Changes in carbon stocks and tree diversity in agro-ecosystems in south western Uganda:
what role for carbon sequestration payments?
C. Nakakaawa•J. Aune•P. Vedeld
Received: 6 March 2009 / Accepted: 30 October 2009 ÓSpringer Science+Business Media B.V. 2009
Abstract Direct payments offered as compensation for adopting practices which increase carbon sequestration have implications for biodiversity conservation. This study analyzed changes in carbon stocks and tree diversity on carbon and non-carbon farmers’ plots in a pilot carbon offset project implemented by smallholder farmers in south western Uganda.
On carbon farms, results indicate a respective decline and an increase in carbon density on farmlands and grasslands. On non-carbon farmers’ plots, there was an increase in both farmland and grassland carbon density. Carbon densities in farmland were significantly (t= -2.38;P=0.023) higher than those in grasslands. There were no significant differ- ences in tree diversity on farmlands but significant differences (species richnesst=2.18;
P=0.04; Shannon Index t=2.92;P value=0.0077) in grasslands. Tree diversity on carbon farmers’ plots decreased in farmlands, but increased in grasslands, while for non- carbon farmers there was a decline in tree diversity in both grasslands and farmlands. There were strong positive correlations between carbon density and tree diversity.
Keywords Carbon densitySpecies richnessShannon indexCarbon incentives GrasslandsFarmlands
Introduction
Humanity today faces important global challenges in the form of climate change [which is a result of increasing greenhouse gases (GHGs), mainly carbon dioxide (CO2)], and loss of biodiversity. The United Nations Framework Convention on Climate Change (UNFCCC)
C. Nakakaawa (&)J. AuneP. Vedeld
Noragric, Department of International Environment and Development Studies, Norwegian University of Life Sciences, P.O. Box 5003, 1432 A˚ s, Norway e-mail: [email protected]
J. Aune
e-mail: [email protected] P. Vedeld
e-mail: [email protected] DOI 10.1007/s11056-009-9180-5
and the Convention on Biological Diversity (CBD) aim at addressing these challenges, which have been attributed to global warming and deforestation (UN1992; WSSD2002).
The existence of potential synergies between the two conventions offers opportunities for implementing practices that aim at achieving the objectives of both conventions simulta- neously (Jacquemont and Caparro´s2002). However, several authors have raised concerns about the potential perverse incentives of the carbon market provided for under the Clean Development Mechanism (CDM) of the UNFCCC Kyoto Protocol (Caparro´s and Jacquemont2003; Huston and Marland2003; Koziell and Swingland2002; Matthews et al.
2002; Niesten et al.2002). The exclusion of avoided deforestation, with a limited focus on afforestation and reforestation (A/R) under the CDM during the first commitment period, has implications for existing carbon stocks and offers limited prospects of contributing substantially to biodiversity conservation (Niesten et al.2002; Totten et al.2003).
Several carbon offset projects targeting the CDM have been implemented, but few studies have been conducted to analyze the actual impacts of A/R project activities on tree diversity. Huston and Marland (2003) provide a qualitative discussion of the potential impacts of carbon sequestration on biodiversity, while Matthews et al. (2002) and recently Nelson et al. (2008) base their analyses on hypothetical landowner responses to carbon sequestration payments for developing land use scenarios and analyzing their potential impacts on biodiversity. A common conclusion is that the potential negative effects of payments for carbon sequestration services on biodiversity outweigh any possible gains.
Furthermore, the relationship between tree species diversity and biomass carbon sequestration is of great concern among land managers interested in sequestering a max- imum amount of carbon over a short time period. The question of whether diversity increases biomass productivity is therefore pertinent to current efforts to increase carbon stocks while ensuring the maintenance and enhancement of biological diversity (Kirby and Potvin2007; Liang et al.2007). Economic incentives for carbon sequestration programs are expected to result in a surplus of plantations containing fast growing exotic species, with potential negative impacts on biodiversity (Caparro´s and Jacquemont2003).
Based on a pilot carbon offset project implemented by smallholder farmers in south western Uganda, we analyze empirical inventory data for changes in carbon stocks and tree diversity, with and without carbon payments. Farmers who are paid for providing carbon sequestration services are referred to as ‘carbon farmers’, and those who are not paid are referred to as ‘non-carbon farmers’. The following research questions were addressed:
1. Are there significant differences in tree vegetation characteristics (tree density and basal area) and carbon densities before (2003), compared to after (2006) the carbon project was implemented on carbon and non-carbon farmers’ plots?
2. Are there significant differences in tree species diversity (richness and composition) in 2003 and 2006, and between carbon and non-carbon farmers’ plots?
3. How do changes in biomass carbon densities relate to changes in tree density, basal area and tree diversity measures?
The trees for global benefits (TGB) carbon sequestration project
The TGB carbon offset project is a pilot reforestation project involving small-scale rural farmers in the Bushenyi district, south western Uganda (Fig.1). The project can be cat- egorized as a development and conservation carbon sequestration project (May et al.
2004). It was introduced to the communities in 2003 as part of sustainable resource use and
biodiversity conservation efforts by the Environmental Conservation Trust of Uganda (ECOTRUST), in collaboration with CARE International. The project was implemented based on the ‘Plan Vivo’ approach1(www.planvivo.org). In addition to generating carbon offsets, this approach emphasizes the enhancement of rural livelihoods through tree planting, in order to restore degraded areas, provide employment, promote sustainable land use and ensure maintenance of sustainable timber resources. The tree planting practices promoted by the pilot project aim to conserve biodiversity by means of integrating indig- enous tree species into existing farming systems. Trees are planted in agroforestry systems, scattered on farmlands, along farm boundaries and/or established as woodlots. The main tree species include Maesopsis eminii, Erythrina abyssinica, Ricinus communis and Markhamia platycalyxwhich are planted at variable densities and managed under different rotation systems ranging from 12 to 50 years. In addition to the carbon benefits, tree growth in the pilot project area is expected to act as a buffer zone for Kashoya Kitomi and Maramagambo Forest Reserves, which have been cleared over time for crop cultivation.
Carbon contracts
Participating farmers enter individually into an agreement with ECOTRUST. Starting with 33 carbon farmers in 2003 (ECOTRUST et al.2007), the number of farmers with contracts who had received payments increased to over 170 by the end of 2008. Some farmers have now received their third payment (Kairu David, personal communication, 2008). Contract terms cover a 10-year period and specify the amount of carbon to be sold, the price per ton
Fig. 1 Location of study area and sample plots in the Bushenyi District, south western Uganda
1 For a detailed description of the Plan Vivo approach and a background to the project implementation see Orrego (2005).
to be paid (negotiated on a case-by-case basis), targets to be met within each of five established monitoring periods, and a schedule of payments. Disbursement of funds to farmers is conditional on the farmer meeting the targets within the specified time. In addition, farmers must set aside an additional 10% of their total carbon offset potential to cover shortfalls in the event that they fail to meet objectives (Assimwe, personal com- munication, 2005). The pilot phase (2003–2006) acted as a platform for expansion of the project to other parts of the country. The project has been extended to Hoima and Masindi Districts, where new ‘Plan Vivos’ are yet to be reviewed and new contracts signed, depending on the availability of more carbon buyers (ECOTRUST 2007; ECOTRUST et al.2007).
Carbon sales
The project has undergone a pilot phase of limited offset sales. The major buyer of the credits generated by the pilot project was an international, UK-based packaging company, TetraPak, who bought the first credits (11,200 tons of CO2) in December 2003 and an additional 9,000 tons of CO2 in 2004. Carbon sales increased steadily between 2004 and 2007. The carbon offset price has oscillated between USD 3.8 and 5.5/tCO2eq, with one particularly high priced transaction of USD 10.45/tCO2eq. The average volume weighted carbon price in the period 2004–2007 was about USD 4.5/tCO2eq (ECOTRUST et al.2007).
Materials and methods
Study site
The study was conducted in the Bushenyi District, south western Uganda (Fig.1). The topography is undulating with broad ridge tops and generally small valleys around central Bushenyi, but more steeply sloping in the southern and northern parts. The altitude of the study area ranges from about 910 to 1,950 m above sea level. The predominant soil types are yellowish brown sandy clay loams developed from phyllite, schist, and gneiss rocks.
Black sandy loams developed from volcanic ash and rift valley sediments are a common occurrence towards the north and north western parts of the district (Harrop 1960).
According to the United Nations Food and Agricultural Organization (FAO) classification, these soils are predominantly feralsols and acrisols (FAO et al.1998; IUSS et al.2006). The rainfall pattern is bimodal with an annual average of 1,200 mm. The average minimum and maximum temperatures are 14 and 26°C. Soil productivity is characterized largely as medium, owing to high annual rainfall, good soil depth, soil structure, and inherent soil fertility. These conditions are favorable for crop cultivation (Ruecker et al.2003).
The dominating land use is subsistence agriculture (80%). Other land use types include woodland (12%) and grasslands (5%) that are used mainly as grazing areas (Ministry of Water Lands and Environment 2003). Agriculture is the major source of income. The cultivated plots are small, ranging from 1 to 4 ha (Naughton-Treves et al.2007). Banana production is the main cash and subsistence crop in the area, and dominates the use of the cropped land (81%), followed by bean (4%), maize (4%), sweet potato (4%), finger millet (3%) and sorghum (2%) as the main food crops. Tea and coffee (bothCoffea arabicaand Coffea canephoraVarrobusta), as well as cotton and Irish potato constitute other common cash crops. Small scale dairy farming, bee keeping andEucalyptusgrowing are common among the wealthy members of the community.
Both anthropogenic and natural disturbances have shaped the tree species composition of the study area over time. The area is characterized by increasing economic growth and high population growth rates, estimated at about 3% per annum (UBOS 2004). These factors exert pressure on the remaining natural tropical high forests in the area. Land clearing for subsistence agriculture and pastures, and agricultural intensification have contributed towards reduced biodiversity (Naughton-Treves et al. 2007). Trees on farm- lands have either been retained on land that has been cleared, planted as boundary markers, scattered in croplands or planted as shade trees in banana-coffee agroforestry systems.
Under the carbon project, the areas allocated for tree planting are mainly those with steep slopes and generally on more marginal agricultural land.
Data sources, sampling procedure and interviews
Data was collected from carbon and non-carbon farmers’ plots in 2003 and 2006 to investigate differences which may be attributed to the TGB project interventions. The activities involved tree planting aimed at increasing carbon stocks and improving biodi- versity. In this study, we included non-carbon farmers as a control group to distinguish project-induced changes from changes induced by other factors, based on recommenda- tions in guidelines for conservation project evaluation (Ferraro and Pattanayak2006).
A reconnaissance survey was undertaken in December 2003 to identify farmers and locate their respective plots. Plots belonging to carbon farmers were identified with the help of CARE field staff, contact farmers and the various farm owners. We included all the carbon farmers who had trees on their farms in 2003. Non-carbon farmers’ plots were chosen randomly based on a household listing obtained from Local Council 1 chairpersons.
Assistance was sought from Local Council 1 chairpersons who were aware of households’
participation/non participation status in the carbon project. The final sample was dependent on the willingness of farmers to participate in the study. Both carbon and non-carbon farmers were interviewed to elicit information on total land holdings, carbon plot area, species preferences and reasons for the preferences, management and use of the various tree species. In addition, two focus group discussions were conducted with carbon and non- carbon farmers. Non-carbon farmers who owned enough land to qualify for enrollment in the carbon project were requested to explain why they had not done so.
Tree inventory
A total of 54 plots (21 carbon farmers and 33 non-carbon farmers) were sampled in 2003, and 60 plots (30 carbon farmers and 30 non-carbon farmers) in 2006. All plots were geo- referenced. Plot boundary global positioning system (GPS) coordinates were digitized in a Geographical Information System environment and the resulting polygons were used to estimate the area of the plots. The estimates were reconciled with data collected during farmer interviews. The size of the carbon plots varied from 0.2 to 7 ha, depending on the area allocated to tree growing by the different farmers, with an average of 1 ha per farmer.
We enumerated all the trees in very small plots, while for big plots, we established square sample plots of 0.25 m2. All trees species with diameter at breast height (DBH)C3 cm were identified. The DBH was measured using calipers. Height, bole length and crown width for all trees were also recorded. All data was collected with assistance from the National Biomass Study (NBS) team from the National Forestry Authority (NFA).
Data analysis
Different land uses generate different quantities and types of ecosystem services. A pre- liminary analysis of tree parameter data (diameter at breast height, bole length, height, and crown width) revealed significant differences between grasslands and farmlands for carbon and non-carbon farmers’ plots (Table1). We then controlled for land use and analyzed data from grasslands and farmlands separately. Tree density/basal area, species diversity, richness and composition, and carbon density were used as proxy measures for timber, biodiversity and carbon sequestration services.
Estimating biomass and carbon density
Above-ground biomass (AGB) for individual trees was computed based on generalized allometric equations stratified by diameter class (Velle1995). The data used for developing the equations was collected through destructive sampling as part of the Uganda National Biomass Study component aiming at estimating relationships between tree characteristics and tree weight. The general form of the equation based on Velle (1995) is:
lnðPWSÞ ¼aþblnðDÞ þclnðHTÞ þdlnðCRÞ
wherelnnatural logarithm,PWSpredicted weight of tree,Ddiameter at breast height,HT tree height,CRcrown width; anda,b,canddare constants which vary for three diameter class levels as indicated in Table2. Use of these generalized allometric equations is
Table 1 Tree growth parameters in grasslands and farmlands for carbon and non-carbon farmers, Bushenyi District, south western Uganda, 2003 and 2006
Tree attribute 2003 2006
Grasslands Farmlands tValue Pvalue Grasslands Farmlands tValue Pvalue Carbon farmers
Diameter at breast height (DBH-cm)
8.72 8.22 1.12 0.2632 8.58 7.43 2.29 0.0223*
Bole length (m)
2.98 3.13 -1.2 0.2322 2.47 2.26 2.08 0.0377*
Height (m) 6.62 6.01 2.76 0.0059** 5.26 4.33 4.24 \.0001**
Crown width (m)
2.11 2.21 -1.03 0.3022 1.97 1.59 3.57 0.0004**
Non-carbon farmers Diameter at breast height (DBH-cm)
9.20 11.41 -3.59 0.0003** 10.64 13.47 -3.09 0.0021**
Bole length (m)
2.85 3.98 -7 \.0001** 5.58 4.56 2.56 0.0107**
Height (m) 5.88 7.38 -5.67 \.0001** 8.85 8.49 0.71 0.4782
Crown width (m)
2.12 2.53 -3.54 0.0004** 2.53 2.83 -1.66 0.0982
* Indicates significance at 0.05, ** indicates significance at 0.01
justified because even in highly diverse systems, more than 95% of the variation in AGB carbon stocks is explained by DBH alone (Brown2002).
Individual tree AGB was then summed by plot (Mg plot-1) and results were stan- dardized to Mg ha-1 by dividing by the plot size. Below-ground root biomass was indi- rectly estimated as 20% of the above-ground tree biomass. This percentage is based on a range of values obtained from a predictive relationship established from extensive litera- ture surveys (Cairns et al. 1997; Mokany et al. 2006). The below-ground biomass was added to the estimated above-ground biomass to obtain an estimate of the total biomass, which was converted to carbon by assuming 50% carbon content (IPCC2003).
Assessing tree diversity
Common measures of diversity include counts of number of species (species richness) and use of indices such as Shannon–Wiener’s index (Shannon and Weaver1949) or the Gini–
Simpson index (Simpson1949), which further on are referred to as Shannon’s and Simpson’s diversity indices, respectively. Use of species richness as a measure of diversity is simple and straightforward, but it ignores the relative frequency of species. Simpson’s diversity index gives more weight to dominant species, while Shannon’s diversity index reflects both evenness and species richness, without favoring either dominant or rare species (Magurran 1988). Thus, we used tree species richness and Shannon’s diversity index based on species count and tree basal area, respectively to (a) quantify and compare temporal changes, and (b) compare carbon and non-carbon farmers’ plots in farmlands and grasslands.
We explored the use of Shannon’s diversity index based on species count, but the results were misleading for plots with high frequencies of trees from a limited number of species.
Since the interest of the carbon project is to increase biomass carbon stocks and to promote diversity, we recalculated Shannon’s diversity index based on species basal area. The explanatory power of Shannon’s diversity index based on basal area is superior to a measure based on species count. Basal area provides a better indication of the degree to which each species occupies a particular site and is a good measure of potential biomass growth (McMinn1992; Nangendo et al.2002).
Species diversity was calculated based on Shannon diversity index using the general formula:
H0¼ X pi lnpi
whereH0=Shannon’s diversity index, pi=species proportion (based either on species count or species basal area), and ln=natural logarithm. Using species basal area, the Shannon index was calculated as:
Table 2 Constants used for the different diameter classes to convert vegetation measures to above-ground biomass in south western Uganda
Diameter class Constants
a b c d
DBH\20 cm -0.85989 1.544561 0.50663 0.333346
20CDBHB60 cm -1.750891 1.943912 0.473731 0.245776
DBH[60 cm -2.166502 2.032931 0.31292 0.436348
Source: Velle (1995)
H0¼ X BAi=BAt
ln BAi=BAt
where BAi basal area of a particular speciesi in a plot, and BAt total basal area of all species in a plot.
Statistical analysis
All statistical analyses were performed using SAS (2003). The statistical significance of the differences between mean tree density, basal area, biomass carbon density, species richness and the Shannon diversity index of plots from different land uses (grassland and farm- lands), farmer categories (carbon and non-carbon farmers), and years (2003 and 2006) were tested using the two samplet-test (Zar1996). Principal Component Analysis (PCA) based on Euclidean distance ordinations was used to assess similarity in species compo- sition in 2003 and 2006 between carbon and non-carbon farmers’ plots. Pearson’s corre- lation analysis was used to explore the relationship between species diversity (species richness and Shannon diversity) and carbon density. Regression analysis was used to examine the relationship between variation in biomass carbon density and tree basal area, tree density, species richness and Shannon diversity.
Results
Tree vegetation characteristics and biomass carbon density Tree density, basal area and biomass carbon density in farmlands
In Table3tree density on carbon farmers’ plots was significantly higher than that on non- carbon farmers’ plots in both 2003 (P\0.01) and 2006 (P\0.05). In 2003, the basal area on carbon farmers’ plots was higher than that on non-carbon farmers’ plots, while in 2006, non-carbon farmers’ plots had higher basal area than carbon farmers’ plots. Carbon density on non-carbon farmers’ plots was significantly higher than that on carbon farmers’ plots in 2006. Comparisons between 2003 and 2006 reveal no significant differences in tree den- sity, although tree density increased on both carbon and non-carbon farmers’ plots. The recorded increase in tree density was higher on carbon farmers’ plots (6.2%) than on non- carbon farmers’ plots (5.3%). The tree basal area in 2006 on carbon farmers’ plots was significantly (P\0.01) lower than that in 2003, while there was no significant difference between 2003 and 2006 on non-carbon farmers’ plots, despite the decrease. From 2003 to 2006 there was a decrease in carbon density (-30%) on carbon farmers’ plots compared to an increase (52%) on non-carbon farmers’ plots (Table3). Most of the trees on carbon farmers’ plots were aged between 1 and 3 years.
Tree density, basal area and carbon density in grasslands
In 2003, the mean tree density and carbon densities on carbon farmers’ plots were higher (but not significantly) than those on non-carbon farmers’ plots located in grasslands, while in 2006, the mean tree density and carbon densities on non-carbon farmers’ plots was higher than that on carbon farmers’ plots (Table4).
There were no significant differences between 2003 and 2006 and between carbon and non-carbon farmers’ plots with respect to tree density, basal area and carbon density in grasslands. There was an increase in tree density and carbon density on both carbon and
Table 3 Tree density, basal area and carbon density in farmlands on carbon and non-carbon farmers’ plots, Bushenyi District, south western Uganda in 2003 and 2006
Variable Carbon
farmers
Non-carbon farmers
tValue Pvalue
Tree density (trees ha-1)
2003 94.73 (18.32) 36.39 (12.29) 2.72 0.0104**
2006 100.95 (18.32) 41.71 (12.63) 2.39 0.0226*
tValue -0.22 -0.29
Pvalue NS NS
Basal area (m2ha-1)
2003 0.41 (0.07) 0.33 (0.10) 0.56 NS
2006 0.18 (0.04) 0.26 (0.07) -1.05 NS
tValue 3.35 0.57
Pvalue 0.0021** NS
Carbon density (Mg ha-1)
2003 3.06 (0.55) 3.37 (1.12) -0.2 NS
2006 2.15 (0.39) 5.13 (1.43) -2.38 0.0231*
tValue 1.37 -0.98
Pvalue NS NS
Means are followed by standard error in parentheses. Means were compared with studentt-test; columns show comparisons between carbon and non-carbon farmers; rows show temporal comparisons;NSindicates not significant,*indicates significance at 0.05,**indicates significance at 0.01
NSindicates not significant
* Indicates significance at 0.05, ** indicates significance at 0.01
Table 4 Tree density, basal area and carbon density in grasslands on carbon and non-carbon farm- ers’ plots, Bushenyi District, south western Uganda in 2003 and 2006
Means are followed by standard error in parentheses. Means were compared with studentt-test;
columns indicate comparisons between carbon and non-carbon farmers; rows indicate temporal comparisons; all comparisons are not significant
Variable Carbon farmers
Non-carbon farmers
tValue
Tree density (trees ha-1)
2003 130.19 (58.05) 95.42 (21.47) 0.61
2006 135.11 (28.11) 136.00 (45.22) -0.01
tValue -0.08 -0.7
Basal area (m2ha-1)
2003 0.48 (0.11) 0.49 (0.11) -0.06
2006 0.38 (0.12) 0.53 (0.19) -0.58
tValue 0.58 -0.19
Carbon density (Mg ha-1)
2003 5.45 (1.79) 3.68 (0.91) 0.93
2006 7.01 (2.71) 9.52 (3.54) -0.49
tValue -0.48 -1.34
non-carbon farmers’ plots while in terms of basal area, there was a decrease on carbon farmers’ plots compared to an increase on non-carbon farmers’ plots. A higher increase in tree density was recorded on non-carbon farmers’ plots (43%) than on carbon farmers’
plots (4%).
Species composition, richness and diversity Tree species composition
A total of 1,862 trees from 62 species and 33 families were enumerated in 2003. Of these, 990 trees from 41 species and 26 families were encountered on carbon farmers’ plots and 872 trees from 50 species and 28 families were found on non-carbon farmers’ plots (Appendix). Most of the species exhibiting only one tree were found on non-carbon farmers’ plots (17 species on non-carbon farmers’ plots had a frequency of 1, compared to 6 species on carbon farmers’ plots). There was a decrease in the number of trees enu- merated in 2006, with a total of 1,524 trees from 68 species and 32 families being recorded.
Of these, 834 trees belonging to 49 species and 24 families were recorded on carbon farmers’ plots and 690 trees belonging to 46 species and 25 families were recorded on non- carbon farmers’ plots.
Euphorbiaceae was the richest family, with 10 species, followed by Fabaceae (5) and Moraceae (5) on carbon farmers’ plots. Despite its richness, Euphorbiaceae species appeared mostly at low frequencies and low basal areas, whereas Myrtaceae and Faba- ceae appeared with high frequencies for some species (Eucalyptus) and large basal areas.
Eucalyptus saligna was the dominant tree species, contributing 61 and 52% of the trees enumerated in 2003 on carbon and non-carbon farmers’ plots, respectively (Appendix). In 2006, the proportion ofEucalyptus salignaspecies on carbon farmers’ plots declined by 53%, while on non-carbon farmers’ plots it still constituted the most dominant tree species, accounting for 64% of the trees enumerated. The dominant tree species in 2006 on carbon farmers’ plots was Erythrina abyssinica, accounting for 17% of the trees enumerated, and there was a more even distribution among species on carbon farmers’
plots. Other common tree species includedRicinus communis,Bridelia micrantha,Ficus natalensis, Markhamia platcalyx, Psidium guava, Annona senegalensis, Mangifera indica,Persea Americana,Thevetia peruviana,Crassocephalum mannii andHarungana madagascariensis. These species constituted over 20 and 26% on carbon farmers’ plots in 2003 and 2006, respectively, while the corresponding percentages were 18 and 8% on non-carbon farmers’ plots.
Tree species richness and diversity in farmlands
Tree species richness and Shannon diversity index on farmlands were higher on carbon farmers’ plots than on non-carbon farmers’ plots in 2003 and 2006. There was a decrease in species richness and Shannon diversity on both carbon and non-carbon farmers’ plots, but there were no significant differences between 2003 and 2006 or between carbon and non- carbon farmers’ plots (Table5). Higher declines in species richness and Shannon diversity were recorded on carbon farmers’ plots (25 and 11%, respectively), compared to non- carbon farmers’ plots (11 and 8%, respectively).
Tree species richness and diversity in grasslands
Comparisons between carbon and non-carbon farmers’ reveal no significant differences in 2003 while in 2006, species richness and Shannon diversity on carbon farmers’ plots were significantly higher than on non-carbon farmers’ plots (species richness P=0.04 and Shannon index P=0.008). In grasslands there was an increase in species richness and Shannon diversity on carbon farmers’ plots compared to a decrease on non-carbon farmers’ plots (Table5), although none of these differences between 2003 and 2006 are significant.
Trends in tree species richness
We observed and recorded the trends in common tree species richness during both inventories and on both carbon and non-carbon farmers’ plots (Fig.2).
The short distance indicated in the PCA plot indicates similarities in species compo- sition on non-carbon farmers’ plots in 2003 and 2006, and differences in species com- position on carbon farmers’ plots. The plot also indicates similarities in species composition between carbon and non-carbon farmers’ plots in 2003.
Table 5 Tree species diversity in farmlands and grasslands on carbon and non-carbon farmers’
plots, Bushenyi District, south western Uganda in 2003 and 2006
Means are followed by standard error in parentheses. Means were compared with studentt-test;
columns indicate comparisons between carbon and non-carbon farmers; rows indicate temporal comparisons
* Indicates significance at 0.05,
** indicates significance at 0.01
Variable Carbon
farmers
Non-carbon farmers
tValue Pvalue
Farmlands
Species richness (count)
2003 6.00 (0.77) 4.50 (0.53) 1.63 0.112 2006 4.48 (0.56) 3.86 (0.48) 0.78 0.439
tValue 1.62 0.83
Pvalue 0.1152 0.4106
Shannon’s diversity index
2003 1.16 (0.18) 1.14 (0.11) 0.11 0.913 2006 1.03 (0.12) 1.05 (0.15) -0.09 0.927
tValue 0.65 0.53
Pvalue 0.5209 0.6001
Grasslands
Species richness (count)
2003 3.67 (0.94) 4.00 (0.67) -0.29 0.7716 2006 5.00 (0.67) 3.06 (0.55) 2.18 0.04*
tValue -1.15 1.08
Pvalue 0.2652 0.2907
Shannon’s diversity index
2003 0.65 (0.22) 0.77 (0.18) -0.43 0.6689 2006 1.14 (0.14) 0.54 (0.13) 2.92 0.0077**
tValue -1.88 1.09
Pvalue 0.0783 0.2854
Overall site diversity Index
2.55–2.79
Twenty-six (26) new species (not recorded in 2003) were identified on carbon farmers’
plots in 2006, of which 8 species were recorded only in grasslands, 15 species were recorded only on farmlands, and 3 species (Croton macrostachyus, Pancovia sp nr tur- binate, and Protea congensis) were recorded on both farmlands and grasslands. On non- carbon farmers’ plots, 18 new species were encountered, of which 9 species were recorded only in grasslands, 8 species were found only in farmlands, and only 1 species (Cordia Africana) was found on both farmlands and grasslands. The dominant tree species in grasslands and farmlands wereEuphorbia candelabrum(15%) andJatropha curcus(20%), respectively.
Relationship between tree diversity, basal area and biomass carbon density
Results from the correlation analysis (Table6) indicate significant positive correlations between carbon density and tree species diversity for data collected in 2006 from carbon farmers’ plots located in farmlands (species richness=0.49, P=0.02; Shannon Index r=0.51, P=0.01). Though statistically insignificant, other correlations indicate strong
Fig. 2 Principal Component Analysis species composition scatter plots for carbon and non-carbon farmers in the Bushenyi District, south western Uganda, 2003 and 2006
Table 6 Correlations between tree species diversity measures and carbon density on farmer plots in south western Uganda, 2003 and 2006
Tree species diversity measure Farmlands carbon density Grasslands carbon density Carbon
farmers
Non-carbon farmers
Carbon farmers
Non-carbon farmers Species richness (count)
2003 -0.29 (0.36) 0.02 (0.92) -0.02 (0.96) 0.14 (0.67)
2006 0.49 (0.02)* 0.52 (0.06) 0.60 (0.09) 0.33 (0.21)
Shannon’s diversity index
2003 -0.53 (0.07) 0.02 (0.93) 0.04 (0.91) -0.15 (0.67)
2006 0.51 (0.01) * 0.36 (0.21) 0.56 (0.12) -0.1 (0.72)
R-squared is in parentheses; significance of thePvalue is shown using * which indicates significance at 0.05
positive relationships between carbon density and tree species diversity (e.g. for grass- lands, species richness=0.60; and Shannon Index=0.56).
Simple regression analysis indicates that biomass carbon density increased linearly with basal area (Fig.3a) and tree density (Fig.3b).
Multiple regression analysis using basal area, tree density, species richness and Shannon index resulted in basal area being the only common significant predictor for carbon density in grasslands, farmlands and the combined model for grasslands and farmlands (Table7). The variation in biomass carbon density explained by tree density in the simple regression analysis is largely accounted for by tree basal area in the multiple regression model. This is supported by a significantly positive correlation between tree basal area and tree density (r=0.52, P\0.0001). In addition to basal area, species count and Shannon index were important predictors in farmlands and in the combined model for grasslands and farmlands.
R2 = 0.8358, P < 0.001
-10 0 10 20 30 40 50
0 1 2 3
Tree basal area
Carbon density, tonnes/ha
R2 = 0.2432, P < 0.001
0 5 10 15 20 25 30 35 40 45
0 200 400 600 800
Tree density, Trees/ha
Carbon density, tonnes/ha
a
b
Fig. 3 Relationships between carbon density (tonnes/ha) andatree basal area (m2/ha) andbtree density (trees/ha)
Table 7 Parameter estimates of the linear regression models of carbon density in grasslands and farmlands, in south western Uganda
Variable Grasslands and
farmlands (n=114)
Grasslands (n=45) Farmlands (n=69)
Intercept -0.805 (0.587) -1.726 (1.069) -0.105 (0.528)
Tree density 0.005 (0.003) 0.003 (0.005) 0.002 (0.004)
Basal area 15.409 (0.746)** 17.330 (1.254)** 11.989 (0.741)**
Species richness -0.602 (0.233)* -0.252 (0.474) -0.689 (0.197)**
Shannon index 2.105 (0.958)* 1.069 (1.860) 2.797 (0.842)**
R2 0.85 0.88 0.82
Standard errors in parentheses NSindicates not significant
* Indicates significance at 0.05, ** indicates significance at 0.01
Discussion
Tree density, basal area and carbon density
The tree densities on carbon farmers’ plots (95–136 trees ha-1) are comparable to those found in agricultural landscapes (147 trees ha-1) with similar farming practices in Tan- zania (Munishi et al. 2008), but where farmers were not receiving cash payments as incentives for tree growing. Besides cash returns or economic considerations, farmers’
decisions to retain or plant trees are framed by an array of social, biophysical and available policy options which have consequences for their livelihoods (Pattanayak et al.2003; Vosti et al.2005). A detailed analysis and consideration of the interplay of these different factors is important, but beyond the scope of this paper.
The mean basal areas in our study (Tables3,4) are very low compared to the mean (21 m2ha-1) reported by Munishi et al. (2008). This can be attributed to the fact that most of the trees in the study area are still young. This fact is also evident from basic vegetation structural characteristics, such as the dominance of trees of small diameter.
The mean annual increment (MAI) in wood carbon density (above and below ground) on non-carbon farmers’ plots in both grasslands (1.9 Mg C ha-1) and farmlands (0.6 Mg C ha-1) was higher than that on carbon farmers’ plots, but similar to figures reported for dry miombo woodlands in Zambia (0.6–1.1 Mg C ha-1) and Mozambique (0.8 Mg C ha-1) after 16–35 years after abandonment (Chidumayo1997; Williams et al.
2008). Carbon sequestration potential in above-ground biomass through agroforestry interventions is estimated at 59 Mg C ha-1 for Sub-Saharan Africa (Houghton et al.
1993). Considering the shorter time interval in this study, there is a potential for further increase in carbon stocks, but this will also depend on the tree management practices employed.
Species composition, richness and diversity
Mean tree diversity in our sample farms is generally low compared to farms in Tanzania where a Shannon diversity index of 2.7 was recorded (Munishi et al.2008). This value is within the range (2.5–2.8) of the overall diversity index for the project area. Compared to our results, this indicates that there is a potential for increasing diversity in the study area, given the similarities in production systems.
Considering the effect of carbon payments, our a prioriexpectation was that carbon farmers’ plots would exhibit an increase in tree species richness and diversity compared to plots owned by non-carbon farmers, due to the respective presence or absence of incentives. However, the results show a decrease in species richness on farmland carbon farmers’ plots. This may be explained by the fact that no deliberate efforts are made to reward carbon farmers for increased diversity. When farmers are paid for carbon sequestration, there is no explicit consideration of diversity in tree species planted. In practice, there has been a tendency for farmers to plant one or two tree species, whose planting material is readily available at a low cost. For instance, the popularity of Maesopsis eminii species was attributed to its fast growth, readily available planting materials, ease of propagation, and compatibility with existing agricultural practices. Traditionally, during land clearing, farmers have retained tree species based on their multiple economic, social and cultural values (Kakudidi 2004;
Katende et al. 1995).
It is not enough to provide incentives for encouraging changes in land use practices aimed at increasing carbon stocks, without accompanying rewards for increasing diversity.
Deliberate efforts by ECOTRUST to promote indigenous valuable timber tree species have shifted some carbon farmers’ focus from exotic species such asEucalyptusto indigenous species. Eligibility to receive carbon payments is now restricted to planting indigenous tree species. Farmers do not earn any payments for the carbon accumulated by their trees when they plant exotic species such as Eucalyptus and Pine. Merely rewarding farmers for incremental carbon stocks may eventually lead to a simple shift of focus from exotic monocultures to indigenous monocultures. The decline in the proportion of Eucalyptus salignaon carbon farmers’ plots compared to an increase on non-carbon farmers’ plots, can be attributed to the project’s focus on indigenous trees species. For instance, from the field observations, the most common tree species planted by carbon farmers isMaesopsis eminii2and this may replaceEucalyptus on carbon farmers’ plots as the most dominant species in the near future.
Farmers do not have access to funds required for establishing tree crops and do not receive any payment for maintaining original tree vegetation. Only new planting is awarded. Since they have no incentive to maintain existing trees, some farmers tend to cut down trees which they referred to as ‘‘non-carbon trees’’, in order to meet cash require- ments for establishing ‘‘carbon trees’’. This tendency is manifested by the decrease in carbon density, despite the increase in tree density (Table3), and the increase in trees in the lower diameter class on carbon farmers’ plots, compared to a decrease and an increase in frequency of trees in higher diameter classes on non-carbon farmers’ plots (results not presented). This trend has implications for diversity conservation and accumulation of biomass carbon stocks. It may be necessary to provide the initial planting material or some financing in the form of baseline payments as in the silvopastoral project, or front loading payments to meet establishment costs as in Costa Rica’s reforestation PES program (Pagiola et al.2007).
With markets for timber and payments for carbon sequestration services, the internal- ized costs are those associated with timber production and carbon sequestration. The cost of increasing biodiversity is not internalized since there is no incentive for enumerating biodiversity value when assessing payments for carbon sequestration. Optimizing for carbon sequestration will in many cases have a negative impact on biodiversity values (Caparros et al.2007). Co-funding to cater for the biodiversity value can be sought through other programs such as the Global Environment Facility (GEF) Operational Program on Integrated Ecosystem Management (OP#12). This program provides a comprehensive framework facilitating inter-sectoral and participatory approaches to natural resource management planning and implementation within the context of sustainable development (Global Environment Facility 2000). In addition, some buyers in other existing carbon markets are willing to preferentially buy or pay higher prices for carbon credits if these are associated with measurable conservation and development benefits (Hamilton et al.2007).
Myers and Kent (2001) note that direct payments may not always constitute ‘‘silver bullets’’.
2 Maesopsisdoes not appear as the most common tree species on carbon farmers’ plots because most of the trees were still young. By the time of data collection, the trees planted by some carbon farmers had not yet attained the minimum DBH of 3 cm considered in our study. As a result, some species planted by carbon farmers are under represented in the species list.
Broader policy interventions aimed at removing perverse incentives or subsidies that encourage loss of biodiversity, may be required. In the current carbon contracts, farmers receive between $3.8 and $5.5/tCO2eq (ECOTRUST et al. 2007), which is lower than the range of observed average weighted prices ($6.8–$8.2/tCO2e) in the CDM market (Hamilton et al. 2008). It may not be possible to increase the current price to cater for biodiversity values, given the low prices in the voluntary target market for this project.
Focus group discussions revealed that other programs in the area, such as the Inter- national Small Group Tree Planting program (TIST) and subsidies for plantation forestry established under the Sawlog Production Grant Scheme (SPGS),3also have an impact on farmers’ decisions. A non-carbon farmer with experience in tree planting, who had ben- efited from SPGS funding, made a simple comparison between carbon payments and the subsidy provided under the SPGS scheme. Basing his argument on the ‘‘time value’’ of money, the reforestation subsidy provided under the SPGS (estimated at half the cost of establishing one hectare of pine plantations, namely Ug.Shs.600,000, equivalent to $3004) is a more attractive incentive than the additional carbon returns received over a period of 10 years, which are conditional on planting indigenous tree species.
Farmers with access to information and resources, who are in a position to make unconstrained decisions, have therefore opted to forego the opportunity to earn income from carbon payments. They have invested instead in Pine and Eucalyptus plantations which offer a faster return compared to the indigenous tree species emphasized by the project. In addition, many farmers lack the required technical knowledge and management experience to engage in carbon sequestration projects.
Relationship between tree diversity, basal area and carbon density
The relationship between biodiversity and carbon sequestration is complicated and depends on the nature of the production system and the reforestation (or afforestation) policies adopted. Some authors argue that biodiversity conservation tends to increase carbon sequestration, and production systems that support more biodiversity are also better at carbon sequestration (Alavalapati et al. 2002). Results in this study do not indicate any such clear relationships. Other studies (Hooper and Vitousek 1997; Tilman et al. 2001) have reported linear increases in productivity with increasing species diversity in grass- lands. In a recent study in Panama, Kirby and Potvin (2007) report a positive correlation between biodiversity and carbon storage across land use types, and emphasize the potential for a single project to optimize both carbon storage and biodiversity conservation. How- ever, if reforestation policies involve replacing multi-species and biodiversity rich forests with mono-species and less biodiverse systems, then a policy oriented towards carbon sequestration may conflict with one oriented toward preserving biodiversity (Caparro´s and Jacquemont2003).
3 The Sawlog Production Grant Scheme (SPGS) aims to increase private sector participation in plantation forestry development and ensure a sustainable timber supply in future. It focuses on the promotion of Pine (70%) andEucalyptus(30%). In some areas, farmers have opted to plant these species, given the readily available planting materials, knowledge of the establishment and management practices and market for the products (mainly timber, poles and firewood).
4 At an exchange rate of 1 USD equivalent to 2000 Uganda Shillings (December 2003).
The Clean Development Mechanism (CDM) framework which provides opportunities for developing countries to participate in carbon trading, requires consideration of the impact of carbon sequestration on biodiversity (UNFCCC2004). However, in its current form, there is no explicit value attached to biodiversity. In order to maximize carbon returns, it is more economically attractive to establish short rotation mono-species plan- tations rather than maintaining or establishing long rotation indigenous multi-species tree production systems.
Conclusions and implications for future biodiversity conservation and policy formulation
Carbon trading through afforestation and reforestation projects presents opportunities for reducing atmospheric carbon emissions and mitigating against climate change. It also has implications for the management of tree diversity. Results from this study were to some extent influenced by the initial project design, which did not provide incentives for maintaining baseline carbon stocks, or any additional rewards for enhancing tree diversity. There was a respective decrease and increase in carbon density on carbon and non-carbon farmers’ plots in farmlands, while there was an increase in carbon density on both carbon and non-carbon farmers’ plots in grasslands. We found no significant dif- ferences in tree diversity for plots located in farmlands, while for grasslands there was a significant difference in species diversity between carbon and non-carbon farmers in 2006. Temporal comparisons (2003–2006) indicate that on carbon farmers’ plots, there was an increase in tree diversity in grasslands and a decrease in farmlands, while for non-carbon farmers’ plots, there was a decline in tree diversity in both grasslands and farmlands. There were strong positive correlations between carbon density and tree diversity.
An increase in species richness and Shannon diversity index on carbon farmers’ plots in grasslands compared to a decline in farmlands is an indication of a high potential tree planting niche. Grassland areas located near protected areas and used mainly for grazing could be given priority in biodiversity and carbon enhancing activities. Grasslands would also serve as important buffers for the remaining natural forests in the area that are experiencing increasing pressure from the increasing population. Integrating trees within the agricultural landscape (grasslands) would mitigate against the effects of degradation from grazing animals and to contribute to increased production through temperature amelioration and fodder provision (Manning et al.2006).
Agricultural land which is already dominated by monoculture production systems is less likely to warrant attention because the chances of reconciling farming and biodiversity conservation needs in such systems are low. The cultivation of bananas (on mainly monoculture plantations), and a few remaining tree-crop mixtures of banana-coffee and/or banana-ficus systems, constitute the major source of income and subsistence livelihood.
Banana-coffee agroforestry is regarded by farmers as an important tree-crop mixture that constitutes a sustainable cash production system. About a third of the tree species found on farmers’ plots in this study were also found in coffee-banana agroforestry systems (Hemp 2006).
The multilayered vegetation structure and the floral composition of coffee-banana agroforestry systems, traditionally common in central Uganda and among the Chagga in Tanzania, mimic a tropical montane forest more than other agricultural systems, and maintain high levels of both wild and cultivated biodiversity (Perfecto et al.1996; Hemp
2006). This land use system provides opportunities for bundling multiple ecosystem ser- vices and offers much greater conservation value than the monoculture banana plantations which currently dominate the agricultural landscape in south western Uganda. Some studies in coffee-banana agroforestry systems have examined their carbon sequestration potential and contribution to biodiversity conservation (Me´ndez et al.2007; Perfecto et al.
2003). Offering payments for bundled ecosystem services (Wendland et al.2009), rather than for one environmental service such as carbon sequestration, can help avoid problems related to loss of other ecosystem services.
A focus on such tree-crop mixtures, which to a certain extent already exist in the area, may achieve better results than the pilot project approach which focused on high value timber tree species which were not currently found on farmers’ plots. However, it is not sufficient to identify systems that fulfill both carbon sequestration and biodiversity con- servation goals; it is also necessary to consider if farmers, from economic and socio- cultural perspectives, are able and/or willing to adopt appropriate land use strategies. This issue is beyond the scope of this paper, but is addressed elsewhere (Nakakaawa and Vedeld 2009). Future research studies should investigate the possibility of carbon losses due to leakage as a result of displaced farmers’ activities and tree cutting, especially by carbon farmers to finance their carbon project activities. When scaling up the project in future, the flaws in incentive design should also be addressed.
The analysis presented in this article is limited by a lack of long-term biomass monitoring data for agro-ecosystems for the period before project implementation (i.e. 2000–2003).
Results are also limited by the fact that it may be too early to arrive at any firm conclusions on the impact of carbon payments in the study area. Sierra and Russman (2006) found no difference in forest cover between non-Payment for Environmental Services (PES) and PES recipient farms after 5 years of PES contracts. The period in this study (3 years) is even shorter. The interpretation of the results should be considered in the light of these limitations;
nevertheless the outcomes have implications for future A/R carbon sequestration project design and implementation.
Acknowledgments This study was funded by the Norwegian Agency for Development Cooperation (NORAD) and SIDA/SAREC through the Faculty of Forestry and Nature Conservation and School of Graduate Studies, respectively, at Makerere University, Uganda. We thank the land owners of the areas involved in this study for allowing us to work on their land. We also thank Hillary Beshekya and Wilson Turyahikayo for field assistance and particularly for putting us in touch with the various farmers. We are grateful to the team which was involved in data collection (Christopher Mukwaya, Grace Agaba, John Ongodia, Patrick Turyatunga, Patrick Byakagaba and Simon Kizza). We acknowledge assistance with data management from John Ayongera, Edward Senyonjo and George Kityo. We are grateful for the comments provided on an earlier draft by Dr. Md. Danesh Miah and an anonymous reviewer.
Appendix See Table8.
Table8Listofspeciesrecordedincarbonandnon-carbonfarmers’plotsintheBushenyiDistrict,southwesternUganda FamilynameScientificnameNumberoftrees(count)Basalarea(m2ha-1)Origin CarbonNon-carbonCarbonNon-carbon 20032006200320062003200620032006 AnacardiaceaeMangiferaindica7101690.130.140.310.32Exo Rhusnatalensis20100000Ind Rhusvulgaris200.010Ind AnnonaceaeAnnonasenegalensis1516100.050.0600Ind ApocynaceaeFuntumiaelastica0100.04Ind Plumeriaalba1000Ind Tabernaemontanaholstii400.030Ind Thevetiaperuviana1124020.020.0800Exo AraliaceaePolysciasfulva0200.04Ind ArecaceaePhoenixreclinata10300.0300.070Ind AsteraceaeCrassocephalummannii111100.0300.010Exo Vernoniaamygdalina2159400.030.060.01Ind Vernoniaauriculifera051000.0100Ind BignoniaceaeKigeliaafricana0200Ind Markhamiaplatycalyx182427140.060.080.290.35Ind Spathodeacampanulata240.110.05Exo BixaceaeBixaorellana400.010Exo BoraginaceaeCordiaafricana0600.07Exo Ehretiacymosa1210.030Exo CaesalpiniaceaeCassiadidymobotrya0100Ind Cassiasiamea100.010Exo Cassiaspectabilis02500.04Exo CasuarinaceaeCasuarinaequisetifolia200.010Exo
Table8continued FamilynameScientificnameNumberoftrees(count)Basalarea(m2ha-1)Origin CarbonNon-carbonCarbonNon-carbon 20032006200320062003200620032006 CecropiaceaeMyrianthusarboreus20900.0400.150Ind CelastraceaeCassineaethiopica01600.07Ind Maytenusovata2000Ind Maytenussenegalensis0400.13Ind ClusiaceaeHarunganamadagascariensis1014430.060.040.010.01Ind CombretaceaeCombretummolle0211000.010Ind Terminaliasuperba0100.02Ind CupressaceaeCupressuslusitanica130.040.13Exo DracaenaceaeDracaenaafromontana10700.0300.020Ind Dracaenasteudneri200.10Ind EuphorbiaceaeBrideliamicrantha2731860.160.130.10.05Ind Brideliascleroneura37200.010.010.010Ind Crotonmacrostachyus061600.0200.04Ind Crotonmegalocarpus40300000Ind Euphorbiacalamiformis0100Ind Euphorbiacandelabrum03000.04Ind Heveabraziliensis013000.030.090Exo Jatrophacurcus0400200.0700Exo Ricinodendronheudelotii1000Ind Ricinuscommunis57292880.110.050.060.02Ind Sapiumellipticum0414800.140.320.02Ind