Citation
Carbon Sequestration Potential of Agroforestry Systems in the West African Sahel

Material Information

Title:
Carbon Sequestration Potential of Agroforestry Systems in the West African Sahel An Assessment of Biological and Socioeconomic Feasibility
Creator:
Takimoto, Asako
Place of Publication:
[Gainesville, Fla.]
Publisher:
University of Florida
Publication Date:
Language:
english
Physical Description:
1 online resource (184 p.)

Thesis/Dissertation Information

Degree:
Doctorate ( Ph.D.)
Degree Grantor:
University of Florida
Degree Disciplines:
Forest Resources and Conservation
Committee Chair:
Nair, Ramachandr P.
Committee Members:
Comerford, Nicholas B.
Schuur, Edward A.
Martin, Timothy A.
Alavalapati, Janaki R.
Nair, Vimala D.
Graduation Date:
12/14/2007

Subjects

Subjects / Keywords:
Forest Resources and Conservation -- Dissertations, Academic -- UF
bank, clean, development, fence, fodder, kyoto, live, mechanism, parkland, protocol
Live fences ( jstor )
Jury sequestration ( jstor )
Soil science ( jstor )
Genre:
Electronic Thesis or Dissertation
born-digital ( sobekcm )
Forest Resources and Conservation thesis, Ph.D.

Notes

Abstract:
In recent years, carbon (C) sequestration potential of agroforestry systems has attracted attention, especially following Kyoto Protocol's recognition of agroforestry as an option for mitigating green house gasses. Although the possible benefits of agroforestry in carbon (C) sequestration have been conceptually discussed, field measurements to validate these concepts have not been undertaken to any significant extent. In addition to the traditional agroforestry systems, improved practices and technologies are now being expanded into the dry regions such as the West African Sahel for perceived benefits such as arresting desertification, reducing water and wind erosion hazards, and improving biodiversity. Thus, it is imperative to investigate C sequestration potential of agroforestry practices in these regions. My research hypothesizes that the tree-based systems will retain more C in the systems both above- and below-ground than tree-less land-use systems. By joining the C credit market, the landowners could sell the C sequestered in their agroforestry systems. My research consisted of three components. The first examined C (biomass + soil) stored in five target land-use systems: two traditional parkland systems involving Faidherbia albida and Vitellaria paradoxa trees as the dominant species, two improved agroforestry systems (live fence and fodder bank), and land that is out of cultivation (abandoned or degraded) in the S?gou Region, Mali. The second component involved a study of soil C dynamics of these systems: the extent of soil C storage/accumulation by trees and stability of the C accumulated were investigated. In the third component, socioeconomic feasibility of the agroforestry systems was examined in the context of C sequestration and C credit sale. Research results show that the selected agroforestry systems have the potential for sequestering more C both above- and belowground than in tree-less land-use systems, and that the trees tend to contribute to storing more stable C in the soil. Among the selected land-use systems, live fence and fodder bank are more suitable to start as agroforestry C sequestration projects than the traditional parkland systems for smallholder farmers in the studied region. Between the two improved systems, live fence has higher C sequestering potential per unit area and is economically less risky than fodder banks. Adopting these systems on cultivated land rather than on abandoned land is likely to sequester more C and be more profitable. Since parklands are traditionally practiced, they are not likely to qualify as a new C sequestration project soon. Nevertheless, F. albida trees are more attractive than V. paradoxa trees in terms of C sequestration potential. These results can be used for development of recommendations and guidelines on selection of land use-systems and species and their management, for planning successful C sequestration projects in the West African Sahel. ( en )
General Note:
In the series University of Florida Digital Collections.
General Note:
Includes vita.
Bibliography:
Includes bibliographical references.
Source of Description:
Description based on online resource; title from PDF title page.
Source of Description:
This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Thesis:
Thesis (Ph.D.)--University of Florida, 2007.
Local:
Adviser: Nair, Ramachandr P.
Statement of Responsibility:
by Asako Takimoto.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Takimoto, Asako. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
663112616 ( OCLC )
Classification:
LD1780 2007 ( lcc )

Downloads

This item has the following downloads:

takimoto_a.pdf

takimoto_a_Page_033.txt

takimoto_a_Page_017.txt

takimoto_a_Page_182.txt

takimoto_a_Page_164.txt

takimoto_a_Page_004.txt

takimoto_a_Page_010.txt

takimoto_a_Page_103.txt

takimoto_a_Page_074.txt

takimoto_a_Page_014.txt

takimoto_a_Page_119.txt

takimoto_a_Page_090.txt

takimoto_a_Page_109.txt

takimoto_a_Page_080.txt

takimoto_a_Page_088.txt

takimoto_a_Page_076.txt

takimoto_a_Page_063.txt

takimoto_a_Page_161.txt

takimoto_a_Page_181.txt

takimoto_a_Page_054.txt

takimoto_a_Page_160.txt

takimoto_a_Page_106.txt

takimoto_a_Page_144.txt

takimoto_a_Page_042.txt

takimoto_a_Page_029.txt

takimoto_a_Page_078.txt

takimoto_a_Page_146.txt

takimoto_a_Page_091.txt

takimoto_a_Page_085.txt

takimoto_a_Page_137.txt

takimoto_a_Page_047.txt

takimoto_a_Page_104.txt

takimoto_a_Page_009.txt

takimoto_a_Page_075.txt

takimoto_a_Page_133.txt

takimoto_a_Page_138.txt

takimoto_a_Page_086.txt

takimoto_a_Page_056.txt

takimoto_a_Page_032.txt

takimoto_a_Page_125.txt

takimoto_a_Page_030.txt

takimoto_a_Page_052.txt

takimoto_a_Page_071.txt

takimoto_a_Page_043.txt

takimoto_a_Page_172.txt

takimoto_a_Page_171.txt

takimoto_a_Page_048.txt

takimoto_a_Page_096.txt

takimoto_a_Page_094.txt

takimoto_a_Page_126.txt

takimoto_a_Page_020.txt

takimoto_a_Page_099.txt

takimoto_a_Page_040.txt

takimoto_a_Page_177.txt

takimoto_a_Page_102.txt

takimoto_a_Page_055.txt

takimoto_a_Page_179.txt

takimoto_a_Page_168.txt

takimoto_a_Page_008.txt

takimoto_a_Page_046.txt

takimoto_a_Page_147.txt

takimoto_a_Page_128.txt

takimoto_a_Page_145.txt

takimoto_a_Page_167.txt

takimoto_a_Page_095.txt

takimoto_a_Page_098.txt

takimoto_a_Page_049.txt

takimoto_a_Page_135.txt

takimoto_a_Page_129.txt

takimoto_a_Page_025.txt

takimoto_a_Page_011.txt

takimoto_a_Page_175.txt

takimoto_a_Page_136.txt

takimoto_a_Page_097.txt

takimoto_a_Page_130.txt

takimoto_a_Page_038.txt

takimoto_a_Page_178.txt

takimoto_a_Page_041.txt

takimoto_a_Page_151.txt

takimoto_a_Page_173.txt

takimoto_a_Page_044.txt

takimoto_a_Page_037.txt

takimoto_a_Page_139.txt

takimoto_a_Page_035.txt

takimoto_a_Page_083.txt

takimoto_a_Page_110.txt

takimoto_a_Page_158.txt

takimoto_a_Page_183.txt

takimoto_a_Page_169.txt

takimoto_a_Page_084.txt

takimoto_a_Page_005.txt

takimoto_a_Page_021.txt

takimoto_a_Page_007.txt

takimoto_a_Page_166.txt

takimoto_a_Page_002.txt

takimoto_a_Page_141.txt

takimoto_a_Page_101.txt

takimoto_a_Page_122.txt

takimoto_a_Page_026.txt

takimoto_a_Page_100.txt

takimoto_a_Page_015.txt

takimoto_a_Page_018.txt

takimoto_a_Page_013.txt

takimoto_a_Page_019.txt

takimoto_a_Page_061.txt

takimoto_a_Page_016.txt

takimoto_a_Page_127.txt

takimoto_a_Page_112.txt

takimoto_a_Page_006.txt

takimoto_a_Page_148.txt

takimoto_a_Page_124.txt

takimoto_a_Page_108.txt

takimoto_a_Page_092.txt

takimoto_a_Page_081.txt

takimoto_a_Page_174.txt

takimoto_a_Page_067.txt

takimoto_a_Page_117.txt

takimoto_a_Page_059.txt

takimoto_a_Page_064.txt

takimoto_a_Page_143.txt

takimoto_a_Page_045.txt

takimoto_a_Page_001.txt

takimoto_a_Page_134.txt

takimoto_a_Page_034.txt

takimoto_a_Page_024.txt

takimoto_a_Page_140.txt

takimoto_a_Page_170.txt

takimoto_a_Page_153.txt

takimoto_a_Page_114.txt

takimoto_a_Page_150.txt

takimoto_a_Page_003.txt

takimoto_a_Page_057.txt

takimoto_a_Page_022.txt

takimoto_a_Page_157.txt

takimoto_a_Page_184.txt

takimoto_a_Page_154.txt

takimoto_a_Page_066.txt

takimoto_a_Page_131.txt

takimoto_a_Page_111.txt

takimoto_a_Page_028.txt

takimoto_a_Page_039.txt

takimoto_a_Page_152.txt

takimoto_a_Page_142.txt

takimoto_a_Page_036.txt

takimoto_a_Page_079.txt

takimoto_a_Page_149.txt

takimoto_a_Page_162.txt

takimoto_a_Page_159.txt

takimoto_a_Page_123.txt

takimoto_a_Page_121.txt

takimoto_a_Page_180.txt

takimoto_a_Page_070.txt

takimoto_a_Page_093.txt

takimoto_a_Page_027.txt

takimoto_a_Page_072.txt

takimoto_a_Page_073.txt

takimoto_a_Page_132.txt

takimoto_a_Page_012.txt

takimoto_a_Page_087.txt

takimoto_a_Page_031.txt

takimoto_a_Page_051.txt

takimoto_a_Page_165.txt

takimoto_a_Page_155.txt

takimoto_a_Page_089.txt

takimoto_a_Page_065.txt

takimoto_a_pdf.txt

takimoto_a_Page_116.txt

takimoto_a_Page_050.txt

takimoto_a_Page_176.txt

takimoto_a_Page_120.txt

takimoto_a_Page_062.txt

takimoto_a_Page_077.txt

takimoto_a_Page_163.txt

takimoto_a_Page_068.txt

takimoto_a_Page_023.txt

takimoto_a_Page_118.txt

takimoto_a_Page_105.txt

takimoto_a_Page_115.txt

takimoto_a_Page_069.txt

takimoto_a_Page_058.txt

takimoto_a_Page_107.txt

takimoto_a_Page_053.txt

takimoto_a_Page_156.txt

takimoto_a_Page_060.txt

takimoto_a_Page_113.txt

takimoto_a_Page_082.txt


Full Text








Table 5-2. 613C ValUeS of whole soil and three fraction sizes from five studied land-use systems,
at Segou Region, Mali. (all values are average of three replicates)


Whole soil Fraction
Land-use system Depth (cm) 250 2000 ptm 53 250 ptm <53 ptm


Faidherbia albida
parkland


near tree


10
-40
-100
10
-40
-100
10
-40
-100

10
-40
-100
10
-40
-100
10
-40
-100

10
-40
-100
10
-40
-100
10
-40
-100


-21.5
-18.7
-16.6
-21.8
-17.5
-15.6
-19.5
-16.9
-15.1

-20.7
-18.8
-18.7
-20.6
-17.8
-17.8
-18.7
-17.6
-16.8

-23.9
-20.1
-17.1
-22.5
-19.3
-16.9
-20.9
-17.9
-16.3

-18.2
-17.9
-17.4

-16.3
-15.7
-16.4


-22.4
-20.3
-18.8
-22.9
-20.7
-18.2
-20.9
-21.4
-20.0

-23.1
-23.3
-22.7
-22.7
-19.9
-20.9
-20.5
-19.2
-19.6

-25.4
-21.1
-19.2
-23.3
-22.2
-19.2
-22.2
-18.9
-19.7

-20.4
-21.3
-22.9

-16.8
-16.9
-18.0


-20.3
-20.2
-18.6
-21.8
-19.7
-17.2
-20.0
-20.6
-18.6

-21.1
-20.1
-20.8
-20.4
-18.3
-18.8
-18.1
-18.9
-20.8

-23.9
-19.8
-18.0
-22.4
-19.3
-18.6
-18.8
-19.1
-18.9

-21.2
-20.4
-18.7

-18.5
-16.5
-16.2


-21.2
-18.6
-18.1
-21.5
-18.3
-17.3
-19.3
-17.3
-15.9

-21.4
-20.1
-19.2
-20.6
-17.8
-18.3
-19.3
-19.0
-18.9

-22.6
-20.3
-17.7
-23.1
-18.8
-17.7
-20.8
-17.7
-16.8

-19.1
-17.4
-16.9

-16.4
-15.7
-16.8


3 m from tree



10 m from tree


Vitellaria paradoxa
parkland


near tree


3 m from tree



10 m from tree


Live fence


near tree


1 m from tree line



3 m from tree line


Fodder bank




Abandoned land


0 -10
10 40
40 100

0 -10
10 40
40 100










Figure 5-8. Linear regression between silt + clay content of soil and whole soil C content in
three depth classes across five land-use systems in Segou region of Mali. The three
data points, one each in each depth class, marked by circles around them, refer to one
of the fodder bank plots the data from which were quite inconsistent with those from
the other fodder bank plots as well as all the other treatments; these points were
therefore considered as outliers and excluded from the regression.












Elevation
Land-use Name of the village Position Size of the plot
(m)

Faidherbia albida Togo N. 13.35, W. 6.31 300 1 ha
parkland

Vitellaria .paradoxa Dakala N. 13.32, W. 6.23 297 1 ha
parkland

Live fence Dougoucouna N. 13.37, W.6.37 298 294m (average)

Fodder bank' Dakala N. 13.32, W. 6.23 297 0.25 ha
Siguila N. 13.28, W. 6.21 305 0.25 ha
Banankoroni N. 13.35, W. 6.38 293 0.22 ha

Abandoned land Diamaribougou N. 13.36, W. 6.19 298 0.5 ha
10ne fodder bank plot from each village

Table 4-2. Characteristics of the experimental plots (three plots average) for five-selected land-
use systems in Segou region, Mali.
Number of
DBH (cm) Height (m) tre 1h' Species composition

Faidherbia albida 59.4 (1.8) 13.0 (0.9) 21 (5.3) Average 88.6% Faidherbia.
parkland albida dominance

Vitellaria paradoxa 41.7 (5.9) 9.9 (0.9) 20 (0.6) Average 80.6% Vitellaria
parkland paradoxa dominance

Live fence 2.5 (0.5) 2.5 (0.4) 3720 (882) Average 67.6% Acacia.
nilotica

Fodder bank 2.2 (0.5) 20(.) 588 (277) Gliricidia sepium average
only

Abandoned land 2.8 (0.6) 1.3 (0.4) 46 (30) Average 47.5% Guiera
diameter at ground senegalensis and 39.5%
Combretum micranthum


Table 4-1. Characteristics of the villages where the experimental plots were
region, Mali.


set up in Segou


Note: Numbers in parentheses are standard deviations.
standing trees in the plot.


Tree dominance means the percentage among the









land area, representing a large part of the agricultural landscape under subsistence farming in the

WAS and it is the predominant agroforestry system. For example, the agroforestry parkland

system occupies about 90 % of the agricultural land area in Mali (Cisse, 1995), and in Burkina

Faso, parklands are found throughout settled zones where agriculture is practiced.

Parklands are most often characterized by the dominance of one or a few tree species.

Species composition is generally more diverse and variable, however, in areas located farther

away from villages and only occasionally cultivated. Common species in the WAS are Acacia

senegal, Adansoniadd~~~ddd~~~ddd~~ digiata, Anogeissus leiocarpus, Balan2ites aegyptiaca, Bombax costatum,

Bora~ssus aethiopum, Ceiba pentandra, Diospyros mespiliformis, Elaeis guineensis, Faidherbia

albida, Hyphaene thebaica, Lannea microcarpa, Parkia biglobosa, Sclerocarya birrea,

Tama~rindus indict, Vitellaria paradoxa, V~1~~11~11 itex doniana, and Ziziphus mauritiana (Table 2-2)

(Boffa 1999).

Improved Agroforestry Practices

The expansion of rain-fed agriculture results in soil erosion through the removal of

vegetative cover and physical disturbance. Wind and water erosion is extensive in many parts of

the WAS. Practically every country of Africa is prone to desertification, but the Sahelian

countries at the southern fringe of the Sahara are particularly vulnerable (Reich et al. 2001). Soil

nutrients are removed through crops, erosion, and leaching by rainfall, without replenishment by

additions or regeneration under natural fallow. Inappropriate tillage and cultural practice reduce

soil infiltration and retention of water, which further degrade the land (de Alwis 1996). Also,

deforestation accelerates the land degradation as trees and shrubs are cut to satisfy the

construction, fuel, and fodder requirements of the cultivators and their livestock. In the WAS,

farmers/pastoralists usually graze their animals in the open area without any control (Figure 2-6).

Degraded land spreads as these animals go further after eating the vegetation around the villages.













No. Price Cost (CFA)
Exchange rate US Dollar 550
Size of the fodder bank 200 m perimeter
Discount rate 15%
Material Costs
Seedling (A. nilotica:) 400 29.0 11600
Seedling (Z. nmauritiana:) 100 28.0 2800
Seedling (A. senegall) 100 25.9 2590
Seedling (B. rufescens) 100 25.9 2590
Seedling (L. inernzis) 100 24.5 2450
Seedling (G. sepium ) 200 35.1 7020
TOTAL 29050
Agriculutral equipment (every year) 1500
Labor Costs
Daily labor wage 1 750 750

Revenues
Yields from live fence products (after 3rd year) 18594
Price of fuelwood (CFA/kg) 18 18
Price of timber (CFA/log) 800 37 29600
Price of C ($, FCFA) kg C 0.042 23


APPENDIX C
COST BENEFIT ANALYSIS (CASH FLOW) OF FODDER BANK









Conclusions

Regarding the overall attractiveness of the selected land use systems, live fence and

fodder bank are more suitable to start as agroforestry C sequestration proj ects than the parkland

systems for small-scale farmers in the studied region. Between the two improved systems, live

fence has higher C sequestering potential per unit area and is less risky than fodder banks. This

situation could change, however, depending on tree management and conditions that affect tree

growth. Adopting these systems on cultivated land rather than on abandoned land is likely to

sequester more C and be more profitable. Since parklands are traditionally practiced, they do not

qualify as a new C sequestration project. Nevertheless, F. albida trees are more attractive than

yK paradoxa trees in terms of C sequestration potential.

Agroforestry Adoption for C sequestration in the Study Region

Based on the findings summarized above and information acquired through fieldwork and

literature review, some factors that either limit or favor agroforestry adoption in the regions can

be identified.

Limiting Factors

With the current price range (and its large fluctuations) for C credits and the amount of C

potentially sequestered, the income from C sale is not likely to be a maj or source of income for

farmers in the WAS and therefore is not likely to be a strong incentive to start the live fence or

fodder bank. In addition, farmers are concerned about other factors as well, such as risks in

undertaking a new farming practice. Farmers might need some support such as technical and/or

material assistance to cover initial costs, and/or insurance or some kind of safety net in case the

trees die due to unexpected causes. As regards parklands, increasing the tree density is difficult

because parkland trees grow relatively slowly. Also, it is technically challenging since parkland

tree species rely on natural regeneration.














Eae4: Chercher du materiel pour la haie morte autour de la haie vive
Personnes impliquees No. de personnel/ No. de personnel/ No. de personnel/
(H/F/E): jours (heures) jours (heures) jours (heures) -
Anl:. An 2: An 3:





Note :




Etape 5: Construire la haie morte autour de la haie vive
Personnes impliquees No. de personnel/ No. de personnel/ No. de personnel/
(H/F/E): jours (heures) jours (heures) jours (heures) -
Anl:. An 2: An 3:





Note :




Etape 6: Entretien des banques fourrageres et la haie vive (suivi, boucher les spaces etc. SANS
recolte)
Personnes impliquees No. de personnel/ No. de personnel/ No. de personnel/
(H/F/E): jours (heures) jours (heures) jours (heures) -
Anl:. An 2: An 3:





Note :
An 4. 5. 6.... (si possible)











ITE1VI/EAR
Material costs
Labor costs
Obtaining & planting
seeds/seedlings
Watering plants
Collecting material for dead
Constructing dead fence
Maintenance of live fence
Collecting products frcanlive
Marketing products from live
Harvesting
TOTAL Labor costs


9 10 11 12 13 14 15 16 17
1000 1000 1000 1000 1000 1000 1000 1000 1000








2,625 2,625 2,625 2,625 2,625 2,625 2,625 2,625 2,625
1,500 1,500 1,500 1,500 1,500 1,500 1,500 1,500 1,500
375 375 375 375 375 375 375 375 375

4,500 4,500 4,500 4,500 4,500 4,500 4,500 4,500 4,500

5500 5500 5500 5500 5500 5500 5500 5500 5500
1563 1360 1182 1028 894 777 676 588 511


IOTALLENDSTS
pv cost


Revenues
Yiehis front 1ive fence products 27054 27054 27054 27054 27054 27054 27054 27054 27054
Yields from fuelwood
Yields from timber


IOTALLIUEVENTIES
py revenue
Net benefit (cash fow)


27054 27054 27054 27054 27054 27054 27054 27054 27054
7690 6687 5815 5057 4397 3824 3325 2891 2514
21554 21554 21554 21554 21554 21554 21554 21554 21554

0.28426 0.24718 0.21494 0.18691 0.16253 0.14133 0.12289 0.10686 0.09293
6126.99 5327.82 4632.89 4028.6 3503.13 3046.2 2648.87 2303.36 2002.92


Discount factor
Present value
NPV
IRR
BCR


Ideal accounting system
C storage (kg)
C sale (FCFA)
Net cash fow
NPV
IRR
pv revenue
BCR

Tonne-year accounting
C sale (FCFA)
Net cash fow
NPV
IRR
py revenue
BCR


50 50 30 30 30 30 30 7 7
1155 1155 693 693 693 693 693 161.7 161.7
22709 22709 22247 22247 22247 22247 22247 21716 21716



8018.76 6972.83 5964.03 5186.11 4509.66 3921.45 3409.95 2908.4 2529.04


25 25 15 15 15 15 15 3 3
21579 21579 21569 21569 21569 21569 21569 21557 21557



7697.49 6693.47 5818.28 5059.37 4399.45 3825.61 3326.62 2891.49 2514.34










Tropical forest conversion contributes as much as 25 % of net annual CO2 emiSSIOnS

globally (Palm et al. 2004). Removing this atmospheric C and storing it in the terrestrial

biosphere is, thus, one option for mitigating the emission of this GHG. A recent assessment of

Rose et al. (2007), referenced by Intergovernmental Panel on Climate Change (IPCC)'s newest

report, suggests that land-based mitigation agriculture, forestry, and biomass liquid and solid

energy substitutes can be cost-effective land mitigation options. And, it can contribute over the

century 94 to 343 Pg C equivalent of greenhouse gas emission abatement, which is 15 to 40

percent of the total abatement required for stabilization.

Agroforestry for C sequestration

Under the Kyoto Protocol's Article 3.3, further defined by Marrakesh Accord in 2001,

agroforestry was recognized as an option of mitigating GHGs. Since then, the C sequestration

potential of agroforestry systems has attracted greater attention from both industrialized and

developing countries. It is attractive because of its applicability to a large number of people and

areas currently in agriculture, as well as its perceived potential for reducing pressure on natural

forests. Also, Clean Development Mechanism (CDM), defined in Article 12 of the Protocol adds

the attractiveness, because the CDM provides for Annex I Parties (industrialized countries which

have emission reduction goals) to implement proj ect activities that reduce emissions in non-

Annex I Parties (developing countries), in return for certified emission reductions (CERs)

(UNFCCC 2007). Since agroforestry is traditionally and widely practiced in developing

countries, it is feasible/easy options for both developing and developed groups of countries to

start as mitigation proj ects under the CDM.

However, as stated by Makundi et al. (2004) and several others, estimating the amount of

C sequestered by agroforestry poses unique challenges. In addition to the complexity caused by

diverse factors such as climate, soil type, tree-planting densities, and tree management as well as














O Biomnass C 0 Soil C 0 10 cm

2 Soil C 10 40 cm I Soil C 40 100 cm


Ab ovegroun d
biomass


40


S20


0


S20


40


60


Lve
fence


Fodder
bank


Aban don ed
land


I I IGround level
b ::: a
bI a


a Soil


-r Faidhrbia
albida
-parkland


Vitala~riaa
pardoxa:
parkl and


Figure 4-7. Aboveground and belowground C stock per ha of five selected land-use systems.
Biomass C is shown above the x-axis, and soil C is shown below the axis with three
soil depth classes. Values of live fence and fodder bank are from UNFCCC
equations.











ITEM/YEAR 0 1 2 3 4 5 6 7 8
Material costs 30050 1000 1000 1000 1000 1000 1000 1000 1000


Labor costs
Obtaining & planting
seeds/seedlings
Watering plants
Collecting material for dead
Constructing dead fence
Maintenance
Collecting products frona live fence
Collecting fodder
Marketing products
Harvesting
ITYTAL Labor costs

ITOTAL COSTS
pv cost
Revenues
Yields front 1ive fence products
Yields front foder
Labor thnesaved
Yields front fuelwood
Yields front timber

ITOTAL REVIR4CES
py revenue
Net benefit (cash fow)

Discount factor
Present value
NWV
IRR
BCR

Ideal accounting systenI
C storage (kg)
C sale (FCFA)
Net cash fow
NWV
IRR
pv revenue
BCR


8946
2384
4,381 4,124 2,062
3,093 2,062 1,289
5,786 5,909 6,169 6,429 6,429 6,429 6,429 6,429 6,429
1031 1031 1031 1031 1031 1031 1031
2679 2679 2679 2679 2679 2679
375 375 375 375 375 375 375

24590 12095 10925 10513 10513 10513 10513 10513 10513
94589
54640 13095 11925 11513 11513 11513 11513 11513 11513
54640 11387 9017 7570 6583 5724 4977 4328 3764

379 18594 18594 18594 18594 18594 18594 18594
3,500 4,000 4,000 4,000 4,500 4,500
16875 19286 19286 19286 19286 19286




0 379 18594 38969 41880 41880 41880 42380 42380
0 330 14060 25623 23945 20822 18106 15932 13854
-54640 -12716 6668 27456 30366 30366 30366 30866 30866

1 0.86957 0.75614 0.65752 0.57175 0.49718 0.43233 0.37594 0.3269
-54640 -11057 5042 18053 17362 15097 13128 11604 10090
87319
29.5%
1.67



0 78 78 78 78 78 64 64 64
0 1801.8 1801.8 1801.8 1801.8 1801.8 1478.4 1478.4 1478.4
-54640 -10914 8470 29258 32168 32168 31845 32345 32345
96394
31.4%
0 1896.68 15422 26807.5 24974.9 21717.3 18744.8 16487.8 14337.2
1.74



0 39 39 39 39 39 32 32 32
-54640 -12677 6707 27495 30405 30405 30398 30898 30898
87523
29.5%
0 363.583 14088.9 25648.3 23966.9 20840.8 18119.4 15944 13864.3
1.67


Tonne-year accounting
C sale (FCFA)
Net cash flow
NWV
IRR
py revenue
BCR









Biomass C Sequestration


Studies in Various Ecoregions

The amount of C sequestered in an agroforestry system depends largely on the nature of

the system put in place, the structure and function of which are determined by environmental and

socioeconomic factors (Albrecht and Kandji. 2003). Other factors influencing C storage include

tree species and system management (Delaney and Roshetkol998; Roshetko et al. 2002). Palm

et al. (2004) compared the amount of C stored (above-ground) in different ecological systems

(Table 3-2). To compare the rotation of the different land-use systems, time-averaged C of each

system was used.

C stocks in the vegetation of the primary forests averaged 300 Mg C ha- and that of

logged or managed forests ranged from a high of 228 Mg C ha-l in Cameroon to a low of 93 Mg

C ha-l in Indonesia. Time-averaged aboveground C for the different land-uses ranged from 50 -

90 Mg C ha-l in long-fallow shifting cultivation and complex agroforestry systems, to 30 60

Mg C ha-l in simple agroforestry systems and most tree plantations and medium-fallow

rotations. These are considerably larger than those for annual crops or pastures.

Studies in West Africa

In a review of C sequestration in tropical agroforestry systems, Albrecht and Kandji (2003)

estimated that agrosilvicultural system could sequester 29 53 Mg C ha-l in humid tropical

Africa. A case study in Cameroon (humid west and central Africa) showed that the cacao

(Theobroma cacao) agroforest is superior to the alternative food crop production system (slash-

and-burn), both in C sequestration and below- and above-ground bio-diversity. Total biomass in

cacao agroforest was 304 Mg ha- compared to crop fields (84 Mg ha- ) (Duguma et al. 2001).

Compared to C gains in the humid tropics, the benefits of agroforestry in the WAS, such as

parklands or improved fallow seem to be less. A simulation study in Senegal compared the C









amount of non-crystalline clays (Powers and Schlesinger 2002). Indeed, the whole soil C

content was related to silt + clay content in all three depth classes (strongest in 10 40 cm)

(Figure 5-8). Silt + clay content also had strong relationship with S fraction C (<53 Cpm), the

strongest being in 40 100 cm (Figure 5-9), but not as well as with whole soil C. This suggests

that silt and clay are mainly associated with soil C in <53 Cpm size, especially in deep soil, but

they are also associated with larger fraction size form of soil C. Based on the data points on

Figure 5-9, abandoned land data do not seem to follow this relationship well, while they follow

the relationship better in Figure 5-8. This could be because silt and clay are more involved in

forming larger than 53 Cpm fraction size C in abandoned land compared with other systems.

After being abandoned for a few years (less than 10: see Chapter 4), the land probably was

subj ected to extensive erosion that took away aggregates and sandy particles from the surface

layer. The remaining soil was higher in silt and clay and formed a hard surface pan, which

prevented further disturbance or leaching. The maj ority of soil C in the abandoned land was of

C4 plant origin (Figure 5-6), suggesting that the stored C in abandoned land was mainly from the

previous land-use (land cultivated with C4 plants) and was well protected.

The other four land-use systems were not significantly different in terms of the soil

characteristics (pH, bulk density, particle size). However, as seen in the soil pit photo (Figure 5-

2), the color was quite different in each soil pit, suggesting soil variations among plots of same

land-use systems in the same village. Still, soil C content variations among these soils can be

explained more as a consequence of the influence of trees and land management than caused by

soil characteristics as in the case of abandoned land. Among the two parklands and live fence

systems, C content was expected to be: near tree > under the crown > outside the crown, as

reported in a similar study in the parkland system in South Mali, which showed the significant





























Figure 4-6. Abandoned land just outside of Diamaribougou village. The land was cultivated
until less than 10 years ago. The surface soil is eroded leading to formation of a hard
surface pan. Only certain bushes such as Guiera senegalensis and Combretum
micranthum can survive on this type of degraded land. (Photographed by author)









remain few, and are mainly in lessening the negative impact on the climate resulting from land-

use change and deforestation.

For C offset projects, however, the risk of shifting from cropping to more C-beneficial

practices seems to be high, especially for subsistence farmers who have little lands. A study of C

sequestration through agroforestry in Senegal found that the costs for resource-poor farmers are

considerably higher than those of intermediate and richer farmers, because the former often lack

the necessary assets (land, labor, and animals) to switch from current to alternative practices

(Tschakert 2007).

Many African policy makers and financial institutions express little interest in controlling

GHG emissions or adapting to changes in climate. This attitude is based on their experience that,

in general, other, more local, environmental problems have more direct influence on their

populations than climate change. Senior government officials and most members of civil society

do not understand the climate issue very well (Denton et al. 2001). Many development

practitioners remain skeptical, arguing that C brokers, national ministries and local leaders rather

than needy rural populations will benefit from C proj ects.

An important challenge for the WAS countries lies in that they need to be more

"attractive" than the other African and developing countries in order to draw and hold

investments for C sequestration projects under CDM. As discussed above, the potential C gains

in the WAS through agroforestry has been considered to be unattractive. Synergies between

development and climate change response, however, can be an answer. Agroforestry projects,

which protect soils and result in C sequestration, also provide employment opportunities for local

farmers (Hardner et al. 2000). Soil C sequestration project through agroforestry could provide a

crucial link between three international conventions: the UN Framework Convention on Climate











6C1


CO

-2



LL








Project year (O 25)
5 =Mean,+1/-1 SD -95%,5%


Figure 6-7. Simulated net benefit (total costs -total revenues in each year) of the fodder bank
project (without C sale). Mean value of each year' s probability distribution is shown
in the yellow line, red range is plus minus 1 standard deviation from the mean, and
green range is the 5 to 95 % likelihood of the value.

































145









gains after 25 years from protecting parkland systems (2.48 Mg C ha- ) with that of rotating

crops with Luecaena spp. fallow planting (6.35 Mg C ha- ) (Tschakert 2004). Although drylands

as whole are believed to provide a substantial opportunity for C offsets especially in soil C,

because of their large area (47.2 % of land in the world) and low human populations (Lal 2004b),

most studies in the Sahel region have concluded that the potential for C gains per unit area

through agroforestry is relatively low (Walker and Desankar 2004;Woomer et al. 2004b)

compared with other ecoregions.

In general, introducing trees into agricultural systems is expected to increase water and

nutrient availability, because trees can fix nitrogen, retrieve water and nutrients from below the

rooting zone of crops, and reduce water and nutrient losses from leaching and erosion (Buresh

and Tian 1997). This tree effect has been demonstrated in various agroforestry systems in the

semiarid region. However, this added value was lowest where it is most needed, in resource-

poor environments: the competition between woody plants and crops is strong (Kater et al. 1992;

Breman and Kessler 1997). Water constraints are the strongest limitations for C sequestration in

the WAS. Several local tree species such as Acacia tortilis, Guiera senegalensis, Pterocarpus

lucens have been planted in grasslands of the region for sequestering C, but their capacity to

grow has been shown to be constrained by moisture availability (Woomer et al. 2004a). The

capacity of exotic dryland tree species to afforest the WAS is also uncertain.

Since the moisture and nutrient levels of the study field are expected to be low, the tree

growth and the consequent C storing will not be extremely high, either, compared with more

moist parts of West Africa. However, the amounts of C sequestered as a result of specific land-

uses are mostly unknown in the WAS, thus, it is worthwhile to conduct the research to have a

reference data for future C sequestration proj ects.










(Gossypium spp.) is the main crop (and product) of the region and the region is well known for it.

A large cotton mill, invested in by Chinese companies, is operating at the edge of the city. Rice

(Oryza glaberrima and Oryza sativa) is grown extensively in the irrigated area around the Niger

River (Republique du Mali 2005). Farmers commonly grow rain-fed pearl millet (Pennisetum

glaucum) and sorghum (Sorghum bicolor) as staple food crops.

Selected Land-use Systems for Field Data Collection

A preliminary survey was first conducted in July 2005 to identify the targeted land-use

systems and possible villages to locate on-farm plots. Five systems were selected: two parkland

agroforestry systems, two improved agroforestry systems, and an "abandoned" land (degraded

land) for comparison.

Parkland systems

The maj or land-use in the Segou region, as in most of other parts of the WAS, is parkland

agroforestry. Two parklands types are common: with Faidherbia albida or Vitellaria paradoxa~11~~1~~11~

as the dominant tree species. These two types occupy more than 60 % of cultivated land in

Segou region (personal communication, August 2005, with Director of Forestry Department,

Segou). Tree density is in the range of 20 to 30 trees/ha in both systems. Crops cultivated

underneath the trees include pearl millet and sorghum, sometimes intercropped with cowpea

(Vigna unguiculata) and/or banbarra groundnut (Vigna subterranea syn. Voandzeia

subterranea).

F. albida has a unique characteristic of "reverse" phenology or foliation (i.e. bearing leaves

during the hot dry season and dropping them before the rainy season), which is quite

advantageous for agroforestry: it reduces shading of crops grown underneath the tree and

possibly reduces competition for water between trees and crops, and enables farmers to grow

crops under the trees with practically very little reduction of cropped area in the intercropping



































1.3 Si vous avez achete les semences ou les plants, quels etaient les cotts?
Unite Prix par unite Quantite Cott total
Semences
Plants

1.4 Outils utilises pour installer et maintenir les banque fourrageres:
Note : Noter tous les outils utilises pour l'installation et l'entretien de les banque fourrageres, pas
les outils uniquement utilises pour le travail champ~tre agricole. (preferer le prix de achete, mais
si difficile, noter quel annee).
Outils(entrer Nombre Prix d'outil Valeur total Nombre
codes): d'outils (prix de d'annees
marched en F d'utilisation
CFA)


APPENDIX A
SOCIAL SURVEY QUESTIONNAIRE FOR FODDER BANK OWNERS


Questionnaire #:


Date:


Duree approximative:


QUESTIONS SUR L'INSTALLATION DES BANQUES FOURRAGERES

1.1 Methode d'installation:
(Entrer code)
1: Semi-direct
2: Plants a racines nues
3: Plants en pots
4: Autres
9: Pas de reponse


1.2 Comment vous avez obtenu les semences ou les plants?


(Entrer code)
1: Cultives en pepiniere par le paysan
2: Don d'une structure d'encadrement
3: Achete d'un autre paysan / au marched


4: Don d'un autre paysan
5: Autres
9: Pas de reponse


Houe
Hache
Pioche
Coupe-coupe
Pelle


6: Piquet/Piquasse
7: Brouette
8: Arrosoir
9: Charrue
10: Multiculteur


11: Charrette
12: Barre a mine
13: Autres
99: Pas de repons









(R2=0.60) with the S size C (contains protected C) content. However, this relationship was not

seen for M and L fractions; the more fraction percentage of M or L size in the soil does not mean

the more soil C of those sizes. Also, the regression between silt + clay content and whole soil C

content was strong, especially in the 10 40 cm soil depth (Figure 5-8).

Sand-, silt-, clay-, or silt + clay contents did not have strong relationship with L fraction C

or M fraction C. Although the regression between silt + clay content and S fraction C was poor

(R2=0.16) in the total data set, it became more pronounced when the data set was divided by

depth class. R2 ValUeS between silt + clay and S fraction C were 0.53 at 0 10 cm, 0.44 at 10 -

40 cm, and 0.67 at 40 100 cm (Figure 5-9). Soil pH or bulk density did not show any strong

relationship with C in whole soil or in each of the three fractions in any depth or for all depths

combined.

Percentages of C3-origin C or C4-origin C did not have a significant relationship with

whole soil C content. There was a strong relationship between C3-origin C content and L

fraction C content (R2=0.67) throughout the data set. The relationship was stronger at 0 10 cm

soil (R2=0.72) when data sets were divided by depth (Figure 5-10), but were not significant in

deeper depths. The relationship between C3-origin C and S fraction C content was also observed

in the 40 100 cm depth (R2=0.45).

Discussion

Contrary to expectations, C content in all soil depths was higher in the "abandoned" land

than in any of the four agroforestry systems (Figure 5-4), although the significance varied

depending on the depth class. Judging from the observation that the abandoned land soil had

significantly more silt and clay fractions than those of soils under the agroforestry systems

(Table 5-1), it seems that the whole soil C content was directly related to the silt and clay

contents of the soil. In general, soil organic C content is known to correlate positively with the










Charlotte Skov, Chrysa Mitraki, Rania Habib, Maitreyi Mandal, Trina Hofreiter, Troy Thomas,

and my fiance, Nick Georgelis.

Last but not least, I express my most profound gratitude to my mother Ayuko Takimoto,

whose endless love and confidence in me made me come this far.









fence and fodder bank), and land that is out of cultivation (abandoned or degraded) in the Segou

Region, Mali. The second component involved a study of soil C dynamics of these systems: the

extent of soil C storage/accumulation by trees and stability of the C accumulated were

investigated. In the third component, socioeconomic feasibility of the agroforestry systems was

examined in the context of C sequestration and C credit sale.

Research results show that the selected agroforestry systems have the potential for

sequestering more C both above- and belowground than in tree-less land-use systems, and that

the trees tend to contribute to storing more stable C in the soil. Among the selected land-use

systems, live fence and fodder bank are more suitable to start as agroforestry C sequestration

proj ects than the traditional parkland systems for smallholder farmers in the studied region.

Between the two improved systems, live fence has higher C sequestering potential per unit area

and is economically less risky than fodder banks. Adopting these systems on cultivated land

rather than on abandoned land is likely to sequester more C and be more profitable. Since

parklands are traditionally practiced, they are not likely to qualify as a new C sequestration

proj ect soon. Nevertheless, F. albida trees are more attractive than y. paradoxa trees in terms of

C sequestration potential.

These results can be used for development of recommendations and guidelines on selection

of land use-systems and species and their management, for planning successful C sequestration

proj ects in the West African Sahel.






















































Figure 6-2. Simulated NPV probability distribution of the live fence project (without C sale).
The distribution is likelihood (y-axis) of the proj ect' s NPV (x-axis): the worst
scenario can be less than -100,000 FCFA in NPV, and the best scenario can be close
to 150,000 FCFA in NPV. The peak of the distribution is most likely (best guess)
scenario of the proj ect.


I


Figure 6-1. Social survey with farmers in Segou, Mali. Based on the ICRAF database, all
farmers who have at least once harvested fodder from the fodder bank were
interviewed. The survey was conducted in Bambara (local language) and translated
to French through the interpreter (man with a jacket in the photos).


in
-150 -100 -50 0 50 100
N PV (FCFA i n tho~usan ds)


100


1 5%


96.5


46


-5 0. 107 8









maj or technique to ameliorate the spreading land degradation in the WAS. Details of the

improved agroforestry practices being introduced in the study region are described in Chapter 4.





























Figure 2-5. Parkland system in Segou, Mali. Trees are scattered in the cultivated land, and
protected for non-timber use. Ox-drawn plows are used to till the land to sow the
crops upon onset of rains. (Photographed by author)


?3;:' ;":""~i; ';;'
r*Il ..~ .


111


Figure 2-6. Allowing the cattle to roam freely on the landscape during the dry season after
seasonal crops have been harvested is a common feature of the WAS land-use system.
This often leads to overgrazing (photo from the Segou region, Mali). When the open
lands near the village are depleted of vegetation, farmers are forced to take the
animals further away from the village. (Photographed by author)










5-8 Linear regression between silt + clay content of soil and whole soil C content in three
depth classes across Hyve land-use systems in Segou region of Mali ............... ... ............1 12

5-9 Linear regression between silt and clay content of soil and C in soil particles of <53
Cpm in three soil-depth classes across Hyve land-use systems in Segou, Mali ................... 114

5-10 Linear regression between C derived from C3 plants and C in the large soil particles
(250 2000 Cpm) at 0 10 cm soil depth across Hyve land-use systems of Segou
region, M ali.................. ...............116................

6-1 Social survey with farmers in Segou, Mali ................. ...............140.............

6-2 Simulated NPV probability distribution of the live fence proj ect (without C sale).........140

6-3 Simulated net benefit (total costs total revenues in each year) of the live fence
project (without C sale)............... ...............141.

6-4 Simulated NPV probability distribution of the live fence proj ect (with C sale by the
ideal accounting method) ........... _... ......... ...............142..

6-5 Regression sensitivity analysis for NPV of the live fence proj ect (with C sale by the
ideal accounting method) ........... _... ......... ...............143..

6-6 Simulated NPV probability distribution of the fodder bank proj ect (without C sale).....144

6-7 Simulated net benefit (total costs total revenues in each year) of the fodder bank
project (without C sale)............... ...............145.

6-8 Simulated NPV probability distribution of the fodder bank proj ect (with C sale by
the ideal accounting method) ................. ...............146......... .....

6-9 Regression sensitivity analysis for NPV of the fodder bank proj ect (with C sale by
the ideal accounting method) ................. ...............147......... .....









expected difference between the measured 613C ValUeS, it is possible to calculate the proportion

of C4 derived material and C3 derived material in biomass or soil C (Balesdent et al. 1998). This

method has been used for soil C research to assess vegetation composition change (Dzurec et al.

1985) or to follow the dynamics (Harris et al. 2001). Mcdonagh et al. (2001) measured how

SOM from original vegetation (forest: C3 plants) were diminished after continuous cultivation of

maize (Zea nzays: C4 plants). In agroforestry system, Jonsson et al. (1999) used this method to

prove the positive influence of trees (C3 plants) on SOM increase at millet (Pennisetunt

glaucunt: C4 plants) cropland.

Soil C in the WAS

In the WAS, most of the soils have low activity clay, with low water retention and are

susceptible to soil erosion and compaction, as described in Chapter 2. Organic matter content of

these soils has been depleted due to overgrazing, agricultural mismanagement, deforestation and

overexploitation of the natural resources. As a result, soil organic C stock density in West Africa

is very low (4.2 4.5 kg C m-2), COmpared with the world average (10.9 11.6 kg C m-2), and

relatively lower even when compared to the average for Africa (6.4 6.7 kg C m-2) (Batj es

2001).

Soil degradation is a maj or obstacle for agricultural productivity and thus sustainable

development of the WAS. The possibility of enhancing C sequestration through improved soil

management has been discussed academically and at international workshops, as part of the

search for agroecosystem sustainability in the region. Among soil-nutrients studies in Africa, tree

integration into croplands is often recommended for soil amelioration (Onim et al. 1990; Tiessen

et al. 1991; Manlay et al. 2002). Kang et al. (1999) reported Grilicidia sepium and Leucaena

leucocephala increased surface soil organic C by 15 % compared to sole crops in a 12-year

hedgerow intercropping trial on a Nigerian Alfisol. Parkland system studies affirm that the soil


































To my parents and grandmother










Lal, R. 1999. Global carbon pools and fluxes and the impact of agricultural intensification and
judicious land use. p. 44-52. hz Prevention of Land Degradation, Enhancement of Carbon
Sequestration and Conservation of Biodiversity through Land Use Change and Sustainable
Land Management with a Focus on Latin America and the Caribbean. FAO, Rome, Italy.

Levasseur, V., M. Djimde, and A. Olivier. 2004. Live fences in Segou, Mali: an evaluation by
their early users. Agrofor. Syst. 60:131-136.

Littmann, T. 1991. Rainfall, temperature, and dust storm anomalities in the African Sahel.
Geogr. J 157:136-160.

MacDicken, K.G. (ed.) 1997. A guide to monitoring carbon storage in forestry and agroforestry
projects. Winrock Intemnational Institute for Agricultural Development, Forest Carbon
Monitoring Program. Arlington, VA.

Makumba, W., B. Janssen, O. Oenema, F.K. Akinnifesi, D. Mweta, and F. Kwesiga. 2006. The
long-term effects of a gliricidia-maize intercropping system in Southemn Malawi, on
gliricidia and maize yields, and soil properties: Nutrient management in tropical
agroecosystems. Agric. Ecosyst. Environ. 116:85-92.

Makundi, W.R. and J.A. Sathaye. 2004. GHG Mitigation potential and cost in tropical forestry -
relative role for agroforestry. Environ. Dev. Sust. 6:235-260.

Manlay, R.J., J.L. Chotte, D. Masse, J.Y. Laurent, and C. Feller. 2002. Carbon, nitrogen and
phosphorus allocation in agro-ecosystems of a West African savanna: III. Plant and soil
components under continuous cultivation. Agric. Ecosyst. Environ. 88:249-269.

Marino, B.D. and M.B. McElroy. 1991. Isotopic composition of atmospheric CO2 inferred from
carbon in C4 plant cellulose. Nature 349:127.

Masera, O.R., J.F. Garza-Caligaris, M. Kanninen, T. Karjalainen, J. Liski, G.J. Nabuurs, A.
Pussinen, B.H.J. de Jong, and G.M.J. Mohren. 2003. Modeling carbon sequestration in
afforestation, agroforestry and forest management proj ects: the CO2FIX V.2 approach.
Ecol. Modell. 164:177-199.

McDonagh, J.F., T.B.Thomsen, and J. Magid. 2001. Soil organic matter decline and
compositional change associated with cereal cropping in southern Tanzania. Land Deg.
Dev. 12:13-26.

McLauchlan, K.K. and S.E. Hobble. 2004. Comparison of labile soil organic matter fractionation
techniques. Soil Sci. Soc. Am. J. 68:1616-1625.

Mitchell, T. 2005. Sahel rainfall index (20-10N, 20W-10E), 1898 2004. [Online] Avaliable at
http://jisao.washington. edu/data~sets/sahel/index2 .html#values (verified 5 Jul. 2007).
University of Washington, Seattle, WA.

Montagnini, F. and P.K.R. Nair. 2004. Carbon sequestration: An underexploited environmental
benefit of agroforestry systems. Agrofor. Syst. 61-62:281-295.









CHAPTER 5
SOIL CARBON SEQUESTRATION IN DIFFERENT PARTICLE-SIZE FRACTIONS AT
VARYING DEPTHS UNDER AGROFORESTRY SYSTEMS IN MALI

Introduction

The measurement of carbon (C) content is part of the basic soil analysis procedure.

Inventory data on soil C content is available in most of the WAS countries. However, to discuss

soil C sequestration as one of the options for mitigation of atmospheric CO2, the stability of soil

C (how well C is "captured" inside the soil) has to be considered. In other words, the soil C that

goes back to the atmosphere after decomposition within a month of its deposition, and that stays

in the soil for decades should not be counted as similar in terms of C credits. Characteri stics

such as the stability of soil C are very controversial issues in estimating and accounting

methodologies (Ingram and Fernandes 2001; Garcia-Oliva and Masera 2004). Also, soil-

sampling depth for these accounting procedures needs to be deeper than for normal soil analysis.

The conventional soil C studies of agricultural systems mostly focus on soil organic matter in the

surface layer of 20 cm for the interests of soil fertility. But sampling of deeper soil horizon is

necessary in efforts to understand the extent of soil C protection and characteristics of various

soil C forms, especially the systems involving deep rooting plants such as trees (Jobbagy and

Jackson 2000).

In general, soil C dynamics regarding C sequestration have not yet been well studied,

although recent technological development and interests towards climate mitigation activities are

contributing to an increased number of this type of studies (Post et al. 2000; Sun et al. 2004).

Still, these studies are limited even in developed countries, and not easily available in the

research- resource limited area such as the studied region or Africa in general. The studies of

this nature that have been conducted so far have been in natural environment such as forest

stands, tundra, or grasslands, probably due to the relatively stable dynamics of soil properties









Soil Taxonomy Orders








L 1


I 8


Histaosls
Spodosols
Andisols
Oxisols
Vertisol~s
Anidis~ols
Ultisols
Molliso~ls
Alfisols
inceptisois
Enrtisols
Ounces, Pans
inlond Water


or Rocklands


Ociober 1996



Figure 2-4. Distribution of soil orders (USDA soil taxonomy) in West Africa. Source: Eswaran
et al. (1996)










Chapter 4. In each land-use system, three on-farm plots (replicates) were chosen for soil

sampling.

Research Design

Soil samples were taken from different distances from trees. In the two parkland systems,

three horizontal distances for soil sampling were chosen:

* Near (bottom of) the tree
* 3 m (about half the crown radius) from the trunk
* 10 m from the trunk (outside of the crown)

The average size tree in each plot was selected based on aboveground inventory data as the

center of the sampling area. Soil samples were taken from four directions (north, south, west,

east) around the tree and mixed before putting in the sample bag (Figure 5-1).

Three horizontal distances for sampling live fence plots were:

* Near (bottom of) the tree
* 1 m (inside) from planted line (root influence zone)
* 3 m (inside) from planted line (outside the crown and rootzone)

Live fences are either rectangular or polygonal shapes; four sampling points on different

sides were randomly chosen. Samples away from the tree line were taken inside the fence,

because outside of the fence were often paths or borders of the cultivated land.

Since fodder bank trees are evenly planted (2m x m) inside the plots and shrubs are

randomly grown in abandoned land plots, horizontal differentiation of sampling was not taken at

these two systems. In each plot, samples were taken from four randomly selected points, and

mixed well to form the composite sample.

Sampling depths at each horizontal distance were as described in Chapter 4: 0 10 cm

(surface soil), 10 40 cm (crop root zone), and 40 100 cm (tree-root zone). This was based on












ITEM/YEAR
Material costs
Labor costs
Obtaining & planting
seeds/seedlings
Watering plants
Collecting material for dead
Constructing dead fence
Maintenance of live fence
Collecting products from live
Marketing products from live
Harvesting
ITOLLL Labor costs


0 1 2 3 4 5 6 7 8
33054 1000 1000 1000 1000 1000 1000 1000 1000



7,500
1,875
6,375 6,000 3,000
4,500 3,000 1,875
1,875 1,875 2,625 2,625 2,625 2,625 2,625 2,625 2,625
1,500 1,500 1,500 1,500 1,500 1,500 1,500
375 375 375 375 375 375 375

22,125 10,875 9,375 4,500 4,500 4,500 4,500 4,500 4,500


TCYEAL COSTS
pv cost
Revenues
Yields fromnlive fence products
Yields from fuelwood
Yields from timber

ITOLLL PEllEFUES
pv revenue
14et benefit (cash fow)

]Discount factor
Present value
11PV
IRR
BCR

Ideal accounting system
C storage (kg)
Csale (FCFA)
14et cash fow
NPV
IRR
pv revenue
BCR


55179 11875 10375 5500
55179 10326 7845 3616


5500 5500 5500
3145 2734 2378


5500 5500
2068 1798


552 27054 27054 27054 27054 27054 27054 27054




0 552 27054 27054 27054 27054 27054 27054 27054
0 480 20457 17788 15468 13451 11696 10171 8844
-55179 -11323 16679 21554 21554 21554 21554 21554 21554

1 0.86957 0.75614 0.65752 0.57175 0.49718 0.43233 0.37594 0.3269
-55179 -9846.1 12611.7 14172.1 12323.6 10716.1 9318.39 8102.95 7046.04
52802
25.5%
1.53



0 70 70 70 70 70 50 50 50
0 1617 1617 1617 1617 1617 1155 1155 1155
-55179 -9706 18296 23171 23171 23171 22709 22709 22709
60465
27.3%
0 1886.09 21679.4 18851.6 16392.7 14254.6 12195.5 10604.8 9221.57
1.60



0 35 35 35 35 35 25 25 25
-55179 -11288 16714 21589 21589 21589 21579 21579 21579
52974
25.5%
0 51C'.231 20483 17811.3 15488.1 13467.9 11706.9 10179.9 8852.12
1.53


Tonne-year accounting
C sale (FCFA)
14et cash fow
11PV
IRR
pv revenue
BCR










To protect the newly planted trees, dead fences (made from dead bush brunches) were

needed for the initial three years. Branches were obtained free of charge in the wild; no cash

costs were involved. To estimate the material cost of the dead fence, the volume of branches and

the price farmers would be willing to pay on the market were asked, and recalculated to the

standard size. The total average tool cost per farmer per year for the average size of live fence

was estimated 1,000 FCFA ($1.82) (van Dorp et al 2005). Since the fodder bank requires more

use of tools based on the data for required labor, the cost was set at 1,500 FCFA ($ 2.73) per

year.

To estimate the labor costs, farmers were asked for the average wage rate they pay for a

hired labor. The most common daily labor wage (7 hours work) was 750 FCFA ($ 1.36), ranging

between 500 and 1,000 FCFA ($ 0.91 $ 1.82) depending on the season. The respondents of the

survey were also asked if they actually hired labor to install their fodder banks. The large

maj ority of them did not; instead they used their family members including children, or

exchanged the labor with neighbors.

Labor tasks are divided into:

1. Obtaining seedlings (or seeds)

2. Planting

3. Watering

4. Collecting materials for the dead fence

5. Constructing the dead fence around the live fence for protecting the seedlings (first three
years)

6. Maintenance of the live fence/fodder bank (weeding, replanting, pruning etc.)

7. Collecting products from the live fence and the fodder bank

8. Marketing live fence products (bringing to local market and selling)

9. Harvesting the timber/fuelwood at the end of the rotation









situation. It is also a nitrogen-fixing tree, and farmers use the foliage for both organic manure

and fodder. y. paradoxa is probably the most common parkland species in the WAS, known as

Karitd (in French) or Shea (in English). Farmers use the fat extracted from the nuts in multiple

ways, such as cooking oil, medicine, and cream for dry skin. This fat, called Karitd butter or

Shea butter, has recently become popular for cosmetic use in the western world. It has a natural

UV protection and moisturizing effect, and is therefore one of the "booming" products for

international cosmetic companies. ICRAF organized a program to study the characteristics of V.

paradoxa physiology for better production and to establish a network for local farmers to market

this newly developing commodity (ProKaritd 2007).

Improved agroforestry systems

To examine the possibility of implementing reforestation/afforestation proj ects by

agroforestry under the Kyoto Protocol for C sale, it is necessary to consider agroforestry systems

with higher tree density than that of parkland systems (where crown cover is about 20 %), or

abandoned land (crown cover is close to 0 %). This is because the definition of "forest" or

afforestationn" of Kyoto Protocol normally refers to higher tree density than parkland and taller

trees than bushes in abandoned land.

In Segou region, ICRAF carried out a study to identify agroforestry needs for the WAS in

general. The study indicated an overall shortage of fodder during the dry season, and that

farmers need to protect their fields, especially during the dry season when cattle roam freely (van

Duijl 1999; Figure 2-6). To address these problems, ICRAF has been introducing the improved

agroforestry technologies such as live fences and fodder banks.

Live fence refers to planting relatively fast-growing trees in very high density around field

plots, orchards, or cultivated land. Trees are planted along plot/field boundaries at 1 m intervals

in two lines 1.5 m apart thus giving a 3-m wide thick fence around the cultivated land. Five tree/




























Daily labo r wage


Stan dard b~ coeffi cients


Figure 6-9. Regression sensitivity analysis for NPV of the fodder bank proj ect (with C sale by
the ideal accounting method). Standard b coefficients show how these input variables
are related to the results (NPV). Positive (negative) value means the input variable
positively (negatively) affect the NPV, and the absolute value represents the extent of
the influence.









Traditional Agroforestry Practices

Bush fallow/shifting cultivation

Shifting cultivation refers to the land-management practice where a period of cropping

(cropping phase) is alternated with a period in which the soil is rested (fallow phase). This

system has been traditionally practiced in the WAS, as well as other tropical and semi-tropical

regions of the world (Nair 1993). First, the clearing is done using axes or machetes and only

herbaceous plants, saplings and undergrowth are cut. When the cut material is dried and burned,

the cleared area is planted with crops like yams, sorghum, millet, maize (Zea mays), and cassava

(Manihot esculenta). The land is cultivated for one to four years after which it returns to fallow.

The regrowth of natural vegetation rejuvenates the soil through nutrient cycling, addition of litter

and suppression of weeds (Ferguson 1983).

In general, the fallow phase is much longer than the cropping phase. However, recent

rapid population growth in the WAS countries (from 2.5 to 3.0 %) requires additional cultivated

land, often at the expense of fallow and pastureland. Over the years, the fallows became greatly

reduced both in area and duration, putting in j eopardy the return of vegetative cover for the

build-up of soil fertility (Kaya 2000).

Parkland system

Another traditional land-use system, sometimes overlapped with tree-combined fallow

system, is known as the 'agroforestry parklands' system. Parklands are generally understood as

landscapes in which mature trees occur scattered in cultivated or recently fallowed fields (Boffa

1999). Farmers grow crops around and underneath of the trees (Figure 2-5). These trees are

selectively left or regenerated by farmers because of the variety of functions (mostly non-timber

use) such as food and medicine (Table 2-2). Parkland trees can also contribute to temperature

amelioration and to prevention of soil erosion (Jonsson et al. 1999). Parklands occupy a vast


















Ic ~~~r;--

:


~f~ ;
i-i
.s~--, L.~
~


-I ~~
L"


Figure 4-1. A: Location map of Mali; B: Map of Mali showing its land-locked nature: C: Map of
Segou region (The sign refers to the city of Segou).










Moody, P.W., S.A. Yo, and R.L. Aitken. 1997. Soil organic carbon, permanganate fractions, and
the chemical properties of acidic soils. Aust. J. Soil Res. 35:1301-1308.

Mosier, A., R. Wassmann, L. Verchot, J. King, and C. Palm. 2004. Methane and nitrogen oxide
fluxes in tropical agricultural soils: Sources, sinks and mechanisms. Environ. Dev. Sust.
6:11-49.

Moura-Costa, P. and C. Wilson. 2000. An equivalence factor between CO2 avoidedemissions
and sequestration description and applications in forestry. Mitig. Adapt. Strat. Glob.
Change 5:51-60.

Nair, P. K. R. and V.D. Nair. 2003. Carbon storage in North American agroforestry systems. In:
Kimble, J., Heath, L.S., Birdsey, R.A., and Lal, R. (eds). The Potential of U.S. Forest Soils
to Sequester Carbon and Mitigate the Greenhouse Effect, pp. 333-346. CRC Press LLC,
Boca Raton, FL.

Nair, P.K.R. (ed.) 1993. An Introduction to Agroforestry. Kluwer Academic Publishers,
Dordrecht, Netherlands.

Niang, A., M. Djimde, B. Kaya, E.G. Bonkoungou, and M. Macalou. 2002. Paper presented at
Improving the quantity and quality of dry season fodder availability in the Sahel. Presented
at the Regional Agroforestry Conference. Pretoria, South Africa. May 20-24, 2002.

Nierop, K.G.J., F.H. Tonneijck, B. Jansen, and J.M. Verstraten. 2007. Organic matter in volcanic
ash soils under forest and paramo along an Ecuadorian altitudinal transect. Soil Sci. Soc.
Am. J. 71:1119-1127.

Oba, N., N.C. Stenseth, and W.J. Lusigi. 2000. New perspectives on sustainable grazing
management in arid zones of sub-Saharan Africa. Bioscience 50:3 5-51.

Onim, J.F.M., M. Mathuva, K. Otieno, and H.A. Fitzhugh. 1990. Soil fertility changes and
response of maize and beans to green manures of leucaena, sesbania and pigeonpea.
Agrofor. Syst. 12:197-215.

Oren, R., D.S. Ellsworth, K.H. Johnsen, N. Phillips, B.E. Ewers, C. Maier, K.V.R Schafer, H.
McCarthy, G. Hendrey, S.G. McNulty, and G.G Katul. 2001. Soil fertility limits carbon
sequestration by forest ecosystems in a CO2-enriched atmosphere. Nature (London)
411:469-472.

Palm, C.A, T. Tomich, M. Van Noordwijk, S. Vosti, J. Alegre, J. Gockowski, and L. Verchot.
2004. Mitigating GHG emissions in the humid tropics: Case studies from the Alternatives
to Slash-and-Burn Program (ASB). Environ. Dev. Sust. 6:145-162.










Cisse, M.I. Les parcs agroforestiers au Mali. 1995. Etat des connaissances et perspectives pour
leur amelioration. Rapport AFRENA. ICRAF, Nairobi, Kenya.

Coughenour, M.B., J.E. Ellis, and R.G. Popp. 1990. Morphometric relationships and
developmental patterns of Acacia tortilis and Acacia reficiens in Southern Turkana, Kenya.
Bull. Torrey Bot. Club 117:8-17.

Dai, A., P.J. Lamb, K.E. Trenberth, M. Hulme, P.D. Jones, and P. Xie. 2004. The recent Sahel
drought is real. Int. J. Climat. 24:1323-1331.

Dalsted, N.L. and P.H. Gutierrez. 2007. Partial budgeting. [Online]. Available at
http ://www.ext.colostate.edu/PUB S/farmmgt/03 760.html (verified 5 Jul. 2007). Colorado
State University Extension-Agriculture, Fort Collins, CO.

De Alwis, K.A. Recapitalization of soil productivity in sub-Saharan Africa. 1996. FAO
Investment Center, Rome, Italy.

De Jong, B.H., S.O. Gaona, S.Q. Montalvo, E.E. Bazan, and N.P. Hemnandez. 2004. Economics
of agroforestry carbon sequestration: A case study from southern Mexico. Chap.4. In
J.R.R. Alavalapati and D.E. Mercer (ed.) Valuing agroforestry systems. Kluwer Academic
Publishers, Netherlands.

De Jong, B.H.J. 2001. Uncertainties in estimating the potential for carbon mitigation of forest
management. For. Ecol. Manage. 154:85-104.

Del Galdo, I., J. Six, A. Peressotti, and M.F. Cotrufo. 2003. Assessing the impact of land-use
change on soil C sequestration in agricultural soils by means of organic matter
fractionation and stable C isotopes. Glob. Chan. Bio. 9:1204-1213.

Delaney, M. and J. Roshetko. 1999. Field test of carbon monitoring methods for home gardens in
Indonesia p.45-5 1. hz Field Tests of Carbon Monitoring Methods in Forestry Proj ects.
Forest Carbon Monitoring Program. Winrock Intemnational, Arlington, VA.

Denton, F., Y. Sokona, and J.P. Thomas. Climate change and sustainable development strategies
in the making: What should West African countries expect? 2001. OECD Climate Change
and Development. Environnement et Developpement du Tiers Monde (ENDA-TM),
Dakar, Senegal.

Dixon, R.K. 1995. Agroforestry system: sources or sinks of greenhouse gases? Agrofor. Syst.
31:99-116.

Dixon, R.K., S Brown, R.A. Houghton, M.C. Solomon, M.C. Trexler, and J. Wisniewski. 1994a
Carbon pools and flux of global forest ecosystems. Science (Washington, DC) 263:185-
190.

Dixon, R.K., J.K. Winjum, K.J. Andrasko, J.J. Lee, and P.E. Schroeder. 1994b. Integrated land-
use systems: assessment of promising agroforest and alternative land-use practices to
enhance carbon conservation and sequestration. Clim. Change 27:71-92.









relative to the emissions that are supposed to be offset may be problematic, since there may be a

time lag of years to decades between the establishment of the offset proj ect and actual uptake of

the C.

Cumulative C storage

The cumulative storage approach is based on an understanding of C cycle dynamics and

radiative forcing of the atmosphere. The total warming effect of a given emission is determined

by the cumulative presence of GHG in the atmosphere; in other words the product of

concentration and time. In the case of CO2, terrestrial and oceanic sinks take up C previously

emitted, over time. Assuming the dynamics of the C cycle remain stable, most CO2 emitted at

the present will be absorbed within 100 years, and the cumulative radiative forcing produced by

this emission will be proportional to the area under the depletion curve, expressed in tC.years.

Calculation of this area provides an estimate of the cumulative C storage that would be required

to offset an emission of 1 tC at the present time. This method avoids both the necessity of

making questionable assumptions about the long-term balance of C in forests/agroforests, and

the practical difficulties of implementing flux-based incentive systems. However, international

agreements on the conversion factor for tC.years per tC emission and the time limit for crediting

the effect of a given proj ect are required (Tipper and De Jong 1998).

Other accounting methods

In addition to these relatively simple conventional methods, alternative approaches have

been proposed to better address the temporal dimension of C storage, such as equivalence-

adjusted average storage, stock change crediting with ton-year liability adjustment, equivalence-

factor yearly crediting, equivalence-delayed full crediting, and ex-ante ton-year crediting (De

Jong 2001). Most of these are based on adopting a two-dimensional measurement unit that

reflects storage and time, i.e., ton-C year. The general concept of the ton-year approach is in the









A conservative estimate of soil C sequestration potential through addition of organic

matter such as plant litter and animal waste to these agricultural soils is in the range of 0. 1 0.3

Mg C ha-' yr' (Batj es 2004). ICRAF is trying to promote live fences and fodder banks for land

amelioration and counts C sequestration potential as one of the potential benefits. However, it is

important to address the possibility of causing net loss of soil C while converting abandoned land

into live fences or fodder banks in this study region at the initial stage, because of the tillage

factor at establishment. And, because the subsequent land-use practices provide low levels of

litter input, especially in fodder banks, it may take a long time to regain the initial loss of soil C.















U n prote ctecd
,soil C
Litter quality


I


Ph ysicallIy
protected
soil C


BiochemicallSy
protected
soil C


Microaggregate- Silt- and clay-
associated soil C associated soil C

Condensation/complexation

N on- hyd ro lyza ble
soil C


Figure 5-3. Model of soil organic matter dynamics
Source: Six et al. 2002. Figure 3 in page 163.












y = 0.64 x + 0.26
R2 = 0.72


3

o
v,

r
o
w


I FA
O VP
A LF
X FB
m AL


0 1


2 3
C3 plant origin C (g C


4
kgl soil)


5 6


Figure 5-10. Linear regression between C derived from C3 plants and C in the large soil
particles (250 2000 Cpm) at 0 10 cm soil depth across five land-use systems of
Segou region, Mali.





























Prclject year (O 25)
A-Mean,*1-1sD II -951u

Figure 6-3. Simulated net benefit (total costs total revenues in each year) of the live fence
project (without C sale). Mean value of each year' s probability distribution is shown
in the yellow line, red range is plus minus 1 standard deviation from the mean, and
green range is the 5 to 95 % likelihood of the value.




























141










crops present, tree density, age, land-use history) were chosen. Villages for two parkland plots

were chosen because of relatively mature and uniform F. albida and y. paradoxa trees (Figure 4-

2, 4-3). Each plot was set inside a different farm. The village for live fence plots was chosen

because it has a group of farmers who participated in ICRAF's live fence program. Three

farmers' live fences with the same age and similar tree growth were selected (Figure 4-4).

Fodder bank plots were more difficult to find. Since fewer fodder banks were adopted and

maintained than live fences, only three comparable fodder banks were found in three different

villages (Figure 4-5). The village for abandoned land plots was chosen near the vast degraded

land spreading east of Segou city. All abandoned land plots were previously cultivated by

farmers (Figure 4-6).

Data Collection

Field data collection was conducted from August to September 2005.

Biomass measurement

The plot size was I ha for parkland systems, while it was the 'whole site' for improved

systems (about 0.25 ha or less), and 0.5 ha for abandoned land (Table 4-1).

Data recorded for aboveground biomass were:

* Species and number of trees in each plot
* Diameter at breast height (DBH) and/or diameter at the ground of each tree/bush
* Tree/bush height
* Crown size for bushes in abandoned land.

Regarding land-use history, age of traditional parkland and abandoned land systems were

difficult to estimate. According to owners of the plots, all parkland plots were at least 35 years

old, and the abandoned land plots had been "abandoned" for less than 10 years. All three live

fence plots were 8 years old (at the time of data sampling they were established in 1997), and

two of the three fodder bank plots were 9 years and one was 6 years old.










Seghieri, J., M. Simier, A. Mahamane, P. Hiernaux, and S. Rambal. 2005. Adaptative above-
ground biomass, stand density and leaf water potential to droughts and clearing in Guiera
senegalensis, a dominant shrub in Sahelian fallows (Niger). J. Trop. Ecol. 21:203-213.

Sharrow, S.H. and S. Ismail. 2004. Carbon and nitrogen storage in agroforests, tree plantations,
and pastures in western Oregon, USA. Agrofor. Syst. 60:123-130.

Shepherd, D. and F. Montagnini. 2001. Above ground carbon sequestration potential in mixed
and pure tree plantation in the humid tropics. J. Trop. For. Sci. 13:450-459.

Six, J., R.T. Conant, E.A. Paul, and K. Paustian. 2002a. Stabilization mechanisms of soil organic
matter: Implications for C-saturation of soils. Plant Soil 241:155-176.

Six, J., P. Callewaert, S. Lenders, S. De Gryze, S.J. Morris, E.G. Gregorich, E.A. Paul, and K.
Paustian. 2002b. Measuring and understanding carbon storage in afforested soils by
physical fractionation. Soil Sci. Soc. Am. J. 66:1981-1987.

Six, J., R. Merckx, K. Kimpe, K. Paustian, and E.T. Elliott. 2000c. A re-evaluation of the
enriched labile soil organic matter fraction. Eur. J. Soil Sci. 51: 283-293.

Six, J., E.T. Elliott, K. Paustian, and J.W. Doran. 1998. Aggregation and soil organic matter
accumulation in cultivated and native grassland soils. Soil Sci. Soc. Am. J. 62:1367-1377.

Smith, P., D.S. Powlson, M.J. Glending, and J.U. Smith. 1998. Preliminary estimates of the
potential for carbon mitigation in European soils through no-till farming. Glob. Change
Bio. 4: 679-685.

Stainback, A.G. and J.R.R. Alavalapati. 2002. Economic analysis of slash pine forest carbon
sequestration in the southern U.S. J. For. Econ. 8:105-117.

Sun, O.J., J. Campbell, B.E. Law, and V. Wolf. 2004. Dynamics of carbon stocks in soils and
detritus across chronosequences of different forest types in the Pacific Northwest, USA.
Glob. Chan. Biol.10:1470-1481.

Tan, Z.X., R. Lal, N.E. Smeck, and F.G. Calhoun. 2004. Relationships between surface soil
organic carbon pool and site variables. Geoderma 121: 187-195.

Tiepolo, G., M. Calmon, and A.R. Feretti. 2002. Paper presented at measuring and monitoring
carbon stocks at the Guaraquecaba Climate Action Proj ect, Parana, Brazil. Proceedings of
the International Symposium on Forest Carbon Sequestration and Monitoring. 11-15
November 2002, Taipei, Taiwan. Taiwan Forestry Research Institute Winrock
International .

Tiessen, H., H.K. Hauffe, and A.R. Mermut. 1991. Deposition of Harmattan dust and its
influence on base saturation of soils in northern Ghana. Geoderma 49:285-299.

Tieszen, L.L. 1991. Natural variations in the carbon isotope values of plants: implications for
archaeology, ecology and paleoecology. J. Archaeol. Sci. 18: 227.









the expectation that the amount of C content would differ by depth class depending on the

presence or absence of tree roots and tillage.

In summary, the numbers of soil samples are:

2 (Parkland systems) x 3 (horizontal dist.) x 3 (depth) x 3 (replicates) = 54
1 (Live fence) x 3 (horizontal dist.) x 3 (depth) x 3 (replicates) = 27
1 (Fodder bank) x 1 (horizontal dist.) x 3 (depth) x 3 (replicates) = 9
1 (Abandoned land) x 1(horizontal dist.) x 3(depth) x 3(replicates) = 9
Total = 99

Soil Preparation and Analyses

Samples were all air-dried and passed through a 2 mm sieve (except samples for bulk

density measurement) at the field station in Segou. There is no visible O horizon or surface

litter, and therefore no analysis was done for that layer (Woomer et al. 2004) (Figure 5-2). Soil

samples were brought back from Mali to University of Florida in October 2005 for analysis.

Samples for bulk density measurement for each depth class were separately collected at

each plot with a 100 cm3 cylinder. Wet weight and air-dry weight were measured in the field.

Samples were oven-dried and analyzed for particle size distribution (USDA Soil Survey Lab

Method) and pH at the University of Florida, Soil and Water Science Department laboratory.

Sub-samples were taken from the 99 air dried samples and ground. Soil C content (g C kg' soil)

of the sub-samples was measured by the dry combustion method on an automated Flash EA 1 112

NC elemental analyzer (Thermo Fisher Scientific, Inc.).

Soil fractionation

Soil samples were fractionated into three aggregate size classes (2000 250 Cpm, 250 53

Cpm, and <53 Cpm) by wet sieving, following the method of Six et al. (2002). A sub-sample of

100 g of the composite soil sample was submerged in deionized water as disruptive forces of

slaking for about 5 minutes prior to placing it on top of a 250 Cpm sieve. The sieving was done

manually by moving the sieve up and down approximately 50 times in 2 minutes. The fraction









soil) of sand, silt and clay were compared statistically over the five land-use systems.

Abandoned land had lower sand content (530 694 g kg-l soil) and higher content of silt and

clay (306 470 g kg-' soil) than the other four systems (sand: 715 935 g kg-' soil, silt + clay:

65 285 g kg-l soil) (Table 5-1). Soils of the four systems in each depth class were not different

in the particle size content.

Whole Soil C

Whole soil C content across all the systems varied from 1 6 g C kg-l soil (Figure 5-4).

Two parklands and live fence had three sampling locations (0, 3, and 10 m from tree base for

parklands; 0, 1, and 3 m from tree lines for live fence). Only the surface soil (0 10 cm) of live

fence showed a difference between the "near tree" and the two zones more distant from the tree,

but other depth classes and two parklands plots did not show difference by horizontal distance

from trees, although the trend of "near tree" > "outside crown" was observed in the surface soil

of both parkland systems. C content decreased with soil depth for all land-use systems except

the fodder bank, where the surface soil (0 10 cm) had less whole soil C than lower depths.

The whole soil C data of five land-use systems were compared statistically using two

factorial (land-use and depth) ANOVA (model 1). Both land-use and depth factors, as well as

land-use~depth interaction were significant for that variables. By Tukey-Kramer multiple

comparison test, abandoned land had higher C content than the other four systems, but the other

four systems were not different from each other. C content was different by depth: 0 10 cm >

10 40 cm > 40 100 cm across all treatments.

Among the data for the two parklands and live fence plots, horizontal distance was another

"factor" (model 2). Land-use and depth factors were significant, but distance was not

significant, although showing the trend (p=0.0884), in the three-way factorial (land-use, depth,

and distance) ANOVA. When land-use factor and distance factor were examined in the fixed









soil amelioration (Breman 1997; Tschakert 2007). The World Agroforestry Centre (ICRAF)

announced that they were confident that establishment of agroforestry, especially in degraded

land, would qualify and play an important role under the Clean Development Mechanism (CDM)

of Kyoto Protocol (ICRAF 2007).

A few maj or problems exist, however, in the context of promoting agroforestry practices

by smallholders for entering CDM market. For example, we need to examine cost per unit of C

sequestration since there are many other options such as emission reduction or sequestration by

large-scale monoculture plantation (De Jong et al. 2004). Furthermore, based on the feasibility

studies, appropriate technical and political assistance should be provided so that smallholder of

agroforestry owners will not always be less competitive than other mitigation proj ects options.

Thus, socioeconomic feasibility of the improved systems is an important consideration in the

context of agroforestry implementation for GHG mitigation, and that is the scope of this chapter.

Under the Kyoto Protocol, only newly sequestered C as a result of the proj ect is

recognized as tradablee" C; the traditional agroforestry systems such as parklands are, thus, not

likely to be counted as C sequestration projects. The improved agroforestry systems that are

currently being introduced in the study region can be qualified for tradable C credits. Therefore,

the target agroforestry systems for this study are live fence and fodder bank in Segou region,

Mali (detailed system description in Chapter 4).

Research Questions

1. What is the relative attractiveness of the two improved agroforestry systems (live fence and
fodder bank) considering their C sequestration potential, economic profitability, and social
acceptability?

2. If C credit markets were introduced under the CDM of Kyoto Protocol, would adoption of
agroforestry provide more profits to land owners? If yes, how much?




































02007 Asako Takimoto












TABLE OF CONTENTS


page

ACKNOWLEDGMENT S .............. ...............4.....


LIST OF TABLES ........._.._ _..... ._ ._ ...............10....


LIST OF FIGURES ........._.. ..... ._ ._ ...............11....


AB S TRAC T ............._. .......... ..............._ 13...


CHAPTER


1 INTRODUCTION ................. ...............15......... .....


Back ground ..........._...__.......... ...............15.....
Rationale and Significance .............. ...............16....
Research Questions and Obj ectives ................. ...............17......... ....
Dissertation Overview ................ ...............18.................


2 THE WEST AFRICAN SAHEL: GENERAL LAND-USE AND AGROFORESTRY ........20


Description of the Region ................. ...............20................
Clim ate .............. ...............20....
Vegetati on ................. ...............21.................
Soil ................. ...... ....... ... ......... .... .... ...........22
Traditional Farming Systems and Agroforestry in the WAS .............. ....................2
Traditional Agroforestry Practices .............. ...............25....
Bush fallow/shifting cultivation ..........._..._ ......_._ ....._._ ...........2
Parkland system............... ...............25.
Improved Agroforestry Practices .............. ...............26....

3 LITERATURE REVIEW: CARBON SEQUESTRATION POTENTIAL OF
AGROFORESTRY SYSTEMS IN THE WEST AFRICAN SAHEL (WAS)..............._.._. ...36

Overvi ew ........._._.. ... .... ._._......... ._._..... .... .. ... ........3

C Sequestration as a Climate-Change-Mitigation Activity ........._._.... ......_._........36
Agroforestry for C sequestration .........._... ... ...............37_._......
Methodologies for C Sequestration Measurements ....._._._ ..... ... .__ ......_._..........3
Direct On-site Measurement............... ..............3
Inventory ............... .... ...............40.
Conversion and estimation ........._._ ...... .__ ...............40...
Indirect Remote Sensing Techniques .............. ...............42....
M odeling.................. ........... .. .... .......4
Default Values for Land/Activity Based Practices ....._._._ ..... ... .__ ........_.......44
Accounting Methods ........._._ ...... .... ...............44...









remaining on the top of the sieve was collected in a hard plastic pan, oven-dried at 65 oC and

weighed. Water plus soil <250 Cpm were poured through a 53 Cpm sieve and the same sieving

procedure repeated. The recovery of mass soil fractions after overall wet sieving procedure

ranged from 97 to 99% of the initial soil mass. Sub-samples for each soil fraction (99 samples x

3 fraction size = 297 samples) were then ground, and C contents were measured by the same dry

combustion instrument as described for whole soil C. Soil C in the large (L) fraction (2000 -

250 Cpm) contains fairly new coarse/fine particulate organic matter (POM) C, although there is

other forms of "protected" C not fully separated by wet sieving. The medium (M) fraction (250

- 53 Cpm) contains both less protected C (within fine POM) and protected C (microaggregate

protected POM C and silt + clay associated C). C in the small (S) fraction (<53 Cpm) contains the

protected form (silt + clay protected C or non-hydrolysable C), although there are less stable

forms of C in the size class, too (Six et al. 1998, Six et al. 2000). Unprotected C involves the

youngest form of SOM, and through the process of aggregate formation, adsorption/desorption,

and condensation/complexation, soil C becomes older and more stable (SOM dynamic model

Figure 5-3).

C isotopic ratio (13C/12C) measurement

613C ValUeS of soil samples (whole soil and fractionated soil) were measured by Thermo

Finnigan MAT Delta Plus XL mass spectrometer (Thermo Fisher Scientific, Inc.). C isotope

ratios are presented in 8-notation:

613C = [(RSample RStd)/RStd] x 103 (Eq. 5-1)

Where RSample is the 13/ 12C ratio of the sample, and RStd is the 13/ 12C ratio of the Vienna

Pee Dee Belemnite (VPDB) standard.

Relative proportions of soil C derived from C4 plants material versus C3 plants material

was estimated by mass balance (Balesdent and Mariotti 1996):









When PV benefits equals to PV costs, the BCR is 1, and NPV is 0. Also, if PV benefits

exceed PV costs, BCR must be greater than 1, and if PV costs exceed PV revenues, BCR<1.

Thus, according to the BCR decision-rule, proj ects are acceptable when the BCR is 1 or greater,

and unacceptable if BCR<1 (Klemperer 1996).

The IRR of the proj ect is the discount rate at which the NPV becomes "O" in the NPV

formula (Eq. 6-1).

B C,
= (Eq. 6-3)
(1 +1RR)Y = (1+1IRR )

The IRR is the rate of return earned on funds invested in a proj ect. The equation 6-3 also

says that the IRR is the interest rate at which PV benefits equals PV costs. A project is

acceptable if its IRR is equal to or greater than the minimum acceptable rate of return

(Klemperer 1996). In this study's case, however, the farmers will not have a specific acceptable

rate of return. So, the rate can be recognized as "acceptable" if it is greater than the interest rate

(when farmers take a loan from a local bank or financial institution).

Several basic budgets were available for calculating the above three decision-rules, such as

whole-farm budget, enterprise budget, partial budget and cash flow budget. Each budget is

specific in its application, and the partial budgets were used in this study. Partial budget is used

to evaluate the economic effect of minor adjustments in some portion of the business. Many

changes that do not require a complete reorganization are possible in a farming business. Given

a fixed set of resources, a farmer can employ these resources in more than one way in response

to changes in product price levels, cropping patterns, or carrying capacity. Partial budgets are

useful to evaluate changes such as expanding an enterprise (e.g. a crop), alternative enterprise,

and different production practices (Dalsted and Gutierrez 2007). Because introducing a live

fence and/or a fodder bank has limited impacts on the costs and returns of a farmer' s budget due









inventory. How to calculate/determine the amount of "sequestered" C over a certain period is

another issue, and discussed in the "Accounting Methods" section.

Inventory

In general, C in forest or agroforestry systems can be divided into four groups; 1)

Aboveground biomass, 2) Belowground biomass, 3) Soil C, 4) Litter fall/crop residue. Methods

to collect and calculate the sample data from proj ect sites have been standardized by many

reports and studies (MacDicken 1997; Roshetko et al. 2002).

Data for the four C categories are collected by timber cruising and sampling of herbaceous

vegetation, soil, and standing litter crop at sample plots (Shepherd and Montagnini 2001; Brown

2002; Tiepolo et al. 2002). Also, for existing forests, many tropical countries have at least one

inventory of all or part of their forest area that could be applied for agroforestry systems,

although many of the inventories are more than 10 years old and very few have repeated

inventories. Data from these inventories can be converted to biomass C depending on the level

of detail reported (Brown, 1997).

Conversion and estimation

For aboveground biomass, trees are divided by compartments: leaves, branches and trunks,

and measured in dry weight (Beer et al. 1990), because each compartment has unique C content

and decomposition rate. Although this is the most accurate method, these inventories are often

too time-consuming and costly.

Alternatively, biomass expansion factors or allometric biomass equations are often used,

because they require only stem wood information such as diameter at breast height (DBH).

These equations exist for practically all forests types of the world, especially in the temperate

zone (Sharrow and Ismail 2004). But, because of the very general nature of these equations, they

lack accuracy; they are, at best, approximations. For an agroforestry system, Shroeder (1994)









The UNFCCC guidelines suggest using these equations when no local species allometric

equations are available.

A second option is to use the equations developed from Acacia tortillas and Acacia

ruficiens in Northwest Kenya (Coughenour et al. 1990). These are:

Loglo (g mass) -2.26+3.98 loglo (mm stem diameter) (Eq. 4-3)
R2 =0.98 (stem diameter<15.7 mm)

Log to (g mass)- -0.68 +2.66 loglo (mm stem diameter) (Eq. 4-4)
R2 =0.98 (stem diameter>15.7 mm)


Although Northwest Kenya is not in the WAS, its climatic condition is much more similar

to that of the study area than to the area where the UNFCCC guideline's equation was developed.

Gonzalez (2001) used these Acacia spp. equations in his research at various parts of Senegal.

These two options were both tried in this study, and the results of estimated biomass C are

compared later.

For abandoned land plots, equations for Guiera senegalensis used by Seghieri et al (2005)

in Niger were adopted. G. senegalensis is the most dominant shrub species in the abandoned

land plots, and the equation was originally developed in fallows of Mali (Cisse 1980, Franklin

and Hiernaux 1991).

Foliage mass of each stem of each shrub: Blstem (g),
Basal circumference of the stem: Cstem (cm):
Blstem =1.09xCstem (all stems of n=20 shrubs, R2 =0. 82, P<0. 001) (Eq. 4-5)
Stem wood dry mass: Bwstem (kg):
Bwstem =0.0037xCstem (36 stems among n=15 shrubs, R2 =0.90, P<0.001) (Eq. 4-6)
Leaf and wood masses (Blstem and Bwstem) were then aggregated for each multi-stemmed shrub.

To calculate the amount of C in the biomass, C fraction rate of 0.5 is suggested in the

UNFCCC guideline. Belowground biomass is also estimated by using the suggested root/shoot









When the components of revenues were examined, a big difference between the two

proj ects' cash flows was the revenue from the saved time of the fodder bank. Before the fodder

bank installation, farmers had to graze animals almost everyday for quite a long time during the

dry season. Since the survey data showed that this grazing time/labor a farmer could save by the

fodder bank was considerably high, the revenue of this component became the significant

difference between the live fence and the fodder bank. Other revenue components, timber and

fuelwood sale occurred at the end of the proj ect year, thus, the revenues from them were

discounted largely when converted to the present values. Another revenue component, C sale,

was added to the cash flow with the ideal accounting method and the tonne-year method

separately. The amount paid in US$ was converted to FCFA, and put into the cash flows.

Three decision rules (NPV, BCR, and IRR) for three different conditions (No C sale, C

sale with the ideal accounting method, and C sale with the tonne-year accounting method) were

calculated (Table 6-2). C sale by the ideal accounting method significantly changed all three

decision rules. NPV of the live fence was 52,802 FCFA ($ 96.0) without C sale, and it increased

to 60,465 FCFA ($ 109.9) with C sale by the ideal accounting method. NPV of the fodder bank

was 87,319 FCFA ($ 158.8) without C sale, and 96,394 FCFA ($179.3) with C sale by the ideal

accounting method. BCR and IRR also increased (economically more profitable) with C sale.

However, C sale by the tonne-year method did not increase the three decision rules much from

those without C sale. In NPV, only 172 FCFA ($0.3) increase for the live fence and 204 FCFA

($0.3) increase for the fodder bank were observed, compared with NPV without C sale. The C

sale profits by the tonne-year method were too small to make a change of BCR and IRR values

for both live fence and fodder bank proj ects.










(Richter et al. 1999; Schuur et al. 2001). And, soil C studies in agricultural croplands have

mostly been in the context of soil productivity management (Phillips et al. 1993; Beare et

al.1994; Alvalez et al. 1995; Blair et al. 1995; Franzluebbers et al. 1995; Rhoton 2000), although

recently more and more soil C studies are considering agricultural soil as C sinks, a mechanism

which removes CO2 from the atmosphere (Smith et al. 1998, Duiker and Lal 1999, Lal 2004b)

Soil C studies in agroforestry systems have been few. Interactions between crops and trees

and the relatively short-term rotation of land management make such studies more complicated

and challenging compared to single-species agricultural and forestry systems. Existing studies in

agroforestry systems measure soil organic matter (SOM) content with other soil nutrients (Kang

et al. 1999; Makumba et al. 2006). Those studies discussed whether trees have positive (e.g.,

nitrogen fixing) or negative (e.g., competition for light, nutrients, or water) impacts on crop

production. In the WAS, parkland trees were found to increase soil C around trees (Jonsson et

al. 1999), and an improved fallow system (planting Gliricida sepium during the non-cropping

phase) was found to increase soil C on the surface compared with natural grass fallow (Kaya and

Nair 2001). These studies support the expectation that agroforestry systems would enhance soil

C sequestration, but there is still little information about trees' influence on C in deeper soil and

stability of various forms of soil C sequestered by trees.

In this study, organic C in soil is assumed to be "protected" from further decomposition in

three ways: 1. physically stabilized, or protected through microaggregation (microaggregate-

associated soil C), 2. intimate association with silt and clay particles (silt- and clay- associated

soil C), and 3. biochemically stabilized through the formation of recalcitrant soil organic matter

compounds (non-hydrolysable soil C) (Six et al. 2002). There are some other ways such as Al or

Fe- SOM complexes, C accumulation resulting from anaerobic conditions, and transfer to









Biomass C stock

Since this study examines the potential of C sales under the Kyoto Protocol's Clean

Development Mechanism (CDM), methodologies used here for estimating biomass C stock are

based on the guideline published by UNFCCC (2006). Although species-specific allometric

equations are ideal, none was available for parkland trees from the study region. As an

alternative, the UNFCCC guideline recommends using the following general equations from

FAO (1997).

Tree biomass (kg) = exp (-2.134 + (2.530 InDBH(cm))) (n = 191, R2 = 0.97) (Eq. 4-1)

In the FAO (1997) paper, there are general equations suggested for drylands. However,

those equations were developed from much smaller sets of trees in India and Mexico, and their

DBH ranges are 3 to 30 cm. E. albida and y. paradoxa trees in parkland plots of this study

greatly exceed the diameter range of these general equations. The average DBH of trees in the

plots were about 59 cm for E. albida and 42 cm for y. paradoxa. Using the dryland general

equations may cause significant over estimation of the biomass. Thus, this study follows a

method proposed by Woomer et al. (2004a) in Senegal, using Equation 4-1. This equation is

also from FAO (1997) for higher rainfall area (1500 4000 mm), but covers the diameter ranges

of E albida and y. paradoxa trees in this study.

There are two options for estimating the biomass of the five live fence and one fodder bank

species. One is following UNFCCC's guideline, using a general equation for areas with <900

mm annual rainfall. The tree sizes are within this equation's DBH limits (3 30 cm). The

equation is:

Tree biomass (kg) = 10 ^'(-0.535 + loglo(nxDBH(cm)2/4)) (R2 = 0.94) (Eq. 4-2)









On the other hand, introducing live fences and/or fodder banks into cultivated land or

abandoned land can sequester C by increasing the tree biomass, but the extent would vary largely

by the baseline and accounting method used. Biomass C sequestered by live fence planting is a

straightforward "addition" in the C equation for the site, since they are normally established on

the tree-less cultivated land (the baseline is nearly 0). The potential of fodder banks, however,

depends on initial plot condition. As in the situation for live fence, when the cultivated land is

converted to fodder bank, most of C sequestered by the fodder trees can be counted. However,

establishing fodder banks on abandoned land, as ICRAF or local government is trying to

promote, may actually result in net loss of C stock during the initial stage, because the biomass

from bushes and grasses in the abandoned land has to be removed at the time of establishment,

and it may take years for fodder trees to accumulate an amount equal to the original biomass.

Further investigations are needed on temporal C dynamics of these systems.

Soil C is not considered in the calculations of the Kyoto Protocol for its first commitment

period (2008-2012). When, rather than if, soil C is taken into account, determining baseline soil

C will be another challenge to determine and compare the C sequestration potentials of land-use

systems. Results suggest that soil-sampling depth makes a large difference in estimating the

amount of C stored per area basis, as well as the potential for C sequestration (Table 4-4). To

compare and discuss the C sequestration potential of different land-use or different ecoregions, it

will be very important to standardize sampling depth. Several studies in Africa reported that

planting trees for C sequestration will not immediately retain soil C equal to the baseline level

nor increase it in the short term (Kaya and Nair 2001; Walker and Desanker 2004). Introducing

live fences or fodder banks may increase the biomass C in the system, but may not increase soil

C. Soil C sequestration potential will be discussed in more detail in the next chapter.









and C accounting methods are very important; but they are also often controversial issues at

international negotiations of climate change.

Approaches to Assessing C Sequestration Performance

Fluxes of C and flow summation

Balancing the annual flux from a source of emissions by uptake in a forestry/agroforestry

proj ect is conceptually the simplest way of providing offsets. In this approach, offsets are

'delivered' to the C credit market on an annual basis, according to the emissions avoided,

relative to the proj ect baseline. However, since fluxes of C associated with forestry/agroforestry

are irregular, it may be difficult to match the uptake by a particular project to an industrial source

of emissions.

Furthermore, national or international authorities must assign permanent C storage status

to proj ect areas, such that the owners are liable for any emissions, as well as eligible for credits.

Without such status C might be accumulated in the growing phase of the forestry cycle, only to

be lost at the end of rotation (Tipper and De Jong 1998).

Average changes in the stocks of C

The pilot phase of most sequestration proj ects is assessed on the basis of the long-term

average increase in the stocks of terrestrial C relative to the baseline (Kursten and Burschel

1993; van Noordwijk et al. 2002), expressed as tC according to the equation: Average net C

storage (tC) = C(C stored in project C stored in baseline) in tC / n (years)

The stock change method calculates the difference in C stocks between a proj ect and its

baseline at a given point in time. A key advantage of both methods is that it focuses on the

sustainability of changing the stock of C stored in vegetation and soils. However, long-term C

storage is not easily defined, and there may be considerable argument over the assumptions

about risks and possible future changes in management. The timing of the emission reduction










(An. 9. 10. si possible)
Etape 3: Vente des products de les banques forrageres
Personnes No. de P/J No. de P/J No. de P/J No. de P/J No. de P/J No. de P/J
impliquees An 3: An 4: An 5: An 6: An 7: An 8:
(H/F/E)
G. Sepium:






Note :




1.12 Quelle est la distance du champ jusqu'au marched ou vous vendez les products de les
banqlues forrageres (et la haie vive, si quelauechose) ?


1.13 Quels sont les moyens de transport utilises pour vendre les products des banques
forrageres au marched?
Note: Noter tous les moyens de transport utilises pour la vente des products de les banques
forrageres, PAS les moyens de transport uniquement utilises pour le travail champ~tre agricole.
Moyen de Nombre Prix (en CFA) Montant total Nombre
transport (entrer d' annees
codes): d'utilisation


Nom du marched


Distance (entrer codes)


1: 0-5 km
2: 5-10 km
3: 10-20 km
4: 20-50 km
5: >50 km
9: Pas de reponse


1: Charrette
2: Bicyclette/mobylette
3: Vehicule
4: Autres
9: Pas de reponse






















Figure 4-2. Faidherbia albida parkland in Togo village. The tree leaves are shed at the
beginning of the rainy season; but they return at the beginning of the dry rainy season.
The understory crop is pearl millet (Pennisetum glaucum). (Photographed by author)


Figure 4-3. Vitellaria paradoxa~11~~1~~11~ parkland in Dakala village. The trees have wide canopies, and
leaves remain during the rainy season. Farmers plant crops (in this photo, pearl
millet) beneath the trees, often very close to the trunk. (Photographed by author)


"~-~-~-~-~14

..
;1~.









Ideal accounting method: In this method, payments for C sequestration occur as the service

is provided and a debit occurs when C is released (i.e. by fire or harvest). Farmers annually

receive the payment according to the amount of C sequestered in their proj ect' s fields (live

fence/fodder bank). The full debit at harvest means that the total amount of C credits sale

received during the life of the proj ect (live fence and/or fodder bank) are paid back to the

investor by farmers.

Tonne-year accounting method: Although the ideal accounting system is "ideal" for the

land owners (i.e. farmers), it is risky for investors because they are not sure the proj ect will last

until the end of the rotational age. The tonne-year method does not require redemption of C

credits upon harvest, because the payment occurs based only on the 'equivalent' amount of

permanently avoided emissions during a given year (Moura-Costa and Wilson 2000). This

method has the advantage that no guarantee is needed if the proj ect will last a required number of

years, as the annual payments are adjusted by the equivalent factor. This is a more favorable

method for the investors, and politically popular (Hardner et al 2000). In this study, the

equivalent factor of 0.0215 (Cacho et al. 2003a) was used. Farmers annually receive C credit

payment only equivalent to "the amount of C sequestered in each year 0.0215", but there is no

payment back to the investors at the end of the proj ect.

These two methods were separately incorporated into the cash flow of both a live fence

project and a fodder bank project (see Appendix B, C). The decision rules (NPV, BCR, and

IRR) were calculated in three different cash flows: 1) Cash flow without C sale, 2) Cash flow

with C sale (ideal accounting), and 3) Cash flow with C sale (tonne-year accounting).

Sensitivity analysis

The calculation of NPV, BCR, or IRR in the CBA described above is based on the best

guess scenario where all variables of costs and benefits included in a cash flow are "most likely"









Soil sampling

Based on discussions with ICRAF researchers, three depth classes were determined for soil

sampling: 0 10 cm (surface soil), 10 40 cm (crop-root zone), and 40 100 cm (tree-root

zone). The average size tree in each plot was selected based on aboveground inventory data as

the center of the soil sampling area. Samples were taken from three horizontal distances from

trees in the two parkland systems and live fence system. At each horizontal distance, samples

were taken from four different points using an auger, and samples from each depth from these

points were well mixed as a composite sample before transferring them into bags. For the fodder

bank, where the bush/tree density was fairly uniform, and the abandoned land, plots, four random

points were chosen to make a composite sample of each depth. More details of soil sampling are

described in Chapter 5. Sampling for bulk density measurements were taken separately for each

depth and land-use using a 100 cm3 stainless steel cylinder. A soil pit (1 m depth) was made for

each land-use plot, and the cylinder was horizontally driven to the center of each depth class to

take the samples for bulk density determination. All samples (total 144 samples: 99 composite

samples and 45 bulk density samples) were air-dried and shipped to University of Florida for

analyses.

Carbon Stock Estimation

Amount of biomass C and soil C (C stock) were estimated respectively, as follows. Total

C stock (Mg C /ha) of each land-use system was calculated by adding biomass C stock and soil C

stock of each plot of each land-use system (all data on per ha basis). Live fences are

conventionally expressed in terms of length of rows. In this study, the "area" under live fence is

calculated based on 3 m width; but in practical terms, the area of the field "serviced" by the live

fence is important. Since live fences are along plot/field boundaries of unequal sizes, it is not

realistic to assign a standard row length per unit area (ha) of plot/field.









Twenty-two farmers from 13 different villages in the Segou region were interviewed (the

live fence survey was conducted on 18 owners from 15 different villages). The language used

was Bambara, the most common local language in Mali, although the questionnaire was made in

French. A translator (French Bambara), an agronomist who had conducted social survey for

ICRAF, was hired to communicate with interviewees (Figure 6-1). To ensure his

survey/interviewing skill, an experienced ICRAF officer went through the survey questionnaire

with him before the real survey started, and made him practice the follow-up explanations in case

farmers did not understand the questions.

The maj ority of fodder banks were 0.25 ha (50 m 50 m) in area, since that was the

default recommendation of ICRAF. Some interviewed farmers turned out to have larger or

smaller sizes by the time of the survey, due to the success or failure of the management. All the

labor data and other costs were converted to per 0.25 ha basis before taking the mean. The live

fence study was based on the average live fence row-length, 291 m (van Dorp et al. 2005).

Because the live fence and/or fodder bank installation was at least a couple of years ago,

farmers seemed to have difficulties recalling the installation costs, especially labor (days and

people) needed for planting and management. Also, it was very difficult to estimate the amounts

of products harvested such as fodder and fruits. The sizes of the bags farmers were using to

collect the harvests varied. Direct measurements of the bag size and the fodder weight (air-

dried) were conducted at several villages to reduce the estimation variability.

Local Market Survey

There are three local markets inside the city of Segou where most of the farmers in the

villages go to buy/sell their products and equipment. Price data were collected from all markets,

although some products such as fodder were not sold in all the markets. The average price of

each item was used for the analysis.










sequestration as well as to develop a land resources information system in the WAS, geared

towards CDM and/or C sale.

Socioeconomic Implications

To analyze the socioeconomic feasibility of the agroforestry practices for mitigation

projects, analysis tools, i.e. models, are needed. Economic models of different scales used for

the studies in various ecoregions are summarized here. Although the number of studies is small,

the case studies and possibility of using these economic models in the WAS are also examined.

Economic Models

In most studies of C sequestration, agroforestry is regarded as one of the forest

management options for potential C sequestration. There are few studies specifically discussing

economic models of C sequestration in agroforestry systems; instead, models designed for

managed forests are usually applied (Masera et al. 2003). These economic models for

accounting C sequestration proj ects can be categorized into two different spatial scales.

National/global scale

Apart from the C sequestration potential per se of agroforestry systems, the potential for

realizing this assumed benefit depends largely on the availability of land which can be changed

to agroforestry from land with less C storage, such as agricultural Hields. Attempts to estimate

the global potential for increasing C sinks through land-use change had been conducted at the

global, national, and regional levels for more than a decade (Dixon et al. 1994b; Sathaye et al.

2001; Godal et al. 2003). These studies use simple integrated model structure, based on

biophysical and economic information. In this kind of large spatial scale, empirical model

schemes such as the Holdridge life zone system (LZS) can be used as a guideline for organizing

vegetation data (Pfaff et al. 2000). For economic factors such as the price of land, cost of land-

use change, and timber price, national census information are generally available. Information










LIST OF TABLES


Table page

2-1 Common tree and shrub species found throughout the West African Sahel .....................29

2-2 Main productive functi on s of agroforestry parklands ................. .......... ................3 1

3-1 Summary of various biomass C measurement approaches used commonly in C
sequestration studies .............. ...............59....

3-2 Aboveground time-averaged C stock in different ecosystems and agroforestry
practices .............. ...............60....

4-1 Characteristics of the villages where the experimental plots were set up in Segou
region, M ali ................. ...............76.................

4-2 Characteristics of the experimental plots (three plots average) for five-selected land-
use systems in Segou region, Mali ................. ...............76...............

4-3 Estimated biomass C (above and below ground) stock values of each plot and three
plots average of five selected land-use systems............... ...............77

4-4 Total C stock (biomass C + soil C of different depth) of five selected land-use
system s. .............. ...............77....

5-1 Soil profile characteristics for plots of the five land-use systems used in the study at
Segou Region, M ali .............. ...............101....

5-2 613C ValUeS of whole soil and three fraction sizes from five studied land-use systems,
at Segou Region, M ali ................. ...............102...............

6-1 Demographic characteristics of the target population in Segou, Mali ................... ..........13 8

6-2 Net Present Value (NPV), Benefit Cost Ratio (BCR), and Internal Rate of Return
(IRR) of the live fence and the fodder bank proj ects in the three different scenarios
(without C sale, with C sale by the ideal accounting method, and with C sale by the
tonne-year accounting method) in Segou, Mali ................ ...............138..............

6-3 NPV sensitivity of the live fence proj ect and the fodder bank proj ect to the change of
an input variable in Segou, Mali ................ ...............139..............










reported regarding C sequestration potential of agroforestry systems in semiarid and arid regions.

In addition to already existing indigenous agroforestry systems, improved practices and

technologies are now being expanded into these dry regions for perceived benefits such as

arresting desertification, reducing water and wind erosion hazards, and improving biodiversity

(Droppelmann et al. 2000; Gordon et al. 2003). In this scenario, it is imperative that C

sequestration potential of agroforestry practices in these regions is investigated. Considering that

the ecological production potential of these dry ecosystems is inherently low compared to that of

"high-potential" areas of better climatic and soil conditions, the extent to which agroforestry

systems can contribute if at all to C sequestration in such regions is in itself an important

issue.

This study was conducted in Mali, situated in the West African Sahel (WAS), one of the

largest semiarid regions of the world. Considering the large extent of area of the region (approx.

5.4 million km2), results of studies of this nature are likely to have wide applicability; yet, such

studies have been rare, possibly because of the relative backwardness of the region in terms of

economic development and therefore research facilities and infrastructure. Needless to say, such

studies are important because of their relevance in the context of C credit sale under CDM. The

WAS is one of the most environmentally vulnerable and poorest areas in the world. If the

maj ority of the people who are subsistence farmers can receive even small amounts of C

payments through their agroforestry practices, it would be a substantial contribution to their

economic welfare and the overall development of the region. Thus, an analysis of the C

sequestration potential of various agroforestry practices (traditional and newly introduced) in the

region is timely.

Research Questions and Objectives

To address the issues discussed above, four research questions are raised:









subsoil by colloidal or soluble C; but they do not seem to occur significantly under the soil and

climatic conditions of the study region (Tan et al. 2004; Nierop et al. 2007; Zinn et al. 2007).

The turnover time for physically protected C (type 1 and 2) is estimated to be 50 1000 years;

for biochemically protected C (type 3), it is 1000 3000 years. The turnover time for less stable

C within macroaggregates is 5 50 years, and for other types C, such as the litter fraction, it is

0. 1 20 years (Batjes 2001). To differentiate the types of soil C, physical fractionation is the

common initial step. The dynamics of soil C in each fraction size can be further investigated by

13C isotopic ratio measurement, which distinguishes between C derived from plants that follow

C3 photosynthetic pathways (all trees) and those that follow C4 pathways (most warm-season

graminaceous plants: in this study pearl millet, Pennisetum glaucum, and sorghum, Sorghum

bicolor). This method has been used for studying the impact of land-use change on soil C and

for comparing the C dynamics in different land-use systems (Balesdent et al. 1998; Potvin et al.

2004).

Research Questions

In this scenario, the present study was undertaken based on the premise that compared with

agricultural and tree-less systems, agroforestry systems will help store more C in soil and offer

better stability of stored C in deeper soil layers due to presence of deep-rooted trees. Specific

research questions are:

1. Do trees contribute to soil C storage in the selected agroforestry systems, and how stable is
the stored C?

2. What is the relative attractiveness of each of the selected agroforestry systems or land-use
change in terms of its soil C sequestration potential?

Materials and Methods

The study was conducted in the seven selected villages of Segou region, Mali, West

Africa. The details of the site and the selected land-use systems (treatments) are described in










1: 0 10 cm, 2: 10 40 cm, 3: 40 100 cm.

Fk is the fraction size, k = L, M, S

L: large fraction (2000 250 Cpm), M: medium fraction (250 53 Cpm), S: small fraction

(<53 Cpm)

It is the isotopic ratio, l = C3, C4

C3: C3 plants origin C, C4: C4 plants origin C

T,, is the distance, m = n, m, f

n: near the tree, m: middle of the canopy, f: far from the tree

eykln is the random variable error within the experiment.

Model 1 and model 2 including all the possible interactions between the factors were run

using PROC MIXED procedure of SAS. Interactions that were not significant were dropped

from the model. The models that were biologically and statistically significant are presented in

the results section. To further examine the interactions, data were sorted (PROC SORT

procedure) with certain factors fixed, and tested again using ANOVA. Based on the outcome of

the ANOVA, factors and other soil characteristics (e.g. percentage of sand, silt, and clay) were

tested for their relationships using linear regression.

All statistical tests were considered significant when p<0.05 unless otherwise specified.

Results

Soil Characteristics

Soils in the sample plots are mostly sandy loam or loamy sand (Table 5-1). Soil colors

varied from whitish or dark gray to reddish brown in different plots, but all are categorized as

Haplustalfs by the regional survey (Doumbia 2000). Abandoned-land soil was extremely hard to

sample with an auger because bedrock was found in some places at less than 1 m depth (Figure

5-2). Most of the time, silt or clay was clearly observed in 70 80 cm depths. Content (g kg









CHAPTER 4
ABOVEGROUND AND BELOWGROUND CARBON STOCKS IN TRADITIONAL AND
IMPROVED AGROFORESTRY SYSTEMS IN MALI, WEST AFRICA

Introduction

Agroforestry is a very common concept of traditional agricultural land-use in most of the

tropics. In the West African Sahel (WAS), the traditional systems such as "bush fallow" and

"parkland" systems involve integration of trees with agricultural crops. The trees provide

subsidiary (famine) food when crops fail by drought; can be sources of oil, wine or other

condiments, and are used for tools, fences or fodder (Boffa 1999). Also, trees can increase water

and nutrient availability through nitrogen-fixation, retrieval of water and nutrients from the

deeper layers of soil, and reduction of water and nutrient losses from leaching and erosion in the

semiarid region (Buresh and Tian 1997; Kang et al. 1999). As described in Chapter 2, parkland

agroforestry systems are currently the most prevalent land-use systems in the WAS. Other

agroforestry practices such as improved fallow, intercropping, tree fodder planting, and boundary

planting have been introduced, but these are still not widely adopted (Niang et al. 2002;

Levasseur et al. 2004).

Most of existing studies on the parkland systems are about the productivity of trees and/or

crops grown underneath, or about the interaction/competition of the trees and crops (Kater et al.

1992; Jonsson et al. 1999). Carbon (C) sequestration, defined by the United Nations Framework

Convention of Climate Change (UNFCCC) as "the process of removing carbon from the

atmosphere and depositing it in a reservoir," has not been a subj ect matter of studies in much of

the WAS region, let alone in parklands and other agroforestry systems of the region.

Nevertheless, it is widely accepted that environmental degradation resulting from long-term

drought and land-use change has adversely affected the terrestrial C stocks in the region (FAO

2000; Reich et al. 2001). Although Mali, where this study was conducted, signed off on the









afforestation projects. The model is a multi-cohort ecosystem-level model based on C

accounting of forest stands, including forest biomass, soils and products (Masera et al. 2003).

Another common methodological approach to estimating mitigation potential more broadly

is known as comprehensive mitigation assessment process (COMAP). The COMAP model

requires the proj section of land-use scenarios for both a baseline and for a mitigation case. It

requires data on a per hectare basis on C sequestration in vegetation, detritus, and forest

products, soils and also on GHG emission avoidance activities (Makundi and Sathaye 2004).

Default Values for Land/Activity Based Practices

This approach is the broadest, nation-level approach, which uses default values for certain

land-based activities for estimating C storage. A land-use based accounting system would focus

on the changes in C stocks on managed lands during a defined time period (Dixon et al. 1994a).

Default values would be assigned to a particular tract of land based upon county or regional level

research on the average sequestration likely to result from specific agricultural or conservation

measures in that area. Various values could be assigned to such broad land management

activities as forest, cropland, or grazing management. Under this approach, field measurements

of C storage changes in individual fields would not be necessary. Land-use monitoring can be

readily measured by remote sensing techniques, eliminating the need for many field inspectors.

However, field plots may need to be set up, representing the average or a range of conditions for

the entire proj ect area, and used as a reference to provide actual estimates to increase the

accuracy of large-scale proj ects.

Accounting Methods

In order to assert that agroforestry systems are an important C sequestration method, the

amounts measured in agroforestry systems must result in long-term changes in terrestrial C

storage and CO2 COncentrations in the atmosphere (Masera et al. 2003). Thus, the time frame











(g C kg"' soil)
O 1 2 3


(g C kg'' soil)
1 2 3


(p m) O


(p m)


m
-
m
m
m
M
m
m
M
m


I I


I





i


I I


250 2000
53 250
<63
S260 2000
53 250
o- <63
250 2000
53 250
<63


250 2UO
63 2j0
63
S260 2U0
5 3 2j0


250 2UO
63 2j0
63


m C3


m C4
m C3


(g C kg-' soil)
O 1 2 3


(g C kg-' soil)
0 1 2 3


(p m)


(p m)


I I


260 2U0
63 2j0
63
E' 250 2UO0
--53 250
8 63
250 2UO
53 250
8 63


250 2000
53 250
<53
E' 250 2000
-c 250


0 250 -2000
58 250
8 <53


IIC4
II C3


m C4


m
-
m


m
M


I
I










Doumbia, O. Final Report on the Soil Resources of the Villages Covered by the "JICA-SEGOU
Project" 2000. Segou, Mali.

Droppelmann, K.J., J. Lehmann, J.E. Ephrath, and P.R. Berliner. 2000. Water use efficiency and
uptake patterns in a runoff agroforestry system in an arid environment. Agrofor. Syst.
49:223-243.

Duguma, B., J. Gockowski, and J. Bakala. 2001. Smallholder cacao (Theobroma cacao Linn.)
cultivation in agroforestry systems of West and Central Africa: challenges and
opportunities. Agrofor. Syst. 51:177-188.

Duiker, S.W. and R. Lal. 1999. Crop residue and tillage effects on carbon sequestration in a
Luvisol in central Ohio. Soil Tillage Res. 52, 73-81.

Dzurec, R.S., T.W. Boutton, M.M. Caldwell, and B.N. Smith. 1985. Carbon isotope ratios of soil
organic matter and their use in assessing community composition changes in Curlew
Valley, Utah. Oecologia (Berlin) 66:17-24.

Eleki, K., R.M. Cruse, and K.A. Albrecht. 2005. Root segregation of C3 and C4 species using
carbon isotope composition. Crop Sci 45:879-882.

Eswaran, H., R. Almaraz, E. Berg, and P. Reich. 1996. An Assessment of the Soil Resources of
Africa in Relation to Productivity. [Online] Available at
http ://soils.usda.gov/use/worldsoils/papers/afia.html (verified 5 Jul. 2007). NRC S,
Washington, DC.

FAO. 2000. Global Forest Resources Assessment 2000. FAO Forestry Paper 140. Rome, Italy.

FAO. 1997. Estimating biomass and biomass change of tropical forests. FAO Forestry Paper
134. Rome, Italy.

FAO. 1991. Feeding dairy cows in the tropics. FAO Animal Production and Health Paper 86.
Rome, Italy.

Ferguson, W. 1983. Integrating crops and livestock in West Africa. FAO Animal Production and
Health Paper 41. Rome, Italy.

Franklin, J. and P. Hiernaux. 1991. Estimating foliage and woody biomass in Sahelian and
Sudanian woodlands using a remote sensing model. Int. J. Remote Sens.12:1387-1404.

Franzluebbers, A.J., F.M. Hons, and D.A. Zuberer. 1995. Tillage-induced seasonal changes in
soil physical properties affecting soil CO2 evolution under intensive cropping. Soil Tillage
Res. 34: 41-60.

Garcia-Oliva, F. and O.R. Masera. 2004. Assessment and measurement issues related to soil
carbon sequestration in Land-Use, Land-Use Change, and Forestry (LULUCF) proj ects
under the Kyoto Protocol. Clim. Change 65:347-364.































L596


5-
4-






-1 U


I I I II I
-50 0 50 1 U 1 50 2U3

N PV (FCFA i n thousands)


-52. 71 33


11 Q C699


Figure 6-4. Simulated NPV probability distribution of the live fence project (with C sale by the
ideal accounting method). The distribution is likelihood (y-axis) of the proj ect' s NPV
(x-axis): the worst scenario can be less than -75,000 FCFA in NPV, and the best
scenario can be close to 150,000 FCFA in NPV. The peak of the distribution is most
likely (best guess) scenario of the proj ect.










specific difficulties arising from requirements for monitoring, verification, leakage assessment

and the establishment of credible baselines, agroforestry estimations are beset by the problem of

estimating the area under agroforestry practices. Nevertheless, the IPCC (2000) estimated that

630 million ha of unproductive croplands and grasslands could be converted to agroforestry

worldwide, with the potential to sequester 391,000 Mg of C per year by 2010 and 586,000 Mg C

per year by 2040.

Although the credibility of conceptual models and theoretical benefits has been

demonstrated, C sequestration potential is still a little-studied characteristic of agroforestry

systems (Nair and Nair 2003). More studies examining how much C can be sequestered/stored

in various agroforestry systems around the world are needed. Several studies and reviews from

different regions of the world have discussed agroforesty' s benefits and limitations for C

sequestration (Schroeder 1994; Dixon 1995; Albrecht and Kandji 2003), but only very few deal

with comprehensive comparisons of different practices in each ecoregion.

Due to the difficult physical environment and lack of research infrastructure, agroforestry

systems in the WAS are one of the least documented topics regarding C sequestration potential.

Lal (1999) estimated the potential for sequestering C in the region was, as in most other

drylands, fairly low, between 0.05 0.3 Mg C ha' yr- The estimate, however, included a

variety of uncertainties related to future shifts in global climate, land-use and land cover, and the

poor performance of trees and crops on poor soils in the region.

In the WAS, impacts of population pressure, over-grazing and continuous drought are

causing severe land degradation. Consequently, biomass C stocks steadily decline within land-

use/land cover. Opportunities for C gains in the region are, thus, often discussed in the context

of agricultural fertility and sustainability of farming systems, which involve agroforestry such as









Further sorting and testing showed that the distance factor was significant in only M fraction C at

the 0 10 cm depth, where "near tree" was higher in content than "outside crown."

Isotope Analysis of Whole Soil C

The measured 813C ValUeS of each depth class of each land-use systems varied from -23.9

to -15.1 (Table 5-2). Based on the values and the mass balance calculation, whole soil C data

was separated into that originating from C3 plants (trees) and C4 plants (crops) (Figure 5-6).

"Near tree" data and "outside crown" data are presented side by side for two parklands (Figure 5-

6 A, B, C, D) and live fence (Figure 5-6, E, F). In the figure, C of tree origin was found more in

surface soil and near the tree, although when they were tested statistically, there was no

significant difference between "near tree" and "outside crown" data except for that of live fence

at the 0 10 cm depth. Fodder bank did not have much C of C3-origin, even with trees growing

in the plots. On abandoned land, C4-origin C was the major form of C, and, as mentioned

earlier, the soil C content was higher in this system compared with other systems.

Three-factorial (land-use, depth, and isotopic ratio) ANOVA was conducted among the

five land-use systems (model 1). All factors were significant, and three-factor interaction and

two-factor interactions including isotopic ratio were also significant. C3-origin C and C4-origin

C were then tested separately using the "SORT" procedure. Land-use was not a significant

factor among C3-origin C data, but depth was. For C4-origin C, both land-use and depth had

significant effect: abandoned land had higher content than other four systems with parklands

higher than the improved systems, and deeper depth had less C content.

Four-factorial (land-use, depth, distance, and isotopic ratio) ANOVA was used to test

differences among the two parklands and live fence systems (model 2). Distance was again not a

significant factor while all others were. Four-factor interaction, as well as three- and two-factor

interactions including distance was significant, suggesting distance was somewhat influential for











Table 6-3. NPV sensitivity of the live fence proj ect and the fodder bank proj ect to the change of
an input variable in Segou, Mali.
Live fence Fodder bank

No C sale With C sale No C sale With C sale
-------- FCFA ------

Base 52,802 60,465 87,319 96,394


Discount rate -5 % 109,367 118,391 166,434 177,304
Discount rate +5 % 20,999 27,478 43,094 50,670

Seedling cost -50 % 68,829 76,492 101,844 110,919
Seedling cost +50 % 36,775 44,438 72,794 81,869

Labor price -50 % 83,171 87,002 88,750 93,287
Labor price +50 % 22,433 26,264 85,888 90,425

Yield of harvests +50 % 129,330 140,825 149,624 163,236
Yield of harvests 50 % -23,727 -19,894 25,014 19,551

C price +50 % 64,297 100,931
C price 50 % 56,633 91,856


Source: Base values are from the best guess scenario cash flow. NPV values of "with
from ideal accounting method scenario.


C sale" are










g C kg ~'soil
01 2 34


g C kg-' soil




B 2
1730
S40

60
70
. +250 -2000 plm BO
-E53 -250 plm 90
<53 lm
100


g C kg' soil
S1 23 4




j~;Tt~C


O

30
40

60
70

BO

90
100


-+250 -2000 plm
-e53 20 plm
<53 plm


-st250 2000 lm
-4E53 10 lm
<53 lm


g C kg-1 soil





E








-5 1OCl


-<63 plm


Sg C kq-' sil 4







- 25 20 l


- 5 j l




-<63 lm


O
10
20

40
.1 4
50
S60
70
80
90
100


O
10
20


r 40
0..0
250

70
80
90
100


Figure 5-5. Soil C content of three particle size fractions in three depth classes (0 10 cm, 10 -
40 cm, and 40 100 cm) under five land-use systems in Segou, Mali. A) Faidherbia
albida parkland, B) Vitellaria paradoxa~11~~1~~11~ parkland, C) Live fence, D) Fodder bank, and
E) Abandoned land. Range of the each depth value is 95% confidence level. Depth
indicated is the mid-point of sampled depth.









because Kyoto Protocol currently admits only C sequestered as a result of newly implemented

mitigation proj ects, and traditional land-use systems such as parklands and forest conservation -

sustaining parklands are not recognized as emission-reduction activities under the Protocol at

least until 2012.

On the other hand, improved agroforestry systems (live fence and fodder bank) were found

to have a better chance to be recognized as C sequestering activities than parkland systems.

Because these systems are newly introduced, most of the biomass stored in the systems can be

counted as "sequestered" C credits, although their potentials as C sequestration proj ects were not

as high as expected. C sequestration potential of a land-use system has to be expressed on a unit-

area basis for a given period of time. From that perspective, some improved agroforestry

systems (live fence and fodder bank, in this study) do not rank high because of the nature of their

planting configurations and/or management requirements. Live fence trees are densely planted

along rows such that the individual trees are not "allowed" to grow fully; moreover, the fence

rows are on plot boundaries and therefore the area occupied or influenced by a fence row in

relation to the total area of the plot it borders is low. As far as the fodder banks are concerned,

the fodder trees that are frequently harvested for their leafy biomass cannot obviously be

expected to store large quantities of biomass C. Therefore, the absolute amounts of C stored in

these systems per unit area would not be as large as that for, say, parklands. While the amounts

of biomass C stored (calculated from general allometric equations following UNFCCC

guidelines) were 54.0 and 22.4 Mg C ha-l respectively for 40-year or older stands ofF. albida

and K paradoxa parkland systems, the amounts were 4.7 Mg C ha-l for a 8-year-old stand of live

fence and 2.2 Mg C ha-l for 6 9 -year-old stand of fodder bank.









Root biomass is often estimated from root:shoot ratios (R/S). It can be calculated by

sample plot measurements, but there are also lists of reference data. A literature review by

Cairns et al. (1997) included more than 160 studies covering tropical, temperate and boreal

forests that reported both root biomass and aboveground biomass. The mean R/S based on these

studies was 0.26, with a range of 0. 18 (lower 25 % quartile) to 0.30 (upper 75 % quartile). The

R/S did not vary significantly with latitudinal zone (tropical, temperate, and boreal), soil texture

(fine, medium and coarse), or tree type (angiosperm and gymnosperm).

Soil C samples should be collected from each layer, dry-weighed and analyzed for its C

contents by recommended laboratory procedures. To calculate C stocks per unit area, the C

content in the soil is multiplied by the bulk density of the respective soil layer. By itself, C

sequestration in agricultural soils is expected to make only modest contributions globally (e.g., 3

- 6 % of total fossil C emissions) (Paustian et al. 1997). However, this amount can be

significantly varied through management such as fallow phase, erosion, tillage, or tree

incorporation.

Indirect Remote Sensing Techniques

Even where field measurement methodologies are established, agricultural/forestry

practices are inherently dispersed over a wide geographic area. Staffing costs for monitoring and

verification of land-use practices over such a wide area could prove to be cost prohibitive.

Because direct field measurements can be expensive, the use of indirect remote sensing

techniques is being considered. A range of remote data collection technologies is now available

including satellite imagery and aerial photo-imagery from low flying airplanes. Sensors that can

measure the height of the canopy or vertical structure will be needed along with the more

traditional sensors on Landsat or Spot satellites in order to improve the ability of remotely

sensing biomass (Brown 2002).










Table 2-1. Common tree and shrub species found throughout the West African Sahel.


Botanic Description
One of the most common species in the
Sahel. Deep root system with feathers
leaves protecting barks from dry winds.
Species often seen in the WAS are A.
nilotica, A. tortillas, A. senegal, and A.
seyal.

This drought-and fire-resistant tree is found
throughout the Sahel. With trunks that are
often 10 15 m wide, it is one of the largest
trees (in terms of trunk width): it grows up
to 25 m high. In the dry season, the baobab
is completely without leaves, and because of
its distinguishable shape of branches that
look like roots, it is called the "upside down"
tree.


Species
Acacia spp.







Adansonia
digiata
(The baobab
tree)


Functional Use
The bark of most acacia produces tannins,
which are used in tanning leather. A.
senegal produces gum arabic, used in
pharmaceuticals and adhesives. Fruits are
sometimes consumed as condiments


The bark can be used for rope and cloth,
and the trunk, when hollowed out, as a
shelter. Fruits and leaves are food
sources; especially leaves are very
important vitamin source for the local
people.


Balanites Multi-branched, spiny shrub or tree up to 10
(,, e mot e,, m tall. Trunk is short and often branching
from near the base. Branches are armed
with stout yellow or green thorns up to 8 cm
long.


The fleshy pulp of both unripe and ripe
fruits is edible and eaten dried or fresh.
The fresh and dried leaves, fruits, and
sprouts are all eaten by livestock.


The branches are quite strong, and are a
useful material for building stools, beds,
tool handles, etc. A tea made by steeping
the leaves of C. micaranthum in boiling
water is a traditional tonic drink and a
decoction of the leaves is sometimes used
as a medication for malaria.

It is a valuable fodder tree for game and
domestic animals during dry season. The
seeds can be boiled and eaten, but first
the skin has to be removed. Also the pods
may be dried and ground into flour,
which is edible.

Leaves and roots are traditionally used to
treat different diseases, particularly
malaria and intestinal disorders.


Combretum
spp.


The genus comprises about 370 species of
trees and shrubs, 300 of which are native to
tropical and southern Africa. C. glutinosum
and C. micaranthum are common in the
WAS. They are bushes branching from
bases, 1-2m tall.


Faidherbia One of the fastest growing trees in the WAS.
albida It is deciduous and has the remarkable
phenolgy of leaves falling off in rainy
season and coming back in the dry season. It
can grow up to 30 m tall. Branching stems
and an erect to roundish crown.


Guiera
senegalensis


Perennial bush which is a maj or component
of disturbed parts of bushland in the WAS.
Also abundant on roadsides and fallowed
lands. The woody part is fragile.










economically affect local households. Chapter 7 gives a synthesis, conclusions and

recommendations for future research and development efforts.









ACKNOWLEDGMENTS

I am indebted and grateful for many individuals and organizations who contributed to this

study and my doctoral program. I thank my chair, Dr. P.K. Nair, for his dedication and guidance

throughout this process, and my committee, Dr. Nick Comerford, Dr. Janaki Alavalapati, Dr.

Tim Martin, Dr. Ted Schuur, and Dr. V.D. Nair, for their support and encouragement.

I recognize and express my sincere gratitude to the individuals and their institutions that

supported me during my doctoral studies: the School of Forest Resources and Conservation

(Cherie Arias, Sherry Tucker, Dr. Tim White), University of Florida International Center (Debby

Anderson), the Center for Tropical Conservation and Development of UF, the World

Agroforestry Centre (especially Dr. Bocary Kaya), the Fulbright Program, and the Joint

Japan/World Bank Graduate Scholarship Program (JJ/WBGSP).

At the Hieldwork in Segou, Mali, I received tremendous support and cooperation from the

farmers, field onfcers, and other collaborators. It was one of the most challenging times of my

life, and I could not go through without them. Thank you to Nicole Demers, Bayo Mounkoro,

Keita, Samake, and other officers in ICRAF Segou office, Kayo Sakaguchi, Takako Uchida, Mr.

Kiyoshi Sakai, and all the farmers in Segou who let me use their fields for data collections and

participated the survey.

I have greatly benefited from my friendship with colleagues in the agroforestry lab at UF.

I thank Solomon Haile, Alyson Dagang, Julie Clingerman, Sam Allen, Eddie Ellis, Brian Becker,

Joyce Leptu, David Howlett, Wendy Francesconi, Subrajit Saha, Shinjiro Sato, and Masaaki

Yamada, for the discussions and supports.

To my precious friends who have been an integral part of the many years of this process,

thank you Gogce Kayihan, Brian Daley, Elli Sugita, Mike Bannister, Jason and Karen Hupp,









Materials and Methods

The World Agroforesty Centre (ICRAF) conducted monitoring surveys for farmers

implementing live fences and fodder banks after introduction of the systems (Hamer et al. 2005,

van Duijl 2000). Data from these studies as well as databases from ICRAF research station were

used for this study. Furthermore, field surveys were conducted during February March 2006

(the dry season after the harvests when farmers were less busy with agricultural activities) to

collect additional data necessary for the analysis.

The target population was composed of farmers living in the Segou region who had

adopted live fences and/or fodder banks with assistance of ICRAF. A comprehensive cost-

benefit study of live fences had already been conducted for ICRAF by van Dorp et al. (2005).

Also, the need and social acceptability of live fences and fodder banks had been discussed in

several previous studies (van Duijl 1999; Levasseur 2003; Yossi et al. 2005). The information

about the fodder bank implementation was much more scarce than that of live fence. Thus, the

survey focused on collecting more data for fodder banks, specifically data to conduct the cost-

benefit analysis (CBA) equivalent to the existing live fence study, as well as data such as the

price of timber and non-timber products from both live fences and fodder banks to conduct risk

simulation analysis.

Social Survey of Fodder Bank Farmers

The questionnaire was designed based on that of the live fence survey (Annex A),

following the protocol of the Institutional Review Board of University of Florida (Protocol #

2005-U-1023). The structured questionnaire consisted of open-ended and/or close-ended

questions of 14 sections, asking for information about materials and labor used for managing and

harvesting fodder banks as well as related benefits from the implementation.









are also used as fodders. In the south, where the savannah replaces the steppe, the tall perennial

grasses such as Andropzogon gavanus as well as annual grasses with long cycles such as

Pennisetum pzedicellatum, Andropzogon pzseudapzricus, and Diheteropzogon hagerupziiare are

common. These grasses grow rapidly up to 2.5 m in height, but natural bush fires control the

reserves. Some of these species are introduced as ornamental or fodder species in the US

(Pennisetum pedicellatum, called Kyasuma grass) and Australia (Andoropogon gayan~~~ggg~~us), and

because of their rigorous spread, they are invasive species.

Although most tree and shrub species are found both in steppe and savannah (Table 2-1),

the woody vegetation become more and more diverse and dense as one goes south. The trees in

the WAS are usually low-branched and may ramify from their base. Crowns are generally very

wide, and much more developed than the bole. The thickness of the bark has been interpreted as

affording protection against repeated bush fires. Spines and thorns on branches are also frequent,

which prevent reducing water loss through evaporation. It may afford some protection against

browsing by large mammals, but does not prevent foliage browsing.

Soil

Detailed information on the soil resource base of the WAS is inadequate for most research

purposes. In most countries, farm-level information and detailed soil maps are non-existent.

Still, in 1977, Food and Agricultural Organization (FAO) of United Nations (UN) and UN

Educational and Scientific Organization (UNESCO) formed soil map of Africa, by aggregating

specific soil mapping units to form soil regions that corresponded roughly to Africa's maj or

ecological regions. Natural Resources Conservation Service (NRCS) of the United States

Department of Agriculture (USDA) had a pedon database with more than 400 pedons from

Africa. With published national soil survey reports, NRCS translated the legend of the UN Soil

Map of the Africa into Soil Taxonomy Map (Figure 2-4).

























I III/I /I I II


r II


i 1 1 1 l i l
1890 1900 10 20 30


i l l
4s0 1950 60


l i i l l
70) 80


l I l
90 200


year


Figure 2-2. Standardized annual Sahel rainfall (June to October) from 1898 to 2004. The rainfall
data are converted to relative values (standardized) with respect to data from 1898
tol993, such that the mean and standard deviation of the series are 0 and 1,
respectively. Positive values (orange) are the years with rainfall more than the mean
of 1898 1993 data, and negative values (blue) are the years with less rainfall.
Source: Mitchell (2005)


A B


Figure 2-3. Seasonal landscape contrast of the WAS. Photos of the same site A) in the dry
season and B) rainy season. Source: USGS (http://edcintl. cr.usgs.gov/sahel .html).


'"r CT"''"il'"l'r










Boffa, J. M. Agroforestry parklands in sub-Saharan Africa 1999. FAO Conservation Guides 34.
FAO, Rome, Italy.

Bouliere, F. 1983. Tropical Savannas. Elsevier Scientifie Publishing Co. New York, NY.

Breman, H. and J.J. Kessler. 1997. The potential benefits of agroforestry in the Sahel and other
semi-arid regions. Euro. J. Agron. 7:25-33.

Brown, S. 1997. Estimating Biomass and Biomass Change of Tropical Forests: a Primer. FAO,
Rome, Italy.

Brown, S. 1999. Guidelines for Inventorying and Monitoring Carbon Offsets in Forest-Based
Projects. Winrock International, Arlington, VA.

Brown, S. 2002. Measuring carbon in forests: current status and future challenges. Environ.
Pollut. 116:363-372.

Buresh, R.J. and G. Tian. 1997. Soil improvement by trees in sub-Saharan Africa. Agrofor. Syst.
38:51-76.

Cacho, O.J., G.R. Marshall, and M. Milne. 2003a. Smallhoder agroforestry projects: potential for
carbon sequestration and poverty alleviation. ESA Working Paper No.03-06. FAO.
www. fao. org/es/esa

Cacho, O.J., R.L. Hean, and R.M. Wise. 2003b. Carbon-accounting methods and reforestation
incentives. Aust. J. Agric. Res. Econ. 47:153-179.

Cairns, M.A., S. Brown, E.H. Helmer, and G.A. Baumgardner. 1997. Root biomass allocation in
the world's upland forests. Oecologia (Berlin) 111:1-11.

Campbell C.A., B.G. McConkey, R.P. Zentner, F. Selles, and D. Curtin. 1996. Tillage and crop
rotation effects on soil organic C and N in a course-textured Typic Haploboroll in
southwestern Saskatchewan. Soil Tillage Res. 37: 3-14.

Campbell, H.A. and P.C. Brown. (ed.) 2003. Benefit-cost analysis: Financial and economic
appraisal using spreadsheets. Cambridge University Press, Port Melbourne, Australia.

Choudhary, M.A., A. Akramkhanov, and S. Saggar. 2002. Nitrous oxide emissions from a New
Zealand cropped soil: tillage effects, spatial and seasonal variability. Agric. Ecosyst.
Environ. 93:33-43.

CIA. 2007. The World Factbook: Mali.[Online] Available at
https://www. cia. gov/library/publications/the-world-factboo/esm.html (verified 5 Jul.
2007). CIA, Washington, DC.

Cissee, M.I. 1980. Production fourragre de quelques arbres saheliens: relations entire la biomasse
foliaire maximale et divers parametresphysiques. p.203-208. In H.N. Le Houerou. (ed.) Les
fourrages ligneux en Afrique, L'etat actuel des connaissances. CiPEAA, Addis Abeba.









conditions, it is generally restricted to a period of three to five months from April to October.

During this period, there is an average of 24 rainfall events, 10 to 12 of which occur in August.

Rainstorms are rarely prolonged, usually lasting no more than one or two hours. Rainfall

intensities range from 5 to more than 50 mm per event (Gritzner 1988). The rainy season is

followed by an extended dry season where the vegetation cover changes drastically (Figure 2-3).

The monthly mean temperature of the region is 26 27 OC, with a monthly mean

maximum of 34 36 OC and monthly mean minimum 21 230C. Temperature abnormalities are

relatively low for the area as a whole (+0.7 OC to -0.6 OC), but may be greater in individual

places (Littmann 1991).

Vegetation

The WAS contains three generalized phytogeographical divisions corresponding to the

climate zones (Figure 2-1): (i) the northerly Sahelo-Saharan zone, or grass steppe, between the

100 and 200mm isohyets; (ii) the Sahel proper, or tree steppe, between the 200 and 400 mm

isohyets; (iii) the southerly Sudano-Sahelian borderlands, or shrub savannah, extending to the

800 mm isohyets.

Savanna plants are renowned for their well-developed root systems, penetrating deeply into

the soil. Herbaceous plants, mostly perennials, always have an extensive root system, often

forming a close mat of rootlets in the upper layers of the soil. Most of the roots are located

within the upper 30 cm of soil (Bourliere 1983). Grasses in the steppe grow in the very short

growing season (60 90 days) with narrow leaves in circles or basal rosettes. One of the most

common grass species throughout the WAS is Cenchrus biflorus. This prickly, short-lived grass

is the food of choice for the herds that graze throughout the Sahel. Mature grass has sharp

bristles; but ensiling softens them, so that it can also be used as silage (FAO 1991). Other

common grass species in steppe such as Schoenefeldia grcilis, Elionorus elegans, Borreria spp.,










Kyoto Protocol, there has been no pilot proj ect to document C sequestration and C credit sale in

the country.

Woomer et al (2004b) conducted a national scale C stock assessment in Senegal

(neighboring country of Mali), and found that there were opportunities for biological C

mitigation, but they were constrained by available knowledge and access to resources.

Compared with large-scale tree plantation, agroforestry is expected to be the most feasible

afforestation/reforestation proj ect that can be conducted by the maj ority of resource-limited land

users (farmers) in the WAS. Because of the scarcity of on-site information, it is important to

directly measure or estimate both biomass C and soil C stocks of various agroforestry systems.

Therefore the study reported in this chapter was undertaken with two research questions:

1. How do different agroforestry systems differ in their potential for C sequestration? How
much C is stored in the traditional and improved agroforestry systems, especially comparing
above-ground and below-ground?

2. What is the overall relative attractiveness of each of the selected agroforestry systems
considering them as biological C sequestration proj ects?

Materials and Methods

C sequestration potential of a specified proj ect is calculated by "C sequestered by the

proj ect" minus "C sequestered by the baseline (without the proj ect)". Since this study is not a

long-term proj ect, it was impossible to monitor both C accumulations by the proj ect

(agroforestry) and by non-proj ect land-use over the time. Instead, the differences of C stock

among selected land-use systems are assumed to represent the potential of C sequestration by the

land-use change.

Study Area

This research was done in Segou, Mali, in cooperation with the ICRAF (World

Agroforestry Centre) Field Station of Sahel Regional Programme.










application of a factor to convert the climatic effect of temporal C storage an equivalent

(equivalent factor) amount of avoided emissions.

Technical Problems and Uncertainties

There are a numb er of shortcomi ngs of conventi onal m ethod s for e sti mati ng/accounti ng

the C in a system that need to be considered. These include the uncertainties related to future

shifts in global climate, land-use and land cover, the poor performance of trees and crops, varied

environments, pests and diseases such as nematodes. For example, the amount of C remaining

belowground at the end of the tree rotation, and the amount of C sequestered in products created

from the harvested wood, including their final disposition are often not included in the

accounting methods discussed above (Johnsen et al. 2001). Oren et al. (2001) reported that after

an initial growth spurt, trees grew more slowly and did not absorb as much C from the

atmosphere as expected. They concluded that assessment of future C sequestration should

consider the limitations imposed by soil fertility as well as interactions with nitrogen deposition.

In addition to these uncertainties, there are some concerns about the impacts of

agroforestry in other GHGs. The wide-scale use of woody legumes, which is common in

agroforestry systems, might result in release of nitrous oxide (N20)(Choudhary et al. 2002),

although it does not seem to be as strong an impact as N- fertilization (Mosier et al. 2004). N20

is known to have a global warming potential 200 300 times higher than that of CO2. Similarly,

pasture and rice paddy cultivation in agroforestry systems can produce significant quantities of

methane (CH4), another strong GHG (20 60 times higher impact than CO2), on a global scale

(Dixon, 1995).



































(gC kg'l soil)
O 1 2 3


(g C kg soil)


O 1 2 3


(p m)


(p m)


S250 --1300
.- 53 250


c, 250-2~000
63 250
<53


Iil


250 -2000
SB 250


E' 250 -2000
--53 250


c, 250-2000
SB 250
<53


iggI


m C4
a C3


m C4
a C3


(gC kg-' soil)


(g C kg soil)


O 12 3


O 12 3


(prm)


(prm)


I I


250 -2000
CD
53 250
ca <53
E' 250 -2000
4-I53 -250
oE'- <53
c,250 -2000
53 250
<53


lill
-
lill
11111111
111
1111
-
iggggggggg
15551
-
lill
1555555555


250 2000
53 250
ca <53
i 20 2000

53 <53

C, 250-23000
CD
"[ 53 250
CD
S<653


SC4
m c3


MC3


lilIM


111111111
lilllli









dynamics, and are actively discussed in the soil science arena (Balesdent et al. 1998; Del Galdo

et al. 2003; Powers and Veldkamp 2005).

Soil fractionation: There are various ways to separate SOM into labile and recalcitrant

pools, and these methods rely on chemical, physical, or biological separation, and many of them

are used sequentially in analyses (McLauchlan and Hobbie 2004). Chemical fractionation

separates soil C into different resistance class to decomposition by using acid,

permanganate(KMnO4), or hot water, (Moody et al. 1997; Ghani et al. 2003). Physical

fractionation separates labile and recalcitrant fractions by either size or density. Sieving is used

to size differentiation and flotation with a dense liquid is usually used to measure light fraction

which is considered to be labile (Six et al. 1998). Biological separation uses microbes to

separate labile SOC from recalcitrant SOC under controlled temperature and moisture conditions,

assuming that microbes will mineralize the most labile C first, with recalcitrant C being

mineralized later (Alvarez and Alvarez 2000). With applying these methods, many examine the

impact of land-use change on soil C storage and dynamics. For example, it is possible to assess

how land-use rotation (including fallow) or change of management such as reduction of tillage

can effectively protect recalcitrant soil C, so that enhance soil C sequestration (Pikul. et al. 2007;

Zibilske and Bradford 2007).

13C isotope measurement: During photosynthesis, CO2 fixation of C3 plants discriminates

against the heavier isotope 13C more than do C4 plants, which result in different stable carbon

isotope composition (613C) 13 /12C ratio relative to that found in the PDB (Pee Dee belemnite),

for their plant material. This composition value of C3 plants is between -23 and -34"90, whereas

C4 plants ranges from -9 to -179Rb (Eleki et al. 2005). Negative values of 813C indicate that the

plant material is depleted in 13C COmpared with the PDB standard. Using this theoretical









CHAPTER 1
INTRODUCTION

Background

It is widely accepted that current global climate change or global warming is "the" most

serious environmental issue affecting human lives. Global warming refers to the increase in the

average temperature of the Earth's near-surface air and oceans in recent decades and its proj ected

continuation. It is brought about primarily by the increase in atmospheric concentrations of the

so-called greenhouse gases (GHGs). GHGs are components of atmosphere contributing to the

"green house effect," the process in which the emission of infrared radiation by the atmosphere

warms a planet's surface. The Intergovernmental Panel on Climate Change (IPCC), established

by the United Nations (UN) to evaluate the risk of climate change concludes in its most recent

report that "most of the observed increase in globally averaged temperatures since the mid-20th

century is very likely due to the observed increase in anthropogenic greenhouse gas

concentrations" (IPCC 2007). The Kyoto Protocol to the United Nations Framework Convention

on Climate Change (UNFCC) is the first and so far the largest international agreement to

stabilize GHG concentrations

Carbon dioxide (CO2) is a maj or GHG and its concentration build-up is accelerated by

human activities such as burning of fossil fuels and deforestation. One of the approaches to

reducing CO2 COncentration in the atmosphere, called biomass carbon (C) sequestration, is to

"store" it in forest and forest soils by trees and other plants through photosynthesis. This concept

became widely known because the Kyoto Protocol has an approach called Land Use, Land Use

Change and Forestry (LULUCF), which allows the use of C sequestration through afforestation

and reforestation as a form of GHG offset activities. The Marrakesh Accords in 2001

determined more detailed rules of LULUCF and added forest management, crop management,









The northern WAS, adj acent to rocky Sahara desert to the north, is dominated by Entisols

and in some parts by Aridisols. Most of the Entisols in the area have an aridic soil moisture

regime and are formed on sandy or loamy deposits. Psamments are present as Eingerings of

Sahara in zones with ustic or udic soil moisture regimes (Eswaran et al. 1996). Vertisols occur

locally in some places along the rift valley of the Niger River and around Lake Chad. At lower

latitudes within the WAS, Alfisols are extensively spread. In general, the wind-blown sand from

the Sahara desert has buried many of the former Oxisols and Alfisols/Ultisols; thus, soils in the

WAS characteristically have very sandy top soils and a low-activity clay subsoil.

In terms of soil quality for agricultural use, soil moisture stress is perhaps the overriding

constraint in much of the WAS. It is not only because of the low and erratic precipitation but

also of the ability of the soil to hold and release water. A large part of northern WAS (Entisols

and Aridisols) has low available water holding capacities (AWHC), <25 mm. And southern part

of the WAS is made up of soils with medium AWHC (24 100 mm), mainly Alfisols and

Ultisols. Salinity and alkalinity are other problems affecting agriculture. The extremely acid

soils, which are mainly the acid sulphate soils, occupy areas around the Niger delta. Some parts

of Alfisols (close to southern Ultisols) have acid surface and subsurface horizons, which,

coupled with the moisture stress conditions, makes these soils extremely difficult to manage for

productive use under low-input conditions. The annual additions of dust from the Sahara

brought by the Harmattan winds (a dry and dusty wind blowing south off the Sahara into the

Gulf of Guinea during the dry season) raise the pH and base saturation of the surface horizons;

although the changes are less acute than the eastern part of the Sahel where subsoil acidity is a

problem (Tiessen et al. 1991).









depth class, the distance factor was still not significant in any depth class. However, at 10 40

cm and 40 100 cm classes, C content was higher for F. albida parkland than in the case of the

live fence.

C in Soil Fractions

Carbon-fraction contents [large (L): (250 2000 Cpm), and medium (M): (53 250 Cpm)] in

different systems were mostly not different among each other, and ranged from 1 to 2 g C kg-

soil (except in fodder bank, where it was from 0 to 1 C kg-l soil) (Figure 5-5). Small (S) fraction

(<53 Cpm) C content did not change much from 0 10 cm to 10 40 cm depth in all systems. In

the live fence treatment, C in the 0 10 cm depth contained more L fraction (1.8 g C kg-l soil)

than the other two size fractions (less than 0.7 g C kg-l soil), whereas the fodder bank treatment

had very low C content of L and M fractions in that soil layer (0.6 g C kg-l soil).

Data of C in the three fractions were analyzed by three-factorial ANOVA (land-use, depth,

and size) (model 1). All three factors were significant, as well as three combinations of two-

factor interactions. Results of multiple comparisons among land-use systems were the same as

for the whole soil C data; abandoned land had higher C content than the other four systems,

which did not differ among each other. Depth class comparison also showed the same result: 0 -

10 cm > 10 40 cm > 40 100 cm. S fraction and L fraction C were both significantly higher in

content than M fraction C when three depth class data are combined. The significance varied

when each depth class was separately tested, but M fraction C content was always the lowest.

When each fraction size data were tested separately, land-use and depth were significant for all

fraction sizes.

Distance from the tree was the only factor that was not significant in four-factorial

ANOVA (land-use, depth, size, and distance) for three systems (two parklands and live fence:

model 2). Interactions of four factors and three factors including distance were not significant.









CARBON SEQUESTRATION POTENTIAL OF AGROFORESTRY SYSTEMS
IN THE WEST AFRICAN SAHEL:
AN ASSESSMENT OF BIOLOGICAL AND SOCIOECONOMIC FEASIBLITY






















By

ASAKO TAKIMOTO


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA

2007









Another issue is that the initial C loss (both in biomass and soil) resulting from land

clearing and tillage for facilitating tree establishment in these improved practices is expected to

be significant; this loss may not be compensated by the planted trees any time soon, given their

slow growth rates owing to poor soil fertility and adverse climatic conditions. Therefore,

introducing these improved systems in abandoned land for land amelioration, as the World

Agroforestry Centre (ICRAF) is promoting, may not make a significant contribution to net C

sequestration in the near term; indeed it is likely to cause net negative C balance in the initial

stage of implementation.

Currently soil C is not considered to be tradable, but the relative portion of soil C in the

studied systems turned out to be comparatively large. For example, the percentages of soil C (0

- 100 cm) in total C (biomass C + soil C 0 100 cm) stock of the studied agroforestry systems

were 38 % in F. albida parkland, 55 % in y. paradoxa parkland, 84 % in live fence, and 94 % in

fodder bank. This cannot be ignored when the potential for long-term storage is considered.

Soil fractionation studies and isotopic ratio measurements showed that tree litter tends to

increase unprotected, relatively new C on the surface soil. In the deeper soil, the parklands that

have had trees in the system for a long time were likely to hold more protected C than the newly

introduced live fence or fodder bank systems. Also, management practices such as tillage and

litter usage seemed to have a large influence on soil C accumulation.

Socioecomic Potential

The cost-benefit analysis suggested that live fence and fodder bank were likely to be

profitable for farmers, whether with or without C sale. C sale changed the profitability: $ 13.9

more in net present value (NPV) of average-size live fence (291 m), and $ 20.5 more in NPV of

average size fodder bank (0.25 ha). These estimations are based on the assumptions of 25-year

rotation, no transaction costs on the farmers, and an accounting method ideal to C sellers




Full Text

PAGE 1

1 CARBON SEQUESTRATION POTENTIAL OF AGROFORESTRY SYSTEMS IN THE WEST AFRICAN SAHEL : AN ASSESSMENT OF BIOLOGICAL AND SOCIOECONOMIC FEASIBILITY By ASAKO TAKIMOTO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSI TY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007

PAGE 2

2 2007 Asako Takimoto

PAGE 3

3 To my parents and grandmother

PAGE 4

4 ACKNOWLEDGMENTS I am indebted and grateful for many individu als and organizations who contributed to this study and my doctoral program. I thank my chair, Dr. P.K. Nair, for his dedication and guidance throughout this process, and my committee, Dr. Nick Comerford, Dr. Janaki Alavalapati, Dr. Tim Martin, Dr. Ted Sc huur, and Dr. V.D. Nair, for their support and encouragement. I recognize and express my sincere gratitude to the individuals and their institutions that supported me during my doctoral studies: the School of Forest Resources and Conservation (Cherie Ari as, Sherry Tucker, Dr. Tim White), University of Florida International Center (Debby Anderson), the Center for Tropical Conservation and Development of UF, the World Agroforestry Centre (especially Dr. Bocary Kaya), the Fulbright Program, and the Joint Jap an/World Bank Graduate Scholarship Program (JJ/WBGSP). At the fieldwork in S gou, Mali, I received tremendous support and cooperation from the farmers, field officers, and other collaborators It was one of the most challenging times of my life, and I c ould not go through without them. Thank you to Nicole Demers, Bayo Mounkoro, Keita, Samake, and other officers in ICRAF S gou office, Kayo Sakaguchi, Takako Uchida, Mr. Kiyoshi Sakai, and all the farmers in S gou who let me use their fields for data colle ctions and participated the survey. I have greatly benefited from my friendship with colleagues in the agroforestry lab at UF. I thank Solomon Haile, Alyson Dagang, Julie Clingerman, Sam Allen, Eddie Ellis, Brian Becker, Joyce Leptu, David Howlett, Wend y Francesconi, Subrajit Saha, Shinjiro Sato, and Masaaki Yamada, for the discussions and supports. To my precious friends who have been an integral part of the many years of this process, thank you Gogce Kayihan, Brian Daley, Elli Sugita, Mike Bannister, Jason and Karen Hupp,

PAGE 5

5 Charlotte Skov, Chrysa Mitraki, Rania Habib, Maitreyi Mandal, Trina Hofreiter, Troy Thomas, and my fianc Nick Georgelis. Last but not least, I express my most profound gratitude to my mother Ayuko Takimoto, whose endless love and confidence in me made me come this far.

PAGE 6

6 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ......................... 10 LIST OF FIGURES ................................ ................................ ................................ ....................... 11 ABSTRACT ................................ ................................ ................................ ................................ ... 13 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .................. 15 Background ................................ ................................ ................................ ............................. 15 Rationale and Significance ................................ ................................ ................................ ..... 16 Research Questions and Objectives ................................ ................................ ........................ 17 Dissertation Overview ................................ ................................ ................................ ............ 18 2 THE WEST AFRICAN SAHEL: GENERAL LAND USE AND AGROFORESTRY ........ 20 Description of the Region ................................ ................................ ................................ ....... 20 Climate ................................ ................................ ................................ ............................ 20 Vegetation ................................ ................................ ................................ ........................ 21 Soil ................................ ................................ ................................ ................................ ... 22 Traditional Farming Systems and Agroforestry in the WAS ................................ ................. 24 Traditional Agroforestry Practices ................................ ................................ .................. 25 Bush fallow/shifting cultivation ................................ ................................ ............... 25 Parkland system ................................ ................................ ................................ ........ 25 Improved Agroforestry Practices ................................ ................................ .................... 26 3 LITERATURE REVIEW: CARBON SEQUESTRATION POTENTIAL OF AGROFORESTRY SYSTEMS IN THE WEST AFRICAN SAHEL (WAS) ....................... 36 Overview ................................ ................................ ................................ ................................ 36 C Sequestration as a Climate Change Mitigation Activity ................................ ............. 36 Agroforestry for C sequestration ................................ ................................ ..................... 37 Methodologies for C Sequestration Measurements ................................ ................................ 39 Direct On site Measurement ................................ ................................ ............................ 39 Inventory ................................ ................................ ................................ .................. 40 Conversion and e stimation ................................ ................................ ....................... 40 Indirect Remote Sensing Techniques ................................ ................................ .............. 42 Modeling ................................ ................................ ................................ .......................... 43 Default Values for Land/Activity Based Practices ................................ .......................... 44 Accounting Methods ................................ ................................ ................................ ............... 44

PAGE 7

7 Approaches to Assessing C Sequestration Performance ................................ ................. 45 Fluxes of C and flow summation ................................ ................................ ............. 45 Average changes in the stocks of C ................................ ................................ ......... 45 Cumulative C storage ................................ ................................ ............................... 46 Other accounting methods ................................ ................................ ........................ 46 Technical Problems and Uncertainties ................................ ................................ ............ 47 Biomass C Sequestration ................................ ................................ ................................ ........ 48 Studies in Various Ecoregions ................................ ................................ ......................... 48 Studies in West Africa ................................ ................................ ................................ ..... 48 Soil C Sequestration ................................ ................................ ................................ ............... 50 Studies of Soil C Stock and Dynamics ................................ ................................ ............ 50 Soil C in the WAS ................................ ................................ ................................ ........... 52 Socioeconomic Implications ................................ ................................ ................................ ... 54 Economic Models ................................ ................................ ................................ ............ 54 National/global scale ................................ ................................ ................................ 54 Micro/site s pecific scale ................................ ................................ ........................... 55 Feasibility in West Africa ................................ ................................ ................................ 56 4 ABOVEGROUND AND BELOWGROUND CARBON STOCKS IN TRADITIONAL AND IMPROVED AGROFORESTRY SYSTEMS IN MALI, WEST AFRICA ................. 61 Introducti on ................................ ................................ ................................ ............................. 61 Materials and Methods ................................ ................................ ................................ ........... 62 Study Area ................................ ................................ ................................ ....................... 62 Republic of Mali ................................ ................................ ................................ ....... 63 Sgou region ................................ ................................ ................................ ............. 63 Select ed Land use Systems for Field Data Collection ................................ .................... 64 Parkland systems ................................ ................................ ................................ ...... 64 Improved a groforestry s ystems ................................ ................................ ................ 65 Abandoned (degraded ) land ................................ ................................ ..................... 66 Research Design ................................ ................................ ................................ .............. 66 Data Collection ................................ ................................ ................................ ................ 67 Biomass measurement ................................ ................................ .............................. 67 Soil sampling ................................ ................................ ................................ ............ 68 C arbon Stock Estimation ................................ ................................ ................................ 68 Biomass C stock ................................ ................................ ................................ ....... 69 Soil C stock ................................ ................................ ................................ .............. 71 Statistical Analysis ................................ ................................ ................................ .......... 71 Results ................................ ................................ ................................ ................................ ..... 72 C Stock in Biomass and Soil ................................ ................................ ........................... 72 Total C Stock ................................ ................................ ................................ ................... 72 Relationship between Biomass C and Soil C ................................ ................................ .. 73 Discussion ................................ ................................ ................................ ............................... 73

PAGE 8

8 5 SOIL CARBON SEQUESTRATION IN DIFFERENT PARTICLE SIZE FRACTIONS AT VARYING DEPTHS UNDER AGROFORESTRY SYSTEMS IN MALI .................... 83 Introduction ................................ ................................ ................................ ............................. 83 Research Questions ................................ ................................ ................................ ................. 85 Materials and Methods ................................ ................................ ................................ ........... 85 Research Design ................................ ................................ ................................ .............. 86 Soil Preparation and Analyses ................................ ................................ ......................... 87 Soil fractionation ................................ ................................ ................................ ...... 87 C isotopic ratio ( 13 C / 12 C) measurement ................................ ................................ ... 88 Statistical Analysis ................................ ................................ ................................ .......... 89 Results ................................ ................................ ................................ ................................ ..... 90 Soil Characteristics ................................ ................................ ................................ .......... 90 Whole Soil C ................................ ................................ ................................ ................... 91 C in Soil Fractions ................................ ................................ ................................ ........... 92 Isotope Analysis of Whole Soil C ................................ ................................ ................... 93 Isotope Analysis of C in Soil Fractions ................................ ................................ ........... 94 Relationship s of Data Sets ................................ ................................ ............................... 94 Discussion ................................ ................................ ................................ ............................... 95 6 SOCIOECONOMIC ANALYSIS OF THE CARBON SEQUESTRATION POTENTIAL OF I MPROVED AGROFORESTRY SYSTEMS IN MALI, WEST AFRICA ................................ ................................ ................................ ................................ 117 Introduction ................................ ................................ ................................ ........................... 117 Research Question s ................................ ................................ ................................ ............... 118 Materials and Methods ................................ ................................ ................................ ......... 119 Social Survey of Fodder Bank Farmers ................................ ................................ ......... 119 Local Market Survey ................................ ................................ ................................ ..... 120 Types of Analysis ................................ ................................ ................................ .......... 121 Cost benefit analysis (CBA) ................................ ................................ .................. 121 Sensitivity analy sis ................................ ................................ ................................ 127 Risk modeling ................................ ................................ ................................ ........ 128 Results ................................ ................................ ................................ ................................ ... 129 Demographic Characteristic s of Target Population ................................ ...................... 129 Cost Benefit Analysis: Best Guess Scenario of the Live Fence and the Fodder Bank 130 Sensitivity Anal ysis ................................ ................................ ................................ ....... 132 Risk Modeling and Simulation ................................ ................................ ...................... 132 Discussion ................................ ................................ ................................ ............................. 134 7 SUMMA RY AND CONCLUSIONS ................................ ................................ ................... 148 C Sequestration P otential ................................ ................................ ................................ ...... 148 Biophysical Potential ................................ ................................ ................................ ..... 148 Socioecomic Potential ................................ ................................ ................................ ... 150 C onclusions ................................ ................................ ................................ ................... 152

PAGE 9

9 Agroforestry Adoption for C sequestration in the Study Region ................................ ......... 152 Limiting Factors ................................ ................................ ................................ ............ 152 Favorable Factors ................................ ................................ ................................ .......... 153 Implications for Agrofores try ................................ ................................ ............................... 154 Future Research ................................ ................................ ................................ .................... 155 APPENDIX A SOCIAL SURVEY QUESTIONNAIRE FOR FODDER BANK OWNERS ...................... 156 B COST BENEFIT ANALYSIS (CASH FLOW) OF LIVE FENCE ................................ ..... 163 C COST BENEFIT ANALYSIS (CASH FLOW) OF FODDER BANK ................................ 167 LIST OF REFERENCES ................................ ................................ ................................ ............. 171 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 184

PAGE 10

10 LIST OF TABLES Table page 2 1 Common tree and shrub species found throughout the West African Sahel ..................... 29 2 2 Main productive functions of agroforestry parklands ................................ ........................ 31 3 1 Summary of various biomass C measurement approaches used commonly in C sequestration studies ................................ ................................ ................................ .......... 59 3 2 Aboveground time averaged C stock in different ecosystems and agroforestry prac tices ................................ ................................ ................................ ............................. 6 0 4 1 Characteristics of the villages where the experimental plots were set up in Sgou region, Mali ................................ ................................ ................................ ........................ 76 4 2 Characteristi cs of the experimental plots (three plots average) for five selected land use system s in Sgou region, Mali ................................ ................................ ..................... 76 4 3 Estimated biomass C (above and below ground) stock values of each plot and th ree plots average of five selected land use systems ................................ ................................ 77 4 4 Total C stock (biomass C + soil C of different depth) of five selected land use systems. ................................ ................................ ................................ .............................. 77 5 1 Soil profile characteristics for plots of the five land use systems used in the study at Sgou Region, Mali ................................ ................................ ................................ ......... 101 5 2 13 C value s of whole soil and three fraction sizes from five studied land use systems, at Sgou Region, Mali ................................ ................................ ................................ ...... 102 6 1 Demographic characteristics of the target population in Sgou, Mali ............................. 138 6 2 N et Present Value (N PV ) B enefit Cost Ratio (B CR ) and Internal Rate of Return ( IRR ) of the live fence and the fodder bank projects in the three different scenarios (without C sale, with C sa le by the ideal accounting method, and with C sale by the tonne year accounting method) in Sgou, Mali ................................ ............................... 138 6 3 NPV sensitivity of the live fence project and the fodder bank project to the chan ge of an input variable in Sgou, Mali ................................ ................................ ...................... 139

PAGE 11

11 LIST OF FIGURES Figure page 2-1 Map of West Africa with ecological zones and isohyetal lines .........................................32 2-2 Standardized annual Sahel rainfall (June to October) from 1898 to 2004 .........................33 2-3 Seasonal landscape contrast of the WAS ........................................................................... 33 2-4 Distribution of soil orders (USDA soil taxonomy) in West Africa ...................................34 2-5 Parkland system in Sgou, Mali.........................................................................................35 2-6 Allowing the cattle to roam freely on the landscape during the dry season ......................35 4-1 Location map of A: Mali; B: Mali showing its land-locked nature: C: Map of Sgou region ......................................................................................................................78 4-2 Faidherbia albida parkland in Togo village ......................................................................79 4-3 Vitellaria paradoxa parkland in Dakala village.................................................................79 4-4 Live fence system in Dougoukouna village .......................................................................80 4-5 Fodder bank in Dakala village ...........................................................................................80 4-6 Abandoned land just outside of Diamaribougou village....................................................81 4-7 Aboveground and belowground C stock per ha of five selected land-use systems ...........82 5-1 Soil sampling, Sgou, Mali ..............................................................................................103 5-2 Soil pits dug in plots of the five land-use systems studied in Sgou region of Mali .......104 5-3 Model of soil organic matter dynamics ...........................................................................105 5-4 Whole soil C content of three depth classes (0 10 cm, 10 40 cm, and 40 100 cm) in different land-use systems in Sgou, Mali ............................................................106 5-5 Soil C content of three particle size fractions in three depth classes (0 10 cm, 10 40 cm, and 40 100 cm) under five land-use systems in Sgou, Mali ...........................107 5-6 Whole soil C, divided into C3 plants (trees)origin and C4 plants (crops)-origin, in different soil layers up to 100 cm depth, in five land-use systems in Sgou, Mali .........108 5-7 Soil C in three fraction sizes divided into C3 plants-origin and C4 plants-origin in different soil particle-size fractions under different land-use systems in Sgou, Mali ....109

PAGE 12

12 5 8 Linear regression between silt + clay content of soil and whole soil C content in three depth classes across five land use systems in Sgou region of Mali ............................... 112 5 9 Linear regression betwe en silt and clay content of soil and C in soil particles of <53 depth classes across five land use systems in Sgou, Mali ................... 114 5 10 Linear regression between C derived from C3 plants and C in the large soil particles ( 250 2000 at 0 10 cm soil depth across five land use systems of Sgou region, Mali. ................................ ................................ ................................ ..................... 116 6 1 Social survey with farmers in Sgou, Mali ................................ ................................ ...... 140 6 2 Simulated NPV probability distribution of the live fence project (without C sale) ......... 140 6 3 Simulated net benefit (total costs total revenues in each year) of the live fence project (wit hout C sale) ................................ ................................ ................................ .... 141 6 4 Simulated NPV probability distribution of the live fence project (with C sale by the ideal accounting method) ................................ ................................ ................................ 142 6 5 Regression sensitivity analysis for NPV of the live fence project (with C sale by the ideal accounting method) ................................ ................................ ................................ 143 6 6 Simulated NPV probability distribution of the fodder bank proj ect (without C sale) ..... 144 6 7 Simulated net benefit (total costs total revenues in each year) of the fodder bank project (without C sale) ................................ ................................ ................................ .... 145 6 8 Simulated NPV probability distribution of the fodder bank project (with C sale by the ideal accounting method) ................................ ................................ ........................... 146 6 9 Regression sensitivity analysis for NPV of the fodder bank project (with C sale by the ideal accounting method) ................................ ................................ ........................... 147

PAGE 13

13 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degr ee of Doctor of Philosophy CARBON SEQUESTRATION POTENTIAL OF AGROFORESTRY SYSTEMS IN THE WEST AFRICAN SAHEL : AN ASSESSMENT OF BIOLOGICA L AND SOCIOECONOMIC FEASIBILITY By Asako Takimoto December 2007 Chair: P. K. Ramachandran Nair Major: Forest Resource s and Conservation In recent years, carbon (C) sequestration potential of agroforestry systems has attracted attention an option for mitigating green house gasses. Although the possible benefits of agroforestry in carbon ( C) sequestration have been conceptually discussed, field measurements to validate these concepts have not been undertaken to any significant extent. I n addition to the traditional agroforestry systems, i mproved practic es and technologies are now being expanded into the dr y regions such as the West African Sahel for perceived benefits such as arresting desertification, reducing water and wind erosion hazards, and improving biodiversity Thus, it is imperative to investi gate C sequestration potential of agroforestry practices in these regions. M y research hypothesizes that the tree based systems will retain more C in the systems both above and below ground than tree less land use systems. By joining the C credit market the landowners c ould sell the C sequestered in their agroforestry systems. My research consisted of th ree components. The first examined C (biomass + soil) stored in five target land use systems: two traditional parkland system s involving Faidherbia a lbida and Vitell a ria paradoxa trees as the dominant species, two improved agroforestry systems (live

PAGE 14

14 fence and fodder bank ) and land that is out of cultivation ( abandoned or degraded) in the S gou Region, Mali. The second component involved a study of so il C dynamics of these systems : the extent of soil C storage/accumulation by trees and stability of the C accumulated were investigated. In the third component, socioeconomic feasibility of the agroforestry systems was examined in the context of C sequest ration and C credit sale. Research results show that t he selected agroforestry systems have the potential for sequestering more C both above and belowground than in tree less land use systems, and that the trees tend to contribute to storing more stable C in the soil. A mong the selected land use systems, live fence and fodder bank are more suitable to start as agroforestry C sequestration projects than the traditional parkland systems for smallholder farmers in the studied region Between the two impro ved systems, live fence has higher C sequestering potential per unit area and is economically less risky than fodder bank s. Adopting these systems on cultivated land rather than on abandoned land is likely to sequester more C and be more profitable. Sinc e parklands are traditionally practiced they are not likely to qualify as a new C sequestration project soon Nevertheless, F. albida trees are more attractive than V. paradoxa trees in terms of C sequestration potential. These results can be used for development of recommendations and guidelines on selection of land use systems and species and their management, for planning successful C sequestration projects in the West African Sahel.

PAGE 15

15 CHAPTER 1 INTRODUCTION Background It is widely accepted that c urrent global climate change or global warming is the most serious environmental issue affecting human lives. G lobal warming refers to the increase in the average temperature of the Earth's near surface air and oceans in recent decades and its projected continuation I t is brought about primarily by the increase in atmospheric concentrations of the so called greenhouse gases (GHGs). GHGs are components of atmosphere contributing to the green house effect, the process in which the emission of infrared radiation by the atmosphere warms a planet s surface. The Intergovernmental Panel on Climate Change (IPCC) established by the U nited N ations (UN) to evaluate the risk of climate change concludes in its most recent report that most of the observed increa se in globally averaged temperatures since the mid 20th century is very likely due to the observed increase in anthropogenic greenhouse gas concentrations (IPCC 2007). The Kyoto Protocol to the United Nations Framework Convention on Climate Change (UNFCC ) is the first and so far the largest international agreement to stabilize GHG concentrations Carbon dioxide (CO 2 ) is a major GHG and its concentration build up is accelerated by human activities such as burning of fossil fuels and deforestation. One of t he approaches to reducing CO 2 concentration in the atmosphere called biomass carbon (C) sequestration is to it in forest and forest soils by trees and other plants through photosynthesis. This concept became widely known because the Kyoto Protoc ol has an approach called Land Use, Land Use Change and Forestry (LULUCF), which allows the use of C sequestration through afforestation and reforestation as a form of GHG offset activities. The Marrakesh Accords in 2001 determined more detailed rules of LULUCF and added forest management, crop management,

PAGE 16

16 grazing land management, and revegetation as LULUCF activities. This enables agroforestry to be an activity of C sequestration under the Kyoto Protocol, and since then, C sequestration potential of agro forestry systems has attracted attention from both industrialized and developing countries (Albrecht and Kandji 2003 ; Makundi and Sathaye 2004; Sharrow and Ismail 2004 ). This became particularly relevant because of an arrangement called Clean Development Mechanism (CDM) under the Kyoto Protocol which allows industrialized countries with a GHGs reduction commitment to invest in mitigation projects in developing countries as an alternative to what is generally more costly in their own countries. Since a gro forestry is mostly practiced by subsist ence farmers in developing countries there is an attractive opportunity for those farmers to benefit economically from agroforestry if the C sequestered through agroforestry activities are sold to developed countr ies ; it will be an environmental benefit to the global community at large as well. Rationale and Significance The IPCC Report (2000) estimates that 630 million ha of unproductive croplands and grasslands could be converted to agroforestry worldwide, with the potential to sequester 0.391 Pg of C (1 Pg = petagram = 10 15 g = 1 billion ton ) per year by 2010 and 0.586 Pg C per year by 2040. The credibility of conceptual models and theoretical foundations of the possible benefits of agroforestry in C sequestration have been suggested: agroforestry ha s C storage potential in its multiple plant species and soil, high applicability in agricultural land, and indirect effects such as decreasing pressure on natural forest or soil erosion (Nair and Nair 200 3; Lal 2004 a ; M ontagnini and Nair 2004 ) Field measurements to validate these concepts and hypotheses, however, have not been undertaken to a significant extent. Some studies of specific agroforestry practices proved the potential of C sequestration and its benefits, s uch as the Indonesian homegarden systems ( Roshetko et al. 2002 ; Schroth et al. 2002 ). B ut very few such studies have been

PAGE 17

17 reported regarding C sequestration potential of agroforestry systems in semiarid and arid region s In addition to already existing i ndigenous agroforestry systems, improved practices and technologies are now being expanded into these dr y regions for perceived benefits such as arresting desertification, reducing water and wind erosion hazards, and improving biodiversity ( Droppelmann et al. 2000 ; Gordon et al. 2003 ). In this scenario, it is imperative that C sequestration potential of agroforestry practices in these regions is investigated. Considering that the ecological production potential of these dry ecosystems is inherently low co mpared to that of systems can contribute if at a l l to C sequestration in such regions is in itself an important issue. This study was conducted in Mali, s ituated in the West African Sahel (WAS) one of the largest semiarid regions of the world Considering the large extent of area of the region (approx. 5.4 million km 2 ), results of studies of this nature are likely to have wide applicability; yet, such stu dies have been rare possibly because of the relative backwardness of the region in terms of economic development and therefore research facilities and infrastructure. Needless to say, such studies are important because of their relevance in the context o f C credit sale under CDM. The WAS is one of the most environmentally vulnerable and poorest area s in the world. If the majority of the people who are subsisten ce farmers can receive even small amount s of C payments through their agroforestry practices, it would be a substantial contribution to their economic welfare and the overall development of the region. Thus, an analysis of the C sequestration potential of v arious agroforestry practices (traditional and newly introduced ) in the region is timely. R esearch Questions and Objectives To address the issues discussed above four research questions are raised :

PAGE 18

18 1. H ow much C is stored in different agroforestry systems aboveground and belowground ? 2. How do trees contribut e to C storage in the soil, and how labil e is th is C? 3. W hat is the overall relative attractiveness of each of the selected agroforestry systems considering its C sequestration potential in the context of its biological potential, economic profitability, and social acceptability? 4. If carbon credit market s were introduced under CDM, would adoption of agroforestry provide more profits to land owners? If yes, how much? Dissertation Overview This dissertation is presented in seven chapters. Following this introductory chapter (Chapter 1), Chapter 2 des cribes the natural environment of the WAS, the study region, in terms of its climate, vegetation, soil taxonomy etc T land use systems in general and agroforestry systems in particular, are also described. Chapter 3 presents the literature re view summarizing the method s used to estimate the C sequestration potential in agroforestry systems, as well as the current stat e of knowledge on C sequestration potential in the WAS The possibilities and limit ation s in the region, current research tren d s and future research needs are also included Chapter 4 pr esents the result s of C stock measurement s and a comparison of five selected land use systems (four agroforestry systems and one degraded land) in the S gou region, Mali. Methodologies and resu lts of measuring both biomass C and soil C are presented. Total C storage of each system is compared and discussed. Chapter 5 examines soil C measurements in more detail based on analyses of soil samples drawn from different depths from each of the five selected land use types, and discusses influence of trees and land management on soil C sequestration and stability of soil C Chapter 6 presents a socioeconomic feasibility analysis of two improved agroforestry systems in the study region ; results of cos t / benefit and sensitivity analysis are presented both with and without C sale scenarios A risk assessment using a simulation program gives insight in to how introducing agroforestry in the study region might

PAGE 19

19 economically affect local household s Chapter 7 gives a synthesis, conclusions and recommendations for future research and development efforts.

PAGE 20

20 CHAPTER 2 THE WEST AFRICAN SAHEL: GENERAL LAND USE AND AGROFORESTRY Description of the Region The Sahel is a transition zone between the hyper arid Sahar a to the north and the more humid savannas and woodlands to the south. The west part of the Sahel region (West African Sahel: WAS) includes nine countries wh o are members of the Interstate Committee for D rought C ontrol in Sahel (CILSS); these are Burkina Faso, Cape Verde, Gambia, Guinea Bissau, Mali, Mauritania Niger, Senegal, and Chad. The area covers about 5.4 million km 2 with over 500 million inhabitants. Its vegetation mostly consists of bushes, herbs and small trees and does not offer year round harvests. The main characteristics of the WAS include : 1) irregular and little predictable rainfall; 2) p redominance of a griculture and animal husbandry: more than half of the inhabitants are farmers and agriculture contributes more than 40 % to the Gros s Domestic Product (GDP) ; and 3) h igh demographic growth (around 3 %) and high urban growth (around 7 %) (USGS 2007 ) Climate The isohyet lines of the region are almost parallel to the latitude and divide the WAS into three sub groups: Sahelo S aharan, Sa helian, and Sudano Sahelian zones ( Figure 2 1 ). R ainfall in the region varies from 200 to 2500 mm per year with the vast majority of the region receiving between 350 to 800 mm, and is characterized by year to year and decadal tim e scale variability; there were extended wet periods in 1905 09 and 1950 69, and extended dry periods in 1910 14 and 1968 1997 ( Figure 2 2 ). The most recent drought that began in late 1960s caused the severe famine in th e 1970s Since 1997, the rainfall recovered somewhat, but the annual rainfall of the recent years was still below the pre 1970 level (= ~ 540mm), except 1994, 1999, and 2003 (Dai et al. 2004). Although the length of the rainy season varies with latitude a nd local

PAGE 21

21 conditions, it is generally restricted to a period of three to five months from April to October During this period, there is an average of 24 rainfall events, 10 to 12 of which occur in August. Rainstorms are rarely prolonged, usually lasting no more than one or two hours. Rainfall intensities range from 5 to more than 50 mm per event (Gritzner 1988). The rainy season is followed by an extended dry season where the vegetation cover changes drastically ( Figure 2 3 ) The monthly mean temperature of the region is 26 27 C with a monthly mean maximum of 34 36 C and monthly mean minimum 21 23 C Temperature abnormalities are relatively low for the area as a whole (+0.7 C to 0.6 C ), but may be greater in indivi dual places (Littmann 1991). Vegetation The WAS contains three generalized phytogeographical divisions corresponding to the climate zones ( Figure 2 1 ) : (i) the northerly Sahelo Saharan zone, or grass steppe, between the 100 and 2 00mm isohyets; (ii) the Sahel proper, or tree steppe, between the 200 and 400 mm isohyets; (iii) the southerly Sudano Sahelian borderlands, or shrub savannah, extending to the 8 00 mm isohyets Savanna plants are renowned for their well developed root syst ems, penetrating deeply into the soil. Herbaceous plants, mostly perennials, always have an extensive root system, often forming a close mat of rootlets in the upper layers of the soil. Most of the roots are located within the upper 30 cm of soil (Bourli ere 1983). Grasses in the steppe grow in the very short growing season (60 90 days) with narrow leaves in circles or basal rosettes. One of the most common grass species throughout the WAS is Cenchrus biflorus This prickly, short lived grass is the fo od of choice for the herds that graze throughout the Sahel. Mature grass has sharp bristles; but ensiling softens them, so that it can also be used as silage (FAO 1991). Other common grass species in steppe such as Schoenefeldia gracilis Elionorus elega ns Borreria spp.,

PAGE 22

22 are also used as fodders In the south, where the savannah replaces the steppe the tall perennial grasse s such as Andropogon gayanus as well as annual grasses with long cycles such as Pennisetum pedicellatum Andropogon pseudapricus a nd Diheteropogon hagerupii are are common. These grasses grow rapidly up to 2.5 m in height, but natural bush fires control the reserves. Some of the se species are introduced as ornamental or fodder species in the US ( Pennisetum pedicellatum, called Kyasu ma grass) and Australia ( Andoropogon gayanus ), and because of their rigorous spread, they are invasive species. Although most tree and shrub species are found both in steppe and savannah ( Table 2 1), t he woody vegetation beco me more and more divers e and dense as one goes south The trees in the WAS are usually low branched and may ramify from their base. Crowns are generally very wide, and much more developed than the bole. The thickness of the bark has been interpreted as affording protection against repeated bush fires. Spines and thorns on branches are also frequent, which prevent reducing water loss through evaporation. It may afford some protection against browsing by large mammals, but does not prevent foliage browsi ng. Soil D etailed information on the soil resource base of the WAS is inadequate for most research purposes. In most countries, farm level information and detailed soil maps are non existent. Still, in 1977, Food and Agricultural Organization (FAO) of United Nations (UN) and UN Educational and Scientific Organization (UNESCO) formed soil map of Africa, by aggregating specific soil mapping units to form soil regions that corresponded roughly to Africa's major ecological regions. Natural Resources Conser vation Service (NRCS) of the United States Department of Agriculture (USDA) had a pedon database with more than 400 pedons from Africa. With published national soil survey reports, NRCS translated the legend of the UN Soil Map of the Africa into Soil Taxo nomy Map ( Figure 2 4 )

PAGE 23

23 The northern WAS adjacent to rocky Sahara desert to the north, is dominated by Entisols and in some part s by Aridisols. Most of the Entisols in the area ha ve an aridic soil moisture regime and are formed on sandy or loamy deposits. Psamments are present as fingerings of Sahara in zones with ustic or udic soil moisture regimes (Eswaran et al.1996) Vertisols occur locally in some places along the rift valley of the Niger River and around Lake Chad. At lo wer latitudes within the WAS Alfisols are extensive ly spread In general, the wind blown sand from the Sahara desert has buried many of the former Oxisols and Alfisols/Ultisols ; thus, soils in the WAS characteristically have very sandy top soils and a lo w activity clay subsoil In terms of soil quality for agricultural use, soil moisture stress is perhaps the overriding constraint in much of the WAS It is not only because of the low and erratic precipitation but also of the ability of the soil to hold and release water. A large part of northern WAS (Entisols and Aridisols) has low available water holding capacities ( AWHC ) <25 mm And southern part of the WAS is made up of soils with medium AWHC (24 100 mm) mainly Alfisols and Ultisols. Salinity and alkalinity are other problem s affecting agriculture. The extremely acid soils, which are mainly the acid sulphate soils occupy area s around the Niger delta. Some part s of Alfisols (close to southern Ultisols) have acid surface and subsurface horizon s, which, coupled with the moisture stress conditions, makes these soils extremely difficult to manage for productive use under low input conditions. The annual additions of dust from the Sahara brought by the Harmattan winds ( a dry and dusty wind blowing south off the Sahara into the Gulf of Guinea during the dry season ) raise the pH and base saturation of the surface horizons; although the changes are less acute than the east ern part of the Sahel where subsoil acidity is a problem (Tiessen et al. 1991)

PAGE 24

24 In addition to the moisture stress and alkalinity /acidity, there are several other soil related constraints common in the WAS contributing to low productivity These include: 1 ) inherently low nutrient storage capacities (cation exchange capacities, CE Cs) due to the low activity kaolintic clay minerals present or the overall low clay contents, 2) low equilibrium soil organic matter levels due to intensive cultivation without adequate biomass return and high surface soil temperatures, 3) the presence of large amounts of free aluminum and iron oxides which reduces the availability of phosphate to plants ( Gritzner 1988 ; de Alwis 1996) Traditional Farming Systems and Agroforestry in the WAS The traditional farming systems in the WAS are rain fed low externa l input operations Farmers use traditional agricultural methods: use of domestic wastes, farmyard manure, crop rotations, and the incorporation of trees on farmlands There is a considerable variety of crops grown in Sahelian agricultural systems, inclu ding: grains, such as millet ( Pennisetum glaucum ) sorghum ( Sorghum bicolor ) fonio ( Digitaria exilis ) rice ( Oryza glaberrima and Oryza sative ) sesame ( Sesamum indicum ), and safflower ( Carthamus tinctorius ); garden crops, such as eggplant ( Solanum melong ena ), broad beans ( Vicia faba ), okra ( Abelmoschus esculentus ), carrots ( Daucus carota ), chick peas ( Cicer arietinum ), pigeon peas ( Cajanus cajan ), cowpeas ( Vigna unguiculata ), ground nut ( Arachis hypogaea ), yams ( Dioscorea spp.), calabash ( Lagenaria sicera ria ), leeks ( Allium ampeloprasum ), melons ( Cucurbitaceae Family ), etc. Cultivated tree crops includ ing dates ( Phoenix dactylifera ), figs ( Ficus spp .), lemons ( Citrus spp .), mulberries ( Morus spp .), and various gums ( Acacia spp .) are also common (Gritzner 1988 ; ICRISAT 2007 ).

PAGE 25

25 Traditional Agroforestry Practices Bush fallow/shifting cultivation Shifting cultivation refers to the land management practice where a period of cropping (cropping phase) is alternated with a period in which the soil is rested (fall ow phase). This system has been traditionally practiced in the WAS, as well as other tropical and semi tropical regions of the world (Nair 1993). First, t he clearing is done using axes or machetes and only herbaceous plants, saplings and undergrowth are cut. When the cut material is dried and burned the cleared area is planted with crops like yams, sorghum, millet, maize ( Zea mays ) and cassava ( Manihot esculenta ) The land is cultivated for one to four years after which it returns to fallow. The regr owth of natural vegetation rejuvenates the soil through nutrient cycling, addition of litter and suppression of weeds (Ferguson 1983) In general, the fallow phase is much longer than the cropping phase. However, recent rapid population growth in the WA S countries ( from 2.5 to 3.0 %) requires additional cultivated land, often at the expense of fallow and pastureland. Over the years, the fallows became greatly reduced both in area and duration, putting in jeopardy the return of vegetative cover for the b uild up of soil fertility (Kaya 2000). Parkland system Another traditional land use system, sometimes overlapped with tree combined fallow system, is Parklands are generally understood as landscapes in which mature trees occur scattered in cultivated or recently fallowed fields ( Boffa 1999). Farmers grow crops around and underneath of the trees ( Figure 2 5 ) These trees are selectively left or regenerated by farmers because of the variety of functions (mostly non timber use) such as food and medicine ( Table 2 2 ) Parkland trees can also contribute to temperature amelioration and to prevention of soil erosion (Jonsson et al 1999). Parklands occupy a v ast

PAGE 26

26 land area, representing a large part of the agricultural landscape under subsistence farming in the WAS and it is the predominant agroforestry system. For example, the agroforestry parkland system occupies about 90 % of the agricultural land area in M ali (Ciss, 1995) and in Burkina Faso, parklands are found throughout settled zones where agriculture is practiced. Parklands are most often characterized by the dominance of one or a few tree species. Species composition is generally more diverse and variable, however, in areas located farther away from villages and only occasionally cultivated. Common species in the WAS are Acacia senegal, Adansonia digiata Anogeissus leiocarpus Balanites aegyptiaca Bombax costatum Borassus aethiopum Ceiba penta ndra Diospyros mespiliformis Elaeis guineensis Faidherbia albida Hyphaene thebaica Lannea microcarpa Parkia biglobosa Sclerocarya birrea Tamarindus indica Vitellaria paradoxa Vitex doniana and Ziziphus mauritiana ( Tab le 2 2 ) ( Boffa 1999) Improved Agroforestry Practice s The expansion of rain fed agriculture results in soil erosion through the removal of vegetative cover and physical disturbance. Wind and water erosion is extensive in many parts of the WAS Practica lly every country of Africa is prone to desertification, but the Sahelian countries at the southern fringe of the Sahara are particularly vulnerable (Reich et al. 2001) Soil nutrients are removed through crops, erosion, and leaching by rainfall, without replenishment by additions or regeneration under natural fallow. Inappropriate tillage and cultural practice reduce soil infiltration and retention of water, which further degrade the land (de Alwis 1996). Also, deforestation accelerate s the land degrada tion as trees and shrubs are cut to satisfy the construction, fuel, and fodder requirements of the cultivators and their livestock. In the WAS, farmers/pastoralists usually graze their animals in the open area without any control ( Figure 2 6 ). Degraded land spreads as these animals go further after eating the vegetation around the villages.

PAGE 27

27 Consequently, forest and woodland areas are rapidly declining by an estimated 1 .5 % per year on average of West African countries (FAO 2000 ). Prevention of land degradation by controlled grazing and afforestation is often discussed and tried sporadically throughout the WAS as project s financed mainly by international donor communities and agencies (Oba et al. 2000) However, local particip ation has often been short lived and management not successful because little consideration was given to why farmers keep brow sing the animals and do not protect or grow trees. Gradually, there has been a growing awareness that trees be regarded as an int egral component of an overall farming system and that a complex decision making environment with interdisciplinary interactions is needed (Boffa 1999) Adoption of improved land use systems such as agroforestry has been recommended and tried for r ehabilita tion of the degraded soils in various parts of the WAS (Roose et al. 1999 ; Lal 2004 a ). No till farming and improved fallow involving short rotation woody and/or other perennial species are increasingly studied Improved fallow rests land from cultivation as in natural fallow s but the vegetation comprises planted and managed species of leguminous trees, shrubs, and herbaceous cover crops. These vegetation and the roots are expected to reduce the soil nutrient loss or even to replenish them both chemical ly and physically, and to sustain crop production with shorter fallow period (Bationo et al. 2000; Kaya and Nair 2001) Farming systems that promote organic manure inputs (including litters from woody plants) and tree cropping systems have also been tried (Breman and Kessler 1997). As such, agroforestry practices involving incorporation of woody plants (both indigenous and exotic species) on cultivated land as intercrops fences, shelter belts, and/or fodder resources are recognized as a

PAGE 28

28 major technique t o ameliorate the s preading land degradation in the WAS. Detail s of the improved agroforestry practices being introduced in the study region are described in Chapter 4.

PAGE 29

29 Table 2 1. Common tree and shrub species found throughout the W est African Sahel S pecies Botanic Description Functional Use Acacia spp. One of the most common species in the Sahel. Deep root system with feathery leaves protecting barks from dry winds. Species often seen in the WAS are A. nilotica, A. tortillas, A. senegal, and A. seya l. The bark of most acacia produces tannins, which are used in tanning leather. A. senegal produces gum arabic, used in pharmaceuticals and adhesives. Fruits are sometimes consumed as condiments Adansonia digiata ( The b aobab tree ) This drought and fire resistant tree is found throughout the Sahel. With trunks that are often 10 15 m wide, it is one of the largest trees (in terms of trunk width); it grows up to 25 m high. In the dry season, the baobab is completely without leaves, and because of its dis tinguishable shape of branches that look like roots, it is called the "upside down" tree. The bark can be used for rope and cloth, and the trunk, when hollowed out, as a shelter. Fruits and leaves are food sources; especially leaves are very important vi tamin source for the local people. Balanites aegyptiaca Multi branched, spiny shrub or tree up to l0 m tall. Trunk is short and often branching from near the base. Branches are armed with stout yellow or green thorns up to 8 cm long. The fleshy pulp of both unripe and ripe fruits is edible and eaten dried or fresh. The fresh and dried leaves, fruits, and sprouts are all eaten by livestock. Combretum spp. The genus comprises about 370 species of trees and shrubs, 300 of which are native to tropical and southern Africa. C. glutinosum and C. micaranthum are common in the WAS. They are bushes branching from bases, 1 2m tall. The branches are quite strong, and are a useful material for building stools, beds, tool handles, etc. A tea made by steeping the le aves of C. micaranthum in boiling water is a traditional tonic drink and a decoction of the leaves is sometimes used as a medication for malaria. Faidherbia albida One of the fastest growing trees in the WAS. It is deciduous and has the remarkable phenol gy of leaves falling off in rainy season and coming back in the dry season. It can grow up to 30 m tall. Branching stems and an erect to roundish crown. It is a valuable fodder tree for game and domestic animals during dry season. The seeds can be boile d and eaten, but first the skin has to be removed. Also the pods may be dried and ground into flour, which is edible. Guiera senegalensis Perennial bush which is a major component of disturbed parts of bushland in the WAS. Also abundant on roadsides and fallowed lands. The woody part is fragile. Leaves and roots are traditionally used to treat different diseases, particularly malaria and intestinal disorders.

PAGE 30

30 Table 2 1. Continued. Species Botanic Description Functional Use Parkia biglobosa Large tree (up to 20 m) with a dense spreading crown, scaly and creviced grey brown bark. Rather slow growth, begins fruiting after 8 years. Trees are usually isolated. Bark, leaves, flowers and pods have innumerable medicinal and food utilizations, the pods, in pa rticular (husk and pulp) are staple food for humans, stored in households. Foliage contains saponins, but is nevertheless considered palatable to cattle, flowers are rich in nectar and beehives are often placed on the branches. Prosopis africana Small to large leguminous spiny trees (4 20 m), with an open canopy and drooping foliage. It thrives in arid soil and is resistant to droughts, on occasion developing extremely deep root systems. The fruit of the tree is used as fodder, while the seeds are ferme nted to make a protein rich condiment. The products from the hard wood, such as some wooden farm implements, kitchen utensils, and planks for construction, are extensively traded. The tree is a good source of firewood and charcoal. The secondary roots are used as medicine. Vitellaria paradoxa Occurs in a wide latitudinal belt between 5 and 15 N from Senegal to the Central African Republic. The size of the mature tree varies from 7 25 m. The bole is short, 3 4 m, sometimes up to 8 m with diameter le ss than 1 m and with thick bark that protects old trees from bush fires The main product is shea butter (karit) which is extracted from the seeds. It is one of the most affordable and widely used vegetable fats in the Sahel and plays an important role in the economy of the region. The timber is of good quality, termite resistant, and generally very durable, but is normally used only when the tree has passed the fruit bearing age. Compiled from USDA plants database, FAO plants database, and other FAO docum ents.

PAGE 31

31 Table 2 2 Main productive functions of agroforestry parklands Parkland tree function Examples Browse Pterocarpus erinaceus, Pterocarpus lucens, Balanites aegyptiaca, Faidherbia albida, Acacia raddiana, Bauhinia rufescens Famine food Parkland p roducts eaten when crops have failed. Young shoots of Borassus aethiopum eaten as vegetables; fruits and leaves of Ficus gnaphalocarpa and other Ficus species. Fat and oil production Butter extracted from Vitellaria paradoxa ; oil produced from Balanites aegyptiaca, Parinari macrophylla, Lophira alata and Elaeis guineensis Food complement Condiments served with staple cereals. Seeds of Parkia biglobosa, Tamarindus indic, Adansonia digitata and Ceiba pentandra leaves. Handicrafts and clothing Borassus aethiopum (baskets, hats, furniture), fibers from Adansonia digitata, Ficus thonningii and Ficus glumosa Soil fertility Faidherbia albida and, to a lesser degree, Prosopis africana (Nitrogen fixing) Wine production The sap of Elaeis guineensis, Bora ssus aethiopum and Hyphaene thebaca is processed into wine. Wood production Ziziphus spp., Anogeissus leiocarpus (firewood), Borassus aethiopum (construction). Source: (Boffa 1999)

PAGE 32

32 Figure 2 1. Map of West Africa with ecological zones and isohyetal lines. The WAS consists of Sahelo Saharan, Sahelian and Sudano Sahelian zones Source: Famine Early Warning Systems Network ( http://www.fews.net/ )

PAGE 33

33 Figure 2 2. Standardized annual Sahel rainfall (June to October ) from 1898 to 2004. The rainfall data are converted to relative values (standardized) with respect to data from 1898 to1993, such that the mean and standard deviation of the series are 0 and 1 respectively Positive values (orange) are the years with ra infall more than the mean of 1898 1993 data, and negative values (blue) are the years with less rainfall. Sou r ce: Mi t chell (2005) ( A B Figure 2 3. Seasonal landscape contrast of the WAS. P hotos of the same site A) in the dr y season and B) rainy season. Source: USGS ( http://edcintl.cr.usgs.gov/sahel.html ).

PAGE 34

34 Figure 2 4. Distribution of soil orders (USDA soil taxonomy) in West Africa Source: Eswaran et al. (1996)

PAGE 35

35 Figure 2 5. P arkland system in S gou, Mali Trees are scattered in the cultivat ed land, and protected for non timber use. Ox drawn plows are used to till the land to sow the crops upon onset of rains ( Photographed by author) Figu re 2 6. Allowing the cattle to roam freely on the landscape during the dry season after seasonal crops have been harvested is a common feature of the WAS land use system. This often leads to overgrazing (photo from the S gou region, Mali ). When the open lands near the village are depleted of vegetation, farmers are forced to take the animals further away from the village. ( Photographed by author)

PAGE 36

36 CHAPTER 3 LITERATURE REVIEW : C ARBON SEQUESTRATION POTENT IAL OF AGROFORESTRY SYSTEMS IN THE WEST AFRICAN S AHEL (WAS) Overview Carbon (C) sequestration has become a hotly debated and widely researched topic during the recent past. Consequently, voluminous literature is available on the subject. The review in this chapter is limited to issues that are most rel evant to the present study. Following a general overview of the topic, the chapter presents brief descriptions of various methodologies that are currently recognized and/or debated for C measurement and accounting, although not all of these were used in t his study Then studies estimating C storage in agroforestry systems (both biomass C and soil C) in the WAS and other ecoregions are presented. Given that the potential of C sequestration cannot be fully evaluated without integrating both biophysical an d socioeconomic sides of acceptability s ocioeconomic issues related to C sequest ration activities through agroforestry are also discussed. C Sequestration as a Climate Change Mitigation Activity The international response to climate change started in fu ll with the establishment of the United Nations Framework Convention on Climate Change (UNFCCC) in 1992. Five years later, 159 countries signed a treaty called the Kyoto Protocol, which commits the 38 signatory developed countries to reduce their collecti ve greenhouse gas (GHG) emissions by at least 5% compared to the 1990 level by the period 2008 2012. The agreement came into force on February 16, 2005, following its ratification by Russia on November 18, 2004. As of April 2007, a total of 169 countri es and other governmental entities have ratified the agreement A unique characteristic of the Kyoto protocol is that it allows the amount of CO 2 sequestered by forests to be counted towards emission targets.

PAGE 37

37 Tropical forest conversion contributes as mu ch as 25 % of net annual CO 2 emissions globally ( Palm et al. 2004). Removing this atmospheric C and storing it in the terrestrial biosphere is thus, one option for mitigating the emission of this GHG. A recent assessment of Rose et al. (2007), reference d by Intergovernmental Panel on Climate Change (IPCC) s newest report, suggests that land based mitigation agriculture, forestry, and biomass liquid and solid energy substitutes can be cost effective land mitigation options And it can contribute ove r the century 94 to 343 Pg C equivalent of greenhouse gas emission abatement, which is 15 to 40 percent of the total abatement required for stabilization. Agroforestry for C sequestration Under the Kyoto Protocol s Article 3.3, further defined by Marrak esh Accord in 2001, a groforestry was recognized as an option of mitigating GHG s. Since then, the C sequestration potential of agroforestry systems has attracted greater attention from both industrialized and developing countries. It is attractive because of its applicability to a large number of people and area s currently in agriculture, a s well as its perceived potential for reducing pressure on natural forests. Also, Clean Development Mechanism (CDM), defined in Article 12 of the Protocol adds the attr activeness because the CDM provides for Annex I Parties (industrialized countries which have emission reduction goals) to implement project activities that reduce emissions in non Annex I Parties (developing countries) in return for certified emission re ductions (CERs) (UNFCCC 2007) Since agroforestry is traditionally and widely practiced in developing countries, it is feasible/easy options for both developing and developed groups of countries to start as mitigation projects under the CDM. However, as stated by Makundi et al. ( 200 4 ) and several others, estimating the amount of C sequestered by agroforestry poses unique challenges In addition to the complexity caused by diverse factors such as climate, soil type, tree planting densities, and tree mana gement as well as

PAGE 38

38 specific difficulties arising from requirements for monitoring, verification, leakage assessment and the establishment of credible baselines agroforestry estimations are beset by the problem of estimating the area under agroforestry prac tices. Nevertheless, the IPCC (2000) estimate d that 630 million ha of unproductive croplands and grasslands could be converted to agroforestry worldwide, with the potential to sequester 391,000 Mg of C per year by 2010 and 586,000 Mg C per year by 2040. Although the credibility of conceptual models and theoretical benefits has been demonstrated, C sequestration potential is still a little studied characteristic of agroforestry systems (Nair and Nair 2003). More studies examining how much C can be seques tered/stored in various agroforestry systems around the world are needed. S everal s tudies and reviews from different regions of the world have discussed agroforesty s benefits and limitations for C sequestration (Schroeder 1994; Dixon 1995; Albrecht and K andji 2003) but only very few deal with comprehensive comparisons of different practices in each ecoregion. Due to the difficult physical environment and lack of research infrastructure, agroforestry systems in the WAS are one of the least documented to pics regarding C s equestration potential. Lal (1999) estimated the potential for sequestering C in the region wa s, as in most other drylands, fairly low, between 0.05 0.3 Mg C ha 1 y r 1 The estimate, however, include d a variety of uncertainties relate d to future shifts in global climate, land use and land cover, and the poor performance of trees and crops on poor soils in the region. In the WAS, impacts of population pressure over grazing and continuous drought are causing severe land degradation. Co nsequently, biomass C stocks steadily decline within land use/land cover. Opportunities for C gains in the region are, thus, often discussed in the context of agricultural fertility and sustainability of farming systems, which involve agroforestry such as

PAGE 39

39 tree crop livestock integration and fallowing practices (Manlay et al. 2002; Woomer et al. 2004 a ). Methodologies for C Sequestration Measurements Efforts to accurately measure C in forests are gaining global attention as countries seek to comply with agre ements under the UNFCCC. Many methodologies have been put forth to quantify the amount of C in forests (Beer et al 1990; MacDicken 1997; Brown 1999), and are best based on permanent sample plots laid out in a statistically sound design s. This is often q uite difficult in agroforestry systems and is one of the reasons why there are few studies that actually measure the amount of C (Montagnini and Nair 2004) Practically, there are four possible approaches to measur ing the amount of C stored as a result of particular land management practice; 1) Direct on site measurements of biomass, soil C, or C flux, 2) Indirect remote sensing techniques, 3) Modeling, 4) Default values for land/activity based practices ( Table 3 1 ) Most of these approaches were originally developed to estimate the amount of C in forest stands. Several pilot projects are ongoing to ensure that C that is sequestered for the long term in economically viable agroforestry systems is reliably measured. T he facto rs that influence which approach is used in a specific project depends on technical availability budget for the measurement, and size of the land to be estimated. Since most of C mitigation projects are either still in the pilot stage or implemented on a small scale, direct measurement approaches are most commonly used and reported. Direct On site Measurement Direct on site measurement includes field sampling and laboratory measurements of total C in the biomass and soil. These measurements (including in ventory data used for the remote sensing, modeling or default values) are in effect snapshots of C stored at the time of the

PAGE 40

40 inventory. How to calculate/determine the amount of sequestered C over a certain period is another issue, and discussed in the Accounting Methods section. Inventory In general, C in forest or agroforestry systems can be divided into four groups; 1) Aboveground biomass, 2) Belowground biomass, 3) Soil C, 4) L itter fall /crop residue Methods to collect and calculate the sample data from project sites have been standardized by many reports and studies (MacDi c ken 1997; Roshetko et al. 2002). Data for the four C categories are collected by timber cruising and sampling of herbaceous vegetation, soil, and standing litter crop at s ample plots (Shepherd and Montagnini 2001; Brown 2002; Tie p olo et al 2002). Also, for existing forests, many tropical countries have at least one inventory of all or part of their forest area that could be applied for agroforestry systems, although many of the inventories are more than 10 years old and very few have repeated inventories. Data from these inventories can be converted to biomass C depending on the level of detail reported (Brown, 1997). Conversion and e stimation For aboveground biomass, tre es are divided by compartments: leaves, branches and trunks, and measured in dry weight (Beer et al 1990), because each compartment has unique C content and decomposition rate. Although this is the most accurate method, these inventories are often too ti me consuming and cost ly Alternatively, biomass expansion factors or allometric biomass equations are often used, because they require only stem wood information such as diameter at breast height (DBH). These equations exist for practically all forests t ypes of the world, especially in the temperate zone (Sharrow and Ismail 2004). But, because of the very general nature of these equations, they lack accuracy; they are, at best, approximations. For an agroforestry system Shroeder (1994)

PAGE 41

41 used a ratio of total aboveground biomass to stem wood biomass of 2.15 derived from many previous studies. Where tree stocking density was high (>500 trees ha 1 ) and the growth cycle or rotation length was relatively long (>10 years), i.e., for conditions more similar to those for a forest plantation, a ratio of 1.6 was used in the study to estimate total aboveground biomass. Total C content is usually estimated based on the assumption that 45 to 50 % of branch and stem dry biomass is C, and that 30 % of dry foliage biom ass is C (Shepherd and Montagnini 2001; Schroth et al. 2002). Herbaceous vegetation and standing litter are also collected from sample plots and weighed to calculate their C content. It is often assumed in inventories that this vegetation type contributes little to the total biomass C of a forest and it is often ignored. However, the contribution of herbaceous vegetations is often larger in agroforestry systems than in forests such as green manure from trees in natural systems. The amount of litterfall, pruning residues, and crops largely depends on the season and rotation period (Beer et al. 1990). Thus, it is difficult to estimate using general ratios as used in the stem biomass estimation. For belowground C, it is divided into two main categories; root biomass, and soil C (mainly organic matter). Although methods for measuring aboveground biomass are well established, measurement of root biomass is difficult and time consuming in any ecosystem and methods are generally not standardized (Ingram and Fernandes 2001) A review of the literature shows that typical methods include spatially distributed soil cores or pits for fine and medium roots and partial to complete excavation and/or allometry for coarse roots. The distinction between live and dead roots is generally not made and root biomass is usually reported as total. Moreover, sampling depths are not standardized, yet the depth selected in a given study is assumed to capture practically all roots (Brown 2002).

PAGE 42

42 Root biomass is often estimated f rom root:shoot ratios (R/S). It can be calculated by sample plot measurements, but there are also lists of reference data. A literature review by Cairns et al. (1997) included more than 160 studies covering tropical, temperate and boreal forests that rep orted both root biomass and aboveground biomass. The mean R/S based on these studies was 0.26, with a range of 0.18 (lower 25 % quartile) to 0.30 (upper 75 % quartile). The R/S did not vary significantly with latitudinal zone (tropical, temperate, and bo real), soil texture (fine, medium and coarse), or tree type (angiosperm and gymnosperm). Soil C samples should be collected from each layer, dry weighed and analyzed for its C contents by recommended laboratory procedures To calculate C stocks per unit area, the C content in the soil is multiplied by the bulk density of the respective soil layer. By itself, C sequestration in agricultural soils is expected to make only modest contributions globally (e.g. 3 6 % of total fossil C emissions) (Paustian et al. 1997) However, this amount can be significantly varied through management such as fallow phase, erosion, tillage or tree incorporation Indirect Remote Sensing Techniques Even where field measurement methodologies are established, agricultural/ forestry practices are inherently dispersed over a wide geographic area. Staffing costs for monitoring and verification of land use practices over such a wide area could prove to be cost prohibitive. Because direct field measurements can be expensive, th e use of indirect remote sensing techniques is being considered. A range of remote data collection technologies is now available including satellite imagery and aerial photo imagery from low flying airplanes. Sensors that can measure the height of the ca nopy or vertical structure will be needed along with the more traditional sensors on Landsat or Spot satellites in order to improve the ability of remotely sensing biomass (Brown 2002)

PAGE 43

43 A promising advance in remote measurements of forest/agroforest biom ass C is a scanning lidar (a pulsed laser), a relatively new type of sensor that explicitly measures canopy height. This sensor is able to monitor 98 % of the earth s closed canopy forests (Brown 2002). Another promising advance in the remote sensing are a, especially at smaller scales (thus, probably more appropriate for agroforestry systems), couples dual camera digital videos (wide angle and zoom) with a pulse laser profiler, data recorders, and differential GPS (geographical positioning system) mounted on a single engine plane (Brown 2002). The plane flies aerial transects across the area with several fixed altitude to take the images data, and these images are used to create 3D models of the terrain. From these measurements, crown area, tree height, o r number of stems per area of agroforestry systems would be much more easily and accurately estimated. Modeling Since total direct inventory is site specific and can be expensive, another way to lower the cost of estimating C amount is developing a model To date, several models have been developed that simulate C budgets and fluxes at the level of the forest stands. These models range from very detailed ecophysiological models used in climate impact assessment, to very general empirical, descriptive mo dels of C budgets within forest stands. None of these models ha s been widely disseminated, and none of them accepted as a possible standard for C crediting projects so far. One of the most recognized and utilized model s by various projects including ag roforestry projects is CO 2FIX which was developed by researchers of Wageningen University, Universidad Nacional Autonoma de Mexico, Centro Agronomico Tropical de Investigacion y Ensenanza (CATIE), and European Forest Institute (EFI). This model is a user friendly tool for dynamically estimating the C sequestration potential of forest management, agroforestry and

PAGE 44

44 afforestation projects. The model is a multi cohort ecosystem level model based on C accounting of forest stands, including forest biomass, soil s and products (Masera et al. 2003). Another common methodological approach to estimat ing mitigation potential more broadly is known as comprehensive mitigation assessment process (COMAP). Th e COMAP model requires the projection of land use scenarios fo r both a baseline and for a mitigation case. It requires data on a per hectare basis on C sequestration in vegetation, detritus, and forest products, soils and also on GHG emission avoidance activities (Makundi and Sathaye 2004). Default Values for Land/A ctivity Based Practices This approach is the broadest, nation level approach which uses default values for certain land based activities for estimating C storage. A land use based accounting system would focus on the changes in C stocks on managed lands during a defined time period (Dixon et al 1994 a ). Default values would be assigned to a particular tract of land based upon county or regional level research on the average sequestration likely to result from specific agricultural or conservation measure s in that area. Various values could be assigned to such broad land management activities as forest, cropland, or grazing management. Under this approach, field measurements of C storage changes in individual fields would not be necessary. Land use moni toring can be readily measured by remote sensing techniques, eliminating the need for many field inspectors. However, field plots may need to be set up, representing the average or a range of conditions for the entire project area, and used as a reference to provide actual estimates to increase the accuracy of large scale projects. Accounting Methods In order to assert that agroforestry systems are an important C sequestration method, the amount s measured in agroforestry systems must result in long term changes in terrestrial C storage and CO 2 concentrations in the atmosphere (Masera et al. 2003). Thus, the time frame

PAGE 45

45 and C accounting methods are very important ; but they are also often controversial issue s at international negotiations of climate change. Approaches to Assess ing C Sequestration Performance Fluxes of C and flow summation Balancing the annual flux from a source of emissions by uptake in a forestry/agroforestry project is conceptually the simplest way of providing offsets. In this approach offsets are to the C credit market on an annual basis, according to the emissions avoided, relative to the project baseline. However, since fluxes of C associated with forestry/agroforestry are irregular, it may be difficult to match the upt ake by a particular project to an industrial source of emissions. Furthermore, national or international authorities must assign permanent C storage status to project areas, such that the owners are liable for any emissions, as well as eligible for credi ts. Without such status C might be accumulated in the growing phase of the forestry cycle, only to be lost at the end of rotation (Tipper and De Jong 1998). Average changes in the stocks of C The pilot phase of most sequestration projects is assessed on the basis of the long term average increase in the stocks of terrestrial C relative to the baseline (Kursten and Burschel 1993; van Noordwijk et al. 2002), expressed as tC according to the equation: Average net C (C stored in projec t C s tored in baseline) in tC / n (y ea rs) The stock change method calculates the difference in C stocks between a project and its baseline at a given point in time. A key advantage of both methods is that it focuses on the sustainability of changing the stock of C stored in vegetation and soils. However, long term C storage is not easily defined, and there may be considerable argument over the assumptions about risks and possible future changes in management. The timing of the emission reduction

PAGE 46

46 relative to the emissions that are supposed to be offset may be problematic, since there may be a time lag of years to decades between the establishment of the offset project and actual uptake of the C. Cumulative C storage The cumulative storage approach is based on an understanding of C cycle dynamics and radiative forcing of the atmosphere. The total warming effect of a given emission is determined by the cumulative presence of GHG in the atmosphere ; in other words the product of concentration and time. In the case of CO 2 terrestrial and oceanic sinks take up C previously emitted, over time. Assuming the dynamics of the C cycle remain stable, most CO 2 emitted at the present will be absorbed within 100 years, and the cumulative radiative forcing produced by thi s emission will be proportional to the area under the depletion curve, expressed in tC.years. Calculation of this area provides an estimate of the cumulative C storage that would be required to offset an emission of 1 tC at the present time. This method avoids both the necessity of making questionable assumptions about the long term balance of C in forests/agroforests, and the practical difficulties of implementing flux based incentive systems. However, international agreements on the conversion factor f or tC.years per tC emission and the time limit for crediting the effect of a given project are required (Tipper and De Jong 1998) Other accounting methods In addition to these relatively simple conventional methods, alternative approaches have been propos ed to better address the temporal dimension of C storage, such as equivalence adjusted average storage, stock change crediting with ton year liability adjustment, equivalence factor yearly crediting, equivalence delayed full crediting, and ex ante ton year crediting (De Jong 2001). Most of these are based on adopting a two dimensional measurement unit that reflects storage and time, i.e., ton C year. The general concept of the ton year approach is in the

PAGE 47

47 application of a factor to convert the climatic eff ect of temporal C storage an equivalent (equivalent factor) amount of avoided emissions. Technical Problems and Uncertainties There are a number of shortcomings of conventional methods for estimating /accounting the C in a system that need to be considered These include the uncertainties related to future shifts in global climate, land use and land cover, the poor performance of trees and crops, varied environments, pests and diseases such as nematodes. For example, the amount of C remaining belowground a t the end of the tree rotation, and the amount of C sequestered in products created from the harvested wood, including their final disposition are often not included in the accounting methods discussed above (Johnsen et al. 2001). Oren et al. ( 2001 ) repor ted that after an initial growth spurt, trees grew more slowly and did not absorb as much C from the atmosphere as expected They concluded that assessment of future C sequestration should consider the limitations imposed by soil fertility as well as inte ractions with nitrogen deposition. In addition to these uncertainties, there are some concern s about the impacts of agroforestry in other GHGs. The wide scale use of woody legumes which is common in agroforestry systems, might result in release of nitr ous oxide ( N 2 O ) (Choudhary et al. 2002) although it does not seem to be as strong an impact as N fertilization (Mosier et al. 2004) N 2 O is known to have a global warming potential 200 300 times higher than that of CO 2 Similarly, pasture and rice pad dy cultivation in agroforestry systems can produce significant quantities of methane ( CH 4 ) another strong GHG (20 60 times higher impact than CO 2 ), on a global scale (Dixon, 1995).

PAGE 48

48 Biomass C Sequestration Studies in Various Ecoregion s The amount of C sequestered in an agroforestry system depends largely on the nature of the system put in place, the structure and function of which are determined by environmental and socioeconomic factors (Albrecht and Kandji 2003). Other factors influencing C storage include tree species and system management (Delaney and Roshetko 1998; Roshetko et al. 2002). Palm et al. (2004) compared the amount of C stored (above ground) in different ecological system s ( Table 3 2 ) To compare the rotati on of the different land use systems, time averaged C of each system was used. C stocks in the vegetation of the primary forests averaged 300 Mg C ha 1 and that of logged or managed forests ranged from a high of 228 Mg C ha 1 in Cameroon to a low of 93 M g C ha 1 in Indonesia. Time averaged aboveground C for the different land uses ranged from 50 90 Mg C ha 1 in long fallow shifting cultivation and complex agroforestry systems to 30 60 Mg C ha 1 in simple agroforestry systems and most tree plantation s and medium fallow rotations. The se are considerably larger than those for annual crops or pastures. Studies in West Africa In a review of C sequestration in tropical agroforestry systems, Albrecht and Kandji (2003) estimated that agrosilvicultural sys tem could sequester 29 53 Mg C ha 1 in humid tropical Africa. A case study in Cameroon (humid west and central Africa) showed that the cacao ( Theobroma cacao ) agroforest is superior to the alternative food crop production system (slash and burn), both i n C sequestration and below and above ground bio diversity. Total biomass in cacao agroforest was 304 Mg ha 1 compared to crop fields ( 84 Mg ha 1 ) (Duguma et al. 2001). Compared to C gains in the humid tropics, the benefits of agroforestry in the WAS, s uch as parklands or improved fallow seem to be less. A simulation study in Senegal compared the C

PAGE 49

49 gains after 25 years from protecting parkland systems (2.48 Mg C ha 1 ) with that of rotating crops with Luecaena spp fallow planting (6.35 Mg C ha 1 ) (Tscha kert 2004). Although drylands as whole are believed to provide a substantial opportunity for C offsets especially in soil C, because of their large area (47.2 % of land in the world) and low human populations (Lal 2004b) most studies in the S ahel region have concluded that the potential for C gains per unit area through agroforestry is relatively low (Walker and Desankar 2004;Woomer et al. 2004b) compared with other ecoregions In general, i ntroducing trees into agricultural systems is expected to increas e water and nutrient availability because trees can fix nitrogen, retrieve water and nutrients from below the rooting zone of crops and reduce water and nutrient losses from leaching and erosion (Buresh and Tian 1997) This tree effect has been demonstr ated in various agroforestry systems in the semiarid region. However, this added value was lowest where it is most needed, in resource poor environments: the competition between woody plants and crops is strong (Kater et al. 1992; Breman and Kessler 1997) Water constraints are the strongest limitations for C sequestration in the WAS. Several local tree species such as Acacia tortilis Guiera senegalensis Pterocarpus lucens have been planted in grasslands of the region for sequestering C, but their capa city to grow has been shown to be constrained by moisture availability (Woomer et al. 2004 a ). The capacity of exotic dryland tree species to afforest the WAS is also uncertain. Since the moisture and nutrient levels of the study field are expected to be low, the tree growth and the consequent C storing will not be extremely high, either, compared with more moist part s of West Africa. However, the amounts of C sequestered as a result of specific land uses are mostly unknown in the WAS, thus, it is worthwh ile to conduct the research to have a reference data for future C sequestration projects.

PAGE 50

50 Soil C Sequestration Recently, research focused on C sequestration potential in terrestrial ecosystems has been shifting from biomass C to soil C, because soil is r ecognized as an important storage (sink) for C and studies estimating biomass have accumulated for various ecosystem (including agroecosystem and plantation forests), while the dynamics of soil C is much less known. Studies of Soil C Stock and Dynamics T he comparison study of C sequestration potential by Palm et al. (2004) ( Table 3 2 ) also measured soil C storages. On average, 45 Mg C ha 1 were found in the forest systems studied (0 20 cm depth), and 80 100 % of that C st orage in agroforestry systems; 80 % in pastures; 90 100 % in long fallow cycles; 65 % in short term fallows, and 50 % or less in annual crops and degraded grasslands. Rosalina et al. (1997) reported a mean of 104 Mg C ha 1 for home gardens in North Lamp u n g, Indonesia, and th at 58 % of the stock is soil C. The soil C sampling depth of these studies is usually on surface, up to 20cm. Although the surface soil is the major part of which soil organic carbon (SOC) is found, it may not be deep enough to capt ure all the tree roots influence on soil C (Jobbagy and Jackson 2000). But because of the labor intensity and the relatively low soil C density, deeper soil C tends to be ignored. Storage of soil C is rather easy to quantify and/or estimate, but detect ing soil C flux including its turn over time is much more difficult and has not been studied much. Most current models of soil organic matter (SOM) dynamics assume that equilibrium C stocks are linearly proportional to C inputs, i.e. there are no assumpti ons of soil C saturation. Six et al. (2002 a ) questioned the validity of this assumption for projecting longer term SOM dynamics, and developed the proposition that physiochemical characteristics inherent to soils define the maximum protective capacity of SOM pools. Methodologies such as fractionation and C isotope measurements are being developed for quantifying and identifying the characteristics of soil C

PAGE 51

51 dynamics, and are actively discussed in the soil science arena (Balesdent et al. 1998; Del Galdo et al. 2003; Powers and Veldkamp 2005). Soil fractionation : There are various ways to separate SOM into labile and recalcitrant pools, and these methods rely on chemical, physical, or biological separation and many of them are used sequentially in analys es (McLauchlan and Hobbie 2004) Chemical fractionation separates soil C into different resistance class to decomposition by using acid, permanganate(KMnO 4 ), or hot water, (Moody et al. 1997; Ghani et al. 2003). Physical fractionation separates labile an d recalcitrant fractions by either size or density. Sieving is used to size differentiation and flotation with a dense liquid is usually used to measure light fraction which is considered to be labile (Six et al. 1998). Biological separation uses microbe s to separate labile SOC from recalcitrant SOC under controlled temperature and moisture conditions assuming that microbes will mineralize the most labile C first, with recalcitrant C being mineralized later (Alvarez and Alvarez 2000) With applying thes e methods, many examine the impact of land use change on soil C storage and dynamics. For example, it is possible to assess how land use rotation (including fallow) or change of management such as reduction of tillage can effectively protect recalcitrant soil C, so that enhance soil C sequestration (Pikul. et al. 2007; Zibilske and Bradford 2007). 13 C isotope measurement : During photosynthesis, CO 2 fixation of C 3 plants discriminates against the heavier isotope 13 C more than do C 4 plants, which result in different stable carbon isotope composition ( 13 C), 13 C/ 12 C ratio relative to that found in the PDB (Pee Dee belemnite) for their plant material. T his composition v alue of C 3 plants is between 23 and 34 whereas C 4 plants range s from 9 to 17 (Eleki et al 2005). Negative values of 13 C indicate that the plant material is depleted in 13 C compared with the PDB standard. Using th is theoretical

PAGE 52

52 expected difference between the measured 13 C value s, it is possible to calculate the proportion of C4 derived material and C3 derived material in biomass or soil C (Balesdent et al. 1998). This method has been used for soil C research to assess vegetation composition change (Dzurec et al. 1985) or to follow the dynamics (Harris et al. 2001). Mcdonagh et al. (2001) measured how SOM from original vegetation (forest: C3 plants) were diminished after continuou s cultivation of maize ( Zea mays : C4 plants). In agroforestry system, Jonsson et al. (1999) used this method to prove the positive influence of trees (C3 plants) on SOM increase at millet ( Pennisetum glaucum : C4 plants) cropland. Soil C in the WAS In th e WAS, most of the soils have low activity clay with low water retention and are susceptible to soil erosion and compaction as described in Chapter 2. Organic matter content of these soils has been depleted due to overgrazing, agricultural mismanagement deforestation and overexploitation of the natural resources. As a result, soil organic C stock density in West Africa is very low ( 4.2 4.5 kg C m 2 ), compared with the world average (10.9 11.6 kg C m 2 ), and relatively lower even when compared to th e average for Africa (6.4 6.7 kg C m 2 ) (Batjes 2001). S oil degradation is a major obstacle for agricultural productivity and thus sustainable development of the WAS The possibility of enhancing C sequestration through improved soil management has been discussed academically and at international workshops, as part of the search for agroecosystem sustainability in the region. Among s oil nutrients studies in Africa, tree integration into croplands is often recommended for soil amelioration (Onim et al. 19 90; Tiessen et al. 1991; Manlay et al. 2002). Kang et al. (1999) reported Grilicidia sepium and Leucaena leucocephala increased surface soil organic C by 15 % compared to sole crops in a 12 year hedgerow intercropping trial on a Nigerian Alfisol. Parklan d system studies affirm that the soil

PAGE 53

53 under the trees is richer in organic matter content and several cations compared to adjacent tree less sites ( Kater et al. 1992; Jonsson et al. 1999). Soil amelioration by land management in the WAS is, however, often met with limited success when it comes to maintaining or increasing soil nutrient and C stocks. The potential for C sequestration in a given soil and agroecological zone is proportional to the original reserves present under undisturbed conditions. Brem an and Kessler (1997) compared the added values of woody plants in croplands or pastures between Sahel zone and wetter Sudan zone. They concluded possibilities to improve the soil organic matter status are more limited in Sahel, due to resource poor envir onments and competition for water between woody plants and crops or pasture. Because of the slow establishment of a woody plant community and the rapid turnover of organic matter, a long transition period is necessary under the conditions in Sahel. Imp roved fallow studies also suggest that a long period would be needed to amend soil physical conditions of the highly degraded soil in the WAS (Buresh and Tian 1998; Ringius 2002; Kaya and Nair 2004). Soil C sequestration is not recognized as a mitigation means during the first commitment period of the Kyoto Protocol (2008 2012 ) although political pressure to reverse this situation has been growing. In his review of soil C sequestration in Africa, Ringius (2002) stated that s ub S aharan Africa would not profit significantly from soil C sequestration under the Clean Development Mechanism (CDM) as long as the land use pressures due to a rapidly growing population and poverty remain unsolved. Profitability of the soil C sequestration project is un certain, s ince cost benefit studies of the sequestration activities have not been conducted. There is a need to launch long term (>10 yr) field experiments and pilot projects for soil C

PAGE 54

54 sequestration as well as to develop a land resources information system in the WAS, geared towards CDM and/or C sale. Socioeconomic Implications To analyze the socioeconomic feasibility of the agroforestry practices for mitigation projects, analysis tools, i.e. models, are needed. Economic models of different scales used for the stu dies in various ecoregions are summarized here. Although the number of stud ies is small, the case studies and possibility of using these economic models in the WAS are also examined. Economic Models In most studies of C sequestration, agroforestry is reg arded as one of the forest management option s for potential C sequestration T here are few stud ies specifically discussing economic models of C sequestration in agroforestry systems; instead, models designed for managed forests are usually applied (Masera et al 2003). These economic models for accounting C sequestration projects can be categorized into two different spatial scales National/global scale Apart from the C sequestration potential per se of agroforestry systems, the potential for realizing t his assumed benefit depends largely on the availability of land which can be changed to agroforestry from land with less C storage such as agricultural field s Attempts to estimate the global potential for increasing C sinks through land use change had b een conducted at the global, national, and regional level s for more than a decade (Dixon et al 1994 b; Sathaye et al. 2001; Godal et al. 2003 ). T hese studies use simple integrated model structure based on biophysical and economic information In this ki nd of large spatial scale empirical model schemes such as the Holdridge life zone system (LZS) can be used as a guideline for organizing vegetation data (Pfaff et al 2000) For economic factors such as the price of land, cost of land use change, and tim ber price national census information are generally available. I nformation

PAGE 55

55 on land use and trend s of change, collected from national survey s or satellite/remote sensing data, are integrated and the overall economic impacts are examined. Results are pres ented in several different ways. Dixon (1995) presented the potential C storage (over 50 year rotation) and initial project financial costs for agroforestry systems for ecoregions of selected nations in terms of $/Mg C. Some studies estimate total invest ment cost in actual dollar amounts for developed countries to achieve their reduction goals (Baron and Lanza 2000, Godal et al 2003), and others estimate each country s investment in $ ha 1 or internal rate of return (IRR) (%) (Dixon 1995 ; Sathaye et al 2001). Micro/site s pecific scale At this level, more detailed economic estimation is conducted based on data from the specific target (project) field. Various factors of benefits and costs are individually counted. There are three different time frames for counting these costs and benefits of C sequestration projects: point estimates, partial equilibrium estimates, and general equilibrium approaches Many of the point estimate studies provide undiscounted private costs and benefits of the project s impl ementation in $ ha 1 These studies usually count only direct inputs and outputs including land cost. Estimating opportunity costs to compare with other land use option s is often done (Tomich et al. 2002). Most of the se studies reveal little about how c osts might change throughout the project with time or if the project were to expand or be repeated. Thus, the estimates tend to be biased towards the low end The Scolel T project, conducted in Mexico, is one of the few long term and comprehensive econo mic impact studies on this subject It serves as an example agroforestry project for calculat ing the costs related to implementing a C sequestration project in rural environments dominated by resource poor small scale farmers wh o are expected to be major player s in agroforestry worldwide. The study accounts for cost s of project design, the time

PAGE 56

56 required to explain to farmers the project objectives, C related inventories, and the cost of baseline setting, which are often ignored in similar estimate analys is (De Jong et al. 2004). These equilibrium studies usually present the results as discounted costs in $/Mg C with long term average sequestered C as Mg C ha 1 For estimating much longer scale s than one cycle of agroforestry rotation, computer based mode ling methods are usually used. There are several models such as CO2FIX and WaNULCAS for estimating the dynamics of C sequestration over decades to centuries and some are applicable to agroforestry ( Van Noordwijk et al. 2002 ; Masera et al 2003 ). Wise and Cacho (2005) used the WaNULCAS model for ecological estimates combined with their economic model including variables such as local discount rate, firewood price, and labor cost, and simulated the long term economic value of switching land use from agricu ltural system to agroforestry system in Indonesia. They include expected C prices into its economic analysis, and presented net present values (NPV) in $ ha 1 of several different setting (management options). Agroforestry systems in temperate area s ar e usually analyzed in a very different way. Compared with developing countries in the tropics, the C credit sale through forestation is not likely to be economically feasible soon in the temperate area Thus, C sequestration tends to be considered as an environmental benefit (non market value, or subject to receive the subsidies). Studies to estimate the benefits of C sequestration using models that are generally used for these environmental commodities seem to be the current research trends ( Stainback a nd Alavalapati 2000; Alavalapati et al. 2004). Feasibility in West Africa West African countries GHG emissions are currently negligible in global terms, due to the low level of development and industrialization. As a result, emission reduction opportunit ies

PAGE 57

57 remain few, and are mainly in lessening the negative impact on the climate resulting from land use change and deforestation. For C offset projects, however, the risk of shifting from cropping to more C beneficial practices seems to be high, especially for subsistence farmers who have little lands. A study of C sequestration through agroforestry in Senegal found that the costs for resource poor farmers are considerably higher than those of intermediate and richer farmers, because the former often lack t he necessary assets (land, labor, and animals) to switch from current to alternative practices (Tschakert 2007). Many African policy makers and financial institutions express little interest in controlling GHG emissions or adapting to changes in climate. This attitude is based on their experience that, in general, other, more local, environmental problems have more direct influence on their populations than climate change. Senior government officials and most members of civil society do not understand the climate issue very well (Denton et al. 2001) Many development practitioners remain skeptical, arguing that C brokers, national ministries and local leaders rather than needy rural populations will benefit from C projects. An important challenge for th e WAS countries lies in that they need to be more attractive than the other African and developing countries in order to draw and hold investments for C sequestration projects under CDM. As discussed above, the potential C gains in the WAS through agrof orestry has been considered to be unattractive. Synergies between development and climate change response, however, can be an answer. Agroforestry projects which protect soils and result in C sequestration, also provide employment opportunities for loca l farmers (Hardner et al. 2000) Soil C sequestration project through agroforestry could provide a crucial link between three international conventions: the UN Framework Convention on Climate

PAGE 58

58 Change (UNFCCC), the UN Convention to Combat Desertification (U NCCD), and the UN Convention on Biodiversity (UNCBD). As such, incentives can come from a much broader area such as development assistance, other multilateral agreements and sectoral policies on energy and agriculture.

PAGE 59

59 Table 3 1. Summary of various bi omass C measurement approaches used commonly in C sequestration studies. Approaches Tools and methods for data collection Remarks Direct on site measurement Plot sampling, tree inventories Using allometric equations, biomass expansion factors, root:shoo t ratios Regarded as most accurate and site specific. Cost for inventory is high. Indirect remote sensing techniques Satellite imagery, aerial photo imagery pulsed laser, dual digital camera. Field inventory for the reference data. Relatively larger s cale Technical availability can be an issue Cost effective. Modeling Ecophysiological study based, ecosystem based, or land use change based Field inventory or data from national surveys Mainly used in academics or pilot projects so far Needs man y assumptions, but can be applicable to various situations Default values for land/activity based practices Land used change based. Focus on the changes in C stocks. Field plots for the reference data Most macro scale approach. Can be used not only for forestry/agroforestry but also other land use.

PAGE 60

60 Table 3 2. Aboveground time averaged C stock in different ecosystems and agroforestry practices. Time average C stock (Mg C ha 1 (C stored in projec t C stored in baseline) in Mg C ha 1 / n (yea rs.) Meta land use systems Country and specific land use Time averaged C of land use system Mg C ha 1 Undisturbed forest Indonesia Peru 306 (207 405) 294 Managed/logged forests Brazil/Peru Cameroon Indonesia 150 (123 185) 228 (221 255) 93.2 (51.9 134) Shifting cultivation and crop fallows Cameroon Shifting cultivation, 23yrs fallow Bush fallow, 9.5 yrs Brazil/Peru Short fallow, 5yrs Improved fallow, 5yrs 23 yrs fallow 77.0 (60.2 107) 28.1 (22.1 38.1) 6.86 (4.27 9.6 1) 11.5 (9.50 13.4) 93 (80.5 101) Complex/extensive agroforests Permanent Rotational Cameroon, Cacao Indonesia Rubber Cameroon, Cacao Indonesia Rubber 88.7 (57.2 120) 89.2 (49.4 129) 61 (40 83) 46.2 (28.9 75.2) Simple agroforests/ i ntensive tree crop Brazil/Peru Coffee monoculture Multistrata system Cameroon, Oil Palm Indonesia, Pulp trees 11.0 (8.73 12.5) 61.2 (47.5 74.7) 36.4 37.2 (23.6 50.7) Grasslands/crops Brazil/Peru Extensive pastures Intensive pastures Indo nesia Cassava/Imperata 2.85 3.06 <2 Numbers in parentheses are range of the mean value. Source: Palm et al ( 2004 ) Table II in page 149

PAGE 61

61 CHAPTER 4 ABOVEGROUND AND BELO WGROUND CARBON STOCK S IN TRADITIONAL AND IMPROVED AGROFORESTR Y SYSTEMS IN MAL I, WEST AFRICA Introduction Agroforestry is a very common concept of traditional agricultural land use in most of the tropics. In the West African Sahel (WAS), the traditional systems such as bush fallow and parkland systems involve in tegration of tre es with agricultural crops. The trees provide subsidiary ( famine ) food when crops fail by drought ; can be sources of oil, wine or other condiments, and are used for tools, fences or fodder (Boffa 1999). Also, trees can increase water and nutrient availab ility through nitrogen fixation retriev al of water and nutrients from the deeper layer s of soil and reduction of water and nutrient losses from leaching and erosion in the semiarid region (Buresh and Tian 1997; Kang et al. 1999) A s described in C hapter 2, parkland agroforestry systems are currently the most prevalent land use systems in the WAS Other agroforestry practices such as improved fallow, intercropping, tree fodder planting, and boundary planting have been introduced but these are still not widely adopted (Niang et al. 2002; Levasseur et al. 2004) Most of existing studies on the parkland systems are about the productivity of trees and/or crops grown underneath, or about the interaction/competition of the trees and crops (Kater et al. 1992; Jonsson et al. 1999) C arbon (C) sequestration defined by the United Nations Framework Convention of Climate Change (UNFCCC) as has not been a subject matter of studi es in much of the WAS region, let alone in parklands and other agroforestry systems of the region. Nevertheless, it is widely accepted that environmental degradation resulting from long term drought and land use change has adversely affect ed the terrestri al C stocks in the region (FAO 2000; Reich et al. 2001). Although Mali, where this study was conducted signed off on the

PAGE 62

62 Kyoto Protocol, there has been no pilot project to document C sequestration and C credit sale in the country. Woomer et al (2004b) co nducted a national scale C stock assessment in Senegal (neighboring country of Mali), and found that there were opportunities for biological C mitigation, but they were constrained by available knowledge and access to resources. Compared with large scale tree plantation, agroforestry is expected to be the most feasible afforestation/reforestation project that can be conducted by the majority of resource limited land users (farmers) in the WAS. Because of the scarcity of on site information, it is importan t to directly measure or estimate both biomass C and soil C stocks of various agroforestry systems Therefore the study reported in this chapter was undertaken with two research questions: 1. How do different agroforestry system s differ in their potential f or C sequestration? H ow much C is stored in the traditional and improved agroforestry systems, e specially comparing above ground and below ground ? 2. W hat is the overall relative attractiveness of each of the selected agroforestry systems considering them as biological C sequestration project s ? Materials and Methods C sequestration potential of a specified project is calculated by C sequestered by the project minus C sequestered by the baseline (without the project) Since this study is not a long term project, it was impossible to monitor both C accumulations by the project (agroforestry) and by non project land use over the time. Instead, the difference s of C stock among selected land use systems are assumed to represent the potential of C sequestrati on by the land use change. Study Area This research was done in Sgou, Mali in cooperation with the ICRAF ( World Agroforestry Centre ) Field Station of Sahel Regional Programme.

PAGE 63

63 Republic of Mali Mali is a landlocked country with an area of 1.24 million k m 2 ; the Sahara desert occupies 60 % of it ( Figure 4 1 ) Mali shares borders with seven countries: Mauritania, Algeria, Burkina Faso, Ivory Coast, Guinea, Niger, and Senegal. It is a vast land of plains fed by two major rivers, the Senegal River on its western edge and the Niger River flowing in a wide arc from southwest to east The p opulation is 12.3 million with a growth rate 2.63 %, one of the highest in the world (World Bank 2007 ). While most people live in rural area s, 1. 2 million people live in the capital city, Bamako ; 90 % of people are Muslim s The major language is Bambara ( original language of a major ethnic group), although the official language is French Agriculture is the main source of livelihood for the peop le, with 80 % of people engaged in agriculture or fishing (CIA 2007) Cotton is the main export product ; gold and phosphate from mines in the north ern area are also traded. T he per capita national income was US$ 380 in 2005 Despite higher economic grow th since 1994, Mali remains one of the world's poorest countries rated 174/177 in 2005, using the UNDP Human Development Index (World Bank 2007). With the impact of current climate change and environmental degradation, the country is vulnerable to droug ht and risks further desertification S gou region The City of S gou is Mali s second largest urban center locat ed on the Niger River about 300 km northeast of Bamako. It is the capital of S gou region one of the eight administrative regions of Mali ( Figure 4 1 ) The region is locate d in the buffer zone of the Sahara, with 60 90 rainy days and 300 700 mm of rainfall annually (the rainfall intensity increasing from the north to the south ) The Sgou region has seven cercles ( administrative sections) and 2,218 villages. The p opulation of the region is about 2 million with 0.3 million living in S gou city. Cotton

PAGE 64

64 ( Gossypium spp.) i s the main crop (and product ) of the region and the region is well known for it A large cotto n mill invested in by Chinese companies is operating at the edge of the city. R ice ( Oryza glaberrima and Oryza sativa ) i s grown extensively in the irrigated area around the Niger River (R publique du Mali 2005). F armers commonly grow rain fed pearl m il let ( Pennisetum glaucum ) and sorghum ( Sorghum bicolor ) as staple food crops Select ed Land use Systems for Field Data Collection A p reliminary survey was first conducted in July 2005 to identify the targeted land use systems and possible villages to loc ate on farm plots. Five systems were selected : two parkland agroforestry systems, two improved agroforestry systems, and abandoned land (degraded land) for comparison. Parkland systems The major land use in the S gou region, as in most of other parts of the WAS, is parkland agroforestry. T wo parklands types are common: with Faidherbia albida or Vitellaria paradoxa as the dominant tree species. These two types occupy more than 60 % of cultivated land in S gou region (personal communication August 20 05, with Director of Forestry Department, S gou). Tree density is in the range of 20 to 30 trees/ha in both systems. Crops cultivated unde rneath the trees include pearl millet and sorghum sometimes intercropped with cowpea ( Vigna unguiculata ) a nd/or ban barra groundnut ( Vigna subterranea syn. Voandzeia subterranea ) F. albida has a during the hot dry season and dropping them before the rain y season ), which is quite advantageous for agroforestry: it reduces shading of crops grown underneath the tree and possibly reduces competition for water between trees and crops, and enables farmers to grow crops under the trees with practically very little reduction of cropped area in the int ercropping

PAGE 65

65 situation. It is also a nitrogen fixing tree, and farmers use the foliage for both organic manure and fodder. V. paradoxa is probably the most common parkland species in the WAS, known as Kari t (in French) or Shea (in English). Farmers use t he fat extracted from the nuts in multiple way s, such as cooking oil, medicine, and cream for dry skin. This fat, called Karit butter or Shea butter, has recently become popular for cosmetic use in the western world. It has a natural UV protection and m oisturizing effect and is therefore one of the booming products for international cosmetic companies. ICRAF organize d a program to study the characteristics of V paradoxa physiology for better production and to establish a network for local farmers to market this newly developing commodity (Pro Karit 2007 ). Improved a groforestry s ystems To examine the possibility of implementing reforestation/afforestation project s by agroforestry under the Kyoto P rotocol for C sale, it is necessary to consider agrof orestry systems with higher tree density than that of parkland systems ( where crown cover is about 20 %), or abandoned land (crown cover is close to 0 %). This is because the definition of forest or afforestation of Kyoto Protocol normally refers to h igher tree density than parkland and taller trees than bushes in abandoned land. In S gou region, ICRAF carried out a study to identify agroforestry needs for the WAS in general. The study indicated an overall shortage of fodder during the dry season, a nd that farmers need to protect their fields, especially during the dry season when cattle roam freely (van Duijl 1999 ; Figure 2 6 ). To address these problems, ICRAF has been introducing the improved agroforestry technologies such as live fence s and fodde r banks. Live fence refers to planting relatively fast growing trees in very high density around field plots, orchard s or cultivated land. Trees are planted along plot/field boundaries at 1 m intervals in two lines 1.5 m apart thus giving a 3 m wide th ick fence around the cultivated land F ive tree/

PAGE 66

66 woody perennial species are commonly used for live fence in the S gou region: Acacia nilotica Acacia senegal Bauhinia rufescens Lawsonia inermis and Ziziphus mauritiana The protected crops inside the fence are mainly cash crops such as cassava ( Manihot esculenta ), watermelon ( Citrull u s lanatus ) calabash ( Lagenaria siceraria ) and groundnuts ( Arachis hypogaea ) Fodder bank is a system of planting exotic and/or indigenous species suitable for animal fod der in relatively high density. ICRAF introduced an exotic species, Gliricidia sepium and two indigenous fodder trees, Pterocarpus lucens and P. erinaceus although these two species did not grow well enough to be harvested in all the experimental plots. The common size of the fodder bank is 0.25 ha (50 m by 50 m) framed in by live fence, and fodder trees are planted 2 m by 1 m in lines. Most of the pilot fodder banks were established on previously cultivated land. Abandoned (degraded ) land Land degra dation is a very severe problem in S gou region and the extent of degradation varies considerably from no vegetation with crusted surface to land covered by bushes and grasses. In this study, lands somewhat vegetated with grasses and bushes were chosen fo r plots. These lands were previously cultivated, but recently (within 10 years) abandoned because of the lack of soil fertility ICRAF is trying to introduce live fence s and fodder bank s to improve this over exploited land In this scenario, the differe nce in the amount of C between these abandoned lands and fodder bank/live fence systems would be the sequestration potential of the improved agroforestry practices. Research Design Since it was impossible to find all the land use systems in the same villag e, plots representing each land use system were set up in different villages in the S gou region ( Table 4 1 ). All the villages are within 30 km from the center of the city. For each land use system (treatment), three on farm plots (replicates) that were as uniform as possible (size, understory

PAGE 67

67 crops present tree density, age, land use history) were chosen. Villages for two parkland plots were chosen because of relatively mature and uniform F. albida and V. paradoxa trees ( Figure 4 2 4 3 ) Each plot was set inside a different farm. The village for live fence plots was chosen because it has a group of farmers who participated in ICRAF s live fence program. Three farmers live fences with the same age and similar tree growth were selected ( Figure 4 4 ) Fodder bank plots were more difficult to find. Since fewer fodder bank s were adopted and maintained than live fence s only three comparable fodd er banks were found in three different villages ( Figure 4 5 ) The village for abandoned land plots was chosen near the vast degraded land spreading east of S gou city. All abandoned land plots were previously cultivated by farm ers ( Figure 4 6 ) Data Collection Field data collection was conducted from August to September 2005. Biomass measurement T he plot size was 1 ha for parkland systems, wh site for improved systems (about 0. 25 ha or less), and 0.5 ha for abandoned land ( Table 4 1 ) D ata r ecorded for aboveground biomass were : Species and n umber of trees in each plot Diameter at breast height (DBH) and/or diameter at the ground of each tree/bush T ree /bush h eight Crown size for bushes in abandoned land. Regarding land use history, age of traditional parkland and abandoned land systems were difficult to estimate. According to owners of the plots, all parkland plots were at least 35 years old, and t he abandoned land plots less than 10 years. All three live fence plots were 8 years old (at the time of data sampling they were established in 1997), and two of the three fodder bank plots were 9 years and one was 6 years old.

PAGE 68

68 S oil sampl ing Based on discussions with ICRAF researchers, three depth classes were determined for soil sampling: 0 10 cm (surface soil) 10 40 cm (crop root zone) and 40 100 cm (tree root zone ). The average size tree in each plot was selected based on aboveground inventory data as the center of the soil sampling area S amples were taken from three horizontal distances from trees in the two parkland systems and live fence system At each horizontal distance samples were taken from four different p oints using an auger, and samples from each depth from these points were well mixed as a composite sample before transferring them i n to bags. F or the fodder bank where the bush/tree density was fairly uniform, and the abandoned land plots, f our random p oints were chosen to make a composite sample of each depth. More details of soil sampling are described in Chapter 5 Sampling for bulk density measurements were taken separately for each depth and land use using a 100 c m 3 stain less steel cylinder. A so il pit (1 m depth) was made for each land use plot, and the cylinder was horizontally driven to the center of each depth class to take the samples for bulk density determination All samples (total 144 samples: 99 composite samples and 45 bulk density sam ples) were air dried and shipped to University of Florida for analyses. C arbon Stock Estimation A m ount of biomass C and soil C (C stock) were estimated respectively, as follow s Total C stock (Mg C /ha) of each land use system was calculated by adding b iomass C stock and soil C stock of each plot of each land use system (all data on per ha basis). Live fences are conventionally expressed in terms of length of rows. calculated based on 3 m width; but in prac fence is important. Since live fences are along plot/field boundaries of unequal sizes, it is not realistic to assign a standard row length per unit area (ha) of plot/field.

PAGE 69

69 Biomass C stock Since this study examin es the potential of C sale s under the Kyoto Protocol s Clean Development Mechanism (CDM), methodologies used here for estimating biomass C stock are based on the guideline published by UNFCCC (2006). Although species specific allometric e quations are ideal, none was available for parkland trees from the study region. As an alternative, the UNFCCC guideline recommends using the following general equation s from FAO (1997) T ree biomass (kg) = exp ( 2.134 + (2.530 lnD BH(cm) )) (n = 191, R 2 = 0.97) (Eq. 4 1) In the FAO (1997) paper there are general equations suggested for drylands. However, th o se equations were developed from much smaller sets of trees in India and Mexico, and their DBH ranges are 3 to 30 cm. F. albida and V. parad oxa trees in parkland plots of this study greatly exceed the diameter range of these general equations. The average DBH of trees in the plots were about 59 cm for F. albida and 42 cm for V. paradoxa Using the dryland general equations may cause signific ant over estimation of the biomass. Thus, this study follows a method proposed by Woomer et al. (2004 a ) in Senegal, using Equation 4 1. This equation is also from FAO (1997) for higher rainfall area (1500 4000 mm), but covers the diameter ranges of F. albida and V. paradoxa trees in this study There are two options for estimating the biomass of the five live fence and one fodder bank species. One is following UNFCCC s guideline, using a general equation for areas with <900 mm annual rainfall. The tre e sizes are within this equation s DBH limits (3 30 cm). The equation is: T ree biomass (kg) = 10 ^ ( 0.535 + log 10 DBH(cm) 2 /4 )) (R 2 = 0.94) (Eq. 4 2)

PAGE 70

70 The UNFCCC guideline s suggest using these equations when no local species allometric equations are available. A s econd option is to us e the equations developed from Acacia tortillas and Acacia ruficiens in Northwest K enya (Coughenour et al. 1990). These are: Log 10 (g mass) = 2.26+3.98 log 10 (mm stem diameter) (Eq. 4 3) R 2 =0.98 (stem diameter<15.7 mm) Log 10 (g mass) = 0.68 +2.66 log 10 (mm stem diameter) (Eq. 4 4) R 2 =0.98 (stem diameter>15.7 mm) Although Northw est Kenya is not in the WAS, its climatic condition is much more similar to that of the study area than to the area where the UNFCCC guideline s equation was developed. Gonzalez (2001) used these Acacia spp. equations in his research at various parts of S enegal. These two options were both tried in this study, and the results of estimated biomass C are compared later. For abandoned land plots, equations for Guiera senegalensis used by Seghieri et al (2005) in Niger were adopted G. senegalensis is the mo st dominant shrub species in the abandoned land plots, and the equation was originally developed in fallows of Mali (Ciss 1980, Franklin and Hiernaux 1991). Foliage mass of each stem of each shrub: Bl stem (g), B asal circumference of the stem: C stem (cm): Bl stem = 1 09 C stem (all stems of n = 20 shrubs, R 2 = 0 82 P < 0 001) (Eq. 4 5) S tem wood dry mass: B w stem (kg): B w stem = 0 0037 C stem (36 stems among n = 15 shrubs, R 2 = 0 90 P < 0 001) (Eq. 4 6) Leaf and wood masses ( Bl stem and B w stem ) were then aggregated for each multi stemmed shrub. To calculate the amount of C in the biomass, C fraction rate of 0.5 is suggested in the UNFCCC guideline. Belowground biomass is also estimated by using the suggested root/shoot

PAGE 71

71 ratios, which are 0.25 for trees and 0.5 for abandoned land bushes. Biomass C stock was calculated by adding aboveground biomass C and belowground biomass C. Soil C stock Soil C stock was estimated from the samples taken at each land use s plots Sub samples were taken from the 99 air d ried composite samples and ground. Soil C percentages of the sub samples (% of C weight in whole soil weight) were measured by the dry combustion method on an automated Flash EA 1112 NC elemental analyzer (Thermo Fisher Scientific, Inc.). Then, the soil C percentage data of each land use of each depth class was converted to the amount of C per ha basis with using bulk density data. Statistic al Analysis Analysis of variance (ANOVA) by SAS PROC MIXED procedure and Turkey Kramer multiple comparison test we re conducted to compare the C stocks of different land use systems and soil depth. The linear model shown below was used. y i = + L i + e i y i is the C concentration in land use i is the population mean, L i is the land use (treatments), i = FA, VP, LF, FB, and AL. FA: F.albida parkland, VP: V.paradoxa parkland, LF: live fence, FB: fodder bank, and AL: abandoned land. e i is the r andom variable error within the experiment Linear correlation was also tested to examine the relationship of biomass C and soil C stock.

PAGE 72

72 Results C Stock in Biomass and Soil The t wo parklands selected for the study were similar in tree density and the d ominance of major tree species ( Table 4 2 ) F. albida trees were generally larger and taller than the V. paradoxa trees in all parkland plots. Although live fence and fodder bank plots ha d large number s of trees per ha, these were proportionally calculated numbers based on small plots. The real size of live fence plots was 0.088 ha on average (293 m length 3 m width) with 327 trees, and that of fodder bank plots was 0.24 ha on average with 145 G. sepium trees. The s oil C specific comparisons will be presented in detail in Chapter 5 ; but estimates of soil C stock of each depth are presented, along with biomass C stock per ha basis of each system ( Figure 4 7 ). Statistical comparison by ANOVA show e d that a bandoned land ha d larger soil C stock than the other four systems in every depth, although the significance varied with depth Estimated biomass C values of live fence and fodder bank plots from UNFCCC guideline equations and those from Acacia sp p. equations developed in Kenya were significantly different ( t test p <0.01) ( Table 4 3 ) However, when each set of estimation was compared with other three systems by ANOVA, the results were the same. The ranking of systems in order of biomass C stock was: F. albida parkland > V. paradoxa park la nd > Live fence > Fodder bank >Abandoned land. However, the last three systems were not significantly different ( Tukey Kramer test ) even when analyzed separately Total C Stock Tot al C ( aboveground biomass C + soil C) stock of each system was calculated and compared at three different soil depth ranges The order among the systems and the significance of difference var ied with the depth of soil ( Table 4 4 ). Overall, F. albida parkland ha d the largest total C stock and was significant ly differen t from the other four systems. V. paradoxa

PAGE 73

73 parkland ha d the second largest total C stock, while the difference between other systems became less significant as deeper soil C stock was included. Also as deeper soil was taken into account, abandoned land ha d more C stock relative to the other systems. Relationship between Biomass C and Soil C All the possible combination s of biomass C stock data and soil C stock data across the five land use systems were tested for their relationship: Biomass C stock data and soil C stock data (0 10cm), Biomass C stock data and soil C stock data (0 40cm), Biomass C stock data and soil C stock data (0 100cm), Biomass C stock dat a and soil C stock data (10 40cm), and Biomass C stock data and soil C stock data (40 100cm). None of the regression was significant. Discussion In terms of total C stock per ha traditional agroforestry systems (parklands) are a larger storage than im proved agroforestry systems (live fence and fodder bank) or abandoned land. Although the improved agroforestry systems are relatively young, they are not likely to store as much as the parklands at the end of their 25 year rotation Because branches of f odder bank trees are annually pruned to control their height (<4 m), and live fence trees are planted in such a high density that woody biomass accumulation per tree will be comparatively less However, h aving a large C sto ck does not necessar il y mean hav ing a large C sequestration potential. Traditional parklands are very stable (long standing) and so is the C stored Farmers in the area are unlikely to increase the tree density of parklands to match the UNFCCC s forest plantation criteria for seques tering more C for sale, because it will produce a negative impact on crop growth. Also it is very difficult to convert nutrient poor abandoned land to parklands, since silvicultural methods are not established for these species (parkland trees are mostly natural regeneration), and abandoned land is not fertile enough to grow crops underneath.

PAGE 74

74 On the other hand, introducing live fence s and/or fodder bank s into cultivated land or abandoned land can sequester C by increasing the tree biomass, but the extent would vary largely by the baseline and accounting method used Biomass C sequestered by live fence planting is a since they are normally established on the tree less cultivated land (the baseline is nearly 0). The potential of fodder bank s however, depends on initial plot condition As in the situation for live fence, w hen the cultivated land is converted to fodder bank most of C sequestered by the fodder trees can be counted. However, establ ishing fodder bank s on abandoned land, as ICRAF or local government is trying to promote, may actually result in net loss of C stock during the initial stage because the biomass from bushes and grasses in the abandoned land has to be removed at the time o f establishment, and it may take years for fodder trees to accumulate an amount equal to the original biomass. Further investigations are needed on temporal C dynamics of these systems. S oil C is not considered in the calculations of the Kyoto Protocol f or its first commitment period (2008 2012) When, rather than if, soil C is taken into account, determining b aseline soil C will be another challenge to determine and compare the C sequestration potentials of land use systems. Results suggest that soil s ampling depth makes a large difference in estimating the amount of C stored per area basis, as well as the potential for C sequestration ( Table 4 4 ) To compare and discuss the C sequestration potential of different land use o r different ecoregions it will be very important to standardize sampling depth. Several studies in Africa reported that planting trees for C sequestration will not immediately retain soil C equal to the baseline level nor increase it in the short term (K aya and Nair 2001; Walker and Desanker 2004). Introducing live fence s or fodder bank s may increase the biomass C in the system but may not in crease soil C. S oil C sequestration potential will be discussed in more detail in the next chapter.

PAGE 75

75 Regarding t he biomass estimation methodology, two sets of allometric equations we re used for live fence and fodder bank. The values calculated from the two methods are significantly different from each other ( UNFCCC s general equation is much more conservative than Acacia species equations from Kenya). However, comparing each method s values with other three land use systems showed similar results: both live fence and fodder bank have not (perhaps not yet) stored significant amount s of C compared with bushes of aban doned land. This is partly because of the young age of the two systems (6 to 9 years old) The UNFCCC guideline suggest s applying the general equation only if it is impossible to find/establish local allometric equations. The Acacia species equations ar e not exactly local but they are from environments more similar to the studied area than were the general equations. Considering the difference between these two method s developing the local allometric equations is likely to increase the profits from C credit sale when the C market is introduced in the area. Obviously, substantial research efforts are warranted in this area.

PAGE 76

76 Table 4 1. Characteristics of the v illages where the experimental plots were set up in S gou region, Mali Land use Name of t he village Position Elevation (m) Size of the plot Faidherbia albida parkland Togo N.13.35, W. 6.31 300 1 ha Vitellaria .paradoxa parkland Dakala N.13.32, W. 6.23 297 1 ha Live fence Dougoucouna N. 13.37, W.6.37 298 294m (average) Fodde r bank 1 Dakala N.13.32, W. 6.23 297 0.25 ha Siguila N.13.28, W. 6.21 305 0.25 ha Banankoroni N.13.35, W. 6.38 293 0.22 ha Abandoned land Diamaribougou N. 13.36, W. 6.19 298 0.5 ha 1 One fodder bank plot from each village Table 4 2. Characte ristics of the experimental plots (three plots average) for five selected land use system s in S gou region, Mali DBH (cm) Height (m) Number of trees (ha 1 ) Species composition Faidherbia albida parkland 59.4 (1.8) 13.0 (0.9) 21 (5.3) A verage 88.6% F aidherbia albida dominance Vitellaria paradoxa parkland 41.7 (5.9) 9.9 (0.9) 20 (0.6) A verage 80.6% V itellaria paradoxa dominance Live fence 2.5 (0.5) 2.5 (0.4) 3720 (882) Average 67.6% A cacia nilotica Fodder bank 2.2 (0.5) 2.0 (1.1) 5 88 (277) G liricidia sepium average only Abandoned land 2.8 (0.6) diameter at ground 1.3 (0.4) 46 (30) A verage 47.5% Guiera senegalensis and 39.5% Combretum micranthum Note: Numbers in parentheses are standard deviations. Tree dominance means the p ercentage among the standing trees in the plot.

PAGE 77

77 Table 4 3. Estimated b iomass C (above and below ground) stock values of each plot and three plots average of five selected land use systems. Two sets of values from different allometric equations are s hown for live fence and fodder bank systems, which are significantly different in t test (values of UNFCCC equations < values of Acacia spp. equations ). Faidherbia albida parkland Vitellaria paradoxa parkland Live fence UNFCCC 1 Acacia 2 Fodder bank UN FCCC 1 Acacia 2 Abandoned land (Mg C ha 1 ) Plot A 51.4 24.2 3.2 5.9 2.1 4.8 0.8 Plot B 55.7 16.5 3.0 4.3 1.8 2.7 0.4 Plot C 54.8 26.6 7.8 14.8 2.7 4.9 1.0 Average 54.0 22.4 4.7 8.3 2.2 4.1 0.7 a b c c 3 c c 3 c a, b, c: Mean separation by Tukey s multiple comparison test at p <= 0.05) 1 Estimation from the UNFCCC guideline's equations 2 Estimation from Acacia spp. equations developed in Northern Kenya 3 Values from UNFCCC equations and Kenyan equations w ere separately compared with other three systems Table 4 4. Total C stock (biomass C + soil C of different depth) of five selected land use systems. Total C stock (Mg C ha 1 ) More C Less C 1 2 3 4 5 Biomass + 0 10cm soil C FA 59.8 a VP 27.7 b LF 9.8 c AL 7.9 c FB 4.8 c Biomass + 0 40cm soil C FA 70.8 a VP 37.1 b AL 24.7 bc LF 17.7 c FB 14.0 c Bi omass + 0 100cm soil C FA 87.3 a AL 56.9 b VP 49.8 b FB 35.6 b LF 28.7 c FA: F aidherbia albida parkland, VP: V itellaria paradoxa parkland, LF: Live fence, FB: Fodder bank, AL: Abandoned land (a, b, c: Mean separation across land use systems by Tuk ey Kramer 's multiple comparison test at p <= 0.05) Data s ource: Biomass C values of live fence and fodder bank are from UNFCCC equations.

PAGE 78

78 A B Figure 4 1. A: Location map of Mali; B: Map of Mali showing its land locked nature: C: Map of S gou region ( T he sign refers to the city of S gou). S gou Niger River Bani Rive r C

PAGE 79

79 Figure 4 2. Faidherbi a albida parkland in Togo village. The tree leaves are shed at the beginning of the rainy season ; but they return at the beginning of the dry rainy season. The understory crop is pearl millet ( Pennisetum glaucum ) (Photographed by author) Figure 4 3 Vitellaria paradoxa parkland in Dakala village. The trees have wide canopies, and leaves remain during the rainy season. Farmers plant crops (in this photo, pearl millet) beneath the trees, often very close to the trunk (Photographed by author)

PAGE 80

80 Figure 4 4. Live fence system in Dougoukouna village. Relatively fast growing (mostly thorny) trees are planted around the crop field to protect crops from free roaming animals. The trees are planted in two lines (1.5m apart). The outside line trees s hown in the photo are mostly Acacia nilotica (Photographed by author) Figure 4 5. Fodder bank in Dakala village. Gliricidia sepium trees are planted at 2 m by 1 m spacing in lines. Towards the end of the dry season when other fodder sources such as fresh grasses or crop residue are scarce, farmers harvest branches of the trees, dry and feed them to their domestic animals. (Photographed by author)

PAGE 81

81 Figure 4 6. Abandoned land just outside of Diamaribougou village. The land was cultivated until l ess than 10 years ago. The surface soil is eroded leading to formation of a hard surface pan. Only certain bushes such as Guiera senegalensis and Combretum micranthum can survive on this type of degraded land. (Photographed by author)

PAGE 82

82 Figure 4 7. Aboveground and belowground C stock per ha of five selected land use systems. Biomass C is shown above the x axis, and soil C is shown below the axis with three soil depth classes. Values of live fence and fodder bank are from UNFCCC equations.

PAGE 83

83 CHAPT ER 5 SOIL CARBON SEQUESTRATION IN DIFFERENT PARTICL E SIZE FRACTION S AT VARYING DEPTHS UNDER AGROFORESTRY SYSTEMS IN MALI Introduction The measurement of carbon ( C ) content is part of the basic soil analysis procedure. Inventory data on soil C content is a vailable in most of the WAS countries. However, to discuss soil C sequestration as one of the options for mitigation of atmospheric CO 2 the stability of soil C ( how well C captured inside the soil) has to be considered. In other words, the soil C t hat goes back to the atmosphere after decomposition within a month of its deposition, and that stays in the soil for decades should not be counted as similar in terms of C credits. Characteristics such as the stability of soil C are very controversial iss ues in estimating and accounting methodologies (Ingram and Fernandes 2001; Garcia Oliva and Masera 2004) Also, soil sampling depth for these accounting procedures needs to be deeper than for normal soil analysis. The conventional soil C studies of agric ultural systems mostly focus on soil organic matter in the surface layer of 20 cm for the interests of soil fertility But sampling of deeper soil horizon is necessary in efforts to understand the extent of soil C protection and characteristics of various s oil C form s especially the systems involving deep rooting plants such as trees (Jobbagy and Jackson 2000) In general, soil C dynamics regarding C sequestration have not yet been well studied, although recent technological development and interests tow ards climate mitigation activities are contributing to an increased number of this type of studies (Post et al. 2000; Sun et al. 2004) Still, these studies are limited even in developed countries, and not easily available in the research resource limite d area such as the studied region or Africa in general. The studies of this nature that have been conducted so far have been in natural environment such as forest stands, tundra, or grasslands, probably due to the relatively stable dynamics of soil proper ties

PAGE 84

84 (Richter et al. 1999; Schuur et al. 2001) And s oil C studies in agricultural croplands have mostly been in the context of soil productivity management ( Phillips et al. 1993; Beare et al.1994; Alvalez et al. 1995 ; Blair et al. 1995; Franzluebbers et al. 1995 ; Rhoton 2000 ), although recently more and more soil C studies are considering agricultural soil as C sinks a mechanism which removes CO 2 from the atmosphere (Smith et al. 1998, Duiker and Lal 1999, Lal 2004 b ) Soil C studies in agroforestry syst ems have been few. Interactions between crops and trees and the relatively short term rotation of land management make such studies more complicated and challenging compared to single species agricultural and forestry systems Existing studies in agrofor estry systems measure soil organic matter (SOM) content with other soil nutrients (Kang et al. 1999; Makumba et al. 2006). Those studies discussed whether trees have positive ( e.g., nitrogen fixing) or negative ( e.g., competition for light, nutrients, or water) impacts on crop production. In the WAS, parkland trees were found to increase soil C around trees (Jonsson et al. 1999), and an improved fallow system (planting Gliricida sepium during the non cropping phase) was found to increase soil C on the sur face compared with natural grass fallow (Kaya and Nair 2001). These studies support the expectation that agroforestry systems would enhance soil C sequestration, but there is still little information about trees influence on C in deeper soil and stabilit y of various forms of soil C sequestered by trees. In this study, o rganic C in soil is assumed three ways: 1 physically stabilized, or protected through microaggregation (microaggregate associated soil C) 2. intimate association with silt and clay particles (silt and clay associated soil C), and 3 b iochemically stabilized through the formation of recalcitrant soil organic matter compounds (non hydrolysable soil C) (Six et al. 2002). There are some othe r ways such as Al or Fe SOM complexes, C accumulation resulting from anaerobic conditions, and transfer to

PAGE 85

85 subsoil by colloidal or soluble C ; but they do not seem to occur significantly under the soil and climatic conditions of the study region (Tan et al 2004; Nierop et al. 2007 ; Zinn et al. 2007) The turnover time for physically protected C (type 1 and 2) is estimated to be 50 1000 years ; for biochemically protected C (type 3) it is 1000 3000 years. The turnover time for less stable C within mac roaggregates is 5 50 years, and for other type s C such as the litter fraction it is 0.1 20 years (Batjes 2001). To differentiate the types of soil C, physical fractionation is the common initial step. The dynamics of soil C in each fraction size ca n be further investigated by 13 C isotopic ratio measurement, which distinguishe s between C derived from plants that follow C3 photosynthetic pathways (all trees) and those that follow C4 pathway s (most warm season graminaceous plants: in this study pearl m illet Pennisetum glaucum and sorghum Sorghum bicolor ) This method has been used for studying the impact of land use change on soil C and for comparing the C dynamics in different land use systems (Balesdent et al. 1998; Potvin et al. 2004). Research Questions In this scenario, the present study was undertaken based on the premise that compared with agricultural and tree less systems, a groforestry systems will help store more C in soil and offer better stability of stored C in deeper soil layers due t o presence of deep rooted trees Specific research questions are: 1. Do trees contribute to soil C storage in the selected agroforestry systems, and how stable is the stored C ? 2. What is the relative attractiveness of each of the selected agroforestry systems or land use change in terms of its soil C sequestration potential? Materials and Methods The study was conducted in the seven selected villages of S gou region, Mali, West Africa. The details of the site and the selected land use systems (treatment s ) are described in

PAGE 86

86 Chapter 4. In each land use system, three on farm plots (replicates) were chosen for soil sampling Research Design Soil s amples were taken from different distances from trees. In the two parkland systems, three horizontal distances for so il sampling were chosen : Near (bottom of) the tree 3 m ( about half the crown radius) from the trunk 10 m from the trunk ( outside of the crown) The average size tree in each plot was selected based on aboveground inventory data as the center of the sampl ing area Soil samples were taken from four direction s (north, south, west, east) around the tree and mixed before putting in the sample bag ( Figure 5 1 ). Three horizontal distances for sampling live fence plots were: Near (bot tom of) the tree 1 m ( inside) from planted line (root influence zone) 3 m (inside) from planted line ( outside the crown and rootzone ) Live fences are either rectangular or polygonal shapes; four sampling points on different sides were randomly chosen. Sa mples away from the tree line were taken inside the fence, because outside of the fence were often paths or borders of the cultivated land. Since fodder bank trees are evenly planted (2m 1m) inside the plots and shrubs are randomly grown in abandoned land plots, horizontal differentiation of sampling was not taken at these two systems. In each plot, samples were taken from four randomly selected points, and mixed well to form the composite sample Sampling depth s at each horizontal distance were as describ ed in Chapter 4 : 0 10 cm (surface soil) 10 40 cm (crop root zone) and 40 100 cm (tree root zone ). This was based on

PAGE 87

87 the expectation that the amount of C content would differ by depth class depending on the presence or absence of tree roots and til lage. In summary, the number s of soil samples are: 2 (P arkland systems) 3 (horizontal dist.) 3 (depth) 3 (replicates) = 54 1 (Live fence) 3 (horizontal dist.) 3 (depth) 3 (replicates) = 27 1 (Fodder bank) 1 (horizontal dist.) 3 (depth) 3 (replicates) = 9 1 (Abandoned land) 1 (horizontal dist.) 3 (depth) 3 (replicates) = 9 Total = 99 Soil Preparation and Analyses Samples were all air dried and passed through a 2 mm sieve (except samples for bulk density measurement) at the field station in S gou. There is no visible O horizon or surface litter and therefore no analysis was done for that layer ( Woomer et al 2004) ( Figure 5 2 ) Soil samples were brought back from Mali to University of Flo rida in October 2005 for analysis Samples for bulk density measurement for each depth class w ere separately collected at each plot with a 100 cm 3 cylinder. Wet weight and air dry weight were measured in the field. S amples were oven dried and analyzed fo r p article size distribution (USDA Soil Survey Lab Method) and pH at the University of Florida, Soil and Water Science Department laboratory Sub samples were taken from the 99 air dried samples and ground. Soil C content ( g C kg 1 soil ) of the sub sampl es w as measured by the dry combustion method on an automated Flash EA 1112 NC elemental analyzer (Thermo Fisher Scientific, Inc.). Soil f ractionation Soil samples were fractionated into three aggregate size classes (2000 250 m, 250 53 m, and <53 m) by wet sieving, following the method of Six et al. ( 2002) A sub sample of 100 g of the composite soil sample was submerged in deionized water as disruptive forces of slaking for about 5 min utes prior to placing it on top of a The sieving was done manually by moving the sieve up and down approximately 50 times in 2 minutes. The fraction

PAGE 88

88 remaining on the top of the sieve was collected in a hard plastic pan, oven dried at 65 o C and weighed. m were poured through a 53 procedure repeated. The recovery of mass soil fractions after overall wet sieving procedure ranged from 97 to 99% of the initial soil mass. S ub samples for e ach soil fraction (99 samples 3 fracti on size = 297 samples) were then ground and C contents were measure d by the same dry combustion instrument as described for whole soil C. Soil C in the large (L) fraction (2000 m) contains fairly new coarse/fine particulate organic matter (POM) C, although there is other forms of protected C not fully separated by wet sieving. The medium (M) fraction (250 m) contains both less protected C (within fine POM) and protected C (microaggregate protected POM C and silt + clay associated C). C in the small (S) fraction (<53 m) contains the protected form (silt + clay protected C or non hydrolysable C) although there are less stable forms of C in the size class, too (Six et al. 1998, Six et al. 2000) Unprotected C involves the youngest form of SOM and through the process of aggregate formation, adsorption/desorption, and condensation/complexation, soil C becomes older and more stable (SOM dynamic model Figure 5 3 ) C isotopic ratio ( 13 C / 12 C ) measurement 13 C values of soil samples (whole soil and fractionated soil) were measured by Thermo Finnigan MAT Delta Plus XL mass spectrometer (Thermo Fisher Scientific, Inc.) C isotope notation: 13 C = [(R Sample R Std )/R Std ] 10 3 ( Eq 5 1) Where R Sample is the 13 C/ 12 C ratio of the sample, and R Std is the 13 C/ 12 C ratio of the Vienna Pee Dee Belemnite ( VPDB ) standard. Relative proportions of soil C derived from C 4 plants material versus C 3 plants material was estimated by mass balance ( Balesdent and Mariotti 1996):

PAGE 89

89 C 4 plants contribution = ( Tr ) / ( Cr Tr ) (Eq. 5 2) Where is the 13 C value of a given sample, Cr is the average 13 C value of C 4 plants tissue ( Tr is the average 13 C value of C 3 plants ( In the studied l and use systems trees and bushes in abandoned land are C3 plants, and crops grown underneath the parklands around the live fence are C4 plants (sorghum and millet) as well as the presumed previous vegetation (crops) of fodder bank and abandoned land. St atistical Analysis Analysis of variance (ANOVA) was used to estimate the variance components. The linear models were applied to the soil C concentration data. Model1 was applied to all five land use systems, and model 2 was applied to three land use syst ems (two parklands and live fence ) that have distance information. The linear models were: Model 1: y ijkl = + L i + D j + F k + I l + L*D ij + L*F ik + L*I il + D*F jk + D*I jl + F*I kl + L*D*F ijk + L*D*I ijl + D*F*I jkl + L*D*F*I ijkl + e ijkl (Eq. 5 3) Model 2 : y ijklm = + L i + D j + F k + I l + T m + L*D ij + L*F ik + L*I il + D*F jk + D*I jl + F*I kl + L*T im + D*T jm + F*T km + I*T lm + L*D*F ijk + L*D*I ijl + D*F*I jkl + L*D*T ijm + L*F*T ikm + L*I*T ilm + D*F*T jkm + D*I*T jlm + F*I*T klm + L*D*F*I ijkl + L*D*F*T ijkm + L*D*I* T ijlm + D*F*I*T jklm + L*D*F*I*T ijklm + e ijklm (Eq. 5 4) y ijklm is the C concentration in land use i at depth of j fraction size of k isotopic ratio of l distance of m is the population mean, L i is the land use (treatments), i = FA, VP, LF, FB, a nd AL. FA: F. albida parkland, VP: V. paradoxa parkland, LF: live fence, FB: fodder bank, and AL: abandoned land. D j is the depth, j = 1 3

PAGE 90

90 1: 0 10 cm, 2: 10 40 cm, 3: 40 100 cm. F k is the fraction size, k = L, M, S L: large fraction ( 2000 m ), M: medium fraction ( 250 53 m), S: small fraction (<53 m ) I l is the isotopic ratio, l = C3, C4 C3: C3 plants origin C, C4: C4 plants origin C T m is the distance, m = n, m, f n: near the tree, m: middle of the canopy, f: far from the tree e ijklm is the random variable error within the experiment M odel 1 and model 2 including all the possible interactions between the factors were run using PROC MIXED procedure of SAS. Interactions that were not significant were dropped from the model. The models th at were biologically and statistically significant are presented in the results section. To further examine the interaction s data were sorted (PROC SORT procedure) with certain factors fixed, and tested again using ANOVA Based on the outcome of the ANO VA, factors and other soil characteristics (e. g percentage of sand, silt, and clay) were tested for their relationships using line ar regression All statistic al tests were considered significan t when p <0.05 unless otherwise specified. Results Soil Charac teristics Soils in the sample plots are mostly sandy loam or loamy sand ( Table 5 1 ) Soil colors varied from whitish or dark gray to reddish brown in different plots, but all are categorized as Haplustalfs by the regional surv ey (Doumbia 2000). Abandoned land soil was extremely hard to sample with an auger because bedrock was found in some places at less than 1 m depth ( Figure 5 2 ). Most of the time silt or clay was clearly observed in 70 80 cm d epths Content (g kg 1

PAGE 91

91 soil) of sand, silt and clay were compared statistically over the five land use systems. Abandoned land ha d lower sand content (530 694 g kg 1 soil) and higher content of silt and clay (306 470 g kg 1 soil) than the other four systems (sand: 715 935 g kg 1 soil, silt + clay: 65 285 g kg 1 soil) ( Table 5 1 ) Soils of the four systems in each depth class were not different in the particle size content Whole Soil C Whole soil C content across a ll the systems varied from 1 6 g C kg 1 soil ( Figure 5 4 ) Two parklands and live fence ha d three sampling locations ( 0, 3, and 10 m from tree base for parklands; 0, 1, and 3 m from tree lines for live fence ) Only the surfac e soil (0 10 cm) of live fence show ed a difference between the near tree and the two zones more distant from the tree but other depth classes and two parklands plots d id not show difference by horizontal distance from trees although the trend of nea r tree > outside crown was observed in the surface soil of both parkland systems C content decreased with soil depth for all land use systems except the fodder bank where the surface soil (0 10 cm) had less whole soil C than lower depths The whole soil C data of five land use systems were compared statistically using two factorial (land use and depth) ANOVA (model 1) Both land use and depth factors, as well as land use depth interaction were significant for that variables By Tukey Kramer multip le comparison test abandoned land ha d higher C con tent than the other four systems b ut the other four systems were not different from each other C con tent was different by depth: 0 10 cm > 10 40 cm > 40 100 cm across all treatments Among the d ata for the two parklands and live fence plots, horizontal distance was another factor (model 2) Land use and depth factors were significant but distance was not significant although showing the trend ( p =0.0884) in the three way factorial (land use, depth, and distance) ANOVA. When land use factor and distance factor were examined in the fixed

PAGE 92

92 depth class, the distance factor was still not significant in any depth class. However, at 10 40 cm and 40 100 cm class es C con tent was higher for F. al bida parkland than in the case of the live fence. C in Soil Fractions Carbon fraction contents [large (L): (250 m) and medium (M): (53 m) ] in different systems were most ly not different among each other and ranged from 1 to 2 g C kg 1 soil (except in fodder bank where it was from 0 to 1 C kg 1 soil) ( Figure 5 5 ). S mall (S) fraction (<53 m) C content d id not change much from 0 10 cm to 10 40 cm depth in all systems. In the live fence treatment, C in the 0 10 cm depth contained more L fraction (1.8 g C kg 1 soil) than the other two size fractions (less than 0.7 g C kg 1 soil) whereas the fodder bank treatment had very low C content of L and M fraction s in that soil layer ( 0.6 g C kg 1 soil) Data of C in t he three fractions were analyzed by three factorial ANOVA (land use, depth, and size) (model 1) All three factors were significant, as well as three combinations of two factor interactions. Results of multiple comparisons among land use systems were the same as for the whole soil C data; abandoned land ha d higher C con tent than the other four system s, which did not differ among each other. Depth class comparison also showed the same result: 0 10 cm > 10 40 cm > 40 100 cm S fraction and L fractio n C were both significantly higher in con tent than M fraction C when three depth class data are combined. The significance varied when each depth class was separately tested, but M fraction C content was always the lowest. When each fraction size data we re tested separately, land use and depth were significant for all fraction sizes. Distance from the tree was the only factor that was not significant in four factorial ANOVA (land use, depth, size, and distance) for three systems (two parklands and live fence : model 2 ). Interactions of four factors and three factors including distance were not significant.

PAGE 93

93 Further sorting and testing show ed that the distance factor was significant in only M fraction C at the 0 10 cm depth, where near tree was higher in con tent than outside crown Isotope Analysis of Whole Soil C The measured 13 C values of each depth class of each land use systems varied from 23.9 to 15.1 ( Table 5 2 ). Based on the values and the mass balance calcula tion, whole soil C data was separated into that originating from C 3 plants (trees) and C 4 plants (crops) ( Figure 5 6 ). Near tree data and outside crown data are presented side by side for two parklands ( Figure 5 6 A, B, C, D ) and live fence ( Figure 5 6, E, F ) I n the figure C of t ree origin was found more in surface soil and near the tree, although when they were tested statistically, there was no significant difference between near tree and outside crown data except for that of live fence at the 0 10 cm depth. Fodder bank d id not have much C of C 3 origin, even with trees growing in the plots. On abandoned land C 4 origin C was the major form of C and, as mention ed earlier, the soil C content was higher in this system compared with other systems Three factorial (land use, depth, and isotopic ratio) ANOVA was conducted among the five land use systems (model 1) All factors were significant, and three factor inte raction and two factor interactions including isotopic ratio were also significant. C 3 origin C and C 4 origin C were then tested separately using the SORT procedure. Land use was not a significant factor among C 3 origin C data, but depth was For C 4 o rigin C, both land use and depth had significant effect: abandoned land had higher content than other four systems with parklands higher than the improved systems, and deeper depth had less C content. Four factorial (land use, depth, distance, and isotop ic ratio) ANOVA was used to test differences among the two parklands and live fence systems (model 2) Distance was again not a significant factor while all others were Four factor interaction, as well as three and two factor interactions including dis tance was significant, suggesting distance was somewhat influential for

PAGE 94

94 C content When only C 3 origin C data was tested separately land use was not significant but depth and distance were Multiple comparison tests show ed that C 3 origin C content from near tree and half crown were higher than that from outside crown throughout the three systems. Among C 4 origin C, land use and depth were significant factors but distance was not. Isotope Analysis of C in Soil Fraction s The measured 13 C values of three fraction sizes varied from 25.4 to 15.7 ( Table 5 2 ). Data of C content in the three size fractions were also separated into those with C originating from C 3 and C 4 plants ( Figure 5 7 ). Four factorial ANOVA (land use, depth, size, isotopic ratio) was used to test differences among the five land use systems (model 1) Land use, depth, and size were significant factors but isotopic ratio was not. Four factor interaction was not significant, but all two factor interaction combinations including isotopic ratio were significant ( p <0.01). When C 3 origin C was tested separately land use was not significant while size and depth were. All three were significant factors among C 4 origin C data. Wh en data were sorted by the fraction size, C 3 origin C was significantly more than C 4 origin C in the L fraction, while C 4 origin C was more than C 3 origin C in the S fraction. Among the two parklands and live fence data sets, five factorial ANOVA (land use, depth, size, isotopic ratio, and distance) was conducted (model 2) Land use, size and depth were significant but isotopic ratio and distance were not. All combinations of three factor interactions including isotopic ratio and distan ce were signifi cant, so was the two factor interaction of distance and isotopic ratio. Distance bec ame a significant factor ( near tree > outside crown ) for C 3 origin C data sets, while it was not for C 4 origin C. Relationship s of Data Sets Linear relationship s were tested between C content data sets and other soil characteristics of the samples. The S fraction (<53 m) percentage in whole soil had significant relationship

PAGE 95

95 (R 2 =0. 60 ) with the S size C (contains protected C) con tent However, this relationship was not seen for M and L fraction s ; the more fraction percentage of M or L size in the soil does not mean the more soil C of those sizes. Also, t he regression between silt + clay content and whole soil C con tent was strong especially in the 10 40 cm soil dept h ( Figure 5 8 ). S and silt clay or silt + clay contents d id not have strong relationship with L fraction C or M fraction C. Although the regression between silt + clay content and S fraction C was poor (R 2 =0. 16 ) in the tota l data set it became more pronounced when the data set was divided by depth class. R 2 values between s ilt + clay and S fraction C were 0.53 at 0 10 cm 0.44 at 10 40 cm and 0.67 at 40 100 cm ( Figure 5 9 ). Soil pH or bul k density d id not show any strong relationship with C in whole soil or in each of the three fraction s in any depth or for all depth s combined. Percentages of C 3 origin C or C4 origin C d id not have a significant relationship with whole soil C con tent The re was a strong relationship between C 3 origin C con tent and L fraction C con tent (R 2 =0.67) throughout the data set. The relationship was stronger at 0 10 cm soil (R 2 =0.72) when data sets were divided by depth ( Figure 5 10 ), but were not significant in deeper depths The relationship between C3 origin C and S fraction C content was also observed in the 40 100 cm depth (R 2 =0.4 5 ). Discussion Contrary to expectations, C content in all soil depth s abandoned land than in any of the four agroforestry systems ( Figure 5 4 ) although the significance varied depending on the depth class Judging from the observation that the abandoned land soil ha d significantly more silt and clay fracti ons than those of soils under the agroforestry systems ( Table 5 1 ) it seems that the whole soil C content was directly related to the silt and clay contents of the soil In general, soil organic C con tent is known to correlat e positively with the

PAGE 96

96 amount of non crystalline clays (Powers and Schlesinger 2002). Indeed, the whole soil C content was related to silt + clay content in all three depth classes (strongest in 10 40 cm) ( Figure 5 8 ) Silt + clay content also had strong relationship with S fraction C (<53 m ), the strongest being in 40 100 cm ( Figure 5 9 ) but not as well as with whole soil C This suggests that silt and clay are mainly associated with soil C in <53 m size, especially in deep soil, but they are also associated with larger fraction size form of soil C. Based on the data points on Figure 5 9 abandoned land data do not seem to follow this relationship well, while they follow the relationship better in Figure 5 8 This could be because silt and clay are more involved in forming larger than 53 m fraction size C in abandoned land compared with other systems. After being abandoned for a few years (less than 10: see Chapter 4 ) the land probably was subjected to extensive erosion that took away aggregates and sandy particles from the surface lay er. The r emain ing soil was higher in silt and clay and formed a hard surface pan, which prevent ed further disturbance or leaching. The m ajority of soil C in the abandoned land was of C 4 plant origin ( Figure 5 6 ) suggesting tha t the stored C in abandoned land was mainly from the previous land use ( land cultivated with C4 plants) and was well protected. The o ther four land use systems were not significantly different in terms of the soil characteristics (pH, bulk density, parti cle size ). However, as seen in the soil pit photo ( Figure 5 2 ) the color was quite different in each soil pit, suggesting soil variations among plots of same land use systems in the same village. Still, s oil C content variatio ns among these soils can be explained more as a consequence of the influence of trees and land management than caused by soil characteristics as in the case of abandoned land Among the two parklands and live fence systems, C con tent was expected to be: n ear tree > under the crown > outside the crown, as reported in a similar study in the parkland system in South Mali, which showed the significant

PAGE 97

97 difference between near tree and outside the crown in 0 20 cm (although not significant in 20 40 cm) (Kate r et al. 1992). However, in this study, th e tree effect (distance factor) was only significant in the live fence but not in the two parkland systems (the trends of near tree > outside crown were observed in both parklands) ( F igure 5 4 ). From the field observation, this seems to be because of the frequent tillage ( a couple of times during the growing season) in the parklands Many studies reported that tillage cause d significant loss of soil organic C especially in the surfa ce (Gebhart et al. 1994; Campbell et al. 1996; Six et al. 1998) T illage using animal traction is done very close to the trunks of parkland trees where tree density is low and this accelerates the decomposition of organic matter in the top soil, and thus reduces soil C accumulation. Live fence trees are planted in high density and their thorny branches spread, making it almost impossible to do tillage near the tree lines. C con tent a round the live fence tree lines is higher due to the higher litter inpu ts from trees ( Figure 5 6 ) The positive correlation of C 3 (tree) derived C and L fraction (contains new SOM) C in the 0 10 cm soil ( Figure 5 10 ) also suggests that the recent planting of live fence trees and the input of litter have already contribute d to the accumulation of C in the topsoil The reason why fodder bank system s did not have much topsoil C ( Figure 5 4 ) could also be explained by the management style. Firs t, the land is tilled before planting the trees, which cause s the loss of aggregated C (L and M fraction C). Then crops are not grown inside the fodder bank after tree planting, and most of the tree leaves and branches are taken away as fodder. The low litter input after establishment also causes the low density of L and M fraction C ( Figure 5 5 ) Since S fraction C contains well protected C its con tent in fodder bank s was not affected much by the tillage, and stayed similar with other systems in all depths

PAGE 98

98 Long term influence of tree presence was observed in parklands data. S fraction C content (involving well protected C) was higher in deeper soil in parklands compared with live fence s or fodder bank s ( Figure 5 5 ). This is probably because of the long term inputs of litter and tree root biomass in the parkland system, compared to the other two systems (live fence and fodder bank) that represent land that was treeless (only crops) until six to nin e years ago. Also, the distance factor is significant among two parklands and live fence in M fraction C. M fraction C contains various forms of SOM including microaggregate associated C, which means trees contribute to increasing the protected form of S OC. This suggests that tree s help increase not only litter input and the content of unprotected C, but also facilitate a variety of protection to soil C from unprotected to protected state. T he content of C 4 origin C was higher than that of C 3 origin C when the whole data (whole treatment, whole depth) was tested. The tendency of higher amount of C 4 origin C in deeper soil layer and/or in S fraction was also observed ( Figure 5 6 5 7 ) In the studie d land use systems trees and bushes in abandoned land are C3 plant s whereas crops grown underneath the parklands and around the live fence as well as the presumed previous vegetation (crops) of fodder bank and abandoned land are C4 plants (sorghum and mi llet). Moreover, there are other isotopic variation/bias that could be considered in the use of the mass balance equation (Eq.5 2). The 13 C of plant (the main source of SOM) is known to var y depending on species and environmental factors or CO 2 concentration in the atmosphere (Tieszen 1991; Marino et al. 1991) These differentiations are relatively small compared with the large differenc e caused by the different photosynthetic pathway. Another considerable isotopic composition change is related to SOM decay. Over the decomposition process, 13 C value of soil organic C tend s to increase (less negative) (Balesdent and Mariotti 1996). The study using 13 C value for tracing

PAGE 99

99 the vegetation change in East African savannah suggests that using the mass balance approach to interpret soil profiles might lead to an under estimation of past C3 plant abundance (Gillson et al. 2004). Thus, the C3 or igin C, especially in deeper soil tend s to be underestimated, and this might be one of the reasons that C3 origin C in the deeper soil (tree root contribution) was not observed as much as expected. It is also probably due to the higher inputs of belowgrou nd biomass from crops in the sampling depth. A boveground crop residue is usually taken away as fodder in fodder bank but crop roots are annually left in the soil and become the source of soil organic C. Crop roots are also expected to have a faster deco mposition rate than tree roots that contain higher proportion of lignin and other substances that slow down the decomposition process. In addition, sampling depth (1 m) might not be enough to see more tree roots influence than crop roots influence. Most of Mali s soil research sets 40 cm as the crop root influence limit, but a study found a significant amount of sorghum roots close to or deeper than 60cm in a similar climatic condition (Jones et al. 1998). Tree roots, especially in semiarid area s are ex pected to go even deeper than 1 m (Jelt sch et al 1 99 6). Overall, soil organic C content in the stud ied systems w ere of relatively lower magnitude ( 1 to 6 g C kg 1 soil ) than in agriculture or agroforestry systems of other ecoregions This is possibly due, at least partly, to the rapid decomposition of organic C, which is known to be facilitated by the high temperature and low silt and clay contents (Schimel et al. 1994; Hassink 1997). The lower amount of M fraction C than S and L fraction C (when test ed with all treatments data combined) also suggests that most of the litter inputs are decomposed rapidly so that little is going to the process of becoming a more protected form of C. In many situations it is probably best for farmers to allow the major ity of the residues to be eaten by cattle in these systems rather than attempt to build soil organic matter.

PAGE 100

100 A conservative estimate of soil C sequestration potential through addition of organic matter such as plant litter and animal waste to these agric ultural soils is in the range of 0.1 0.3 Mg C ha 1 yr 1 (Batjes 2004). ICRAF is trying to promote live fence s and fodder bank s for land amelioration and count s C sequestration potential as one of the potential benefits. However, it is important to addr ess the possibility of causing net loss of soil C while converting abandoned land into live fence s or fodder bank s in this study region at the initial stage, because of the tillage factor at establishment. And because the subsequent land use practice s pr ovide low levels of litter input especially in fodder bank s, it may take a long time to regain the initial loss of soil C.

PAGE 101

101 Table 5 1. Soil profile characteristics for plots of the five land use systems used in the study at Sgou Region, Mali. Depth ( cm) Sand (g kg 1 soil) Silt (g kg 1 soil) Clay (g kg 1 soil) pH Bulk density (g cm 3 ) Faidherbia albida 0 10 865 69 66 6.7 1.42 Parkland 10 40 805 81 114 6.3 1.50 40 100 715 86 199 5.9 1.50 Vitellaria paradoxa 0 10 857 86 57 6.6 1.48 Parkland 10 40 809 96 95 5.9 1.38 40 100 798 82 120 5.6 1.47 Live fence 0 10 935 22 43 5.8 1.44 10 40 900 22 78 5.3 1.51 40 100 846 31 123 5.0 1.39 Fodder bank 0 10 830 72 98 5.2 1.42 10 40 797 91 112 5.3 1 .46 40 100 752 95 153 5.5 1.55 Abandoned land 0 10 694 129 177 5.4 1.36 10 40 576 164 260 5.3 1.21 40 100 530 164 306 4.9 1.43

PAGE 102

102 Table 5 2. 13 C value s of whole soil and three fraction sizes from five studied land use systems, at Sgou Region, Mali. (all values are average of three replicates)

PAGE 103

103 Figure 5 1. S oil sampling, S gou, Mali The samples were drawn with an auger fro m the defined soil depths and horizontal distances from trees ; each sample was a composite of four sub samples drawn from different points within a plot. (Photographed by author)

PAGE 104

104 A B C D E Figure 5 2. S oil pits dug in plots of the five land use systems studied in Sgou region of Mali. The red stick is marked (in black) at 10 cm intervals. A) F aidherbia albida parkland. B) Vitellaria paradoxa parkland. C) Live fence. D) Fodder bank. E) Abandoned land. (Photographed by author)

PAGE 105

105 Figure 5 3. Model of soil organic matter dynamics Source: Six et al. 2002 Figure 3 in page 163.

PAGE 106

106 Figure 5 4 Whole soil C content of three depth classes (0 10 cm, 10 40 cm, and 40 100 cm) in different land use systems in Sgou, Mali: A) Faidherbia albida parkland B ) Vitellaria paradoxa parkland C ) Live fence D ) Fodder bank and E) Abandoned land. Rang e of the each depth value is 95 % confidence level. Depth indicated is the mid point of sampled depth.

PAGE 107

107 Figure 5 5. S oil C content of three particle size fractions in three depth classes (0 10 cm, 10 40 cm, and 40 100 cm) under f ive l and use systems in S gou, Mali. A) Faidherbia albida parkland B ) Vitellaria paradoxa parkland C) Live fence D ) Fodder bank and E) Abandoned land. Range of the each depth value is 95% confidence level. Depth indicated is the mid point of sampled depth D E A B C

PAGE 108

1 08 Figure 5 6. Whole s oil C, d ivided into C 3 plants (trees) origin and C 4 plants (crops) origin in different soil layers up to 100 cm depth, in five land use systems in S gou, Mali. A) Faidherbia albida parkland near tree trunk B ) Faidherb ia albida parkland outside crown, C) Vitellaria paradoxa parkland near tree trunk D) Vitellaria paradoxa parkland, outside crown, E) Live fence near trees, F) Live fence, 3m away from tree lines, G ) Fodder bank and H) Abandoned land. A B C D D D E D D F D D G D D H D D

PAGE 109

109 Figure 5 7. So il C in three fraction sizes divided into C3 plants origin and C4 plants origin in different soil particle size fractions under different land use systems in Sgou, Mali. A) Faidherbia albida parkland near tree trunk B ) Faidherbia albida parkland outsid e crown, C) Vitellaria paradoxa parkland near tree trunk D) Vitellaria paradoxa parkland, outside crown, E) Live fence near trees, F) Live fence, 3m away from tree lines, G ) Fodder bank and H) Abandoned land.

PAGE 110

110

PAGE 111

111

PAGE 112

112 F igure 5 8. Linear regression betw een silt + clay content of soil and whole soil C content in three depth class es across five land use systems in Sgou region of Mali The three data points, one each in each depth class, marked by circles around them, refer to one of the fodder bank plots the data from which were quite inconsistent with those from the other fodder bank plots as well as all the other treatments; these points were therefore considered as outliers and excluded from the regression.

PAGE 113

113

PAGE 114

114 Figure 5 9. Linear regression between sil t and clay content of soil and C in soil particles of <53 m in three soil depth classes across five land use systems in Sgou, Mali. The three data points, one each in each depth class, marked by circles around them, refer to one of the fodder bank plots the data from which were quite inconsistent with those from the other fodder bank plots as well as all the other treatments; these points were therefore considered as outliers and excluded from the regression.

PAGE 115

115

PAGE 116

116 Figure 5 10. Linear regression between C derived from C3 plant s and C in the large soil particles ( 250 2000 m ) at 0 10 cm soil depth across five land use systems of Sgou region, Mali.

PAGE 117

117 CHAPTER 6 SOCIOECONOMIC ANALYSIS OF THE CARBON SEQUESTRATION POTENTIAL OF IMPROVED AGROFORESTRY SYSTEMS IN MALI, WEST AFRICA Introduction T he success in the implementation of any project for greenhouse gas (GHG) mitigation This is particularly so in a region such as this study site where the vast majority of inhabitants rely o n the outputs from their crop fields and animals for subsistence and cash income rarely exists in their household budgets. These farmers might be encouraged to plant trees in their croplands for potential carbon (C) benefits (cash payments) considering t hat agroforestry is a recognized GHG mitigation strategy according to the Kyoto Protocol. Several major reasons have been recognized as favoring introduction of C sequestration benefits into smallholders agroforestry practices in developing countries O ne is that this sequestration service does not need to be transported, thus, it can benefit people in remote areas, many of whom are poor. Secondly, there are no quality differences: a molecule of C is the same where ver it is located; so the problem often faced by smallholders in not being able to achieve the quality required by international markets in agricultural commodities does not apply here (Cacho et al. 2003a) Furthermore, even small amount s of additional income would make a great difference for t hese subsistence farmers who have practically no opportunity to make such additional cash income The political environment involvement in GHG mitigation projects. The U nited N ations F ramework C onvention on C limate C hange (UNFCCC) included development, equity and sustainability as conditions to be met when setting its principles for stabilizing GHG concentrations through mitigation policy (UNFCCC 2007) Large sca l e adoption of C sequestration activities by agroforestry in African countries could contribute to these objectives through biodiversity conservation, rural employment, and

PAGE 118

118 soil amelioration (Breman 1997; Tschakert 2007) The World Agroforestry Centre (ICRAF) announced that they were confident that establishment of agroforestry, especially in degraded land, would qualify and play an important role under the Clean Development Mechanism (CDM) of Kyoto Protocol (ICRAF 2007) A few major problems exist, however, in the context of promoting agroforestry p ractices by smallholders for enter ing CDM market F or example we need to examine cost per unit of C sequestration since there are many other options such as emission reduction or sequestration by large scale monoculture plantation (De Jong et al. 2004) Furthermore based on the feasibility studies, appropriate technical and political assistance should be provided so that smallholder of agroforestry owners will not always be less competitive than other mitigation projects options. Thus, socioeconomic fe asibility of the improved systems is an important consideration in the context of agroforestry implementation for GHG mitigation, and that is the scope of this chapter. Under the Kyoto Protocol, only newly sequestered C as a result of the project is reco gnized as tradable C ; t he traditional agroforestry systems such as parklands are, thus, not likely to be counted as C sequestration projects. The improved agroforestry systems that are currently being introduced in the study region can be qualified for tradable C credits The refore, the target agroforestry systems for this study are live fence and fodder bank in S gou region, Mali (detail ed system description in Chapter 4) Research Question s 1. W hat is the relative attractiveness of the two improved agr oforestry systems (live fence and fodder bank) considering their C sequestration potential, economic profitability, and social acceptability? 2. If C credit market s were introduced under the CDM of Kyoto Protocol would adoption of agroforestry provide more profits to land owners? If yes, how much?

PAGE 119

119 Materials and Methods The World Agroforesty Centre (ICRAF) conducted monitoring surveys for farmers implementing live fences and fodder banks after introduction of the systems (Hamer et al. 2005, van Duijl 2000). Data from these studies as well as databases from ICRAF research station were used for this study. Furthermore, field surveys were conducted during February March 2006 (the dry season after the harvests when farmers were less busy with agricultural act ivities) to collect additional data necessary for th e analysis. The target population was composed of farmers living in the S gou region who had adopted live fences and/or fodder banks with assistance of ICRAF. A comprehensive cost benefit study of live f ences had already been conducted for ICRAF by van Dorp et al (2005) A lso, the need and social acceptability of live fences and fodder banks had been discussed in several previous studies ( van Duijl 1999; Levasseur 2003; Yossi et al 2005). The informati on about the fodder bank implementation was much more scarce than that of live fence Thus, t he survey focus ed on collecting more data for fodder banks, specifically data to conduct the cost benefit analysis (CBA) equivalent to the existing live fence stu dy, as well as data such as the price of timber and non timber products from both live fences and fodder banks to conduct risk simulation analysis Social Survey of Fodder Bank Farmers The questionnaire was designed based on that of the live fence survey (Annex A) following the protocol of the Institutional Review Board of University of Florida (Protocol # 2005 U 1023). The structured questionnaire consisted of open ended and/ or close ended questions of 14 sections, asking for information about material s and labor used for managing and harvesting fodder banks as well as related benefits from the implementation.

PAGE 120

120 Twenty two farmers from 13 different villages in the S gou region were interviewed (the live fence survey was conducted on 18 owners from 15 di fferent villages). The language used was Bambara, the most common local language in Mali, although the questionnaire was made in French. A translator (French Bambara) an agronomist who had conducted social survey for ICRAF was hired to communicate wi th interviewees ( Figure 6 1 ). T o ensure his survey/interviewing skill, an experienced ICRAF officer went through the survey questionnaire with him before the real survey started and made him practice the follow up explanations in case farmers did not understand the questions The majority of fodder banks were 0.25 ha (50 m 50 m) in area since that was the default recommendation of ICRAF. Some interviewed farmers turned out to have larger or smaller sizes by the time of the s urvey due to the success or failure of the management. All the labor data and other costs were converted to per 0.25 ha basis before taking the mean. The live fence study was based on the average live fence row length, 291 m (van Dorp et al. 2005). Be cause the live fence and/or fodder bank installation was at least a couple of years ago, farmers seemed to have difficulties recalling the installation costs, especially labor (days and people) needed for planting and management. Also, it was very difficu lt to estimate the amount s of products harvested such as f odder and fruits. The sizes of the bags farmers were using to collect the harvests varied. Direct measurements of the bag size and the fodder weight (air dried) were conducted at several villages to reduce the estimation variability. Local Market Survey There are three local markets inside the city of S gou where most of the farmers in the villages go to buy/sell their products and equipment. Price data were collected from all markets although some products such as fodder were not sold in all the markets. T he average price of each item w as used for the analysis.

PAGE 121

121 Types of Analysis Cost benefit analysis (CBA) CBA involves weighing the total expected costs against the total expected benefits of one or more actions in order to choose the best or most profitable project, taking time into consideration (Campbell and Brown 2003) In this case, the it is profitable or not for farmers to start a live fence and/ or a fodder b ank. To appraise these projects, three decision rules were used: net present value (NPV) ; benefit cost ratio (BCR) and internal rate of return (IRR). The NPV of a project simply expresses the difference between the discounted present value (PV) of futu re benefits and the discounted present value of future costs In other words, that NPV is the sum of revenues in each year, y, discounted to year 0 minus the sum of costs in each year discounted to year 0. NPV = PV(Benefits) PV(Costs) = (Eq. 6 1) B y : Project benefits (revenues) of a given year y C y : Project costs of a given year y r: Discount rate/interest rate n: Project life, years According to the NPV guideline, a project is acceptable (profitable) if NPV is zero or greater. Projects with a negative NPV are unacceptable (Klemperer 1996). BCR is another way of determining whether the project should be accepted or rejected as an inve st ment. It is the present value of benefits divided by the present value of costs. BCR = = (Eq. 6 2)

PAGE 122

122 When PV benefits equals to PV costs, the BCR is 1, and NPV is 0. Also, if PV benefits exceed PV costs, BCR must be greater than 1 and if PV costs exceed PV revenues, BCR<1 Thus, according to the BCR decision rule, projects are acceptable when the BCR is 1 or greater, and unacceptable if BCR<1 (Klemperer 1996). The IRR of the project is the discount rate at which the NPV becomes 0 in the NPV formula (Eq. 6 1) (Eq. 6 3) The IRR is the rate of return earned on funds invested in a project. The equation 6 3 also says that the IRR is the interest rate at which PV benefits equals PV costs. A project is acceptable if its IRR is equal to or greater than the minimum acceptable rate of return (Klemperer 1996). In this study s case, however, the farmers will not have a specific acceptable rate of return. So, the rate can be recognized as acceptable if it is greater than the interest rate (when farmers t ake a loan from a local bank or financial institution ) Several basic budgets were available for calculating the above three decision rules such as whole farm budget, enterprise budget, partial budget and cash flow budget. Each budget is specific in its application, and the partial budgets were used in this study. Partial budget is used to evaluate the economic effect of minor adjustments in some portion of the business. Many changes that do not require a complete reorganization are pos sible in a farming business. Given a fixed set of resources, a farmer can employ these resources in more than one way in response to changes in product price levels, cropping patterns or carrying capacity. Partial budgets are useful to evaluate changes such as expanding an enterprise ( e.g. a crop), alternative enterprise, and different production practices (Dalsted and Gutierrez 2007) B ecause introducing a live fence and/or a fodder bank has limited impacts on the costs and returns of a farmer s budget due

PAGE 123

123 to their small size s, the partial budget was appropriate to use. Partial budgeting is based on the princip le that a small change in a farming business will have one or more of the following effects: 1 Eliminate or reduce some costs, 2 Eliminate or reduce some benefits, 3 Cause additional costs to be incurred, and 4 Cause additional returns to be received (Dalsted and Gutierrez 2007) The net effects, i.e. NPV, BCR, and IRR, can be calculated from those four components. Only the changes in costs and benefits that resulted directly from starting a live fence and/or a fodder bank were collected/ extracted from the survey data and taken to account in the analysis. The project cycle was set to 25 years. This is the expected rotation time for both li ve fence and fodder bank tree species (personal communication with Dr. Bocary Kaya, 2006 ). The growth curve was estimated based on the available data from biomass measurement. A discount rate of 15 % was used, drawing on the information about standard in terest rates applied by the several local institutions for micro credits (available to local farmers) in the study region (van Dorp et al. 2005). Data collection and analysis were done with local currency, FCFA. It is called CFA (Communaut Financiaire Af ricaine) franc, which is fixed against the euro at 1 = 655.96 FCFA in 2006. When the results are shown in US dollar for the reference, exchange rate of US $ 1 = 550 FCFA (average exchange rate during the field survey) was used Cost structure : Cost of seedlings w as calculated following the method of Tr aor et al. (2003): an aggregation of the cost of seeds and the labor cost needed to grow seedlings in the nursery. The average size of the fodder bank was 0.25 ha (50 m*50 m ) the length of live fence rows was 200 m (50 m*4 with 800 trees ) around to prote ct the fodder trees. Thus, planting and maintaining these trees were included in the costs of fodder bank

PAGE 124

124 To protect the newly planted trees, dead fences (made from dead bush brunches) were needed for the initial three years. Branches were obtained fr ee of charge in the wild; no cash costs were involved. To estimate the material cost of the dead fence, the volume of branches and the price farmers would be willing to pay on the market were asked, and recalculated to the standard size. The total averag e tool cost per farmer per year for the average size of live fence was estimated 1,000 FCFA ($1.82) (van Dorp et al 2005). Since the fodder bank requires more use of tools based on the data for required labor, the cost was set at 1,500 FCFA ( $ 2.73 ) per y ear To estimate the labor costs, farmers were asked for the average wage rate they pay for a hired labor. The most common daily labor wage (7 hours work) was 750 FCFA ( $ 1.36), ranging between 500 and 1 000 FCFA ($ 0.91 $ 1.82) depending on the seaso n. The respondents of the survey were also asked if they actually hired labor to install their fodder banks. The large majority of them did not; instead they used their family members including children, or exchanged the labor with neighbors. Labor tas ks are divided into: 1. Obtaining seedlings (or seeds) 2. Planting 3. Watering 4. Collecting materials for the dead fence 5. Constructing the dead fence around the live fence for protecting the seedlings (first three years) 6. Maintenance of the live fence/fodder bank ( weeding, replanting, pruning etc.) 7. Collecting products from the live fence and the fodder bank 8. Marketing live fence products (bringing to local market and selling) 9. Harvesting the timber/fuelwood at the end of the rotation

PAGE 125

125 From the survey, the time farmers spent on each of the above tasks were estimated for every year, and converted to the standard size and an average or median value was taken for the CBA analysis. Median values were used when the sample population had extreme outliers. In the cash flow ch art, the labor time was calculated to monetary value, using the average labor wage (Appendix B, C) Benefit structure : Yields from live fence trees are: Acacia nilotica : Fruits for tanning agent for the treatment of leather and traditional medicines are s old in local markets. Branches are used as material for dead fences. Acacia senegal : Bark is used as a traditional medicine, although it is not easily harvested, and not sold in market. Bauhinia rufescens : Leaves can be used as medicine, but not sold in the market. Lawsonia inermis : Leaves are transformed into powder to be used for the dying of hands and feet of women (cosmetic use), and highly valued in local markets. Ziziphus mauritiana : Fruits are edible, but mostly for home consumption. Benefits fro m these products were calculated in monetary value with local market price in van Dorp study (2005). That data were used in this study s cash flow. From the third year, farmers started harvesting the fodder (branches and leaves). Since the fodder tree, Gl ir icidia sepium is an exotic species and had not been commonly used in the study region no market value was available for its fodder; none of the farmers interviewed had sold its fodder in the market. T hus, the expected price of G. sepium fodder, if i t is sold in the market, was asked for in the survey, and the average of the answers w as used to estimate the fodder value. In addition fodder bank provides another source of revenue i.e. the saved time The owners were asked how many people and days t hey used to spend looking for fodder in the wild before the fodder bank installation, as well as after they started harvesting from their fodder banks. The difference is the saved time/labor, which they can use for other activities. The saved

PAGE 126

126 time was th en converted to monetary value using the labor wage, and counted as benefits of the fodder bank in the cash flow. Timber harvests at the end of rotation period (25 years) were also estimated by the projected growth lines. The local market prices of timbe r and fuelwood were used to estimate the monetary values. E ighty percent of trees planted in a live fence and a fodder bank is assumed to produce a small log per tree which can be sold in the local market at 700 1,500 FCFA ($ 1.27 $ 2.73) per one car t (about 40 logs). The rest of trees and all b ranches foliages etc. (about 40% of the total expected biomass) were assumed to be sold as fuelwood which is 4,000 6,000 FCFA ($ 7.27 $ 9.10) per one cart (250 300 kg) (personal communication with ICRAF o fficers and local merchants 2006 ). C sale : Price of C varies quite largely at the international market. This study used the average price for C emission trading in 2006, $42 per Mg C (World Bank 2006). Transaction costs [which are the costs of arranging a contract ( i.e. C sequestration project and consequent C credit sale ) and monitoring and enforcing the contract, as opposed to production costs (implementation costs of the project) ] were considered to be 0 in this cash flow. This is because the transac tion costs of agroforestry projects for C sequestration are usually covered by the third party such as the project s trust fund (Scolel T 2007). The trust fund deals with C buyers (companies or individuals in developed countries) for the trade, monitors the project performance, and provides the C payment and technical assistance to farmers. The payment method of C credit is also a long debated issue in the negotiation of the Kyoto Protocol and related meetings as discussed in Chapter 3 Two major paymen t methods (Cacho et al. 2003 b ) were tried in this study for comparison.

PAGE 127

127 Ideal accounting method : In this method, payments for C sequestration occur as the service is provided and a debit occurs when C is released (i.e. by fire or harvest). Farmers annuall y receive the payment according to the amount of C sequestered in their project s fields (live fence/fodder bank) The full debit at harvest means that the total amount of C credits sale received during the life of the project ( live fence and/or fodder ba nk ) are paid back to the investor by farmers. Tonne year accounting method : Although the ideal accounting system is ideal for the land owners (i.e. farmers), it is risky for investors because they are not sure the project will last until the end of the rotational age. The tonne year method does not require redemption of C credits upon harvest, because the payment occurs based only on the equivalent amount of permanently avoided emissions during a given year (Moura Costa and Wilson 2000). This method has the advantage that no guarantee is needed if the project will last a required number of years, as the annual payments are adjusted by the equivalent factor. This is a more favorable method for the investors, and politically popular (Hardner et al 2000 ). In this study, the equivalent factor of 0.0215 (Cacho et al. 2003a) was used. Farmers annually receive C credit payment only equivalent to the amount of C sequestered in each year 0.0215 but there is no payment back to the investors at the end of the project. These two methods were separately incorporated into the cash flow of both a live fence project and a fodder bank project (see Appendix B, C). The decision rules (NPV, BCR, and IRR) were calculated in three different cash flows: 1) Cash flo w without C sale, 2) Cash flow with C sale (ideal accounting), and 3) Cash flow with C sale (tonne year accounting). Sensitivity analysis The calculation of NPV, BCR, or IRR in the CBA described above is based on the best guess scenario where all variabl es of costs and benefits included in a cash flow are most likely

PAGE 128

128 values such as an average of a data set. However, the future is uncertain: we do not know with certainty what the future values of a project s costs and benefits will be. Sensitivity anal ysis is the simple process of establishing the extent to which the outcome of the benefit cost analysis is sensitive to the assumed values of the inputs used in the analysis (Campbell and Brown 2003). In this study, sensitivity of NPV to the change of maj or five input variable were tested and presented in the results section. The tested variables were : 1) discount rate, 2) seedling cost, 3) labor price (wage), 4) yield of harvests, and 5) C credit price. All variables except the discount rate were change d +/ 50 % from the best guess scenario to compare which variable would affect the NPV most. Discount rate was changed only +/ 5% because it was unreasonable to assume the local discount rate to change largely ( such as 50 % ) Risk modeling Risk analysi s refers to the identification and description of the nature of uncertainty surrounding the project variables using probability distributions. When there is no appropriate information on the expected range of values of input variable (= risks), only sensi tivity analysis can be done to observe the uncertainty of output projection. However, if some expectations are available for the occurrence of the variability of the input variable, it is possible to conduct the risk analysis. Each input variable has a r ange of possible values; high, medium, low; or, maximum, mean, minimum. Risk modeling describes the likelihood of occurrence of these input variables within the given range (probability distribution). The probability distribution for the out put (NPV) wil l then depend on the aggregation of probability distributions for these individual input variables into a joint probability distribution. The NPV probability distribution was simulated by a computer program called @RISK (Palisade Corporation). The prog ram performs a simulation known as Monte Carlo analysis, whereby the NPV of the projects is recalculated over and over again, each time using a different,

PAGE 129

129 randomly chosen, set of values of input variables The random selection of values is based on the ch aracteristics of each input variable s probability distribution. In this study, there was insufficient information for the distribution of each four input variable (labor wage, seedling cost, yield, and C credit price) Campbell and Brown (2003) recommen d ed the use of a triangular or three point distribution for this kind of analysis. This is the distribution described by a high, low and best guess estimate, which provide the maximum, minimum, and modal values of the distribution respectively. Each in put variable s distribution was set based on the surveys data, observation, and personal communication with ICRAF field officers. Labor wage distribution was likely to be between 500 and 1,000 FCFA (0.91 1.82 US$), while 750 FCFA ($ 1.36)/man/day was the most prevalent labor wage. Seedling cost distribution was assumed to be 50 % to +50 % from the best guess scenario. Harvests or yield distribution was assumed to be 0 % to +50 % from the best guess scenario. C price was assumed to have a range of $ 3. 7 to $ 99 per Mg C which was the minimum and maximum price traded in international C market in 2006 (World Bank 2006). Mean price was $ 42/Mg C as used in the best guess scenario. Results Demographic Characteristics of Target Population Demographic chara cteristics ( Table 6 1 ) were not based on data collected in the social survey conducted in this study However, the information on live fence farmer s was already available in the previous studies of ICRAF (Levasseur 2003; van D orp et al. 2005). T he fodder bank owners who were interviewed in this study were mostly in the same village or neighboring villages with similar condition s, and therefore their demographic characteristics were considered to be similar. Average household size is 27.7 persons, consisting o f 6.8 male members, 6.7 female members and 14.2 children. An active household member is a person actively

PAGE 130

130 contributing to agricultural activities (male 4.9, female 5.2, children 5.7 on average). Non active members are ge nerally sick or elderly or children under 10 years old (9.2 on average). Also, about 10 % of the household members have migrated from the village, usually for temporary labor in the town of S gou or other urban areas, to work or study. The average area of cultivated land owned by a household is 15.3 ha. The main enterprise is millet, occupying more than 50 % of total cultivating area. Other major crops include sorghum (2.3 ha), rice (2.1 ha) and groundnut (0.9 ha). On average, each household possesses 3.6 work oxen (for agricultural labor such as tillage), 5.8 cows (for breeding purposes), 2.1 donkeys, and other small livestock including sheep, goats, and poultry. Cost Benefit Analysis: Best Guess Scenario of the Live Fence and the Fodder Bank Based on the collected data, all costs and revenues (benefits) consisting of cash flow (year 0 to 25) were put into the spreadsheet. The cash flows of the live fence project and the fodder bank project in the best guess scenario are shown in Appendix B, C. The net benefit (total revenue s total costs) of each year was negative at the beginning and turned positive from the second year for both projects. Among components of costs, labor cost was the largest. Total labor cost throughout the project year conv erted to present value was 60,738 FCFA ($ 110.4) for the 291 m live fence project and 94,589 FCFA ($ 172.0) for the 0.25 ha fodder bank project Labor cost was high in the first three years for the live fence project compared with the rest of the project term, because of the initial management such as construction of dead fence for protecting seedlings. T he fodder bank project needed more consistent management practices (labor) than for the live fence project due to the fodder tree management such as weed ing and pruning. Seedling cost was a relatively large component of the costs on net cash flow of both projects, since it was initial investment (at year 0) and was not discounted to calculate the present value.

PAGE 131

131 When the components of revenues were exami ned, a big difference between the two projects cash flows was the revenue from the saved time of the fodder bank. Before the fodder bank installation, farmers had to graze animals almost everyday for quite a long time during the dry season. Since the su rvey data showed that this grazing time/labor a farmer could save by the fodder bank was con siderably high, the revenue of this component became the significant difference between the live fence and the fodder bank. Other revenue components, timber and fu elwood sale occurred at the end of the project year, thus, the revenues from them were discounted largely when converted to the present values Another revenue component, C sale was added to the cash flow with the ideal accounting method and the tonne ye ar method separately. The amount paid in US$ was converted to FCFA, and put into the cash flows. Three decision rules (NPV, BCR, and IRR ) for three different conditions ( No C sale, C sale with the ideal accounting method, and C sale with the tonne year accounting method ) were calculated ( Table 6 2 ). C sale by the ideal accounting method significantly changed all three decision rules. NPV of the live fence was 52 802 FCFA ($ 96.0) without C sale, and it increased to 60 465 F CFA ($ 109.9) with C sale by the ideal accounting method. NPV of the fodder bank was 87 319 FCFA ($ 158.8) without C sale, and 96 394 FCFA ($179.3) with C sale by the ideal accounting method BCR and IRR also increased (economically more profitable) with C sale. However C sale by the tonne year method did not increase the three decision rules much from those without C sale In NPV, only 172 FCFA ($0.3) increase for the live fence and 204 FCFA ($0.3) increase for the fodder bank were observed compared with NPV without C sale. The C sale profits by the tonne year method were too small to make a change of BCR and IRR values for both live fence and fodder bank projects.

PAGE 132

132 Sensitivity Analysis Sensitivity analysis was conducted, changing five major input v ariables separately : when one variable is changed, others are not changed ( Table 6 3 ) Two scenarios, with or without C sale were tested to see the NPV sensitivity. The ideal accounting method was used for calculating the NP V of with C sale since the CBA results ( Table 6 2 ) showed that the tonne year accounting method hardly changed the NPV or other two values from No C sale scenario. When the discount rate was changed from 15 %, the NPV cha nged greatly in both live fence and fodder bank projects Seedling costs change (+/ 50 %), on the other hand, did not fluctuate the NPV much compared with other variables in both projects Labor price changes (+/ 50%) affect ed the NPV of the live fence project and the fodder bank project differently, causing a large change in the NPV of live fence while causing very little in the NPV of fodder bank. This is because the labor wage variable was used for calculating both costs (labor) and benefits (revenu es from the saved time) of the fodder bank project cash flow. The variable, which has the largest impact on NPV values, was yield (harvest of fodder, live fence products and timber). When the yield was tested with +/ 50 % the range of NPV was largest in both live fence and fodder bank projects. T h e NPV values became even negative ( meaning: the project is economically unacceptable) for the live fence project when yield is 50 % from the best guess scenario C price change (+/ 50 %) did not change the N PV of both projects largely, suggesting C price is not an influential factor to change the projects profitability Risk Modeling and Simulation The risk simulation program @RISK was run based on the best guess scenario cash flow ( shown in Appendix B, C) w ith the major four variable s range described in the Materials and Methods section. The mean of the NPV distribution of the live fence project without C sale was 28,730 FCFA ($ 52.2), and the 90 % confidence range was from 50,178 FCFA ($ 91.1) to

PAGE 133

133 96,546 FCFA ($ 175.7) ( Figure 6 2 ) The chance of the NPV being negative (meaning: the project is economically not acceptable ) was 26.38 %. The net cash flow, each year s total revenue minus total costs, of the project was also simul ated ( Figure 6 3 ) This shows the probability distribution of the net benefit in each year from year 0 to 25. According to this simulation, the net benefit will be positive with 95 % likelihood from the third year of the projec t Then the net benefit is likely to be stable, and will increase significantly because of the harvests of timber and firewood at the end of the rotation. C sale was added to the live fence project cash flow, and the risk simulation was run again. I t w as conducted only with the ideal accounting method, because C sale by tonne year accounting method change d NPV very little from that of No C sale scenario Adding C sale changed the NPV probability distribution T he mean NPV of the distribution increas ed to 36,058 FCFA ($ 65.6), and 90 % confidence range was from 52,713 FCFA ($ 95.8) to 110,069 FCFA ($ 200.1) ( Fig ure 6 4 ). The chance of the NPV becoming negative is 24.47%, slightly less than that without C sale. The two pr obability distributions (with or without C sale) were found to be significantly different, when compared using t test ( p <0.01). The @RISK program also conduct ed a regression sensitivity analysis, which is able to show how each input variable is influentia l for the NPV simulation ( Figure 6 5 ). Yield has a positive as well as the largest impact among the input variables. Both labor wage and seedling cost had negative coefficients (when the variable increase s the NPV would decrea se s ), but relatively small extent. C price although positive, had the smallest influence on the NPV projection. Mean of the NPV distribution of the fodder bank project without C sale was 63,153 FCFA ($ 114.8), and its 90 % confidence range was from 53, 386 FCFA ($ 97.1) to 161,313 FCFA ($ 293.3) ( Figure 6 6 ) The chance that the NPV becomes negative (economically not acceptable)

PAGE 134

134 was 19.15 %. P robability distribution of the net benefit of each project year (0 to 25) was simul ated ( Figure 6 7 ). The net benefit turned positive in 95 % confidence from the fourth year of the project. The distribution is much more largely spread from the mean than the same simulation of the live fence ( Figure 6 3 ). C sale was added to the fodder bank cash flow and the simulation was run again. The NPV probability distribution of the fodder bank project with C sale (by the ideal accounting method) is shown in Figure 6 8 T he mean NPV became 63 289 FCFA ( $115.1 ), slightly more than that of without C sale simulation T he 90 % confidence range was from 54,305 FCFA ($ 98.7) to 159,301 FCFA ($ 289.6), and the chance of NPV being negative was 19.12 %. The distribut ions of with C sale and without C sale were compared; they were not significantly different in t test. The sensitivity regression ( Figure 6 9 ) showed again that the yield was the most influential and positive factor for the fodder bank NPV simulation. C price remained to be relatively small and positive variable. Seedling cost and labor wage were both negative variables, same as the results of the live fence project, but the impact of labor wage was much smaller in the fodd er bank simulation. This is because labor wage variable was used for calculating both labor costs and benefits (the saved time) of the fodder bank project, while it was used only for calculating labor costs in the live fence project. Discussion Overall, C sale seemed to increase the profitability of both live fence and fodder bank systems for farmers. However, if the tonne year accounting method is applied, the amount of C payment per farmer will be too little; it will not be attractive for farmers to p articipate the C sale program. Even with the ideal accounting system, the benefits from C sale will not be the major part of the farmer s income. The live fence or the fodder bank project can provide multiple benefits, and the best guess scenario shows i t is likely to be profitable without C sale. If farmers

PAGE 135

135 can gain the C payment without changing the use of the live fence or the fodder bank and without paying some of the transaction costs, which were the assumption s of this analysis, the payment does in crease expected NPV. And there is no reason for farmers not to participate the C sale program, if it exists As mentioned in the beginning of this chapter, even if the amount of cash income is very small (in the perspective of C credit buyers in develope d countries) it will make a big difference to the economic situation and well being of the farmers in the studied region. For investors, these agroforestry projects might not be very attractive regarding the amount of C credits they can purchase. However the price used in the simulation was the same as in any large scale C mitigation project. In addition, contributing to the socioeconomic development of communities in Africa could provide an environmental friendly image to the companies/corporations who are often blamed as responsible for large amount s of GHG emission; this image value may add more attractiveness for the companies to invest in this type of C sequestration project. The added C sale increased the mean of the simulated NPV significantly for the live fence project, but not for the fodder bank project, which suggests that C sale is likely to have more economic impacts for the live fence owners than for the fodder bank owners. In both the best guess scenario analysis and the risk simulation analysis yield was the most influential and uncertain factor among the input variables. A ccording to the survey data farmers who had larger trees and more harvests tend ed to spend more time for watering and weeding during the initial years of the proje ct Thus, proper training for the initial year treatment could significantly increase the expected yield for both live fence and fodder bank. On the other hand, the annual precipitation, which influence s the tree growth greatly varies largely in the stu dy region and it risks the expected yield Future climate shifts is unknown, but it would change the

PAGE 136

136 amount of labor for the management ( especially watering) required for each project, which will change the project s attractiveness. A nother issue is th at the assumed labor wage (750 FCFA/man/day) is a major factor affecting the project s profitability. From the interviews, it was clear that the rate used in the analysis was prevalent in the region but the occasions when farmers were hired for the manual labor and received this amount of wages were rather very few, for example during busy farming season s such as harvesting. Thus, the real opportunity cost of the labor may be considerably less than the assumption, substantially lower. As shown in the NPV probability distribution of the simulations (Figure 6 2 6 4 6 6 6 8 ), the mean NPV of each distributi on was much smaller than that of the best guess scenario ( Table 6 2 ). The best guess scenario s NPVs were at the highest peak of each probability distribution. I t means that the risk and uncertainty of the project were somewh at taken into account in the simulation s and suggests that evaluating the project s profitability only by the best guess scenario may overestimate the project s expected profitability. Both the best guess scenario and the risk simulation suggest that th e fodder bank project has larger expected profits than the live fence project, although the scale of these two systems are not same and cannot be compared as two options. In reality, much more farmers adopted the live fence than the fodder bank in the reg ion. This is probably because live fences already exist ed somewhat in the form of traditional live fences or dead fences, and farmers do not need large parcels of land or extra labor inputs. The fodder bank is a rather new concept that introduces exotic species and requires larger areas of land and extra labor. These factors seem to be the heavy burden for farmers, especially for relatively poor ones. In addition, the probability distribution of the fodder bank project was more horizontally spread than that of the live fence

PAGE 137

137 project (Figure 6 2 6 6 ) and the net benefit flow of the fodder bank had wider probability range than that of the live fence (Figure 6 3 6 7 ) This suggests that the fodder bank has a higher chance of NPV fluctuation. Overall, the fodder bank seems to be a high risk, high return project than the live fence. The majority of the farmers in the region are subsistence oriented and are expected to be very risk averse. Profitability may not be the first consideration in the ir adoption process. Other factors, such as water or labor availability and the presence of fodder in the open land nearby might be the more important factor s in the adoption of the fodder bank, if they are to be promoted.

PAGE 138

138 Table 6 1. Demographic characteristics of the target population in S gou, Mali. Average household size (Number of people) Active 1 male 4.9 Active female 5.2 Active children 5. 7 Non active male 0.7 Non active female 1.4 Non active children 7.1 Migrated 2 2.7 Total 27.7 Cultivating area (ha) Millet (Pennisetum glaucum) 6.4 Sorghum ( Sorghum bicolor) 2.3 Rice ( Oryza glaberrima and Oryza sativa ) 2.1 Groundnut ( Arachis hypogaea ) 0.9 Chickpea ( Cicer arietinum ) 0.6 Cassava ( Manihot esculenta ) 0.6 Fonio ( Digitaria exilis ) 0.6 Watermelon ( Cucuribitaceae ) and other fruits 0.9 Vegetable and others 0.9 Total 15.3 Cattle possession per household Work oxen 3.6 Cows 5 .8 Donkeys 2.1 Sheep 4.5 Goats 4.7 Poultry 21.7 Horses 0.1 1. Actively working in agricultural activities. 2. Temporarily moving out from the village. Data from van Dorp et al. 2005. Table 6 2 N et P resent V alue (N PV ) B enefit C ost R atio (B CR ) an d Internal R ate of R eturn ( IRR ) of the live fence and the fodder bank projects in the three different scenarios (without C sale, with C sale by the ideal accounting method, and with C sale by the tonne year accounting method) in Sgou, Mali Live Fence Fodder bank No C sale Ideal accounting Tonne year accounting No C sale Ideal accounting Tonne year accounting NPV (FCFA) 52 802 60 465 52,974 87,319 96,394 87,523 BCR 1.53 1.60 1.53 1.67 1.74 1.67 IRR 25.5% 27.3% 25.5% 29.5% 31.4% 29.5%

PAGE 139

139 Table 6 3 NPV sensitivity of the live fence project and the fodder bank project to the change of an input variable in Sgou, Mali Live fence Fodder bank No C sale With C sale No C sale With C sale ----------FCFA ----------Ba se 52 802 60,465 87,319 96,394 Discount rate 5 % 109 367 118 391 166 434 177 304 Discount rate +5 % 20 999 27 478 43 094 50 670 Seedling cost 50 % 68 829 76 492 101 844 110 919 Seedling cost +50 % 36 775 44 438 72 794 81 869 Lab or price 50 % 83 171 87 002 88 750 93 287 Labor price +50 % 22 433 26 264 85 888 90 425 Yield of harvests + 50 % 129 330 140 825 149 624 163 236 Yield of harvests 50 % 23 727 19 894 25 014 19 551 C price +50 % 64 297 100 931 C price 50 % 56 633 91 856 Source: Base values are from the best guess scenario cash flow. NPV values of with C sale are from ideal accounting method scenario.

PAGE 140

140 Figure 6 1. S ocial survey with farmers in S gou, Mali. Based on the ICRAF da tabase, all farmers who have at least once harvested fodder from the fodder bank were interviewed. The survey was conducted in Bambara (local language) and translated to French through the interpreter (man with a jacket in the photos) Figure 6 2. Simu lated NPV probability distribution of the live fence project (without C sale). The distribution is likelihood (y axis) of the project s NPV (x axis): the worst scenario can be less than 100,000 FCFA in NPV, and the best scenario can be close to 150,000 F CFA in NPV. The peak of the distribution is most likely (best guess) scenario of the project.

PAGE 141

141 Figure 6 3. Simulated net benefit (total costs total revenues in each year) of the live fence project (without C sale). Mean value of each year s probabi lity distribution is shown in the yellow line, red range is plus minus 1 standard deviation from the mean, and green range is the 5 to 95 % likelihood of the value.

PAGE 142

142 Figure 6 4. Simulated NPV probability distribution of the live fence project (with C sale by the ideal accounting method). The distribution is likelihood (y axis) of the project s NPV (x axis): the worst scenario can be less than 75,000 FCFA in NPV, and the best scenario can be close to 150,000 FCFA in NPV. The peak of the distribution is most likely (best guess) scenario of the project.

PAGE 143

143 Figure 6 5. Regression sensitivity analysis for NPV of the live fence project (with C sale by the ideal accounting method). Standard b coefficients show how these input variables are related to the results (NPV). Positive (negative) value means the input variable positively (negatively) affect the NPV, and the absolute value represents the extent of the influence.

PAGE 144

144 Figure 6 6. Simulated NPV probability distribution of the fodder bank project (without C sale). The distribution is likelihood (y axis) of the project s NPV (x axis): the worst scenario can be less than 100,000 FCFA in NPV, and the best scenario can be up to 200,000 FCFA in NPV. The peak of the distribution is most likely (best g uess) scenario of the project.

PAGE 145

145 Figure 6 7. Simulated net benefit (total costs total revenues in each year) of the fodder bank project (without C sale). Mean value of each year s probability distribution is shown in the yellow line, red range is pl us minus 1 standard deviation from the mean, and green range is the 5 to 95 % likelihood of the value.

PAGE 146

146 Figure 6 8. Simulated NPV probability distribution of the fodder bank project (with C sale by the ideal accounting method). The distribution is lik elihood (y axis) of the project s NPV (x axis): the worst scenario can be close to 100,000 FCFA in NPV, and the best scenario can be up to 200,000 FCFA in NPV. The peak of the distribution is most likely (best guess) scenario of the project.

PAGE 147

147 Figure 6 9. Regression sensitivity analysis for NPV of the fodder bank project (with C sale by the ideal accounting method). Standard b coefficients show how these input variables are related to the results (NPV). Positive (negative) value means the input vari able positively (negatively) affect the NPV, and the absolute value represents the extent of the influence.

PAGE 148

148 CHAPTER 7 SUMMARY AND CONCLUSION S This dissertation study examine d the carbon (C) sequestration potential of major agroforestry practices in S gou Region, Mali of the West African Sahel (WAS), and analyze d the feasibility and socioeconomic characteristics of the selected agroforestry systems in the context of C sequestration service s The selected systems were two traditional parkland agrofores try systems with Faidherbia albida or Vite ll aria paradoxa as the d ominant tree species two newly introduced (improved) agroforestry systems (live fence and fodder bank), and a so abandoned (degraded) land. The research revolved around four major questions. 1. How much C is stored in different agroforestry systems aboveground and belowground? 2. Do trees contribut e to store C in soil, and how stable is that C ? 3. W hat is the overall relative attractiveness of each of the selected agroforestry systems in te rms of its C sequestration potential ? 4. If C credit market s were available, would adopting agroforestry provide more profits to land owners? C Sequestration P otential Biophysical Potential T he selected agroforestry systems proved to have potential s for sequ estering more C both above and belowground than the tree less cultivated land in the study region However, the estimated amount s of C stored in the se systems and sequestered after the systems establishment are quite vari able depending on the baseline ( without project) status a s well as the accounting method used T he t wo traditional parklands store significant amount s of C in the biomass C. Especially, the large F. albida trees ( average DBH 59.4 cm, height 13 m) store considerable amounts of C. Howe ver, parklands are not likely to be considered for C sequestration projects any time soon

PAGE 149

149 because Kyoto Protocol currently admits only C sequestered as a result of newly im p lemented mitigation project s, and traditional land use systems such as parklands an d f orest conservation sustaining parklands are not recognized as emission reduction activities under the Protocol at least until 2012 On the other hand, i mproved agroforestry systems ( live fence and fodder bank) were found to have a better chance to be recognized as C sequestering activities than parkland systems. Because these systems are newly introduced, most of the biomass stored in the systems can be counted as sequestered C credits although their potential s as C sequestration projects were not as high as expected. C sequestration potential of a land use system has to be expressed on a unit area basis for a given period of time. From that perspective, some i mproved agroforestry systems ( live fence and fodder bank, in this study) do not rank high because of the nature of their planting configurations and/or management requirements. Live fence trees are densely planted rows are on plot boundaries an d therefore the area occupied or influenced by a fence row in relation to the total area of the plot it borders is low. As far as the fodder banks are concerned, the f odder trees that are frequently harvested for their leafy biomass cannot obviously be ex pected to store large quantities of biomass C. Therefore, the absolute amounts of C stored in these systems per unit area would not be as large as that for, say, parklands While the amounts of biomass C stored (calculated from general allometric equatio ns following UNFCC C guidelines) were 54.0 and 22.4 Mg C ha 1 respectively for 40 year or older stands of F. a lbida and V. p aradoxa parkland systems, the amounts were 4.7 Mg C ha 1 for a 8 year old stand of live fence and 2.2 Mg C ha 1 for 6 9 year old s tand of fodder bank.

PAGE 150

150 Another issue is that the initial C loss (both in biomass and soil) resulting from land clearing and tillage for facilitating tree establishment in these improved practices is expected to be significant ; this loss may not be compensat ed by the planted tree s any time soon given their slow growth rates owing to poor soil fertility and adverse climatic conditions. Therefore, introducing the se improved systems in abandoned land for land amelioration as the World Agroforestry Centre (ICR AF) is promoting, m ay not make a significant contribution to net C sequestration in the near term; indeed it is likely to cause net negative C balance in the initial stage of implementation. Currently soil C is not considered to be tradable but the rela tive portion of soil C in the studied systems turned out to be comparatively large For example, the percentages of soil C (0 100 cm) in total C (biomass C + soil C 0 100 cm) stock of the studied agroforestry systems were 38 % in F. albida parkland, 5 5 % in V. paradoxa parkland, 84 % in live fence, and 94 % in fodder bank. This can not be ignor ed when the potential for long term storage is considered. Soil fractionation studies and isotopic ratio measurement s showed that tree litter tend s to increase unprotected, relatively new C on the surface soil. In the deeper soil, t he parklands that have had trees in the system for a long time were likely to h o ld more protected C than the newly introduced live fence or fodder bank systems. Also, management pra ctices such as tillage and litter usage seemed to have a large influence on soil C accumulation. Socioecomic Potential T he cost benefit analysis suggested that live fence and fodder bank were likely to be profitable for farmers, whether with or without C sale. C sale changed the profitability: $ 13.9 more in net present value (NPV) of average size live fence (291 m), and $ 20.5 more in NPV of average size fodder bank (0.25 ha). These estimations are based on the assumptions of 25 year rotation, no trans action costs on the farmers and an accounting method ideal to C sellers

PAGE 151

151 (farmers) ; all of these assumptions are subject to change. With the accounting method that is in favor for the investors, the expected profits from C sale in the same model produced an increase in NPV of only about $ 0.3 in both systems. Even in the local currency with farmer s monetary values $ 0. 3 increase in NPV is almost nothing for a 25 year project. Thus, it is clear that the accounting method is a very strong factor to deter mine whether C sale through agroforestry should be introduced to the region. Also, sensitivity analysis and r isk analysis showed that C price did not have a major influence on changing the cost and benefit flow of both systems. I t would contribute to th e increase of profitability but had relatively small effect compared with other major variables such as yield, labor cost, and seedling cost. On the other hand, t ree growth (yield) ha d a strong influence on the project s profitability. Regression sensiti vity analysis showed that the effect of yield was 10 times or more strong than that of other factors. Since natural environment such as rainfall and pests greatly affects tree growth, the uncertainty (risk) regarding yield ( whether tree would grow expecte dly or not) is quite difficult to control and is a major discouraging factor for applying the improved agroforestry in general, whether with or without C sale. Regarding relative attractiveness of live fence and fodder bank, it was difficult to compare b ecause the scale of the project (the land needed) was different in the simulation of two systems. The f odder bank project that needs about three times more land (and correspondingly higher labor cost) than live fence project showed larger range of expecte d NPV ($ 98.7 to $ 289.6 in fodder bank vs. $ 95.8 to $ 65.6 in live fence ) Considering that improved agroforestry systems require additional work for farmer s cultivati on practices and that their resources (land, money for buying seeds etc.) are ve ry limited live fence would be easier and less risk y project for them to implement.

PAGE 152

152 C onclusion s Regarding the overall attractiveness of the selected land use systems, live fence and fodder bank are more suitable to start as agroforestry C sequestration p rojects than the parkland systems for small scale farmers in the studied region Between the two improved systems, live fence has higher C sequestering potential per unit area and is less risky than fodder bank s. T his situation could change, however, dep ending on tree management and conditions that affect tree growth. Adopting these systems on cultivated land rather than on abandoned land is likely to sequester more C and be more profitable. Since parklands are traditionally practiced they do not quali fy as a new C sequestration project. Nevertheless, F. albida trees are more attractive than V. paradoxa trees in terms of C sequestration potential. Agroforestry Adoption for C sequestration in the Study Region Based on the findings summarized above and information acquired through fieldwork and literature review, some factors that either limit or favor agroforestry adoption in the regions can be identified. Limiting Factors With the current price range (and its large fluctuation s ) for C credit s and the amount of C potentially sequestered the income from C sale is not likely to be a major source of income for farmers in the WAS and therefore is not likely to be a strong incentive to start the live fence or fodder bank. I n addition, farmers are concern e d about other factors as well, such as risks in undertaking a new farming practice. Farmers might need some support such as technical and/or material assistance to cover initial costs, and/or insurance or some kind of safety net in case the trees die due to unexpected causes. As regards parklands increasing the tree density is difficult because parkland trees grow relatively slowly Also, it is technically challenging since parkland tree species rely on natural regeneration.

PAGE 153

153 Soil C estimation is far m ore labor intensive and costly than biomass C estimation, and methods of measurements/accounting are still under discussion. Even in the same land use system, the amount of soil C per unit area can vary depending on the depth of sampling and/or the instru ments used to measure C content. C storage is related to s oil properties which makes it difficult to standardize soil C sequestration potential for any land use. Whether all C or only protected C should be counted is another issue I f only protected C is to be counted its method and ease of determination would become an issue. Based on farmer interviews observations, it appeared that relatively rich farmers were the ones who tried the improved agroforestry systems as ICR AF recommended, and succeed ed which is not different from the experience with many (or most) agricultural development initiatives In order to achieve poverty alleviation through C credit sale, it is important that the poorest poor of the region can ado pt the technology Involving farmers with little resource s needs naturally, extra support. Since C sale is not likely to provide much income under current condition s covering the cost of assistance and transaction costs for C trade would be a large fin ancial burden. Institutions such as international NGOs or national/local governments will have to be encouraged to bear these costs. Favorable Factors Some of the successful live fence and fodder bank owners started their projects by themselves without ICRAF s support. These successful projects had demonstration value too in that farmers of o ther villages were interested in starting live fence and fodder bank by looking at those pilot plots and came to request ICRAF s support in their villages This st rongly indicates the interests in the products and effects of live fence and/or fodder bank. As the land degradation proceeds and more farm fields are abandoned, the ability of live fence and

PAGE 154

154 fodder bank for pr otecting farmlands from free r oaming animals and producing fodder is expected to make them even more attractive for farmers. In the economic analysis, labor cost was converted to a monetary value using the local labor wage. However, since there are very few employment opportunities for farmers most occurring only during the harvesting or planting season, the real opportunity cost (labor cost) might be much lower than the assumed rate. This would lower the cost of live fence and fodder bank project s than the model used in the analys is and consequently, increase their profitability Thus, profitability of the improved system could be larger than the values shown in C hapter 6, although the extent is unknown The situation or understandings of C sequestration project as a mitigation activity is also changing. Climate change is a major global issue these days. Both price and amount of traded C are rising rapidly every year. After the first commitment period (2008 2012) of the Kyoto Protocol soil C m ay be counted for sequestration projects, which would increase the C sale income for the studied agroforestry projects. Also, at the international negotiation regarding rules of the Kyoto Protocol, conservationists such as the Nature Conservancy and researchers point out the importance of the forest conservation effect prevent ing CO 2 emission from deforestation and suggest the conservation cost to be shared internationally T hus, in future, the system steadily storing certain amount of C, such as parkland systems, m ight also be r ecognized as mitigation projects Implications for Agroforestry Economic benefits of establishing the improved agroforestry practice s were clearly found in the studied region. V arious social and environmental benefits were also found such as increasing soil organic matter and preventing soil erosion through introducing trees in the agricultural practices (in both traditional and improved agroforestry). Some of the non

PAGE 155

155 marketable benefits such as C sequestration are not directly perceived by the individ ual farmer. From C credit payment farmer s can at least receive some rewards for the environmental service they provide (internalize the non market values). In this sense, C credit sale is an option to charge for many of the non market services agrofores try systems provide. When the C sale through CDM becomes more popular in the future, agroforestry systems will definitely have potential to promote economic development of subsistence farmers as well as environmental improvement in developing countries. Future Research For the biophysical aspect s of the C sequestration project study, there is a strong need for more studies on soil C dynamic s. One of the reasons why soil C is not recognized as tradable C currently is the lack of information. To start the small scale C sequestration projects such as agroforestry, it is necessary to develop guidelines similar to the biomass C estimation guideline by UNFCCC. Conduct ing direct soil C measurement for each small scale project will be too costly for the projec t to be attractive. E stablish ing guideline s or default values would however, be quite challenging and controversial both academically and politically. In order to administ er agroforestry projects for C sequestration, an organization such as a project trust fund is needed to bear the transaction costs such as costs of monitoring and certifying the C sequestered, providing technical assistance and C payments, and selling accumulated C credits to the buyers at the international C market. It will be worth while to launch such a new pilot project the type of which has never been attempted in the study region or entire Africa, to understand a trust fund s responsibilities including designing the project, explaining to farmers the project objectives and provi ding technical/material assistance conducting inventories related to C, and setting the baseline Indeed, such a research project would be essential for promoting C sequestration through agroforestry in the WAS

PAGE 156

156 APPENDIX A SOCIAL SURVEY QUESTIONNAIRE FOR FODDER BANK OWNERS Questionnaire #: Date: Duree approximative: BANQUES FOURRAGERES 1.1 (Entrer code)_____________________________________ 1: Semi direct 2 : Plants a ra cines nues 3: Plants en pots 4: Autres 9: Pas de rponse 1.2 Comment vous avez obtenu les semences ou les plants? (Entrer code)_____________________________________ 1: Cultivs en ppinire par le paysan tre paysan / au march 5: Autres 9: Pas de rponse 1.3 Si vous avez achet les semences ou les plants, quels taient les cots? Unit Prix par unit Quantit Cot total Semences Plants 1.4 Outils utiliss pour installer et maintenir les banque fo ur rageres: Note es banque fourrageres pas les outils uniquement utiliss pour le travail champtre agricole. (preferer le prix de achete, mais si difficile, not er quel annee). Outils(entrer codes): Nombre (prix de march en F CFA) Valeur total Nombre 1: Houe 2: Hache 3: Pioche 4: Coupe coupe 5: Pelle 6: Piquet/Piquasse 7: Brouett e 8: Arrosoir 9: Charrue 10: Multiculteur 11: Charrette 12: Barre mine 13: Autres 99: Pas de rpons

PAGE 157

157 1.5 est ure? 1.6 personne/jour)? Note: Spcifier si le repas est compris dans le taux. 1.7 Personnes impliqus dans les diffrentes activits pour de l es banques fourrageres (pendant les premires trois annes): Notes: Expliquer comment vous tes arriv au nombre de personne/jours. le An 3: 1999; An 4: 2000 etc.) tape 1: Obtenir les plants ou les semences (PAS la production des plants en ppinire!) Personnes impliques (H/F/E): No. de personne/ jours (heures) An 1:___ Note: tape 2: Transplanter les plants ou semences au champ (inclure la haie vive) Personnes impliques (H/F/E): No. de personne/ jours (heures) An 1:____ Note: tape 3: Arroser le s banque s fourragere s Personnes impliques (H/F/E): No. de personne/ jour s (heures) An 1:____ No. de personne/ jours (heures) An 2:____ No. de personne/ jours (heures) An 3:___ Note :

PAGE 158

158 tape 4: Chercher du matriel pour la haie morte autour de la haie vive Personnes impliques (H/F/E): No. de personne/ jours (heure s) An 1:____ No. de personne/ jours (heures) An 2:____ No. de personne/ jours (heures) An 3:___ Note : tape 5: Construire la haie morte autour de la haie vive Personnes impliques (H/F/E): No. de personne/ jours (heures) An 1:____ No. de p ersonne/ jours (heures) An 2:____ No. de personne/ jours (heures) An 3:___ Note : tape 6: Entretien de s banques fourrageres et la haie vive (suivi, boucher les espaces etc. SANS r colte ) Personnes impliques (H/F/E): No. de personne/ jours (heu res) An 1:____ No. de personne/ jours (heures) An 2:____ No. de personne/ jours (heures) An 3:___ Note : An 4, 5, 6.... (si possible)

PAGE 159

159 1.8 Utilisation des produits des banques fourrageres et la haie vive pendant la dernire anne: Espce Produit (en trer codes) Utilisation de produit (entrer codes) Unit de rcolte (par ex. 1 sac) unit (en kg.) Production annuelle (en units) G.Sepium P. Lucens Produits: 1: Feuilles 2: Branches 3: Bois 4: Fruits 5: Ecorces 6: Racines 7: Fleures 8: Semences 9: Pas de rponse Utilisation: 1: Alimentation 2: Mdicaments 3: Bois de chauffe 4: Bois de service 5: Parure (beaut) 6: Fourrage 7: Tannage 8: Autres 9: Pas de rponse 1.9 Distribution des produits des banque s f orrageres et la haie vive pendant la dernire anne (en Espce Produit (voir ci dessus pour les codes) Auto consommati on (en units) Dons (en units) Vente (en units) Rserv (en units) G. Sepium P. Lucens

PAGE 160

160 1.10 L es prix et la valeur de la vente au march (en F CFA pour 2006) Note: Si le produit se ne vend pas au march, estimer le prix. Espce Produit (voir ci dessus pour les codes) Prix minimum (par unit) Prix maximum (par unit) Prix moyen Valeur de la vente to tale (en 2006) G. Sepium 1.11 Personnes impliqus dans les diffrentes activits pour la rcolte, la transformation et la vente des produits de les banques fo u r r ageres ou/et la haie vive ( partir de la troisime anne) Notes: Spcifier pour les diffrents espece Expliquer comment vous tes arriv au nombre de personne/jours (heures) ; ta pe 1: Collecte des produits de les banques forrageres Personnes impliques (H/F/E) No. de P/J (H) An 3:_____ No. de P/J (H) An 4: _____ No. de P/J (H) An 5: _____ No. de P/J (H) An 6: _____ No. de P/J (H) An 7: _____ No. de P/J (H) An 8: _____ G. sepium : Note : (An. 9, 10, si possible) tape 2: Transformation des produits de les banques forrageres Personnes impliques (H/F/E) No. de P/J An 3:_____ No. de P/J An 4: _____ No. de P/J An 5: _____ No. de P/J An 6: _____ No. de P/J An 7: ____ No. de P/J An 8: _____ G. Sepium : Note :

PAGE 161

161 (An. 9, 10, si possible) tape 3: Vente des produits de les banques forrageres Personnes impliques (H/F/E) No. de P/J An 3:_____ No. de P/J An 4: _____ No. de P/J An 5: _____ No. de P/J An 6: _____ No. de P/J An 7: _____ No. de P/J An 8: _____ G. Sepium : Note : 1.12 banques forrageres (et la haie vive, si quelquechose) ? Nom d u march Distance (entrer codes) 1: 0 5 km 2: 5 10 km 3: 10 20 km 4: 20 50 km 5: >50 km 9: Pas de rponse 1.13 Quels sont les moyens de transport utiliss pour vendre les produits des banques forrageres au march? Note: Noter tous les moyens de t ransport utiliss pour la vente des produits de les banques forrageres, PAS les moyens de transport uniquement utiliss pour le travail champtre agricole. Moyen de transport (entrer codes): Nombre Prix (en CFA) Montant total Nombre n 1: Charrette 2: Bicyclette/mobylette 3: Vhicule 4: Autres 9: Pas de rponse

PAGE 162

162 1. 14 Personnes impliqus dans le activit pour le rasembler forrage ligneux de champs ou brousse: Notes: Expliquer comment vous tes arriv au nombre de personne/jours; Personnes impliques (H/F/E): No. de personne/ jours (heures) An. 2005 Note : Si possible, avant commencer les banques forrageres Personnes implique s (H/F/E): No. de personne/ jours (heures) An. Note:

PAGE 163

163 APPENDIX B COST BENEFIT ANALYSIS (CASH FLOW) OF LIVE FENCE

PAGE 164

164

PAGE 165

165

PAGE 166

166

PAGE 167

167 APPENDIX C COST BENEFIT ANALYSIS (CASH FLOW) OF FODDER BANK

PAGE 168

168

PAGE 169

169

PAGE 170

170

PAGE 171

171 LIST OF REFERENCES Alavalapati, J.R.R., R.K. Shrestha, G.A. Stainback, and J.R. Matta. 2004. Agroforestry development: An environmental economic perspective. Agrofor. Syst. 61 62: 299 310 Albrecht, A. and S.T. Kandji. 2003. Carbon sequestration in tropical agroforestry systems. Agric. Ecosyst. Envir on. 99:15 27. Alvarez R., R.A. Diaz, N. Barbero, O.J. Santanatoglia, and L. Blotta. 1995. Soil organic carbon, microbial biomass and CO2 C production from three tillage systems. Soil Tillage Res. 33: 17 28. Alvarez, R. and C. R. Alvarez. 2000. Soil organic matter pools and their associations with carbon mineralization kinetics. Soil Sci Soc Am J 64:184 189. Balesdent, J. and A. Mariotti. 1996. Measurement of soil organic matter turnover using 13C natural abundance. p.83 112. In T.W. Boutton and S.I. Yam asaki (ed.) Mass spectrometry of soils. Marcel Dekker, New York. Balesdent, J., E. Besnard, D. Arrouays, and C. Chenu. 1998. The dynamics of carbon in particle size fractions of soil in a forest cultivation sequence. Plant Soil 201:49 57. Baron, R. and A. Lanza. 2000. Kyoto commitments: macro and micro insights on trading and the Clean Development Mechanism. Integrated Assessment 1:137 144. Bationo, A. and B.R. Ntare. 2000. Rotation and nitrogen fertilizer effects on pearl millet, cowpea and groundnut yield and soil chemical properties in a sandy soil in the semi arid tropics, West Africa. J. Agric. Sci. (Cambridge) 134:277 284. Batjes, N.H. 2004. Estimation of soil carbon gains upon improved management within croplands and grasslands of Africa. Environ. Dev Sust. 6:133 143. Batjes, N.H. 2001. Options for increasing carbon sequestration in West African soils: An exploratory study with special focus on Senegal. Land Deg. Dev. 12:131 142. Beare, M. H. M. L. Cabrera, P. F. Hendrix, and D. C. Coleman. 1994. Ag gregate protected and unprotected organic matter pools in conventional and no tillage soils. Soil Sci Soc Am J. 58:787 795. Beer, J., A. Bonnemann, W. Chavez, H.W. Fassbender, A.C. Imbach, and I. Martel. 1990. Modelling agroforestry systems of cacao ( T heobroma cacao ) with laurel ( Cordia alliodora ) or poro ( Erythrina poeppigiana ) in Costa Rica. Agrofor. Syst. 12:229 249. Blair, G.J., R.D.B. Lefroy, and L. Lisle. 1995. Soil carbon fractions based on their degree of oxidation, and the development of a carb on management index for agricultural systems. Aust. J. Agric. Res. 46:1459 1466.

PAGE 172

172 Boffa, J. M. Agroforestry parklands in sub Saharan Africa 1999. FAO Conservation Guides 34. FAO, Rome, Italy. Bouliere, F. 1983. Tropical Savannas. Elsevier Scientific Publi shing Co. New York, NY. Breman, H. and J.J. Kessler. 1997. The potential benefits of agroforestry in the Sahel and other semi arid regions. Euro. J. Agron. 7:25 33. Brown, S. 1997. Estimating Biomass and Biomass Change of Tropical Forests: a Primer. F AO, R ome, Italy. Brown, S. 1999. Guidelines for Inventorying and Monitoring Carbon Offsets in Forest Based Projects Winrock International Arlington, VA Brown, S. 2002. Measuring carbon in forests: current status and future challenges. Environ. Pollut. 116: 363 372. Buresh, R.J. and G. Tian. 1997. Soil improvement by trees in sub Saharan Africa. Agrofor. Syst. 38:51 76. Cacho, O.J., G.R. Marshall, and M. Milne. 2003a. Smallhoder agroforestry projects: potential for carbon sequestration and poverty alleviation ESA Working Paper No.03 06 FAO. www.fao.org/es/esa Cacho, O.J., R.L. Hean, and R.M. Wise. 2003b. Carbon accounting methods and reforestation incentives. Aust. J. Agric. Res. Econ. 47:153 179. Cairns, M.A., S. Brown, E.H. Helmer, and G.A. Baumgardner. 19 97. Root biomass allocation in the world's upland forests. Oecologia (Berlin) 111:1 11. Campbell C.A., B.G. McConkey, R.P. Zentner, F. Selles, and D. Curtin. 1996. Tillage and crop rotation effects on soil organic C and N in a course textured Typic Haplobor oll in southwestern Saskatchewan. Soil Tillage Res. 37: 3 14. Campbell, H.A. and P.C. Brown. (ed.) 2003. Benefit cost analysis: Financial and economic appraisal using spreadsheets. Cambridge University Press, Port Melbourne, Australia. Choudhary, M.A., A. Akramkhanov, and S. Saggar. 2002. Nitrous oxide emissions from a New Zealand cropped soil: tillage effects, spatial and seasonal variability. Agric. Ecosys t Environ. 93:33 43. CIA 2007 The World Factbook: Mali. [Online] Available at https://www.cia.gov/library/publications/the world factbook/geos/ml.html (verified 5 Jul. 2007). CIA, Washington, DC. Cisse, M.I. 1980. Production fourragre de quelques arbres saheliens: relations entre la biomasse foliaire maximale et divers parametresphysiques. p.203 208. In H.N. Le Houerou. (ed.) Les fourrages ligneux en Afrique, L etat actuel des connaissances. C iPEAA, Addis Abeba.

PAGE 173

173 Ciss, M.I. Les parcs agroforestiers au Mali. 1995. E tat des connaissances et perspectives pour leur amlioration Rapport AFRENA. ICRAF, Nairobi, Kenya. Coughenour, M.B., J.E. Ellis, and R.G. Popp. 1990. Morphometric relationships and developmental patterns of Acacia tortilis and Acacia reficiens in Southe rn Turkana, Kenya. Bull. Torrey Bot Club 117:8 17. Dai, A., P.J. Lamb, K.E. Trenberth, M. Hulme, P.D. Jones, and P. Xie. 2004. The recent Sahel drought is real. Int. J. Climat. 24:1323 1331. Dalsted, N.L. and P.H. Gutierrez. 2007. Partial budgeting. [Onli ne]. Available at http://www.ext.colostate.edu/PUBS/farmmgt/03760.html (verified 5 Jul. 2007). Colorado State University Extension Agriculture, Fort Collins, CO. D e Alwis, K.A. Recapital ization of soil productivity in sub Saharan Africa. 1996. FAO Investment Center, Rome, Italy. De Jong, B.H., S.O. Gaona, S.Q. Montalvo, E.E. Bazan, and N.P. Hernandez. 2004. Economics of agroforestry carbon sequestration: A case study from southern Mexico Chap.4. In J.R.R. Alavalapati and D.E. Mercer (ed.) Valuing agroforestry systems. Kluwer Academic Publishers, Netherlands. De Jong, B.H.J. 2001. Uncertainties in estimating the potential for carbon mitigation of forest management. For. Ecol. Manage. 154: 85 104. Del Galdo, I., J. Six, A. Peressotti, and M. F. Cotrufo. 2003. Assessing the impact of land use change on soil C sequestration in agricultural soils by means of organic matter fractionation and stable C isotopes. Glob. Chan. Bio. 9:1204 1213. Delane y, M. and J. Roshetko. 1999. Field test of carbon monitoring methods for home gardens in Indonesia p.45 51. In Field Tests of Carbon Monitoring Methods in Forestry Projects. Forest Carbon Monitoring Program. Winrock International, Arlington, VA. Denton, F ., Y. Sokona, and J.P. Thomas. Climate change and sustainable development strategies in the making: What should West African countries expect? 2001. OECD Climate Change and Development. Environnement et Dveloppement du Tiers Monde (ENDA TM) Dakar, Senega l. Dixon, R.K. 1995. Agroforestry system: sources or sinks of greenhouse gases? Agrofor. Syst. 31:99 116. Dixon, R.K., S Brown, R.A. Houghton, M.C. Solomon, M.C. Trexler, and J. Wisniewski. 1994 a Carbon pools and flux of global forest ecosystems. Science ( Washington, DC) 263:185 190. Dixon, R.K., J.K. Winjum, K.J. Andrasko, J.J. Lee, and P.E. Schroeder. 1994b. Integrated land use systems: assessment of promising agroforest and alternative land use practices to enhance carbon conservation and sequestration. Clim. Change 27:71 92.

PAGE 174

174 SEGOU 2000. Sgou, Mali Droppelmann, K.J., J. Lehmann, J.E. Ephrath, and P.R. Berliner. 2000. Water use efficiency and uptake patterns in a runoff agroforestry system in an arid environment. Agrofor. Syst. 49:223 243. Duguma, B., J. Gockowski, and J. Bakala. 2001. Smallholder cacao ( Theobroma cacao Linn .) cultivation in agroforestry systems of West and Central Africa: challenges and opportuni ties. Agrofor. Syst. 51:177 188. Duiker S.W. and R. Lal. 1999. Crop residue and tillage effects on carbon sequestration in a Luvisol in central Ohio. Soil Tillage Res. 52, 73 81. Dzurec, R.S., T.W. Boutton, M.M. Caldwell, and B.N. Smith. 1985. Carbon isot ope ratios of soil organic matter and their use in assessing community composition changes in Curlew Valley, Utah. Oecologia (Berlin) 66:17 24. Eleki, K., R.M. Cruse, and K.A. Albrecht. 2005. Root segregation of C3 and C4 species using carbon isotope compo sition. Crop Sci 45:879 882. Eswaran, H., R. Almaraz, E. Berg, and P. Reich. 1996. An Assessment of the Soil Resources of Africa in Relation to Productivity. [ Online] Available at htt p://soils.usda.gov/use/worldsoils/papers/africa1.html (verified 5 Jul. 2007). N RCS, Washington, DC. FAO 2000. Global Forest Resources Assessment 2000. FAO Forestry Paper 140. Rome, Italy. FAO 1997. Estimating biomass and biomass change of tropical fore sts. FAO Forestry Paper 134. Rome, Italy. FAO. 1991. Feeding dairy cows in the tropics. FAO Animal Production and Health Paper 86. Rome, Italy. Ferguson, W. 1 983. Integrating crops and livestock in West Africa. FAO Animal Production and Health Paper 41 R ome, Italy. Franklin, J. and P. Hiernaux. 1991. Estimating foliage and woody biomass in Sahelian and Sudanian woodlands using a remote sensing model. Int. J. Remote Sens.12:1387 1404. Franzluebbers, A.J., F.M. Hons, and D.A. Zuberer. 1995. Tillage induce d seasonal changes in soil physical properties affecting soil CO2 evolution under intensive cropping. Soil Tillage Res. 34: 41 60. Garcia Oliva, F. and O.R. Masera. 2004. Assessment and measurement issues related to soil carbon sequestration in Land Use, L and Use Change, and Forestry (LULUCF) projects under the Kyoto Protocol. Clim. Chan ge 65:347 364.

PAGE 175

175 Gebhart, D.L., H.B. Johnson, H.S. Mayeux, and H.W. Polley. 1994. The CRP increases soil organic carbon. 49: 488 492. J. Soil Water Conserv. Ghani, A., M. Dext er, and K.W. Perrott. 2003. Hot water extractable carbon in soils: a sensitive measurement for determining impacts of fertilisation, grazing and cultivation. Soil Biol. Biochem. 35:1231 1243. Godal, O., Y. Ermoliev, G. Klaassen, and M. Obersteiner. 2003. C arbon trading with imperfectly observable emissions. Environ. Res. Econ. 25:151 169. Gonzalez, P. 2001. Desertification and a shift of forest species in the West African Sahel Clim. Res. 17:217 228. Gordon, J.E., W.D. Hawthorne, G. Sandoval, and A.J. Barr ance. 2003. Trees and farming in the dry zone of southern Honduras II: the potential for tree diversity conservation. Agrofor. Syst. 59:107 117. Gritzner, J.A. 1988. The West African Sahel. University of Chicago, Committee on Geographical Studies Chicago, IL. Hamer, A., S. Franzel, B. Mounkoro, A. Niang, and C.O. Traore. 2005. Fodder banks in Mali. ICRAF, Bamako, Mali. Hardner, J.J., P.C. Frumhoff, and D.C. Goetze. 2000. Prospects for mitigating carbon, conserving biodiversity, and promoting socioeconomi c development objectives through the clean development mechanism. Mitig. Adapt. Strat. Glob. Change 5:61 80. Harris, D., R.H. William, and C. van Kessel. 2001. Acid fumigation of soils to remove carbonates prior to total organic carbon or CARBON 13 isotop ic analysis. Soil Sci Soc Am J 65:1853 1856. Hassink, J. 1997. The capacity of soils to preserve organic C and N by their association with clay and silt particles. Plant Soil 191:77 87. ICRAF. 2007. ICRAF Position on the Clean Development Mechanism in the Land Use, Land Use Change and Forestry Sector [Online] Available at http://www.worldagroforestry.org/climatechange/documents/ICRAFPositiononCDM.pdf (verif ied 5 Jul. 2007 ) Nairobi, Kenya. ICRISAT Crops. 2007 [Online] Available at http://www.icrisat.org/ (verified 5 Jul. 2007). Ingram, J.S.I. and E.C.M. Fernandes. 2001. Managing carbon sequestration in soils: conce pts and terminology. Agric. Ecosys. Environ. 87:111 117. IPCC 2007. Climate Change 2007: Mitigation of Climate Change. Working Group III contribution to the Intergovernmental Panel on Climate Change Fourth Assessment Report Bangkok, Thailand.

PAGE 176

176 IPCC 200 0. Land use, Land use Change, and Forestry. A Special Report of the IPCC Cambridge University Press Cambridge, UK. Jeltsch, F., S.J. Milton, W.R.J. Dean, and N. van Rooyen. 1996. Tree spacing and coexistence in semiarid savannas. J. Ecol. 84:583 595. Job bagy, E.G. and R.B. Jackson. 2000. The vertical distribution of soil organic carbon and its relation to climate and vegetation. Ecol. Applic. 10:423 436. Johnsen, K.H., D. Wear, R. Oren, R.O. Teskey, F. Sanchez, R. Will, J. Butnor, D. Markewitz, D. Richter T. Rials, H.L. Allen, J. Seiler, D. Ellsworth, C. Maier, G. Katul, and P.M. Dougherty. 2001. Meeting global policy commitments: Carbon sequestration and southern pine forests. J. For. 99:14 21. Jones, M., M.L. Sinclair, and V.L. Grime. 1998. Effect of tr ee species and crown pruning on root length and soil water content in semi arid agroforestry. Plant Soil. 201:197 207. Jonsson, K., C.K. Ong, and J.C.W. Odongo. 1999. Influence of scattered Nere and Karite trees on microclimate, soil fertility and millet y ield in Burkina Faso. Exp. Agric. 35:39 53. Kang, B.T., F.E. Caveness, G. Tian, and G.O. Kolawole. 1999. Longterm alley cropping with four hedgerow species on an Alfisol in southwestern Nigeria effect on crop performance, soil chemical properties and nem atode population. Nutr. Cycl. Agroecosyst. 54:145 155. Kater, L.J.M., S. Kante, and A Budelman. 1992. Karite ( Vitellaria paradoxa ) and nere( Parkia biglobosa ) associated with crops in South Mali. Agrofor. Syst. 18:89 105. Kaya, B. 2000. Soil fertility regen eration through improved fallow systems in southern Mali. Ph.D. diss. Univ. of Florida Gainesville, FL Kaya, B. and P.K.R. Nair. 2004. Dynamics of Particulate Organic Matter following biomass addition from fallow improvement species in southern Mali. Agro for. Syst. 60:267 276. Kaya, B. and P.K.R. Nair. 2001. Soil fertility and crop yields under improved fallow systems in southern Mali. Agrofor. Syst. 52:1 11. Klemperer, W.D. (ed.) 1996. Forest resource economics and finance. McGraw Hill, Inc USA Kursten, E. and P. Burschel. 1993. CO2 mitigation by agroforestry. Water Air Soil Pollut. 70:533 544. Lal, R. 2004a. Carbon sequestration in dryland ecosystems. Environ. Manag. 33:528 544 Lal, R. 2004b. Soil carbon sequestration impacts on global climate change and food security. Science (Washington, DC) 304:1623 1627.

PAGE 177

177 Lal, R. 1999. Global carbon pools and fluxes and the impact of agricultural intensification and judicious land use. p. 44 52. In Prevention of Land Degradation, Enhancement of Carbon Sequestration and Conservation of Biodiversity through Land Use Change and Sustainable Land Management with a Focus on Latin America and the Caribbean. FAO, Rome, Italy. Levasseur, V., M. Djimde, and A. Olivier. 2004. Live fences in Sgou, Mali: an evaluation by their early users. Agrofor. Syst. 60:131 136. Littmann, T. 1991. Rainfall, temperature, and dust storm anomalities in the African Sahel. Geogr. J 157:136 160. MacDicken, K.G. (ed.) 1997. A guide to monitoring carbon storage in forestry and agroforestry projects. Winrock International Institute for Agricultural Development, Forest Carbon Monitoring Program. Arlington, VA. Makumba, W., B. Janssen, O. Oenema, F.K. Akinnifesi, D. Mweta, and F. Kwesiga. 2006. The long term effects of a gliricidia maize intercropping s ystem in Southern Malawi, on gliricidia and maize yields, and soil properties: Nutrient management in tropical agroecosystems. Agric. Ecosyst. Environ. 116:85 92. Makundi, W.R. and J.A. Sathaye. 2004. GHG Mitigation potential and cost in tropical forestry relative role for agroforestry. Environ. Dev. Sust. 6:235 260. Manlay, R.J., J.L. Chotte, D. Masse, J.Y. Laurent, and C. Feller. 2002. Carbon, nitrogen and phosphorus allocation in agro ecosystems of a West African savanna: III. Plant and soil components under continuous cultivation. Agric. Ecosyst. Environ. 88:249 269. Marino, B.D. and M.B. McElroy. 1991. Isotopic composition of atmospheric CO2 inferred from carbon in C4 plant cellulose. Nature 349:127. Masera, O.R., J.F. Garza Caligaris, M. Kanninen, T. Karjalainen, J. Liski, G.J. Nabuurs, A. Pussinen, B.H.J. de Jong, and G.M.J. Mohren. 2003. Modeling carbon sequestration in afforestation, agroforestry and forest management projects: the CO2FIX V.2 approach. Ecol. Modell 164:177 199. McDonagh, J.F., T.B .Thomsen, and J. Magid. 2001. Soil organic matter decline and compositional change associated with cereal cropping in southern Tanzania. Land Deg. Dev. 12:13 26. McLauchlan, K.K. and S.E. Hobbie. 2004. Comparison of labile soil organic matter fractionation techniques. Soil Sci Soc Am J 68:1616 1625. Mitchell, T. 2005. Sahel rainfall index (20 10N, 20W 10E), 1898 2004. [Online] Avaliable at http://jisao.washington.edu/data_s ets/sahel/index2.html#values (verified 5 Jul. 2007) University of Washington, Seattle, WA. Montagnini, F. and P.K.R. Nair. 2004. Carbon sequestration: An underexploited environmental benefit of agroforestry systems. Agrofor. Syst. 61 62:281 295.

PAGE 178

178 Moody, P.W., S.A. Yo, and R.L. Aitken. 1997. Soil organic carbon, permanganate fractions, and the chemical properties of acidic soils. Aust. J. Soil Res. 35:1301 1308. Mosier, A., R. Wassmann, L. Verchot, J. King, and C. Palm. 2004. Methane and nitrogen oxide flu xes in tropical agricultural soils: Sources, sinks and mechanisms. Environ. Dev. Sust. 6:11 49. Moura Costa, P. and C. Wilson. 2000. An equivalence factor between CO2 avoidedemissions and sequestration description and applications in forestry. Mitig. Ada pt. Strat. Glob. Change 5:51 60. Nair, P. K. R. and V.D. Nair. 2003. Carbon storage in North American agroforestry systems. In: Kimble, J., Heath, L.S., Birdsey, R.A., and Lal, R. (eds). The Potential of U.S. Forest Soils to Sequester Carbon and Mitigate t he Greenhouse Effect, pp. 333 346. CRC Press LLC, Boca Raton, FL. Nair, P.K.R. (ed.) 1993. An Introduction to Agroforestry. Kluwer Academic Publishers, Dordrecht, Netherlands Niang, A., M. Djimde, B. Kaya, E.G. Bonkoungou, and M. Macalou. 2002. Paper pre sented at Improving the quantity and quality of dry season fodder av a ilability in the Sahel. Presented at the Regional Agroforestry Conference. Pretoria, South Africa. May 20 24, 2002. Nierop, K.G.J., F.H. Tonneijck, B. Jansen, and J.M. Verstraten. 2007. Organic matter in volcanic ash soils under forest and paramo along an Ecuadorian altitudinal transect. Soil Sci Soc Am J 71:1119 1127. Oba, N., N.C. Stenseth, and W.J. Lusigi. 2000. New perspectives on sustainable grazing management in arid zones of su b Saharan Africa. Bioscience 50:35 51. Onim, J.F.M., M. Mathuva, K. Otieno, and H.A. Fitzhugh. 1990. Soil fertility changes and response of maize and beans to green manures of leucaena, sesbania and pigeonpea. Agrofor. Syst. 12:197 215. Oren, R., D.S. Ells worth, K.H. Johnsen, N. Phillips, B.E. Ewers, C. Maier, K.V.R Schafer, H. McCarthy, G. Hendrey, S.G. McNulty, and G.G Katul. 2001. Soil fertility limits carbon sequestration by forest ecosystems in a CO2 enriched atmosphere. Nature (London) 411:469 472. Pa lm, C. A T. Tomich, M. Van Noordwijk, S. Vosti, J. Alegre, J. Gockowski, and L. Verchot. 2004. Mitigating GHG emissions in the humid tropics: Case studies from the Alternatives to Slash and Burn Program (ASB). Environ. Dev. Sust. 6:145 162.

PAGE 179

179 Palm, C.A., P.L Woomer, J. Alegre, L. Arevalo, C. Castilla, D.G. Cordeiro, B. Feigl, K. Hairiah, J. Kotto Same, A. Mendes, A. Moukam, D. Murdiyarso, R. Njomgang, W.J. Parton, A. Ricse, V. Rodrigues, S.M. Sitompul, and M van Noordwijk. 1999. Carbon sequestration and trac e gas emissions in slash and burn and alternative land use in the humid tropics. ASB Climate Change Working Group Final Report, Phase II. ASB Coordination Office, ICRAF Nairobi, Kenya. Paustian, K., O. Andren, H.H. Janzen, R. Lal, P. Smith, G. Tian, H. T iessen, M. van Noordwijk, and P. L. Woomer. 1997. Agricultural soils as a sink to mitigate CO2 emissions. Soil Use and Manage 13: 230 244. Pfaff, A.S.P., S. Kerr, R.F. Hughes, S. Liu, G.A. Sanchez Azofeifa, D. Schimel, J. Tosi, and V. Watson. 2000. The Ky oto protocol and payments for tropical forest: An interdisciplinary method for estimating carbon offset supply and increasing the feasibility of a carbon market under the CDM. Ecol. Econ. 35:203 221. Phillips, D L ., P.D. Hardin, V. W. Benson, J.V. Baglio. 1 993. Nonpoint source pollution impacts of alternative agricultural management practices in Illinois: A simulation study. J. Soil Water Consev. 48: 449 457. Pikul, J.L., S. Osborne, M. Ellsbury, and W. Riedell. 2007. Particulate organic matter and water st able aggregation of soil under contrasting management. Soil Sci Soc Am J 71:766 776. Post, W. M. and K.C. Kwon. 2000. Soil carbon sequestration and land use change: processes and potential. Glob. Change Biol. 6:317 327. Potvin, C., E. Whidden, and T. M oore. 2004. A case study of carbon pools under three different land uses in Panama. Clim. Change 67:291 307. Powers, J.S. and E. Veldkamp. 2005. Regional variation in soil carbon and 13C in forests and pastures of northeastern Costa Rica. Biogeochem. 72:31 5 336. Powers, J.S. and W.H. Schlesinger. 2002. Relationships among soil carbon distributions and biophysical factors at nested spatial scales in rain forests of northeastern Costa Rica. Geoderma 109:165 190. ProKarite. Projet d Appui Technique la Filli re Karit 2007. [Online] Available at http://www.prokarite.org/index eng.html (verified 5 Jul.2007). Bamako, Mali. Reich, P.F., S.T. Numbrem, R.A. Almaraz, and H. Eswaran. 2001. Land resource stress es and desertification in Africa. In E.M. Bridges, I.D. Hannam, L.R. Oldeman, F.W.T. Pening de Vries, S.J. Scherr, and S. Sompatpanit (eds.) Responses to Land Degradation. Proc. 2nd. International Conference on Land Degradation and Desertification, Khon K aen, Thailand. Oxford Press, New Delhi, India. Rpublique du Mali. 2005. Minitere du Plan et de L amenagement du Territoire and Ministere de L agriculture de L elevage et de la Peche. Enquete Agricole de Conjoncture Campagne 2003/2004. Bamako, Mali.

PAGE 180

180 Rhoto n, F.E. 2000. Influence of time on soil response to no till practices. Soil Sci Soc Am J 64: 700 709. Richter, D.D., D. Markewitz, S.E. Trumbore, and C.G. Wells. 1999. Rapid accumulation and turnover of soil carbon in a re establishing forest. Nature ( London) 400:56 58. Ringius, L. 2002. Soil carbon sequestration and the CDM: Opportunities and challenges for Africa. Clim. Change 54:471 495. Roose, E., V. Kabore, and C. Guenat. 1999. Zai Practice: A w est African traditional rehabilitation system for semi arid degraded lands, a case study in Burkina Faso Arid Soil Res. Rehabil. 13: 343 355. Rosalina, U., Setiabudhi, and A.E. Putra. Vegetation analysis and database management system in Lampung and Jambi. 1997. ICRAF Southeast Asia Regional Office, Bogor, Indonesia. Rose, S., H. Ahammad, B. Eickhout, B. Fisher, A. Kurosawa, S. Rao, K. Riahi, and D. van Vuuren. 2007. Land in climate stabilization modeling: Initial observations. [Online] Available at http://www.stanford.edu/group/EMF/projects/group21/Landuse.pdf (verified 5 Jul. 2007). Stanford Energy Modeling Forum (EMF), Stanford University, Stanford, CA. Roshetko, J.M., M. Delaney, K. Hairiah, and P Purnomosidhi. 2002. Carbon stocks in Indonesian homegarden systems: Can smallholder systems be targeted for increased carbon storage? Am. J. Alt. Agric. 17:138 148. Sathaye, J.A., W.R. Makundi, K. Andrasko, R. Boer, N.H. Ravindranath, P. Sudha, S. Rao, R. Lasco, F. Pulhin, O Masera, A. Ceron, J. Ordonez, X. Deying, X. Zhang, and S. Zuomin. 2001. Carbon mitigation potential and costs of forestry options in Brazil, China, India, Indonesia, Mexico, the Philippines and Tanzania. Mitig. Adapt. Strat. Glob. Change 6:185 211. Schi mel, D.S., B.H. Braswell, E.A. Holland, R. McKeown, D.S. Ojima, T.H. Painter, W.J. Parton, and A.R. Townsend. 1994. Climatic, edaphic, and biotic controls over storage and turnover of carbon in soils. Glob. Biogeochem. Cycl. 8:279 293. Schroeder, P. 1994. Carbon storage benefits of agroforestry systems. Agrofor Syst. 27:89 97. Schroth, G., S.A. D'Angelo, W.G. Teixeira, D. Haag, and R. Lieberei. 2002. Conversion of secondary forest into agroforestry and monoculture plantations in Amazonia: consequences for biomass, litter and soil carbon stocks after 7 years. For. Ecol. Manag. 163:131 150. Schuur, E.A.G. 2001. The effect of water on decomposition dynamics in mesic to wet Hawaiian montane forests. Ecosystems 4:259 273. Scolel Te. 2007. [Online] Available at http://www.eccm.uk.com/scolelte/ (verified 5 Jul. 2007). Edinburgh, UK.

PAGE 181

181 Seghieri, J., M. Simier, A. Mahamane, P. Hiernaux, and S. Rambal. 2005. Adaptative above ground biomass, stand density and leaf water potential to droughts and clearing in Guiera senegalensis a dominant shrub in Sahelian fallows (Niger). J. Trop. Ecol. 21:203 213. Sharrow, S.H. and S. Ismail. 2004. Carbon and nitrogen storage in agroforests, tree plantations, and pastures in western Ore gon, USA. Agrofor. Syst. 60:123 130. Shepherd, D. and F. Montagnini. 2001. Above ground carbon sequestration potential in mixed and pure tree plantation in the humid tropics. J Trop. For. Sci. 13:450 459. Six, J., R.T. Conant, E.A. Paul, and K. Paustian. 2002a. Stabilization mechanisms of soil organic matter: Implications for C saturation of soils. Plant Soil 241:155 176. Six, J., P. Callewaert, S. Lenders, S. De Gryze, S.J. Morris, E.G. Gregorich, E.A. Paul, and K. Paustian. 2002b. Measuring and understan ding carbon storage in afforested soils by physical fractionation. Soil Sci Soc Am J 66:1981 1987. Six, J., R. Merckx, K. Kimpe, K. Paustian, and E.T. Elliott. 2000c. A re evaluation of the enriched labile soil organic matter fraction. Eur. J. Soil Sci 51: 283 293. Six, J., E.T. Elliott, K. Paustian, and J.W. Doran. 1998. Aggregation and soil organic matter accumulation in cultivated and native grassland soils. Soil Sci Soc Am J 62:1367 1377. Sm ith P., D.S. P owlson M J. G lending and J.U. S mith 1998. Preliminary estimates of the potential for carbon mitigation in European soils through no till farming. Glob. Change Bio. 4: 679 685. Stainback, A.G. and J.R.R. Alavalapati. 2002. Economic analysis of slash pine forest carbon sequestration in the so uthern U.S. J. For. Econ. 8:105 117. Sun, O.J., J. Campbell, B.E. Law, and V. Wolf. 2004. Dynamics of carbon stocks in soils and detritus across chronosequences of different forest types in the Pacific Northwest, USA. Glob. Chan. Bio l .10:1470 1481. Tan, Z. X., R. Lal, N.E. Smeck, and F.G. Calhoun. 2004. Relationships between surface soil organic carbon pool and site variables. Geoderma 121: 187 195. Tiepolo, G., M. Calmon, and A.R. Feretti. 2002. Paper presented at measuring and monitoring carbon stocks at t he Guaraquecaba Climate Action Project, Parana, Brazil Proceedings of the International Symposium on Forest Carbon Sequestration and Monitoring. 11 15 November 2002, Taipei, Taiwan. Taiwan Forestry Research Institute Winrock International. Tiessen, H., H. K. Hauffe, and A.R. Mermut. 1991. Deposition of Harmattan dust and its influence on base saturation of soils in northern Ghana. Geoderma 49:285 299. Tieszen, L.L. 1991. Natural variations in the carbon isotope values of plants: implicaations for archaeolog y, ecology and paleoecology. J. Archaeol. Sci. 18: 227.

PAGE 182

182 Tipper, R. and B.H. De Jong. 1998. Quantification and regulation of carbon offsets from forestry: comparison of alternative methodologies, with special reference to Chiapas, Mexico. Commonwealth For. Rev. 77:219 228. Tomich, T.P., H. de Foresta, R. Dennis, Q. Ketterings D. Murdiyarso, C. Palm, F. Stolle, Suyanto, and M. van Noordwijk. 2002. Carbon offsets for conservation and development in Indonesia? Am. J. Alt. Agric. 17:125 137. Traore, C.O., F. P lace, and A. Niang. 2002. Evaluation economique des couts de production des plants d'especes haies vives a la pepiniere de Sgou. Mali. ICRAF, Bamako, Mali. Tschakert, P. 2004. Carbon for farmers: Assessing the potential for soil carbon sequestration in th e Old Peanut Basin of Senegal. Clim. Change 67:273 290. Tschakert, P. 2007. Environmental services and poverty reduction: Options for smallholders in the Sahel: Making carbon sequestration work for Africa's rural poor Opportunities and constraints. Agric Syst. 94:75 86. UNFCCC. 2006. Revised simplified baseline and monitoring methodologies for selected small scale afforestation and reforestation project activities under the clean development mechanism. [Online] Available at http://cdm.unfccc.int/UserManagement/FileStorage/CDMWF_AM_A3II6AX6KGW5GBB 7M6AI98UD3W59X4 (verified 5 Jul.2007). Bonn, Germany. UNFCCC 2007. [Online] Available at http://unfccc.int/2860.php (verified 5 Jul.2007). Bonn, Germany. USGS. 2007. International Program Sahel Land Use. [Online] Available at http://edcintl.cr.usgs.gov/sahel.html (verified 5 Jul. 2007). Reston, VA. V an Dorp, M., B. Mounkoro, S. Soumana, C.O. Traore, S. Franzel, F. Place, and A. Niang. 2005. Economic analysis of improved live fences as an agroforestry technology as compared to traditional live fences and dead fen ces in the Sgou Region, Mali. ICRAF, Bamako, Mali. V an Duijl, E. 2000. Feeding livestock in the dry season: towards identifying potential adopters for fodderbanks in Sgou, Mali. Draft Report. ICRAF, Bamako, Mali. Van Duijl, E.C. 1999. Characterisation of potential adopters for live fences in Sgou, Mali. ICRAF, Bamako, Mali. Van Noordwijk, M., S. Rahayu, K. Hairiah, Y.C. Wulan, A. Farida, and B. Verbist. 2002. Carbon stock assessment for a forest to coffee conversion landscape in Sumber Jaya (Lampung, I ndonesia): from allometric equations to land use change analysis. Science China 45:75 86. Walker, S.M. and P.V. Desanker. 2004. The impact of land use on soil carbon in Miombo Woodlands of Malawi. For. Ecol. Manag. 203:345 360.

PAGE 183

183 Wise, R. and O. Cacho. 2005. Tree crop interactions and their environmental and economic implications in the presence of carbon sequestration payments. Environ Modell. Software 20:1139 1148. Woomer, P.L., L.L. Tieszen, G. Tappan, A. Toure, and M. Sall. 2004a. Land use change and ter restrial carbon stocks in Senegal. J. Arid Environ. 59:625 642. Woomer, P.L., A. Toure, and M. Sall. 2004b. Carbon stocks in Senegal's Sahel transition zone. J. Arid Environ. 59:499 510. World Bank. 2007. Mali Country Brief. [Online] Available at www.worldbank.org/mali (verified 5 Jul. 2007). Washington, DC. World Bank. 2006. State and trends of the carbon market 2006. W orld Bank, Washington DC. Yossi, H., B. Kaya, C.O. Traore, A. Niang, I. Butare, V. Levass eur, and D. Sanogo. 2005. Les haies vives au Sahel: Etat des connaissances et reccommandations pour la recherche et le developpement. ICRAF Programme Regional Sahel, Bamako, Mali. Zibilske, L.M. and J.M. Bradford. 2007. Soil aggregation, aggregate carbon and nitrogen, and moisture retention induced by conservation tillage. Soil Sci Soc Am J 71:793 802. Zinn, Y L., R. Lal, J.M. Bigham, and D.V.S. Resck. 2007. Edaphic controls on soil organic carbon retention in the Brazilian Cerrado: Texture and mineral ogy. Soil Sci Soc Am J 71: 1204 1214.

PAGE 184

184 BIOGRAPHICAL SKETCH Asako Takimoto was born in Ashiya, Hyogo, Japan in 1975. She graduated from the Kyoto University in Kyoto, Japan, in 1998 where she earned her B.Sc. in Agricultural Science and Forestry. In 1997, she received the Rotary International Scholarship and entered the graduate program of the Nicholas School of the Environment, Duke University, where she finished her Master of Forestry. After graduation in 1999, she returned to Japan and worked for the Japan International Cooperation Agency. In 2003, she received the Fulbright Scholarship to start her PhD study in agroforestry at University of Florida.