of organic matter, can result in the heterogeneous distribution of animal excreta in the landscapes and hence increase spatial heterogeneities in soil properties, available nutrients andbiomasses, most likely leading to intensification of GHG emissions via enhanced microbiological processes (Smith et al., 2003).
The purpose of this study was to evaluate the magnitude of the spatial heterogeneity of the emissions of CO2, N2O and CH4 from the soils and the surface water bodies in a pastoral landscape, and to identify possible links between the emissions and the livestock in ecosystem functioning. Our specific objectives were to: (1) quantify GHG fluxes in the different landscape units; (2) assess the spatial heterogeneities of GHG fluxes; and (3) link the heterogeneities to livestock habits and movements. To achieve these objectives, CO2, N2O and CH4 fluxes from soils and water surfaces were monitored for a whole year across different landscape units within the area influenced by the Widou borehole. This area is assumed to be representative of sylvo-pastoral ecosystems in northern Senegal, West Africa.
The study was carried out in the sylvo-pastoral Ferlo region of Senegal, West Africa. The Ferlo region is characterized by high seasonal, annual and decadal rainfall variabilities and relatively high mean annual temperatures (Martí nez et al., 2011). Three main seasons, each lasting four months, can be distinguished: a hot-wet season from July to October, a cold-dry season from November to February of next year and a hot-dry season from March to June. The mean annualprecipitation is rather low (296 mm) and highly variable (Miehe et al., 2010). Mean annual temperature is 27.7° C and monthly averages fluctuate between a maximum of 30.2° C in October and a minimum of 24.5° C in January (Ndiaye et al., 2014).
The Ferlo region is organized around a network of boreholes at 30 km intervals. The boreholes were dug in the 1950s and 1960s (Manoli et al., 2014). This study focused on a 706 km² area around the Widou borehole (15° 59′ N, 15° 19′ W), a circular zone with a radius of 15 km from the borehole (Fig. 1). The area is representative of the semi-arid sylvo-pastoral ecosystems in the region. We selected this particular borehole because of the availability of a comprehensive database createdas part of the organized survey activities(Ancey et al., 2008; Bah et al., 2010). The ‘ enclosures’ in Figure 1 refer to the enclosed experimental grazing plots created over 30 years ago as a part of a project implemented by a German agency (Miehe et al., 2010). An exhaustive survey of the 354 settlements located in the vicinity of the Widou borehole was made in December 2013 to quantify the size of the herds of the different animal species. The landscape supports a stocking rate ranging from 0.11 to 0.39 tropical livestock units per hectare(or TLU/hm2) depending on the seasonal movements of the herds.
 | Fig. 1 Map of the study area around Widou borehole |
A combination of the GIS technique and field observations was used to describe and mapthe main landscape units (Fig. 1). Six landscape units were defined based on land uses and topographic variations using a Landsat ETL+204-049 image acquired on 3November 2010.
(I) Vicinityof theborehole (< 1 km² , or 0.1% of the total 706 km2 area) corresponds to the part of the landscape in which the borehole is located. The borehole is used to water herds and its vicinity is the place where the animals rest after being watered. The soilsnear the water tank of the borehole are waterlogged all year round. We investigated both stagnant surface water and the soils around the tank.
(II) Settlements (44 km² , or 6.3% of the total area) correspond to the areas occupied by herder’ s homestead and a corral for the herd. These areas are characterized by high soil organic matter (SOM)contents due to the long-term accumulation of animal manureduring night corralling.
(III) Naturalponds (19 km² , 2.7%) correspond to areas of clayey soils where stagnant water accumulates temporarily during the wetseason and the ponds normallydry up soon after the wet season is over. We investigated both the water from the ponds and the soilsalong the shorelines of the ponds.
(IV) Forest plantations (6 km² , 0.9%)were established by the government of Senegal as a part of the Great Green Wall project. They are mainly composed ofAcacia spp.
(V) Native rangelands (635 km² , 89.9%) are occupied by natural vegetation (grass, shrubs, and trees) and are actively pastured by livestock.
(VI) Enclosures (< 1 km² , 0.03%)refer to the enclosed experimental grazing plots created over 30 years ago.
To investigate seasonal variabilities of GHG fluxes, we conducted field campaigns on six occasions between May 2014 and January 2015, once during the hot dry-season (May 2014), once a month during the hot-wet season (from July to October 2014), and once during the cold-dry season (January 2015). Observations were made at 13 sites (native rangelands, n=5; natural ponds, n=2; settlements, n=2; forest plantations, n=1; enclosures, n=1; and in the vicinity of the borehole, n=2). The selection of the sampling sites took into consideration both the areal extent of the area occupied by each landscape unit and the potential ability to emit GHGs.
2.2.1CO2, N2O and CH4fluxes
GHG fluxes from each of the 13 sites were quantified using rectangular stainless steel static chambers described in Serç a et al. (1994). All the measurements taken at each site were made in duplicate under parasols during the day.
Air samples (20 mL) were collected immediately after the closure of the chamber and subsequently at 30-minute intervals during the following 90 minutes. Samples for CO2 analysis were stored in serum vials pre-flushed with N2(Gué rin et al., 2007). Samples for CH4 and N2O analysis were stored in serum vials initially filled with a salt-saturated solution (Deshmukh et al., (2014). The analyses were performed by SRI 8610C gas chromatography (SRI Instruments, Torrance, CA, USA) equipped with a 63Ni electron detector (ECD) and a Flame Ionization Detector (FID)methanizer.
The diffusive fluxes were calculated from the slope of the linear regression of the concentration of gas in the chamber versus time. The measured fluxes were discarded when the correlation coefficient r2 was < 0.8 (Deshmukh et al., 2014).
Samples were collected in the ponds to measure the concentrations of dissolved GHGs in the water. The water samples were collected and poisoned according to Gué rin and Abril (2007), and analyzed by gas chromatography after the creation of a headspace with N2. Dissolved concentrations of CO2, CH4 and N2O were computed with the solubility coefficient of Weiss (1974), Yamamoto et al. (1976) and Wanninkhof (1992) for CO2, CH4 and N2O, respectively. Diffusive fluxes were calculated from surface concentrations using the turbulent boundary layer model (Liss and Slater, 1974) with a gas transfer velocity (k600) of 2 cm/h and the in situ water temperature.
2.2.2Soil chemical properties
Soils were sampled in triplicate to a depth of 10 cm at all 13 sites in January 2015. The samples were used to determine soil total carbon, nitrogen, and phosphorus, NH4+, NO3-, bulk density, and particle sizes in the certified ISO9001 LAMA laboratory (2008 by Euro-Quality System) according to the methods described by Pansu and Gautheyrou (2006).
2.2.3Collection of solid manure and monitoring of livestock
Each month, the solid manure deposited on the surface soils was collected manually in four 0.25 hm2 plots in all the landscape units. The total quantity of solid manure in each livestock category (cattle, small ruminants, and non-ruminants) was weighed. The samples were then dried in a stove for three days at 65° C to measure dry matter content. Three ruminant herds were also monitored for 1-2 day(s) each month to estimate the time the animals spent in each landscape unit.
2.3.1 Correlation between fluxes and environmental parameters
All statistical analyses were conducted with R software (RDevelopment CoreTeam, 2015)using the ‘ ‘ lme4’ ’ package for linear regression analysis and ‘ ‘ ade4’ ’ (Dray and Dufour, 2007) for ANOVA. Data were checked for normality and variance analyses. Significant differences in means between the six landscape units and between the three seasons were identified with Tukey’ s honest significant differences (HSD) test. Pearson correlation analysis was used to test the relationship between soil GHG emissions and the quantity of solid manure deposited on the surface soils.
2.3.2Scaling up measurements for temporal and spatial estimations
For temporal upscaling (monthly/season average), the observations made on a specific day of the month were generalized to the month/season assuming that the magnitude of the emissions was the same every day of the month/season. For spatial upscaling (landscape unit average), a mean and a standard deviation were calculated for each landscape unit based on all observations of the landscape unit concerned.
Two different approaches were used to extrapolate emissions to the whole landscape. The “ only native rangelands” approach is a simplified and also a widely practiced method (Valentini et al., 2014). It only takes account of emissions from rangelands, as the entire landscape comprises this unit (in reality 90% of the borehole territory is covered by native rangelands). The second approach, called “ all landscape units” approach, accountsfor the spatial heterogeneity of GHG emissions. In both cases, each flux measurement was weighted according to the length of the sampling season and also to the relative proportions of the areal extent of the different landscape units in the context of the entire landscape concerned to calculate the total GHG emissions at the year time scale and at the landscape spatial scale.
Extrapolations at year and landscape levels were converted into CO2 equivalents based on theglobal warming potentials of the three gases proposed by IPCC (2013). These global warming potentials are 1, 34, and 298, for CO2, CH4, and N2O, respectively.
Table 1 lists the livestock stocking rate in a given landscape unit, the quantity of solid manure deposited on the soil per month and the physical and chemical characteristics of the soil in the different landscape units. The livestock stocking rate is expressed in tropical livestock units (TLU) per unit area (hm2), whereone TLU is equivalent to an animal of 250kg live weight. The average livestock rate in a landscape unit ranged from 0.08 (± 0.02)TLU/(hm² • month)in rangelands to 10.84 (± 6.96)TLU/(hm² • month)in the vicinity of the borehole. The resulting excretion rates ranged from 13.7 (± 6.9) kgDM/(hm² • month) in rangelands to 68.8 (± 63.6) kgDM/(hm² • month) in the vicinity of the borehole. It should be noted that the dung excretion was quantified in kilogram of dry matter (DM). The soils in the vicinity of the borehole were richer in nutrients and organic matter (includingtotal carbon, total nitrogen, mineral nitrogen, and total phosphorus). However, the soils near settlements are richer in mineral nitrogen than the soils in the vicinity of the borehole.
 | Table 1 Livestock stocking rate, manure deposition, physical and chemical characteristics of the soil (0-10cm) |
Table 2 lists soil GHG emissions in the different landscape units. Soil CO2 fluxes ranged from 1148.2 (± 91.6) mg/(m² • d) in rangelands to 97, 144.1 (± 52, 450.1) mg/(m² • d) in the vicinity of the borehole. The mean CO2 flux across all landscape units was 4352.7 (± 1220.1)mg/(m² • d). Soil N2O fluxes ranged from 0.6 (± 0.1) mg/(m² • d) in forest plantation to 22.6 (± 10.8) mg/(m² • d) in the vicinity of the borehole. The mean N2O flux across all landscape units was 2.5 (± 0.9) mg/(m² • d). Soil CH4 fluxes ranged from -3.2 (± 0.3) mg/(m² • d) in forest plantationsto 691.3 (± 352.6) mg/(m² • d) in the vicinity of the borehole. The mean CH4 flux across all landscape units was 2.6 (± 2.5) mg/(m² • d).
| Table 2 Soil greenhouse gas emissions in the six different landscapes |
CO2 and N2O fluxes exhibited similar trends. CO2 and N2O emissions from the soils in the vicinity of the borehole and in the settlements were significantly higher (P< 0.05) than the fluxesfrom forest plantations, enclosures, rangelands and the shoreline of the ponds. But, there were no significant differences in CO2 and N2O emissions between the vicinity of the borehole and the settlements. The differences in CO2 and N2O emissions among forest plantations, enclosures, rangelands and the shorelines of the ponds were also not significant.
The spatial pattern ofCH4 emissions from the soils slightly differed from those of CO2 and N2O. The highestCH4 emissionswere measured in the vicinity of the borehole and theshorelines of the ponds, being significantly higher (P< 0.05) than the fluxes from forest plantations, enclosures, rangelands and settlements. There were no significant differences between the vicinity of the borehole and the shorelines of the ponds and also no significant differences among forest plantations, enclosures, rangelands, and the settlements.
Concerning GHG exchanges between water surfaces and the atmosphere (in the natural ponds and in the vicinity of the borehole), all the water samples were under-saturated in N2O relative to the atmosphere, and N2O concentrations were below the limit of detection. CO2 fluxes ranged from 5855.1 (± 970.5) mg/(m² • d) in natural pond water to 97, 980.2 (± 14, 861.6) mg/(m² • d) in the stagnant water in the vicinity of the borehole. The calculated overall mean (for both landscape units) of CO2 fluxes was 30, 640.9 (± 5481.2) mg/(m² • d). CH4 fluxes ranged from 142.2 (± 1.1) mg/(m² • d) in natural pond water to 8788.5 (± 2295.9) mg/(m² • d) in the stagnant water around the borehole. The calculated overall mean (for both landscape units) of CH4fluxes was 3170.8 (± 659.4) mg/(m² • d). The pond CO2 fluxes were not significantly different from the soilCO2 fluxes, whereas pond CH4 fluxes were significantly higher (P< 0.05) than the soilCH4 fluxes. On average, CH4 fluxes from ponds were 45times higher than those from the soils.
Figure 2 shows the contributions of the three GHGs to total GHG emissions at landscape level. CO2 was the main contributor and accounted for 82% of total emissions. N2O accounted for 15% and CH4for 3%.Rangelands, which represent 90% of the entire territory, were the main contributorstoCO2 andN2O emissions despite of the fact that the fluxper unit area in the rangelands is low. The vicinity of the borehole and the shorelines of the pondswere the main sourcesof CH4 emissions (93% of CH4 emissions) despite the fact that the borehole and the pondsaccounted for less than 3% of the entire study area.
 | Fig. 2 Contributions of the three greenhouse gas (CO2, N2O, and CH4) emissions from soil and water at the landscape level |
Figure 3 shows the contributions of the three seasons (wet, cold-dry, and hot-dry seasons) to average daily emissions of GHG from the soils and water at the landscape level. Most emissions (65%) occurred during the hot-wet season. The cold-dry and hot-dry seasons represented 14% and 21% of total emissions, respectively. Rangelands were the main sources of GHG emissions, especially during the cold-dry and hot-dry seasons, accounting for more than 90% of total emissions. The contributions of settlements, ponds, and the vicinity of the borehole increased during the hot-wet season mainly due to the higher CH4 emissions rates (see Section 3.1).
 | Fig. 3 Contributions of the greenhouse gas emissions in three seasons (hot-wet, cold-dry, and hot-dry seasons) from soil and water at the landscape level |
This study illustrates the high spatial heterogeneity in GHG exchanges with the atmosphere.In the case of CH4, the highest emissions per surface unit were recorded in permanentlyor temporarilyflooded landscape units (borehole and ponds) where hydromorphic conditions favor methanogenesis (Serrano-Silva et al., 2014). These units are also frequently visited by animals that contribute to the accumulation of manure (Table 1). When only soil CH4emissions were taken into account, ponds and the vicinity of the borehole emitted 58.8 (± 74.3) and 277.9 (± 72.3) mg/(m² • d), respectively. When surface water CH4emissions were also taken into account, these two landscape units emitted 147.7 (± 92.9)and 656.7 (± 145.7) mg/(m² • d), respectively. CH4emissions from the remaining landscape units were very low or negative especially from units withsandy soils and limited manure inputs (rangelands and forest plantations). This isquite consistent with reports in the literature that consider non-flooded grassland soils to be sinks for atmospheric CH4(Mosier et al., 2004)and that CH4 emissions are negligible in rain-fed agro-ecosystems (Nyamadzawo et al., 2014).
The highest CO2 and N2O emissions per surface unit mainly came from the landscape units in which the animals spent the longest time, i.e., the vicinity of the borehole, settlements and around the ponds. This can be explained by the high inputs of manure that stimulate microbial activities. Figure 4 shows the positive effect of inputs of manure on GHG emissions as reported byLin et al. (2009). The correlation is stronger for CO2 and N2O fluxes (R² > 0.5, P< 0.05; Figs. 4a and b) than for CH4 (Fig. 4c). Manure is a direct source of CO2 emissionsand also a direct source of N2O emissions (Yamulki et al., 1999; Clemens and Ahlgrimm, 2001; Li and Kelliher, 2007; Saggar et al., 2007). The elevated emissions were normally accomplished through increasingsoil C and N contents (Hiernaux et al., 1999) and also through stimulating soil microbiological activities(Chotte et al., 2012; Petersen et al., 2013).
 | Fig. 4 Regressionrelationships between soil greenhouse gas emissions and inputs of manure within the study area including all landscape units |
The spatial distribution of manure deposition certainly explains a large proportion of the spatial heterogeneity of the GHG emissions observed in this study. Other studies (e.g., Schlecht et al., 2004) have also demonstrated the important role played by livestock in the spatial distribution of soil nutrients and carbon stocks. Inthe Sudanian savanna agro-pastoral ecosystemsin West Africa, traditional practices like free grazing and night corralling lead to nutrient and carbon transfers from the periphery to the core of the landscapes (Manlay et al., 2004a), resulting in ring-like organization of the landscape with a positive gradient of nutrients and carbon storage in the soil from the rangelands to the dwellings(Diarisso et al., 2015). This organization may explain the marked differences in CO2 emissions rates between rangelands and croplands in Sudanian savanna agro-pastoral ecosystems (Brü mmer et al., 2009). Similarly in Sahelian drier sylvo-pastoral ecosystems, quite like the ecosystem studied here, the spatial heterogeneity of GHG emissions can be explained by livestock-related nitrogen and carbon transfers. That is, animals mainly intake biomasses in certain landscape units (e.g., rangelands, tree plantations) while excretion preferentially occurs in others while resting (in settlements) or drinking (stagnant water around the borehole and in natural ponds). Consequently, organic matter is concentrated in particular landscape units.
Figure 5 compares the total annual emissions (in CO2 equivalent) of the three main GHGs using two approaches described in Section 2.3.2. The estimate of total annual emissions (in CO2 equivalents) using “ only native rangelands” approachwas 1.2× 106t C-CO2eq/aand the estimate using the “ all landscape units” approach was 1.4× 106t C-CO2eq/ain the Widou territory (706 km2). CH4 emissions are not taken into account in the “ only native rangelands” approach because they are negligible in rangelands, and mainly occur in the vicinity of the borehole and the natural ponds. The “ only native rangelands” approach leads to a 14% underestimation of total emissions. These results question the fact that extrapolations at landscape and sometimes at regional levels are frequently based on spatially limited measurements because data on tropical regions are scarce (Merbold et al., 2009; Galy-Lacaux and Delon, 2014; Tagesson et al., 2015b). For instance, most regional African GHG balances are based on data provided by the regional network of flux towers (Bombelli et al., 2009; Mbow, 2014; Valentini et al., 2014).
 | Fig. 5 Annual total greenhouse gas emissions from soil and water at landscape level calculated using two different approaches |
Another way to increase the robustness of the assessment of the yearly total GHG emissions from soil in tropical sylvo-pastoral ecosystems is the frequency of the GHG measurements. In the dry seasons (both hot-dry and cold-dry seasons), emissions fluxes remain stably low over time because the soil is dry and the microbiological activity is considerably slowed down, justifying less frequent measurements (see Section 3.2). But during the wet season (i.e., the hot-wet season), microbiological activity and the resulting GHG fluxes are extremely high. The occurrence of scaling up errors is certainly higher in the wet season despite the fact more frequent measurements were made in this season in the present study. Indeed, the microbiological activities and other GHG emission-related processes in the soils of tropical arid and semi-arid areas go through a dormant period during the dry season and are subjected tomineralization flushes after rainfall events at the onset of the wet season (Kim et al., 2012). It is thus difficult to capture the resulting variability of GHG emissions. It would require a dedicated monitoring program with very frequent measurements. Unfortunately, the present study did not have the required measurement frequency during the wet season because our primary objective was to analyze the spatial variability of GHG emissions. To better represent the mineralization flushes and peak emissions, modeling would be required to correlate emissions with the dynamics of soil humidity (Saggar et al., 2007; Song et al., 2015). To date, modelling efforts have been constrainedby the lack of available data mainly due to the inaccessible nature of tropical rangeland ecosystems (Krü ger et al., 2013).
To our knowledge, the data set on the carbon dioxide, nitrous oxide, and methane fluxes in northern Senegal reported here is the first from a tropical sylvo-pastoral ecosystem. Our results provide evidence for quite notable spatial heterogeneity of GHG emissions.The spatial heterogeneity of GHG emissions can be explained by livestock movements within the study area, and the resulting stocking rate variations, which affect the spatial distribution of solid manure and result in hotspots of GHG emissions. The vicinity of the borehole and the settlements had the highest manure input rates and thus the highest CO2 and N2O emissions. The stagnant water in natural ponds and in the vicinity of the borehole was the main contributor of CH4 emissions due to permanent or seasonal hydromorphic conditions. The high CH4 emissions were further enhanced by the large quantity of manure excreted by animals while they are drinking.
At the landscape level, CO2 is the main GHG (83.1% of all GHG emissions) and native rangelands arethe main landscape unit (78.8% of all GHG emissions) contributing to total GHGemissions. It should be stressed that the “ only native rangelands” approach underestimated the total annual emissions by 14% because most of CH4 emissions occur elsewhere than in rangelands and yet only rangelands were considered in the “ only native rangelands” approach. Further investigations are needed to refine the GHG footprint of sylvo-pastoral systems in the region including more frequent monitoring during the seasonal transitions with a focus on the onset of the wet season, known to be a period in which a flush of mineralization of organic matter occurs in the tropics. Long-term efforts should also be made to assess possible inter-annual variability of GHG emissions, since climate fluctuations can have impacts on the presence, movements, and excretion rates of livestock and since the impacts can modify the spatial heterogeneity ofthe GHG balance.
The study was funded by the Islamic Development Bank (the Merit Scholarship Program for High Technology) and the European Union (the FP7-KBBE-2010 Animal Change Project).
The authors have declared that no competing interests exist.
| 1 |
|
| 2 |
|
| 3 |
|
| 4 |
|
| 5 |
|
| 6 |
|
| 7 |
|
| 8 |
|
| 9 |
|
| 10 |
|
| 11 |
|
| 12 |
|
| 13 |
|
| 14 |
|
| 15 |
|
| 16 |
|
| 17 |
|
| 18 |
|
| 19 |
|
| 20 |
|
| 21 |
|
| 22 |
|
| 23 |
|
| 24 |
|
| 25 |
|
| 26 |
|
| 27 |
|
| 28 |
|
| 29 |
|
| 30 |
|
| 31 |
|
| 32 |
|
| 33 |
|
| 34 |
|
| 35 |
|
| 36 |
|
| 37 |
|
| 38 |
|
| 39 |
|
| 40 |
|
| 41 |
|
| 42 |
|
| 43 |
|
| 44 |
|
| 45 |
|
| 46 |
|
| 47 |
|
| 48 |
|
| 49 |
|
| 50 |
|
| 51 |
|
| 52 |
|
| 53 |
|
| 54 |
|
| 55 |
|
| 56 |
|
| 57 |
|
