Carbon for Farmers: Assessing the Potential for Soil Carbon Sequestration in the Old Peanut Basin of Senegal

Page 1

CARBON FOR FARMERS: ASSESSING THE POTENTIALFOR SOIL CARBON SEQUESTRATION IN THE OLD PEANUT

BASIN OF SENEGAL

PETRA TSCHAKERT1

Arid Lands Resource Sciences, University of Arizona, 1955 E. 6th Street, Tucson, AZ 85719, U.S.A.E-mail: [email protected]

Abstract. Carbon sequestration in soil organic matter of degraded Sahelian agro-ecosystems couldplay a significant role in the global carbon (C) uptake through terrestrial sinks while, simultaneously,contributing to sustainable agriculture and desertification control. The paper documents the resultsof a two-year pilot project in Senegal assessing real project opportunities with main emphasis on theWest-Central Agricultural Region (“Old Peanut Basin”). Current total system C content in this region,calculated on the basis of in situ soil and biomass carbon measurements, amounted to 28 t ha–1 with11 t C ha–1 in soils (0–20 cm) and 6.3 t C ha–1 in trees. Potential changes in soil C, simulated withCENTURY for a 25-year period, ranged from –0.13 t C ha–1 yr–1 under poor management to +0.43 tC ha–1 yr–1 under optimum agricultural intensification. Simulated changes in crop yields varied from–62% to +200% under worst and best management scenarios respectively. Best management practicesthat generate the highest sequestration rates are economically not feasible for the majority of localsmallholders, unless considerable financial support is provided. Especially when applied on a largerscale, such packages risk to undermine local, opportunistic management regimes and, in the long run,also the beneficiaries’ capacity to successfully adapt to their constantly changing environment.

1. Introduction

Carbon sequestration in small-scale farming systems in drylands is increasinglypromoted as a win-win strategy. It is assumed to simultaneously increase the car-bon uptake through terrestrial sinks, thus playing a crucial role in global climatechange mitigation, and to contribute to improved well being among local small-holders through more sustainable land use and management practices (Lal, 1999;Lal et al., 1999; Woomer et al., 1997). According to the United Nations Environ-ment Programme (UNEP), 90% of the African drylands are degraded. Thus, theirrestoration seems of imminent importance to those who inhabit and depend on theselands for a living.

Soil organic carbon (SOC) and CO2 fixing capacity of semi-arid lands are fairlylow. In Senegal, SOC estimates for the West-Central Agricultural Region rangefrom 4.5 t C ha–1 for continuously cultivated areas with short-term fallowing to18 t C ha–1 for non-degraded savannas (Tiessen et al., 1998; Ringius, 2002). It is

1Present address: McGill University, Department of Biology, 1205 Ave Dr. Penfield, Montreal, PQH3A 1B1, Canada. Tel: (514) 398-6726, Fax: (514) 398-5069, E-mail: [email protected]

Climatic Change 67: 273–290, 2004.C© 2004 Kluwer Academic Publishers. Printed in the Netherlands.

Page 2

274 PETRA TSCHAKERT

assumed that in semi-arid regions 0.05–0.3 t C ha–1 yr–1 can be sequestered on crop-lands and 0.05–0.1 t C ha–1 yr–1 on grasslands and pastures (Lal, 1999). Improvedmanagement practices such as reduced tillage, manure application, mulching, com-posting, fallowing, crop rotations, and agro-forestry (Lal et al., 1999; Bruce et al.,1998) as well as changes in land use, including the conversion of degraded crop-lands to grasslands or pasture, will not only increase the rate of CO2 uptake fromthe atmosphere but also contribute to erosion control and enriched biodiversity.Since soil organic matter is usually lower in areas where degradation is severe, thepotential to increase the soil C through land rehabilitation is high. Thus, soil carbonsequestration could provide a crucial link between three international conventions:the UN Framework Convention on Climate Change (UNFCC), the UN Conven-tion to Combat Desertification (UNCCD), and the UN Convention on Biodiversity(UNCBD).

Promoting degraded agro-ecosystems as potential carbon sinks has several ad-vantages. Especially in contrast to forest sinks, carbon sequestration in degradedagro-ecosystems are more likely to secure carbon storage in the form of SOM in thelong run. This is due to longer residence time of carbon in soils and the potential in-terest of local farmers in its preservation, since their livelihoods most often directlydepend on it (Olsson and Ardo, 2002). Also, such carbon sinks are anticipated toprovide direct economic, environmental, and social benefits for local populations.In areas where most smallholders rely, at least to a certain extent, on subsistenceagriculture, increased soil fertility and crop yields, improved water-holding ca-pacity in soils, less animal pressure on crop and grazing lands, and enhanced foodsecurity could all contribute directly to enhanced well being. Lastly, promoting car-bon sinks in degraded African drylands, either through the Kyoto Protocol or otherinternational agreements, provides a promising avenue for addressing north-southequity issues combined with necessary support for the rural poor.

Despite these seeming advantages, controversy surrounding the notion of carbonsequestration in soil remains. Net effects on the local and regional carbon balance asa result of nitrogen (N) fertilizer application and the use of manure are still disputed(Schlesinger, 2000; Izaurralde et al., 2000). Many development practitioners remainskeptical, arguing that carbon brokers, national ministries, and local leaders ratherthan needy rural populations will benefit from carbon projects. Most importantly,soil carbon sequestration will not be eligible during the first commitment period ofthe Kyoto Protocol, although political pressure has been growing (Ringius, 2002).

This paper aims to address some of the current gaps and limitations in soil carbonsequestration research in dryland environments. It evaluates both the biophysicaland socio-economic opportunities and constraints for soil carbon sequestration inSenegal’s semi-arid regions. It attempts to shed light on the current C status, thepotential for C gains through improved management practices and land use changes,and anticipated costs and benefits related to a carbon offset scheme in cooperationwith smallholders.

Page 3

CARBON FOR FARMERS 275

2. A Case Study in Senegal: The SOCSOM Project

In order to assess the potential for soil carbon sequestration in semi-arid envi-ronments in the Sahel, a two-year pilot project was conducted in Senegal. TheSOCSOM project (Sequestration of Carbon in Soil Organic Matter), funded byUSAID and implemented as a collaborative research activity between EROS DataCenter/USGS, Colorado State University, University of Arizona, Centre de SuiviEcologique (CSE) and Institut Senegalais de Recherches Agricoles (ISRA), hasfour major components:

1. Assessment of current carbon status through soil and biomass carbon mea-surements;

2. Evaluation of options for carbon sequestration through simulation of man-agement practices and impacts on carbon pools using CENTURY, a biogeo-chemical model;

3. Quantification of costs and benefits of improved management practices;4. Assessment of the institutional and policy requirements necessary to design

and implement a future carbon sequestration program in Senegal.

2.1. SITE DESCRIPTION

The SOCSOM study was conducted in three distinct agro-ecological zones(Figure 1), reflecting the north-south stratification in terms of precipitation, soiltypes, vegetation, and ethnic groups. One zone, the ‘Old Peanut Basin’ (West–Central Agricultural Region), was to receive special emphasis and, thus, will bedescribed in more detail in this paper.

The Old Peanut Basin is characterized by soils with large portions of aeo-lian material. The dominant soil types are luvic Arenosols, ferric Luvisols, andchromic Vertisols (FAO, 1974), corresponding to Lamellic Ustipsamments, Ul-tic Haplustalfs, and Chromic Haplusterts, respectively, following the USDA SoilTaxonomy. Farmers distinguish mainly between “dior” and “deck”. The first arecommon on former dune slopes and usually contain >95% sand and <0.2% organiccarbon. The second are hydromorphic with 85–90% sand and slightly more organiccarbon (Badiane et al., 2000).

The climate is semi-arid with annual precipitation ranging from 350–700 mm,making this part of the country barely suitable for rain-fed agriculture with mil-let (Pennisetum typhoıdes), groundnuts (Arachis hypogaea), sorghum (Sorghumbicolor), and cowpeas (Vigna unguiculata). The rainy season lasts from July toSeptember or October, although both spatial and temporal variation of rainfall arehigh and episodic crop failures well known. Over 90% of arable lands are used forcultivation. The main woody species are Faidherbia albida (White Thorn), Acacianilotica (Black Thorn), Acacia raddiana (Twisted Acacia), Acacia seyal (Gum

Page 4

276 PETRA TSCHAKERT

Figure 1. Major agro-ecological zones in Senegal with the three SOCSOM study areas: (I) theNorthern Pastoral Sandy Region and the Senegal River Valley around Podor; (II) the West–CentralAgricultural Region (‘Old Peanut Basin’) around Bambey; and III) the Sudan–Guinean TransitionZone around Velingara. Map source: EROS Data Center/USGS.

Page 5

CARBON FOR FARMERS 277

Thala Tree), Adansonia digitata (Baobab), Balanites Aegyptica (Desert date), andGuiera senegalensis (Guiera).

The Old Peanut Basin is known for its high population densities, ranging from150 to 225 inhabitants per km2 (Centre du Suivi Ecologique, 2000). Eighty eightpercent of the population identify themselves as members of one of two ethnicgroups, the Wolof and Serer (Centre de Suivi Ecologique, 2000). The variabil-ity of millet and sorghum production is high, making smallholders vulnerable tofluctuations in precipitation.

3. Current C Status

3.1. SOIL AND BIOMASS C MEASUREMENTS

3.1.1. MethodologyBiomass and soil carbon measurements were taken in three villages within the studyarea (Thilla Ounte, Ngodjileme, and Thiaytou), in collaboration with teams of farm-ers trained on the spot. Seven sample fields per village were selected based on alottery, following a stratified random sampling. Trees were counted and measured(diameter at breast height) on a 0.25-ha sampling area within each field. Under-storey, litter, and root samples were taken respectively from 1 m2, 0.25 m2, and 0.04m2 replicate subquadrats, all within the main quadrat, with the origin and directionestablished at random. Soil samples were collected at 0–20 cm and 20–40 cm andbulk density at 10 cm and 30 cm depth within the same replicate subquadrats. Allsamples were analyzed at the ISRA/CNRA research station in Bambey. Biomasssamples were oven-dried at 70 ◦C to constant weight for dry matter. Soil sampleswere analyzed following Black & Walkley for C (Walkley, 1947), Kjeldahl for N(Bremner, 1996), Olson modified by Dabin (1967) for soil available P, and Robinsonfor granulometric analysis of clay and silt (Day, 1965). Bulk density was measuredusing volume infill procedures.

Data was compiled in an Excel spreadsheet and standardized on a hectare basis.Following Woomer et al. (in press), the proportion of C in all biomass pools wasassumed to be 0.47. An allometric equation, proposed by FAO (1997) for drywoodlands was used to calculate the tree biomass.

3.1.2. ResultsThe results from the soil and biomass carbon analysis showed significant variationbetween sampling sites, both for total system carbon and distinct carbon pools(Table I). Total system carbon varied between 12.7 t ha–1 and 59.3 t ha–1 (mean27.8, SD ± 12). This variation is most likely due to high clay and silt contentsobserved on one site (#9) and considerable differences in presence and densities offield trees.

Page 6

278 PETRA TSCHAKERTTA

BL

EI

Bio

mas

san

dso

ilca

rbon

for

21si

tes

(fiel

ds)

inth

eW

est–

Cen

tral

Agr

icul

tura

lReg

ion,

2001

Soil

CSo

ilC

Tota

lsoi

lCTo

tal

long

DD

latD

DC

lay

+si

lt%C

lay

+si

lt%T

ree

CH

erb.

CR

ootC

Litt

erC

0–20

cm20

–40

cm0–

40cm

syst

emC

Bul

kde

nsity

Bul

kde

nsity

Site

Lan

dus

eV

illag

e—

-deg

rees

—–

0–20

cm20

–40

cm—

——

——

——

——

——

——

-ca

rbon

(kg

ha−1

)—

——

——

——

——

——

—–

0–20

cm20

–40

cm

1C

1−1

6.68

714

.797

2.7

4.6

078

799

734

10,8

655,

988

16,8

5318

,464

1.64

1.57

2C

1−1

6.68

514

.801

4.2

5.5

553

1,01

862

93,

002

13,5

618,

399

21,9

5927

,162

1.63

1.56

3C

1−1

6.67

814

.801

5.8

6.9

14,4

9733

335

32,

580

18,6

286,

549

25,1

7742

,940

1.61

1.54

4C

1−1

6.68

414

.791

2.9

3.5

8,00

816

717

61,

290

9,31

43,

275

12,5

8922

,230

1.7

1.62

5C

1−1

6.68

214

.805

5.1

5.9

17,2

8526

166

41,

571

13,4

217,

508

20,9

3040

,710

1.67

1.60

6C

2−1

6.45

614

.625

4.1

6.1

2,34

856

270

531

10,1

746,

157

16,3

3119

,536

1.79

1.79

7C

2−1

6.45

514

.623

4.7

6.8

1,32

311

658

21,

907

9,97

46,

421

16,3

9620

,324

1.65

1.62

8C

2−1

6.45

914

.615

3.8

6.6

16,5

0717

978

72,

700

10,3

815,

846

16,2

2736

,400

1.69

1.58

9C

2−1

6.44

514

.595

20.2

28.4

6,46

220

81,

058

446

21,3

4929

,822

51,1

7159

,344

1.72

1.57

10C

2−1

6.45

514

.614

4.8

7.3

5,95

332

267

64,

771

7,60

35,

152

12,7

5524

,476

1.64

1.66

11C

2−1

6.45

714

.629

7.8

13.4

308

134

922

780

6,19

94,

400

10,5

9912

,744

1.65

1.57

12C

3−1

6.41

014

.935

3.9

4.9

5,34

757

523

975

17,0

776,

102

23,1

7930

,080

1.59

1.52

13C

3−1

6.40

914

.931

3.7

4.8

19,2

6913

61,

721

1,36

17,

896

4,27

212

,168

34,6

551.

571.

50

14C

3−1

6.40

214

.931

3.6

4.3

6,27

52

858

870

6,61

44,

356

10,9

6918

,974

1.62

1.63

15C

3−1

6.39

914

.923

5.2

7.2

1,94

468

42,

356

2,10

94,

367

2,81

17,

178

14,2

711.

631.

51

16C

3−1

6.41

514

.916

6.8

8.7

675

386

3,12

014

,223

16,5

044,

725

21,2

2939

,633

1.59

1.56

17F

1−1

6.68

514

.793

4.7

5.4

5353

342

31,

810

10,9

907,

922

18,9

1221

,730

1.67

1.60

18F

3−1

6.40

914

.913

5.7

7.3

481

174

494

1,69

66,

451

3,63

310

,084

12,9

281.

581.

57

19F

3−1

6.40

714

.946

5.0

6.7

1,19

520

41,

240

2,83

06,

529

5,53

712

,067

17,5

351.

601.

54

20P

1−1

6.68

714

.799

5.5

6.6

13,2

8440

579

91,

673

15,8

877,

265

23,1

5239

,313

1.67

1.60

21P

2−1

6.45

414

.617

5.9

6.9

10,1

2247

235

35,

572

8,30

05,

192

13,4

9330

,011

1.67

1.67

Mea

n5.

57.

56,

280

282

895

2,54

411

,052

6,73

017

,782

27,7

841.

641.

59SD

3.6

5.8

6,55

524

271

52,

979

4,63

25,

499

9,16

512

,040

0.05

0.06

Sour

ce:S

oila

ndbi

omas

sC

mea

sure

men

ts(2

001)

.L

and

use:

C=

culti

vate

dfie

lds,

F=

field

sin

fallo

w,P

=pa

rkla

nds

(“ch

amps

arbo

res

”).

Vill

age:

1=

Thi

llaO

unte

,2=

Ngo

djil

em

e,3

=T

hiay

tou

Page 7

CARBON FOR FARMERS 279

On average, combined soil and tree C accounted for 86% of total system carbon(SD ± 11). Soil C in the upper 20 cm amounted to 63% of total measured C while37% were found in the 20–40 cm horizon. The mean value of 11 t ha–1 for soil C inthe upper level was later used as a base comparison for simulations with CENTURY.Nevertheless, variation between sites was large. In addition to the impact of highclay and silt contents on soil carbon (Pieri, 1992), field histories recorded duringmeasurements show that sites with mean values of 4–8 t C ha–1 were relativelypoorly managed. Values >14 t ha–1 can be largely correlated with major organicmatter inputs. Carbon in herbaceous biomass, roots and litter accounted for 1, 3.2,and 9.1% of total system carbon, respectively.

Grouping the data by land-use type confirms Ringius’ assertion (2002) that shortfallow periods in degraded savannas of West Africa are insufficient to restore soilorganic matter. Fallow fields in this study showed a total of 12.9–21.7 t C ha–1, withvalues for both tree and soil C lower than that on cultivated fields. This is mainlydue to the fact that fallow lands are used as open-access grazing grounds withslow regrowth. New parklands (fields with trees planted during the last decade)yielded 30–39 t C ha–1, the highest total system C values. Average amounts incultivated lands were slightly lower (29.5 t C ha–1). However, the mean obscuresthe differences between poorly and well-managed fields.

Total system carbon values for the Old Peanut Basin ranked between those ofadjacent SOCSOM sites to the north and the south, with highest combined treeand soil C (48 t C ha–1) in the Casamance (Woomer et al., in press). Soil C valuesfor cultivated and new parklands in the center matched those of woodlands farthersouth, indicating fairly poor protection of the latter against cutting, browsing andother sources of degradation.

3.2. BIOGEOCHEMICAL SIMULATIONS USING CENTURY

3.2.1. MethodologyIn order to estimate past and future carbon levels as well as to confirm currentC contents as obtained through ground measurements, a biogeochemical model,CENTURY, was used (Parton et al., 1994). CENTURY is an ecosystem model thatsimulates fluxes of carbon (C), nitrogen (N), phosphorus (K), as well as sulphur(S). Here, the model’s prime function was to evaluate the impact of a series ofmanagement practices on soil C in the upper 20-cm horizon (g m–2) and millet andgroundnuts crop yields (g m–2).

Charcoal CENTURY was used instead of the standard monthly CENTURY tobetter simulate stable soil C in highly sandy soils, following Skjemstad et al. (1996)who found that the majority of protected soil C, analogous to the passive pool inCENTURY, can be in the form of charcoal in systems that are or were frequentlyburned. Due to the absence of pristine sites, historic C levels were obtained byreconstructing land use and management changes, using remotely sensed data,expert estimates, field interviews, and, for periods prior to 1945, approximations

Page 8

280 PETRA TSCHAKERT

Figure 2. A CENTURY model simulation for tree and soil C in the Old Peanut Basin, with undisturbedsavanna grasslands (1800–1850), past cultivation (1851–2001), and future management practices(2002–2050): 1 = agricultural intensification; 2 = plantation with Faidherbia albida; 3 = millet–sorghum rotation.

based on similar and undisturbed environments (Appendix). The model was runto equilibrium for 1850 years before introducing the first scenario. Input data forthe model were obtained from field measurements, the literature, and the DirectionNationale de la Meteorologie in Dakar. Real temperature and precipitation datawere used for the period 1960–2001. The same climate data were used for futurescenarios while historic periods prior to 1960 were based on long-term means.

3.2.2. Simulation Results for Past and Current C LevelsCENTURY results suggest an initial total system carbon content in the pre-cultivation savanna of 60 t ha–1, with mainly tree C (34.7 t C ha–1) and soil C(20.1 t C ha–1). With the clearing of land for agriculture, tree C decreases rapidlyand soil C more slowly (Figure 2). By 1900, tree C has decreased by 65% (0.43 tC ha–1 yr–1) and soil C by 21% (0.083 t C ha–1 yr–1). Both parameters continue todecrease until the present, due to a combination of agricultural expansion, biomassremoval, pruning, browsing, episodic droughts, periodic fires, and insufficient or-ganic matter inputs. A short-term increase in tree and soil C from the 1950s tothe mid 1970s can be related to years of favorable precipitation and intensifiedagriculture due to subsidies through the state’s agricultural policies.

Current (2001) C values simulated by CENTURY correlate well with the resultsfrom the ground measurements. Modeled soil C for the upper horizon amounted to11.9 t ha–1 compared to 11 t ha–1 observed on the sites. These values match thosecited in the literature (Ringius, 2002). Simulated tree C amounted to 4.2 t ha–1,which is comparable to a mean of 4.3 t C ha–1 measured on the ground. Overall,the model suggests that total system C has decreased by 71% (42.7 t ha–1) from1850 to the present day. Losses of soil C amounted to 8.2 t ha–1 (–41%) and thosefor tree C to 29.4 t ha–1 (–87%).

Page 9

CARBON FOR FARMERS 281

In terms of crop yields, CENTURY simulated millet yields at an average of653 kg/ha for the period 1980–2001. This is higher than the mean reported bythe Direction de l’Agriculture for the same time period (527 kg/ha). Simulatedgroundnut yields were 707 kg/ha, which is only slightly higher than the officialmean (665 kg/ha).

4. Biophysical Potential for C Sequestration in Soils

A total of 25 management scenarios were used in CENTURY to simulate thebiophysical potential for carbon sequestration for Old Peanut Basin. Input variablesfor all scenarios were defined on the basis of field measurements and values fromthe literature.

4.1. SIMULATION OF SOIL CARBON SEQUESTRATION POTENTIAL

The model results suggest that, after 50 years, total system C could be increasedfrom current 17.3 t ha–1 to 32.2 t ha–1 under an optimum intensification scenarioand 40.8 t ha–1 if all agricultural land was converted into a grassland plantationwith Faidherbia albida (kad), a highly valued nitrogen-fixing tree (Figure 2). As-suming a worst case scenario with an annual millet–sorghum rotation, no inputs,and continuous browsing and pruning, soil C would decline to 7.9 t ha–1 (–33%)and tree C to a minimum of 0.6 t C ha–1 (–86%).

Next, all 25 scenarios were compared to the 2001 baseline value (11.9 t C ha–1).As illustrated in Table II, 14 practices are expected to result in increases in soil Cover the next 25–50 years, ranging from roughly 0.3–13.5 t ha–1. These practicesinclude the application of cattle and sheep manure (4–10 t ha–1) with or withoutfertilizer, 3–10-year fallow periods with organic matter input in rotation with 4–6-year cropping cycles, plantations with Faidherbia albida (kad), and the optimumintensification scenario.

With exception of the tree-planting scenario, the majority of C gains are achievedin the first 25 years. During this period, simulated soil C increases ranged from 0.02to 0.43 t C ha–1 yr–1, which is higher than estimated by Lal (1999). However, underthe fallow options, gains will be outweighed by losses from cropping years in thelong run. Annual increases in soil C due to the conversion of croplands into grass-lands were less significant (0.06–0.1 t C ha–1 yr–1 during 25 years for grasslandswith and without grazing and grasslands with protected Faidherbia albida). Thiscorresponds well with Lal’s estimates. Thus, refraining from cultivation alone isnot sufficient to restore soil C levels.

On the negative side, the model predicts further decline of soil C if continu-ous cropping was practiced without external inputs. Nutrient mining and C lossesare expected to be slightly greater and to occur faster without groundnuts in the

Page 10

282 PETRA TSCHAKERT

TABLE IIImpact of management practices on soil C (0–20 cm horizon) compared to 2001 C status in theWest–Central Agricultural Region, simulated with CENTURY for two 25-year periods (2002–2026and 2027–2050)

Changes in soil Changes in soilC (t ha−1) C (t ha−1)First 25 years Second 25 years

Management practices (2002–2026) (2027–2050)

Rotation millet–sorghum (1:1)∗, no inputs −3.18 −0.74

Rotation millet–groundnuts (4:1), no inputs −3.12 −0.74

Rotation millet–groundnuts (2:1), no inputs −2.99 −0.78

Rotation millet–groundnuts (1:1), no inputs −1.62 −0.42

Stubble grazing (20 cows) on millet–groundnuts rotation −1.24 −0.42

Rotation crops–fallow + 30cows summer grazing (4:1) −1.00 −0.65

Rotation crops–fallow + mixed animals (2:2) −0.95 −0.63

Rotation crops–fallow + 80sheep summer grazing (4:3) −0.87 −0.58

Rotaion crops–fallow (4:3) with millet and groundnuts −0.37 −1.00

Horse manure 1.5 t on millet, no imputs on groundnuts −0.77 −0.38

Rotation millet–groundnuts (1:1), protection of kad ∗∗ −0.64 −0.30

Compost 2 t on millet, no input on groundnuts 0.51 −0.17

Conversion of cropland to grassland with summer grazing 1.46 0.60

Rotation crops–fallow + 2 t manure (4:3) 3.43 −1.23

Cow manure 4 t on millet, no inputs on groundnuts 2.42 0.20

Conversion of cropland to grassland 1.87 0.94

Cow manure 4 t on millet, 150 kg fertilizer on groundnuts 2.99 0.18

Sheep manure 5 t on millet, no inputs on groundnuts 3.19 0.20

Rotation crops–fallow + leucaena prunings (4:3) 4.56 −1.15

Conversion of cropland to grassland+protection kad 2.48 0.94

Sheep manure 10 t on millet, no inputs on groundnuts 4.25 0.37

Rotation crops–fallow + 2 t manure (6:10) 6.17 −0.92

Rotation crops–fallow + 2 t leucaena prunings (6:10) 6.35 −0.95

Kad plantation (250-300 trees) 5.81 5.30

Optimum agricultural intensification: crops–fallow (2:1), 10.83 2.70150 kg fertilizer on groundnuts, 4 t manure on fallow, 5tsheep manure + 2 t leucaena prunings on millet

Source: CENTURY simulations (2002).∗Years of rotational cycle in parenthesis∗∗Kad = Faidherbia albida.

Page 11

CARBON FOR FARMERS 283

Figure 3. CENTURY simulation of historic average crop yields (1980–2000) compared to potentialfuture crop yields (2002–2026) expected as result of selected management practices; all values inkg/ha.

rotational cropping cycle (3.2 t C ha–1 compared to 3 t C ha–1with groundnuts).Scenarios based on rotational cycles of fallow-crops and grazed fallow-crops aswell as stubble grazing during the dry season also revealed losses during the first25 years, although not as severe as under continuous cultivation. Given farmers’preferences for maintaining soil fertility through grazing–cropping schemes, thisparticular simulation result was rather surprising.

4.2. SIMULATED CHANGES IN CROP YIELDS

CENTURY simulated changes in crop yields due to improved management prac-tices, ranging from –62% to +200% for millet and –45% to +133% for groundnutsover a period of 25 years, under worst and best management scenarios respectively.These values correlate well with changes predicted for soil carbon. A comparison ofsimulated historic (1980–2001) crop yields with values from various managementpractices (2002–2026) indicates that the maximum average from the optimum in-tensification scenario was 1.650 kg/ha for groundnuts and 1.960 kg/ha for millet(Figure 3). Under the worst-case scenario, the model suggests a drop in averagemillet yields to a low of 295 kg/ha. This amount would undoubtedly be insufficientto satisfy basic household food needs.

5. Economic Considerations at the Household Level

5.1. METHODOLOGY

Whether or not smallholders have the financial means to implement improved car-bon management practices will depend to a large extent on their economic situation.Here, an ‘improved’ practice is defined as yielding at least 1.5 t C ha–1 over a period

Page 12

284 PETRA TSCHAKERT

of 25 years, which corresponds to the minimum sequestration rate as estimated byLal (1999). The economic analysis for the Old Peanut Basin is based on two parts:(1) a farmer-centered cost-benefit analysis, described in detail in Tschakert (2004);and (2) an assessment of potential cash payments through C trading compared to ac-tual household budgets. Both rely on three distinct resource endowment categories(poor, medium, and rich), as defined through participatory wealth ranking, to ac-count for diverse household resources and differential economic potential amongthe sample population. For instance, annual revenues in 2001 ranged from $30 to$395 per adult equivalent (Tschakert, in press). Such large variation suggests thatit would be oversimplified and highly misleading to assume an ‘average’ farmerfor the purpose of a carbon cost-benefit analysis for drylands. The unit of analysisassumed for both parts was one hectare and the time frame 25 years.

5.2. RESULTS

5.2.1. Local Costs and BenefitsInitial (year 1) investment costs for all tested carbon management practices rangedfrom $0–54 (conversion of cropland to grassland for poor and rich households,respectively) to roughly $3,000 for the 10-year fallow scenario with organic matterinput and the fattening of three cows as an income-generating activity. The lowestundiscounted net costs were calculated for the 10-year fallow option without animalfattening ($104–315) and the highest for the conversion of cropland to grasslandscenario ($1,100–1,400).

On the benefit side, certain practices were assumed to produce no financialgains during the first year while others, mainly those including the sale of animalsand animal products, generated more than $2,000 per hectare during the sametime. Undiscounted net benefits proved to be highest under the long-term fallowscenario with animal fattening ($3,000), suggesting that carbon offset practicescombined with income-generating activities might be the most beneficial option.Practices with high initial investment costs tended to result in negative net presentvalues (NPV) at the end of the investment cycle, meaning that the practice wasnot profitable. Overall, only one practice (converting cropland to grassland withAcacia leatea hedges and highly valuable seeds) proved profitable for poor farmers,while half of all the tested options showed positive NPV for the better-endowedhouseholds.

To date, few in depth cost estimates exist for C sequestration activities on Africandrylands. Those that do exist (Squires et al., 1997; Ringius, 2002) provide a singlenumber cost estimate for land restoration, even though, as shown here, variationsbetween proposed practices and resource endowment groups can be fairly signif-icant. As suggested by Ringius (2002), there is an urgent need for more detailedfinancial assessments, also because earlier studies have most likely underestimatedoverall costs.

Page 13

CARBON FOR FARMERS 285

5.2.2. Benefits From Carbon TradingIn addition to benefits anticipated to occur directly at the local level, either throughsales of animals, animal and wood products, or increased crop yields, potentialcash payments from international carbon trading were calculated for all ‘best’management practices Based on Robert (2001), an average price of $15 for 1 t Csequestered was assumed.

Economic gains per hectare, shown in Table III, ranged from $28 to $162 (<$2to <$7 per year). For reasons of comparison, yearly gains ($) from potential Ctrading were weighed against household income from crops (%). These values,ranging from 1–4.5% for the three resource endowment groups, are extremely lowand, thus, not likely to constitute a significant financial incentive for the adoptionof ‘best’ practices. This is particularly true for poor households since they are alsothe ones who command very scarce resources, face high risks, and thus are lessflexible when it comes to innovations. The same analysis performed with $25 for1 t C sequestered resulted in only slightly higher annual cash benefits ($2–11) or1.6–7.2% of farmers’ annual income from crops.

Finally, a household cash-flow analysis, performed with STELLA and describedin detail in Tschakert (2004), illustrated that in all simulated cases, except forone, farmers would not have the extra investment capital necessary to cover first-year costs of proposed management practices. A simple comparison of householdbudgets, as recorded from field surveys, and initial investment costs underlines thedilemma. In 2001, poor and rich households counted a mean income of $577 and$2,293, respectively. In most cases, annual expenditures were at least as high. Giventhe initial investment costs of $0–3,000 (mean of $ 682) for C management andthe poor economic status of most farmers, it can be concluded that ample financialsupport would be required to allow these smallholders to participate in any C offsetprogram.

6. Discussion

The results from the soil and biomass measurements and the simulations performedwith CENTURY for the Old Peanut Basin constitute a first step in a total carbonassessment for Senegal. To date, they are integrated into a national database that,as it is anticipated, will allow the Senegalese experts to better determine their ownnational, regional, and local C offset options to be discussed during internationalclimate mitigation negotiations.

Nevertheless, four major limitations to the biophysical components of this studywarrant consideration. First, relatively pristine and undisturbed control sites, re-quired to accurately estimate historic soil and biomass carbon, no longer exist inthe Old Peanut Basin. Thus, maximum historic C levels had to be approximatedon the basis of sites farther south. Second, so far, all CENTURY simulations werebased on real-time climate data. It certainly would be of interest to also include moreextreme climate change scenarios in order to explore the impact on the overall C

Page 14

286 PETRA TSCHAKERT

TAB

LE

III

Ant

icip

ated

econ

omic

bene

fits

from

Ctr

adin

gon

ly(a

llva

lues

in$

and

for

one

hect

are)

Cga

ins

Eco

nom

icga

inE

cono

mic

%of

inco

me

%of

inco

me

%of

inco

me

afte

raf

ter

gain

afr

omcr

ops

from

crop

sfr

omcr

ops

Prac

tices

25yr

s(t

ha−1

)25

yrs

($ha

−1)a

($ha

−1yr

−1)

inpo

orH

Hsb

inav

erag

eH

Hsb

inri

chH

Hsb

Gra

ssla

nd1.

8728

.05

1.12

1.62

0.51

0.39

Cow

man

ure

4t

2.42

36.2

41.

452.

300.

660.

50

Gra

ssla

nd+

prot

ectio

nka

d2.

4837

.14

1.49

2.36

0.67

0.51

Cow

man

ure

4t+

fert

ilize

r2.

9944

.91

1.80

2.85

0.81

0.62

Shee

pm

anur

e5

t3.

1947

.88

1.92

3.04

0.87

0.66

Rot

atio

ncr

ops–

fallo

w+

man

ure

(4:3

)3.

4351

.51

2.06

3.27

0.93

0.71

Shee

pm

anur

e10

t4.

2563

.74

2.55

4.05

1.15

0.88

Rot

atio

ncr

ops–

fallo

w+

leuc

aena

prun

ings

(4:3

)4.

5668

.23

2.73

4.33

1.23

0.94

Kad

plan

tatio

n(2

50–3

00tr

ees)

5.81

87.1

63.

495.

531.

581.

21

Rot

atio

ncr

ops–

fallo

w+

man

ure

(6:1

0)6.

1792

.16

3.69

5.85

1.67

1.28

Rot

atio

ncr

ops–

fallo

w+

leuc

aena

prun

ings

(6:1

0)6.

3595

.18

3.81

6.04

1.72

1.32

Opt

imum

agri

cultu

rali

nten

sific

atio

n10

.83

162.

416.

5010

.31

2.94

2.25

Mea

n4.

5367

.88

2.72

4.30

1.23

0.94

Sour

ce:C

EN

TU

RY

sim

ulat

ions

,cos

t-be

nefit

anal

ysis

,and

field

wor

k(2

001)

.a A

ssum

ing

1tC

=$1

5.bA

vera

gein

com

efr

omcr

ops/

year

:$63

(poo

rho

useh

olds

),$2

21(a

vera

geho

useh

olds

),$2

89(r

iche

rho

useh

olds

).

Page 15

CARBON FOR FARMERS 287

sequestration potential for the study area. Third, the role of manure for the local andregional carbon balance remains ambiguous. Since increases of organic carbon infields through the application of manure are likely to involve losses on rangelands,the net affect of manure is difficult to assess. Due to this ambiguity, practices involv-ing manure might be excluded from eligible sink options. Fourth, the CENTURYmodel has been tested extensively in cropped soils. In the Sahel, however, only fewstudies so far have used CENTURY or other models and performance verificationremains difficult. Olsson and Ardo (2002) report underestimated soil carbon valuesfor Sudan. In this study, crop yields were slightly underestimated compared to theexisting agricultural data. Also, spatial aspects remain difficult to capture.

As for the economic assessment, the SOCSOM project has identified two ma-jor limitations. First, for the purpose of analysis, the same one hectare of land isassumed to undergo the same improved management practice or land-use changefor the duration of the simulation. From the perspective of carbon buyers, this isthe easiest approach, not only for evaluating potential carbon gains, costs, and ben-efits, but also for the purpose of measuring and monitoring carbon recoveries andremunerating the participating stakeholders. From the perspective of smallholders,however, this viewpoint is flawed. Due to the very “local, complex, diverse, dynamicand unpredictable” realities of poor people’s lives (Chambers, 1997), farmers aremore concerned with adapting to their constantly changing environment, seizingopportunities and evading hazards when they emerge, rather than to strive for a newequilibrium as assumed by the model. Decisions regarding land use and manage-ment are characterized by high spatial and temporal variability and determined by avariety of factors. These include the availability of and access to land, labor, seeds,cash, organic matter, social networks, means of transportation, and side benefits ofmanagement practices as well as credits and land tenure arrangements.

Hence, favoring a prescribed package of ‘best’ management practices that scorehighest on sequestration rates, C storage, crop yields, and economic gains, es-pecially on a large scale, might in fact undermine farmers’ dynamic and diverseadaptation mechanisms and, thus, increase rather than reduce their vulnerabilityto risk. Although flexible sequestration packages are likely to be less appealing toforeign investors, they satisfy local people’s needs by strengthening their responsesto shocks and uncertainty.

Second, economic profitability is only one, although a highly critical criterionthose farmers consider when adopting new practices. Other factors, such as com-patibility with overall livelihood strategies, social learning, and state policies mightbe equally important. Carbon brokers and buyers will seek a combination of highestC gains and lowest costs when evaluating potential C projects. Generally, costs willbe lower when working with high resource endowment farmers who already ownthe inputs necessary to implement an improved practice and, hence, face the lowestrisks of adoption. In the Old Peanut Basin, these farmers are rare. Those groups andindividuals who most need the anticipated benefits are least likely to adopt, simplybecause they face much higher risks. For this reason, poorer farmers are also less

Page 16

288 PETRA TSCHAKERT

attractive to potential buyers. If improving poor people’s livelihoods is indeed cen-tral to carbon sinks development in African drylands, this “innovativeness-needsparadox” (Rogers, 1995) will have to be taken into account and special incentivesbe found to also include the economically weakest groups.

Finally, it should be stressed again that carbon sequestration in agricultural soilsis currently not eligible for project activities under the Kyoto protocol. Thus, poten-tial cash payments through carbon trading, as calculated here, remain speculative.Nevertheless, they clearly indicate that direct economic benefits for smallhold-ers in drylands are likely to be limited and that more emphasis should be put onother economic and livelihood incentives for promoting carbon management. Onlythrough a realistic and holistic appraisal of all possibilities and constraints can thecarbon sequestration programs in African drylands be appropriately designed and,eventually, be successful and sustainable.

Appendix: Historic Management Scenarios (0–2001) as Specified inCENTURY

Period Description

0–1850 Initialization; Cenchrus as native grass and a West African, drought-adapted,deciduous Acacia as major tree; summer grazing of grassland at moderate levels;70% of grassland burnt every 5th year and goats browsing shrubs also every 5thyear;

1851–1900 Introduction of agriculture; 7-year rotational cycle with 1 year M, 1 year S and 5years of fallow; winter grazing after cropping years; 70% of trees removed due tofire during cropping years; browsing by goats; M straw mulching on S and M;Faidherbia albida (“ kad ”) as dominant tree from this period on (reverse growingcycle, nitrogen-fixing);

1901–1945 Introduction of GN; 7-year rotational cycle with 1 year M, 1 year GN, 1 year S,and 4 years of fallow; tree burning and shrub browsing; winter grazing after Mand S years; 2 t of GN shells as organic matter input on M; tree pruning forpreparing savanna for cultivation;

1946–1965 Five-year rotational cycle with 1 year M, 1 year GN, 1 year S, and 2 years offallow; pruning and shrub browsing; 2 t of manure on M, M straw mulchingbefore GN and S, and 2 t of GN shells also before S;

1966–1981 Agricultural intensification through heavily state-supported “ProgrammeAgricole”; 2-year rotational cycle with 1 year GN and 1 year improved M; nofallow; animal traction instead of hand weeding; fertilizer on both crops;

1982–1997 Seven-year rotational cycle reflecting the disengagement of the state under the“Nouvelle Politique Agricole” with 1 year of “forced” fallow (lack of seeds), 1year GN, 2 years M, 1 year GN, and 2 years M; fertilizer on GN only one out oftwo cropping seasons, 2 t of manure and a combination of 1 t manure + 1 thousehold waste alternating on M three out of 4 cropping seasons; browsing bygoats and tree pruning continues;

1998–2001 Two-year rotational cycle with 1 year GN and 1 year M; no fallow; 0.5 t manure+ 0.5 t household waste on M; no inputs on GN.

M = millet, GN = groundnuts, S = sorghum.

Page 17

CARBON FOR FARMERS 289

Acknowledgements

Mamadou Khouma and Modou Sene of ISRA, Senegal, provided assistance in soiland biomass C measurements and the analysis of laboratory results. Bill Partonand Steve DelGrosso, Colorado State University, are thanked for their help with theCENTURY simulations. Larry Tieszen of the EROS Data Center, Sioux Falls,Chuck Hutchinson, University of Arizona, Tucson, and Lennart Olsson, LundUniversity, Sweden provided encouragement and means to conduct this study.

References

Badiane, A. N., Khouma, M., and Sene, M.: 2000, Region de Diourbel: Gestion des sols, DrylandsResearch Working Paper 15, Drylands Research, Somerset, England, p. 25.

Batjes, N. H.: 2001, ‘Options for increasing carbon sequestration in West African soils: an exploratorystudy with special focus on Senegal’, Land Degrad. Dev. 12, 131–142.

Bremner, J. M.: 1996, ‘Nitrogen-Total’, in Sparks, D. L. (ed.), Methods of Soil Analysis, Part3. Chemical Methods, American Society of Agronomy, Madison, Wisconsin, pp. 1085–1121.

Bruce, J. P., Frome, M., Haites, E., Janzen, H., Lal, R., and Paustian, K.: 1998, ‘Carbon sequestrationin soils’, Soil Water Conserv. 54, 382–389.

Centre de Suivi Ecologique: 2000, Annuaire sur l’environnement et les ressources naturelles duSenegal, Centre de Suivi Ecologique, Dakar, Senegal, p. 268.

Chambers, R.: 1997, Whose Reality Counts ? Putting the First Last, Intermediate Technology Publi-cations, London, p. 249.

Dabin, B.: 1967, ‘Application des dosages automatiques a l’analyse des sols. 3eme partie’, CahiersORSTOM, Serie Pedologie 5(3), 257–286.

Day, P. R.: 1965, ‘Particle fraction and particle-size analysis’, in Black, C. A. (ed.), Methods of SoilAnalysis, Part I, American Society of Agronomy, Madison, Wisconsin, pp. 545–567.

Food and Agriculture Organization of the United Nations (FAO)—UNESCO: 1974, Soil map of theworld, 1:5000000, Vol.1, Legend, UNESCO, Paris.

Food and Agriculture Organization of the United Nations (FAO): 1997, Estimating Biomass andBiomass Change of Tropical Forests: A Primer, FAO Forestry Paper 134, FAO, Rome, p. 55.

Izaurralde, R. C., McGill, W. B., and Rosenberg, N. J.: 2000, ‘Carbon cost of applying nitrogenfertilizer’, Science 288, 811–812.

Lal, R.: 1999, ‘Soil management and restoration for C sequestration to mitigate the acceleratedgreenhouse effect’, Progress Environ. Sci. 1, 307–326.

Lal, R., Hassan, H., and Dumanski, J.: 1999, ‘Desertification control to sequester C and mitigate thegreenhouse effect’, in Rosenberg, N., Izaurralde, R. C., Malone, E. L. (eds.), Carbon Sequestrationin Soils: Science, Monitoring, and Beyond, Proceedings of the St. Michaels Workshop, December1998, Batelle Press, Columbus, Ohio, pp. 83–136.

Olsson, L. and Ardo, J.: 2002, ‘Soil carbon sequestration in degraded semiarid agro-ecosystems:Perils and potentials’, Ambio 31(6), 471–477.

Parton, W. J., Ojima, D. S., Cole, C. V., and Schimel, D. S.: 1994, ‘A general model for soil organicmatter dynamics: Sensitivity to litter chemistry, texture and management’, in Proceedings Quan-titative Modeling of Soil Forming Processes, Minneapolis, Minnesota, 2 November 1992, SoilScience Society of America, Madison, Wisconsin, pp. 147–167.

Pieri, C. J. M. G.: 1992, Fertility of Soils: A Future for Farming in the West African Savanna, SpringerVerlag, Berlin, Heidelberg, p. 348.

Page 18

290 PETRA TSCHAKERT

Ringius, L.: 2002, ‘Soil carbon sequestration and the CDM: Opportunities and challenges for Africa’,Climatic Change, 54, 471–495.

Robert, M.: 2001, Soil Carbon Sequestration for Improved Land Management, World Soil ResourcesReports 96, FAO, Rome, p. 62.

Rogers, E. M.: 1995, Diffusion of Innovations, 4th edition, The Free Press, New York, p. 518.Schlesinger, W. H.: 2000, ‘Carbon sequestration in soils: Some cautions amidst optimism’, Agric.

Ecosys. Environ. 82, 121–127.Skjemstad, J. O., Clarke, P., Taylor, J. A., Oades, J. M., and McClure, S. G.: 1996, ‘The chemistry

and nature of protected carbon in soil’, Aust. J. Soil Res. 34, 251–271.Squires, V. R., Glenn, E. P., and Ayoub, A. T. (eds.): 1997, Combating Global Climate Change by

Combating Land Degradation, Nairobi, UNEP.Tiessen, H., Feller, C., Sampaio, E. V. S. B., and Garin, P.: 1998, ‘Carbon sequestration and turnover

in semiarid savannas and dry forest’, Climatic Change 40, 105–117.Tschakert, P.: (2004), ‘The costs of soil carbon sequestration: An economic analysis for small-scale

farming systems in Senegal’, Agric. Sys. 81, 227–253.Tschakert, P.: (in press b), ‘More food, less poverty? The potential role of carbon sequestration in

smallholder farming systems in Senegal’, in Lal, R., Uphoff, N., Stewart, B. A. and Hansen,D. O. (eds.), Climate Change and Global Food Security, Marcel Dekker, New York.

Walkley, A.: 1947, ‘A critical examination of a rapid method for determining organic carbon insoils: Effect of variations in digestion conditions and of inorganic soil constituents’, Soil Sci. 63,251–263.

Woomer, P. L., Palm, C. A., Qureshi, J. N., and Kotto-Same, J.: 1997, ‘Carbon sequestration andorganic resources management in African smallholder agriculture’, in Lal, R., Kimble, J. M.,Follett, R. F., and Stewart, B. A. (eds.), Management of Carbon Sequestration in Soil, CRC Press,Boca Raton, pp. 153–173.

Woomer, P. L., Tieszen L., Tappan, G., Toure, A., and Sall, M.: (in press), ‘Land use change andterrestrial carbon stocks in Senegal’, J. Arid Environ.

(Received 13 May 2003; in revised form 26 May 2004)